From fc3849b8bee5bb513333dec83482d2cfbb9e6408 Mon Sep 17 00:00:00 2001
From: Dan Helfman <dan@luminotes.com>
Date: Fri, 21 Nov 2008 14:07:23 -0800
Subject: [PATCH] HTML entities/characters like "&quot;" are now stripped from
 notebook names before turning them into friendly ids. This means that blog
 posts now have better URLs if they the post titles contain HTML
 entities/characters.

---
 model/Notebook.py           |  8 ++++++--
 model/delta/1.5.8.sql       | 15 +++++++++++++++
 model/schema.sql            | 13 ++++++++++++-
 model/test/Test_notebook.py |  8 ++++++++
 4 files changed, 41 insertions(+), 3 deletions(-)
 create mode 100644 model/delta/1.5.8.sql

diff --git a/model/Notebook.py b/model/Notebook.py
index 58b5a18..5b3839b 100644
--- a/model/Notebook.py
+++ b/model/Notebook.py
@@ -360,11 +360,15 @@ class Notebook( Persistent ):
     self.__name = name
     self.update_revision()
 
+  HTML_REFERENCE_PATTERN = re.compile( "&[a-zA-Z]+;|&#\d+;" )
   FRIENDLY_ID_STRIP_PATTERN = re.compile( "[^a-zA-Z0-9\-]+" )
 
   def __friendly_id( self ):
-    friendly_id = self.WHITESPACE_PATTERN.sub( u"-", self.__name.lower() )
-    return self.FRIENDLY_ID_STRIP_PATTERN.sub( u"", friendly_id )
+    # convert to lowercase, remove HTML character/entity refs, collapse whitespace to dashes, strip
+    # other punctuation. strip leading/trailing dashes
+    friendly_id = self.HTML_REFERENCE_PATTERN.sub( u" ", self.__name.lower() )
+    friendly_id = self.WHITESPACE_PATTERN.sub( u"-", friendly_id )
+    return self.FRIENDLY_ID_STRIP_PATTERN.sub( u"", friendly_id ).strip( "-" )
 
   def __set_read_write( self, read_write ):
     # The read_write member isn't actually saved to the database, so setting it doesn't need to
diff --git a/model/delta/1.5.8.sql b/model/delta/1.5.8.sql
new file mode 100644
index 0000000..a0de041
--- /dev/null
+++ b/model/delta/1.5.8.sql
@@ -0,0 +1,15 @@
+CREATE OR REPLACE FUNCTION friendly_id(text) RETURNS text
+    AS $_$select trim( both '-' from
+      regexp_replace(
+        regexp_replace(
+          regexp_replace(
+            lower( $1 ),
+            '&[a-zA-Z]+;|&#\\d+;', ' ', 'g'
+          ),
+          '\\s+', '-', 'g'
+        ),
+        '[^a-zA-Z0-9\\-]', '', 'g'
+      )
+    );$_$
+    LANGUAGE sql IMMUTABLE;
+reindex index notebook_friendly_id_index;
diff --git a/model/schema.sql b/model/schema.sql
index 2e7543a..8072ab9 100644
--- a/model/schema.sql
+++ b/model/schema.sql
@@ -26,7 +26,18 @@ create function log_note_revision() returns trigger as $_$
   $_$ language plpgsql;
 ALTER FUNCTION public.log_note_revision() OWNER TO luminotes;
 CREATE FUNCTION friendly_id(text) RETURNS text
-    AS $_$select regexp_replace( regexp_replace( lower( $1 ), '\\s+', '-', 'g' ), '[^a-zA-Z0-9\\-]', '', 'g' );$_$
+    AS $_$select trim( both '-' from
+      regexp_replace(
+        regexp_replace(
+          regexp_replace(
+            lower( $1 ),
+            '&[a-zA-Z]+;|&#\\d+;', ' ', 'g'
+          ),
+          '\\s+', '-', 'g'
+        ),
+        '[^a-zA-Z0-9\\-]', '', 'g'
+      )
+    );$_$
     LANGUAGE sql IMMUTABLE;
 ALTER FUNCTION public.friendly_id(text) OWNER TO luminotes;
 CREATE TABLE file (
diff --git a/model/test/Test_notebook.py b/model/test/Test_notebook.py
index 994160a..1cbb076 100644
--- a/model/test/Test_notebook.py
+++ b/model/test/Test_notebook.py
@@ -177,6 +177,14 @@ class Test_notebook( object ):
     self.notebook.name = u"This is Bob's  notebook!"
     assert self.notebook.friendly_id == u"this-is-bobs-notebook"
 
+  def test_friendly_id_with_html_entity_reference( self ):
+    self.notebook.name = u"This is Bob's &quot;notebook&quot;!"
+    assert self.notebook.friendly_id == u"this-is-bobs-notebook"
+
+  def test_friendly_id_with_html_character_reference( self ):
+    self.notebook.name = u"This is Bob's &#165; notebook!"
+    assert self.notebook.friendly_id == u"this-is-bobs-notebook"
+
   def test_set_read_write( self ):
     original_revision = self.notebook.revision
     self.notebook.read_write = Notebook.READ_WRITE_FOR_OWN_NOTES