Fix a traceback with "check --only spot" when the "spot" check is unconfigured (#857).

Reviewed-on: #852
Reviewed-by: Dan Helfman <witten@torsion.org>

.dockerignore

@@ -0,0 +1,2 @@
+.git
+.tox

.drone.yml

@@ -1,18 +0,0 @@
-pipeline:
-  build:
-    image: python:${PYTHON_VERSION}-alpine${ALPINE_VERSION}
-    pull: true
-    commands:
-      - pip install tox
-      - tox
-      - apk add --no-cache borgbackup
-      - tox -e end-to-end
-
-matrix:
-  ALPINE_VERSION:
-    - 3.7
-    - 3.8
-  PYTHON_VERSION:
-    - 3.5
-    - 3.6
-    - 3.7

.eleventy.js

@@ -0,0 +1,50 @@
+const pluginSyntaxHighlight = require("@11ty/eleventy-plugin-syntaxhighlight");
+const codeClipboard = require("eleventy-plugin-code-clipboard");
+const inclusiveLangPlugin = require("@11ty/eleventy-plugin-inclusive-language");
+const navigationPlugin = require("@11ty/eleventy-navigation");
+
+module.exports = function(eleventyConfig) {
+    eleventyConfig.addPlugin(pluginSyntaxHighlight);
+    eleventyConfig.addPlugin(inclusiveLangPlugin);
+    eleventyConfig.addPlugin(navigationPlugin);
+    eleventyConfig.addPlugin(codeClipboard);
+
+    let markdownIt = require("markdown-it");
+    let markdownItAnchor = require("markdown-it-anchor");
+    let markdownItReplaceLink = require("markdown-it-replace-link");
+
+    let markdownItOptions = {
+        html: true,
+        breaks: false,
+        linkify: true,
+        replaceLink: function (link, env) {
+            if (process.env.NODE_ENV == "production") {
+                return link;
+            }
+            return link.replace('https://torsion.org/borgmatic/', 'http://localhost:8080/');
+        }
+    };
+    let markdownItAnchorOptions = {
+        permalink: markdownItAnchor.permalink.headerLink()
+    };
+
+    eleventyConfig.setLibrary(
+        "md",
+        markdownIt(markdownItOptions)
+            .use(markdownItAnchor, markdownItAnchorOptions)
+            .use(markdownItReplaceLink)
+            .use(codeClipboard.markdownItCopyButton)
+    );
+
+    eleventyConfig.addPassthroughCopy({"docs/static": "static"});
+
+    eleventyConfig.setLiquidOptions({dynamicPartials: false});
+
+    return {
+        templateFormats: [
+          "md",
+          "txt"
+        ]
+    }
+};
+

.flake8

@@ -0,0 +1 @@
+select = Q0

.gitea/issue_template/bug_template.yaml

@@ -0,0 +1,77 @@
+name: "Bug or question/support"
+about: "For filing a bug or getting support"
+body:
+  - type: textarea
+    id: problem
+    attributes:
+      label: What I'm trying to do and why
+    validations:
+      required: true
+  - type: textarea
+    id: repro_steps
+    attributes:
+      label: Steps to reproduce
+      description: Include (sanitized) borgmatic configuration files if applicable.
+    validations:
+      required: false
+  - type: textarea
+    id: actual_behavior
+    attributes:
+      label: Actual behavior
+      description: Include (sanitized) `--verbosity 2` output if applicable.
+    validations:
+      required: false
+  - type: textarea
+    id: expected_behavior
+    attributes:
+      label: Expected behavior
+    validations:
+      required: false
+  - type: textarea
+    id: notes
+    attributes:
+      label: Other notes / implementation ideas
+    validations:
+      required: false
+  - type: input
+    id: borgmatic_version
+    attributes:
+      label: borgmatic version
+      description: Use `sudo borgmatic --version` or `sudo pip show borgmatic | grep ^Version`
+    validations:
+      required: false
+  - type: input
+    id: borgmatic_install_method
+    attributes:
+      label: borgmatic installation method
+      description: e.g., pip install, Debian package, container, etc.
+    validations:
+      required: false
+  - type: input
+    id: borg_version
+    attributes:
+      label: Borg version
+      description: Use `sudo borg --version`
+    validations:
+      required: false
+  - type: input
+    id: python_version
+    attributes:
+      label: Python version
+      description: Use `python3 --version`
+    validations:
+      required: false
+  - type: input
+    id: database_version
+    attributes:
+      label: Database version (if applicable)
+      description: Use `psql --version` / `mysql --version` / `mongodump --version` / `sqlite3 --version`
+    validations:
+      required: false
+  - type: input
+    id: operating_system_version
+    attributes:
+      label: Operating system and version
+      description: On Linux, use `cat /etc/os-release`
+    validations:
+      required: false

.gitea/issue_template/config.yaml

@@ -0,0 +1 @@
+blank_issues_enabled: true

.gitea/issue_template/feature_template.yaml

@@ -0,0 +1,15 @@
+name: "Feature"
+about: "For filing a feature request or idea"
+body:
+  - type: textarea
+    id: request
+    attributes:
+      label: What I'd like to do and why
+    validations:
+      required: true
+  - type: textarea
+    id: notes
+    attributes:
+      label: Other notes / implementation ideas
+    validations:
+      required: false

.gitea/workflows/build.yaml

@@ -0,0 +1,28 @@
+name: build
+run-name: ${{ gitea.actor }} is building
+on:
+  push:
+    branches: [main]
+
+jobs:
+  test:
+    runs-on: host
+
+    steps:
+      - uses: actions/checkout@v4
+      - run: scripts/run-end-to-end-tests
+
+  docs:
+    needs: [test]
+    runs-on: host
+    env:
+      IMAGE_NAME: projects.torsion.org/borgmatic-collective/borgmatic:docs
+
+    steps:
+      - uses: actions/checkout@v4
+      - run: podman login --username "$USERNAME" --password "$PASSWORD" projects.torsion.org
+        env:
+          USERNAME: "${{ secrets.REGISTRY_USERNAME }}"
+          PASSWORD: "${{ secrets.REGISTRY_PASSWORD }}"
+      - run: podman build --tag "$IMAGE_NAME" --file docs/Dockerfile --storage-opt "overlay.mount_program=/usr/bin/fuse-overlayfs" .
+      - run: podman push "$IMAGE_NAME"

.gitignore

@@ -2,8 +2,10 @@
 *.pyc
 *.swp
 .cache
-.coverage
+.coverage*
 .pytest_cache
 .tox
-build
-dist
+__pycache__
+build/
+dist/
+pip-wheel-metadata/

AUTHORS

@@ -7,6 +7,8 @@ Johannes Feichtner: Support for user hooks
 Michele Lazzeri: Custom archive names
 Nick Whyte: Support prefix filtering for archive consistency checks
 newtonne: Read encryption password from external file
-Robin `ypid` Schneider: Support additional options of Borg
+Robin `ypid` Schneider: Support additional options of Borg and add validate-borgmatic-config command
 Scott Squires: Custom archive names
 Thomas LÉVEIL: Support for a keep_minutely prune option. Support for the --json option
+
+And many others! See the output of "git log".

MANIFEST.in

@@ -1 +1,2 @@
 include borgmatic/config/schema.yaml
+graft sample/systemd

README.md

@@ -1,534 +1,156 @@
 ---
 title: borgmatic
-permalink: borgmatic/index.html
+permalink: index.html
 ---
-## Overview
 
-<img src="https://projects.torsion.org/witten/borgmatic/raw/branch/master/static/borgmatic.png" width="150px" style="float: right; padding-left: 1em;">
+## It's your data. Keep it that way.
 
-borgmatic is a simple Python wrapper script for the
-[Borg](https://www.borgbackup.org/) backup software that initiates a backup,
-prunes any old backups according to a retention policy, and validates backups
-for consistency. The script supports specifying your settings in a declarative
-configuration file rather than having to put them all on the command-line, and
-handles common errors.
+<img src="docs/static/borgmatic.png" alt="borgmatic logo" width="150px" style="float: right; padding-left: 1em;">
 
-Here's an example config file:
+borgmatic is simple, configuration-driven backup software for servers and
+workstations. Protect your files with client-side encryption. Backup your
+databases too. Monitor it all with integrated third-party services.
+
+The canonical home of borgmatic is at <a href="https://torsion.org/borgmatic">https://torsion.org/borgmatic/</a>
+
+Here's an example configuration file:
 
 ```yaml
-location:
-    # List of source directories to backup. Globs are expanded.
-    source_directories:
-        - /home
-        - /etc
-        - /var/log/syslog*
+# List of source directories to backup.
+source_directories:
+    - /home
+    - /etc
 
-    # Paths to local or remote repositories.
-    repositories:
-        - user@backupserver:sourcehostname.borg
+# Paths of local or remote repositories to backup to.
+repositories:
+    - path: ssh://k8pDxu32@k8pDxu32.repo.borgbase.com/./repo
+      label: borgbase
+    - path: /var/lib/backups/local.borg
+      label: local
 
-    # Any paths matching these patterns are excluded from backups.
-    exclude_patterns:
-        - /home/*/.cache
+# Retention policy for how many backups to keep.
+keep_daily: 7
+keep_weekly: 4
+keep_monthly: 6
 
-retention:
-    # Retention policy for how many backups to keep in each category.
-    keep_daily: 7
-    keep_weekly: 4
-    keep_monthly: 6
+# List of checks to run to validate your backups.
+checks:
+    - name: repository
+    - name: archives
+      frequency: 2 weeks
 
-consistency:
-    # List of consistency checks to run: "repository", "archives", or both.
-    checks:
-        - repository
-        - archives
+# Custom preparation scripts to run.
+before_backup:
+    - prepare-for-backup.sh
+
+# Databases to dump and include in backups.
+postgresql_databases:
+    - name: users
+
+# Third-party services to notify you if backups aren't happening.
+healthchecks:
+    ping_url: https://hc-ping.com/be067061-cf96-4412-8eae-62b0c50d6a8c
 ```
 
-borgmatic is hosted at <https://torsion.org/borgmatic> with [source code
-available](https://projects.torsion.org/witten/borgmatic). It's also mirrored
-on [GitHub](https://github.com/witten/borgmatic) for convenience.
+borgmatic is powered by [Borg Backup](https://www.borgbackup.org/).
 
-Want to see borgmatic in action? Check out the <a
-href="https://asciinema.org/a/203761" target="_blank">screencast</a>.
+## Integrations
 
-<script src="https://asciinema.org/a/203761.js" id="asciicast-203761" async></script>
+<a href="https://www.postgresql.org/"><img src="docs/static/postgresql.png" alt="PostgreSQL" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
+<a href="https://www.mysql.com/"><img src="docs/static/mysql.png" alt="MySQL" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
+<a href="https://mariadb.com/"><img src="docs/static/mariadb.png" alt="MariaDB" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
+<a href="https://www.mongodb.com/"><img src="docs/static/mongodb.png" alt="MongoDB" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
+<a href="https://sqlite.org/"><img src="docs/static/sqlite.png" alt="SQLite" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
+<a href="https://healthchecks.io/"><img src="docs/static/healthchecks.png" alt="Healthchecks" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
+<a href="https://cronitor.io/"><img src="docs/static/cronitor.png" alt="Cronitor" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
+<a href="https://cronhub.io/"><img src="docs/static/cronhub.png" alt="Cronhub" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
+<a href="https://www.pagerduty.com/"><img src="docs/static/pagerduty.png" alt="PagerDuty" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
+<a href="https://ntfy.sh/"><img src="docs/static/ntfy.png" alt="ntfy" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
+<a href="https://grafana.com/oss/loki/"><img src="docs/static/loki.png" alt="Loki" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
+<a href="https://github.com/caronc/apprise/wiki"><img src="docs/static/apprise.png" alt="Apprise" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
+<a href="https://www.borgbase.com/?utm_source=borgmatic"><img src="docs/static/borgbase.png" alt="BorgBase" height="60px" style="margin-bottom:20px; margin-right:20px;"></a>
 
 
-## Installation
+## Getting started
 
-To get up and running, first [install
-Borg](https://borgbackup.readthedocs.io/en/latest/installation.html), at
-least version 1.1.
+Your first step is to [install and configure
+borgmatic](https://torsion.org/borgmatic/docs/how-to/set-up-backups/).
 
-Then, run the following command to download and install borgmatic:
+For additional documentation, check out the links above (left panel on wide screens)
+for <a href="https://torsion.org/borgmatic/#documentation">borgmatic how-to and
+reference guides</a>.
 
-```bash
-sudo pip3 install --upgrade borgmatic
-```
 
-Note that your pip binary may have a different name than "pip3". Make sure
-you're using Python 3, as borgmatic does not support Python 2.
+## Hosting providers
 
+Need somewhere to store your encrypted off-site backups? The following hosting
+providers include specific support for Borg/borgmatic—and fund borgmatic
+development and hosting when you use these referral links to sign up:
 
-### Other ways to install
-
- * [Docker image](https://hub.docker.com/r/monachus/borgmatic/)
- * [Another Docker image](https://hub.docker.com/r/b3vis/borgmatic/)
- * [Debian](https://tracker.debian.org/pkg/borgmatic)
- * [Ubuntu](https://launchpad.net/ubuntu/+source/borgmatic)
- * [Fedora](https://bodhi.fedoraproject.org/updates/?search=borgmatic)
- * [Arch Linux](https://aur.archlinux.org/packages/borgmatic/)
- * [OpenBSD](http://ports.su/sysutils/borgmatic)
-<br><br>
-
-
-## Configuration
-
-After you install borgmatic, generate a sample configuration file:
-
-```bash
-sudo generate-borgmatic-config
-```
-
-If that command is not found, then it may be installed in a location that's
-not in your system `PATH`. Try looking in `/usr/local/bin/`.
-
-This generates a sample configuration file at /etc/borgmatic/config.yaml (by
-default). You should edit the file to suit your needs, as the values are just
-representative. All fields are optional except where indicated, so feel free
-to ignore anything you don't need.
-
-You can also have a look at the [full configuration
-schema](https://projects.torsion.org/witten/borgmatic/src/master/borgmatic/config/schema.yaml)
-for the authoritative set of all configuration options. This is handy if
-borgmatic has added new options since you originally created your
-configuration file.
-
-
-### Encryption
-
-Note that if you plan to run borgmatic on a schedule with cron, and you
-encrypt your Borg repository with a passphrase instead of a key file, you'll
-either need to set the borgmatic `encryption_passphrase` configuration
-variable or set the `BORG_PASSPHRASE` environment variable. See the
-[repository encryption
-section](https://borgbackup.readthedocs.io/en/latest/quickstart.html#repository-encryption)
-of the Quick Start for more info.
-
-Alternatively, the passphrase can be specified programatically by setting
-either the borgmatic `encryption_passcommand` configuration variable or the
-`BORG_PASSCOMMAND` environment variable. See the [Borg Security
-FAQ](http://borgbackup.readthedocs.io/en/stable/faq.html#how-can-i-specify-the-encryption-passphrase-programmatically)
-for more info.
-
-
-## Usage
-
-### Initialization
-
-Before you can create backups with borgmatic, you first need to initialize a
-Borg repository so you have a destination for your backup archives. (But skip
-this step if you already have a Borg repository.) To create a repository, run
-a command like the following:
-
-```bash
-borgmatic --init --encryption repokey
-```
-
-This uses the borgmatic configuration file you created above to determine
-which local or remote repository to create, and encrypts it with the
-encryption passphrase specified there if one is provided. Read about [Borg
-encryption
-modes](https://borgbackup.readthedocs.io/en/latest/usage/init.html#encryption-modes)
-for the menu of available encryption modes.
-
-Also, optionally check out the [Borg Quick
-Start](https://borgbackup.readthedocs.org/en/latest/quickstart.html) for more
-background about repository initialization.
-
-Note that borgmatic skips repository initialization if the repository already
-exists. This supports use cases like ensuring a repository exists prior to
-performing a backup.
-
-If the repository is on a remote host, make sure that your local user has
-key-based SSH access to the desired user account on the remote host.
-
-
-### Backups
-
-You can run borgmatic and start a backup simply by invoking it without
-arguments:
-
-```bash
-borgmatic
-```
-
-This will also prune any old backups as per the configured retention policy,
-and check backups for consistency problems due to things like file damage.
-
-If you'd like to see the available command-line arguments, view the help:
-
-```bash
-borgmatic --help
-```
-
-Note that borgmatic prunes archives *before* creating an archive, so as to
-free up space for archiving. This means that when a borgmatic run finishes,
-there may still be prune-able archives. Not to worry, as they will get cleaned
-up at the start of the next run.
-
-
-### Verbosity
-
-By default, the backup will proceed silently except in the case of errors. But
-if you'd like to to get additional information about the progress of the
-backup as it proceeds, use the verbosity option:
-
-```bash
-borgmatic --verbosity 1
-```
-
-Or, for even more progress spew:
-
-```bash
-borgmatic --verbosity 2
-```
-
-### À la carte
-
-If you want to run borgmatic with only pruning, creating, or checking enabled,
-the following optional flags are available:
-
-```bash
-borgmatic --prune
-borgmatic --create
-borgmatic --check
-```
-
-You can run with only one of these flags provided, or you can mix and match
-any number of them. This supports use cases like running consistency checks
-from a different cron job with a different frequency, or running pruning with
-a different verbosity level.
-
-Additionally, borgmatic provides convenient flags for Borg's
-[list](https://borgbackup.readthedocs.io/en/stable/usage/list.html) and
-[info](https://borgbackup.readthedocs.io/en/stable/usage/info.html)
-functionality:
-
-
-```bash
-borgmatic --list
-borgmatic --info
-```
-
-You can include an optional `--json` flag with `--create`, `--list`, or
-`--info` to get the output formatted as JSON.
-
-
-## Autopilot
-
-If you want to run borgmatic automatically, say once a day, the you can
-configure a job runner to invoke it periodically.
-
-### cron
-
-If you're using cron, download the [sample cron
-file](https://projects.torsion.org/witten/borgmatic/src/master/sample/cron/borgmatic).
-Then, from the directory where you downloaded it:
-
-```bash
-sudo mv borgmatic /etc/cron.d/borgmatic
-sudo chmod +x /etc/cron.d/borgmatic
-```
-
-You can modify the cron file if you'd like to run borgmatic more or less frequently.
-
-### systemd
-
-If you're using systemd instead of cron to run jobs, download the [sample
-systemd service
-file](https://projects.torsion.org/witten/borgmatic/src/master/sample/systemd/borgmatic.service)
-and the [sample systemd timer
-file](https://projects.torsion.org/witten/borgmatic/src/master/sample/systemd/borgmatic.timer).
-Then, from the directory where you downloaded them:
-
-```bash
-sudo mv borgmatic.service borgmatic.timer /etc/systemd/system/
-sudo systemctl enable borgmatic.timer
-sudo systemctl start borgmatic.timer
-```
-
-Feel free to modify the timer file based on how frequently you'd like
-borgmatic to run.
-
-
-## Advanced configuration
-
-### Multiple configuration files
-
-A more advanced usage is to create multiple separate configuration files and
-place each one in an /etc/borgmatic.d directory. For instance:
-
-```bash
-sudo mkdir /etc/borgmatic.d
-sudo generate-borgmatic-config --destination /etc/borgmatic.d/app1.yaml
-sudo generate-borgmatic-config --destination /etc/borgmatic.d/app2.yaml
-```
-
-With this approach, you can have entirely different backup policies for
-different applications on your system. For instance, you may want one backup
-configuration for your database data directory, and a different configuration
-for your user home directories.
-
-When you set up multiple configuration files like this, borgmatic will run
-each one in turn from a single borgmatic invocation. This includes, by
-default, the traditional /etc/borgmatic/config.yaml as well.
-
-And if you need even more customizability, you can specify alternate
-configuration paths on the command-line with borgmatic's `--config` option.
-See `borgmatic --help` for more information.
-
-
-### Hooks
-
-If you find yourself performing prepraration tasks before your backup runs, or
-cleanup work afterwards, borgmatic hooks may be of interest. They're simply
-shell commands that borgmatic executes for you at various points, and they're
-configured in the `hooks` section of your configuration file.
-
-For instance, you can specify `before_backup` hooks to dump a database to file
-before backing it up, and specify `after_backup` hooks to delete the temporary
-file afterwards.
-
-borgmatic hooks run once per configuration file. `before_backup` hooks run
-prior to backups of all repositories. `after_backup` hooks run afterwards, but
-not if an error occurs in a previous hook or in the backups themselves. And
-borgmatic runs `on_error` hooks if an error occurs.
-
-An important security note about hooks: borgmatic executes all hook commands
-with the user permissions of borgmatic itself. So to prevent potential shell
-injection or privilege escalation, do not forget to set secure permissions
-(`chmod 0700`) on borgmatic configuration files and scripts invoked by hooks.
-
-See the sample generated configuration file mentioned above for specifics
-about hook configuration syntax.
-
-
-## Upgrading
-
-In general, all you should need to do to upgrade borgmatic is run the
-following:
-
-```bash
-sudo pip3 install --upgrade borgmatic
-```
-
-However, see below about special cases.
-
-
-### Upgrading from borgmatic 1.0.x
-
-borgmatic changed its configuration file format in version 1.1.0 from
-INI-style to YAML. This better supports validation, and has a more natural way
-to express lists of values. To upgrade your existing configuration, first
-upgrade to the new version of borgmatic.
-
-As of version 1.1.0, borgmatic no longer supports Python 2. If you were
-already running borgmatic with Python 3, then you can simply upgrade borgmatic
-in-place:
-
-```bash
-sudo pip3 install --upgrade borgmatic
-```
-
-But if you were running borgmatic with Python 2, uninstall and reinstall instead:
-
-```bash
-sudo pip uninstall borgmatic
-sudo pip3 install borgmatic
-```
-
-The pip binary names for different versions of Python can differ, so the above
-commands may need some tweaking to work on your machine.
-
-
-Once borgmatic is upgraded, run:
-
-```bash
-sudo upgrade-borgmatic-config
-```
-
-That will generate a new YAML configuration file at /etc/borgmatic/config.yaml
-(by default) using the values from both your existing configuration and
-excludes files. The new version of borgmatic will consume the YAML
-configuration file instead of the old one.
-
-
-### Upgrading from atticmatic
-
-You can ignore this section if you're not an atticmatic user (the former name
-of borgmatic).
-
-borgmatic only supports Borg now and no longer supports Attic. So if you're
-an Attic user, consider switching to Borg. See the [Borg upgrade
-command](https://borgbackup.readthedocs.io/en/stable/usage.html#borg-upgrade)
-for more information. Then, follow the instructions above about setting up
-your borgmatic configuration files.
-
-If you were already using Borg with atticmatic, then you can easily upgrade
-from atticmatic to borgmatic. Simply run the following commands:
-
-```bash
-sudo pip3 uninstall atticmatic
-sudo pip3 install borgmatic
-```
-
-That's it! borgmatic will continue using your /etc/borgmatic configuration
-files.
+<ul>
+ <li class="referral"><a href="https://www.borgbase.com/?utm_source=borgmatic">BorgBase</a>: Borg hosting service with support for monitoring, 2FA, and append-only repos</li>
+ <li class="referral"><a href="https://hetzner.cloud/?ref=v9dOJ98Ic9I8">Hetzner</a>: A "storage box" that includes support for Borg</li>
+</ul>
 
+Additionally, rsync.net has a compatible storage offering, but does not fund
+borgmatic development or hosting.
 
 ## Support and contributing
 
 ### Issues
 
-You've got issues? Or an idea for a feature enhancement? We've got an [issue
-tracker](https://projects.torsion.org/witten/borgmatic/issues). In order to
-create a new issue or comment on an issue, you'll need to [login
-first](https://projects.torsion.org/user/login). Note that you can login with
-an existing GitHub account if you prefer.
+Are you experiencing an issue with borgmatic? Or do you have an idea for a
+feature enhancement? Head on over to our [issue
+tracker](https://projects.torsion.org/borgmatic-collective/borgmatic/issues).
+In order to create a new issue or add a comment, you'll need to
+[register](https://projects.torsion.org/user/sign_up?invite_code=borgmatic)
+first. If you prefer to use an existing GitHub account, you can skip account
+creation and [login directly](https://projects.torsion.org/user/login).
 
-Other questions or comments? Contact <mailto:witten@torsion.org>.
+Also see the [security
+policy](https://torsion.org/borgmatic/docs/security-policy/) for any security
+issues.
+
+
+### Social
+
+Follow [borgmatic on Mastodon](https://fosstodon.org/@borgmatic).
+
+
+### Chat
+
+To chat with borgmatic developers or users, check out the `#borgmatic`
+IRC channel on Libera Chat, either via <a
+href="https://web.libera.chat/#borgmatic">web chat</a> or a native <a
+href="ircs://irc.libera.chat:6697">IRC client</a>. If you don't get a response
+right away, please hang around a while—or file a ticket instead.
+
+
+### Other
+
+Other questions or comments? Contact
+[witten@torsion.org](mailto:witten@torsion.org).
 
 
 ### Contributing
 
+borgmatic [source code is
+available](https://projects.torsion.org/borgmatic-collective/borgmatic) and is also mirrored
+on [GitHub](https://github.com/borgmatic-collective/borgmatic) for convenience.
+
+borgmatic is licensed under the GNU General Public License version 3 or any
+later version.
+
 If you'd like to contribute to borgmatic development, please feel free to
-submit a [Pull Request](https://projects.torsion.org/witten/borgmatic/pulls)
-or open an [issue](https://projects.torsion.org/witten/borgmatic/issues) first
-to discuss your idea. We also accept Pull Requests on GitHub, if that's more
-your thing. In general, contributions are very welcome. We don't bite! 
+submit a [Pull
+Request](https://projects.torsion.org/borgmatic-collective/borgmatic/pulls) or
+open an
+[issue](https://projects.torsion.org/borgmatic-collective/borgmatic/issues) to
+discuss your idea. Note that you'll need to
+[register](https://projects.torsion.org/user/sign_up?invite_code=borgmatic)
+first. We also accept Pull Requests on GitHub, if that's more your thing. In
+general, contributions are very welcome. We don't bite!
 
-
-### Code style
-
-Start with [PEP 8](https://www.python.org/dev/peps/pep-0008/). But then, apply
-the following deviations from it:
-
- * For strings, prefer single quotes over double quotes.
- * Limit all lines to a maximum of 100 characters.
- * Use trailing commas within multiline values or argument lists.
- * For multiline constructs, put opening and closing delimeters on lines
-   separate from their contents.
- * Within multiline constructs, use standard four-space indentation. Don't align
-   indentation with an opening delimeter.
-
-borgmatic code uses the [Black](https://black.readthedocs.io/en/stable/) code
-formatter and [Flake8](http://flake8.pycqa.org/en/latest/) code checker, so
-certain code style requirements will be enforced when running automated tests.
-See the Black and Flake8 documentation for more information.
-
-
-### Development
-
-To get set up to hack on borgmatic, first clone master via HTTPS or SSH:
-
-```bash
-git clone https://projects.torsion.org/witten/borgmatic.git
-```
-
-Or:
-
-```bash
-git clone ssh://git@projects.torsion.org:3022/witten/borgmatic.git
-```
-
-Then, install borgmatic
-"[editable](https://pip.pypa.io/en/stable/reference/pip_install/#editable-installs)"
-so that you can easily run borgmatic commands while you're hacking on them to
-make sure your changes work.
-
-```bash
-cd borgmatic/
-pip3 install --editable --user .
-```
-
-Note that this will typically install the borgmatic commands into
-`~/.local/bin`, which may or may not be on your PATH. There are other ways to
-install borgmatic editable as well, for instance into the system Python
-install (so without `--user`, as root), or even into a
-[virtualenv](https://virtualenv.pypa.io/en/stable/). How or where you install
-borgmatic is up to you, but generally an editable install makes development
-and testing easier.
-
-
-### Running tests
-
-Assuming you've cloned the borgmatic source code as described above, and
-you're in the `borgmatic/` working copy, install tox, which is used for
-setting up testing environments:
-
-```bash
-sudo pip3 install tox
-```
-
-Finally, to actually run tests, run:
-
-```bash
-cd borgmatic
-tox
-```
-
-If when running tests, you get an error from the
-[Black](https://black.readthedocs.io/en/stable/) code formatter about files
-that would be reformatted, you can ask Black to format them for you via the
-following:
-
-```bash
-tox -e black
-```
-
-### End-to-end tests
-
-borgmatic additionally includes some end-to-end tests that integration test
-with Borg for a few representative scenarios. These tests don't run by default
-because they're relatively slow and depend on Borg. If you would like to run
-them:
-
-```bash
-tox -e end-to-end
-```
-
-## Troubleshooting
-
-### Broken pipe with remote repository
-
-When running borgmatic on a large remote repository, you may receive errors
-like the following, particularly while "borg check" is validating backups for
-consistency:
-
-```text
-    Write failed: Broken pipe
-    borg: Error: Connection closed by remote host
-```
-
-This error can be caused by an ssh timeout, which you can rectify by adding
-the following to the `~/.ssh/config` file on the client:
-
-```text
-    Host *
-        ServerAliveInterval 120
-```
-
-This should make the client keep the connection alive while validating
-backups.
-
-
-### libyaml compilation errors
-
-borgmatic depends on a Python YAML library (ruamel.yaml) that will optionally
-use a C YAML library (libyaml) if present. But if it's not installed, then
-when installing or upgrading borgmatic, you may see errors about compiling the
-YAML library. If so, not to worry. borgmatic should install and function
-correctly even without the C YAML library. And borgmatic won't be any faster
-with the C library present, so you don't need to go out of your way to install
-it.
+Also, please check out the [borgmatic development
+how-to](https://torsion.org/borgmatic/docs/how-to/develop-on-borgmatic/) for
+info on cloning source code, running tests, etc.

SECURITY.md

@@ -0,0 +1,18 @@
+---
+title: Security policy
+permalink: security-policy/index.html
+---
+
+## Supported versions
+
+While we want to hear about security vulnerabilities in all versions of
+borgmatic, security fixes are only made to the most recently released version.
+It's not practical for our small volunteer effort to maintain multiple release
+branches and put out separate security patches for each.
+
+## Reporting a vulnerability
+
+If you find a security vulnerability, please [file a
+ticket](https://torsion.org/borgmatic/#issues) or [send email
+directly](mailto:witten@torsion.org) as appropriate. You should expect to hear
+back within a few days at most and generally sooner.

borgmatic/actions/arguments.py

@@ -0,0 +1,9 @@
+import argparse
+
+
+def update_arguments(arguments, **updates):
+    '''
+    Given an argparse.Namespace instance of command-line arguments and one or more keyword argument
+    updates to perform, return a copy of the arguments with those updates applied.
+    '''
+    return argparse.Namespace(**dict(vars(arguments), **updates))

borgmatic/actions/borg.py

@@ -0,0 +1,45 @@
+import logging
+
+import borgmatic.borg.borg
+import borgmatic.borg.rlist
+import borgmatic.config.validate
+
+logger = logging.getLogger(__name__)
+
+
+def run_borg(
+    repository,
+    config,
+    local_borg_version,
+    borg_arguments,
+    global_arguments,
+    local_path,
+    remote_path,
+):
+    '''
+    Run the "borg" action for the given repository.
+    '''
+    if borg_arguments.repository is None or borgmatic.config.validate.repositories_match(
+        repository, borg_arguments.repository
+    ):
+        logger.info(
+            f'{repository.get("label", repository["path"])}: Running arbitrary Borg command'
+        )
+        archive_name = borgmatic.borg.rlist.resolve_archive_name(
+            repository['path'],
+            borg_arguments.archive,
+            config,
+            local_borg_version,
+            global_arguments,
+            local_path,
+            remote_path,
+        )
+        borgmatic.borg.borg.run_arbitrary_borg(
+            repository['path'],
+            config,
+            local_borg_version,
+            options=borg_arguments.options,
+            archive=archive_name,
+            local_path=local_path,
+            remote_path=remote_path,
+        )

borgmatic/actions/break_lock.py

@@ -0,0 +1,34 @@
+import logging
+
+import borgmatic.borg.break_lock
+import borgmatic.config.validate
+
+logger = logging.getLogger(__name__)
+
+
+def run_break_lock(
+    repository,
+    config,
+    local_borg_version,
+    break_lock_arguments,
+    global_arguments,
+    local_path,
+    remote_path,
+):
+    '''
+    Run the "break-lock" action for the given repository.
+    '''
+    if break_lock_arguments.repository is None or borgmatic.config.validate.repositories_match(
+        repository, break_lock_arguments.repository
+    ):
+        logger.info(
+            f'{repository.get("label", repository["path"])}: Breaking repository and cache locks'
+        )
+        borgmatic.borg.break_lock.break_lock(
+            repository['path'],
+            config,
+            local_borg_version,
+            global_arguments,
+            local_path=local_path,
+            remote_path=remote_path,
+        )

borgmatic/actions/check.py

@@ -0,0 +1,672 @@
+import datetime
+import hashlib
+import itertools
+import logging
+import os
+import pathlib
+import random
+
+import borgmatic.borg.check
+import borgmatic.borg.create
+import borgmatic.borg.environment
+import borgmatic.borg.extract
+import borgmatic.borg.list
+import borgmatic.borg.rlist
+import borgmatic.borg.state
+import borgmatic.config.validate
+import borgmatic.execute
+import borgmatic.hooks.command
+
+DEFAULT_CHECKS = (
+    {'name': 'repository', 'frequency': '1 month'},
+    {'name': 'archives', 'frequency': '1 month'},
+)
+
+
+logger = logging.getLogger(__name__)
+
+
+def parse_checks(config, only_checks=None):
+    '''
+    Given a configuration dict with a "checks" sequence of dicts and an optional list of override
+    checks, return a tuple of named checks to run.
+
+    For example, given a config of:
+
+        {'checks': ({'name': 'repository'}, {'name': 'archives'})}
+
+    This will be returned as:
+
+        ('repository', 'archives')
+
+    If no "checks" option is present in the config, return the DEFAULT_CHECKS. If a checks value
+    has a name of "disabled", return an empty tuple, meaning that no checks should be run.
+    '''
+    checks = only_checks or tuple(
+        check_config['name'] for check_config in (config.get('checks', None) or DEFAULT_CHECKS)
+    )
+    checks = tuple(check.lower() for check in checks)
+
+    if 'disabled' in checks:
+        logger.warning(
+            'The "disabled" value for the "checks" option is deprecated and will be removed from a future release; use "skip_actions" instead'
+        )
+        if len(checks) > 1:
+            logger.warning(
+                'Multiple checks are configured, but one of them is "disabled"; not running any checks'
+            )
+        return ()
+
+    return checks
+
+
+def parse_frequency(frequency):
+    '''
+    Given a frequency string with a number and a unit of time, return a corresponding
+    datetime.timedelta instance or None if the frequency is None or "always".
+
+    For instance, given "3 weeks", return datetime.timedelta(weeks=3)
+
+    Raise ValueError if the given frequency cannot be parsed.
+    '''
+    if not frequency:
+        return None
+
+    frequency = frequency.strip().lower()
+
+    if frequency == 'always':
+        return None
+
+    try:
+        number, time_unit = frequency.split(' ')
+        number = int(number)
+    except ValueError:
+        raise ValueError(f"Could not parse consistency check frequency '{frequency}'")
+
+    if not time_unit.endswith('s'):
+        time_unit += 's'
+
+    if time_unit == 'months':
+        number *= 30
+        time_unit = 'days'
+    elif time_unit == 'years':
+        number *= 365
+        time_unit = 'days'
+
+    try:
+        return datetime.timedelta(**{time_unit: number})
+    except TypeError:
+        raise ValueError(f"Could not parse consistency check frequency '{frequency}'")
+
+
+def filter_checks_on_frequency(
+    config,
+    borg_repository_id,
+    checks,
+    force,
+    archives_check_id=None,
+):
+    '''
+    Given a configuration dict with a "checks" sequence of dicts, a Borg repository ID, a sequence
+    of checks, whether to force checks to run, and an ID for the archives check potentially being
+    run (if any), filter down those checks based on the configured "frequency" for each check as
+    compared to its check time file.
+
+    In other words, a check whose check time file's timestamp is too new (based on the configured
+    frequency) will get cut from the returned sequence of checks. Example:
+
+    config = {
+        'checks': [
+            {
+                'name': 'archives',
+                'frequency': '2 weeks',
+            },
+        ]
+    }
+
+    When this function is called with that config and "archives" in checks, "archives" will get
+    filtered out of the returned result if its check time file is newer than 2 weeks old, indicating
+    that it's not yet time to run that check again.
+
+    Raise ValueError if a frequency cannot be parsed.
+    '''
+    if not checks:
+        return checks
+
+    filtered_checks = list(checks)
+
+    if force:
+        return tuple(filtered_checks)
+
+    for check_config in config.get('checks', DEFAULT_CHECKS):
+        check = check_config['name']
+        if checks and check not in checks:
+            continue
+
+        frequency_delta = parse_frequency(check_config.get('frequency'))
+        if not frequency_delta:
+            continue
+
+        check_time = probe_for_check_time(config, borg_repository_id, check, archives_check_id)
+        if not check_time:
+            continue
+
+        # If we've not yet reached the time when the frequency dictates we're ready for another
+        # check, skip this check.
+        if datetime.datetime.now() < check_time + frequency_delta:
+            remaining = check_time + frequency_delta - datetime.datetime.now()
+            logger.info(
+                f'Skipping {check} check due to configured frequency; {remaining} until next check (use --force to check anyway)'
+            )
+            filtered_checks.remove(check)
+
+    return tuple(filtered_checks)
+
+
+def make_archives_check_id(archive_filter_flags):
+    '''
+    Given a sequence of flags to filter archives, return a unique hash corresponding to those
+    particular flags. If there are no flags, return None.
+    '''
+    if not archive_filter_flags:
+        return None
+
+    return hashlib.sha256(' '.join(archive_filter_flags).encode()).hexdigest()
+
+
+def make_check_time_path(config, borg_repository_id, check_type, archives_check_id=None):
+    '''
+    Given a configuration dict, a Borg repository ID, the name of a check type ("repository",
+    "archives", etc.), and a unique hash of the archives filter flags, return a path for recording
+    that check's time (the time of that check last occurring).
+    '''
+    borgmatic_source_directory = os.path.expanduser(
+        config.get(
+            'borgmatic_source_directory', borgmatic.borg.state.DEFAULT_BORGMATIC_SOURCE_DIRECTORY
+        )
+    )
+
+    if check_type in ('archives', 'data'):
+        return os.path.join(
+            borgmatic_source_directory,
+            'checks',
+            borg_repository_id,
+            check_type,
+            archives_check_id if archives_check_id else 'all',
+        )
+
+    return os.path.join(
+        borgmatic_source_directory,
+        'checks',
+        borg_repository_id,
+        check_type,
+    )
+
+
+def write_check_time(path):  # pragma: no cover
+    '''
+    Record a check time of now as the modification time of the given path.
+    '''
+    logger.debug(f'Writing check time at {path}')
+
+    os.makedirs(os.path.dirname(path), mode=0o700, exist_ok=True)
+    pathlib.Path(path, mode=0o600).touch()
+
+
+def read_check_time(path):
+    '''
+    Return the check time based on the modification time of the given path. Return None if the path
+    doesn't exist.
+    '''
+    logger.debug(f'Reading check time from {path}')
+
+    try:
+        return datetime.datetime.fromtimestamp(os.stat(path).st_mtime)
+    except FileNotFoundError:
+        return None
+
+
+def probe_for_check_time(config, borg_repository_id, check, archives_check_id):
+    '''
+    Given a configuration dict, a Borg repository ID, the name of a check type ("repository",
+    "archives", etc.), and a unique hash of the archives filter flags, return a the corresponding
+    check time or None if such a check time does not exist.
+
+    When the check type is "archives" or "data", this function probes two different paths to find
+    the check time, e.g.:
+
+      ~/.borgmatic/checks/1234567890/archives/9876543210
+      ~/.borgmatic/checks/1234567890/archives/all
+
+    ... and returns the maximum modification time of the files found (if any). The first path
+    represents a more specific archives check time (a check on a subset of archives), and the second
+    is a fallback to the last "all" archives check.
+
+    For other check types, this function reads from a single check time path, e.g.:
+
+      ~/.borgmatic/checks/1234567890/repository
+    '''
+    check_times = (
+        read_check_time(group[0])
+        for group in itertools.groupby(
+            (
+                make_check_time_path(config, borg_repository_id, check, archives_check_id),
+                make_check_time_path(config, borg_repository_id, check),
+            )
+        )
+    )
+
+    try:
+        return max(check_time for check_time in check_times if check_time)
+    except ValueError:
+        return None
+
+
+def upgrade_check_times(config, borg_repository_id):
+    '''
+    Given a configuration dict and a Borg repository ID, upgrade any corresponding check times on
+    disk from old-style paths to new-style paths.
+
+    Currently, the only upgrade performed is renaming an archive or data check path that looks like:
+
+      ~/.borgmatic/checks/1234567890/archives
+
+    to:
+
+      ~/.borgmatic/checks/1234567890/archives/all
+    '''
+    for check_type in ('archives', 'data'):
+        new_path = make_check_time_path(config, borg_repository_id, check_type, 'all')
+        old_path = os.path.dirname(new_path)
+        temporary_path = f'{old_path}.temp'
+
+        if not os.path.isfile(old_path) and not os.path.isfile(temporary_path):
+            continue
+
+        logger.debug(f'Upgrading archives check time from {old_path} to {new_path}')
+
+        try:
+            os.rename(old_path, temporary_path)
+        except FileNotFoundError:
+            pass
+
+        os.mkdir(old_path)
+        os.rename(temporary_path, new_path)
+
+
+def collect_spot_check_source_paths(
+    repository, config, local_borg_version, global_arguments, local_path, remote_path
+):
+    '''
+    Given a repository configuration dict, a configuration dict, the local Borg version, global
+    arguments as an argparse.Namespace instance, the local Borg path, and the remote Borg path,
+    collect the source paths that Borg would use in an actual create (but only include files and
+    symlinks).
+    '''
+    stream_processes = any(
+        borgmatic.hooks.dispatch.call_hooks(
+            'use_streaming',
+            config,
+            repository['path'],
+            borgmatic.hooks.dump.DATA_SOURCE_HOOK_NAMES,
+        ).values()
+    )
+
+    (create_flags, create_positional_arguments, pattern_file, exclude_file) = (
+        borgmatic.borg.create.make_base_create_command(
+            dry_run=True,
+            repository_path=repository['path'],
+            config=config,
+            config_paths=(),
+            local_borg_version=local_borg_version,
+            global_arguments=global_arguments,
+            borgmatic_source_directories=(),
+            local_path=local_path,
+            remote_path=remote_path,
+            list_files=True,
+            stream_processes=stream_processes,
+        )
+    )
+    borg_environment = borgmatic.borg.environment.make_environment(config)
+
+    try:
+        working_directory = os.path.expanduser(config.get('working_directory'))
+    except TypeError:
+        working_directory = None
+
+    paths_output = borgmatic.execute.execute_command_and_capture_output(
+        create_flags + create_positional_arguments,
+        capture_stderr=True,
+        working_directory=working_directory,
+        extra_environment=borg_environment,
+        borg_local_path=local_path,
+        borg_exit_codes=config.get('borg_exit_codes'),
+    )
+
+    paths = tuple(
+        path_line.split(' ', 1)[1]
+        for path_line in paths_output.split('\n')
+        if path_line and path_line.startswith('- ') or path_line.startswith('+ ')
+    )
+
+    return tuple(path for path in paths if os.path.isfile(path) or os.path.islink(path))
+
+
+BORG_DIRECTORY_FILE_TYPE = 'd'
+
+
+def collect_spot_check_archive_paths(
+    repository, archive, config, local_borg_version, global_arguments, local_path, remote_path
+):
+    '''
+    Given a repository configuration dict, the name of the latest archive, a configuration dict, the
+    local Borg version, global arguments as an argparse.Namespace instance, the local Borg path, and
+    the remote Borg path, collect the paths from the given archive (but only include files and
+    symlinks).
+    '''
+    borgmatic_source_directory = os.path.expanduser(
+        config.get(
+            'borgmatic_source_directory', borgmatic.borg.state.DEFAULT_BORGMATIC_SOURCE_DIRECTORY
+        )
+    )
+
+    return tuple(
+        path
+        for line in borgmatic.borg.list.capture_archive_listing(
+            repository['path'],
+            archive,
+            config,
+            local_borg_version,
+            global_arguments,
+            path_format='{type} /{path}{NL}',  # noqa: FS003
+            local_path=local_path,
+            remote_path=remote_path,
+        )
+        for (file_type, path) in (line.split(' ', 1),)
+        if file_type != BORG_DIRECTORY_FILE_TYPE
+        if pathlib.Path(borgmatic_source_directory) not in pathlib.Path(path).parents
+    )
+
+
+def compare_spot_check_hashes(
+    repository,
+    archive,
+    config,
+    local_borg_version,
+    global_arguments,
+    local_path,
+    remote_path,
+    log_label,
+    source_paths,
+):
+    '''
+    Given a repository configuration dict, the name of the latest archive, a configuration dict, the
+    local Borg version, global arguments as an argparse.Namespace instance, the local Borg path, the
+    remote Borg path, a log label, and spot check source paths, compare the hashes for a sampling of
+    the source paths with hashes from corresponding paths in the given archive. Return a sequence of
+    the paths that fail that hash comparison.
+    '''
+    # Based on the configured sample percentage, come up with a list of random sample files from the
+    # source directories.
+    spot_check_config = next(check for check in config['checks'] if check['name'] == 'spot')
+    sample_count = max(
+        int(len(source_paths) * (min(spot_check_config['data_sample_percentage'], 100) / 100)), 1
+    )
+    source_sample_paths = tuple(random.sample(source_paths, sample_count))
+    existing_source_sample_paths = {
+        source_path for source_path in source_sample_paths if os.path.exists(source_path)
+    }
+    logger.debug(
+        f'{log_label}: Sampling {sample_count} source paths (~{spot_check_config["data_sample_percentage"]}%) for spot check'
+    )
+
+    # Hash each file in the sample paths (if it exists).
+    hash_output = borgmatic.execute.execute_command_and_capture_output(
+        (spot_check_config.get('xxh64sum_command', 'xxh64sum'),)
+        + tuple(path for path in source_sample_paths if path in existing_source_sample_paths)
+    )
+
+    source_hashes = dict(
+        (reversed(line.split('  ', 1)) for line in hash_output.splitlines()),
+        **{path: '' for path in source_sample_paths if path not in existing_source_sample_paths},
+    )
+
+    archive_hashes = dict(
+        reversed(line.split(' ', 1))
+        for line in borgmatic.borg.list.capture_archive_listing(
+            repository['path'],
+            archive,
+            config,
+            local_borg_version,
+            global_arguments,
+            list_paths=source_sample_paths,
+            path_format='{xxh64} /{path}{NL}',  # noqa: FS003
+            local_path=local_path,
+            remote_path=remote_path,
+        )
+        if line
+    )
+
+    # Compare the source hashes with the archive hashes to see how many match.
+    failing_paths = []
+
+    for path, source_hash in source_hashes.items():
+        archive_hash = archive_hashes.get(path)
+
+        if archive_hash is not None and archive_hash == source_hash:
+            continue
+
+        failing_paths.append(path)
+
+    return tuple(failing_paths)
+
+
+def spot_check(
+    repository,
+    config,
+    local_borg_version,
+    global_arguments,
+    local_path,
+    remote_path,
+):
+    '''
+    Given a repository dict, a loaded configuration dict, the local Borg version, global arguments
+    as an argparse.Namespace instance, the local Borg path, and the remote Borg path, perform a spot
+    check for the latest archive in the given repository.
+
+    A spot check compares file counts and also the hashes for a random sampling of source files on
+    disk to those stored in the latest archive. If any differences are beyond configured tolerances,
+    then the check fails.
+    '''
+    log_label = f'{repository.get("label", repository["path"])}'
+    logger.debug(f'{log_label}: Running spot check')
+
+    try:
+        spot_check_config = next(
+            check for check in config.get('checks', ()) if check.get('name') == 'spot'
+        )
+    except StopIteration:
+        raise ValueError('Cannot run spot check because it is unconfigured')
+
+    if spot_check_config['data_tolerance_percentage'] > spot_check_config['data_sample_percentage']:
+        raise ValueError(
+            'The data_tolerance_percentage must be less than or equal to the data_sample_percentage'
+        )
+
+    source_paths = collect_spot_check_source_paths(
+        repository,
+        config,
+        local_borg_version,
+        global_arguments,
+        local_path,
+        remote_path,
+    )
+    logger.debug(f'{log_label}: {len(source_paths)} total source paths for spot check')
+
+    archive = borgmatic.borg.rlist.resolve_archive_name(
+        repository['path'],
+        'latest',
+        config,
+        local_borg_version,
+        global_arguments,
+        local_path,
+        remote_path,
+    )
+    logger.debug(f'{log_label}: Using archive {archive} for spot check')
+
+    archive_paths = collect_spot_check_archive_paths(
+        repository,
+        archive,
+        config,
+        local_borg_version,
+        global_arguments,
+        local_path,
+        remote_path,
+    )
+    logger.debug(f'{log_label}: {len(archive_paths)} total archive paths for spot check')
+
+    # Calculate the percentage delta between the source paths count and the archive paths count, and
+    # compare that delta to the configured count tolerance percentage.
+    count_delta_percentage = abs(len(source_paths) - len(archive_paths)) / len(source_paths) * 100
+
+    if count_delta_percentage > spot_check_config['count_tolerance_percentage']:
+        logger.debug(
+            f'{log_label}: Paths in source paths but not latest archive: {", ".join(set(source_paths) - set(archive_paths)) or "none"}'
+        )
+        logger.debug(
+            f'{log_label}: Paths in latest archive but not source paths: {", ".join(set(archive_paths) - set(source_paths)) or "none"}'
+        )
+        raise ValueError(
+            f'Spot check failed: {count_delta_percentage:.2f}% file count delta between source paths and latest archive (tolerance is {spot_check_config["count_tolerance_percentage"]}%)'
+        )
+
+    failing_paths = compare_spot_check_hashes(
+        repository,
+        archive,
+        config,
+        local_borg_version,
+        global_arguments,
+        local_path,
+        remote_path,
+        log_label,
+        source_paths,
+    )
+
+    # Error if the percentage of failing hashes exceeds the configured tolerance percentage.
+    logger.debug(f'{log_label}: {len(failing_paths)} non-matching spot check hashes')
+    data_tolerance_percentage = spot_check_config['data_tolerance_percentage']
+    failing_percentage = (len(failing_paths) / len(source_paths)) * 100
+
+    if failing_percentage > data_tolerance_percentage:
+        logger.debug(
+            f'{log_label}: Source paths with data not matching the latest archive: {", ".join(failing_paths)}'
+        )
+        raise ValueError(
+            f'Spot check failed: {failing_percentage:.2f}% of source paths with data not matching the latest archive (tolerance is {data_tolerance_percentage}%)'
+        )
+
+    logger.info(
+        f'{log_label}: Spot check passed with a {count_delta_percentage:.2f}% file count delta and a {failing_percentage:.2f}% file data delta'
+    )
+
+
+def run_check(
+    config_filename,
+    repository,
+    config,
+    hook_context,
+    local_borg_version,
+    check_arguments,
+    global_arguments,
+    local_path,
+    remote_path,
+):
+    '''
+    Run the "check" action for the given repository.
+
+    Raise ValueError if the Borg repository ID cannot be determined.
+    '''
+    if check_arguments.repository and not borgmatic.config.validate.repositories_match(
+        repository, check_arguments.repository
+    ):
+        return
+
+    borgmatic.hooks.command.execute_hook(
+        config.get('before_check'),
+        config.get('umask'),
+        config_filename,
+        'pre-check',
+        global_arguments.dry_run,
+        **hook_context,
+    )
+
+    logger.info(f'{repository.get("label", repository["path"])}: Running consistency checks')
+    repository_id = borgmatic.borg.check.get_repository_id(
+        repository['path'],
+        config,
+        local_borg_version,
+        global_arguments,
+        local_path=local_path,
+        remote_path=remote_path,
+    )
+    upgrade_check_times(config, repository_id)
+    configured_checks = parse_checks(config, check_arguments.only_checks)
+    archive_filter_flags = borgmatic.borg.check.make_archive_filter_flags(
+        local_borg_version, config, configured_checks, check_arguments
+    )
+    archives_check_id = make_archives_check_id(archive_filter_flags)
+    checks = filter_checks_on_frequency(
+        config,
+        repository_id,
+        configured_checks,
+        check_arguments.force,
+        archives_check_id,
+    )
+    borg_specific_checks = set(checks).intersection({'repository', 'archives', 'data'})
+
+    if borg_specific_checks:
+        borgmatic.borg.check.check_archives(
+            repository['path'],
+            config,
+            local_borg_version,
+            check_arguments,
+            global_arguments,
+            borg_specific_checks,
+            archive_filter_flags,
+            local_path=local_path,
+            remote_path=remote_path,
+        )
+        for check in borg_specific_checks:
+            write_check_time(make_check_time_path(config, repository_id, check, archives_check_id))
+
+    if 'extract' in checks:
+        borgmatic.borg.extract.extract_last_archive_dry_run(
+            config,
+            local_borg_version,
+            global_arguments,
+            repository['path'],
+            config.get('lock_wait'),
+            local_path,
+            remote_path,
+        )
+        write_check_time(make_check_time_path(config, repository_id, 'extract'))
+
+    if 'spot' in checks:
+        spot_check(
+            repository,
+            config,
+            local_borg_version,
+            global_arguments,
+            local_path,
+            remote_path,
+        )
+        write_check_time(make_check_time_path(config, repository_id, 'spot'))
+
+    borgmatic.hooks.command.execute_hook(
+        config.get('after_check'),
+        config.get('umask'),
+        config_filename,
+        'post-check',
+        global_arguments.dry_run,
+        **hook_context,
+    )

borgmatic/actions/compact.py

@@ -0,0 +1,66 @@
+import logging
+
+import borgmatic.borg.compact
+import borgmatic.borg.feature
+import borgmatic.config.validate
+import borgmatic.hooks.command
+
+logger = logging.getLogger(__name__)
+
+
+def run_compact(
+    config_filename,
+    repository,
+    config,
+    hook_context,
+    local_borg_version,
+    compact_arguments,
+    global_arguments,
+    dry_run_label,
+    local_path,
+    remote_path,
+):
+    '''
+    Run the "compact" action for the given repository.
+    '''
+    if compact_arguments.repository and not borgmatic.config.validate.repositories_match(
+        repository, compact_arguments.repository
+    ):
+        return
+
+    borgmatic.hooks.command.execute_hook(
+        config.get('before_compact'),
+        config.get('umask'),
+        config_filename,
+        'pre-compact',
+        global_arguments.dry_run,
+        **hook_context,
+    )
+    if borgmatic.borg.feature.available(borgmatic.borg.feature.Feature.COMPACT, local_borg_version):
+        logger.info(
+            f'{repository.get("label", repository["path"])}: Compacting segments{dry_run_label}'
+        )
+        borgmatic.borg.compact.compact_segments(
+            global_arguments.dry_run,
+            repository['path'],
+            config,
+            local_borg_version,
+            global_arguments,
+            local_path=local_path,
+            remote_path=remote_path,
+            progress=compact_arguments.progress,
+            cleanup_commits=compact_arguments.cleanup_commits,
+            threshold=compact_arguments.threshold,
+        )
+    else:  # pragma: nocover
+        logger.info(
+            f'{repository.get("label", repository["path"])}: Skipping compact (only available/needed in Borg 1.2+)'
+        )
+    borgmatic.hooks.command.execute_hook(
+        config.get('after_compact'),
+        config.get('umask'),
+        config_filename,
+        'post-compact',
+        global_arguments.dry_run,
+        **hook_context,
+    )

borgmatic/actions/config/bootstrap.py

@@ -0,0 +1,103 @@
+import json
+import logging
+import os
+
+import borgmatic.borg.extract
+import borgmatic.borg.rlist
+import borgmatic.config.validate
+import borgmatic.hooks.command
+from borgmatic.borg.state import DEFAULT_BORGMATIC_SOURCE_DIRECTORY
+
+logger = logging.getLogger(__name__)
+
+
+def get_config_paths(bootstrap_arguments, global_arguments, local_borg_version):
+    '''
+    Given the bootstrap arguments as an argparse.Namespace (containing the repository and archive
+    name, borgmatic source directory, destination directory, and whether to strip components), the
+    global arguments as an argparse.Namespace (containing the dry run flag and the local borg
+    version), return the config paths from the manifest.json file in the borgmatic source directory
+    after extracting it from the repository.
+
+    Raise ValueError if the manifest JSON is missing, can't be decoded, or doesn't contain the
+    expected configuration path data.
+    '''
+    borgmatic_source_directory = (
+        bootstrap_arguments.borgmatic_source_directory or DEFAULT_BORGMATIC_SOURCE_DIRECTORY
+    )
+    borgmatic_manifest_path = os.path.expanduser(
+        os.path.join(borgmatic_source_directory, 'bootstrap', 'manifest.json')
+    )
+    config = {'ssh_command': bootstrap_arguments.ssh_command}
+
+    extract_process = borgmatic.borg.extract.extract_archive(
+        global_arguments.dry_run,
+        bootstrap_arguments.repository,
+        borgmatic.borg.rlist.resolve_archive_name(
+            bootstrap_arguments.repository,
+            bootstrap_arguments.archive,
+            config,
+            local_borg_version,
+            global_arguments,
+        ),
+        [borgmatic_manifest_path],
+        config,
+        local_borg_version,
+        global_arguments,
+        extract_to_stdout=True,
+    )
+    manifest_json = extract_process.stdout.read()
+
+    if not manifest_json:
+        raise ValueError(
+            'Cannot read configuration paths from archive due to missing bootstrap manifest'
+        )
+
+    try:
+        manifest_data = json.loads(manifest_json)
+    except json.JSONDecodeError as error:
+        raise ValueError(
+            f'Cannot read configuration paths from archive due to invalid bootstrap manifest JSON: {error}'
+        )
+
+    try:
+        return manifest_data['config_paths']
+    except KeyError:
+        raise ValueError(
+            'Cannot read configuration paths from archive due to invalid bootstrap manifest'
+        )
+
+
+def run_bootstrap(bootstrap_arguments, global_arguments, local_borg_version):
+    '''
+    Run the "bootstrap" action for the given repository.
+
+    Raise ValueError if the bootstrap configuration could not be loaded.
+    Raise CalledProcessError or OSError if Borg could not be run.
+    '''
+    manifest_config_paths = get_config_paths(
+        bootstrap_arguments, global_arguments, local_borg_version
+    )
+    config = {'ssh_command': bootstrap_arguments.ssh_command}
+
+    logger.info(f"Bootstrapping config paths: {', '.join(manifest_config_paths)}")
+
+    borgmatic.borg.extract.extract_archive(
+        global_arguments.dry_run,
+        bootstrap_arguments.repository,
+        borgmatic.borg.rlist.resolve_archive_name(
+            bootstrap_arguments.repository,
+            bootstrap_arguments.archive,
+            config,
+            local_borg_version,
+            global_arguments,
+        ),
+        [config_path.lstrip(os.path.sep) for config_path in manifest_config_paths],
+        config,
+        local_borg_version,
+        global_arguments,
+        extract_to_stdout=False,
+        destination_path=bootstrap_arguments.destination,
+        strip_components=bootstrap_arguments.strip_components,
+        progress=bootstrap_arguments.progress,
+    )

borgmatic/actions/config/generate.py

@@ -0,0 +1,48 @@
+import logging
+
+import borgmatic.config.generate
+import borgmatic.config.validate
+import borgmatic.logger
+
+logger = logging.getLogger(__name__)
+
+
+def run_generate(generate_arguments, global_arguments):
+    '''
+    Given the generate arguments and the global arguments, each as an argparse.Namespace instance,
+    run the "generate" action.
+
+    Raise FileExistsError if a file already exists at the destination path and the generate
+    arguments do not have overwrite set.
+    '''
+    borgmatic.logger.add_custom_log_levels()
+    dry_run_label = ' (dry run; not actually writing anything)' if global_arguments.dry_run else ''
+
+    logger.answer(
+        f'Generating a configuration file at: {generate_arguments.destination_filename}{dry_run_label}'
+    )
+
+    borgmatic.config.generate.generate_sample_configuration(
+        global_arguments.dry_run,
+        generate_arguments.source_filename,
+        generate_arguments.destination_filename,
+        borgmatic.config.validate.schema_filename(),
+        overwrite=generate_arguments.overwrite,
+    )
+
+    if generate_arguments.source_filename:
+        logger.answer(
+            f'''
+Merged in the contents of configuration file at: {generate_arguments.source_filename}
+To review the changes made, run:
+
+    diff --unified {generate_arguments.source_filename} {generate_arguments.destination_filename}'''
+        )
+
+    logger.answer(
+        '''
+This includes all available configuration options with example values, the few
+required options as indicated. Please edit the file to suit your needs.
+
+If you ever need help: https://torsion.org/borgmatic/#issues'''
+    )

borgmatic/actions/config/validate.py

@@ -0,0 +1,25 @@
+import logging
+
+import borgmatic.config.generate
+import borgmatic.logger
+
+logger = logging.getLogger(__name__)
+
+
+def run_validate(validate_arguments, configs):
+    '''
+    Given the validate arguments as an argparse.Namespace instance and a dict of configuration
+    filename to corresponding parsed configuration, run the "validate" action.
+
+    Most of the validation is actually performed implicitly by the standard borgmatic configuration
+    loading machinery prior to here, so this function mainly exists to support additional validate
+    flags like "--show".
+    '''
+    borgmatic.logger.add_custom_log_levels()
+
+    if validate_arguments.show:
+        for config_path, config in configs.items():
+            if len(configs) > 1:
+                logger.answer('---')
+
+            logger.answer(borgmatic.config.generate.render_configuration(config))

borgmatic/actions/create.py

@@ -0,0 +1,132 @@
+import importlib.metadata
+import json
+import logging
+import os
+
+import borgmatic.actions.json
+import borgmatic.borg.create
+import borgmatic.borg.state
+import borgmatic.config.validate
+import borgmatic.hooks.command
+import borgmatic.hooks.dispatch
+import borgmatic.hooks.dump
+
+logger = logging.getLogger(__name__)
+
+
+def create_borgmatic_manifest(config, config_paths, dry_run):
+    '''
+    Create a borgmatic manifest file to store the paths to the configuration files used to create
+    the archive.
+    '''
+    if dry_run:
+        return
+
+    borgmatic_source_directory = config.get(
+        'borgmatic_source_directory', borgmatic.borg.state.DEFAULT_BORGMATIC_SOURCE_DIRECTORY
+    )
+
+    borgmatic_manifest_path = os.path.expanduser(
+        os.path.join(borgmatic_source_directory, 'bootstrap', 'manifest.json')
+    )
+
+    if not os.path.exists(borgmatic_manifest_path):
+        os.makedirs(os.path.dirname(borgmatic_manifest_path), exist_ok=True)
+
+    with open(borgmatic_manifest_path, 'w') as config_list_file:
+        json.dump(
+            {
+                'borgmatic_version': importlib.metadata.version('borgmatic'),
+                'config_paths': config_paths,
+            },
+            config_list_file,
+        )
+
+
+def run_create(
+    config_filename,
+    repository,
+    config,
+    config_paths,
+    hook_context,
+    local_borg_version,
+    create_arguments,
+    global_arguments,
+    dry_run_label,
+    local_path,
+    remote_path,
+):
+    '''
+    Run the "create" action for the given repository.
+
+    If create_arguments.json is True, yield the JSON output from creating the archive.
+    '''
+    if create_arguments.repository and not borgmatic.config.validate.repositories_match(
+        repository, create_arguments.repository
+    ):
+        return
+
+    borgmatic.hooks.command.execute_hook(
+        config.get('before_backup'),
+        config.get('umask'),
+        config_filename,
+        'pre-backup',
+        global_arguments.dry_run,
+        **hook_context,
+    )
+    logger.info(f'{repository.get("label", repository["path"])}: Creating archive{dry_run_label}')
+    borgmatic.hooks.dispatch.call_hooks_even_if_unconfigured(
+        'remove_data_source_dumps',
+        config,
+        repository['path'],
+        borgmatic.hooks.dump.DATA_SOURCE_HOOK_NAMES,
+        global_arguments.dry_run,
+    )
+    active_dumps = borgmatic.hooks.dispatch.call_hooks(
+        'dump_data_sources',
+        config,
+        repository['path'],
+        borgmatic.hooks.dump.DATA_SOURCE_HOOK_NAMES,
+        global_arguments.dry_run,
+    )
+    if config.get('store_config_files', True):
+        create_borgmatic_manifest(
+            config,
+            config_paths,
+            global_arguments.dry_run,
+        )
+    stream_processes = [process for processes in active_dumps.values() for process in processes]
+
+    json_output = borgmatic.borg.create.create_archive(
+        global_arguments.dry_run,
+        repository['path'],
+        config,
+        config_paths,
+        local_borg_version,
+        global_arguments,
+        local_path=local_path,
+        remote_path=remote_path,
+        progress=create_arguments.progress,
+        stats=create_arguments.stats,
+        json=create_arguments.json,
+        list_files=create_arguments.list_files,
+        stream_processes=stream_processes,
+    )
+    if json_output:
+        yield borgmatic.actions.json.parse_json(json_output, repository.get('label'))
+
+    borgmatic.hooks.dispatch.call_hooks_even_if_unconfigured(
+        'remove_data_source_dumps',
+        config,
+        config_filename,
+        borgmatic.hooks.dump.DATA_SOURCE_HOOK_NAMES,
+        global_arguments.dry_run,
+    )
+    borgmatic.hooks.command.execute_hook(
+        config.get('after_backup'),
+        config.get('umask'),
+        config_filename,
+        'post-backup',
+        global_arguments.dry_run,
+        **hook_context,
+    )

borgmatic/actions/export_key.py

@@ -0,0 +1,33 @@
+import logging
+
+import borgmatic.borg.export_key
+import borgmatic.config.validate
+
+logger = logging.getLogger(__name__)
+
+
+def run_export_key(
+    repository,
+    config,
+    local_borg_version,
+    export_arguments,
+    global_arguments,
+    local_path,
+    remote_path,
+):
+    '''
+    Run the "key export" action for the given repository.
+    '''
+    if export_arguments.repository is None or borgmatic.config.validate.repositories_match(
+        repository, export_arguments.repository
+    ):
+        logger.info(f'{repository.get("label", repository["path"])}: Exporting repository key')
+        borgmatic.borg.export_key.export_key(
+            repository['path'],
+            config,
+            local_borg_version,
+            export_arguments,
+            global_arguments,
+            local_path=local_path,
+            remote_path=remote_path,
+        )

borgmatic/actions/export_tar.py

@@ -0,0 +1,50 @@
+import logging
+
+import borgmatic.borg.export_tar
+import borgmatic.borg.rlist
+import borgmatic.config.validate
+
+logger = logging.getLogger(__name__)
+
+
+def run_export_tar(
+    repository,
+    config,
+    local_borg_version,
+    export_tar_arguments,
+    global_arguments,
+    local_path,
+    remote_path,
+):
+    '''
+    Run the "export-tar" action for the given repository.
+    '''
+    if export_tar_arguments.repository is None or borgmatic.config.validate.repositories_match(
+        repository, export_tar_arguments.repository
+    ):
+        logger.info(
+            f'{repository["path"]}: Exporting archive {export_tar_arguments.archive} as tar file'
+        )
+        borgmatic.borg.export_tar.export_tar_archive(
+            global_arguments.dry_run,
+            repository['path'],
+            borgmatic.borg.rlist.resolve_archive_name(
+                repository['path'],
+                export_tar_arguments.archive,
+                config,
+                local_borg_version,
+                global_arguments,
+                local_path,
+                remote_path,
+            ),
+            export_tar_arguments.paths,
+            export_tar_arguments.destination,
+            config,
+            local_borg_version,
+            global_arguments,
+            local_path=local_path,
+            remote_path=remote_path,
+            tar_filter=export_tar_arguments.tar_filter,
+            list_files=export_tar_arguments.list_files,
+            strip_components=export_tar_arguments.strip_components,
+        )

borgmatic/actions/extract.py

@@ -0,0 +1,68 @@
+import logging
+
+import borgmatic.borg.extract
+import borgmatic.borg.rlist
+import borgmatic.config.validate
+import borgmatic.hooks.command
+
+logger = logging.getLogger(__name__)
+
+
+def run_extract(
+    config_filename,
+    repository,
+    config,
+    hook_context,
+    local_borg_version,
+    extract_arguments,
+    global_arguments,
+    local_path,
+    remote_path,
+):
+    '''
+    Run the "extract" action for the given repository.
+    '''
+    borgmatic.hooks.command.execute_hook(
+        config.get('before_extract'),
+        config.get('umask'),
+        config_filename,
+        'pre-extract',
+        global_arguments.dry_run,
+        **hook_context,
+    )
+    if extract_arguments.repository is None or borgmatic.config.validate.repositories_match(
+        repository, extract_arguments.repository
+    ):
+        logger.info(
+            f'{repository.get("label", repository["path"])}: Extracting archive {extract_arguments.archive}'
+        )
+        borgmatic.borg.extract.extract_archive(
+            global_arguments.dry_run,
+            repository['path'],
+            borgmatic.borg.rlist.resolve_archive_name(
+                repository['path'],
+                extract_arguments.archive,
+                config,
+                local_borg_version,
+                global_arguments,
+                local_path,
+                remote_path,
+            ),
+            extract_arguments.paths,
+            config,
+            local_borg_version,
+            global_arguments,
+            local_path=local_path,
+            remote_path=remote_path,
+            destination_path=extract_arguments.destination,
+            strip_components=extract_arguments.strip_components,
+            progress=extract_arguments.progress,
+        )
+    borgmatic.hooks.command.execute_hook(
+        config.get('after_extract'),
+        config.get('umask'),
+        config_filename,
+        'post-extract',
+        global_arguments.dry_run,
+        **hook_context,
+    )

borgmatic/actions/info.py

@@ -0,0 +1,52 @@
+import logging
+
+import borgmatic.actions.arguments
+import borgmatic.actions.json
+import borgmatic.borg.info
+import borgmatic.borg.rlist
+import borgmatic.config.validate
+
+logger = logging.getLogger(__name__)
+
+
+def run_info(
+    repository,
+    config,
+    local_borg_version,
+    info_arguments,
+    global_arguments,
+    local_path,
+    remote_path,
+):
+    '''
+    Run the "info" action for the given repository and archive.
+
+    If info_arguments.json is True, yield the JSON output from the info for the archive.
+    '''
+    if info_arguments.repository is None or borgmatic.config.validate.repositories_match(
+        repository, info_arguments.repository
+    ):
+        if not info_arguments.json:
+            logger.answer(
+                f'{repository.get("label", repository["path"])}: Displaying archive summary information'
+            )
+        archive_name = borgmatic.borg.rlist.resolve_archive_name(
+            repository['path'],
+            info_arguments.archive,
+            config,
+            local_borg_version,
+            global_arguments,
+            local_path,
+            remote_path,
+        )
+        json_output = borgmatic.borg.info.display_archives_info(
+            repository['path'],
+            config,
+            local_borg_version,
+            borgmatic.actions.arguments.update_arguments(info_arguments, archive=archive_name),
+            global_arguments,
+            local_path,
+            remote_path,
+        )
+        if json_output:
+            yield borgmatic.actions.json.parse_json(json_output, repository.get('label'))

borgmatic/actions/json.py

@@ -0,0 +1,30 @@
+import json
+import logging
+
+logger = logging.getLogger(__name__)
+
+
+def parse_json(borg_json_output, label):
+    '''
+    Given a Borg JSON output string, parse it as JSON into a dict. Inject the given borgmatic
+    repository label into it and return the dict.
+
+    Raise JSONDecodeError if the JSON output cannot be parsed.
+    '''
+    lines = borg_json_output.splitlines()
+    start_line_index = 0
+
+    # Scan forward to find the first line starting with "{" and assume that's where the JSON starts.
+    for line_index, line in enumerate(lines):
+        if line.startswith('{'):
+            start_line_index = line_index
+            break
+
+    json_data = json.loads('\n'.join(lines[start_line_index:]))
+
+    if 'repository' not in json_data:
+        return json_data
+
+    json_data['repository']['label'] = label or ''
+
+    return json_data

borgmatic/actions/list.py

@@ -0,0 +1,53 @@
+import logging
+
+import borgmatic.actions.arguments
+import borgmatic.actions.json
+import borgmatic.borg.list
+import borgmatic.config.validate
+
+logger = logging.getLogger(__name__)
+
+
+def run_list(
+    repository,
+    config,
+    local_borg_version,
+    list_arguments,
+    global_arguments,
+    local_path,
+    remote_path,
+):
+    '''
+    Run the "list" action for the given repository and archive.
+
+    If list_arguments.json is True, yield the JSON output from listing the archive.
+    '''
+    if list_arguments.repository is None or borgmatic.config.validate.repositories_match(
+        repository, list_arguments.repository
+    ):
+        if not list_arguments.json:
+            if list_arguments.find_paths:  # pragma: no cover
+                logger.answer(f'{repository.get("label", repository["path"])}: Searching archives')
+            elif not list_arguments.archive:  # pragma: no cover
+                logger.answer(f'{repository.get("label", repository["path"])}: Listing archives')
+
+        archive_name = borgmatic.borg.rlist.resolve_archive_name(
+            repository['path'],
+            list_arguments.archive,
+            config,
+            local_borg_version,
+            global_arguments,
+            local_path,
+            remote_path,
+        )
+        json_output = borgmatic.borg.list.list_archive(
+            repository['path'],
+            config,
+            local_borg_version,
+            borgmatic.actions.arguments.update_arguments(list_arguments, archive=archive_name),
+            global_arguments,
+            local_path,
+            remote_path,
+        )
+        if json_output:
+            yield borgmatic.actions.json.parse_json(json_output, repository.get('label'))

borgmatic/actions/mount.py

@@ -0,0 +1,49 @@
+import logging
+
+import borgmatic.borg.mount
+import borgmatic.borg.rlist
+import borgmatic.config.validate
+
+logger = logging.getLogger(__name__)
+
+
+def run_mount(
+    repository,
+    config,
+    local_borg_version,
+    mount_arguments,
+    global_arguments,
+    local_path,
+    remote_path,
+):
+    '''
+    Run the "mount" action for the given repository.
+    '''
+    if mount_arguments.repository is None or borgmatic.config.validate.repositories_match(
+        repository, mount_arguments.repository
+    ):
+        if mount_arguments.archive:
+            logger.info(
+                f'{repository.get("label", repository["path"])}: Mounting archive {mount_arguments.archive}'
+            )
+        else:  # pragma: nocover
+            logger.info(f'{repository.get("label", repository["path"])}: Mounting repository')
+
+        borgmatic.borg.mount.mount_archive(
+            repository['path'],
+            borgmatic.borg.rlist.resolve_archive_name(
+                repository['path'],
+                mount_arguments.archive,
+                config,
+                local_borg_version,
+                global_arguments,
+                local_path,
+                remote_path,
+            ),
+            mount_arguments,
+            config,
+            local_borg_version,
+            global_arguments,
+            local_path=local_path,
+            remote_path=remote_path,
+        )

borgmatic/actions/prune.py

@@ -0,0 +1,56 @@
+import logging
+
+import borgmatic.borg.prune
+import borgmatic.config.validate
+import borgmatic.hooks.command
+
+logger = logging.getLogger(__name__)
+
+
+def run_prune(
+    config_filename,
+    repository,
+    config,
+    hook_context,
+    local_borg_version,
+    prune_arguments,
+    global_arguments,
+    dry_run_label,
+    local_path,
+    remote_path,
+):
+    '''
+    Run the "prune" action for the given repository.
+    '''
+    if prune_arguments.repository and not borgmatic.config.validate.repositories_match(
+        repository, prune_arguments.repository
+    ):
+        return
+
+    borgmatic.hooks.command.execute_hook(
+        config.get('before_prune'),
+        config.get('umask'),
+        config_filename,
+        'pre-prune',
+        global_arguments.dry_run,
+        **hook_context,
+    )
+    logger.info(f'{repository.get("label", repository["path"])}: Pruning archives{dry_run_label}')
+    borgmatic.borg.prune.prune_archives(
+        global_arguments.dry_run,
+        repository['path'],
+        config,
+        local_borg_version,
+        prune_arguments,
+        global_arguments,
+        local_path=local_path,
+        remote_path=remote_path,
+    )
+    borgmatic.hooks.command.execute_hook(
+        config.get('after_prune'),
+        config.get('umask'),
+        config_filename,
+        'post-prune',
+        global_arguments.dry_run,
+        **hook_context,
+    )

borgmatic/actions/rcreate.py

@@ -0,0 +1,41 @@
+import logging
+
+import borgmatic.borg.rcreate
+import borgmatic.config.validate
+
+logger = logging.getLogger(__name__)
+
+
+def run_rcreate(
+    repository,
+    config,
+    local_borg_version,
+    rcreate_arguments,
+    global_arguments,
+    local_path,
+    remote_path,
+):
+    '''
+    Run the "rcreate" action for the given repository.
+    '''
+    if rcreate_arguments.repository and not borgmatic.config.validate.repositories_match(
+        repository, rcreate_arguments.repository
+    ):
+        return
+
+    logger.info(f'{repository.get("label", repository["path"])}: Creating repository')
+    borgmatic.borg.rcreate.create_repository(
+        global_arguments.dry_run,
+        repository['path'],
+        config,
+        local_borg_version,
+        global_arguments,
+        rcreate_arguments.encryption_mode,
+        rcreate_arguments.source_repository,
+        rcreate_arguments.copy_crypt_key,
+        rcreate_arguments.append_only,
+        rcreate_arguments.storage_quota,
+        rcreate_arguments.make_parent_dirs,
+        local_path=local_path,
+        remote_path=remote_path,
+    )

borgmatic/actions/restore.py

@@ -0,0 +1,380 @@
+import copy
+import logging
+import os
+
+import borgmatic.borg.extract
+import borgmatic.borg.list
+import borgmatic.borg.mount
+import borgmatic.borg.rlist
+import borgmatic.borg.state
+import borgmatic.config.validate
+import borgmatic.hooks.dispatch
+import borgmatic.hooks.dump
+
+logger = logging.getLogger(__name__)
+
+
+UNSPECIFIED_HOOK = object()
+
+
+def get_configured_data_source(
+    config,
+    archive_data_source_names,
+    hook_name,
+    data_source_name,
+    configuration_data_source_name=None,
+):
+    '''
+    Find the first data source with the given hook name and data source name in the configuration
+    dict and the given archive data source names dict (from hook name to data source names contained
+    in a particular backup archive). If UNSPECIFIED_HOOK is given as the hook name, search all data
+    source hooks for the named data source. If a configuration data source name is given, use that
+    instead of the data source name to lookup the data source in the given hooks configuration.
+
+    Return the found data source as a tuple of (found hook name, data source configuration dict) or
+    (None, None) if not found.
+    '''
+    if not configuration_data_source_name:
+        configuration_data_source_name = data_source_name
+
+    if hook_name == UNSPECIFIED_HOOK:
+        hooks_to_search = {
+            hook_name: value
+            for (hook_name, value) in config.items()
+            if hook_name in borgmatic.hooks.dump.DATA_SOURCE_HOOK_NAMES
+        }
+    else:
+        try:
+            hooks_to_search = {hook_name: config[hook_name]}
+        except KeyError:
+            return (None, None)
+
+    return next(
+        (
+            (name, hook_data_source)
+            for (name, hook) in hooks_to_search.items()
+            for hook_data_source in hook
+            if hook_data_source['name'] == configuration_data_source_name
+            and data_source_name in archive_data_source_names.get(name, [])
+        ),
+        (None, None),
+    )
+
+
+def restore_single_data_source(
+    repository,
+    config,
+    local_borg_version,
+    global_arguments,
+    local_path,
+    remote_path,
+    archive_name,
+    hook_name,
+    data_source,
+    connection_params,
+):  # pragma: no cover
+    '''
+    Given (among other things) an archive name, a data source hook name, the hostname, port,
+    username/password as connection params, and a configured data source configuration dict, restore
+    that data source from the archive.
+    '''
+    logger.info(
+        f'{repository.get("label", repository["path"])}: Restoring data source {data_source["name"]}'
+    )
+
+    dump_pattern = borgmatic.hooks.dispatch.call_hooks(
+        'make_data_source_dump_pattern',
+        config,
+        repository['path'],
+        borgmatic.hooks.dump.DATA_SOURCE_HOOK_NAMES,
+        data_source['name'],
+    )[hook_name]
+
+    # Kick off a single data source extract to stdout.
+    extract_process = borgmatic.borg.extract.extract_archive(
+        dry_run=global_arguments.dry_run,
+        repository=repository['path'],
+        archive=archive_name,
+        paths=borgmatic.hooks.dump.convert_glob_patterns_to_borg_patterns([dump_pattern]),
+        config=config,
+        local_borg_version=local_borg_version,
+        global_arguments=global_arguments,
+        local_path=local_path,
+        remote_path=remote_path,
+        destination_path='/',
+        # A directory format dump isn't a single file, and therefore can't extract
+        # to stdout. In this case, the extract_process return value is None.
+        extract_to_stdout=bool(data_source.get('format') != 'directory'),
+    )
+
+    # Run a single data source restore, consuming the extract stdout (if any).
+    borgmatic.hooks.dispatch.call_hooks(
+        function_name='restore_data_source_dump',
+        config=config,
+        log_prefix=repository['path'],
+        hook_names=[hook_name],
+        data_source=data_source,
+        dry_run=global_arguments.dry_run,
+        extract_process=extract_process,
+        connection_params=connection_params,
+    )
+
+
+def collect_archive_data_source_names(
+    repository,
+    archive,
+    config,
+    local_borg_version,
+    global_arguments,
+    local_path,
+    remote_path,
+):
+    '''
+    Given a local or remote repository path, a resolved archive name, a configuration dict, the
+    local Borg version, global_arguments an argparse.Namespace, and local and remote Borg paths,
+    query the archive for the names of data sources it contains as dumps and return them as a dict
+    from hook name to a sequence of data source names.
+    '''
+    borgmatic_source_directory = os.path.expanduser(
+        config.get(
+            'borgmatic_source_directory', borgmatic.borg.state.DEFAULT_BORGMATIC_SOURCE_DIRECTORY
+        )
+    ).lstrip('/')
+    dump_paths = borgmatic.borg.list.capture_archive_listing(
+        repository,
+        archive,
+        config,
+        local_borg_version,
+        global_arguments,
+        list_paths=[
+            os.path.expanduser(
+                borgmatic.hooks.dump.make_data_source_dump_path(borgmatic_source_directory, pattern)
+            )
+            for pattern in ('*_databases/*/*',)
+        ],
+        local_path=local_path,
+        remote_path=remote_path,
+    )
+
+    # Determine the data source names corresponding to the dumps found in the archive and
+    # add them to restore_names.
+    archive_data_source_names = {}
+
+    for dump_path in dump_paths:
+        try:
+            (hook_name, _, data_source_name) = dump_path.split(
+                borgmatic_source_directory + os.path.sep, 1
+            )[1].split(os.path.sep)[0:3]
+        except (ValueError, IndexError):
+            logger.warning(
+                f'{repository}: Ignoring invalid data source dump path "{dump_path}" in archive {archive}'
+            )
+        else:
+            if data_source_name not in archive_data_source_names.get(hook_name, []):
+                archive_data_source_names.setdefault(hook_name, []).extend([data_source_name])
+
+    return archive_data_source_names
+
+
+def find_data_sources_to_restore(requested_data_source_names, archive_data_source_names):
+    '''
+    Given a sequence of requested data source names to restore and a dict of hook name to the names
+    of data sources found in an archive, return an expanded sequence of data source names to
+    restore, replacing "all" with actual data source names as appropriate.
+
+    Raise ValueError if any of the requested data source names cannot be found in the archive.
+    '''
+    # A map from data source hook name to the data source names to restore for that hook.
+    restore_names = (
+        {UNSPECIFIED_HOOK: requested_data_source_names}
+        if requested_data_source_names
+        else {UNSPECIFIED_HOOK: ['all']}
+    )
+
+    # If "all" is in restore_names, then replace it with the names of dumps found within the
+    # archive.
+    if 'all' in restore_names[UNSPECIFIED_HOOK]:
+        restore_names[UNSPECIFIED_HOOK].remove('all')
+
+        for hook_name, data_source_names in archive_data_source_names.items():
+            restore_names.setdefault(hook_name, []).extend(data_source_names)
+
+            # If a data source is to be restored as part of "all", then remove it from restore names
+            # so it doesn't get restored twice.
+            for data_source_name in data_source_names:
+                if data_source_name in restore_names[UNSPECIFIED_HOOK]:
+                    restore_names[UNSPECIFIED_HOOK].remove(data_source_name)
+
+    if not restore_names[UNSPECIFIED_HOOK]:
+        restore_names.pop(UNSPECIFIED_HOOK)
+
+    combined_restore_names = set(
+        name for data_source_names in restore_names.values() for name in data_source_names
+    )
+    combined_archive_data_source_names = set(
+        name
+        for data_source_names in archive_data_source_names.values()
+        for name in data_source_names
+    )
+
+    missing_names = sorted(set(combined_restore_names) - combined_archive_data_source_names)
+    if missing_names:
+        joined_names = ', '.join(f'"{name}"' for name in missing_names)
+        raise ValueError(
+            f"Cannot restore data source{'s' if len(missing_names) > 1 else ''} {joined_names} missing from archive"
+        )
+
+    return restore_names
+
+
+def ensure_data_sources_found(restore_names, remaining_restore_names, found_names):
+    '''
+    Given a dict from hook name to data source names to restore, a dict from hook name to remaining
+    data source names to restore, and a sequence of found (actually restored) data source names,
+    raise ValueError if requested data source to restore were missing from the archive and/or
+    configuration.
+    '''
+    combined_restore_names = set(
+        name
+        for data_source_names in tuple(restore_names.values())
+        + tuple(remaining_restore_names.values())
+        for name in data_source_names
+    )
+
+    if not combined_restore_names and not found_names:
+        raise ValueError('No data sources were found to restore')
+
+    missing_names = sorted(set(combined_restore_names) - set(found_names))
+    if missing_names:
+        joined_names = ', '.join(f'"{name}"' for name in missing_names)
+        raise ValueError(
+            f"Cannot restore data source{'s' if len(missing_names) > 1 else ''} {joined_names} missing from borgmatic's configuration"
+        )
+
+
+def run_restore(
+    repository,
+    config,
+    local_borg_version,
+    restore_arguments,
+    global_arguments,
+    local_path,
+    remote_path,
+):
+    '''
+    Run the "restore" action for the given repository, but only if the repository matches the
+    requested repository in restore arguments.
+
+    Raise ValueError if a configured data source could not be found to restore.
+    '''
+    if restore_arguments.repository and not borgmatic.config.validate.repositories_match(
+        repository, restore_arguments.repository
+    ):
+        return
+
+    logger.info(
+        f'{repository.get("label", repository["path"])}: Restoring data sources from archive {restore_arguments.archive}'
+    )
+
+    borgmatic.hooks.dispatch.call_hooks_even_if_unconfigured(
+        'remove_data_source_dumps',
+        config,
+        repository['path'],
+        borgmatic.hooks.dump.DATA_SOURCE_HOOK_NAMES,
+        global_arguments.dry_run,
+    )
+
+    archive_name = borgmatic.borg.rlist.resolve_archive_name(
+        repository['path'],
+        restore_arguments.archive,
+        config,
+        local_borg_version,
+        global_arguments,
+        local_path,
+        remote_path,
+    )
+    archive_data_source_names = collect_archive_data_source_names(
+        repository['path'],
+        archive_name,
+        config,
+        local_borg_version,
+        global_arguments,
+        local_path,
+        remote_path,
+    )
+    restore_names = find_data_sources_to_restore(
+        restore_arguments.data_sources, archive_data_source_names
+    )
+    found_names = set()
+    remaining_restore_names = {}
+    connection_params = {
+        'hostname': restore_arguments.hostname,
+        'port': restore_arguments.port,
+        'username': restore_arguments.username,
+        'password': restore_arguments.password,
+        'restore_path': restore_arguments.restore_path,
+    }
+
+    for hook_name, data_source_names in restore_names.items():
+        for data_source_name in data_source_names:
+            found_hook_name, found_data_source = get_configured_data_source(
+                config, archive_data_source_names, hook_name, data_source_name
+            )
+
+            if not found_data_source:
+                remaining_restore_names.setdefault(found_hook_name or hook_name, []).append(
+                    data_source_name
+                )
+                continue
+
+            found_names.add(data_source_name)
+            restore_single_data_source(
+                repository,
+                config,
+                local_borg_version,
+                global_arguments,
+                local_path,
+                remote_path,
+                archive_name,
+                found_hook_name or hook_name,
+                dict(found_data_source, **{'schemas': restore_arguments.schemas}),
+                connection_params,
+            )
+
+    # For any data sources that weren't found via exact matches in the configuration, try to
+    # fallback to "all" entries.
+    for hook_name, data_source_names in remaining_restore_names.items():
+        for data_source_name in data_source_names:
+            found_hook_name, found_data_source = get_configured_data_source(
+                config, archive_data_source_names, hook_name, data_source_name, 'all'
+            )
+
+            if not found_data_source:
+                continue
+
+            found_names.add(data_source_name)
+            data_source = copy.copy(found_data_source)
+            data_source['name'] = data_source_name
+
+            restore_single_data_source(
+                repository,
+                config,
+                local_borg_version,
+                global_arguments,
+                local_path,
+                remote_path,
+                archive_name,
+                found_hook_name or hook_name,
+                dict(data_source, **{'schemas': restore_arguments.schemas}),
+                connection_params,
+            )
+
+    borgmatic.hooks.dispatch.call_hooks_even_if_unconfigured(
+        'remove_data_source_dumps',
+        config,
+        repository['path'],
+        borgmatic.hooks.dump.DATA_SOURCE_HOOK_NAMES,
+        global_arguments.dry_run,
+    )
+
+    ensure_data_sources_found(restore_names, remaining_restore_names, found_names)

borgmatic/actions/rinfo.py

@@ -0,0 +1,42 @@
+import logging
+
+import borgmatic.actions.json
+import borgmatic.borg.rinfo
+import borgmatic.config.validate
+
+logger = logging.getLogger(__name__)
+
+
+def run_rinfo(
+    repository,
+    config,
+    local_borg_version,
+    rinfo_arguments,
+    global_arguments,
+    local_path,
+    remote_path,
+):
+    '''
+    Run the "rinfo" action for the given repository.
+
+    If rinfo_arguments.json is True, yield the JSON output from the info for the repository.
+    '''
+    if rinfo_arguments.repository is None or borgmatic.config.validate.repositories_match(
+        repository, rinfo_arguments.repository
+    ):
+        if not rinfo_arguments.json:
+            logger.answer(
+                f'{repository.get("label", repository["path"])}: Displaying repository summary information'
+            )
+
+        json_output = borgmatic.borg.rinfo.display_repository_info(
+            repository['path'],
+            config,
+            local_borg_version,
+            rinfo_arguments=rinfo_arguments,
+            global_arguments=global_arguments,
+            local_path=local_path,
+            remote_path=remote_path,
+        )
+        if json_output:
+            yield borgmatic.actions.json.parse_json(json_output, repository.get('label'))

borgmatic/actions/rlist.py

@@ -0,0 +1,40 @@
+import logging
+
+import borgmatic.actions.json
+import borgmatic.borg.rlist
+import borgmatic.config.validate
+
+logger = logging.getLogger(__name__)
+
+
+def run_rlist(
+    repository,
+    config,
+    local_borg_version,
+    rlist_arguments,
+    global_arguments,
+    local_path,
+    remote_path,
+):
+    '''
+    Run the "rlist" action for the given repository.
+
+    If rlist_arguments.json is True, yield the JSON output from listing the repository.
+    '''
+    if rlist_arguments.repository is None or borgmatic.config.validate.repositories_match(
+        repository, rlist_arguments.repository
+    ):
+        if not rlist_arguments.json:
+            logger.answer(f'{repository.get("label", repository["path"])}: Listing repository')
+
+        json_output = borgmatic.borg.rlist.list_repository(
+            repository['path'],
+            config,
+            local_borg_version,
+            rlist_arguments=rlist_arguments,
+            global_arguments=global_arguments,
+            local_path=local_path,
+            remote_path=remote_path,
+        )
+        if json_output:
+            yield borgmatic.actions.json.parse_json(json_output, repository.get('label'))

borgmatic/actions/transfer.py

@@ -0,0 +1,32 @@
+import logging
+
+import borgmatic.borg.transfer
+
+logger = logging.getLogger(__name__)
+
+
+def run_transfer(
+    repository,
+    config,
+    local_borg_version,
+    transfer_arguments,
+    global_arguments,
+    local_path,
+    remote_path,
+):
+    '''
+    Run the "transfer" action for the given repository.
+    '''
+    logger.info(
+        f'{repository.get("label", repository["path"])}: Transferring archives to repository'
+    )
+    borgmatic.borg.transfer.transfer_archives(
+        global_arguments.dry_run,
+        repository['path'],
+        config,
+        local_borg_version,
+        transfer_arguments,
+        global_arguments,
+        local_path=local_path,
+        remote_path=remote_path,
+    )

borgmatic/borg/borg.py

@@ -0,0 +1,72 @@
+import logging
+import shlex
+
+import borgmatic.commands.arguments
+import borgmatic.logger
+from borgmatic.borg import environment, flags
+from borgmatic.execute import DO_NOT_CAPTURE, execute_command
+
+logger = logging.getLogger(__name__)
+
+
+BORG_SUBCOMMANDS_WITH_SUBCOMMANDS = {'key', 'debug'}
+
+
+def run_arbitrary_borg(
+    repository_path,
+    config,
+    local_borg_version,
+    options,
+    archive=None,
+    local_path='borg',
+    remote_path=None,
+):
+    '''
+    Given a local or remote repository path, a configuration dict, the local Borg version, a
+    sequence of arbitrary command-line Borg options, and an optional archive name, run an arbitrary
+    Borg command, passing in REPOSITORY and ARCHIVE environment variables for optional use in the
+    command.
+    '''
+    borgmatic.logger.add_custom_log_levels()
+    lock_wait = config.get('lock_wait', None)
+
+    try:
+        options = options[1:] if options[0] == '--' else options
+
+        # Borg commands like "key" have a sub-command ("export", etc.) that must follow it.
+        command_options_start_index = 2 if options[0] in BORG_SUBCOMMANDS_WITH_SUBCOMMANDS else 1
+        borg_command = tuple(options[:command_options_start_index])
+        command_options = tuple(options[command_options_start_index:])
+
+        if borg_command and borg_command[0] in borgmatic.commands.arguments.ACTION_ALIASES.keys():
+            logger.warning(
+                f"Borg's {borg_command[0]} subcommand is supported natively by borgmatic. Try this instead: borgmatic {borg_command[0]}"
+            )
+    except IndexError:
+        borg_command = ()
+        command_options = ()
+
+    full_command = (
+        (local_path,)
+        + borg_command
+        + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
+        + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
+        + flags.make_flags('remote-path', remote_path)
+        + flags.make_flags('lock-wait', lock_wait)
+        + command_options
+    )
+
+    return execute_command(
+        tuple(shlex.quote(part) for part in full_command),
+        output_file=DO_NOT_CAPTURE,
+        shell=True,
+        extra_environment=dict(
+            (environment.make_environment(config) or {}),
+            **{
+                'BORG_REPO': repository_path,
+                'ARCHIVE': archive if archive else '',
+            },
+        ),
+        borg_local_path=local_path,
+        borg_exit_codes=config.get('borg_exit_codes'),
+    )

borgmatic/borg/break_lock.py

@@ -0,0 +1,42 @@
+import logging
+
+from borgmatic.borg import environment, flags
+from borgmatic.execute import execute_command
+
+logger = logging.getLogger(__name__)
+
+
+def break_lock(
+    repository_path,
+    config,
+    local_borg_version,
+    global_arguments,
+    local_path='borg',
+    remote_path=None,
+):
+    '''
+    Given a local or remote repository path, a configuration dict, the local Borg version, an
+    argparse.Namespace of global arguments, and optional local and remote Borg paths, break any
+    repository and cache locks leftover from Borg aborting.
+    '''
+    umask = config.get('umask', None)
+    lock_wait = config.get('lock_wait', None)
+
+    full_command = (
+        (local_path, 'break-lock')
+        + (('--remote-path', remote_path) if remote_path else ())
+        + (('--umask', str(umask)) if umask else ())
+        + (('--log-json',) if global_arguments.log_json else ())
+        + (('--lock-wait', str(lock_wait)) if lock_wait else ())
+        + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
+        + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
+        + flags.make_repository_flags(repository_path, local_borg_version)
+    )
+
+    borg_environment = environment.make_environment(config)
+    execute_command(
+        full_command,
+        extra_environment=borg_environment,
+        borg_local_path=local_path,
+        borg_exit_codes=config.get('borg_exit_codes'),
+    )

borgmatic/borg/check.py

@@ -1,44 +1,59 @@
+import argparse
+import json
 import logging
-import os
-import subprocess
-
-from borgmatic.borg import extract
-
-
-DEFAULT_CHECKS = ('repository', 'archives')
-DEFAULT_PREFIX = '{hostname}-'
 
+from borgmatic.borg import environment, feature, flags, rinfo
+from borgmatic.execute import DO_NOT_CAPTURE, execute_command
 
 logger = logging.getLogger(__name__)
 
 
-def _parse_checks(consistency_config):
+def make_archive_filter_flags(local_borg_version, config, checks, check_arguments):
     '''
-    Given a consistency config with a "checks" list, transform it to a tuple of named checks to run.
+    Given the local Borg version, a configuration dict, a parsed sequence of checks, and check
+    arguments as an argparse.Namespace instance, transform the checks into tuple of command-line
+    flags for filtering archives in a check command.
 
-    For example, given a retention config of:
-
-        {'checks': ['repository', 'archives']}
-
-    This will be returned as:
-
-        ('repository', 'archives')
-
-    If no "checks" option is present, return the DEFAULT_CHECKS. If the checks value is the string
-    "disabled", return an empty tuple, meaning that no checks should be run.
+    If "check_last" is set in the configuration and "archives" is in checks, then include a "--last"
+    flag. And if "prefix" is set in configuration and "archives" is in checks, then include a
+    "--match-archives" flag.
     '''
-    checks = consistency_config.get('checks', [])
-    if checks == ['disabled']:
-        return ()
+    check_last = config.get('check_last', None)
+    prefix = config.get('prefix')
 
-    return (
-        tuple(check for check in checks if check.lower() not in ('disabled', '')) or DEFAULT_CHECKS
-    )
+    if 'archives' in checks or 'data' in checks:
+        return (('--last', str(check_last)) if check_last else ()) + (
+            (
+                ('--match-archives', f'sh:{prefix}*')
+                if feature.available(feature.Feature.MATCH_ARCHIVES, local_borg_version)
+                else ('--glob-archives', f'{prefix}*')
+            )
+            if prefix
+            else (
+                flags.make_match_archives_flags(
+                    check_arguments.match_archives or config.get('match_archives'),
+                    config.get('archive_name_format'),
+                    local_borg_version,
+                )
+            )
+        )
+
+    if check_last:
+        logger.warning(
+            'Ignoring check_last option, as "archives" or "data" are not in consistency checks'
+        )
+    if prefix:
+        logger.warning(
+            'Ignoring consistency prefix option, as "archives" or "data" are not in consistency checks'
+        )
+
+    return ()
 
 
-def _make_check_flags(checks, check_last=None, prefix=None):
+def make_check_flags(checks, archive_filter_flags):
     '''
-    Given a parsed sequence of checks, transform it into tuple of command-line flags.
+    Given a parsed checks set and a sequence of flags to filter archives,
+    transform the checks into tuple of command-line check flags.
 
     For example, given parsed checks of:
 
@@ -49,76 +64,106 @@ def _make_check_flags(checks, check_last=None, prefix=None):
         ('--repository-only',)
 
     However, if both "repository" and "archives" are in checks, then omit them from the returned
-    flags because Borg does both checks by default.
-
-    Additionally, if a check_last value is given and "archives" is in checks, then include a
-    "--last" flag. And if a prefix value is given and "archives" is in checks, then include a
-    "--prefix" flag.
+    flags because Borg does both checks by default. If "data" is in checks, that implies "archives".
     '''
-    if 'archives' in checks:
-        last_flags = ('--last', str(check_last)) if check_last else ()
-        prefix_flags = ('--prefix', prefix) if prefix else ('--prefix', DEFAULT_PREFIX)
+    if 'data' in checks:
+        data_flags = ('--verify-data',)
+        checks.update({'archives'})
     else:
-        last_flags = ()
-        prefix_flags = ()
-        if check_last:
-            logger.warning(
-                'Ignoring check_last option, as "archives" is not in consistency checks.'
-            )
-        if prefix:
-            logger.warning(
-                'Ignoring consistency prefix option, as "archives" is not in consistency checks.'
-            )
+        data_flags = ()
 
-    if set(DEFAULT_CHECKS).issubset(set(checks)):
-        return last_flags + prefix_flags
+    common_flags = (archive_filter_flags if 'archives' in checks else ()) + data_flags
+
+    if {'repository', 'archives'}.issubset(checks):
+        return common_flags
 
     return (
-        tuple('--{}-only'.format(check) for check in checks if check in DEFAULT_CHECKS)
-        + last_flags
-        + prefix_flags
+        tuple(f'--{check}-only' for check in checks if check in ('repository', 'archives'))
+        + common_flags
     )
 
 
-def check_archives(
-    repository, storage_config, consistency_config, local_path='borg', remote_path=None
+def get_repository_id(
+    repository_path, config, local_borg_version, global_arguments, local_path, remote_path
 ):
     '''
-    Given a local or remote repository path, a storage config dict, a consistency config dict,
-    and a local/remote commands to run, check the contained Borg archives for consistency.
+    Given a local or remote repository path, a configuration dict, the local Borg version, global
+    arguments, and local/remote commands to run, return the corresponding Borg repository ID.
 
-    If there are no consistency checks to run, skip running them.
+    Raise ValueError if the Borg repository ID cannot be determined.
     '''
-    checks = _parse_checks(consistency_config)
-    check_last = consistency_config.get('check_last', None)
-    lock_wait = None
+    try:
+        return json.loads(
+            rinfo.display_repository_info(
+                repository_path,
+                config,
+                local_borg_version,
+                argparse.Namespace(json=True),
+                global_arguments,
+                local_path,
+                remote_path,
+            )
+        )['repository']['id']
+    except (json.JSONDecodeError, KeyError):
+        raise ValueError(f'Cannot determine Borg repository ID for {repository_path}')
 
-    if set(checks).intersection(set(DEFAULT_CHECKS)):
-        remote_path_flags = ('--remote-path', remote_path) if remote_path else ()
-        lock_wait = storage_config.get('lock_wait', None)
-        lock_wait_flags = ('--lock-wait', str(lock_wait)) if lock_wait else ()
 
-        verbosity_flags = ()
-        if logger.isEnabledFor(logging.INFO):
-            verbosity_flags = ('--info',)
-        if logger.isEnabledFor(logging.DEBUG):
-            verbosity_flags = ('--debug', '--show-rc')
+def check_archives(
+    repository_path,
+    config,
+    local_borg_version,
+    check_arguments,
+    global_arguments,
+    checks,
+    archive_filter_flags,
+    local_path='borg',
+    remote_path=None,
+):
+    '''
+    Given a local or remote repository path, a configuration dict, the local Borg version, check
+    arguments as an argparse.Namespace instance, global arguments, a set of named Borg checks to run
+    (some combination "repository", "archives", and/or "data"), archive filter flags, and
+    local/remote commands to run, check the contained Borg archives for consistency.
+    '''
+    lock_wait = config.get('lock_wait')
+    extra_borg_options = config.get('extra_borg_options', {}).get('check', '')
 
-        prefix = consistency_config.get('prefix')
+    verbosity_flags = ()
+    if logger.isEnabledFor(logging.INFO):
+        verbosity_flags = ('--info',)
+    if logger.isEnabledFor(logging.DEBUG):
+        verbosity_flags = ('--debug', '--show-rc')
 
-        full_command = (
-            (local_path, 'check', repository)
-            + _make_check_flags(checks, check_last, prefix)
-            + remote_path_flags
-            + lock_wait_flags
-            + verbosity_flags
+    full_command = (
+        (local_path, 'check')
+        + (('--repair',) if check_arguments.repair else ())
+        + make_check_flags(checks, archive_filter_flags)
+        + (('--remote-path', remote_path) if remote_path else ())
+        + (('--log-json',) if global_arguments.log_json else ())
+        + (('--lock-wait', str(lock_wait)) if lock_wait else ())
+        + verbosity_flags
+        + (('--progress',) if check_arguments.progress else ())
+        + (tuple(extra_borg_options.split(' ')) if extra_borg_options else ())
+        + flags.make_repository_flags(repository_path, local_borg_version)
+    )
+
+    borg_environment = environment.make_environment(config)
+    borg_exit_codes = config.get('borg_exit_codes')
+
+    # The Borg repair option triggers an interactive prompt, which won't work when output is
+    # captured. And progress messes with the terminal directly.
+    if check_arguments.repair or check_arguments.progress:
+        execute_command(
+            full_command,
+            output_file=DO_NOT_CAPTURE,
+            extra_environment=borg_environment,
+            borg_local_path=local_path,
+            borg_exit_codes=borg_exit_codes,
+        )
+    else:
+        execute_command(
+            full_command,
+            extra_environment=borg_environment,
+            borg_local_path=local_path,
+            borg_exit_codes=borg_exit_codes,
         )
-
-        # The check command spews to stdout/stderr even without the verbose flag. Suppress it.
-        stdout = None if verbosity_flags else open(os.devnull, 'w')
-
-        logger.debug(' '.join(full_command))
-        subprocess.check_call(full_command, stdout=stdout, stderr=subprocess.STDOUT)
-
-    if 'extract' in checks:
-        extract.extract_last_archive_dry_run(repository, lock_wait, local_path, remote_path)

borgmatic/borg/compact.py

@@ -0,0 +1,54 @@
+import logging
+
+from borgmatic.borg import environment, flags
+from borgmatic.execute import execute_command
+
+logger = logging.getLogger(__name__)
+
+
+def compact_segments(
+    dry_run,
+    repository_path,
+    config,
+    local_borg_version,
+    global_arguments,
+    local_path='borg',
+    remote_path=None,
+    progress=False,
+    cleanup_commits=False,
+    threshold=None,
+):
+    '''
+    Given dry-run flag, a local or remote repository path, a configuration dict, and the local Borg
+    version, compact the segments in a repository.
+    '''
+    umask = config.get('umask', None)
+    lock_wait = config.get('lock_wait', None)
+    extra_borg_options = config.get('extra_borg_options', {}).get('compact', '')
+
+    full_command = (
+        (local_path, 'compact')
+        + (('--remote-path', remote_path) if remote_path else ())
+        + (('--umask', str(umask)) if umask else ())
+        + (('--log-json',) if global_arguments.log_json else ())
+        + (('--lock-wait', str(lock_wait)) if lock_wait else ())
+        + (('--progress',) if progress else ())
+        + (('--cleanup-commits',) if cleanup_commits else ())
+        + (('--threshold', str(threshold)) if threshold else ())
+        + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
+        + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
+        + (tuple(extra_borg_options.split(' ')) if extra_borg_options else ())
+        + flags.make_repository_flags(repository_path, local_borg_version)
+    )
+
+    if dry_run:
+        logging.info(f'{repository_path}: Skipping compact (dry run)')
+        return
+
+    execute_command(
+        full_command,
+        output_log_level=logging.INFO,
+        extra_environment=environment.make_environment(config),
+        borg_local_path=local_path,
+        borg_exit_codes=config.get('borg_exit_codes'),
+    )

borgmatic/borg/create.py

@@ -2,14 +2,23 @@ import glob
 import itertools
 import logging
 import os
-import subprocess
+import pathlib
+import stat
 import tempfile
 
+import borgmatic.logger
+from borgmatic.borg import environment, feature, flags, state
+from borgmatic.execute import (
+    DO_NOT_CAPTURE,
+    execute_command,
+    execute_command_and_capture_output,
+    execute_command_with_processes,
+)
 
 logger = logging.getLogger(__name__)
 
 
-def _expand_directory(directory):
+def expand_directory(directory):
     '''
     Given a directory path, expand any tilde (representing a user's home directory) and any globs
     therein. Return a list of one or more resulting paths.
@@ -19,7 +28,7 @@ def _expand_directory(directory):
     return glob.glob(expanded_directory) or [expanded_directory]
 
 
-def _expand_directories(directories):
+def expand_directories(directories):
     '''
     Given a sequence of directory paths, expand tildes and globs in each one. Return all the
     resulting directories as a single flattened tuple.
@@ -28,31 +37,121 @@ def _expand_directories(directories):
         return ()
 
     return tuple(
-        itertools.chain.from_iterable(_expand_directory(directory) for directory in directories)
+        itertools.chain.from_iterable(expand_directory(directory) for directory in directories)
     )
 
 
-def _write_pattern_file(patterns=None):
+def expand_home_directories(directories):
     '''
-    Given a sequence of patterns, write them to a named temporary file and return it. Return None
-    if no patterns are provided.
+    Given a sequence of directory paths, expand tildes in each one. Do not perform any globbing.
+    Return the results as a tuple.
     '''
-    if not patterns:
+    if directories is None:
+        return ()
+
+    return tuple(os.path.expanduser(directory) for directory in directories)
+
+
+def map_directories_to_devices(directories):
+    '''
+    Given a sequence of directories, return a map from directory to an identifier for the device on
+    which that directory resides or None if the path doesn't exist.
+
+    This is handy for determining whether two different directories are on the same filesystem (have
+    the same device identifier).
+    '''
+    return {
+        directory: os.stat(directory).st_dev if os.path.exists(directory) else None
+        for directory in directories
+    }
+
+
+def deduplicate_directories(directory_devices, additional_directory_devices):
+    '''
+    Given a map from directory to the identifier for the device on which that directory resides,
+    return the directories as a sorted tuple with all duplicate child directories removed. For
+    instance, if paths is ('/foo', '/foo/bar'), return just: ('/foo',)
+
+    The one exception to this rule is if two paths are on different filesystems (devices). In that
+    case, they won't get de-duplicated in case they both need to be passed to Borg (e.g. the
+    location.one_file_system option is true).
+
+    The idea is that if Borg is given a parent directory, then it doesn't also need to be given
+    child directories, because it will naturally spider the contents of the parent directory. And
+    there are cases where Borg coming across the same file twice will result in duplicate reads and
+    even hangs, e.g. when a database hook is using a named pipe for streaming database dumps to
+    Borg.
+
+    If any additional directory devices are given, also deduplicate against them, but don't include
+    them in the returned directories.
+    '''
+    deduplicated = set()
+    directories = sorted(directory_devices.keys())
+    additional_directories = sorted(additional_directory_devices.keys())
+    all_devices = {**directory_devices, **additional_directory_devices}
+
+    for directory in directories:
+        deduplicated.add(directory)
+        parents = pathlib.PurePath(directory).parents
+
+        # If another directory in the given list (or the additional list) is a parent of current
+        # directory (even n levels up) and both are on the same filesystem, then the current
+        # directory is a duplicate.
+        for other_directory in directories + additional_directories:
+            for parent in parents:
+                if (
+                    pathlib.PurePath(other_directory) == parent
+                    and all_devices[directory] is not None
+                    and all_devices[other_directory] == all_devices[directory]
+                ):
+                    if directory in deduplicated:
+                        deduplicated.remove(directory)
+                    break
+
+    return tuple(sorted(deduplicated))
+
+
+def write_pattern_file(patterns=None, sources=None, pattern_file=None):
+    '''
+    Given a sequence of patterns and an optional sequence of source directories, write them to a
+    named temporary file (with the source directories as additional roots) and return the file.
+    If an optional open pattern file is given, overwrite it instead of making a new temporary file.
+    Return None if no patterns are provided.
+    '''
+    if not patterns and not sources:
         return None
 
-    pattern_file = tempfile.NamedTemporaryFile('w')
-    pattern_file.write('\n'.join(patterns))
+    if pattern_file is None:
+        pattern_file = tempfile.NamedTemporaryFile('w')
+    else:
+        pattern_file.seek(0)
+
+    pattern_file.write(
+        '\n'.join(tuple(patterns or ()) + tuple(f'R {source}' for source in (sources or [])))
+    )
     pattern_file.flush()
 
     return pattern_file
 
 
-def _make_pattern_flags(location_config, pattern_filename=None):
+def ensure_files_readable(*filename_lists):
     '''
-    Given a location config dict with a potential pattern_from option, and a filename containing any
+    Given a sequence of filename sequences, ensure that each filename is openable. This prevents
+    unreadable files from being passed to Borg, which in certain situations only warns instead of
+    erroring.
+    '''
+    for file_object in itertools.chain.from_iterable(
+        filename_list for filename_list in filename_lists if filename_list
+    ):
+        open(file_object).close()
+
+
+def make_pattern_flags(config, pattern_filename=None):
+    '''
+    Given a configuration dict with a potential patterns_from option, and a filename containing any
     additional patterns, return the corresponding Borg flags for those files as a tuple.
     '''
-    pattern_filenames = tuple(location_config.get('patterns_from') or ()) + (
+    pattern_filenames = tuple(config.get('patterns_from') or ()) + (
         (pattern_filename,) if pattern_filename else ()
     )
 
@@ -63,12 +162,12 @@ def _make_pattern_flags(location_config, pattern_filename=None):
     )
 
 
-def _make_exclude_flags(location_config, exclude_filename=None):
+def make_exclude_flags(config, exclude_filename=None):
     '''
-    Given a location config dict with various exclude options, and a filename containing any exclude
+    Given a configuration dict with various exclude options, and a filename containing any exclude
     patterns, return the corresponding Borg flags as a tuple.
     '''
-    exclude_filenames = tuple(location_config.get('exclude_from') or ()) + (
+    exclude_filenames = tuple(config.get('exclude_from') or ()) + (
         (exclude_filename,) if exclude_filename else ()
     )
     exclude_from_flags = tuple(
@@ -76,72 +175,405 @@ def _make_exclude_flags(location_config, exclude_filename=None):
             ('--exclude-from', exclude_filename) for exclude_filename in exclude_filenames
         )
     )
-    caches_flag = ('--exclude-caches',) if location_config.get('exclude_caches') else ()
-    if_present = location_config.get('exclude_if_present')
-    if_present_flags = ('--exclude-if-present', if_present) if if_present else ()
+    caches_flag = ('--exclude-caches',) if config.get('exclude_caches') else ()
+    if_present_flags = tuple(
+        itertools.chain.from_iterable(
+            ('--exclude-if-present', if_present)
+            for if_present in config.get('exclude_if_present', ())
+        )
+    )
+    keep_exclude_tags_flags = ('--keep-exclude-tags',) if config.get('keep_exclude_tags') else ()
+    exclude_nodump_flags = ('--exclude-nodump',) if config.get('exclude_nodump') else ()
 
-    return exclude_from_flags + caches_flag + if_present_flags
+    return (
+        exclude_from_flags
+        + caches_flag
+        + if_present_flags
+        + keep_exclude_tags_flags
+        + exclude_nodump_flags
+    )
+
+
+def make_list_filter_flags(local_borg_version, dry_run):
+    '''
+    Given the local Borg version and whether this is a dry run, return the corresponding flags for
+    passing to "--list --filter". The general idea is that excludes are shown for a dry run or when
+    the verbosity is debug.
+    '''
+    base_flags = 'AME'
+    show_excludes = logger.isEnabledFor(logging.DEBUG)
+
+    if feature.available(feature.Feature.EXCLUDED_FILES_MINUS, local_borg_version):
+        if show_excludes or dry_run:
+            return f'{base_flags}+-'
+        else:
+            return base_flags
+
+    if show_excludes:
+        return f'{base_flags}x-'
+    else:
+        return f'{base_flags}-'
+
+
+def collect_borgmatic_source_directories(borgmatic_source_directory):
+    '''
+    Return a list of borgmatic-specific source directories used for state like database backups.
+    '''
+    if not borgmatic_source_directory:
+        borgmatic_source_directory = state.DEFAULT_BORGMATIC_SOURCE_DIRECTORY
+
+    return (
+        [borgmatic_source_directory]
+        if os.path.exists(os.path.expanduser(borgmatic_source_directory))
+        else []
+    )
+
+
+ROOT_PATTERN_PREFIX = 'R '
+
+
+def pattern_root_directories(patterns=None):
+    '''
+    Given a sequence of patterns, parse out and return just the root directories.
+    '''
+    if not patterns:
+        return []
+
+    return [
+        pattern.split(ROOT_PATTERN_PREFIX, maxsplit=1)[1]
+        for pattern in patterns
+        if pattern.startswith(ROOT_PATTERN_PREFIX)
+    ]
+
+
+def special_file(path):
+    '''
+    Return whether the given path is a special file (character device, block device, or named pipe
+    / FIFO).
+    '''
+    try:
+        mode = os.stat(path).st_mode
+    except (FileNotFoundError, OSError):
+        return False
+
+    return stat.S_ISCHR(mode) or stat.S_ISBLK(mode) or stat.S_ISFIFO(mode)
+
+
+def any_parent_directories(path, candidate_parents):
+    '''
+    Return whether any of the given candidate parent directories are an actual parent of the given
+    path. This includes grandparents, etc.
+    '''
+    for parent in candidate_parents:
+        if pathlib.PurePosixPath(parent) in pathlib.PurePath(path).parents:
+            return True
+
+    return False
+
+
+def collect_special_file_paths(
+    create_command, config, local_path, working_directory, borg_environment, skip_directories
+):
+    '''
+    Given a Borg create command as a tuple, a configuration dict, a local Borg path, a working
+    directory, a dict of environment variables to pass to Borg, and a sequence of parent directories
+    to skip, collect the paths for any special files (character devices, block devices, and named
+    pipes / FIFOs) that Borg would encounter during a create. These are all paths that could cause
+    Borg to hang if its --read-special flag is used.
+    '''
+    # Omit "--exclude-nodump" from the Borg dry run command, because that flag causes Borg to open
+    # files including any named pipe we've created.
+    paths_output = execute_command_and_capture_output(
+        tuple(argument for argument in create_command if argument != '--exclude-nodump')
+        + ('--dry-run', '--list'),
+        capture_stderr=True,
+        working_directory=working_directory,
+        extra_environment=borg_environment,
+        borg_local_path=local_path,
+        borg_exit_codes=config.get('borg_exit_codes'),
+    )
+
+    paths = tuple(
+        path_line.split(' ', 1)[1]
+        for path_line in paths_output.split('\n')
+        if path_line and path_line.startswith('- ') or path_line.startswith('+ ')
+    )
+
+    return tuple(
+        path
+        for path in paths
+        if special_file(path) and not any_parent_directories(path, skip_directories)
+    )
+
+
+def check_all_source_directories_exist(source_directories):
+    '''
+    Given a sequence of source directories, check that they all exist. If any do not, raise an
+    exception.
+    '''
+    missing_directories = [
+        source_directory
+        for source_directory in source_directories
+        if not all([os.path.exists(directory) for directory in expand_directory(source_directory)])
+    ]
+    if missing_directories:
+        raise ValueError(f"Source directories do not exist: {', '.join(missing_directories)}")
+
+
+def make_base_create_command(
+    dry_run,
+    repository_path,
+    config,
+    config_paths,
+    local_borg_version,
+    global_arguments,
+    borgmatic_source_directories,
+    local_path='borg',
+    remote_path=None,
+    progress=False,
+    json=False,
+    list_files=False,
+    stream_processes=None,
+):
+    '''
+    Given vebosity/dry-run flags, a local or remote repository path, a configuration dict, a
+    sequence of loaded configuration paths, the local Borg version, global arguments as an
+    argparse.Namespace instance, and a sequence of borgmatic source directories, return a tuple of
+    (base Borg create command flags, Borg create command positional arguments, open pattern file
+    handle, open exclude file handle).
+    '''
+    if config.get('source_directories_must_exist', False):
+        check_all_source_directories_exist(config.get('source_directories'))
+
+    sources = deduplicate_directories(
+        map_directories_to_devices(
+            expand_directories(
+                tuple(config.get('source_directories', ()))
+                + borgmatic_source_directories
+                + tuple(config_paths if config.get('store_config_files', True) else ())
+            )
+        ),
+        additional_directory_devices=map_directories_to_devices(
+            expand_directories(pattern_root_directories(config.get('patterns')))
+        ),
+    )
+
+    ensure_files_readable(config.get('patterns_from'), config.get('exclude_from'))
+
+    pattern_file = (
+        write_pattern_file(config.get('patterns'), sources)
+        if config.get('patterns') or config.get('patterns_from')
+        else None
+    )
+    exclude_file = write_pattern_file(expand_home_directories(config.get('exclude_patterns')))
+    checkpoint_interval = config.get('checkpoint_interval', None)
+    checkpoint_volume = config.get('checkpoint_volume', None)
+    chunker_params = config.get('chunker_params', None)
+    compression = config.get('compression', None)
+    upload_rate_limit = config.get('upload_rate_limit', None)
+    umask = config.get('umask', None)
+    lock_wait = config.get('lock_wait', None)
+    list_filter_flags = make_list_filter_flags(local_borg_version, dry_run)
+    files_cache = config.get('files_cache')
+    archive_name_format = config.get('archive_name_format', flags.DEFAULT_ARCHIVE_NAME_FORMAT)
+    extra_borg_options = config.get('extra_borg_options', {}).get('create', '')
+
+    if feature.available(feature.Feature.ATIME, local_borg_version):
+        atime_flags = ('--atime',) if config.get('atime') is True else ()
+    else:
+        atime_flags = ('--noatime',) if config.get('atime') is False else ()
+
+    if feature.available(feature.Feature.NOFLAGS, local_borg_version):
+        noflags_flags = ('--noflags',) if config.get('flags') is False else ()
+    else:
+        noflags_flags = ('--nobsdflags',) if config.get('flags') is False else ()
+
+    if feature.available(feature.Feature.NUMERIC_IDS, local_borg_version):
+        numeric_ids_flags = ('--numeric-ids',) if config.get('numeric_ids') else ()
+    else:
+        numeric_ids_flags = ('--numeric-owner',) if config.get('numeric_ids') else ()
+
+    if feature.available(feature.Feature.UPLOAD_RATELIMIT, local_borg_version):
+        upload_ratelimit_flags = (
+            ('--upload-ratelimit', str(upload_rate_limit)) if upload_rate_limit else ()
+        )
+    else:
+        upload_ratelimit_flags = (
+            ('--remote-ratelimit', str(upload_rate_limit)) if upload_rate_limit else ()
+        )
+
+    create_flags = (
+        tuple(local_path.split(' '))
+        + ('create',)
+        + make_pattern_flags(config, pattern_file.name if pattern_file else None)
+        + make_exclude_flags(config, exclude_file.name if exclude_file else None)
+        + (('--checkpoint-interval', str(checkpoint_interval)) if checkpoint_interval else ())
+        + (('--checkpoint-volume', str(checkpoint_volume)) if checkpoint_volume else ())
+        + (('--chunker-params', chunker_params) if chunker_params else ())
+        + (('--compression', compression) if compression else ())
+        + upload_ratelimit_flags
+        + (('--one-file-system',) if config.get('one_file_system') or stream_processes else ())
+        + numeric_ids_flags
+        + atime_flags
+        + (('--noctime',) if config.get('ctime') is False else ())
+        + (('--nobirthtime',) if config.get('birthtime') is False else ())
+        + (('--read-special',) if config.get('read_special') or stream_processes else ())
+        + noflags_flags
+        + (('--files-cache', files_cache) if files_cache else ())
+        + (('--remote-path', remote_path) if remote_path else ())
+        + (('--umask', str(umask)) if umask else ())
+        + (('--log-json',) if global_arguments.log_json else ())
+        + (('--lock-wait', str(lock_wait)) if lock_wait else ())
+        + (
+            ('--list', '--filter', list_filter_flags)
+            if list_files and not json and not progress
+            else ()
+        )
+        + (('--dry-run',) if dry_run else ())
+        + (tuple(extra_borg_options.split(' ')) if extra_borg_options else ())
+    )
+
+    create_positional_arguments = flags.make_repository_archive_flags(
+        repository_path, archive_name_format, local_borg_version
+    ) + (sources if not pattern_file else ())
+
+    # If database hooks are enabled (as indicated by streaming processes), exclude files that might
+    # cause Borg to hang. But skip this if the user has explicitly set the "read_special" to True.
+    if stream_processes and not config.get('read_special'):
+        logger.warning(
+            f'{repository_path}: Ignoring configured "read_special" value of false, as true is needed for database hooks.'
+        )
+        try:
+            working_directory = os.path.expanduser(config.get('working_directory'))
+        except TypeError:
+            working_directory = None
+
+        borg_environment = environment.make_environment(config)
+
+        logger.debug(f'{repository_path}: Collecting special file paths')
+        special_file_paths = collect_special_file_paths(
+            create_flags + create_positional_arguments,
+            config,
+            local_path,
+            working_directory,
+            borg_environment,
+            skip_directories=borgmatic_source_directories,
+        )
+
+        if special_file_paths:
+            logger.warning(
+                f'{repository_path}: Excluding special files to prevent Borg from hanging: {", ".join(special_file_paths)}'
+            )
+            exclude_file = write_pattern_file(
+                expand_home_directories(
+                    tuple(config.get('exclude_patterns') or ()) + special_file_paths
+                ),
+                pattern_file=exclude_file,
+            )
+            create_flags += make_exclude_flags(config, exclude_file.name)
+
+    return (create_flags, create_positional_arguments, pattern_file, exclude_file)
 
 
 def create_archive(
     dry_run,
-    repository,
-    location_config,
-    storage_config,
+    repository_path,
+    config,
+    config_paths,
+    local_borg_version,
+    global_arguments,
     local_path='borg',
     remote_path=None,
     progress=False,
     stats=False,
     json=False,
+    list_files=False,
+    stream_processes=None,
 ):
     '''
-    Given vebosity/dry-run flags, a local or remote repository path, a location config dict, and a
-    storage config dict, create a Borg archive.
+    Given vebosity/dry-run flags, a local or remote repository path, a configuration dict, a
+    sequence of loaded configuration paths, the local Borg version, and global arguments as an
+    argparse.Namespace instance, create a Borg archive and return Borg's JSON output (if any).
+
+    If a sequence of stream processes is given (instances of subprocess.Popen), then execute the
+    create command while also triggering the given processes to produce output.
     '''
-    sources = _expand_directories(location_config['source_directories'])
+    borgmatic.logger.add_custom_log_levels()
+    borgmatic_source_directories = expand_directories(
+        collect_borgmatic_source_directories(config.get('borgmatic_source_directory'))
+    )
 
-    pattern_file = _write_pattern_file(location_config.get('patterns'))
-    exclude_file = _write_pattern_file(_expand_directories(location_config.get('exclude_patterns')))
-    checkpoint_interval = storage_config.get('checkpoint_interval', None)
-    chunker_params = storage_config.get('chunker_params', None)
-    compression = storage_config.get('compression', None)
-    remote_rate_limit = storage_config.get('remote_rate_limit', None)
-    umask = storage_config.get('umask', None)
-    lock_wait = storage_config.get('lock_wait', None)
-    files_cache = location_config.get('files_cache')
-    default_archive_name_format = '{hostname}-{now:%Y-%m-%dT%H:%M:%S.%f}'
-    archive_name_format = storage_config.get('archive_name_format', default_archive_name_format)
-
-    full_command = (
-        (
+    (create_flags, create_positional_arguments, pattern_file, exclude_file) = (
+        make_base_create_command(
+            dry_run,
+            repository_path,
+            config,
+            config_paths,
+            local_borg_version,
+            global_arguments,
+            borgmatic_source_directories,
             local_path,
-            'create',
-            '{repository}::{archive_name_format}'.format(
-                repository=repository, archive_name_format=archive_name_format
-            ),
+            remote_path,
+            progress,
+            json,
+            list_files,
+            stream_processes,
         )
-        + sources
-        + _make_pattern_flags(location_config, pattern_file.name if pattern_file else None)
-        + _make_exclude_flags(location_config, exclude_file.name if exclude_file else None)
-        + (('--checkpoint-interval', str(checkpoint_interval)) if checkpoint_interval else ())
-        + (('--chunker-params', chunker_params) if chunker_params else ())
-        + (('--compression', compression) if compression else ())
-        + (('--remote-ratelimit', str(remote_rate_limit)) if remote_rate_limit else ())
-        + (('--one-file-system',) if location_config.get('one_file_system') else ())
-        + (('--read-special',) if location_config.get('read_special') else ())
-        + (('--nobsdflags',) if location_config.get('bsd_flags') is False else ())
-        + (('--files-cache', files_cache) if files_cache else ())
-        + (('--remote-path', remote_path) if remote_path else ())
-        + (('--umask', str(umask)) if umask else ())
-        + (('--lock-wait', str(lock_wait)) if lock_wait else ())
-        + (('--list', '--filter', 'AME-') if logger.isEnabledFor(logging.INFO) else ())
-        + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
-        + (('--stats',) if not dry_run and (logger.isEnabledFor(logging.INFO) or stats) else ())
-        + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
-        + (('--dry-run',) if dry_run else ())
+    )
+
+    if json:
+        output_log_level = None
+    elif list_files or (stats and not dry_run):
+        output_log_level = logging.ANSWER
+    else:
+        output_log_level = logging.INFO
+
+    # The progress output isn't compatible with captured and logged output, as progress messes with
+    # the terminal directly.
+    output_file = DO_NOT_CAPTURE if progress else None
+
+    try:
+        working_directory = os.path.expanduser(config.get('working_directory'))
+    except TypeError:
+        working_directory = None
+
+    borg_environment = environment.make_environment(config)
+
+    create_flags += (
+        (('--info',) if logger.getEffectiveLevel() == logging.INFO and not json else ())
+        + (('--stats',) if stats and not json and not dry_run else ())
+        + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) and not json else ())
         + (('--progress',) if progress else ())
         + (('--json',) if json else ())
     )
+    borg_exit_codes = config.get('borg_exit_codes')
 
-    logger.debug(' '.join(full_command))
-    subprocess.check_call(full_command)
+    if stream_processes:
+        return execute_command_with_processes(
+            create_flags + create_positional_arguments,
+            stream_processes,
+            output_log_level,
+            output_file,
+            working_directory=working_directory,
+            extra_environment=borg_environment,
+            borg_local_path=local_path,
+            borg_exit_codes=borg_exit_codes,
+        )
+    elif output_log_level is None:
+        return execute_command_and_capture_output(
+            create_flags + create_positional_arguments,
+            working_directory=working_directory,
+            extra_environment=borg_environment,
+            borg_local_path=local_path,
+            borg_exit_codes=borg_exit_codes,
+        )
+    else:
+        execute_command(
+            create_flags + create_positional_arguments,
+            output_log_level,
+            output_file,
+            working_directory=working_directory,
+            extra_environment=borg_environment,
+            borg_local_path=local_path,
+            borg_exit_codes=borg_exit_codes,
+        )

borgmatic/borg/environment.py

@@ -1,15 +1,57 @@
-import os
+OPTION_TO_ENVIRONMENT_VARIABLE = {
+    'borg_base_directory': 'BORG_BASE_DIR',
+    'borg_config_directory': 'BORG_CONFIG_DIR',
+    'borg_cache_directory': 'BORG_CACHE_DIR',
+    'borg_files_cache_ttl': 'BORG_FILES_CACHE_TTL',
+    'borg_security_directory': 'BORG_SECURITY_DIR',
+    'borg_keys_directory': 'BORG_KEYS_DIR',
+    'encryption_passcommand': 'BORG_PASSCOMMAND',
+    'encryption_passphrase': 'BORG_PASSPHRASE',
+    'ssh_command': 'BORG_RSH',
+    'temporary_directory': 'TMPDIR',
+}
+
+DEFAULT_BOOL_OPTION_TO_DOWNCASE_ENVIRONMENT_VARIABLE = {
+    'relocated_repo_access_is_ok': 'BORG_RELOCATED_REPO_ACCESS_IS_OK',
+    'unknown_unencrypted_repo_access_is_ok': 'BORG_UNKNOWN_UNENCRYPTED_REPO_ACCESS_IS_OK',
+}
+
+DEFAULT_BOOL_OPTION_TO_UPPERCASE_ENVIRONMENT_VARIABLE = {
+    'check_i_know_what_i_am_doing': 'BORG_CHECK_I_KNOW_WHAT_I_AM_DOING',
+}
 
 
-def initialize(storage_config):
-    passcommand = storage_config.get('encryption_passcommand')
-    if passcommand:
-        os.environ['BORG_PASSCOMMAND'] = passcommand
+def make_environment(config):
+    '''
+    Given a borgmatic configuration dict, return its options converted to a Borg environment
+    variable dict.
+    '''
+    environment = {}
 
-    passphrase = storage_config.get('encryption_passphrase')
-    if passphrase:
-        os.environ['BORG_PASSPHRASE'] = passphrase
+    for option_name, environment_variable_name in OPTION_TO_ENVIRONMENT_VARIABLE.items():
+        value = config.get(option_name)
 
-    ssh_command = storage_config.get('ssh_command')
-    if ssh_command:
-        os.environ['BORG_RSH'] = ssh_command
+        if value:
+            environment[environment_variable_name] = str(value)
+
+    for (
+        option_name,
+        environment_variable_name,
+    ) in DEFAULT_BOOL_OPTION_TO_DOWNCASE_ENVIRONMENT_VARIABLE.items():
+        value = config.get(option_name)
+        if value is not None:
+            environment[environment_variable_name] = 'yes' if value else 'no'
+
+    for (
+        option_name,
+        environment_variable_name,
+    ) in DEFAULT_BOOL_OPTION_TO_UPPERCASE_ENVIRONMENT_VARIABLE.items():
+        value = config.get(option_name)
+        if value is not None:
+            environment[environment_variable_name] = 'YES' if value else 'NO'
+
+    # On Borg 1.4.0a1+, take advantage of more specific exit codes. No effect on
+    # older versions of Borg.
+    environment['BORG_EXIT_CODES'] = 'modern'
+
+    return environment

borgmatic/borg/export_key.py

@@ -0,0 +1,71 @@
+import logging
+import os
+
+import borgmatic.logger
+from borgmatic.borg import environment, flags
+from borgmatic.execute import DO_NOT_CAPTURE, execute_command
+
+logger = logging.getLogger(__name__)
+
+
+def export_key(
+    repository_path,
+    config,
+    local_borg_version,
+    export_arguments,
+    global_arguments,
+    local_path='borg',
+    remote_path=None,
+):
+    '''
+    Given a local or remote repository path, a configuration dict, the local Borg version, and
+    optional local and remote Borg paths, export the repository key to the destination path
+    indicated in the export arguments.
+
+    If the destination path is empty or "-", then print the key to stdout instead of to a file.
+
+    Raise FileExistsError if a path is given but it already exists on disk.
+    '''
+    borgmatic.logger.add_custom_log_levels()
+    umask = config.get('umask', None)
+    lock_wait = config.get('lock_wait', None)
+
+    if export_arguments.path and export_arguments.path != '-':
+        if os.path.exists(export_arguments.path):
+            raise FileExistsError(
+                f'Destination path {export_arguments.path} already exists. Aborting.'
+            )
+
+        output_file = None
+    else:
+        output_file = DO_NOT_CAPTURE
+
+    full_command = (
+        (local_path, 'key', 'export')
+        + (('--remote-path', remote_path) if remote_path else ())
+        + (('--umask', str(umask)) if umask else ())
+        + (('--log-json',) if global_arguments.log_json else ())
+        + (('--lock-wait', str(lock_wait)) if lock_wait else ())
+        + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
+        + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
+        + flags.make_flags('paper', export_arguments.paper)
+        + flags.make_flags('qr-html', export_arguments.qr_html)
+        + flags.make_repository_flags(
+            repository_path,
+            local_borg_version,
+        )
+        + ((export_arguments.path,) if output_file is None else ())
+    )
+
+    if global_arguments.dry_run:
+        logging.info(f'{repository_path}: Skipping key export (dry run)')
+        return
+
+    execute_command(
+        full_command,
+        output_file=output_file,
+        output_log_level=logging.ANSWER,
+        extra_environment=environment.make_environment(config),
+        borg_local_path=local_path,
+        borg_exit_codes=config.get('borg_exit_codes'),
+    )

borgmatic/borg/export_tar.py

@@ -0,0 +1,75 @@
+import logging
+
+import borgmatic.logger
+from borgmatic.borg import environment, flags
+from borgmatic.execute import DO_NOT_CAPTURE, execute_command
+
+logger = logging.getLogger(__name__)
+
+
+def export_tar_archive(
+    dry_run,
+    repository_path,
+    archive,
+    paths,
+    destination_path,
+    config,
+    local_borg_version,
+    global_arguments,
+    local_path='borg',
+    remote_path=None,
+    tar_filter=None,
+    list_files=False,
+    strip_components=None,
+):
+    '''
+    Given a dry-run flag, a local or remote repository path, an archive name, zero or more paths to
+    export from the archive, a destination path to export to, a configuration dict, the local Borg
+    version, optional local and remote Borg paths, an optional filter program, whether to include
+    per-file details, and an optional number of path components to strip, export the archive into
+    the given destination path as a tar-formatted file.
+
+    If the destination path is "-", then stream the output to stdout instead of to a file.
+    '''
+    borgmatic.logger.add_custom_log_levels()
+    umask = config.get('umask', None)
+    lock_wait = config.get('lock_wait', None)
+
+    full_command = (
+        (local_path, 'export-tar')
+        + (('--remote-path', remote_path) if remote_path else ())
+        + (('--umask', str(umask)) if umask else ())
+        + (('--log-json',) if global_arguments.log_json else ())
+        + (('--lock-wait', str(lock_wait)) if lock_wait else ())
+        + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
+        + (('--list',) if list_files else ())
+        + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
+        + (('--dry-run',) if dry_run else ())
+        + (('--tar-filter', tar_filter) if tar_filter else ())
+        + (('--strip-components', str(strip_components)) if strip_components else ())
+        + flags.make_repository_archive_flags(
+            repository_path,
+            archive,
+            local_borg_version,
+        )
+        + (destination_path,)
+        + (tuple(paths) if paths else ())
+    )
+
+    if list_files:
+        output_log_level = logging.ANSWER
+    else:
+        output_log_level = logging.INFO
+
+    if dry_run:
+        logging.info(f'{repository_path}: Skipping export to tar file (dry run)')
+        return
+
+    execute_command(
+        full_command,
+        output_file=DO_NOT_CAPTURE if destination_path == '-' else None,
+        output_log_level=output_log_level,
+        extra_environment=environment.make_environment(config),
+        borg_local_path=local_path,
+        borg_exit_codes=config.get('borg_exit_codes'),
+    )

borgmatic/borg/extract.py

@@ -1,52 +1,173 @@
 import logging
-import sys
+import os
 import subprocess
 
+import borgmatic.config.validate
+from borgmatic.borg import environment, feature, flags, rlist
+from borgmatic.execute import DO_NOT_CAPTURE, execute_command
 
 logger = logging.getLogger(__name__)
 
 
-def extract_last_archive_dry_run(repository, lock_wait=None, local_path='borg', remote_path=None):
+def extract_last_archive_dry_run(
+    config,
+    local_borg_version,
+    global_arguments,
+    repository_path,
+    lock_wait=None,
+    local_path='borg',
+    remote_path=None,
+):
     '''
-    Perform an extraction dry-run of just the most recent archive. If there are no archives, skip
-    the dry-run.
+    Perform an extraction dry-run of the most recent archive. If there are no archives, skip the
+    dry-run.
     '''
-    remote_path_flags = ('--remote-path', remote_path) if remote_path else ()
-    lock_wait_flags = ('--lock-wait', str(lock_wait)) if lock_wait else ()
     verbosity_flags = ()
     if logger.isEnabledFor(logging.DEBUG):
         verbosity_flags = ('--debug', '--show-rc')
     elif logger.isEnabledFor(logging.INFO):
         verbosity_flags = ('--info',)
 
-    full_list_command = (
-        (local_path, 'list', '--short', repository)
-        + remote_path_flags
-        + lock_wait_flags
-        + verbosity_flags
-    )
-
-    list_output = subprocess.check_output(full_list_command).decode(sys.stdout.encoding)
-
-    last_archive_name = list_output.strip().split('\n')[-1]
-    if not last_archive_name:
+    try:
+        last_archive_name = rlist.resolve_archive_name(
+            repository_path,
+            'latest',
+            config,
+            local_borg_version,
+            global_arguments,
+            local_path,
+            remote_path,
+        )
+    except ValueError:
+        logger.warning('No archives found. Skipping extract consistency check.')
         return
 
     list_flag = ('--list',) if logger.isEnabledFor(logging.DEBUG) else ()
+    borg_environment = environment.make_environment(config)
     full_extract_command = (
-        (
-            local_path,
-            'extract',
-            '--dry-run',
-            '{repository}::{last_archive_name}'.format(
-                repository=repository, last_archive_name=last_archive_name
-            ),
-        )
-        + remote_path_flags
-        + lock_wait_flags
+        (local_path, 'extract', '--dry-run')
+        + (('--remote-path', remote_path) if remote_path else ())
+        + (('--log-json',) if global_arguments.log_json else ())
+        + (('--lock-wait', str(lock_wait)) if lock_wait else ())
         + verbosity_flags
         + list_flag
+        + flags.make_repository_archive_flags(
+            repository_path, last_archive_name, local_borg_version
+        )
     )
 
-    logger.debug(' '.join(full_extract_command))
-    subprocess.check_call(full_extract_command)
+    execute_command(
+        full_extract_command,
+        working_directory=None,
+        extra_environment=borg_environment,
+        borg_local_path=local_path,
+        borg_exit_codes=config.get('borg_exit_codes'),
+    )
+
+
+def extract_archive(
+    dry_run,
+    repository,
+    archive,
+    paths,
+    config,
+    local_borg_version,
+    global_arguments,
+    local_path='borg',
+    remote_path=None,
+    destination_path=None,
+    strip_components=None,
+    progress=False,
+    extract_to_stdout=False,
+):
+    '''
+    Given a dry-run flag, a local or remote repository path, an archive name, zero or more paths to
+    restore from the archive, the local Borg version string, an argparse.Namespace of global
+    arguments, a configuration dict, optional local and remote Borg paths, and an optional
+    destination path to extract to, extract the archive into the current directory.
+
+    If extract to stdout is True, then start the extraction streaming to stdout, and return that
+    extract process as an instance of subprocess.Popen.
+    '''
+    umask = config.get('umask', None)
+    lock_wait = config.get('lock_wait', None)
+
+    if progress and extract_to_stdout:
+        raise ValueError('progress and extract_to_stdout cannot both be set')
+
+    if feature.available(feature.Feature.NUMERIC_IDS, local_borg_version):
+        numeric_ids_flags = ('--numeric-ids',) if config.get('numeric_ids') else ()
+    else:
+        numeric_ids_flags = ('--numeric-owner',) if config.get('numeric_ids') else ()
+
+    if strip_components == 'all':
+        if not paths:
+            raise ValueError('The --strip-components flag with "all" requires at least one --path')
+
+        # Calculate the maximum number of leading path components of the given paths. "if piece"
+        # ignores empty path components, e.g. those resulting from a leading slash. And the "- 1"
+        # is so this doesn't count the final path component, e.g. the filename itself.
+        strip_components = max(
+            0,
+            *(len(tuple(piece for piece in path.split(os.path.sep) if piece)) - 1 for path in paths)
+        )
+
+    full_command = (
+        (local_path, 'extract')
+        + (('--remote-path', remote_path) if remote_path else ())
+        + numeric_ids_flags
+        + (('--umask', str(umask)) if umask else ())
+        + (('--log-json',) if global_arguments.log_json else ())
+        + (('--lock-wait', str(lock_wait)) if lock_wait else ())
+        + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
+        + (('--debug', '--list', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
+        + (('--dry-run',) if dry_run else ())
+        + (('--strip-components', str(strip_components)) if strip_components else ())
+        + (('--progress',) if progress else ())
+        + (('--stdout',) if extract_to_stdout else ())
+        + flags.make_repository_archive_flags(
+            # Make the repository path absolute so the working directory changes below don't
+            # prevent Borg from finding the repo.
+            borgmatic.config.validate.normalize_repository_path(repository),
+            archive,
+            local_borg_version,
+        )
+        + (tuple(paths) if paths else ())
+    )
+
+    borg_environment = environment.make_environment(config)
+    borg_exit_codes = config.get('borg_exit_codes')
+
+    # The progress output isn't compatible with captured and logged output, as progress messes with
+    # the terminal directly.
+    if progress:
+        return execute_command(
+            full_command,
+            output_file=DO_NOT_CAPTURE,
+            working_directory=destination_path,
+            extra_environment=borg_environment,
+            borg_local_path=local_path,
+            borg_exit_codes=borg_exit_codes,
+        )
+        return None
+
+    if extract_to_stdout:
+        return execute_command(
+            full_command,
+            output_file=subprocess.PIPE,
+            working_directory=destination_path,
+            run_to_completion=False,
+            extra_environment=borg_environment,
+            borg_local_path=local_path,
+            borg_exit_codes=borg_exit_codes,
+        )
+
+    # Don't give Borg local path so as to error on warnings, as "borg extract" only gives a warning
+    # if the restore paths don't exist in the archive.
+    execute_command(
+        full_command,
+        working_directory=destination_path,
+        extra_environment=borg_environment,
+        borg_local_path=local_path,
+        borg_exit_codes=borg_exit_codes,
+    )

borgmatic/borg/feature.py

@@ -0,0 +1,40 @@
+from enum import Enum
+
+from packaging.version import parse
+
+
+class Feature(Enum):
+    COMPACT = 1
+    ATIME = 2
+    NOFLAGS = 3
+    NUMERIC_IDS = 4
+    UPLOAD_RATELIMIT = 5
+    SEPARATE_REPOSITORY_ARCHIVE = 6
+    RCREATE = 7
+    RLIST = 8
+    RINFO = 9
+    MATCH_ARCHIVES = 10
+    EXCLUDED_FILES_MINUS = 11
+
+
+FEATURE_TO_MINIMUM_BORG_VERSION = {
+    Feature.COMPACT: parse('1.2.0a2'),  # borg compact
+    Feature.ATIME: parse('1.2.0a7'),  # borg create --atime
+    Feature.NOFLAGS: parse('1.2.0a8'),  # borg create --noflags
+    Feature.NUMERIC_IDS: parse('1.2.0b3'),  # borg create/extract/mount --numeric-ids
+    Feature.UPLOAD_RATELIMIT: parse('1.2.0b3'),  # borg create --upload-ratelimit
+    Feature.SEPARATE_REPOSITORY_ARCHIVE: parse('2.0.0a2'),  # --repo with separate archive
+    Feature.RCREATE: parse('2.0.0a2'),  # borg rcreate
+    Feature.RLIST: parse('2.0.0a2'),  # borg rlist
+    Feature.RINFO: parse('2.0.0a2'),  # borg rinfo
+    Feature.MATCH_ARCHIVES: parse('2.0.0b3'),  # borg --match-archives
+    Feature.EXCLUDED_FILES_MINUS: parse('2.0.0b5'),  # --list --filter uses "-" for excludes
+}
+
+
+def available(feature, borg_version):
+    '''
+    Given a Borg Feature constant and a Borg version string, return whether that feature is
+    available in that version of Borg.
+    '''
+    return FEATURE_TO_MINIMUM_BORG_VERSION[feature] <= parse(borg_version)

borgmatic/borg/flags.py

@@ -0,0 +1,118 @@
+import itertools
+import json
+import logging
+import re
+
+from borgmatic.borg import feature
+
+logger = logging.getLogger(__name__)
+
+
+def make_flags(name, value):
+    '''
+    Given a flag name and its value, return it formatted as Borg-compatible flags.
+    '''
+    if not value:
+        return ()
+
+    flag = f"--{name.replace('_', '-')}"
+
+    if value is True:
+        return (flag,)
+
+    return (flag, str(value))
+
+
+def make_flags_from_arguments(arguments, excludes=()):
+    '''
+    Given borgmatic command-line arguments as an instance of argparse.Namespace, and optionally a
+    list of named arguments to exclude, generate and return the corresponding Borg command-line
+    flags as a tuple.
+    '''
+    return tuple(
+        itertools.chain.from_iterable(
+            make_flags(name, value=getattr(arguments, name))
+            for name in sorted(vars(arguments))
+            if name not in excludes and not name.startswith('_')
+        )
+    )
+
+
+def make_repository_flags(repository_path, local_borg_version):
+    '''
+    Given the path of a Borg repository and the local Borg version, return Borg-version-appropriate
+    command-line flags (as a tuple) for selecting that repository.
+    '''
+    return (
+        ('--repo',)
+        if feature.available(feature.Feature.SEPARATE_REPOSITORY_ARCHIVE, local_borg_version)
+        else ()
+    ) + (repository_path,)
+
+
+def make_repository_archive_flags(repository_path, archive, local_borg_version):
+    '''
+    Given the path of a Borg repository, an archive name or pattern, and the local Borg version,
+    return Borg-version-appropriate command-line flags (as a tuple) for selecting that repository
+    and archive.
+    '''
+    return (
+        ('--repo', repository_path, archive)
+        if feature.available(feature.Feature.SEPARATE_REPOSITORY_ARCHIVE, local_borg_version)
+        else (f'{repository_path}::{archive}',)
+    )
+
+
+DEFAULT_ARCHIVE_NAME_FORMAT = '{hostname}-{now:%Y-%m-%dT%H:%M:%S.%f}'  # noqa: FS003
+
+
+def make_match_archives_flags(match_archives, archive_name_format, local_borg_version):
+    '''
+    Return match archives flags based on the given match archives value, if any. If it isn't set,
+    return match archives flags to match archives created with the given (or default) archive name
+    format. This is done by replacing certain archive name format placeholders for ephemeral data
+    (like "{now}") with globs.
+    '''
+    if match_archives:
+        if match_archives in {'*', 're:.*', 'sh:*'}:
+            return ()
+
+        if feature.available(feature.Feature.MATCH_ARCHIVES, local_borg_version):
+            return ('--match-archives', match_archives)
+        else:
+            return ('--glob-archives', re.sub(r'^sh:', '', match_archives))
+
+    derived_match_archives = re.sub(
+        r'\{(now|utcnow|pid)([:%\w\.-]*)\}', '*', archive_name_format or DEFAULT_ARCHIVE_NAME_FORMAT
+    )
+
+    if derived_match_archives == '*':
+        return ()
+
+    if feature.available(feature.Feature.MATCH_ARCHIVES, local_borg_version):
+        return ('--match-archives', f'sh:{derived_match_archives}')
+    else:
+        return ('--glob-archives', f'{derived_match_archives}')
+
+
+def warn_for_aggressive_archive_flags(json_command, json_output):
+    '''
+    Given a JSON archives command and the resulting JSON string output from running it, parse the
+    JSON and warn if the command used an archive flag but the output indicates zero archives were
+    found.
+    '''
+    archive_flags_used = {'--glob-archives', '--match-archives'}.intersection(set(json_command))
+
+    if not archive_flags_used:
+        return
+
+    try:
+        if len(json.loads(json_output)['archives']) == 0:
+            logger.warning('An archive filter was applied, but no matching archives were found.')
+            logger.warning(
+                'Try adding --match-archives "*" or adjusting archive_name_format/match_archives in configuration.'
+            )
+    except json.JSONDecodeError as error:
+        logger.debug(f'Cannot parse JSON output from archive command: {error}')
+    except (TypeError, KeyError):
+        logger.debug('Cannot parse JSON output from archive command: No "archives" key found')

borgmatic/borg/info.py

@@ -1,29 +1,118 @@
+import argparse
 import logging
-import subprocess
 
+import borgmatic.logger
+from borgmatic.borg import environment, feature, flags
+from borgmatic.execute import execute_command, execute_command_and_capture_output
 
 logger = logging.getLogger(__name__)
 
 
-def display_archives_info(
-    repository, storage_config, local_path='borg', remote_path=None, json=False
+def make_info_command(
+    repository_path,
+    config,
+    local_borg_version,
+    info_arguments,
+    global_arguments,
+    local_path,
+    remote_path,
 ):
     '''
-    Given a local or remote repository path, and a storage config dict,
-    display summary information for Borg archives in the repository.
+    Given a local or remote repository path, a configuration dict, the local Borg version, the
+    arguments to the info action as an argparse.Namespace, and global arguments, return a command
+    as a tuple to display summary information for archives in the repository.
     '''
-    lock_wait = storage_config.get('lock_wait', None)
-
-    full_command = (
-        (local_path, 'info', repository)
-        + (('--remote-path', remote_path) if remote_path else ())
-        + (('--lock-wait', str(lock_wait)) if lock_wait else ())
-        + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
-        + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
-        + (('--json',) if json else ())
+    return (
+        (local_path, 'info')
+        + (
+            ('--info',)
+            if logger.getEffectiveLevel() == logging.INFO and not info_arguments.json
+            else ()
+        )
+        + (
+            ('--debug', '--show-rc')
+            if logger.isEnabledFor(logging.DEBUG) and not info_arguments.json
+            else ()
+        )
+        + flags.make_flags('remote-path', remote_path)
+        + flags.make_flags('log-json', global_arguments.log_json)
+        + flags.make_flags('lock-wait', config.get('lock_wait'))
+        + (
+            (
+                flags.make_flags('match-archives', f'sh:{info_arguments.prefix}*')
+                if feature.available(feature.Feature.MATCH_ARCHIVES, local_borg_version)
+                else flags.make_flags('glob-archives', f'{info_arguments.prefix}*')
+            )
+            if info_arguments.prefix
+            else (
+                flags.make_match_archives_flags(
+                    info_arguments.match_archives
+                    or info_arguments.archive
+                    or config.get('match_archives'),
+                    config.get('archive_name_format'),
+                    local_borg_version,
+                )
+            )
+        )
+        + flags.make_flags_from_arguments(
+            info_arguments, excludes=('repository', 'archive', 'prefix', 'match_archives')
+        )
+        + flags.make_repository_flags(repository_path, local_borg_version)
     )
 
-    logger.debug(' '.join(full_command))
 
-    output = subprocess.check_output(full_command)
-    return output.decode() if output is not None else None
+def display_archives_info(
+    repository_path,
+    config,
+    local_borg_version,
+    info_arguments,
+    global_arguments,
+    local_path='borg',
+    remote_path=None,
+):
+    '''
+    Given a local or remote repository path, a configuration dict, the local Borg version, the
+    arguments to the info action as an argparse.Namespace, and global arguments, display summary
+    information for Borg archives in the repository or return JSON summary information.
+    '''
+    borgmatic.logger.add_custom_log_levels()
+
+    main_command = make_info_command(
+        repository_path,
+        config,
+        local_borg_version,
+        info_arguments,
+        global_arguments,
+        local_path,
+        remote_path,
+    )
+    json_command = make_info_command(
+        repository_path,
+        config,
+        local_borg_version,
+        argparse.Namespace(**dict(info_arguments.__dict__, json=True)),
+        global_arguments,
+        local_path,
+        remote_path,
+    )
+    borg_exit_codes = config.get('borg_exit_codes')
+
+    json_info = execute_command_and_capture_output(
+        json_command,
+        extra_environment=environment.make_environment(config),
+        borg_local_path=local_path,
+        borg_exit_codes=borg_exit_codes,
+    )
+
+    if info_arguments.json:
+        return json_info
+
+    flags.warn_for_aggressive_archive_flags(json_command, json_info)
+
+    execute_command(
+        main_command,
+        output_log_level=logging.ANSWER,
+        extra_environment=environment.make_environment(config),
+        borg_local_path=local_path,
+        borg_exit_codes=borg_exit_codes,
+    )

borgmatic/borg/init.py

@@ -1,39 +0,0 @@
-import logging
-import subprocess
-
-
-logger = logging.getLogger(__name__)
-
-
-def initialize_repository(
-    repository,
-    encryption_mode,
-    append_only=None,
-    storage_quota=None,
-    local_path='borg',
-    remote_path=None,
-):
-    '''
-    Given a local or remote repository path, a Borg encryption mode, whether the repository should
-    be append-only, and the storage quota to use, initialize the repository. If the repository
-    already exists, then log and skip initialization.
-    '''
-    info_command = (local_path, 'info', repository)
-    logger.debug(' '.join(info_command))
-
-    if subprocess.call(info_command, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL) == 0:
-        logger.info('Repository already exists. Skipping initialization.')
-        return
-
-    init_command = (
-        (local_path, 'init', repository)
-        + (('--encryption', encryption_mode) if encryption_mode else ())
-        + (('--append-only',) if append_only else ())
-        + (('--storage-quota', storage_quota) if storage_quota else ())
-        + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
-        + (('--debug',) if logger.isEnabledFor(logging.DEBUG) else ())
-        + (('--remote-path', remote_path) if remote_path else ())
-    )
-
-    logger.debug(' '.join(init_command))
-    subprocess.check_call(init_command)

borgmatic/borg/list.py

@@ -1,26 +1,263 @@
+import argparse
+import copy
 import logging
-import subprocess
+import re
 
+import borgmatic.logger
+from borgmatic.borg import environment, feature, flags, rlist
+from borgmatic.execute import execute_command, execute_command_and_capture_output
 
 logger = logging.getLogger(__name__)
 
 
-def list_archives(repository, storage_config, local_path='borg', remote_path=None, json=False):
-    '''
-    Given a local or remote repository path, and a storage config dict,
-    list Borg archives in the repository.
-    '''
-    lock_wait = storage_config.get('lock_wait', None)
+ARCHIVE_FILTER_FLAGS_MOVED_TO_RLIST = ('prefix', 'match_archives', 'sort_by', 'first', 'last')
+MAKE_FLAGS_EXCLUDES = (
+    'repository',
+    'archive',
+    'paths',
+    'find_paths',
+) + ARCHIVE_FILTER_FLAGS_MOVED_TO_RLIST
 
-    full_command = (
-        (local_path, 'list', repository)
-        + (('--remote-path', remote_path) if remote_path else ())
-        + (('--lock-wait', str(lock_wait)) if lock_wait else ())
-        + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
-        + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
-        + (('--json',) if json else ())
+
+def make_list_command(
+    repository_path,
+    config,
+    local_borg_version,
+    list_arguments,
+    global_arguments,
+    local_path='borg',
+    remote_path=None,
+):
+    '''
+    Given a local or remote repository path, a configuration dict, the arguments to the list action,
+    and local and remote Borg paths, return a command as a tuple to list archives or paths within an
+    archive.
+    '''
+    lock_wait = config.get('lock_wait', None)
+
+    return (
+        (local_path, 'list')
+        + (
+            ('--info',)
+            if logger.getEffectiveLevel() == logging.INFO and not list_arguments.json
+            else ()
+        )
+        + (
+            ('--debug', '--show-rc')
+            if logger.isEnabledFor(logging.DEBUG) and not list_arguments.json
+            else ()
+        )
+        + flags.make_flags('remote-path', remote_path)
+        + flags.make_flags('log-json', global_arguments.log_json)
+        + flags.make_flags('lock-wait', lock_wait)
+        + flags.make_flags_from_arguments(list_arguments, excludes=MAKE_FLAGS_EXCLUDES)
+        + (
+            flags.make_repository_archive_flags(
+                repository_path, list_arguments.archive, local_borg_version
+            )
+            if list_arguments.archive
+            else flags.make_repository_flags(repository_path, local_borg_version)
+        )
+        + (tuple(list_arguments.paths) if list_arguments.paths else ())
     )
-    logger.debug(' '.join(full_command))
 
-    output = subprocess.check_output(full_command)
-    return output.decode() if output is not None else None
+
+def make_find_paths(find_paths):
+    '''
+    Given a sequence of path fragments or patterns as passed to `--find`, transform all path
+    fragments into glob patterns. Pass through existing patterns untouched.
+
+    For example, given find_paths of:
+
+      ['foo.txt', 'pp:root/somedir']
+
+    ... transform that into:
+
+      ['sh:**/*foo.txt*/**', 'pp:root/somedir']
+    '''
+    if not find_paths:
+        return ()
+
+    return tuple(
+        (
+            find_path
+            if re.compile(r'([-!+RrPp] )|(\w\w:)').match(find_path)
+            else f'sh:**/*{find_path}*/**'
+        )
+        for find_path in find_paths
+    )
+
+
+def capture_archive_listing(
+    repository_path,
+    archive,
+    config,
+    local_borg_version,
+    global_arguments,
+    list_paths=None,
+    path_format=None,
+    local_path='borg',
+    remote_path=None,
+):
+    '''
+    Given a local or remote repository path, an archive name, a configuration dict, the local Borg
+    version, global arguments as an argparse.Namespace, the archive paths in which to list files,
+    the Borg path format to use for the output, and local and remote Borg paths, capture the output
+    of listing that archive and return it as a list of file paths.
+    '''
+    borg_environment = environment.make_environment(config)
+
+    return tuple(
+        execute_command_and_capture_output(
+            make_list_command(
+                repository_path,
+                config,
+                local_borg_version,
+                argparse.Namespace(
+                    repository=repository_path,
+                    archive=archive,
+                    paths=[f'sh:{path}' for path in list_paths] if list_paths else None,
+                    find_paths=None,
+                    json=None,
+                    format=path_format or '{path}{NL}',  # noqa: FS003
+                ),
+                global_arguments,
+                local_path,
+                remote_path,
+            ),
+            extra_environment=borg_environment,
+            borg_local_path=local_path,
+            borg_exit_codes=config.get('borg_exit_codes'),
+        )
+        .strip('\n')
+        .split('\n')
+    )
+
+
+def list_archive(
+    repository_path,
+    config,
+    local_borg_version,
+    list_arguments,
+    global_arguments,
+    local_path='borg',
+    remote_path=None,
+):
+    '''
+    Given a local or remote repository path, a configuration dict, the local Borg version, global
+    arguments as an argparse.Namespace, the arguments to the list action as an argparse.Namespace,
+    and local and remote Borg paths, display the output of listing the files of a Borg archive (or
+    return JSON output). If list_arguments.find_paths are given, list the files by searching across
+    multiple archives. If neither find_paths nor archive name are given, instead list the archives
+    in the given repository.
+    '''
+    borgmatic.logger.add_custom_log_levels()
+
+    if not list_arguments.archive and not list_arguments.find_paths:
+        if feature.available(feature.Feature.RLIST, local_borg_version):
+            logger.warning(
+                'Omitting the --archive flag on the list action is deprecated when using Borg 2.x+. Use the rlist action instead.'
+            )
+
+        rlist_arguments = argparse.Namespace(
+            repository=repository_path,
+            short=list_arguments.short,
+            format=list_arguments.format,
+            json=list_arguments.json,
+            prefix=list_arguments.prefix,
+            match_archives=list_arguments.match_archives,
+            sort_by=list_arguments.sort_by,
+            first=list_arguments.first,
+            last=list_arguments.last,
+        )
+        return rlist.list_repository(
+            repository_path,
+            config,
+            local_borg_version,
+            rlist_arguments,
+            global_arguments,
+            local_path,
+            remote_path,
+        )
+
+    if list_arguments.archive:
+        for name in ARCHIVE_FILTER_FLAGS_MOVED_TO_RLIST:
+            if getattr(list_arguments, name, None):
+                logger.warning(
+                    f"The --{name.replace('_', '-')} flag on the list action is ignored when using the --archive flag."
+                )
+
+    if list_arguments.json:
+        raise ValueError(
+            'The --json flag on the list action is not supported when using the --archive/--find flags.'
+        )
+
+    borg_environment = environment.make_environment(config)
+    borg_exit_codes = config.get('borg_exit_codes')
+
+    # If there are any paths to find (and there's not a single archive already selected), start by
+    # getting a list of archives to search.
+    if list_arguments.find_paths and not list_arguments.archive:
+        rlist_arguments = argparse.Namespace(
+            repository=repository_path,
+            short=True,
+            format=None,
+            json=None,
+            prefix=list_arguments.prefix,
+            match_archives=list_arguments.match_archives,
+            sort_by=list_arguments.sort_by,
+            first=list_arguments.first,
+            last=list_arguments.last,
+        )
+
+        # Ask Borg to list archives. Capture its output for use below.
+        archive_lines = tuple(
+            execute_command_and_capture_output(
+                rlist.make_rlist_command(
+                    repository_path,
+                    config,
+                    local_borg_version,
+                    rlist_arguments,
+                    global_arguments,
+                    local_path,
+                    remote_path,
+                ),
+                extra_environment=borg_environment,
+                borg_local_path=local_path,
+                borg_exit_codes=borg_exit_codes,
+            )
+            .strip('\n')
+            .split('\n')
+        )
+    else:
+        archive_lines = (list_arguments.archive,)
+
+    # For each archive listed by Borg, run list on the contents of that archive.
+    for archive in archive_lines:
+        logger.answer(f'{repository_path}: Listing archive {archive}')
+
+        archive_arguments = copy.copy(list_arguments)
+        archive_arguments.archive = archive
+
+        # This list call is to show the files in a single archive, not list multiple archives. So
+        # blank out any archive filtering flags. They'll break anyway in Borg 2.
+        for name in ARCHIVE_FILTER_FLAGS_MOVED_TO_RLIST:
+            setattr(archive_arguments, name, None)
+
+        main_command = make_list_command(
+            repository_path,
+            config,
+            local_borg_version,
+            archive_arguments,
+            global_arguments,
+            local_path,
+            remote_path,
+        ) + make_find_paths(list_arguments.find_paths)
+
+        execute_command(
+            main_command,
+            output_log_level=logging.ANSWER,
+            extra_environment=borg_environment,
+            borg_local_path=local_path,
+            borg_exit_codes=borg_exit_codes,
+        )

borgmatic/borg/mount.py

@@ -0,0 +1,79 @@
+import logging
+
+from borgmatic.borg import environment, feature, flags
+from borgmatic.execute import DO_NOT_CAPTURE, execute_command
+
+logger = logging.getLogger(__name__)
+
+
+def mount_archive(
+    repository_path,
+    archive,
+    mount_arguments,
+    config,
+    local_borg_version,
+    global_arguments,
+    local_path='borg',
+    remote_path=None,
+):
+    '''
+    Given a local or remote repository path, an optional archive name, a filesystem mount point,
+    zero or more paths to mount from the archive, extra Borg mount options, a storage configuration
+    dict, the local Borg version, global arguments as an argparse.Namespace instance, and optional
+    local and remote Borg paths, mount the archive onto the mount point.
+    '''
+    umask = config.get('umask', None)
+    lock_wait = config.get('lock_wait', None)
+
+    full_command = (
+        (local_path, 'mount')
+        + (('--remote-path', remote_path) if remote_path else ())
+        + (('--umask', str(umask)) if umask else ())
+        + (('--log-json',) if global_arguments.log_json else ())
+        + (('--lock-wait', str(lock_wait)) if lock_wait else ())
+        + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
+        + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
+        + flags.make_flags_from_arguments(
+            mount_arguments,
+            excludes=('repository', 'archive', 'mount_point', 'paths', 'options'),
+        )
+        + (('-o', mount_arguments.options) if mount_arguments.options else ())
+        + (
+            (
+                flags.make_repository_flags(repository_path, local_borg_version)
+                + (
+                    ('--match-archives', archive)
+                    if feature.available(feature.Feature.MATCH_ARCHIVES, local_borg_version)
+                    else ('--glob-archives', archive)
+                )
+            )
+            if feature.available(feature.Feature.SEPARATE_REPOSITORY_ARCHIVE, local_borg_version)
+            else (
+                flags.make_repository_archive_flags(repository_path, archive, local_borg_version)
+                if archive
+                else flags.make_repository_flags(repository_path, local_borg_version)
+            )
+        )
+        + (mount_arguments.mount_point,)
+        + (tuple(mount_arguments.paths) if mount_arguments.paths else ())
+    )
+
+    borg_environment = environment.make_environment(config)
+
+    # Don't capture the output when foreground mode is used so that ctrl-C can work properly.
+    if mount_arguments.foreground:
+        execute_command(
+            full_command,
+            output_file=DO_NOT_CAPTURE,
+            extra_environment=borg_environment,
+            borg_local_path=local_path,
+            borg_exit_codes=config.get('borg_exit_codes'),
+        )
+        return
+
+    execute_command(
+        full_command,
+        extra_environment=borg_environment,
+        borg_local_path=local_path,
+        borg_exit_codes=config.get('borg_exit_codes'),
+    )

borgmatic/borg/prune.py

@@ -1,14 +1,16 @@
 import logging
-import subprocess
 
+import borgmatic.logger
+from borgmatic.borg import environment, feature, flags
+from borgmatic.execute import execute_command
 
 logger = logging.getLogger(__name__)
 
 
-def _make_prune_flags(retention_config):
+def make_prune_flags(config, local_borg_version):
     '''
-    Given a retention config dict mapping from option name to value, tranform it into an iterable of
-    command-line name-value flag pairs.
+    Given a configuration dict mapping from option name to value, transform it into an sequence of
+    command-line flags.
 
     For example, given a retention config of:
 
@@ -21,44 +23,78 @@ def _make_prune_flags(retention_config):
             ('--keep-monthly', '6'),
         )
     '''
-    if not retention_config.get('prefix'):
-        retention_config['prefix'] = '{hostname}-'
+    flag_pairs = (
+        ('--' + option_name.replace('_', '-'), str(value))
+        for option_name, value in config.items()
+        if option_name.startswith('keep_') and option_name != 'keep_exclude_tags'
+    )
+    prefix = config.get('prefix')
 
-    return (
-        ('--' + option_name.replace('_', '-'), str(retention_config[option_name]))
-        for option_name, value in retention_config.items()
+    return tuple(element for pair in flag_pairs for element in pair) + (
+        (
+            ('--match-archives', f'sh:{prefix}*')
+            if feature.available(feature.Feature.MATCH_ARCHIVES, local_borg_version)
+            else ('--glob-archives', f'{prefix}*')
+        )
+        if prefix
+        else (
+            flags.make_match_archives_flags(
+                config.get('match_archives'),
+                config.get('archive_name_format'),
+                local_borg_version,
+            )
+        )
     )
 
 
 def prune_archives(
     dry_run,
-    repository,
-    storage_config,
-    retention_config,
+    repository_path,
+    config,
+    local_borg_version,
+    prune_arguments,
+    global_arguments,
     local_path='borg',
     remote_path=None,
-    stats=False,
 ):
     '''
-    Given dry-run flag, a local or remote repository path, a storage config dict, and a
-    retention config dict, prune Borg archives according to the retention policy specified in that
-    configuration.
+    Given dry-run flag, a local or remote repository path, and a configuration dict, prune Borg
+    archives according to the retention policy specified in that configuration.
     '''
-    umask = storage_config.get('umask', None)
-    lock_wait = storage_config.get('lock_wait', None)
+    borgmatic.logger.add_custom_log_levels()
+    umask = config.get('umask', None)
+    lock_wait = config.get('lock_wait', None)
+    extra_borg_options = config.get('extra_borg_options', {}).get('prune', '')
 
     full_command = (
-        (local_path, 'prune', repository)
-        + tuple(element for pair in _make_prune_flags(retention_config) for element in pair)
+        (local_path, 'prune')
+        + make_prune_flags(config, local_borg_version)
         + (('--remote-path', remote_path) if remote_path else ())
         + (('--umask', str(umask)) if umask else ())
+        + (('--log-json',) if global_arguments.log_json else ())
         + (('--lock-wait', str(lock_wait)) if lock_wait else ())
-        + (('--stats',) if not dry_run and logger.isEnabledFor(logging.INFO) else ())
+        + (('--stats',) if prune_arguments.stats and not dry_run else ())
         + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
-        + (('--debug', '--list', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
+        + flags.make_flags_from_arguments(
+            prune_arguments,
+            excludes=('repository', 'stats', 'list_archives'),
+        )
+        + (('--list',) if prune_arguments.list_archives else ())
+        + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
         + (('--dry-run',) if dry_run else ())
-        + (('--stats',) if stats else ())
+        + (tuple(extra_borg_options.split(' ')) if extra_borg_options else ())
+        + flags.make_repository_flags(repository_path, local_borg_version)
     )
 
-    logger.debug(' '.join(full_command))
-    subprocess.check_call(full_command)
+    if prune_arguments.stats or prune_arguments.list_archives:
+        output_log_level = logging.ANSWER
+    else:
+        output_log_level = logging.INFO
+
+    execute_command(
+        full_command,
+        output_log_level=output_log_level,
+        extra_environment=environment.make_environment(config),
+        borg_local_path=local_path,
+        borg_exit_codes=config.get('borg_exit_codes'),
+    )

borgmatic/borg/rcreate.py

@@ -0,0 +1,101 @@
+import argparse
+import json
+import logging
+import subprocess
+
+from borgmatic.borg import environment, feature, flags, rinfo
+from borgmatic.execute import DO_NOT_CAPTURE, execute_command
+
+logger = logging.getLogger(__name__)
+
+
+RINFO_REPOSITORY_NOT_FOUND_EXIT_CODES = {2, 13}
+
+
+def create_repository(
+    dry_run,
+    repository_path,
+    config,
+    local_borg_version,
+    global_arguments,
+    encryption_mode,
+    source_repository=None,
+    copy_crypt_key=False,
+    append_only=None,
+    storage_quota=None,
+    make_parent_dirs=False,
+    local_path='borg',
+    remote_path=None,
+):
+    '''
+    Given a dry-run flag, a local or remote repository path, a configuration dict, the local Borg
+    version, a Borg encryption mode, the path to another repo whose key material should be reused,
+    whether the repository should be append-only, and the storage quota to use, create the
+    repository. If the repository already exists, then log and skip creation.
+
+    Raise ValueError if the requested encryption mode does not match that of the repository.
+    Raise json.decoder.JSONDecodeError if the "borg info" JSON outputcannot be decoded.
+    Raise subprocess.CalledProcessError if "borg info" returns an error exit code.
+    '''
+    try:
+        info_data = json.loads(
+            rinfo.display_repository_info(
+                repository_path,
+                config,
+                local_borg_version,
+                argparse.Namespace(json=True),
+                global_arguments,
+                local_path,
+                remote_path,
+            )
+        )
+        repository_encryption_mode = info_data.get('encryption', {}).get('mode')
+
+        if repository_encryption_mode != encryption_mode:
+            raise ValueError(
+                f'Requested encryption mode "{encryption_mode}" does not match existing repository encryption mode "{repository_encryption_mode}"'
+            )
+
+        logger.info(f'{repository_path}: Repository already exists. Skipping creation.')
+        return
+    except subprocess.CalledProcessError as error:
+        if error.returncode not in RINFO_REPOSITORY_NOT_FOUND_EXIT_CODES:
+            raise
+
+    lock_wait = config.get('lock_wait')
+    extra_borg_options = config.get('extra_borg_options', {}).get('rcreate', '')
+
+    rcreate_command = (
+        (local_path,)
+        + (
+            ('rcreate',)
+            if feature.available(feature.Feature.RCREATE, local_borg_version)
+            else ('init',)
+        )
+        + (('--encryption', encryption_mode) if encryption_mode else ())
+        + (('--other-repo', source_repository) if source_repository else ())
+        + (('--copy-crypt-key',) if copy_crypt_key else ())
+        + (('--append-only',) if append_only else ())
+        + (('--storage-quota', storage_quota) if storage_quota else ())
+        + (('--make-parent-dirs',) if make_parent_dirs else ())
+        + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
+        + (('--debug',) if logger.isEnabledFor(logging.DEBUG) else ())
+        + (('--log-json',) if global_arguments.log_json else ())
+        + (('--lock-wait', str(lock_wait)) if lock_wait else ())
+        + (('--remote-path', remote_path) if remote_path else ())
+        + (tuple(extra_borg_options.split(' ')) if extra_borg_options else ())
+        + flags.make_repository_flags(repository_path, local_borg_version)
+    )
+
+    if dry_run:
+        logging.info(f'{repository_path}: Skipping repository creation (dry run)')
+        return
+
+    # Do not capture output here, so as to support interactive prompts.
+    execute_command(
+        rcreate_command,
+        output_file=DO_NOT_CAPTURE,
+        extra_environment=environment.make_environment(config),
+        borg_local_path=local_path,
+        borg_exit_codes=config.get('borg_exit_codes'),
+    )

borgmatic/borg/rinfo.py

@@ -0,0 +1,68 @@
+import logging
+
+import borgmatic.logger
+from borgmatic.borg import environment, feature, flags
+from borgmatic.execute import execute_command, execute_command_and_capture_output
+
+logger = logging.getLogger(__name__)
+
+
+def display_repository_info(
+    repository_path,
+    config,
+    local_borg_version,
+    rinfo_arguments,
+    global_arguments,
+    local_path='borg',
+    remote_path=None,
+):
+    '''
+    Given a local or remote repository path, a configuration dict, the local Borg version, the
+    arguments to the rinfo action, and global arguments as an argparse.Namespace, display summary
+    information for the Borg repository or return JSON summary information.
+    '''
+    borgmatic.logger.add_custom_log_levels()
+    lock_wait = config.get('lock_wait', None)
+
+    full_command = (
+        (local_path,)
+        + (
+            ('rinfo',)
+            if feature.available(feature.Feature.RINFO, local_borg_version)
+            else ('info',)
+        )
+        + (
+            ('--info',)
+            if logger.getEffectiveLevel() == logging.INFO and not rinfo_arguments.json
+            else ()
+        )
+        + (
+            ('--debug', '--show-rc')
+            if logger.isEnabledFor(logging.DEBUG) and not rinfo_arguments.json
+            else ()
+        )
+        + flags.make_flags('remote-path', remote_path)
+        + flags.make_flags('log-json', global_arguments.log_json)
+        + flags.make_flags('lock-wait', lock_wait)
+        + (('--json',) if rinfo_arguments.json else ())
+        + flags.make_repository_flags(repository_path, local_borg_version)
+    )
+
+    extra_environment = environment.make_environment(config)
+    borg_exit_codes = config.get('borg_exit_codes')
+
+    if rinfo_arguments.json:
+        return execute_command_and_capture_output(
+            full_command,
+            extra_environment=extra_environment,
+            borg_local_path=local_path,
+            borg_exit_codes=borg_exit_codes,
+        )
+    else:
+        execute_command(
+            full_command,
+            output_log_level=logging.ANSWER,
+            extra_environment=extra_environment,
+            borg_local_path=local_path,
+            borg_exit_codes=borg_exit_codes,
+        )

borgmatic/borg/rlist.py

@@ -0,0 +1,171 @@
+import argparse
+import logging
+
+import borgmatic.logger
+from borgmatic.borg import environment, feature, flags
+from borgmatic.execute import execute_command, execute_command_and_capture_output
+
+logger = logging.getLogger(__name__)
+
+
+def resolve_archive_name(
+    repository_path,
+    archive,
+    config,
+    local_borg_version,
+    global_arguments,
+    local_path='borg',
+    remote_path=None,
+):
+    '''
+    Given a local or remote repository path, an archive name, a configuration dict, the local Borg
+    version, global arguments as an argparse.Namespace, a local Borg path, and a remote Borg path,
+    return the archive name. But if the archive name is "latest", then instead introspect the
+    repository for the latest archive and return its name.
+
+    Raise ValueError if "latest" is given but there are no archives in the repository.
+    '''
+    if archive != 'latest':
+        return archive
+
+    full_command = (
+        (
+            local_path,
+            'rlist' if feature.available(feature.Feature.RLIST, local_borg_version) else 'list',
+        )
+        + flags.make_flags('remote-path', remote_path)
+        + flags.make_flags('log-json', global_arguments.log_json)
+        + flags.make_flags('lock-wait', config.get('lock_wait'))
+        + flags.make_flags('last', 1)
+        + ('--short',)
+        + flags.make_repository_flags(repository_path, local_borg_version)
+    )
+
+    output = execute_command_and_capture_output(
+        full_command,
+        extra_environment=environment.make_environment(config),
+        borg_local_path=local_path,
+        borg_exit_codes=config.get('borg_exit_codes'),
+    )
+    try:
+        latest_archive = output.strip().splitlines()[-1]
+    except IndexError:
+        raise ValueError('No archives found in the repository')
+
+    logger.debug(f'{repository_path}: Latest archive is {latest_archive}')
+
+    return latest_archive
+
+
+MAKE_FLAGS_EXCLUDES = ('repository', 'prefix', 'match_archives')
+
+
+def make_rlist_command(
+    repository_path,
+    config,
+    local_borg_version,
+    rlist_arguments,
+    global_arguments,
+    local_path='borg',
+    remote_path=None,
+):
+    '''
+    Given a local or remote repository path, a configuration dict, the local Borg version, the
+    arguments to the rlist action, global arguments as an argparse.Namespace instance, and local and
+    remote Borg paths, return a command as a tuple to list archives with a repository.
+    '''
+    return (
+        (
+            local_path,
+            'rlist' if feature.available(feature.Feature.RLIST, local_borg_version) else 'list',
+        )
+        + (
+            ('--info',)
+            if logger.getEffectiveLevel() == logging.INFO and not rlist_arguments.json
+            else ()
+        )
+        + (
+            ('--debug', '--show-rc')
+            if logger.isEnabledFor(logging.DEBUG) and not rlist_arguments.json
+            else ()
+        )
+        + flags.make_flags('remote-path', remote_path)
+        + flags.make_flags('log-json', global_arguments.log_json)
+        + flags.make_flags('lock-wait', config.get('lock_wait'))
+        + (
+            (
+                flags.make_flags('match-archives', f'sh:{rlist_arguments.prefix}*')
+                if feature.available(feature.Feature.MATCH_ARCHIVES, local_borg_version)
+                else flags.make_flags('glob-archives', f'{rlist_arguments.prefix}*')
+            )
+            if rlist_arguments.prefix
+            else (
+                flags.make_match_archives_flags(
+                    rlist_arguments.match_archives or config.get('match_archives'),
+                    config.get('archive_name_format'),
+                    local_borg_version,
+                )
+            )
+        )
+        + flags.make_flags_from_arguments(rlist_arguments, excludes=MAKE_FLAGS_EXCLUDES)
+        + flags.make_repository_flags(repository_path, local_borg_version)
+    )
+
+
+def list_repository(
+    repository_path,
+    config,
+    local_borg_version,
+    rlist_arguments,
+    global_arguments,
+    local_path='borg',
+    remote_path=None,
+):
+    '''
+    Given a local or remote repository path, a configuration dict, the local Borg version, the
+    arguments to the list action, global arguments as an argparse.Namespace instance, and local and
+    remote Borg paths, display the output of listing Borg archives in the given repository (or
+    return JSON output).
+    '''
+    borgmatic.logger.add_custom_log_levels()
+    borg_environment = environment.make_environment(config)
+
+    main_command = make_rlist_command(
+        repository_path,
+        config,
+        local_borg_version,
+        rlist_arguments,
+        global_arguments,
+        local_path,
+        remote_path,
+    )
+    json_command = make_rlist_command(
+        repository_path,
+        config,
+        local_borg_version,
+        argparse.Namespace(**dict(rlist_arguments.__dict__, json=True)),
+        global_arguments,
+        local_path,
+        remote_path,
+    )
+    borg_exit_codes = config.get('borg_exit_codes')
+
+    json_listing = execute_command_and_capture_output(
+        json_command,
+        extra_environment=borg_environment,
+        borg_local_path=local_path,
+        borg_exit_codes=borg_exit_codes,
+    )
+
+    if rlist_arguments.json:
+        return json_listing
+
+    flags.warn_for_aggressive_archive_flags(json_command, json_listing)
+
+    execute_command(
+        main_command,
+        output_log_level=logging.ANSWER,
+        extra_environment=borg_environment,
+        borg_local_path=local_path,
+        borg_exit_codes=borg_exit_codes,
+    )

borgmatic/borg/state.py

@@ -0,0 +1 @@
+DEFAULT_BORGMATIC_SOURCE_DIRECTORY = '~/.borgmatic'

borgmatic/borg/transfer.py

@@ -0,0 +1,61 @@
+import logging
+
+import borgmatic.logger
+from borgmatic.borg import environment, flags
+from borgmatic.execute import DO_NOT_CAPTURE, execute_command
+
+logger = logging.getLogger(__name__)
+
+
+def transfer_archives(
+    dry_run,
+    repository_path,
+    config,
+    local_borg_version,
+    transfer_arguments,
+    global_arguments,
+    local_path='borg',
+    remote_path=None,
+):
+    '''
+    Given a dry-run flag, a local or remote repository path, a configuration dict, the local Borg
+    version, the arguments to the transfer action, and global arguments as an argparse.Namespace
+    instance, transfer archives to the given repository.
+    '''
+    borgmatic.logger.add_custom_log_levels()
+
+    full_command = (
+        (local_path, 'transfer')
+        + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
+        + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
+        + flags.make_flags('remote-path', remote_path)
+        + flags.make_flags('log-json', global_arguments.log_json)
+        + flags.make_flags('lock-wait', config.get('lock_wait', None))
+        + (
+            flags.make_flags_from_arguments(
+                transfer_arguments,
+                excludes=('repository', 'source_repository', 'archive', 'match_archives'),
+            )
+            or (
+                flags.make_match_archives_flags(
+                    transfer_arguments.match_archives
+                    or transfer_arguments.archive
+                    or config.get('match_archives'),
+                    config.get('archive_name_format'),
+                    local_borg_version,
+                )
+            )
+        )
+        + flags.make_repository_flags(repository_path, local_borg_version)
+        + flags.make_flags('other-repo', transfer_arguments.source_repository)
+        + flags.make_flags('dry-run', dry_run)
+    )
+
+    return execute_command(
+        full_command,
+        output_log_level=logging.ANSWER,
+        output_file=DO_NOT_CAPTURE if transfer_arguments.progress else None,
+        borg_local_path=local_path,
+        borg_exit_codes=config.get('borg_exit_codes'),
+        extra_environment=environment.make_environment(config),
+    )

borgmatic/borg/umount.py

@@ -0,0 +1,22 @@
+import logging
+
+from borgmatic.execute import execute_command
+
+logger = logging.getLogger(__name__)
+
+
+def unmount_archive(config, mount_point, local_path='borg'):
+    '''
+    Given a mounted filesystem mount point, and an optional local Borg paths, umount the filesystem
+    from the mount point.
+    '''
+    full_command = (
+        (local_path, 'umount')
+        + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
+        + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
+        + (mount_point,)
+    )
+
+    execute_command(
+        full_command, borg_local_path=local_path, borg_exit_codes=config.get('borg_exit_codes')
+    )

borgmatic/borg/version.py

@@ -0,0 +1,31 @@
+import logging
+
+from borgmatic.borg import environment
+from borgmatic.execute import execute_command_and_capture_output
+
+logger = logging.getLogger(__name__)
+
+
+def local_borg_version(config, local_path='borg'):
+    '''
+    Given a configuration dict and a local Borg binary path, return a version string for it.
+
+    Raise OSError or CalledProcessError if there is a problem running Borg.
+    Raise ValueError if the version cannot be parsed.
+    '''
+    full_command = (
+        (local_path, '--version')
+        + (('--info',) if logger.getEffectiveLevel() == logging.INFO else ())
+        + (('--debug', '--show-rc') if logger.isEnabledFor(logging.DEBUG) else ())
+    )
+    output = execute_command_and_capture_output(
+        full_command,
+        extra_environment=environment.make_environment(config),
+        borg_local_path=local_path,
+        borg_exit_codes=config.get('borg_exit_codes'),
+    )
+
+    try:
+        return output.split(' ')[1].strip()
+    except IndexError:
+        raise ValueError('Could not parse Borg version string')

borgmatic/commands/completion/actions.py

@@ -0,0 +1,36 @@
+import borgmatic.commands.arguments
+
+
+def upgrade_message(language: str, upgrade_command: str, completion_file: str):
+    return f'''
+Your {language} completions script is from a different version of borgmatic than is
+currently installed. Please upgrade your script so your completions match the
+command-line flags in your installed borgmatic! Try this to upgrade:
+
+    {upgrade_command}
+    source {completion_file}
+'''
+
+
+def available_actions(subparsers, current_action=None):
+    '''
+    Given subparsers as an argparse._SubParsersAction instance and a current action name (if
+    any), return the actions names that can follow the current action on a command-line.
+
+    This takes into account which sub-actions that the current action supports. For instance, if
+    "bootstrap" is a sub-action for "config", then "bootstrap" should be able to follow a current
+    action of "config" but not "list".
+    '''
+    action_to_subactions = borgmatic.commands.arguments.get_subactions_for_actions(
+        subparsers.choices
+    )
+    current_subactions = action_to_subactions.get(current_action)
+
+    if current_subactions:
+        return current_subactions
+
+    all_subactions = set(
+        subaction for subactions in action_to_subactions.values() for subaction in subactions
+    )
+
+    return tuple(action for action in subparsers.choices.keys() if action not in all_subactions)

borgmatic/commands/completion/bash.py

@@ -0,0 +1,66 @@
+import borgmatic.commands.arguments
+import borgmatic.commands.completion.actions
+
+
+def parser_flags(parser):
+    '''
+    Given an argparse.ArgumentParser instance, return its argument flags in a space-separated
+    string.
+    '''
+    return ' '.join(option for action in parser._actions for option in action.option_strings)
+
+
+def bash_completion():
+    '''
+    Return a bash completion script for the borgmatic command. Produce this by introspecting
+    borgmatic's command-line argument parsers.
+    '''
+    (
+        unused_global_parser,
+        action_parsers,
+        global_plus_action_parser,
+    ) = borgmatic.commands.arguments.make_parsers()
+    global_flags = parser_flags(global_plus_action_parser)
+
+    # Avert your eyes.
+    return '\n'.join(
+        (
+            'check_version() {',
+            '    local this_script="$(cat "$BASH_SOURCE" 2> /dev/null)"',
+            '    local installed_script="$(borgmatic --bash-completion 2> /dev/null)"',
+            '    if [ "$this_script" != "$installed_script" ] && [ "$installed_script" != "" ];'
+            f'''        then cat << EOF\n{borgmatic.commands.completion.actions.upgrade_message(
+                    'bash',
+                    'sudo sh -c "borgmatic --bash-completion > $BASH_SOURCE"',
+                    '$BASH_SOURCE',
+                )}\nEOF''',
+            '    fi',
+            '}',
+            'complete_borgmatic() {',
+        )
+        + tuple(
+            '''    if [[ " ${COMP_WORDS[*]} " =~ " %s " ]]; then
+        COMPREPLY=($(compgen -W "%s %s %s" -- "${COMP_WORDS[COMP_CWORD]}"))
+        return 0
+    fi'''
+            % (
+                action,
+                parser_flags(action_parser),
+                ' '.join(
+                    borgmatic.commands.completion.actions.available_actions(action_parsers, action)
+                ),
+                global_flags,
+            )
+            for action, action_parser in reversed(action_parsers.choices.items())
+        )
+        + (
+            '    COMPREPLY=($(compgen -W "%s %s" -- "${COMP_WORDS[COMP_CWORD]}"))'  # noqa: FS003
+            % (
+                ' '.join(borgmatic.commands.completion.actions.available_actions(action_parsers)),
+                global_flags,
+            ),
+            '    (check_version &)',
+            '}',
+            '\ncomplete -o bashdefault -o default -F complete_borgmatic borgmatic',
+        )
+    )

borgmatic/commands/completion/fish.py

@@ -0,0 +1,176 @@
+import shlex
+from argparse import Action
+from textwrap import dedent
+
+import borgmatic.commands.arguments
+import borgmatic.commands.completion.actions
+
+
+def has_file_options(action: Action):
+    '''
+    Given an argparse.Action instance, return True if it takes a file argument.
+    '''
+    return action.metavar in (
+        'FILENAME',
+        'PATH',
+    ) or action.dest in ('config_paths',)
+
+
+def has_choice_options(action: Action):
+    '''
+    Given an argparse.Action instance, return True if it takes one of a predefined set of arguments.
+    '''
+    return action.choices is not None
+
+
+def has_unknown_required_param_options(action: Action):
+    '''
+    A catch-all for options that take a required parameter, but we don't know what the parameter is.
+    This should be used last. These are actions that take something like a glob, a list of numbers, or a string.
+
+    Actions that match this pattern should not show the normal arguments, because those are unlikely to be valid.
+    '''
+    return (
+        action.required is True
+        or action.nargs
+        in (
+            '+',
+            '*',
+        )
+        or action.metavar in ('PATTERN', 'KEYS', 'N')
+        or (action.type is not None and action.default is None)
+    )
+
+
+def has_exact_options(action: Action):
+    return (
+        has_file_options(action)
+        or has_choice_options(action)
+        or has_unknown_required_param_options(action)
+    )
+
+
+def exact_options_completion(action: Action):
+    '''
+    Given an argparse.Action instance, return a completion invocation that forces file completions, options completion,
+    or just that some value follow the action, if the action takes such an argument and was the last action on the
+    command line prior to the cursor.
+
+    Otherwise, return an empty string.
+    '''
+
+    if not has_exact_options(action):
+        return ''
+
+    args = ' '.join(action.option_strings)
+
+    if has_file_options(action):
+        return f'''\ncomplete -c borgmatic -Fr -n "__borgmatic_current_arg {args}"'''
+
+    if has_choice_options(action):
+        return f'''\ncomplete -c borgmatic -f -a '{' '.join(map(str, action.choices))}' -n "__borgmatic_current_arg {args}"'''
+
+    if has_unknown_required_param_options(action):
+        return f'''\ncomplete -c borgmatic -x -n "__borgmatic_current_arg {args}"'''
+
+    raise ValueError(
+        f'Unexpected action: {action} passes has_exact_options but has no choices produced'
+    )
+
+
+def dedent_strip_as_tuple(string: str):
+    '''
+    Dedent a string, then strip it to avoid requiring your first line to have content, then return a tuple of the string.
+    Makes it easier to write multiline strings for completions when you join them with a tuple.
+    '''
+    return (dedent(string).strip('\n'),)
+
+
+def fish_completion():
+    '''
+    Return a fish completion script for the borgmatic command. Produce this by introspecting
+    borgmatic's command-line argument parsers.
+    '''
+    (
+        unused_global_parser,
+        action_parsers,
+        global_plus_action_parser,
+    ) = borgmatic.commands.arguments.make_parsers()
+
+    all_action_parsers = ' '.join(action for action in action_parsers.choices.keys())
+
+    exact_option_args = tuple(
+        ' '.join(action.option_strings)
+        for action_parser in action_parsers.choices.values()
+        for action in action_parser._actions
+        if has_exact_options(action)
+    ) + tuple(
+        ' '.join(action.option_strings)
+        for action in global_plus_action_parser._actions
+        if len(action.option_strings) > 0
+        if has_exact_options(action)
+    )
+
+    # Avert your eyes.
+    return '\n'.join(
+        dedent_strip_as_tuple(
+            f'''
+            function __borgmatic_check_version
+                set -fx this_filename (status current-filename)
+                fish -c '
+                    if test -f "$this_filename"
+                        set this_script (cat $this_filename 2> /dev/null)
+                        set installed_script (borgmatic --fish-completion 2> /dev/null)
+                        if [ "$this_script" != "$installed_script" ] && [ "$installed_script" != "" ]
+                            echo "{borgmatic.commands.completion.actions.upgrade_message(
+                            'fish',
+                            'borgmatic --fish-completion | sudo tee $this_filename',
+                            '$this_filename',
+                        )}"
+                        end
+                    end
+                ' &
+            end
+            __borgmatic_check_version
+
+            function __borgmatic_current_arg --description 'Check if any of the given arguments are the last on the command line before the cursor'
+                set -l all_args (commandline -poc)
+                # premature optimization to avoid iterating all args if there aren't enough
+                # to have a last arg beyond borgmatic
+                if [ (count $all_args) -lt 2 ]
+                    return 1
+                end
+                for arg in $argv
+                    if [ "$arg" = "$all_args[-1]" ]
+                        return 0
+                    end
+                end
+                return 1
+            end
+
+            set --local action_parser_condition "not __fish_seen_subcommand_from {all_action_parsers}"
+            set --local exact_option_condition "not __borgmatic_current_arg {' '.join(exact_option_args)}"
+            '''
+        )
+        + ('\n# action_parser completions',)
+        + tuple(
+            f'''complete -c borgmatic -f -n "$action_parser_condition" -n "$exact_option_condition" -a '{action_name}' -d {shlex.quote(action_parser.description)}'''
+            for action_name, action_parser in action_parsers.choices.items()
+        )
+        + ('\n# global flags',)
+        + tuple(
+            # -n is checked in order, so put faster / more likely to be true checks first
+            f'''complete -c borgmatic -f -n "$exact_option_condition" -a '{' '.join(action.option_strings)}' -d {shlex.quote(action.help)}{exact_options_completion(action)}'''
+            for action in global_plus_action_parser._actions
+            # ignore the noargs action, as this is an impossible completion for fish
+            if len(action.option_strings) > 0
+            if 'Deprecated' not in action.help
+        )
+        + ('\n# action_parser flags',)
+        + tuple(
+            f'''complete -c borgmatic -f -n "$exact_option_condition" -a '{' '.join(action.option_strings)}' -d {shlex.quote(action.help)} -n "__fish_seen_subcommand_from {action_name}"{exact_options_completion(action)}'''
+            for action_name, action_parser in action_parsers.choices.items()
+            for action in action_parser._actions
+            if 'Deprecated' not in (action.help or ())
+        )
+    )

borgmatic/commands/convert_config.py

@@ -1,109 +0,0 @@
-from argparse import ArgumentParser
-import os
-import sys
-import textwrap
-
-from ruamel import yaml
-
-from borgmatic.config import convert, generate, legacy, validate
-
-
-DEFAULT_SOURCE_CONFIG_FILENAME = '/etc/borgmatic/config'
-DEFAULT_SOURCE_EXCLUDES_FILENAME = '/etc/borgmatic/excludes'
-DEFAULT_DESTINATION_CONFIG_FILENAME = '/etc/borgmatic/config.yaml'
-
-
-def parse_arguments(*arguments):
-    '''
-    Given command-line arguments with which this script was invoked, parse the arguments and return
-    them as an ArgumentParser instance.
-    '''
-    parser = ArgumentParser(
-        description='''
-            Convert legacy INI-style borgmatic configuration and excludes files to a single YAML
-            configuration file. Note that this replaces any comments from the source files.
-        '''
-    )
-    parser.add_argument(
-        '-s',
-        '--source-config',
-        dest='source_config_filename',
-        default=DEFAULT_SOURCE_CONFIG_FILENAME,
-        help='Source INI-style configuration filename. Default: {}'.format(
-            DEFAULT_SOURCE_CONFIG_FILENAME
-        ),
-    )
-    parser.add_argument(
-        '-e',
-        '--source-excludes',
-        dest='source_excludes_filename',
-        default=DEFAULT_SOURCE_EXCLUDES_FILENAME
-        if os.path.exists(DEFAULT_SOURCE_EXCLUDES_FILENAME)
-        else None,
-        help='Excludes filename',
-    )
-    parser.add_argument(
-        '-d',
-        '--destination-config',
-        dest='destination_config_filename',
-        default=DEFAULT_DESTINATION_CONFIG_FILENAME,
-        help='Destination YAML configuration filename. Default: {}'.format(
-            DEFAULT_DESTINATION_CONFIG_FILENAME
-        ),
-    )
-
-    return parser.parse_args(arguments)
-
-
-TEXT_WRAP_CHARACTERS = 80
-
-
-def display_result(args):  # pragma: no cover
-    result_lines = textwrap.wrap(
-        'Your borgmatic configuration has been upgraded. Please review the result in {}.'.format(
-            args.destination_config_filename
-        ),
-        TEXT_WRAP_CHARACTERS,
-    )
-
-    delete_lines = textwrap.wrap(
-        'Once you are satisfied, you can safely delete {}{}.'.format(
-            args.source_config_filename,
-            ' and {}'.format(args.source_excludes_filename)
-            if args.source_excludes_filename
-            else '',
-        ),
-        TEXT_WRAP_CHARACTERS,
-    )
-
-    print('\n'.join(result_lines))
-    print()
-    print('\n'.join(delete_lines))
-
-
-def main():  # pragma: no cover
-    try:
-        args = parse_arguments(*sys.argv[1:])
-        schema = yaml.round_trip_load(open(validate.schema_filename()).read())
-        source_config = legacy.parse_configuration(
-            args.source_config_filename, legacy.CONFIG_FORMAT
-        )
-        source_config_file_mode = os.stat(args.source_config_filename).st_mode
-        source_excludes = (
-            open(args.source_excludes_filename).read().splitlines()
-            if args.source_excludes_filename
-            else []
-        )
-
-        destination_config = convert.convert_legacy_parsed_config(
-            source_config, source_excludes, schema
-        )
-
-        generate.write_configuration(
-            args.destination_config_filename, destination_config, mode=source_config_file_mode
-        )
-
-        display_result(args)
-    except (ValueError, OSError) as error:
-        print(error, file=sys.stderr)
-        sys.exit(1)

borgmatic/commands/generate_config.py

@@ -1,45 +1,17 @@
-from argparse import ArgumentParser
+import logging
 import sys
 
-from borgmatic.config import generate, validate
+import borgmatic.commands.borgmatic
 
 
-DEFAULT_DESTINATION_CONFIG_FILENAME = '/etc/borgmatic/config.yaml'
-
-
-def parse_arguments(*arguments):
-    '''
-    Given command-line arguments with which this script was invoked, parse the arguments and return
-    them as an ArgumentParser instance.
-    '''
-    parser = ArgumentParser(description='Generate a sample borgmatic YAML configuration file.')
-    parser.add_argument(
-        '-d',
-        '--destination',
-        dest='destination_filename',
-        default=DEFAULT_DESTINATION_CONFIG_FILENAME,
-        help='Destination YAML configuration filename. Default: {}'.format(
-            DEFAULT_DESTINATION_CONFIG_FILENAME
-        ),
+def main():
+    warning_log = logging.makeLogRecord(
+        dict(
+            levelno=logging.WARNING,
+            levelname='WARNING',
+            msg='generate-borgmatic-config is deprecated and will be removed from a future release. Please use "borgmatic config generate" instead.',
+        )
     )
 
-    return parser.parse_args(arguments)
-
-
-def main():  # pragma: no cover
-    try:
-        args = parse_arguments(*sys.argv[1:])
-
-        generate.generate_sample_configuration(
-            args.destination_filename, validate.schema_filename()
-        )
-
-        print('Generated a sample configuration file at {}.'.format(args.destination_filename))
-        print()
-        print('Please edit the file to suit your needs. The values are just representative.')
-        print('All fields are optional except where indicated.')
-        print()
-        print('If you ever need help: https://torsion.org/borgmatic/#issues')
-    except (ValueError, OSError) as error:
-        print(error, file=sys.stderr)
-        sys.exit(1)
+    sys.argv = ['borgmatic', 'config', 'generate'] + sys.argv[1:]
+    borgmatic.commands.borgmatic.main([warning_log])

borgmatic/commands/hook.py

@@ -1,24 +0,0 @@
-import logging
-import subprocess
-
-
-logger = logging.getLogger(__name__)
-
-
-def execute_hook(commands, config_filename, description):
-    if not commands:
-        logger.debug('{}: No commands to run for {} hook'.format(config_filename, description))
-        return
-
-    if len(commands) == 1:
-        logger.info('{}: Running command for {} hook'.format(config_filename, description))
-    else:
-        logger.info(
-            '{}: Running {} commands for {} hook'.format(
-                config_filename, len(commands), description
-            )
-        )
-
-    for command in commands:
-        logger.debug('{}: Hook command: {}'.format(config_filename, command))
-        subprocess.check_call(command, shell=True)

borgmatic/commands/validate_config.py

@@ -0,0 +1,17 @@
+import logging
+import sys
+
+import borgmatic.commands.borgmatic
+
+
+def main():
+    warning_log = logging.makeLogRecord(
+        dict(
+            levelno=logging.WARNING,
+            levelname='WARNING',
+            msg='validate-borgmatic-config is deprecated and will be removed from a future release. Please use "borgmatic config validate" instead.',
+        )
+    )
+
+    sys.argv = ['borgmatic', 'config', 'validate'] + sys.argv[1:]
+    borgmatic.commands.borgmatic.main([warning_log])

borgmatic/config/checks.py

@@ -1,9 +1,9 @@
-def repository_enabled_for_checks(repository, consistency):
+def repository_enabled_for_checks(repository, config):
     '''
-    Given a repository name and a consistency configuration dict, return whether the repository
-    is enabled to have consistency checks run.
+    Given a repository name and a configuration dict, return whether the
+    repository is enabled to have consistency checks run.
     '''
-    if not consistency.get('check_repositories'):
+    if not config.get('check_repositories'):
         return True
 
-    return repository in consistency['check_repositories']
+    return repository in config['check_repositories']

borgmatic/config/collect.py

@@ -1,29 +1,32 @@
 import os
 
 
-def get_default_config_paths():
+def get_default_config_paths(expand_home=True):
     '''
     Based on the value of the XDG_CONFIG_HOME and HOME environment variables, return a list of
     default configuration paths. This includes both system-wide configuration and configuration in
     the current user's home directory.
+
+    Don't expand the home directory ($HOME) if the expand home flag is False.
     '''
-    user_config_directory = os.getenv('XDG_CONFIG_HOME') or os.path.expandvars(
-        os.path.join('$HOME', '.config')
-    )
+    user_config_directory = os.getenv('XDG_CONFIG_HOME') or os.path.join('$HOME', '.config')
+    if expand_home:
+        user_config_directory = os.path.expandvars(user_config_directory)
 
     return [
         '/etc/borgmatic/config.yaml',
         '/etc/borgmatic.d',
-        '%s/borgmatic/config.yaml' % user_config_directory,
+        os.path.join(user_config_directory, 'borgmatic/config.yaml'),
+        os.path.join(user_config_directory, 'borgmatic.d'),
     ]
 
 
 def collect_config_filenames(config_paths):
     '''
-    Given a sequence of config paths, both filenames and directories, resolve that to just an
-    iterable of files. Accomplish this by listing any given directories looking for contained config
-    files (ending with the ".yaml" extension). This is non-recursive, so any directories within the
-    given directories are ignored.
+    Given a sequence of config paths, both filenames and directories, resolve that to an iterable
+    of absolute files. Accomplish this by listing any given directories looking for contained config
+    files (ending with the ".yaml" or ".yml" extension). This is non-recursive, so any directories
+    within the given directories are ignored.
 
     Return paths even if they don't exist on disk, so the user can find out about missing
     configuration paths. However, skip a default config path if it's missing, so the user doesn't
@@ -38,10 +41,14 @@ def collect_config_filenames(config_paths):
             continue
 
         if not os.path.isdir(path) or not exists:
-            yield path
+            yield os.path.abspath(path)
             continue
 
-        for filename in os.listdir(path):
+        if not os.access(path, os.R_OK):
+            continue
+
+        for filename in sorted(os.listdir(path)):
             full_filename = os.path.join(path, filename)
-            if full_filename.endswith('.yaml') and not os.path.isdir(full_filename):
-                yield full_filename
+            matching_filetype = full_filename.endswith('.yaml') or full_filename.endswith('.yml')
+            if matching_filetype and not os.path.isdir(full_filename):
+                yield os.path.abspath(full_filename)

borgmatic/config/constants.py

@@ -0,0 +1,61 @@
+import shlex
+
+
+def coerce_scalar(value):
+    '''
+    Given a configuration value, coerce it to an integer or a boolean as appropriate and return the
+    result.
+    '''
+    try:
+        return int(value)
+    except (TypeError, ValueError):
+        pass
+
+    if value == 'true' or value == 'True':
+        return True
+    if value == 'false' or value == 'False':
+        return False
+
+    return value
+
+
+def apply_constants(value, constants, shell_escape=False):
+    '''
+    Given a configuration value (bool, dict, int, list, or string) and a dict of named constants,
+    replace any configuration string values of the form "{constant}" (or containing it) with the
+    value of the correspondingly named key from the constants. Recurse as necessary into nested
+    configuration to find values to replace.
+
+    For instance, if a configuration value contains "{foo}", replace it with the value of the "foo"
+    key found within the configuration's "constants".
+
+    If shell escape is True, then escape the constant's value before applying it.
+
+    Return the configuration value and modify the original.
+    '''
+    if not value or not constants:
+        return value
+
+    if isinstance(value, str):
+        for constant_name, constant_value in constants.items():
+            value = value.replace(
+                '{' + constant_name + '}',
+                shlex.quote(str(constant_value)) if shell_escape else str(constant_value),
+            )
+
+        # Support constants within non-string scalars by coercing the value to its appropriate type.
+        value = coerce_scalar(value)
+    elif isinstance(value, list):
+        for index, list_value in enumerate(value):
+            value[index] = apply_constants(list_value, constants, shell_escape)
+    elif isinstance(value, dict):
+        for option_name, option_value in value.items():
+            shell_escape = (
+                shell_escape
+                or option_name.startswith('before_')
+                or option_name.startswith('after_')
+                or option_name == 'on_error'
+            )
+            value[option_name] = apply_constants(option_value, constants, shell_escape)
+
+    return value

borgmatic/config/convert.py

@@ -1,113 +0,0 @@
-import os
-
-from ruamel import yaml
-
-from borgmatic.config import generate
-
-
-def _convert_section(source_section_config, section_schema):
-    '''
-    Given a legacy Parsed_config instance for a single section, convert it to its corresponding
-    yaml.comments.CommentedMap representation in preparation for actual serialization to YAML.
-
-    Where integer types exist in the given section schema, convert their values to integers.
-    '''
-    destination_section_config = yaml.comments.CommentedMap(
-        [
-            (
-                option_name,
-                int(option_value)
-                if section_schema['map'].get(option_name, {}).get('type') == 'int'
-                else option_value,
-            )
-            for option_name, option_value in source_section_config.items()
-        ]
-    )
-
-    return destination_section_config
-
-
-def convert_legacy_parsed_config(source_config, source_excludes, schema):
-    '''
-    Given a legacy Parsed_config instance loaded from an INI-style config file and a list of exclude
-    patterns, convert them to a corresponding yaml.comments.CommentedMap representation in
-    preparation for serialization to a single YAML config file.
-
-    Additionally, use the given schema as a source of helpful comments to include within the
-    returned CommentedMap.
-    '''
-    destination_config = yaml.comments.CommentedMap(
-        [
-            (section_name, _convert_section(section_config, schema['map'][section_name]))
-            for section_name, section_config in source_config._asdict().items()
-        ]
-    )
-
-    # Split space-seperated values into actual lists, make "repository" into a list, and merge in
-    # excludes.
-    location = destination_config['location']
-    location['source_directories'] = source_config.location['source_directories'].split(' ')
-    location['repositories'] = [location.pop('repository')]
-    location['exclude_patterns'] = source_excludes
-
-    if source_config.consistency.get('checks'):
-        destination_config['consistency']['checks'] = source_config.consistency['checks'].split(' ')
-
-    # Add comments to each section, and then add comments to the fields in each section.
-    generate.add_comments_to_configuration(destination_config, schema)
-
-    for section_name, section_config in destination_config.items():
-        generate.add_comments_to_configuration(
-            section_config, schema['map'][section_name], indent=generate.INDENT
-        )
-
-    return destination_config
-
-
-class LegacyConfigurationNotUpgraded(FileNotFoundError):
-    def __init__(self):
-        super(LegacyConfigurationNotUpgraded, self).__init__(
-            '''borgmatic changed its configuration file format in version 1.1.0 from INI-style
-to YAML. This better supports validation, and has a more natural way to express
-lists of values. To upgrade your existing configuration, run:
-
-    sudo upgrade-borgmatic-config
-
-That will generate a new YAML configuration file at /etc/borgmatic/config.yaml
-(by default) using the values from both your existing configuration and excludes
-files. The new version of borgmatic will consume the YAML configuration file
-instead of the old one.'''
-        )
-
-
-def guard_configuration_upgraded(source_config_filename, destination_config_filenames):
-    '''
-    If legacy source configuration exists but no destination upgraded configs do, raise
-    LegacyConfigurationNotUpgraded.
-
-    The idea is that we want to alert the user about upgrading their config if they haven't already.
-    '''
-    destination_config_exists = any(
-        os.path.exists(filename) for filename in destination_config_filenames
-    )
-
-    if os.path.exists(source_config_filename) and not destination_config_exists:
-        raise LegacyConfigurationNotUpgraded()
-
-
-class LegacyExcludesFilenamePresent(FileNotFoundError):
-    def __init__(self):
-        super(LegacyExcludesFilenamePresent, self).__init__(
-            '''borgmatic changed its configuration file format in version 1.1.0 from INI-style
-to YAML. This better supports validation, and has a more natural way to express
-lists of values. The new configuration file incorporates excludes, so you no
-longer need to provide an excludes filename on the command-line with an
-"--excludes" argument.
-
-Please remove the "--excludes" argument and run borgmatic again.'''
-        )
-
-
-def guard_excludes_filename_omitted(excludes_filename):
-    if excludes_filename is not None:
-        raise LegacyExcludesFilenamePresent()

borgmatic/config/environment.py

@@ -0,0 +1,51 @@
+import os
+import re
+
+VARIABLE_PATTERN = re.compile(
+    r'(?P<escape>\\)?(?P<variable>\$\{(?P<name>[A-Za-z0-9_]+)((:?-)(?P<default>[^}]+))?\})'
+)
+
+
+def resolve_string(matcher):
+    '''
+    Given a matcher containing a name and an optional default value, get the value from environment.
+
+    Raise ValueError if the variable is not defined in environment and no default value is provided.
+    '''
+    if matcher.group('escape') is not None:
+        # In the case of an escaped environment variable, unescape it.
+        return matcher.group('variable')
+
+    # Resolve the environment variable.
+    name, default = matcher.group('name'), matcher.group('default')
+    out = os.getenv(name, default=default)
+
+    if out is None:
+        raise ValueError(f'Cannot find variable {name} in environment')
+
+    return out
+
+
+def resolve_env_variables(item):
+    '''
+    Resolves variables like or ${FOO} from given configuration with values from process environment.
+
+    Supported formats:
+
+     * ${FOO} will return FOO env variable
+     * ${FOO-bar} or ${FOO:-bar} will return FOO env variable if it exists, else "bar"
+
+    Raise if any variable is missing in environment and no default value is provided.
+    '''
+    if isinstance(item, str):
+        return VARIABLE_PATTERN.sub(resolve_string, item)
+
+    if isinstance(item, list):
+        for index, subitem in enumerate(item):
+            item[index] = resolve_env_variables(subitem)
+
+    if isinstance(item, dict):
+        for key, value in item.items():
+            item[key] = resolve_env_variables(value)
+
+    return item

borgmatic/config/generate.py

@@ -1,106 +1,137 @@
+import collections
+import io
 import os
+import re
 
-from ruamel import yaml
+import ruamel.yaml
 
+from borgmatic.config import load, normalize
 
 INDENT = 4
+SEQUENCE_INDENT = 2
 
 
-def _insert_newline_before_comment(config, field_name):
+def insert_newline_before_comment(config, field_name):
     '''
     Using some ruamel.yaml black magic, insert a blank line in the config right before the given
     field and its comments.
     '''
     config.ca.items[field_name][1].insert(
-        0, yaml.tokens.CommentToken('\n', yaml.error.CommentMark(0), None)
+        0, ruamel.yaml.tokens.CommentToken('\n', ruamel.yaml.error.CommentMark(0), None)
     )
 
 
-def _schema_to_sample_configuration(schema, level=0):
+def get_properties(schema):
+    '''
+    Given a schema dict, return its properties. But if it's got sub-schemas with multiple different
+    potential properties, returned their merged properties instead.
+    '''
+    if 'oneOf' in schema:
+        return dict(
+            collections.ChainMap(*[sub_schema['properties'] for sub_schema in schema['oneOf']])
+        )
+
+    return schema['properties']
+
+
+def schema_to_sample_configuration(schema, level=0, parent_is_sequence=False):
     '''
     Given a loaded configuration schema, generate and return sample config for it. Include comments
-    for each section based on the schema "desc" description.
+    for each option based on the schema "description".
     '''
+    schema_type = schema.get('type')
     example = schema.get('example')
     if example is not None:
         return example
 
-    config = yaml.comments.CommentedMap(
-        [
-            (section_name, _schema_to_sample_configuration(section_schema, level + 1))
-            for section_name, section_schema in schema['map'].items()
-        ]
-    )
-
-    add_comments_to_configuration(config, schema, indent=(level * INDENT))
+    if schema_type == 'array':
+        config = ruamel.yaml.comments.CommentedSeq(
+            [schema_to_sample_configuration(schema['items'], level, parent_is_sequence=True)]
+        )
+        add_comments_to_configuration_sequence(config, schema, indent=(level * INDENT))
+    elif schema_type == 'object':
+        config = ruamel.yaml.comments.CommentedMap(
+            [
+                (field_name, schema_to_sample_configuration(sub_schema, level + 1))
+                for field_name, sub_schema in get_properties(schema).items()
+            ]
+        )
+        indent = (level * INDENT) + (SEQUENCE_INDENT if parent_is_sequence else 0)
+        add_comments_to_configuration_object(
+            config, schema, indent=indent, skip_first=parent_is_sequence
+        )
+    else:
+        raise ValueError(f'Schema at level {level} is unsupported: {schema}')
 
     return config
 
 
-def _comment_out_line(line):
+def comment_out_line(line):
     # If it's already is commented out (or empty), there's nothing further to do!
     stripped_line = line.lstrip()
     if not stripped_line or stripped_line.startswith('#'):
         return line
 
-    # Comment out the names of optional sections.
-    one_indent = ' ' * INDENT
-    if not line.startswith(one_indent):
-        return '#' + line
+    # Comment out the names of optional options, inserting the '#' after any indent for aesthetics.
+    matches = re.match(r'(\s*)', line)
+    indent_spaces = matches.group(0) if matches else ''
+    count_indent_spaces = len(indent_spaces)
 
-    # Otherwise, comment out the line, but insert the "#" after the first indent for aesthetics.
-    return '#'.join((one_indent, line[INDENT:]))
+    return '# '.join((indent_spaces, line[count_indent_spaces:]))
 
 
-REQUIRED_KEYS = {'source_directories', 'repositories', 'keep_daily'}
-REQUIRED_SECTION_NAMES = {'location', 'retention'}
-
-
-def _comment_out_optional_configuration(rendered_config):
+def comment_out_optional_configuration(rendered_config):
     '''
-    Post-process a rendered configuration string to comment out optional key/values. The idea is
-    that this prevents the user from having to comment out a bunch of configuration they don't care
-    about to get to a minimal viable configuration file.
+    Post-process a rendered configuration string to comment out optional key/values, as determined
+    by a sentinel in the comment before each key.
 
-    Ideally ruamel.yaml would support this during configuration generation, but it's not terribly
-    easy to accomplish that way.
+    The idea is that the pre-commented configuration prevents the user from having to comment out a
+    bunch of configuration they don't care about to get to a minimal viable configuration file.
+
+    Ideally ruamel.yaml would support commenting out keys during configuration generation, but it's
+    not terribly easy to accomplish that way.
     '''
     lines = []
-    required = False
+    optional = False
 
     for line in rendered_config.split('\n'):
-        key = line.strip().split(':')[0]
-
-        if key in REQUIRED_SECTION_NAMES:
-            lines.append(line)
+        # Upon encountering an optional configuration option, comment out lines until the next blank
+        # line.
+        if line.strip().startswith(f'# {COMMENTED_OUT_SENTINEL}'):
+            optional = True
             continue
 
-        # Upon encountering a required configuration option, skip commenting out lines until the
-        # next blank line.
-        if key in REQUIRED_KEYS:
-            required = True
-        elif not key:
-            required = False
+        # Hit a blank line, so reset commenting.
+        if not line.strip():
+            optional = False
 
-        lines.append(_comment_out_line(line) if not required else line)
+        lines.append(comment_out_line(line) if optional else line)
 
     return '\n'.join(lines)
 
 
-def _render_configuration(config):
+def render_configuration(config):
     '''
     Given a config data structure of nested OrderedDicts, render the config as YAML and return it.
     '''
-    return yaml.round_trip_dump(config, indent=INDENT, block_seq_indent=INDENT)
+    dumper = ruamel.yaml.YAML(typ='rt')
+    dumper.indent(mapping=INDENT, sequence=INDENT + SEQUENCE_INDENT, offset=INDENT)
+    rendered = io.StringIO()
+    dumper.dump(config, rendered)
+
+    return rendered.getvalue()
 
 
-def write_configuration(config_filename, rendered_config, mode=0o600):
+def write_configuration(config_filename, rendered_config, mode=0o600, overwrite=False):
     '''
     Given a target config filename and rendered config YAML, write it out to file. Create any
-    containing directories as needed.
+    containing directories as needed. But if the file already exists and overwrite is False,
+    abort before writing anything.
     '''
-    if os.path.exists(config_filename):
-        raise FileExistsError('{} already exists. Aborting.'.format(config_filename))
+    if not overwrite and os.path.exists(config_filename):
+        raise FileExistsError(
+            f'{config_filename} already exists. Aborting. Use --overwrite to replace the file.'
+        )
 
     try:
         os.makedirs(os.path.dirname(config_filename), mode=0o700)
@@ -113,33 +144,167 @@ def write_configuration(config_filename, rendered_config, mode=0o600):
     os.chmod(config_filename, mode)
 
 
-def add_comments_to_configuration(config, schema, indent=0):
+def add_comments_to_configuration_sequence(config, schema, indent=0):
     '''
-    Using descriptions from a schema as a source, add those descriptions as comments to the given
-    config before each field. This function only adds comments for the top-most config map level.
-    Indent the comment the given number of characters.
+    If the given config sequence's items are object, then mine the schema for the description of the
+    object's first item, and slap that atop the sequence. Indent the comment the given number of
+    characters.
+
+    Doing this for sequences of maps results in nice comments that look like:
+
+    ```
+    things:
+        # First key description. Added by this function.
+        - key: foo
+          # Second key description. Added by add_comments_to_configuration_object().
+          other: bar
+    ```
     '''
-    for index, field_name in enumerate(config.keys()):
-        field_schema = schema['map'].get(field_name, {})
-        description = field_schema.get('desc')
+    if schema['items'].get('type') != 'object':
+        return
+
+    for field_name in config[0].keys():
+        field_schema = get_properties(schema['items']).get(field_name, {})
+        description = field_schema.get('description')
 
         # No description to use? Skip it.
         if not field_schema or not description:
+            return
+
+        config[0].yaml_set_start_comment(description, indent=indent)
+
+        # We only want the first key's description here, as the rest of the keys get commented by
+        # add_comments_to_configuration_object().
+        return
+
+
+REQUIRED_KEYS = {'source_directories', 'repositories', 'keep_daily'}
+COMMENTED_OUT_SENTINEL = 'COMMENT_OUT'
+
+
+def add_comments_to_configuration_object(config, schema, indent=0, skip_first=False):
+    '''
+    Using descriptions from a schema as a source, add those descriptions as comments to the given
+    config mapping, before each field. Indent the comment the given number of characters.
+    '''
+    for index, field_name in enumerate(config.keys()):
+        if skip_first and index == 0:
+            continue
+
+        field_schema = get_properties(schema).get(field_name, {})
+        description = field_schema.get('description', '').strip()
+
+        # If this is an optional key, add an indicator to the comment flagging it to be commented
+        # out from the sample configuration. This sentinel is consumed by downstream processing that
+        # does the actual commenting out.
+        if field_name not in REQUIRED_KEYS:
+            description = (
+                '\n'.join((description, COMMENTED_OUT_SENTINEL))
+                if description
+                else COMMENTED_OUT_SENTINEL
+            )
+
+        # No description to use? Skip it.
+        if not field_schema or not description:  # pragma: no cover
             continue
 
         config.yaml_set_comment_before_after_key(key=field_name, before=description, indent=indent)
+
         if index > 0:
-            _insert_newline_before_comment(config, field_name)
+            insert_newline_before_comment(config, field_name)
 
 
-def generate_sample_configuration(config_filename, schema_filename):
+RUAMEL_YAML_COMMENTS_INDEX = 1
+
+
+def remove_commented_out_sentinel(config, field_name):
     '''
-    Given a target config filename and the path to a schema filename in pykwalify YAML schema
-    format, write out a sample configuration file based on that schema.
+    Given a configuration CommentedMap and a top-level field name in it, remove any "commented out"
+    sentinel found at the end of its YAML comments. This prevents the given field name from getting
+    commented out by downstream processing that consumes the sentinel.
     '''
-    schema = yaml.round_trip_load(open(schema_filename))
-    config = _schema_to_sample_configuration(schema)
+    try:
+        last_comment_value = config.ca.items[field_name][RUAMEL_YAML_COMMENTS_INDEX][-1].value
+    except KeyError:
+        return
+
+    if last_comment_value == f'# {COMMENTED_OUT_SENTINEL}\n':
+        config.ca.items[field_name][RUAMEL_YAML_COMMENTS_INDEX].pop()
+
+
+def merge_source_configuration_into_destination(destination_config, source_config):
+    '''
+    Deep merge the given source configuration dict into the destination configuration CommentedMap,
+    favoring values from the source when there are collisions.
+
+    The purpose of this is to upgrade configuration files from old versions of borgmatic by adding
+    new configuration keys and comments.
+    '''
+    if not source_config:
+        return destination_config
+    if not destination_config or not isinstance(source_config, collections.abc.Mapping):
+        return source_config
+
+    for field_name, source_value in source_config.items():
+        # Since this key/value is from the source configuration, leave it uncommented and remove any
+        # sentinel that would cause it to get commented out.
+        remove_commented_out_sentinel(
+            ruamel.yaml.comments.CommentedMap(destination_config), field_name
+        )
+
+        # This is a mapping. Recurse for this key/value.
+        if isinstance(source_value, collections.abc.Mapping):
+            destination_config[field_name] = merge_source_configuration_into_destination(
+                destination_config[field_name], source_value
+            )
+            continue
+
+        # This is a sequence. Recurse for each item in it.
+        if isinstance(source_value, collections.abc.Sequence) and not isinstance(source_value, str):
+            destination_value = destination_config[field_name]
+            destination_config[field_name] = ruamel.yaml.comments.CommentedSeq(
+                [
+                    merge_source_configuration_into_destination(
+                        destination_value[index] if index < len(destination_value) else None,
+                        source_item,
+                    )
+                    for index, source_item in enumerate(source_value)
+                ]
+            )
+            continue
+
+        # This is some sort of scalar. Set it into the destination.
+        destination_config[field_name] = source_config[field_name]
+
+    return destination_config
+
+
+def generate_sample_configuration(
+    dry_run, source_filename, destination_filename, schema_filename, overwrite=False
+):
+    '''
+    Given an optional source configuration filename, and a required destination configuration
+    filename, the path to a schema filename in a YAML rendition of the JSON Schema format, and
+    whether to overwrite a destination file, write out a sample configuration file based on that
+    schema. If a source filename is provided, merge the parsed contents of that configuration into
+    the generated configuration.
+    '''
+    schema = ruamel.yaml.YAML(typ='safe').load(open(schema_filename))
+    source_config = None
+
+    if source_filename:
+        source_config = load.load_configuration(source_filename)
+        normalize.normalize(source_filename, source_config)
+
+    destination_config = merge_source_configuration_into_destination(
+        schema_to_sample_configuration(schema), source_config
+    )
+
+    if dry_run:
+        return
 
     write_configuration(
-        config_filename, _comment_out_optional_configuration(_render_configuration(config))
+        destination_filename,
+        comment_out_optional_configuration(render_configuration(destination_config)),
+        overwrite=overwrite,
     )

borgmatic/config/legacy.py

@@ -1,153 +0,0 @@
-from collections import OrderedDict, namedtuple
-from configparser import RawConfigParser
-
-
-Section_format = namedtuple('Section_format', ('name', 'options'))
-Config_option = namedtuple('Config_option', ('name', 'value_type', 'required'))
-
-
-def option(name, value_type=str, required=True):
-    '''
-    Given a config file option name, an expected type for its value, and whether it's required,
-    return a Config_option capturing that information.
-    '''
-    return Config_option(name, value_type, required)
-
-
-CONFIG_FORMAT = (
-    Section_format(
-        'location',
-        (
-            option('source_directories'),
-            option('one_file_system', value_type=bool, required=False),
-            option('remote_path', required=False),
-            option('repository'),
-        ),
-    ),
-    Section_format(
-        'storage',
-        (
-            option('encryption_passphrase', required=False),
-            option('compression', required=False),
-            option('umask', required=False),
-        ),
-    ),
-    Section_format(
-        'retention',
-        (
-            option('keep_within', required=False),
-            option('keep_hourly', int, required=False),
-            option('keep_daily', int, required=False),
-            option('keep_weekly', int, required=False),
-            option('keep_monthly', int, required=False),
-            option('keep_yearly', int, required=False),
-            option('prefix', required=False),
-        ),
-    ),
-    Section_format(
-        'consistency', (option('checks', required=False), option('check_last', required=False))
-    ),
-)
-
-
-def validate_configuration_format(parser, config_format):
-    '''
-    Given an open RawConfigParser and an expected config file format, validate that the parsed
-    configuration file has the expected sections, that any required options are present in those
-    sections, and that there aren't any unexpected options.
-
-    A section is required if any of its contained options are required.
-
-    Raise ValueError if anything is awry.
-    '''
-    section_names = set(parser.sections())
-    required_section_names = tuple(
-        section.name
-        for section in config_format
-        if any(option.required for option in section.options)
-    )
-
-    unknown_section_names = section_names - set(
-        section_format.name for section_format in config_format
-    )
-    if unknown_section_names:
-        raise ValueError(
-            'Unknown config sections found: {}'.format(', '.join(unknown_section_names))
-        )
-
-    missing_section_names = set(required_section_names) - section_names
-    if missing_section_names:
-        raise ValueError('Missing config sections: {}'.format(', '.join(missing_section_names)))
-
-    for section_format in config_format:
-        if section_format.name not in section_names:
-            continue
-
-        option_names = parser.options(section_format.name)
-        expected_options = section_format.options
-
-        unexpected_option_names = set(option_names) - set(
-            option.name for option in expected_options
-        )
-
-        if unexpected_option_names:
-            raise ValueError(
-                'Unexpected options found in config section {}: {}'.format(
-                    section_format.name, ', '.join(sorted(unexpected_option_names))
-                )
-            )
-
-        missing_option_names = tuple(
-            option.name
-            for option in expected_options
-            if option.required
-            if option.name not in option_names
-        )
-
-        if missing_option_names:
-            raise ValueError(
-                'Required options missing from config section {}: {}'.format(
-                    section_format.name, ', '.join(missing_option_names)
-                )
-            )
-
-
-def parse_section_options(parser, section_format):
-    '''
-    Given an open RawConfigParser and an expected section format, return the option values from that
-    section as a dict mapping from option name to value. Omit those options that are not present in
-    the parsed options.
-
-    Raise ValueError if any option values cannot be coerced to the expected Python data type.
-    '''
-    type_getter = {str: parser.get, int: parser.getint, bool: parser.getboolean}
-
-    return OrderedDict(
-        (option.name, type_getter[option.value_type](section_format.name, option.name))
-        for option in section_format.options
-        if parser.has_option(section_format.name, option.name)
-    )
-
-
-def parse_configuration(config_filename, config_format):
-    '''
-    Given a config filename and an expected config file format, return the parsed configuration
-    as a namedtuple with one attribute for each parsed section.
-
-    Raise IOError if the file cannot be read, or ValueError if the format is not as expected.
-    '''
-    parser = RawConfigParser()
-    if not parser.read(config_filename):
-        raise ValueError('Configuration file cannot be opened: {}'.format(config_filename))
-
-    validate_configuration_format(parser, config_format)
-
-    # Describes a parsed configuration, where each attribute is the name of a configuration file
-    # section and each value is a dict of that section's parsed options.
-    Parsed_config = namedtuple(
-        'Parsed_config', (section_format.name for section_format in config_format)
-    )
-
-    return Parsed_config(
-        *(parse_section_options(parser, section_format) for section_format in config_format)
-    )

borgmatic/config/load.py

@@ -0,0 +1,384 @@
+import functools
+import itertools
+import logging
+import operator
+import os
+
+import ruamel.yaml
+
+logger = logging.getLogger(__name__)
+
+
+def probe_and_include_file(filename, include_directories, config_paths):
+    '''
+    Given a filename to include, a list of include directories to search for matching files, and a
+    set of configuration paths, probe for the file, load it, and return the loaded configuration as
+    a data structure of nested dicts, lists, etc. Add the filename to the given configuration paths.
+
+    Raise FileNotFoundError if the included file was not found.
+    '''
+    expanded_filename = os.path.expanduser(filename)
+
+    if os.path.isabs(expanded_filename):
+        return load_configuration(expanded_filename, config_paths)
+
+    candidate_filenames = {
+        os.path.join(directory, expanded_filename) for directory in include_directories
+    }
+
+    for candidate_filename in candidate_filenames:
+        if os.path.exists(candidate_filename):
+            return load_configuration(candidate_filename, config_paths)
+
+    raise FileNotFoundError(
+        f'Could not find include {filename} at {" or ".join(candidate_filenames)}'
+    )
+
+
+def include_configuration(loader, filename_node, include_directory, config_paths):
+    '''
+    Given a ruamel.yaml.loader.Loader, a ruamel.yaml.nodes.ScalarNode containing the included
+    filename (or a list containing multiple such filenames), an include directory path to search for
+    matching files, and a set of configuration paths, load the given YAML filenames (ignoring the
+    given loader so we can use our own) and return their contents as data structure of nested dicts,
+    lists, etc. Add the names of included files to the given configuration paths. If the given
+    filename node's value is a scalar string, then the return value will be a single value. But if
+    the given node value is a list, then the return value will be a list of values, one per loaded
+    configuration file.
+
+    If a filename is relative, probe for it within: 1. the current working directory and 2. the
+    given include directory.
+
+    Raise FileNotFoundError if an included file was not found.
+    '''
+    include_directories = [os.getcwd(), os.path.abspath(include_directory)]
+
+    if isinstance(filename_node.value, str):
+        return probe_and_include_file(filename_node.value, include_directories, config_paths)
+
+    if (
+        isinstance(filename_node.value, list)
+        and len(filename_node.value)
+        and isinstance(filename_node.value[0], ruamel.yaml.nodes.ScalarNode)
+    ):
+        # Reversing the values ensures the correct ordering if these includes are subsequently
+        # merged together.
+        return [
+            probe_and_include_file(node.value, include_directories, config_paths)
+            for node in reversed(filename_node.value)
+        ]
+
+    raise ValueError(
+        '!include value is not supported; use a single filename or a list of filenames'
+    )
+
+
+def raise_retain_node_error(loader, node):
+    '''
+    Given a ruamel.yaml.loader.Loader and a YAML node, raise an error about "!retain" usage.
+
+    Raise ValueError if a mapping or sequence node is given, as that indicates that "!retain" was
+    used in a configuration file without a merge. In configuration files with a merge, mapping and
+    sequence nodes with "!retain" tags are handled by deep_merge_nodes() below.
+
+    Also raise ValueError if a scalar node is given, as "!retain" is not supported on scalar nodes.
+    '''
+    if isinstance(node, (ruamel.yaml.nodes.MappingNode, ruamel.yaml.nodes.SequenceNode)):
+        raise ValueError(
+            'The !retain tag may only be used within a configuration file containing a merged !include tag.'
+        )
+
+    raise ValueError('The !retain tag may only be used on a mapping or list.')
+
+
+def raise_omit_node_error(loader, node):
+    '''
+    Given a ruamel.yaml.loader.Loader and a YAML node, raise an error about "!omit" usage.
+
+    Raise ValueError unconditionally, as an "!omit" node here indicates it was used in a
+    configuration file without a merge. In configuration files with a merge, nodes with "!omit"
+    tags are handled by deep_merge_nodes() below.
+    '''
+    raise ValueError(
+        'The !omit tag may only be used on a scalar (e.g., string) or list element within a configuration file containing a merged !include tag.'
+    )
+
+
+class Include_constructor(ruamel.yaml.SafeConstructor):
+    '''
+    A YAML "constructor" (a ruamel.yaml concept) that supports a custom "!include" tag for including
+    separate YAML configuration files. Example syntax: `option: !include common.yaml`
+    '''
+
+    def __init__(
+        self, preserve_quotes=None, loader=None, include_directory=None, config_paths=None
+    ):
+        super(Include_constructor, self).__init__(preserve_quotes, loader)
+        self.add_constructor(
+            '!include',
+            functools.partial(
+                include_configuration,
+                include_directory=include_directory,
+                config_paths=config_paths,
+            ),
+        )
+
+        # These are catch-all error handlers for tags that don't get applied and removed by
+        # deep_merge_nodes() below.
+        self.add_constructor('!retain', raise_retain_node_error)
+        self.add_constructor('!omit', raise_omit_node_error)
+
+    def flatten_mapping(self, node):
+        '''
+        Support the special case of deep merging included configuration into an existing mapping
+        using the YAML '<<' merge key. Example syntax:
+
+        ```
+        option:
+            sub_option: 1
+
+        <<: !include common.yaml
+        ```
+
+        These includes are deep merged into the current configuration file. For instance, in this
+        example, any "option" with sub-options in common.yaml will get merged into the corresponding
+        "option" with sub-options in the example configuration file.
+        '''
+        representer = ruamel.yaml.representer.SafeRepresenter()
+
+        for index, (key_node, value_node) in enumerate(node.value):
+            if key_node.tag == u'tag:yaml.org,2002:merge' and value_node.tag == '!include':
+                # Replace the merge include with a sequence of included configuration nodes ready
+                # for merging. The construct_object() call here triggers include_configuration()
+                # among other constructors.
+                node.value[index] = (
+                    key_node,
+                    representer.represent_data(self.construct_object(value_node)),
+                )
+
+        # This super().flatten_mapping() call actually performs "<<" merges.
+        super(Include_constructor, self).flatten_mapping(node)
+
+        node.value = deep_merge_nodes(node.value)
+
+
+def load_configuration(filename, config_paths=None):
+    '''
+    Load the given configuration file and return its contents as a data structure of nested dicts
+    and lists. Add the filename to the given configuration paths set, and also add any included
+    configuration filenames.
+
+    Raise ruamel.yaml.error.YAMLError if something goes wrong parsing the YAML, or RecursionError
+    if there are too many recursive includes.
+    '''
+    if config_paths is None:
+        config_paths = set()
+
+    # Use an embedded derived class for the include constructor so as to capture the include
+    # directory and configuration paths values. (functools.partial doesn't work for this use case
+    # because yaml.Constructor has to be an actual class.)
+    class Include_constructor_with_extras(Include_constructor):
+        def __init__(self, preserve_quotes=None, loader=None):
+            super(Include_constructor_with_extras, self).__init__(
+                preserve_quotes,
+                loader,
+                include_directory=os.path.dirname(filename),
+                config_paths=config_paths,
+            )
+
+    yaml = ruamel.yaml.YAML(typ='safe')
+    yaml.Constructor = Include_constructor_with_extras
+    config_paths.add(filename)
+
+    with open(filename) as file:
+        return yaml.load(file.read())
+
+
+def filter_omitted_nodes(nodes, values):
+    '''
+    Given a nested borgmatic configuration data structure as a list of tuples in the form of:
+
+    [
+        (
+            ruamel.yaml.nodes.ScalarNode as a key,
+            ruamel.yaml.nodes.MappingNode or other Node as a value,
+        ),
+        ...
+    ]
+
+    ... and a combined list of all values for those nodes, return a filtered list of the values,
+    omitting any that have an "!omit" tag (or with a value matching such nodes).
+
+    But if only a single node is given, bail and return the given values unfiltered, as "!omit" only
+    applies when there are merge includes (and therefore multiple nodes).
+    '''
+    if len(nodes) <= 1:
+        return values
+
+    omitted_values = tuple(node.value for node in values if node.tag == '!omit')
+
+    return [node for node in values if node.value not in omitted_values]
+
+
+def merge_values(nodes):
+    '''
+    Given a nested borgmatic configuration data structure as a list of tuples in the form of:
+
+    [
+        (
+            ruamel.yaml.nodes.ScalarNode as a key,
+            ruamel.yaml.nodes.MappingNode or other Node as a value,
+        ),
+        ...
+    ]
+
+    ... merge its sequence or mapping node values and return the result. For sequence nodes, this
+    means appending together its contained lists. For mapping nodes, it means merging its contained
+    dicts.
+    '''
+    return functools.reduce(operator.add, (value.value for key, value in nodes))
+
+
+def deep_merge_nodes(nodes):
+    '''
+    Given a nested borgmatic configuration data structure as a list of tuples in the form of:
+
+    [
+        (
+            ruamel.yaml.nodes.ScalarNode as a key,
+            ruamel.yaml.nodes.MappingNode or other Node as a value,
+        ),
+        ...
+    ]
+
+    ... deep merge any node values corresponding to duplicate keys and return the result. The
+    purpose of merging like this is to support, for instance, merging one borgmatic configuration
+    file into another for reuse, such that a configuration option with sub-options does not
+    completely replace the corresponding option in a merged file.
+
+    If there are colliding keys with scalar values (e.g., integers or strings), the last of the
+    values wins.
+
+    For instance, given node values of:
+
+        [
+            (
+                ScalarNode(tag='tag:yaml.org,2002:str', value='option'),
+                MappingNode(tag='tag:yaml.org,2002:map', value=[
+                    (
+                        ScalarNode(tag='tag:yaml.org,2002:str', value='sub_option1'),
+                        ScalarNode(tag='tag:yaml.org,2002:int', value='1')
+                    ),
+                    (
+                        ScalarNode(tag='tag:yaml.org,2002:str', value='sub_option2'),
+                        ScalarNode(tag='tag:yaml.org,2002:int', value='2')
+                    ),
+                ]),
+            ),
+            (
+                ScalarNode(tag='tag:yaml.org,2002:str', value='option'),
+                MappingNode(tag='tag:yaml.org,2002:map', value=[
+                    (
+                        ScalarNode(tag='tag:yaml.org,2002:str', value='sub_option2'),
+                        ScalarNode(tag='tag:yaml.org,2002:int', value='5')
+                    ),
+                ]),
+            ),
+        ]
+
+    ... the returned result would be:
+
+        [
+            (
+                ScalarNode(tag='tag:yaml.org,2002:str', value='option'),
+                MappingNode(tag='tag:yaml.org,2002:map', value=[
+                    (
+                        ScalarNode(tag='tag:yaml.org,2002:str', value='sub_option1'),
+                        ScalarNode(tag='tag:yaml.org,2002:int', value='1')
+                    ),
+                    (
+                        ScalarNode(tag='tag:yaml.org,2002:str', value='sub_option2'),
+                        ScalarNode(tag='tag:yaml.org,2002:int', value='5')
+                    ),
+                ]),
+            ),
+        ]
+
+    This function supports multi-way merging, meaning that if the same option name exists three or
+    more times (at the same scope level), all of those instances get merged together.
+
+    If a mapping or sequence node has a YAML "!retain" tag, then that node is not merged.
+
+    Raise ValueError if a merge is implied using multiple incompatible types.
+    '''
+    merged_nodes = []
+
+    def get_node_key_name(node):
+        return node[0].value
+
+    # Bucket the nodes by their keys. Then merge all of the values sharing the same key.
+    for key_name, grouped_nodes in itertools.groupby(
+        sorted(nodes, key=get_node_key_name), get_node_key_name
+    ):
+        grouped_nodes = list(grouped_nodes)
+
+        # The merged node inherits its attributes from the final node in the group.
+        (last_node_key, last_node_value) = grouped_nodes[-1]
+        value_types = set(type(value) for (_, value) in grouped_nodes)
+
+        if len(value_types) > 1:
+            raise ValueError(
+                f'Incompatible types found when trying to merge "{key_name}:" values across configuration files: {", ".join(value_type.id for value_type in value_types)}'
+            )
+
+        # If we're dealing with MappingNodes, recurse and merge its values as well.
+        if ruamel.yaml.nodes.MappingNode in value_types:
+            # A "!retain" tag says to skip deep merging for this node. Replace the tag so
+            # downstream schema validation doesn't break on our application-specific tag.
+            if last_node_value.tag == '!retain' and len(grouped_nodes) > 1:
+                last_node_value.tag = 'tag:yaml.org,2002:map'
+                merged_nodes.append((last_node_key, last_node_value))
+            else:
+                merged_nodes.append(
+                    (
+                        last_node_key,
+                        ruamel.yaml.nodes.MappingNode(
+                            tag=last_node_value.tag,
+                            value=deep_merge_nodes(merge_values(grouped_nodes)),
+                            start_mark=last_node_value.start_mark,
+                            end_mark=last_node_value.end_mark,
+                            flow_style=last_node_value.flow_style,
+                            comment=last_node_value.comment,
+                            anchor=last_node_value.anchor,
+                        ),
+                    )
+                )
+
+            continue
+
+        # If we're dealing with SequenceNodes, merge by appending sequences together.
+        if ruamel.yaml.nodes.SequenceNode in value_types:
+            if last_node_value.tag == '!retain' and len(grouped_nodes) > 1:
+                last_node_value.tag = 'tag:yaml.org,2002:seq'
+                merged_nodes.append((last_node_key, last_node_value))
+            else:
+                merged_nodes.append(
+                    (
+                        last_node_key,
+                        ruamel.yaml.nodes.SequenceNode(
+                            tag=last_node_value.tag,
+                            value=filter_omitted_nodes(grouped_nodes, merge_values(grouped_nodes)),
+                            start_mark=last_node_value.start_mark,
+                            end_mark=last_node_value.end_mark,
+                            flow_style=last_node_value.flow_style,
+                            comment=last_node_value.comment,
+                            anchor=last_node_value.anchor,
+                        ),
+                    )
+                )
+
+            continue
+
+        merged_nodes.append((last_node_key, last_node_value))
+
+    return merged_nodes

borgmatic/config/normalize.py

@@ -0,0 +1,269 @@
+import logging
+import os
+
+
+def normalize_sections(config_filename, config):
+    '''
+    Given a configuration filename and a configuration dict of its loaded contents, airlift any
+    options out of sections ("location:", etc.) to the global scope and delete those sections.
+    Return any log message warnings produced based on the normalization performed.
+
+    Raise ValueError if the "prefix" option is set in both "location" and "consistency" sections.
+    '''
+    try:
+        location = config.get('location') or {}
+    except AttributeError:
+        raise ValueError('Configuration does not contain any options')
+
+    storage = config.get('storage') or {}
+    consistency = config.get('consistency') or {}
+    hooks = config.get('hooks') or {}
+
+    if (
+        location.get('prefix')
+        and consistency.get('prefix')
+        and location.get('prefix') != consistency.get('prefix')
+    ):
+        raise ValueError(
+            'The retention prefix and the consistency prefix cannot have different values (unless one is not set).'
+        )
+
+    if storage.get('umask') and hooks.get('umask') and storage.get('umask') != hooks.get('umask'):
+        raise ValueError(
+            'The storage umask and the hooks umask cannot have different values (unless one is not set).'
+        )
+
+    any_section_upgraded = False
+
+    # Move any options from deprecated sections into the global scope.
+    for section_name in ('location', 'storage', 'retention', 'consistency', 'output', 'hooks'):
+        section_config = config.get(section_name)
+
+        if section_config is not None:
+            any_section_upgraded = True
+            del config[section_name]
+            config.update(section_config)
+
+    if any_section_upgraded:
+        return [
+            logging.makeLogRecord(
+                dict(
+                    levelno=logging.WARNING,
+                    levelname='WARNING',
+                    msg=f'{config_filename}: Configuration sections (like location:, storage:, retention:, consistency:, and hooks:) are deprecated and support will be removed from a future release. To prepare for this, move your options out of sections to the global scope.',
+                )
+            )
+        ]
+
+    return []
+
+
+def normalize(config_filename, config):
+    '''
+    Given a configuration filename and a configuration dict of its loaded contents, apply particular
+    hard-coded rules to normalize the configuration to adhere to the current schema. Return any log
+    message warnings produced based on the normalization performed.
+
+    Raise ValueError the configuration cannot be normalized.
+    '''
+    logs = normalize_sections(config_filename, config)
+
+    # Upgrade exclude_if_present from a string to a list.
+    exclude_if_present = config.get('exclude_if_present')
+    if isinstance(exclude_if_present, str):
+        logs.append(
+            logging.makeLogRecord(
+                dict(
+                    levelno=logging.WARNING,
+                    levelname='WARNING',
+                    msg=f'{config_filename}: The exclude_if_present option now expects a list value. String values for this option are deprecated and support will be removed from a future release.',
+                )
+            )
+        )
+        config['exclude_if_present'] = [exclude_if_present]
+
+    # Upgrade various monitoring hooks from a string to a dict.
+    healthchecks = config.get('healthchecks')
+    if isinstance(healthchecks, str):
+        logs.append(
+            logging.makeLogRecord(
+                dict(
+                    levelno=logging.WARNING,
+                    levelname='WARNING',
+                    msg=f'{config_filename}: The healthchecks hook now expects a key/value pair with "ping_url" as a key. String values for this option are deprecated and support will be removed from a future release.',
+                )
+            )
+        )
+        config['healthchecks'] = {'ping_url': healthchecks}
+
+    cronitor = config.get('cronitor')
+    if isinstance(cronitor, str):
+        logs.append(
+            logging.makeLogRecord(
+                dict(
+                    levelno=logging.WARNING,
+                    levelname='WARNING',
+                    msg=f'{config_filename}: The healthchecks hook now expects key/value pairs. String values for this option are deprecated and support will be removed from a future release.',
+                )
+            )
+        )
+        config['cronitor'] = {'ping_url': cronitor}
+
+    pagerduty = config.get('pagerduty')
+    if isinstance(pagerduty, str):
+        logs.append(
+            logging.makeLogRecord(
+                dict(
+                    levelno=logging.WARNING,
+                    levelname='WARNING',
+                    msg=f'{config_filename}: The healthchecks hook now expects key/value pairs. String values for this option are deprecated and support will be removed from a future release.',
+                )
+            )
+        )
+        config['pagerduty'] = {'integration_key': pagerduty}
+
+    cronhub = config.get('cronhub')
+    if isinstance(cronhub, str):
+        logs.append(
+            logging.makeLogRecord(
+                dict(
+                    levelno=logging.WARNING,
+                    levelname='WARNING',
+                    msg=f'{config_filename}: The healthchecks hook now expects key/value pairs. String values for this option are deprecated and support will be removed from a future release.',
+                )
+            )
+        )
+        config['cronhub'] = {'ping_url': cronhub}
+
+    # Upgrade consistency checks from a list of strings to a list of dicts.
+    checks = config.get('checks')
+    if isinstance(checks, list) and len(checks) and isinstance(checks[0], str):
+        logs.append(
+            logging.makeLogRecord(
+                dict(
+                    levelno=logging.WARNING,
+                    levelname='WARNING',
+                    msg=f'{config_filename}: The checks option now expects a list of key/value pairs. Lists of strings for this option are deprecated and support will be removed from a future release.',
+                )
+            )
+        )
+        config['checks'] = [{'name': check_type} for check_type in checks]
+
+    # Rename various configuration options.
+    numeric_owner = config.pop('numeric_owner', None)
+    if numeric_owner is not None:
+        logs.append(
+            logging.makeLogRecord(
+                dict(
+                    levelno=logging.WARNING,
+                    levelname='WARNING',
+                    msg=f'{config_filename}: The numeric_owner option has been renamed to numeric_ids. numeric_owner is deprecated and support will be removed from a future release.',
+                )
+            )
+        )
+        config['numeric_ids'] = numeric_owner
+
+    bsd_flags = config.pop('bsd_flags', None)
+    if bsd_flags is not None:
+        logs.append(
+            logging.makeLogRecord(
+                dict(
+                    levelno=logging.WARNING,
+                    levelname='WARNING',
+                    msg=f'{config_filename}: The bsd_flags option has been renamed to flags. bsd_flags is deprecated and support will be removed from a future release.',
+                )
+            )
+        )
+        config['flags'] = bsd_flags
+
+    remote_rate_limit = config.pop('remote_rate_limit', None)
+    if remote_rate_limit is not None:
+        logs.append(
+            logging.makeLogRecord(
+                dict(
+                    levelno=logging.WARNING,
+                    levelname='WARNING',
+                    msg=f'{config_filename}: The remote_rate_limit option has been renamed to upload_rate_limit. remote_rate_limit is deprecated and support will be removed from a future release.',
+                )
+            )
+        )
+        config['upload_rate_limit'] = remote_rate_limit
+
+    # Upgrade remote repositories to ssh:// syntax, required in Borg 2.
+    repositories = config.get('repositories')
+    if repositories:
+        if any(isinstance(repository, str) for repository in repositories):
+            logs.append(
+                logging.makeLogRecord(
+                    dict(
+                        levelno=logging.WARNING,
+                        levelname='WARNING',
+                        msg=f'{config_filename}: The repositories option now expects a list of key/value pairs. Lists of strings for this option are deprecated and support will be removed from a future release.',
+                    )
+                )
+            )
+            config['repositories'] = [
+                {'path': repository} if isinstance(repository, str) else repository
+                for repository in repositories
+            ]
+            repositories = config['repositories']
+
+        config['repositories'] = []
+
+        for repository_dict in repositories:
+            repository_path = repository_dict['path']
+            if '~' in repository_path:
+                logs.append(
+                    logging.makeLogRecord(
+                        dict(
+                            levelno=logging.WARNING,
+                            levelname='WARNING',
+                            msg=f'{config_filename}: Repository paths containing "~" are deprecated in borgmatic and support will be removed from a future release.',
+                        )
+                    )
+                )
+            if ':' in repository_path:
+                if repository_path.startswith('file://'):
+                    updated_repository_path = os.path.abspath(
+                        repository_path.partition('file://')[-1]
+                    )
+                    config['repositories'].append(
+                        dict(
+                            repository_dict,
+                            path=updated_repository_path,
+                        )
+                    )
+                elif repository_path.startswith('ssh://'):
+                    config['repositories'].append(repository_dict)
+                else:
+                    rewritten_repository_path = f"ssh://{repository_path.replace(':~', '/~').replace(':/', '/').replace(':', '/./')}"
+                    logs.append(
+                        logging.makeLogRecord(
+                            dict(
+                                levelno=logging.WARNING,
+                                levelname='WARNING',
+                                msg=f'{config_filename}: Remote repository paths without ssh:// syntax are deprecated and support will be removed from a future release. Interpreting "{repository_path}" as "{rewritten_repository_path}"',
+                            )
+                        )
+                    )
+                    config['repositories'].append(
+                        dict(
+                            repository_dict,
+                            path=rewritten_repository_path,
+                        )
+                    )
+            else:
+                config['repositories'].append(repository_dict)
+
+    if config.get('prefix'):
+        logs.append(
+            logging.makeLogRecord(
+                dict(
+                    levelno=logging.WARNING,
+                    levelname='WARNING',
+                    msg=f'{config_filename}: The prefix option is deprecated and support will be removed from a future release. Use archive_name_format or match_archives instead.',
+                )
+            )
+        )
+
+    return logs

borgmatic/config/override.py

@@ -0,0 +1,139 @@
+import io
+
+import ruamel.yaml
+
+
+def set_values(config, keys, value):
+    '''
+    Given a hierarchy of configuration dicts, a sequence of parsed key strings, and a string value,
+    descend into the hierarchy based on the keys to set the value into the right place.
+    '''
+    if not keys:
+        return
+
+    first_key = keys[0]
+    if len(keys) == 1:
+        if isinstance(config, list):
+            raise ValueError(
+                'When overriding a list option, the value must use list syntax (e.g., "[foo, bar]" or "[{key: value}]" as appropriate)'
+            )
+
+        config[first_key] = value
+        return
+
+    if first_key not in config:
+        config[first_key] = {}
+
+    set_values(config[first_key], keys[1:], value)
+
+
+def convert_value_type(value, option_type):
+    '''
+    Given a string value and its schema type as a string, determine its logical type (string,
+    boolean, integer, etc.), and return it converted to that type.
+
+    If the option type is a string, leave the value as a string so that special characters in it
+    don't get interpreted as YAML during conversion.
+
+    Raise ruamel.yaml.error.YAMLError if there's a parse issue with the YAML.
+    '''
+    if option_type == 'string':
+        return value
+
+    return ruamel.yaml.YAML(typ='safe').load(io.StringIO(value))
+
+
+LEGACY_SECTION_NAMES = {'location', 'storage', 'retention', 'consistency', 'output', 'hooks'}
+
+
+def strip_section_names(parsed_override_key):
+    '''
+    Given a parsed override key as a tuple of option and suboption names, strip out any initial
+    legacy section names, since configuration file normalization also strips them out.
+    '''
+    if parsed_override_key[0] in LEGACY_SECTION_NAMES:
+        return parsed_override_key[1:]
+
+    return parsed_override_key
+
+
+def type_for_option(schema, option_keys):
+    '''
+    Given a configuration schema and a sequence of keys identifying an option, e.g.
+    ('extra_borg_options', 'init'), return the schema type of that option as a string.
+
+    Return None if the option or its type cannot be found in the schema.
+    '''
+    option_schema = schema
+
+    for key in option_keys:
+        try:
+            option_schema = option_schema['properties'][key]
+        except KeyError:
+            return None
+
+    try:
+        return option_schema['type']
+    except KeyError:
+        return None
+
+
+def parse_overrides(raw_overrides, schema):
+    '''
+    Given a sequence of configuration file override strings in the form of "option.suboption=value"
+    and a configuration schema dict, parse and return a sequence of tuples (keys, values), where
+    keys is a sequence of strings. For instance, given the following raw overrides:
+
+        ['my_option.suboption=value1', 'other_option=value2']
+
+    ... return this:
+
+        (
+            (('my_option', 'suboption'), 'value1'),
+            (('other_option'), 'value2'),
+        )
+
+    Raise ValueError if an override can't be parsed.
+    '''
+    if not raw_overrides:
+        return ()
+
+    parsed_overrides = []
+
+    for raw_override in raw_overrides:
+        try:
+            raw_keys, value = raw_override.split('=', 1)
+            keys = tuple(raw_keys.split('.'))
+            option_type = type_for_option(schema, keys)
+
+            parsed_overrides.append(
+                (
+                    keys,
+                    convert_value_type(value, option_type),
+                )
+            )
+        except ValueError:
+            raise ValueError(
+                f"Invalid override '{raw_override}'. Make sure you use the form: OPTION=VALUE or OPTION.SUBOPTION=VALUE"
+            )
+        except ruamel.yaml.error.YAMLError as error:
+            raise ValueError(f"Invalid override '{raw_override}': {error.problem}")
+
+    return tuple(parsed_overrides)
+
+
+def apply_overrides(config, schema, raw_overrides):
+    '''
+    Given a configuration dict, a corresponding configuration schema dict, and a sequence of
+    configuration file override strings in the form of "option.suboption=value", parse each override
+    and set it into the configuration dict.
+
+    Set the overrides into the configuration both with and without deprecated section names (if
+    used), so that the overrides work regardless of whether the configuration is also using
+    deprecated section names.
+    '''
+    overrides = parse_overrides(raw_overrides, schema)
+
+    for keys, value in overrides:
+        set_values(config, keys, value)
+        set_values(config, strip_section_names(keys), value)

borgmatic/config/validate.py

@@ -1,39 +1,68 @@
-import logging
+import os
 
-import pkg_resources
-import pykwalify.core
-import pykwalify.errors
-from ruamel import yaml
+import jsonschema
+import ruamel.yaml
 
-
-logger = logging.getLogger(__name__)
+import borgmatic.config
+from borgmatic.config import constants, environment, load, normalize, override
 
 
 def schema_filename():
     '''
     Path to the installed YAML configuration schema file, used to validate and parse the
     configuration.
+
+    Raise FileNotFoundError when the schema path does not exist.
     '''
-    return pkg_resources.resource_filename('borgmatic', 'config/schema.yaml')
+    schema_path = os.path.join(os.path.dirname(borgmatic.config.__file__), 'schema.yaml')
+
+    with open(schema_path):
+        return schema_path
+
+
+def format_json_error_path_element(path_element):
+    '''
+    Given a path element into a JSON data structure, format it for display as a string.
+    '''
+    if isinstance(path_element, int):
+        return str(f'[{path_element}]')
+
+    return str(f'.{path_element}')
+
+
+def format_json_error(error):
+    '''
+    Given an instance of jsonschema.exceptions.ValidationError, format it for display as a string.
+    '''
+    if not error.path:
+        return f'At the top level: {error.message}'
+
+    formatted_path = ''.join(format_json_error_path_element(element) for element in error.path)
+    return f"At '{formatted_path.lstrip('.')}': {error.message}"
 
 
 class Validation_error(ValueError):
     '''
-    A collection of error message strings generated when attempting to validate a particular
-    configurartion file.
+    A collection of error messages generated when attempting to validate a particular
+    configuration file.
     '''
 
-    def __init__(self, config_filename, error_messages):
+    def __init__(self, config_filename, errors):
+        '''
+        Given a configuration filename path and a sequence of string error messages, create a
+        Validation_error.
+        '''
         self.config_filename = config_filename
-        self.error_messages = error_messages
+        self.errors = errors
 
     def __str__(self):
         '''
         Render a validation error as a user-facing string.
         '''
-        return 'An error occurred while parsing a configuration file at {}:\n'.format(
-            self.config_filename
-        ) + '\n'.join(self.error_messages)
+        return (
+            f'An error occurred while parsing a configuration file at {self.config_filename}:\n'
+            + '\n'.join(error for error in self.errors)
+        )
 
 
 def apply_logical_validation(config_filename, parsed_configuration):
@@ -42,68 +71,141 @@ def apply_logical_validation(config_filename, parsed_configuration):
     below), run through any additional logical validation checks. If there are any such validation
     problems, raise a Validation_error.
     '''
-    archive_name_format = parsed_configuration.get('storage', {}).get('archive_name_format')
-    prefix = parsed_configuration.get('retention', {}).get('prefix')
-
-    if archive_name_format and not prefix:
-        raise Validation_error(
-            config_filename,
-            ('If you provide an archive_name_format, you must also specify a retention prefix.',),
-        )
-
-    location_repositories = parsed_configuration.get('location', {}).get('repositories')
-    check_repositories = parsed_configuration.get('consistency', {}).get('check_repositories', [])
+    repositories = parsed_configuration.get('repositories')
+    check_repositories = parsed_configuration.get('check_repositories', [])
     for repository in check_repositories:
-        if repository not in location_repositories:
+        if not any(
+            repositories_match(repository, config_repository) for config_repository in repositories
+        ):
             raise Validation_error(
                 config_filename,
-                (
-                    'Unknown repository in the consistency section\'s check_repositories: {}'.format(
-                        repository
-                    ),
-                ),
+                (f'Unknown repository in "check_repositories": {repository}',),
             )
 
-    consistency_prefix = parsed_configuration.get('consistency', {}).get('prefix')
-    if archive_name_format and not consistency_prefix:
-        logger.warning(
-            'Since version 1.1.16, if you provide `archive_name_format`, you should also'
-            ' specify `consistency.prefix`.'
-        )
 
-
-def parse_configuration(config_filename, schema_filename):
+def parse_configuration(config_filename, schema_filename, overrides=None, resolve_env=True):
     '''
-    Given the path to a config filename in YAML format and the path to a schema filename in
-    pykwalify YAML schema format, return the parsed configuration as a data structure of nested
-    dicts and lists corresponding to the schema. Example return value:
+    Given the path to a config filename in YAML format, the path to a schema filename in a YAML
+    rendition of JSON Schema format, a sequence of configuration file override strings in the form
+    of "option.suboption=value", return the parsed configuration as a data structure of nested dicts
+    and lists corresponding to the schema. Example return value:
 
-       {'location': {'source_directories': ['/home', '/etc'], 'repository': 'hostname.borg'},
-       'retention': {'keep_daily': 7}, 'consistency': {'checks': ['repository', 'archives']}}
+        {
+            'source_directories': ['/home', '/etc'],
+            'repository': 'hostname.borg',
+            'keep_daily': 7,
+            'checks': ['repository', 'archives'],
+        }
+
+    Also return a set of loaded configuration paths and a sequence of logging.LogRecord instances
+    containing any warnings about the configuration.
 
     Raise FileNotFoundError if the file does not exist, PermissionError if the user does not
     have permissions to read the file, or Validation_error if the config does not match the schema.
     '''
-    logging.getLogger('pykwalify').setLevel(logging.ERROR)
+    config_paths = set()
 
     try:
-        config = yaml.safe_load(open(config_filename))
-        schema = yaml.safe_load(open(schema_filename))
-    except yaml.error.YAMLError as error:
+        config = load.load_configuration(config_filename, config_paths)
+        schema = load.load_configuration(schema_filename)
+    except (ruamel.yaml.error.YAMLError, RecursionError) as error:
         raise Validation_error(config_filename, (str(error),))
 
-    # pykwalify gets angry if the example field is not a string. So rather than bend to its will,
-    # simply remove all examples before passing the schema to pykwalify.
-    for section_name, section_schema in schema['map'].items():
-        for field_name, field_schema in section_schema['map'].items():
-            field_schema.pop('example', None)
+    override.apply_overrides(config, schema, overrides)
+    constants.apply_constants(config, config.get('constants') if config else {})
 
-    validator = pykwalify.core.Core(source_data=config, schema_data=schema)
-    parsed_result = validator.validate(raise_exception=False)
+    if resolve_env:
+        environment.resolve_env_variables(config)
 
-    if validator.validation_errors:
-        raise Validation_error(config_filename, validator.validation_errors)
+    logs = normalize.normalize(config_filename, config)
 
-    apply_logical_validation(config_filename, parsed_result)
+    try:
+        validator = jsonschema.Draft7Validator(schema)
+    except AttributeError:  # pragma: no cover
+        validator = jsonschema.Draft4Validator(schema)
+    validation_errors = tuple(validator.iter_errors(config))
 
-    return parsed_result
+    if validation_errors:
+        raise Validation_error(
+            config_filename, tuple(format_json_error(error) for error in validation_errors)
+        )
+
+    apply_logical_validation(config_filename, config)
+
+    return config, config_paths, logs
+
+
+def normalize_repository_path(repository):
+    '''
+    Given a repository path, return the absolute path of it (for local repositories).
+    '''
+    # A colon in the repository could mean that it's either a file:// URL or a remote repository.
+    # If it's a remote repository, we don't want to normalize it. If it's a file:// URL, we do.
+    if ':' not in repository:
+        return os.path.abspath(repository)
+    elif repository.startswith('file://'):
+        return os.path.abspath(repository.partition('file://')[-1])
+    else:
+        return repository
+
+
+def repositories_match(first, second):
+    '''
+    Given two repository dicts with keys 'path' (relative and/or absolute),
+    and 'label', or two repository paths, return whether they match.
+    '''
+    if isinstance(first, str):
+        first = {'path': first, 'label': first}
+    if isinstance(second, str):
+        second = {'path': second, 'label': second}
+    return (first.get('label') == second.get('label')) or (
+        normalize_repository_path(first.get('path'))
+        == normalize_repository_path(second.get('path'))
+    )
+
+
+def guard_configuration_contains_repository(repository, configurations):
+    '''
+    Given a repository path and a dict mapping from config filename to corresponding parsed config
+    dict, ensure that the repository is declared at least once in all of the configurations. If no
+    repository is given, skip this check.
+
+    Raise ValueError if the repository is not found in any configurations.
+    '''
+    if not repository:
+        return
+
+    count = len(
+        tuple(
+            config_repository
+            for config in configurations.values()
+            for config_repository in config['repositories']
+            if repositories_match(config_repository, repository)
+        )
+    )
+
+    if count == 0:
+        raise ValueError(f'Repository "{repository}" not found in configuration files')
+
+
+def guard_single_repository_selected(repository, configurations):
+    '''
+    Given a repository path and a dict mapping from config filename to corresponding parsed config
+    dict, ensure either a single repository exists across all configuration files or a repository
+    path was given.
+    '''
+    if repository:
+        return
+
+    count = len(
+        tuple(
+            config_repository
+            for config in configurations.values()
+            for config_repository in config['repositories']
+        )
+    )
+
+    if count != 1:
+        raise ValueError(
+            "Can't determine which repository to use. Use --repository to disambiguate"
+        )

borgmatic/execute.py

@@ -0,0 +1,433 @@
+import collections
+import enum
+import logging
+import os
+import select
+import subprocess
+import textwrap
+
+logger = logging.getLogger(__name__)
+
+
+ERROR_OUTPUT_MAX_LINE_COUNT = 25
+BORG_ERROR_EXIT_CODE_START = 2
+BORG_ERROR_EXIT_CODE_END = 99
+
+
+class Exit_status(enum.Enum):
+    STILL_RUNNING = 1
+    SUCCESS = 2
+    WARNING = 3
+    ERROR = 4
+
+
+def interpret_exit_code(command, exit_code, borg_local_path=None, borg_exit_codes=None):
+    '''
+    Return an Exit_status value (e.g. SUCCESS, ERROR, or WARNING) based on interpreting the given
+    exit code. If a Borg local path is given and matches the process' command, then interpret the
+    exit code based on Borg's documented exit code semantics. And if Borg exit codes are given as a
+    sequence of exit code configuration dicts, then take those configured preferences into account.
+    '''
+    if exit_code is None:
+        return Exit_status.STILL_RUNNING
+    if exit_code == 0:
+        return Exit_status.SUCCESS
+
+    if borg_local_path and command[0] == borg_local_path:
+        # First try looking for the exit code in the borg_exit_codes configuration.
+        for entry in borg_exit_codes or ():
+            if entry.get('code') == exit_code:
+                treat_as = entry.get('treat_as')
+
+                if treat_as == 'error':
+                    logger.error(
+                        f'Treating exit code {exit_code} as an error, as per configuration'
+                    )
+                    return Exit_status.ERROR
+                elif treat_as == 'warning':
+                    logger.warning(
+                        f'Treating exit code {exit_code} as a warning, as per configuration'
+                    )
+                    return Exit_status.WARNING
+
+        # If the exit code doesn't have explicit configuration, then fall back to the default Borg
+        # behavior.
+        return (
+            Exit_status.ERROR
+            if (
+                exit_code < 0
+                or (
+                    exit_code >= BORG_ERROR_EXIT_CODE_START
+                    and exit_code <= BORG_ERROR_EXIT_CODE_END
+                )
+            )
+            else Exit_status.WARNING
+        )
+
+    return Exit_status.ERROR
+
+
+def command_for_process(process):
+    '''
+    Given a process as an instance of subprocess.Popen, return the command string that was used to
+    invoke it.
+    '''
+    return process.args if isinstance(process.args, str) else ' '.join(process.args)
+
+
+def output_buffer_for_process(process, exclude_stdouts):
+    '''
+    Given a process as an instance of subprocess.Popen and a sequence of stdouts to exclude, return
+    either the process's stdout or stderr. The idea is that if stdout is excluded for a process, we
+    still have stderr to log.
+    '''
+    return process.stderr if process.stdout in exclude_stdouts else process.stdout
+
+
+def append_last_lines(last_lines, captured_output, line, output_log_level):
+    '''
+    Given a rolling list of last lines, a list of captured output, a line to append, and an output
+    log level, append the line to the last lines and (if necessary) the captured output. Then log
+    the line at the requested output log level.
+    '''
+    last_lines.append(line)
+
+    if len(last_lines) > ERROR_OUTPUT_MAX_LINE_COUNT:
+        last_lines.pop(0)
+
+    if output_log_level is None:
+        captured_output.append(line)
+    else:
+        logger.log(output_log_level, line)
+
+
+def log_outputs(processes, exclude_stdouts, output_log_level, borg_local_path, borg_exit_codes):
+    '''
+    Given a sequence of subprocess.Popen() instances for multiple processes, log the output for each
+    process with the requested log level. Additionally, raise a CalledProcessError if a process
+    exits with an error (or a warning for exit code 1, if that process does not match the Borg local
+    path).
+
+    If output log level is None, then instead of logging, capture output for each process and return
+    it as a dict from the process to its output. Use the given Borg local path and exit code
+    configuration to decide what's an error and what's a warning.
+
+    For simplicity, it's assumed that the output buffer for each process is its stdout. But if any
+    stdouts are given to exclude, then for any matching processes, log from their stderr instead.
+
+    Note that stdout for a process can be None if output is intentionally not captured. In which
+    case it won't be logged.
+    '''
+    # Map from output buffer to sequence of last lines.
+    buffer_last_lines = collections.defaultdict(list)
+    process_for_output_buffer = {
+        output_buffer_for_process(process, exclude_stdouts): process
+        for process in processes
+        if process.stdout or process.stderr
+    }
+    output_buffers = list(process_for_output_buffer.keys())
+    captured_outputs = collections.defaultdict(list)
+    still_running = True
+
+    # Log output for each process until they all exit.
+    while True:
+        if output_buffers:
+            (ready_buffers, _, _) = select.select(output_buffers, [], [])
+
+            for ready_buffer in ready_buffers:
+                ready_process = process_for_output_buffer.get(ready_buffer)
+
+                # The "ready" process has exited, but it might be a pipe destination with other
+                # processes (pipe sources) waiting to be read from. So as a measure to prevent
+                # hangs, vent all processes when one exits.
+                if ready_process and ready_process.poll() is not None:
+                    for other_process in processes:
+                        if (
+                            other_process.poll() is None
+                            and other_process.stdout
+                            and other_process.stdout not in output_buffers
+                        ):
+                            # Add the process's output to output_buffers to ensure it'll get read.
+                            output_buffers.append(other_process.stdout)
+
+                while True:
+                    line = ready_buffer.readline().rstrip().decode()
+                    if not line or not ready_process:
+                        break
+
+                    # Keep the last few lines of output in case the process errors, and we need the output for
+                    # the exception below.
+                    append_last_lines(
+                        buffer_last_lines[ready_buffer],
+                        captured_outputs[ready_process],
+                        line,
+                        output_log_level,
+                    )
+
+        if not still_running:
+            break
+
+        still_running = False
+
+        for process in processes:
+            exit_code = process.poll() if output_buffers else process.wait()
+
+            if exit_code is None:
+                still_running = True
+                command = process.args.split(' ') if isinstance(process.args, str) else process.args
+                continue
+
+            command = process.args.split(' ') if isinstance(process.args, str) else process.args
+            exit_status = interpret_exit_code(command, exit_code, borg_local_path, borg_exit_codes)
+
+            if exit_status in (Exit_status.ERROR, Exit_status.WARNING):
+                # If an error occurs, include its output in the raised exception so that we don't
+                # inadvertently hide error output.
+                output_buffer = output_buffer_for_process(process, exclude_stdouts)
+                last_lines = buffer_last_lines[output_buffer] if output_buffer else []
+
+                # Collect any straggling output lines that came in since we last gathered output.
+                while output_buffer:  # pragma: no cover
+                    line = output_buffer.readline().rstrip().decode()
+                    if not line:
+                        break
+
+                    append_last_lines(
+                        last_lines, captured_outputs[process], line, output_log_level=logging.ERROR
+                    )
+
+                if len(last_lines) == ERROR_OUTPUT_MAX_LINE_COUNT:
+                    last_lines.insert(0, '...')
+
+                # Something has gone wrong. So vent each process' output buffer to prevent it from
+                # hanging. And then kill the process.
+                for other_process in processes:
+                    if other_process.poll() is None:
+                        other_process.stdout.read(0)
+                        other_process.kill()
+
+                if exit_status == Exit_status.ERROR:
+                    raise subprocess.CalledProcessError(
+                        exit_code, command_for_process(process), '\n'.join(last_lines)
+                    )
+
+                still_running = False
+                break
+
+    if captured_outputs:
+        return {
+            process: '\n'.join(output_lines) for process, output_lines in captured_outputs.items()
+        }
+
+
+SECRET_COMMAND_FLAG_NAMES = {'--password'}
+
+
+def mask_command_secrets(full_command):
+    '''
+    Given a command as a sequence, mask secret values for flags like "--password" in preparation for
+    logging.
+    '''
+    masked_command = []
+    previous_piece = None
+
+    for piece in full_command:
+        masked_command.append('***' if previous_piece in SECRET_COMMAND_FLAG_NAMES else piece)
+        previous_piece = piece
+
+    return tuple(masked_command)
+
+
+MAX_LOGGED_COMMAND_LENGTH = 1000
+
+
+def log_command(full_command, input_file=None, output_file=None, environment=None):
+    '''
+    Log the given command (a sequence of command/argument strings), along with its input/output file
+    paths and extra environment variables (with omitted values in case they contain passwords).
+    '''
+    logger.debug(
+        textwrap.shorten(
+            ' '.join(
+                tuple(f'{key}=***' for key in (environment or {}).keys())
+                + mask_command_secrets(full_command)
+            ),
+            width=MAX_LOGGED_COMMAND_LENGTH,
+            placeholder=' ...',
+        )
+        + (f" < {getattr(input_file, 'name', '')}" if input_file else '')
+        + (f" > {getattr(output_file, 'name', '')}" if output_file else '')
+    )
+
+
+# A sentinel passed as an output file to execute_command() to indicate that the command's output
+# should be allowed to flow through to stdout without being captured for logging. Useful for
+# commands with interactive prompts or those that mess directly with the console.
+DO_NOT_CAPTURE = object()
+
+
+def execute_command(
+    full_command,
+    output_log_level=logging.INFO,
+    output_file=None,
+    input_file=None,
+    shell=False,
+    extra_environment=None,
+    working_directory=None,
+    borg_local_path=None,
+    borg_exit_codes=None,
+    run_to_completion=True,
+):
+    '''
+    Execute the given command (a sequence of command/argument strings) and log its output at the
+    given log level. If an open output file object is given, then write stdout to the file and only
+    log stderr. If an open input file object is given, then read stdin from the file. If shell is
+    True, execute the command within a shell. If an extra environment dict is given, then use it to
+    augment the current environment, and pass the result into the command. If a working directory is
+    given, use that as the present working directory when running the command. If a Borg local path
+    is given, and the command matches it (regardless of arguments), treat exit code 1 as a warning
+    instead of an error. But if Borg exit codes are given as a sequence of exit code configuration
+    dicts, then use that configuration to decide what's an error and what's a warning. If run to
+    completion is False, then return the process for the command without executing it to completion.
+
+    Raise subprocesses.CalledProcessError if an error occurs while running the command.
+    '''
+    log_command(full_command, input_file, output_file, extra_environment)
+    environment = {**os.environ, **extra_environment} if extra_environment else None
+    do_not_capture = bool(output_file is DO_NOT_CAPTURE)
+    command = ' '.join(full_command) if shell else full_command
+
+    process = subprocess.Popen(
+        command,
+        stdin=input_file,
+        stdout=None if do_not_capture else (output_file or subprocess.PIPE),
+        stderr=None if do_not_capture else (subprocess.PIPE if output_file else subprocess.STDOUT),
+        shell=shell,
+        env=environment,
+        cwd=working_directory,
+    )
+    if not run_to_completion:
+        return process
+
+    log_outputs(
+        (process,),
+        (input_file, output_file),
+        output_log_level,
+        borg_local_path,
+        borg_exit_codes,
+    )
+
+
+def execute_command_and_capture_output(
+    full_command,
+    capture_stderr=False,
+    shell=False,
+    extra_environment=None,
+    working_directory=None,
+    borg_local_path=None,
+    borg_exit_codes=None,
+):
+    '''
+    Execute the given command (a sequence of command/argument strings), capturing and returning its
+    output (stdout). If capture stderr is True, then capture and return stderr in addition to
+    stdout. If shell is True, execute the command within a shell. If an extra environment dict is
+    given, then use it to augment the current environment, and pass the result into the command. If
+    a working directory is given, use that as the present working directory when running the
+    command. If a Borg local path is given, and the command matches it (regardless of arguments),
+    treat exit code 1 as a warning instead of an error. But if Borg exit codes are given as a
+    sequence of exit code configuration dicts, then use that configuration to decide what's an error
+    and what's a warning.
+
+    Raise subprocesses.CalledProcessError if an error occurs while running the command.
+    '''
+    log_command(full_command, environment=extra_environment)
+    environment = {**os.environ, **extra_environment} if extra_environment else None
+    command = ' '.join(full_command) if shell else full_command
+
+    try:
+        output = subprocess.check_output(
+            command,
+            stderr=subprocess.STDOUT if capture_stderr else None,
+            shell=shell,
+            env=environment,
+            cwd=working_directory,
+        )
+    except subprocess.CalledProcessError as error:
+        if (
+            interpret_exit_code(command, error.returncode, borg_local_path, borg_exit_codes)
+            == Exit_status.ERROR
+        ):
+            raise
+        output = error.output
+
+    return output.decode() if output is not None else None
+
+
+def execute_command_with_processes(
+    full_command,
+    processes,
+    output_log_level=logging.INFO,
+    output_file=None,
+    input_file=None,
+    shell=False,
+    extra_environment=None,
+    working_directory=None,
+    borg_local_path=None,
+    borg_exit_codes=None,
+):
+    '''
+    Execute the given command (a sequence of command/argument strings) and log its output at the
+    given log level. Simultaneously, continue to poll one or more active processes so that they
+    run as well. This is useful, for instance, for processes that are streaming output to a named
+    pipe that the given command is consuming from.
+
+    If an open output file object is given, then write stdout to the file and only log stderr. But
+    if output log level is None, instead suppress logging and return the captured output for (only)
+    the given command. If an open input file object is given, then read stdin from the file. If
+    shell is True, execute the command within a shell. If an extra environment dict is given, then
+    use it to augment the current environment, and pass the result into the command. If a working
+    directory is given, use that as the present working directory when running the command. If a
+    Borg local path is given, then for any matching command or process (regardless of arguments),
+    treat exit code 1 as a warning instead of an error. But if Borg exit codes are given as a
+    sequence of exit code configuration dicts, then use that configuration to decide what's an error
+    and what's a warning.
+
+    Raise subprocesses.CalledProcessError if an error occurs while running the command or in the
+    upstream process.
+    '''
+    log_command(full_command, input_file, output_file, extra_environment)
+    environment = {**os.environ, **extra_environment} if extra_environment else None
+    do_not_capture = bool(output_file is DO_NOT_CAPTURE)
+    command = ' '.join(full_command) if shell else full_command
+
+    try:
+        command_process = subprocess.Popen(
+            command,
+            stdin=input_file,
+            stdout=None if do_not_capture else (output_file or subprocess.PIPE),
+            stderr=(
+                None if do_not_capture else (subprocess.PIPE if output_file else subprocess.STDOUT)
+            ),
+            shell=shell,
+            env=environment,
+            cwd=working_directory,
+        )
+    except (subprocess.CalledProcessError, OSError):
+        # Something has gone wrong. So vent each process' output buffer to prevent it from hanging.
+        # And then kill the process.
+        for process in processes:
+            if process.poll() is None:
+                process.stdout.read(0)
+                process.kill()
+        raise
+
+    captured_outputs = log_outputs(
+        tuple(processes) + (command_process,),
+        (input_file, output_file),
+        output_log_level,
+        borg_local_path,
+        borg_exit_codes,
+    )
+
+    if output_log_level is None:
+        return captured_outputs.get(command_process)

borgmatic/hooks/apprise.py

@@ -0,0 +1,109 @@
+import logging
+import operator
+
+import borgmatic.hooks.logs
+import borgmatic.hooks.monitor
+
+logger = logging.getLogger(__name__)
+
+
+DEFAULT_LOGS_SIZE_LIMIT_BYTES = 100000
+HANDLER_IDENTIFIER = 'apprise'
+
+
+def initialize_monitor(hook_config, config, config_filename, monitoring_log_level, dry_run):
+    '''
+    Add a handler to the root logger that stores in memory the most recent logs emitted. That way,
+    we can send them all to an Apprise notification service upon a finish or failure state. But skip
+    this if the "send_logs" option is false.
+    '''
+    if hook_config.get('send_logs') is False:
+        return
+
+    logs_size_limit = max(
+        hook_config.get('logs_size_limit', DEFAULT_LOGS_SIZE_LIMIT_BYTES)
+        - len(borgmatic.hooks.logs.PAYLOAD_TRUNCATION_INDICATOR),
+        0,
+    )
+
+    borgmatic.hooks.logs.add_handler(
+        borgmatic.hooks.logs.Forgetful_buffering_handler(
+            HANDLER_IDENTIFIER, logs_size_limit, monitoring_log_level
+        )
+    )
+
+
+def ping_monitor(hook_config, config, config_filename, state, monitoring_log_level, dry_run):
+    '''
+    Ping the configured Apprise service URLs. Use the given configuration filename in any log
+    entries. If this is a dry run, then don't actually ping anything.
+    '''
+    try:
+        import apprise
+        from apprise import NotifyFormat, NotifyType
+    except ImportError:  # pragma: no cover
+        logger.warning('Unable to import Apprise in monitoring hook')
+        return
+
+    state_to_notify_type = {
+        'start': NotifyType.INFO,
+        'finish': NotifyType.SUCCESS,
+        'fail': NotifyType.FAILURE,
+        'log': NotifyType.INFO,
+    }
+
+    run_states = hook_config.get('states', ['fail'])
+
+    if state.name.lower() not in run_states:
+        return
+
+    state_config = hook_config.get(
+        state.name.lower(),
+        {
+            'title': f'A borgmatic {state.name} event happened',
+            'body': f'A borgmatic {state.name} event happened',
+        },
+    )
+
+    if not hook_config.get('services'):
+        logger.info(f'{config_filename}: No Apprise services to ping')
+        return
+
+    dry_run_string = ' (dry run; not actually pinging)' if dry_run else ''
+    labels_string = ', '.join(map(operator.itemgetter('label'), hook_config.get('services')))
+    logger.info(f'{config_filename}: Pinging Apprise services: {labels_string}{dry_run_string}')
+
+    apprise_object = apprise.Apprise()
+    apprise_object.add(list(map(operator.itemgetter('url'), hook_config.get('services'))))
+
+    if dry_run:
+        return
+
+    body = state_config.get('body')
+
+    if state in (
+        borgmatic.hooks.monitor.State.FINISH,
+        borgmatic.hooks.monitor.State.FAIL,
+        borgmatic.hooks.monitor.State.LOG,
+    ):
+        formatted_logs = borgmatic.hooks.logs.format_buffered_logs_for_payload(HANDLER_IDENTIFIER)
+        if formatted_logs:
+            body += f'\n\n{formatted_logs}'
+
+    result = apprise_object.notify(
+        title=state_config.get('title', ''),
+        body=body,
+        body_format=NotifyFormat.TEXT,
+        notify_type=state_to_notify_type[state.name.lower()],
+    )
+
+    if result is False:
+        logger.warning(f'{config_filename}: Error sending some Apprise notifications')
+
+
+def destroy_monitor(hook_config, config, config_filename, monitoring_log_level, dry_run):
+    '''
+    Remove the monitor handler that was added to the root logger. This prevents the handler from
+    getting reused by other instances of this monitor.
+    '''
+    borgmatic.hooks.logs.remove_handler(HANDLER_IDENTIFIER)

borgmatic/hooks/command.py

@@ -0,0 +1,97 @@
+import logging
+import os
+import re
+import shlex
+
+from borgmatic import execute
+
+logger = logging.getLogger(__name__)
+
+
+SOFT_FAIL_EXIT_CODE = 75
+
+
+def interpolate_context(config_filename, hook_description, command, context):
+    '''
+    Given a config filename, a hook description, a single hook command, and a dict of context
+    names/values, interpolate the values by "{name}" into the command and return the result.
+    '''
+    for name, value in context.items():
+        command = command.replace(f'{{{name}}}', shlex.quote(str(value)))
+
+    for unsupported_variable in re.findall(r'{\w+}', command):
+        logger.warning(
+            f"{config_filename}: Variable '{unsupported_variable}' is not supported in {hook_description} hook"
+        )
+
+    return command
+
+
+def execute_hook(commands, umask, config_filename, description, dry_run, **context):
+    '''
+    Given a list of hook commands to execute, a umask to execute with (or None), a config filename,
+    a hook description, and whether this is a dry run, run the given commands. Or, don't run them
+    if this is a dry run.
+
+    The context contains optional values interpolated by name into the hook commands.
+
+    Raise ValueError if the umask cannot be parsed.
+    Raise subprocesses.CalledProcessError if an error occurs in a hook.
+    '''
+    if not commands:
+        logger.debug(f'{config_filename}: No commands to run for {description} hook')
+        return
+
+    dry_run_label = ' (dry run; not actually running hooks)' if dry_run else ''
+
+    context['configuration_filename'] = config_filename
+    commands = [
+        interpolate_context(config_filename, description, command, context) for command in commands
+    ]
+
+    if len(commands) == 1:
+        logger.info(f'{config_filename}: Running command for {description} hook{dry_run_label}')
+    else:
+        logger.info(
+            f'{config_filename}: Running {len(commands)} commands for {description} hook{dry_run_label}',
+        )
+
+    if umask:
+        parsed_umask = int(str(umask), 8)
+        logger.debug(f'{config_filename}: Set hook umask to {oct(parsed_umask)}')
+        original_umask = os.umask(parsed_umask)
+    else:
+        original_umask = None
+
+    try:
+        for command in commands:
+            if not dry_run:
+                execute.execute_command(
+                    [command],
+                    output_log_level=(
+                        logging.ERROR if description == 'on-error' else logging.WARNING
+                    ),
+                    shell=True,
+                )
+    finally:
+        if original_umask:
+            os.umask(original_umask)
+
+
+def considered_soft_failure(config_filename, error):
+    '''
+    Given a configuration filename and an exception object, return whether the exception object
+    represents a subprocess.CalledProcessError with a return code of SOFT_FAIL_EXIT_CODE. If so,
+    that indicates that the error is a "soft failure", and should not result in an error.
+    '''
+    exit_code = getattr(error, 'returncode', None)
+    if exit_code is None:
+        return False
+
+    if exit_code == SOFT_FAIL_EXIT_CODE:
+        logger.info(
+            f'{config_filename}: Command hook exited with soft failure exit code ({SOFT_FAIL_EXIT_CODE}); skipping remaining actions',
+        )
+        return True
+
+    return False

borgmatic/hooks/cronhub.py

@@ -0,0 +1,63 @@
+import logging
+
+import requests
+
+from borgmatic.hooks import monitor
+
+logger = logging.getLogger(__name__)
+
+MONITOR_STATE_TO_CRONHUB = {
+    monitor.State.START: 'start',
+    monitor.State.FINISH: 'finish',
+    monitor.State.FAIL: 'fail',
+}
+
+
+def initialize_monitor(
+    ping_url, config, config_filename, monitoring_log_level, dry_run
+):  # pragma: no cover
+    '''
+    No initialization is necessary for this monitor.
+    '''
+    pass
+
+
+def ping_monitor(hook_config, config, config_filename, state, monitoring_log_level, dry_run):
+    '''
+    Ping the configured Cronhub URL, modified with the monitor.State. Use the given configuration
+    filename in any log entries. If this is a dry run, then don't actually ping anything.
+    '''
+    if state not in MONITOR_STATE_TO_CRONHUB:
+        logger.debug(
+            f'{config_filename}: Ignoring unsupported monitoring {state.name.lower()} in Cronhub hook'
+        )
+        return
+
+    dry_run_label = ' (dry run; not actually pinging)' if dry_run else ''
+    formatted_state = f'/{MONITOR_STATE_TO_CRONHUB[state]}/'
+    ping_url = (
+        hook_config['ping_url']
+        .replace('/start/', formatted_state)
+        .replace('/ping/', formatted_state)
+    )
+
+    logger.info(f'{config_filename}: Pinging Cronhub {state.name.lower()}{dry_run_label}')
+    logger.debug(f'{config_filename}: Using Cronhub ping URL {ping_url}')
+
+    if not dry_run:
+        logging.getLogger('urllib3').setLevel(logging.ERROR)
+        try:
+            response = requests.get(ping_url)
+            if not response.ok:
+                response.raise_for_status()
+        except requests.exceptions.RequestException as error:
+            logger.warning(f'{config_filename}: Cronhub error: {error}')
+
+
+def destroy_monitor(
+    ping_url_or_uuid, config, config_filename, monitoring_log_level, dry_run
+):  # pragma: no cover
+    '''
+    No destruction is necessary for this monitor.
+    '''
+    pass

borgmatic/hooks/cronitor.py

@@ -0,0 +1,58 @@
+import logging
+
+import requests
+
+from borgmatic.hooks import monitor
+
+logger = logging.getLogger(__name__)
+
+MONITOR_STATE_TO_CRONITOR = {
+    monitor.State.START: 'run',
+    monitor.State.FINISH: 'complete',
+    monitor.State.FAIL: 'fail',
+}
+
+
+def initialize_monitor(
+    ping_url, config, config_filename, monitoring_log_level, dry_run
+):  # pragma: no cover
+    '''
+    No initialization is necessary for this monitor.
+    '''
+    pass
+
+
+def ping_monitor(hook_config, config, config_filename, state, monitoring_log_level, dry_run):
+    '''
+    Ping the configured Cronitor URL, modified with the monitor.State. Use the given configuration
+    filename in any log entries. If this is a dry run, then don't actually ping anything.
+    '''
+    if state not in MONITOR_STATE_TO_CRONITOR:
+        logger.debug(
+            f'{config_filename}: Ignoring unsupported monitoring {state.name.lower()} in Cronitor hook'
+        )
+        return
+
+    dry_run_label = ' (dry run; not actually pinging)' if dry_run else ''
+    ping_url = f"{hook_config['ping_url']}/{MONITOR_STATE_TO_CRONITOR[state]}"
+
+    logger.info(f'{config_filename}: Pinging Cronitor {state.name.lower()}{dry_run_label}')
+    logger.debug(f'{config_filename}: Using Cronitor ping URL {ping_url}')
+
+    if not dry_run:
+        logging.getLogger('urllib3').setLevel(logging.ERROR)
+        try:
+            response = requests.get(ping_url)
+            if not response.ok:
+                response.raise_for_status()
+        except requests.exceptions.RequestException as error:
+            logger.warning(f'{config_filename}: Cronitor error: {error}')
+
+
+def destroy_monitor(
+    ping_url_or_uuid, config, config_filename, monitoring_log_level, dry_run
+):  # pragma: no cover
+    '''
+    No destruction is necessary for this monitor.
+    '''
+    pass

borgmatic/hooks/dispatch.py

@@ -0,0 +1,91 @@
+import logging
+
+from borgmatic.hooks import (
+    apprise,
+    cronhub,
+    cronitor,
+    healthchecks,
+    loki,
+    mariadb,
+    mongodb,
+    mysql,
+    ntfy,
+    pagerduty,
+    postgresql,
+    sqlite,
+)
+
+logger = logging.getLogger(__name__)
+
+HOOK_NAME_TO_MODULE = {
+    'apprise': apprise,
+    'cronhub': cronhub,
+    'cronitor': cronitor,
+    'healthchecks': healthchecks,
+    'mariadb_databases': mariadb,
+    'mongodb_databases': mongodb,
+    'mysql_databases': mysql,
+    'ntfy': ntfy,
+    'pagerduty': pagerduty,
+    'postgresql_databases': postgresql,
+    'sqlite_databases': sqlite,
+    'loki': loki,
+}
+
+
+def call_hook(function_name, config, log_prefix, hook_name, *args, **kwargs):
+    '''
+    Given a configuration dict and a prefix to use in log entries, call the requested function of
+    the Python module corresponding to the given hook name. Supply that call with the configuration
+    for this hook (if any), the log prefix, and any given args and kwargs. Return any return value.
+
+    Raise ValueError if the hook name is unknown.
+    Raise AttributeError if the function name is not found in the module.
+    Raise anything else that the called function raises.
+    '''
+    hook_config = config.get(hook_name, {})
+
+    try:
+        module = HOOK_NAME_TO_MODULE[hook_name]
+    except KeyError:
+        raise ValueError(f'Unknown hook name: {hook_name}')
+
+    logger.debug(f'{log_prefix}: Calling {hook_name} hook function {function_name}')
+    return getattr(module, function_name)(hook_config, config, log_prefix, *args, **kwargs)
+
+
+def call_hooks(function_name, config, log_prefix, hook_names, *args, **kwargs):
+    '''
+    Given a configuration dict and a prefix to use in log entries, call the requested function of
+    the Python module corresponding to each given hook name. Supply each call with the configuration
+    for that hook, the log prefix, and any given args and kwargs. Collect any return values into a
+    dict from hook name to return value.
+
+    If the hook name is not present in the hooks configuration, then don't call the function for it
+    and omit it from the return values.
+
+    Raise ValueError if the hook name is unknown.
+    Raise AttributeError if the function name is not found in the module.
+    Raise anything else that a called function raises. An error stops calls to subsequent functions.
+    '''
+    return {
+        hook_name: call_hook(function_name, config, log_prefix, hook_name, *args, **kwargs)
+        for hook_name in hook_names
+        if config.get(hook_name)
+    }
+
+
+def call_hooks_even_if_unconfigured(function_name, config, log_prefix, hook_names, *args, **kwargs):
+    '''
+    Given a configuration dict and a prefix to use in log entries, call the requested function of
+    the Python module corresponding to each given hook name. Supply each call with the configuration
+    for that hook, the log prefix, and any given args and kwargs. Collect any return values into a
+    dict from hook name to return value.
+
+    Raise AttributeError if the function name is not found in the module.
+    Raise anything else that a called function raises. An error stops calls to subsequent functions.
+    '''
+    return {
+        hook_name: call_hook(function_name, config, log_prefix, hook_name, *args, **kwargs)
+        for hook_name in hook_names
+    }

borgmatic/hooks/dump.py

@@ -0,0 +1,80 @@
+import logging
+import os
+import shutil
+
+from borgmatic.borg.state import DEFAULT_BORGMATIC_SOURCE_DIRECTORY
+
+logger = logging.getLogger(__name__)
+
+DATA_SOURCE_HOOK_NAMES = (
+    'mariadb_databases',
+    'mysql_databases',
+    'mongodb_databases',
+    'postgresql_databases',
+    'sqlite_databases',
+)
+
+
+def make_data_source_dump_path(borgmatic_source_directory, data_source_hook_name):
+    '''
+    Given a borgmatic source directory (or None) and a data source hook name, construct a data
+    source dump path.
+    '''
+    if not borgmatic_source_directory:
+        borgmatic_source_directory = DEFAULT_BORGMATIC_SOURCE_DIRECTORY
+
+    return os.path.join(borgmatic_source_directory, data_source_hook_name)
+
+
+def make_data_source_dump_filename(dump_path, name, hostname=None):
+    '''
+    Based on the given dump directory path, data source name, and hostname, return a filename to use
+    for the data source dump. The hostname defaults to localhost.
+
+    Raise ValueError if the data source name is invalid.
+    '''
+    if os.path.sep in name:
+        raise ValueError(f'Invalid data source name {name}')
+
+    return os.path.join(os.path.expanduser(dump_path), hostname or 'localhost', name)
+
+
+def create_parent_directory_for_dump(dump_path):
+    '''
+    Create a directory to contain the given dump path.
+    '''
+    os.makedirs(os.path.dirname(dump_path), mode=0o700, exist_ok=True)
+
+
+def create_named_pipe_for_dump(dump_path):
+    '''
+    Create a named pipe at the given dump path.
+    '''
+    create_parent_directory_for_dump(dump_path)
+    os.mkfifo(dump_path, mode=0o600)
+
+
+def remove_data_source_dumps(dump_path, data_source_type_name, log_prefix, dry_run):
+    '''
+    Remove all data source dumps in the given dump directory path (including the directory itself).
+    If this is a dry run, then don't actually remove anything.
+    '''
+    dry_run_label = ' (dry run; not actually removing anything)' if dry_run else ''
+
+    logger.debug(f'{log_prefix}: Removing {data_source_type_name} data source dumps{dry_run_label}')
+
+    expanded_path = os.path.expanduser(dump_path)
+
+    if dry_run:
+        return
+
+    if os.path.exists(expanded_path):
+        shutil.rmtree(expanded_path)
+
+
+def convert_glob_patterns_to_borg_patterns(patterns):
+    '''
+    Convert a sequence of shell glob patterns like "/etc/*" to the corresponding Borg archive
+    patterns like "sh:etc/*".
+    '''
+    return [f'sh:{pattern.lstrip(os.path.sep)}' for pattern in patterns]

borgmatic/hooks/healthchecks.py

@@ -0,0 +1,102 @@
+import logging
+import re
+
+import requests
+
+import borgmatic.hooks.logs
+from borgmatic.hooks import monitor
+
+logger = logging.getLogger(__name__)
+
+MONITOR_STATE_TO_HEALTHCHECKS = {
+    monitor.State.START: 'start',
+    monitor.State.FINISH: None,  # Healthchecks doesn't append to the URL for the finished state.
+    monitor.State.FAIL: 'fail',
+    monitor.State.LOG: 'log',
+}
+
+DEFAULT_PING_BODY_LIMIT_BYTES = 1500
+HANDLER_IDENTIFIER = 'healthchecks'
+
+
+def initialize_monitor(hook_config, config, config_filename, monitoring_log_level, dry_run):
+    '''
+    Add a handler to the root logger that stores in memory the most recent logs emitted. That way,
+    we can send them all to Healthchecks upon a finish or failure state. But skip this if the
+    "send_logs" option is false.
+    '''
+    if hook_config.get('send_logs') is False:
+        return
+
+    ping_body_limit = max(
+        hook_config.get('ping_body_limit', DEFAULT_PING_BODY_LIMIT_BYTES)
+        - len(borgmatic.hooks.logs.PAYLOAD_TRUNCATION_INDICATOR),
+        0,
+    )
+
+    borgmatic.hooks.logs.add_handler(
+        borgmatic.hooks.logs.Forgetful_buffering_handler(
+            HANDLER_IDENTIFIER, ping_body_limit, monitoring_log_level
+        )
+    )
+
+
+def ping_monitor(hook_config, config, config_filename, state, monitoring_log_level, dry_run):
+    '''
+    Ping the configured Healthchecks URL or UUID, modified with the monitor.State. Use the given
+    configuration filename in any log entries, and log to Healthchecks with the giving log level.
+    If this is a dry run, then don't actually ping anything.
+    '''
+    ping_url = (
+        hook_config['ping_url']
+        if hook_config['ping_url'].startswith('http')
+        else f"https://hc-ping.com/{hook_config['ping_url']}"
+    )
+    dry_run_label = ' (dry run; not actually pinging)' if dry_run else ''
+
+    if 'states' in hook_config and state.name.lower() not in hook_config['states']:
+        logger.info(
+            f'{config_filename}: Skipping Healthchecks {state.name.lower()} ping due to configured states'
+        )
+        return
+
+    ping_url_is_uuid = re.search(r'\w{8}-\w{4}-\w{4}-\w{4}-\w{12}$', ping_url)
+
+    healthchecks_state = MONITOR_STATE_TO_HEALTHCHECKS.get(state)
+    if healthchecks_state:
+        ping_url = f'{ping_url}/{healthchecks_state}'
+
+    if hook_config.get('create_slug'):
+        if ping_url_is_uuid:
+            logger.warning(
+                f'{config_filename}: Healthchecks UUIDs do not support auto provisionning; ignoring'
+            )
+        else:
+            ping_url = f'{ping_url}?create=1'
+
+    logger.info(f'{config_filename}: Pinging Healthchecks {state.name.lower()}{dry_run_label}')
+    logger.debug(f'{config_filename}: Using Healthchecks ping URL {ping_url}')
+
+    if state in (monitor.State.FINISH, monitor.State.FAIL, monitor.State.LOG):
+        payload = borgmatic.hooks.logs.format_buffered_logs_for_payload(HANDLER_IDENTIFIER)
+    else:
+        payload = ''
+
+    if not dry_run:
+        logging.getLogger('urllib3').setLevel(logging.ERROR)
+        try:
+            response = requests.post(
+                ping_url, data=payload.encode('utf-8'), verify=hook_config.get('verify_tls', True)
+            )
+            if not response.ok:
+                response.raise_for_status()
+        except requests.exceptions.RequestException as error:
+            logger.warning(f'{config_filename}: Healthchecks error: {error}')
+
+
+def destroy_monitor(hook_config, config, config_filename, monitoring_log_level, dry_run):
+    '''
+    Remove the monitor handler that was added to the root logger. This prevents the handler from
+    getting reused by other instances of this monitor.
+    '''
+    borgmatic.hooks.logs.remove_handler(HANDLER_IDENTIFIER)

borgmatic/hooks/logs.py

@@ -0,0 +1,91 @@
+import logging
+
+PAYLOAD_TRUNCATION_INDICATOR = '...\n'
+
+
+class Forgetful_buffering_handler(logging.Handler):
+    '''
+    A buffering log handler that stores log messages in memory, and throws away messages (oldest
+    first) once a particular capacity in bytes is reached. But if the given byte capacity is zero,
+    don't throw away any messages.
+
+    The given identifier is used to distinguish the instance of this handler used for one monitoring
+    hook from those instances used for other monitoring hooks.
+    '''
+
+    def __init__(self, identifier, byte_capacity, log_level):
+        super().__init__()
+
+        self.identifier = identifier
+        self.byte_capacity = byte_capacity
+        self.byte_count = 0
+        self.buffer = []
+        self.forgot = False
+        self.setLevel(log_level)
+
+    def emit(self, record):
+        message = record.getMessage() + '\n'
+        self.byte_count += len(message)
+        self.buffer.append(message)
+
+        if not self.byte_capacity:
+            return
+
+        while self.byte_count > self.byte_capacity and self.buffer:
+            self.byte_count -= len(self.buffer[0])
+            self.buffer.pop(0)
+            self.forgot = True
+
+
+def add_handler(handler):  # pragma: no cover
+    '''
+    Add the given handler to the global logger.
+    '''
+    logging.getLogger().addHandler(handler)
+
+
+def get_handler(identifier):
+    '''
+    Given the identifier for an existing Forgetful_buffering_handler instance, return the handler.
+
+    Raise ValueError if the handler isn't found.
+    '''
+    try:
+        return next(
+            handler
+            for handler in logging.getLogger().handlers
+            if isinstance(handler, Forgetful_buffering_handler) and handler.identifier == identifier
+        )
+    except StopIteration:
+        raise ValueError(f'A buffering handler for {identifier} was not found')
+
+
+def format_buffered_logs_for_payload(identifier):
+    '''
+    Get the handler previously added to the root logger, and slurp buffered logs out of it to
+    send to Healthchecks.
+    '''
+    try:
+        buffering_handler = get_handler(identifier)
+    except ValueError:
+        # No handler means no payload.
+        return ''
+
+    payload = ''.join(message for message in buffering_handler.buffer)
+
+    if buffering_handler.forgot:
+        return PAYLOAD_TRUNCATION_INDICATOR + payload
+
+    return payload
+
+
+def remove_handler(identifier):
+    '''
+    Given the identifier for an existing Forgetful_buffering_handler instance, remove it.
+    '''
+    logger = logging.getLogger()
+
+    try:
+        logger.removeHandler(get_handler(identifier))
+    except ValueError:
+        pass

borgmatic/hooks/loki.py

@@ -0,0 +1,154 @@
+import json
+import logging
+import os
+import platform
+import time
+
+import requests
+
+from borgmatic.hooks import monitor
+
+logger = logging.getLogger(__name__)
+
+MONITOR_STATE_TO_LOKI = {
+    monitor.State.START: 'Started',
+    monitor.State.FINISH: 'Finished',
+    monitor.State.FAIL: 'Failed',
+}
+
+# Threshold at which logs get flushed to loki
+MAX_BUFFER_LINES = 100
+
+
+class Loki_log_buffer:
+    '''
+    A log buffer that allows to output the logs as loki requests in json. Allows
+    adding labels to the log stream and takes care of communication with loki.
+    '''
+
+    def __init__(self, url, dry_run):
+        self.url = url
+        self.dry_run = dry_run
+        self.root = {'streams': [{'stream': {}, 'values': []}]}
+
+    def add_value(self, value):
+        '''
+        Add a log entry to the stream.
+        '''
+        timestamp = str(time.time_ns())
+        self.root['streams'][0]['values'].append((timestamp, value))
+
+    def add_label(self, label, value):
+        '''
+        Add a label to the logging stream.
+        '''
+        self.root['streams'][0]['stream'][label] = value
+
+    def to_request(self):
+        return json.dumps(self.root)
+
+    def __len__(self):
+        '''
+        Gets the number of lines currently in the buffer.
+        '''
+        return len(self.root['streams'][0]['values'])
+
+    def flush(self):
+        if self.dry_run:
+            # Just empty the buffer and skip
+            self.root['streams'][0]['values'] = []
+            logger.info('Skipped uploading logs to loki due to dry run')
+            return
+
+        if len(self) == 0:
+            # Skip as there are not logs to send yet
+            return
+
+        request_body = self.to_request()
+        self.root['streams'][0]['values'] = []
+        request_header = {'Content-Type': 'application/json'}
+
+        try:
+            result = requests.post(self.url, headers=request_header, data=request_body, timeout=5)
+            result.raise_for_status()
+        except requests.RequestException:
+            logger.warning('Failed to upload logs to loki')
+
+
+class Loki_log_handler(logging.Handler):
+    '''
+    A log handler that sends logs to loki.
+    '''
+
+    def __init__(self, url, dry_run):
+        super().__init__()
+        self.buffer = Loki_log_buffer(url, dry_run)
+
+    def emit(self, record):
+        '''
+        Add a log record from the logging module to the stream.
+        '''
+        self.raw(record.getMessage())
+
+    def add_label(self, key, value):
+        '''
+        Add a label to the logging stream.
+        '''
+        self.buffer.add_label(key, value)
+
+    def raw(self, msg):
+        '''
+        Add an arbitrary string as a log entry to the stream.
+        '''
+        self.buffer.add_value(msg)
+
+        if len(self.buffer) > MAX_BUFFER_LINES:
+            self.buffer.flush()
+
+    def flush(self):
+        '''
+        Send the logs to loki and empty the buffer.
+        '''
+        self.buffer.flush()
+
+
+def initialize_monitor(hook_config, config, config_filename, monitoring_log_level, dry_run):
+    '''
+    Add a handler to the root logger to regularly send the logs to loki.
+    '''
+    url = hook_config.get('url')
+    loki = Loki_log_handler(url, dry_run)
+
+    for key, value in hook_config.get('labels').items():
+        if value == '__hostname':
+            loki.add_label(key, platform.node())
+        elif value == '__config':
+            loki.add_label(key, os.path.basename(config_filename))
+        elif value == '__config_path':
+            loki.add_label(key, config_filename)
+        else:
+            loki.add_label(key, value)
+
+    logging.getLogger().addHandler(loki)
+
+
+def ping_monitor(hook_config, config, config_filename, state, monitoring_log_level, dry_run):
+    '''
+    Add an entry to the loki logger with the current state.
+    '''
+    for handler in tuple(logging.getLogger().handlers):
+        if isinstance(handler, Loki_log_handler):
+            if state in MONITOR_STATE_TO_LOKI.keys():
+                handler.raw(f'{config_filename}: {MONITOR_STATE_TO_LOKI[state]} backup')
+
+
+def destroy_monitor(hook_config, config, config_filename, monitoring_log_level, dry_run):
+    '''
+    Remove the monitor handler that was added to the root logger.
+    '''
+    logger = logging.getLogger()
+
+    for handler in tuple(logger.handlers):
+        if isinstance(handler, Loki_log_handler):
+            handler.flush()
+            logger.removeHandler(handler)

borgmatic/hooks/mariadb.py

@@ -0,0 +1,257 @@
+import copy
+import logging
+import os
+import shlex
+
+from borgmatic.execute import (
+    execute_command,
+    execute_command_and_capture_output,
+    execute_command_with_processes,
+)
+from borgmatic.hooks import dump
+
+logger = logging.getLogger(__name__)
+
+
+def make_dump_path(config):  # pragma: no cover
+    '''
+    Make the dump path from the given configuration dict and the name of this hook.
+    '''
+    return dump.make_data_source_dump_path(
+        config.get('borgmatic_source_directory'), 'mariadb_databases'
+    )
+
+
+SYSTEM_DATABASE_NAMES = ('information_schema', 'mysql', 'performance_schema', 'sys')
+
+
+def database_names_to_dump(database, extra_environment, log_prefix, dry_run):
+    '''
+    Given a requested database config, return the corresponding sequence of database names to dump.
+    In the case of "all", query for the names of databases on the configured host and return them,
+    excluding any system databases that will cause problems during restore.
+    '''
+    if database['name'] != 'all':
+        return (database['name'],)
+    if dry_run:
+        return ()
+
+    mariadb_show_command = tuple(
+        shlex.quote(part) for part in shlex.split(database.get('mariadb_command') or 'mariadb')
+    )
+    show_command = (
+        mariadb_show_command
+        + (tuple(database['list_options'].split(' ')) if 'list_options' in database else ())
+        + (('--host', database['hostname']) if 'hostname' in database else ())
+        + (('--port', str(database['port'])) if 'port' in database else ())
+        + (('--protocol', 'tcp') if 'hostname' in database or 'port' in database else ())
+        + (('--user', database['username']) if 'username' in database else ())
+        + ('--skip-column-names', '--batch')
+        + ('--execute', 'show schemas')
+    )
+    logger.debug(f'{log_prefix}: Querying for "all" MariaDB databases to dump')
+    show_output = execute_command_and_capture_output(
+        show_command, extra_environment=extra_environment
+    )
+
+    return tuple(
+        show_name
+        for show_name in show_output.strip().splitlines()
+        if show_name not in SYSTEM_DATABASE_NAMES
+    )
+
+
+def execute_dump_command(
+    database, log_prefix, dump_path, database_names, extra_environment, dry_run, dry_run_label
+):
+    '''
+    Kick off a dump for the given MariaDB database (provided as a configuration dict) to a named
+    pipe constructed from the given dump path and database name. Use the given log prefix in any
+    log entries.
+
+    Return a subprocess.Popen instance for the dump process ready to spew to a named pipe. But if
+    this is a dry run, then don't actually dump anything and return None.
+    '''
+    database_name = database['name']
+    dump_filename = dump.make_data_source_dump_filename(
+        dump_path, database['name'], database.get('hostname')
+    )
+
+    if os.path.exists(dump_filename):
+        logger.warning(
+            f'{log_prefix}: Skipping duplicate dump of MariaDB database "{database_name}" to {dump_filename}'
+        )
+        return None
+
+    mariadb_dump_command = tuple(
+        shlex.quote(part)
+        for part in shlex.split(database.get('mariadb_dump_command') or 'mariadb-dump')
+    )
+    dump_command = (
+        mariadb_dump_command
+        + (tuple(database['options'].split(' ')) if 'options' in database else ())
+        + (('--add-drop-database',) if database.get('add_drop_database', True) else ())
+        + (('--host', database['hostname']) if 'hostname' in database else ())
+        + (('--port', str(database['port'])) if 'port' in database else ())
+        + (('--protocol', 'tcp') if 'hostname' in database or 'port' in database else ())
+        + (('--user', database['username']) if 'username' in database else ())
+        + ('--databases',)
+        + database_names
+        + ('--result-file', dump_filename)
+    )
+
+    logger.debug(
+        f'{log_prefix}: Dumping MariaDB database "{database_name}" to {dump_filename}{dry_run_label}'
+    )
+    if dry_run:
+        return None
+
+    dump.create_named_pipe_for_dump(dump_filename)
+
+    return execute_command(
+        dump_command,
+        extra_environment=extra_environment,
+        run_to_completion=False,
+    )
+
+
+def use_streaming(databases, config, log_prefix):
+    '''
+    Given a sequence of MariaDB database configuration dicts, a configuration dict (ignored), and a
+    log prefix (ignored), return whether streaming will be using during dumps.
+    '''
+    return any(databases)
+
+
+def dump_data_sources(databases, config, log_prefix, dry_run):
+    '''
+    Dump the given MariaDB databases to a named pipe. The databases are supplied as a sequence of
+    dicts, one dict describing each database as per the configuration schema. Use the given
+    configuration dict to construct the destination path and the given log prefix in any log
+    entries.
+
+    Return a sequence of subprocess.Popen instances for the dump processes ready to spew to a named
+    pipe. But if this is a dry run, then don't actually dump anything and return an empty sequence.
+    '''
+    dry_run_label = ' (dry run; not actually dumping anything)' if dry_run else ''
+    processes = []
+
+    logger.info(f'{log_prefix}: Dumping MariaDB databases{dry_run_label}')
+
+    for database in databases:
+        dump_path = make_dump_path(config)
+        extra_environment = {'MYSQL_PWD': database['password']} if 'password' in database else None
+        dump_database_names = database_names_to_dump(
+            database, extra_environment, log_prefix, dry_run
+        )
+
+        if not dump_database_names:
+            if dry_run:
+                continue
+
+            raise ValueError('Cannot find any MariaDB databases to dump.')
+
+        if database['name'] == 'all' and database.get('format'):
+            for dump_name in dump_database_names:
+                renamed_database = copy.copy(database)
+                renamed_database['name'] = dump_name
+                processes.append(
+                    execute_dump_command(
+                        renamed_database,
+                        log_prefix,
+                        dump_path,
+                        (dump_name,),
+                        extra_environment,
+                        dry_run,
+                        dry_run_label,
+                    )
+                )
+        else:
+            processes.append(
+                execute_dump_command(
+                    database,
+                    log_prefix,
+                    dump_path,
+                    dump_database_names,
+                    extra_environment,
+                    dry_run,
+                    dry_run_label,
+                )
+            )
+
+    return [process for process in processes if process]
+
+
+def remove_data_source_dumps(databases, config, log_prefix, dry_run):  # pragma: no cover
+    '''
+    Remove all database dump files for this hook regardless of the given databases. Use the given
+    configuration dict to construct the destination path and the log prefix in any log entries. If
+    this is a dry run, then don't actually remove anything.
+    '''
+    dump.remove_data_source_dumps(make_dump_path(config), 'MariaDB', log_prefix, dry_run)
+
+
+def make_data_source_dump_pattern(databases, config, log_prefix, name=None):  # pragma: no cover
+    '''
+    Given a sequence of configurations dicts, a configuration dict, a prefix to log with, and a
+    database name to match, return the corresponding glob patterns to match the database dump in an
+    archive.
+    '''
+    return dump.make_data_source_dump_filename(make_dump_path(config), name, hostname='*')
+
+
+def restore_data_source_dump(
+    hook_config, config, log_prefix, data_source, dry_run, extract_process, connection_params
+):
+    '''
+    Restore a database from the given extract stream. The database is supplied as a data source
+    configuration dict, but the given hook configuration is ignored. The given configuration dict is
+    used to construct the destination path, and the given log prefix is used for any log entries. If
+    this is a dry run, then don't actually restore anything. Trigger the given active extract
+    process (an instance of subprocess.Popen) to produce output to consume.
+    '''
+    dry_run_label = ' (dry run; not actually restoring anything)' if dry_run else ''
+    hostname = connection_params['hostname'] or data_source.get(
+        'restore_hostname', data_source.get('hostname')
+    )
+    port = str(
+        connection_params['port'] or data_source.get('restore_port', data_source.get('port', ''))
+    )
+    username = connection_params['username'] or data_source.get(
+        'restore_username', data_source.get('username')
+    )
+    password = connection_params['password'] or data_source.get(
+        'restore_password', data_source.get('password')
+    )
+
+    mariadb_restore_command = tuple(
+        shlex.quote(part) for part in shlex.split(data_source.get('mariadb_command') or 'mariadb')
+    )
+    restore_command = (
+        mariadb_restore_command
+        + ('--batch',)
+        + (
+            tuple(data_source['restore_options'].split(' '))
+            if 'restore_options' in data_source
+            else ()
+        )
+        + (('--host', hostname) if hostname else ())
+        + (('--port', str(port)) if port else ())
+        + (('--protocol', 'tcp') if hostname or port else ())
+        + (('--user', username) if username else ())
+    )
+    extra_environment = {'MYSQL_PWD': password} if password else None
+
+    logger.debug(f"{log_prefix}: Restoring MariaDB database {data_source['name']}{dry_run_label}")
+    if dry_run:
+        return
+
+    # Don't give Borg local path so as to error on warnings, as "borg extract" only gives a warning
+    # if the restore paths don't exist in the archive.
+    execute_command_with_processes(
+        restore_command,
+        [extract_process],
+        output_log_level=logging.DEBUG,
+        input_file=extract_process.stdout,
+        extra_environment=extra_environment,
+    )

borgmatic/hooks/mongodb.py

@@ -0,0 +1,185 @@
+import logging
+import shlex
+
+from borgmatic.execute import execute_command, execute_command_with_processes
+from borgmatic.hooks import dump
+
+logger = logging.getLogger(__name__)
+
+
+def make_dump_path(config):  # pragma: no cover
+    '''
+    Make the dump path from the given configuration dict and the name of this hook.
+    '''
+    return dump.make_data_source_dump_path(
+        config.get('borgmatic_source_directory'), 'mongodb_databases'
+    )
+
+
+def use_streaming(databases, config, log_prefix):
+    '''
+    Given a sequence of MongoDB database configuration dicts, a configuration dict (ignored), and a
+    log prefix (ignored), return whether streaming will be using during dumps.
+    '''
+    return any(database.get('format') != 'directory' for database in databases)
+
+
+def dump_data_sources(databases, config, log_prefix, dry_run):
+    '''
+    Dump the given MongoDB databases to a named pipe. The databases are supplied as a sequence of
+    dicts, one dict describing each database as per the configuration schema. Use the configuration
+    dict to construct the destination path and the given log prefix in any log entries.
+
+    Return a sequence of subprocess.Popen instances for the dump processes ready to spew to a named
+    pipe. But if this is a dry run, then don't actually dump anything and return an empty sequence.
+    '''
+    dry_run_label = ' (dry run; not actually dumping anything)' if dry_run else ''
+
+    logger.info(f'{log_prefix}: Dumping MongoDB databases{dry_run_label}')
+
+    processes = []
+    for database in databases:
+        name = database['name']
+        dump_filename = dump.make_data_source_dump_filename(
+            make_dump_path(config), name, database.get('hostname')
+        )
+        dump_format = database.get('format', 'archive')
+
+        logger.debug(
+            f'{log_prefix}: Dumping MongoDB database {name} to {dump_filename}{dry_run_label}',
+        )
+        if dry_run:
+            continue
+
+        command = build_dump_command(database, dump_filename, dump_format)
+
+        if dump_format == 'directory':
+            dump.create_parent_directory_for_dump(dump_filename)
+            execute_command(command, shell=True)
+        else:
+            dump.create_named_pipe_for_dump(dump_filename)
+            processes.append(execute_command(command, shell=True, run_to_completion=False))
+
+    return processes
+
+
+def build_dump_command(database, dump_filename, dump_format):
+    '''
+    Return the mongodump command from a single database configuration.
+    '''
+    all_databases = database['name'] == 'all'
+
+    return (
+        ('mongodump',)
+        + (('--out', shlex.quote(dump_filename)) if dump_format == 'directory' else ())
+        + (('--host', shlex.quote(database['hostname'])) if 'hostname' in database else ())
+        + (('--port', shlex.quote(str(database['port']))) if 'port' in database else ())
+        + (('--username', shlex.quote(database['username'])) if 'username' in database else ())
+        + (('--password', shlex.quote(database['password'])) if 'password' in database else ())
+        + (
+            ('--authenticationDatabase', shlex.quote(database['authentication_database']))
+            if 'authentication_database' in database
+            else ()
+        )
+        + (('--db', shlex.quote(database['name'])) if not all_databases else ())
+        + (
+            tuple(shlex.quote(option) for option in database['options'].split(' '))
+            if 'options' in database
+            else ()
+        )
+        + (('--archive', '>', shlex.quote(dump_filename)) if dump_format != 'directory' else ())
+    )
+
+
+def remove_data_source_dumps(databases, config, log_prefix, dry_run):  # pragma: no cover
+    '''
+    Remove all database dump files for this hook regardless of the given databases. Use the log
+    prefix in any log entries. Use the given configuration dict to construct the destination path.
+    If this is a dry run, then don't actually remove anything.
+    '''
+    dump.remove_data_source_dumps(make_dump_path(config), 'MongoDB', log_prefix, dry_run)
+
+
+def make_data_source_dump_pattern(databases, config, log_prefix, name=None):  # pragma: no cover
+    '''
+    Given a sequence of database configurations dicts, a configuration dict, a prefix to log with,
+    and a database name to match, return the corresponding glob patterns to match the database dump
+    in an archive.
+    '''
+    return dump.make_data_source_dump_filename(make_dump_path(config), name, hostname='*')
+
+
+def restore_data_source_dump(
+    hook_config, config, log_prefix, data_source, dry_run, extract_process, connection_params
+):
+    '''
+    Restore a database from the given extract stream. The database is supplied as a data source
+    configuration dict, but the given hook configuration is ignored. The given configuration dict is
+    used to construct the destination path, and the given log prefix is used for any log entries. If
+    this is a dry run, then don't actually restore anything. Trigger the given active extract
+    process (an instance of subprocess.Popen) to produce output to consume.
+
+    If the extract process is None, then restore the dump from the filesystem rather than from an
+    extract stream.
+    '''
+    dry_run_label = ' (dry run; not actually restoring anything)' if dry_run else ''
+    dump_filename = dump.make_data_source_dump_filename(
+        make_dump_path(config), data_source['name'], data_source.get('hostname')
+    )
+    restore_command = build_restore_command(
+        extract_process, data_source, dump_filename, connection_params
+    )
+
+    logger.debug(f"{log_prefix}: Restoring MongoDB database {data_source['name']}{dry_run_label}")
+    if dry_run:
+        return
+
+    # Don't give Borg local path so as to error on warnings, as "borg extract" only gives a warning
+    # if the restore paths don't exist in the archive.
+    execute_command_with_processes(
+        restore_command,
+        [extract_process] if extract_process else [],
+        output_log_level=logging.DEBUG,
+        input_file=extract_process.stdout if extract_process else None,
+    )
+
+
+def build_restore_command(extract_process, database, dump_filename, connection_params):
+    '''
+    Return the mongorestore command from a single database configuration.
+    '''
+    hostname = connection_params['hostname'] or database.get(
+        'restore_hostname', database.get('hostname')
+    )
+    port = str(connection_params['port'] or database.get('restore_port', database.get('port', '')))
+    username = connection_params['username'] or database.get(
+        'restore_username', database.get('username')
+    )
+    password = connection_params['password'] or database.get(
+        'restore_password', database.get('password')
+    )
+
+    command = ['mongorestore']
+    if extract_process:
+        command.append('--archive')
+    else:
+        command.extend(('--dir', dump_filename))
+    if database['name'] != 'all':
+        command.extend(('--drop',))
+    if hostname:
+        command.extend(('--host', hostname))
+    if port:
+        command.extend(('--port', str(port)))
+    if username:
+        command.extend(('--username', username))
+    if password:
+        command.extend(('--password', password))
+    if 'authentication_database' in database:
+        command.extend(('--authenticationDatabase', database['authentication_database']))
+    if 'restore_options' in database:
+        command.extend(database['restore_options'].split(' '))
+    if database.get('schemas'):
+        for schema in database['schemas']:
+            command.extend(('--nsInclude', schema))
+
+    return command

borgmatic/hooks/monitor.py

@@ -0,0 +1,10 @@
+from enum import Enum
+
+MONITOR_HOOK_NAMES = ('apprise', 'healthchecks', 'cronitor', 'cronhub', 'pagerduty', 'ntfy', 'loki')
+
+
+class State(Enum):
+    START = 1
+    FINISH = 2
+    FAIL = 3
+    LOG = 4

borgmatic/hooks/mysql.py

@@ -0,0 +1,255 @@
+import copy
+import logging
+import os
+import shlex
+
+from borgmatic.execute import (
+    execute_command,
+    execute_command_and_capture_output,
+    execute_command_with_processes,
+)
+from borgmatic.hooks import dump
+
+logger = logging.getLogger(__name__)
+
+
+def make_dump_path(config):  # pragma: no cover
+    '''
+    Make the dump path from the given configuration dict and the name of this hook.
+    '''
+    return dump.make_data_source_dump_path(
+        config.get('borgmatic_source_directory'), 'mysql_databases'
+    )
+
+
+SYSTEM_DATABASE_NAMES = ('information_schema', 'mysql', 'performance_schema', 'sys')
+
+
+def database_names_to_dump(database, extra_environment, log_prefix, dry_run):
+    '''
+    Given a requested database config, return the corresponding sequence of database names to dump.
+    In the case of "all", query for the names of databases on the configured host and return them,
+    excluding any system databases that will cause problems during restore.
+    '''
+    if database['name'] != 'all':
+        return (database['name'],)
+    if dry_run:
+        return ()
+
+    mysql_show_command = tuple(
+        shlex.quote(part) for part in shlex.split(database.get('mysql_command') or 'mysql')
+    )
+    show_command = (
+        mysql_show_command
+        + (tuple(database['list_options'].split(' ')) if 'list_options' in database else ())
+        + (('--host', database['hostname']) if 'hostname' in database else ())
+        + (('--port', str(database['port'])) if 'port' in database else ())
+        + (('--protocol', 'tcp') if 'hostname' in database or 'port' in database else ())
+        + (('--user', database['username']) if 'username' in database else ())
+        + ('--skip-column-names', '--batch')
+        + ('--execute', 'show schemas')
+    )
+    logger.debug(f'{log_prefix}: Querying for "all" MySQL databases to dump')
+    show_output = execute_command_and_capture_output(
+        show_command, extra_environment=extra_environment
+    )
+
+    return tuple(
+        show_name
+        for show_name in show_output.strip().splitlines()
+        if show_name not in SYSTEM_DATABASE_NAMES
+    )
+
+
+def execute_dump_command(
+    database, log_prefix, dump_path, database_names, extra_environment, dry_run, dry_run_label
+):
+    '''
+    Kick off a dump for the given MySQL/MariaDB database (provided as a configuration dict) to a
+    named pipe constructed from the given dump path and database name. Use the given log prefix in
+    any log entries.
+
+    Return a subprocess.Popen instance for the dump process ready to spew to a named pipe. But if
+    this is a dry run, then don't actually dump anything and return None.
+    '''
+    database_name = database['name']
+    dump_filename = dump.make_data_source_dump_filename(
+        dump_path, database['name'], database.get('hostname')
+    )
+
+    if os.path.exists(dump_filename):
+        logger.warning(
+            f'{log_prefix}: Skipping duplicate dump of MySQL database "{database_name}" to {dump_filename}'
+        )
+        return None
+
+    mysql_dump_command = tuple(
+        shlex.quote(part) for part in shlex.split(database.get('mysql_dump_command') or 'mysqldump')
+    )
+    dump_command = (
+        mysql_dump_command
+        + (tuple(database['options'].split(' ')) if 'options' in database else ())
+        + (('--add-drop-database',) if database.get('add_drop_database', True) else ())
+        + (('--host', database['hostname']) if 'hostname' in database else ())
+        + (('--port', str(database['port'])) if 'port' in database else ())
+        + (('--protocol', 'tcp') if 'hostname' in database or 'port' in database else ())
+        + (('--user', database['username']) if 'username' in database else ())
+        + ('--databases',)
+        + database_names
+        + ('--result-file', dump_filename)
+    )
+
+    logger.debug(
+        f'{log_prefix}: Dumping MySQL database "{database_name}" to {dump_filename}{dry_run_label}'
+    )
+    if dry_run:
+        return None
+
+    dump.create_named_pipe_for_dump(dump_filename)
+
+    return execute_command(
+        dump_command,
+        extra_environment=extra_environment,
+        run_to_completion=False,
+    )
+
+
+def use_streaming(databases, config, log_prefix):
+    '''
+    Given a sequence of MySQL database configuration dicts, a configuration dict (ignored), and a
+    log prefix (ignored), return whether streaming will be using during dumps.
+    '''
+    return any(databases)
+
+
+def dump_data_sources(databases, config, log_prefix, dry_run):
+    '''
+    Dump the given MySQL/MariaDB databases to a named pipe. The databases are supplied as a sequence
+    of dicts, one dict describing each database as per the configuration schema. Use the given
+    configuration dict to construct the destination path and the given log prefix in any log entries.
+
+    Return a sequence of subprocess.Popen instances for the dump processes ready to spew to a named
+    pipe. But if this is a dry run, then don't actually dump anything and return an empty sequence.
+    '''
+    dry_run_label = ' (dry run; not actually dumping anything)' if dry_run else ''
+    processes = []
+
+    logger.info(f'{log_prefix}: Dumping MySQL databases{dry_run_label}')
+
+    for database in databases:
+        dump_path = make_dump_path(config)
+        extra_environment = {'MYSQL_PWD': database['password']} if 'password' in database else None
+        dump_database_names = database_names_to_dump(
+            database, extra_environment, log_prefix, dry_run
+        )
+
+        if not dump_database_names:
+            if dry_run:
+                continue
+
+            raise ValueError('Cannot find any MySQL databases to dump.')
+
+        if database['name'] == 'all' and database.get('format'):
+            for dump_name in dump_database_names:
+                renamed_database = copy.copy(database)
+                renamed_database['name'] = dump_name
+                processes.append(
+                    execute_dump_command(
+                        renamed_database,
+                        log_prefix,
+                        dump_path,
+                        (dump_name,),
+                        extra_environment,
+                        dry_run,
+                        dry_run_label,
+                    )
+                )
+        else:
+            processes.append(
+                execute_dump_command(
+                    database,
+                    log_prefix,
+                    dump_path,
+                    dump_database_names,
+                    extra_environment,
+                    dry_run,
+                    dry_run_label,
+                )
+            )
+
+    return [process for process in processes if process]
+
+
+def remove_data_source_dumps(databases, config, log_prefix, dry_run):  # pragma: no cover
+    '''
+    Remove all database dump files for this hook regardless of the given databases. Use the given
+    configuration dict to construct the destination path and the log prefix in any log entries. If
+    this is a dry run, then don't actually remove anything.
+    '''
+    dump.remove_data_source_dumps(make_dump_path(config), 'MySQL', log_prefix, dry_run)
+
+
+def make_data_source_dump_pattern(databases, config, log_prefix, name=None):  # pragma: no cover
+    '''
+    Given a sequence of configurations dicts, a configuration dict, a prefix to log with, and a
+    database name to match, return the corresponding glob patterns to match the database dump in an
+    archive.
+    '''
+    return dump.make_data_source_dump_filename(make_dump_path(config), name, hostname='*')
+
+
+def restore_data_source_dump(
+    hook_config, config, log_prefix, data_source, dry_run, extract_process, connection_params
+):
+    '''
+    Restore a database from the given extract stream. The database is supplied as a data source
+    configuration dict, but the given hook configuration is ignored. The given configuration dict is
+    used to construct the destination path, and the given log prefix is used for any log entries. If
+    this is a dry run, then don't actually restore anything. Trigger the given active extract
+    process (an instance of subprocess.Popen) to produce output to consume.
+    '''
+    dry_run_label = ' (dry run; not actually restoring anything)' if dry_run else ''
+    hostname = connection_params['hostname'] or data_source.get(
+        'restore_hostname', data_source.get('hostname')
+    )
+    port = str(
+        connection_params['port'] or data_source.get('restore_port', data_source.get('port', ''))
+    )
+    username = connection_params['username'] or data_source.get(
+        'restore_username', data_source.get('username')
+    )
+    password = connection_params['password'] or data_source.get(
+        'restore_password', data_source.get('password')
+    )
+
+    mysql_restore_command = tuple(
+        shlex.quote(part) for part in shlex.split(data_source.get('mysql_command') or 'mysql')
+    )
+    restore_command = (
+        mysql_restore_command
+        + ('--batch',)
+        + (
+            tuple(data_source['restore_options'].split(' '))
+            if 'restore_options' in data_source
+            else ()
+        )
+        + (('--host', hostname) if hostname else ())
+        + (('--port', str(port)) if port else ())
+        + (('--protocol', 'tcp') if hostname or port else ())
+        + (('--user', username) if username else ())
+    )
+    extra_environment = {'MYSQL_PWD': password} if password else None
+
+    logger.debug(f"{log_prefix}: Restoring MySQL database {data_source['name']}{dry_run_label}")
+    if dry_run:
+        return
+
+    # Don't give Borg local path so as to error on warnings, as "borg extract" only gives a warning
+    # if the restore paths don't exist in the archive.
+    execute_command_with_processes(
+        restore_command,
+        [extract_process],
+        output_log_level=logging.DEBUG,
+        input_file=extract_process.stdout,
+        extra_environment=extra_environment,
+    )

borgmatic/hooks/ntfy.py

@@ -0,0 +1,90 @@
+import logging
+
+import requests
+
+logger = logging.getLogger(__name__)
+
+
+def initialize_monitor(
+    ping_url, config, config_filename, monitoring_log_level, dry_run
+):  # pragma: no cover
+    '''
+    No initialization is necessary for this monitor.
+    '''
+    pass
+
+
+def ping_monitor(hook_config, config, config_filename, state, monitoring_log_level, dry_run):
+    '''
+    Ping the configured Ntfy topic. Use the given configuration filename in any log entries.
+    If this is a dry run, then don't actually ping anything.
+    '''
+
+    run_states = hook_config.get('states', ['fail'])
+
+    if state.name.lower() in run_states:
+        dry_run_label = ' (dry run; not actually pinging)' if dry_run else ''
+
+        state_config = hook_config.get(
+            state.name.lower(),
+            {
+                'title': f'A borgmatic {state.name} event happened',
+                'message': f'A borgmatic {state.name} event happened',
+                'priority': 'default',
+                'tags': 'borgmatic',
+            },
+        )
+
+        base_url = hook_config.get('server', 'https://ntfy.sh')
+        topic = hook_config.get('topic')
+
+        logger.info(f'{config_filename}: Pinging ntfy topic {topic}{dry_run_label}')
+        logger.debug(f'{config_filename}: Using Ntfy ping URL {base_url}/{topic}')
+
+        headers = {
+            'X-Title': state_config.get('title'),
+            'X-Message': state_config.get('message'),
+            'X-Priority': state_config.get('priority'),
+            'X-Tags': state_config.get('tags'),
+        }
+
+        username = hook_config.get('username')
+        password = hook_config.get('password')
+        access_token = hook_config.get('access_token')
+        auth = None
+
+        if access_token is not None:
+            if username or password:
+                logger.warning(
+                    f'{config_filename}: ntfy access_token is set but so is username/password, only using access_token'
+                )
+            auth = requests.auth.HTTPBasicAuth('', access_token)
+        elif (username and password) is not None:
+            auth = requests.auth.HTTPBasicAuth(username, password)
+            logger.info(f'{config_filename}: Using basic auth with user {username} for ntfy')
+        elif username is not None:
+            logger.warning(
+                f'{config_filename}: Password missing for ntfy authentication, defaulting to no auth'
+            )
+        elif password is not None:
+            logger.warning(
+                f'{config_filename}: Username missing for ntfy authentication, defaulting to no auth'
+            )
+
+        if not dry_run:
+            logging.getLogger('urllib3').setLevel(logging.ERROR)
+            try:
+                response = requests.post(f'{base_url}/{topic}', headers=headers, auth=auth)
+                if not response.ok:
+                    response.raise_for_status()
+            except requests.exceptions.RequestException as error:
+                logger.warning(f'{config_filename}: ntfy error: {error}')
+
+
+def destroy_monitor(
+    ping_url_or_uuid, config, config_filename, monitoring_log_level, dry_run
+):  # pragma: no cover
+    '''
+    No destruction is necessary for this monitor.
+    '''
+    pass

borgmatic/hooks/pagerduty.py

@@ -0,0 +1,83 @@
+import datetime
+import json
+import logging
+import platform
+
+import requests
+
+from borgmatic.hooks import monitor
+
+logger = logging.getLogger(__name__)
+
+EVENTS_API_URL = 'https://events.pagerduty.com/v2/enqueue'
+
+
+def initialize_monitor(
+    integration_key, config, config_filename, monitoring_log_level, dry_run
+):  # pragma: no cover
+    '''
+    No initialization is necessary for this monitor.
+    '''
+    pass
+
+
+def ping_monitor(hook_config, config, config_filename, state, monitoring_log_level, dry_run):
+    '''
+    If this is an error state, create a PagerDuty event with the configured integration key. Use
+    the given configuration filename in any log entries. If this is a dry run, then don't actually
+    create an event.
+    '''
+    if state != monitor.State.FAIL:
+        logger.debug(
+            f'{config_filename}: Ignoring unsupported monitoring {state.name.lower()} in PagerDuty hook',
+        )
+        return
+
+    dry_run_label = ' (dry run; not actually sending)' if dry_run else ''
+    logger.info(f'{config_filename}: Sending failure event to PagerDuty {dry_run_label}')
+
+    if dry_run:
+        return
+
+    hostname = platform.node()
+    local_timestamp = (
+        datetime.datetime.utcnow().replace(tzinfo=datetime.timezone.utc).astimezone().isoformat()
+    )
+    payload = json.dumps(
+        {
+            'routing_key': hook_config['integration_key'],
+            'event_action': 'trigger',
+            'payload': {
+                'summary': f'backup failed on {hostname}',
+                'severity': 'error',
+                'source': hostname,
+                'timestamp': local_timestamp,
+                'component': 'borgmatic',
+                'group': 'backups',
+                'class': 'backup failure',
+                'custom_details': {
+                    'hostname': hostname,
+                    'configuration filename': config_filename,
+                    'server time': local_timestamp,
+                },
+            },
+        }
+    )
+    logger.debug(f'{config_filename}: Using PagerDuty payload: {payload}')
+
+    logging.getLogger('urllib3').setLevel(logging.ERROR)
+    try:
+        response = requests.post(EVENTS_API_URL, data=payload.encode('utf-8'))
+        if not response.ok:
+            response.raise_for_status()
+    except requests.exceptions.RequestException as error:
+        logger.warning(f'{config_filename}: PagerDuty error: {error}')
+
+
+def destroy_monitor(
+    ping_url_or_uuid, config, config_filename, monitoring_log_level, dry_run
+):  # pragma: no cover
+    '''
+    No destruction is necessary for this monitor.
+    '''
+    pass