Completed Html_differ unit tests.

2008-05-02 23:59:03 +00:00 · 2008-05-02 23:59:03 +00:00 · 42ae3e0ba1
parent ba89c8116b
commit 42ae3e0ba1
2 changed files with 275 additions and 30 deletions
--- a/controller/Html_differ.py
+++ b/controller/Html_differ.py
@ -17,6 +17,7 @@ class Html_differ( HTMLParser ):
  WORD_AND_WHITESPACE_PATTERN = re.compile( "\S*\s*" )

  def handle_data( self, data ):
+    # this turns "foo bar baz" into [ "foo ", "bar ", "baz" ] and extends the result with it
    self.result.extend( self.WORD_AND_WHITESPACE_PATTERN.findall( data ) )

  def handle_charref( self, ref ):
@ -33,8 +34,7 @@ class Html_differ( HTMLParser ):
      
  def handle_endtag( self, tag, attrs ):
    if tag not in self.requires_no_close:
-      bracketed = "</%s>" % tag
-      self.result.append( bracketed )
+      self.result.append( "</%s>" % tag )
      
  def unknown_starttag( self, tag, attr ):
    self.handle_starttag( tag, None, attr )
@ -45,35 +45,56 @@ class Html_differ( HTMLParser ):
  # used to replace, for instance, "<br/>" with "<br />"
  INVALID_TAG_PATTERN = re.compile( "(\S)/>" )
  INVALID_TAG_FIX = "\\1 />"
-  START_TAG_PATTERN = re.compile( "<([^/][^>]*)>" )
-  END_TAG_PATTERN = re.compile( "</([^>]+)>" )
+
+  def convert_html_to_list( self, html ):
+    """
+    Given an HTML string, produce a list of its constituent elements (tags and text).
+
+    @type html: unicode
+    @param html: HTML string to parse
+    @rtype: [ unicode, ... ]
+    @return: parsed list of HTML elements
+    """
+    self.reset()
+    self.result = []
+    html = self.INVALID_TAG_PATTERN.sub( self.INVALID_TAG_FIX, html )
+    self.feed( html )
+    return [ x for x in self.result if x != "" ]

  def diff( self, html_a, html_b ):
    """
-    Return a composite HTML diff of the given HTML input strings.
+    Return a composite HTML diff of the given HTML input strings. The returned string contains the
+    entirety of the input strings, but with deleted/modified text from html_a wrapped in <del> tags,
+    and inserted/modified text from html_b wrapped in <ins> tags.
+
+    @type html_a: unicode
+    @param html_a: original HTML string
+    @type html_b: unicode
+    @param html-b: modified HTML string
+    @rtype: unicode
+    @return: composite HTML diff
    """
-    # parse html_a into a list
-    self.reset()
-    self.result = []
-    html_a = self.INVALID_TAG_PATTERN.sub( self.INVALID_TAG_FIX, html_a )
-    self.feed( html_a )
-    a = [ x for x in self.result if x != "" ]
+    # parse the two html strings into lists
+    a = self.convert_html_to_list( html_a )
+    b = self.convert_html_to_list( html_b )

-    # parse html_b into a list
-    self.reset()
-    self.result = []
-    html_b = self.INVALID_TAG_PATTERN.sub( self.INVALID_TAG_FIX, html_b )
-    self.feed( html_b )
-    b = [ x for x in self.result if x != "" ]
+    # prepare the two lists for diffing, and then diff 'em
+    ( a, b ) = self.prepare_lists( a, b )
+    return self.diff_lists( a, b )

-    ( a, b ) = self.__prepare_lists( a, b )
-    return self.__diff_lists( a, b )
+  START_TAG_PATTERN = re.compile( "<([^/][^>]*)>" )
+  END_TAG_PATTERN = re.compile( "</([^>]+)>" )

  @staticmethod
-  def __track_open_tags( item, open_tags ):
+  def track_open_tags( item, open_tags ):
    """
-    Add or remove from the open_tags list based on whether the given item is a start or end
-    tag.
+    Add or remove from the open_tags list based on whether the given item contains a start or end
+    tag. If item does not contain any tag, then open_tags remains unchanged.
+
+    @type item: unicode
+    @param item: chunk of HTML, containing either an HTML tag or just text
+    @type open_tags: [ unicode, ... ]
+    @param open_tags: list of open tags
    """
    match = Html_differ.START_TAG_PATTERN.search( item )
    if match:
@ -87,10 +108,26 @@ class Html_differ( HTMLParser ):
    if match and tag in open_tags:
      open_tags.remove( tag )

-  def __prepare_lists( self, a, b ):
+  def prepare_lists( self, a, b ):
    """
-    Prepare the two lists for diffing by merging together adjacent elements within
-    modified/inserted/deleted start and end HTML tags.
+    Prepare the two lists for diffing by merging together adjacent elements that occur within
+    modified start and end HTML tags.
+
+    For instance, if:
+      a = [ 'foo ', 'bar ', 'baz ', 'quux' ]
+      b = [ 'foo ', '<i>', 'bar ', 'baz', '</i> ', 'quux' ]
+    then the returned lists are as follows:
+      a = [ 'foo ', 'bar baz ', 'quux' ]
+      b = [ 'foo ', '<i>bar baz</i> ', 'quux' ]
+
+    Merging these elements together ensures that they're diffed as a single unit. Failing to perform
+    this step would mean that when a phrase in list a becomes italicized in list b, then it wouldn't
+    show up as modified in the resulting diff.
+
+    @type a: [ unicode, ... ]
+    @type b: [ unicode, ... ]
+    @rtype: ( [ unicode, ... ], [ unicode, ... ] )
+    @return: tuple of resulting list a and list b
    """
    matcher = SequenceMatcher( None, a, b )
    result_a = []
@ -112,9 +149,9 @@ class Html_differ( HTMLParser ):

      # go through the altered items looking for start and end tags
      for i in range( i1, i2 ):
-        Html_differ.__track_open_tags( a[ i ], open_tags )
+        Html_differ.track_open_tags( a[ i ], open_tags )
      for j in range( j1, j2 ):
-        Html_differ.__track_open_tags( b[ j ], open_tags )
+        Html_differ.track_open_tags( b[ j ], open_tags )

      if change_type == "replace":
        open_del_items.extend( a[ i1:i2 ] )
@ -125,16 +162,23 @@ class Html_differ( HTMLParser ):
        open_ins_items.extend( b[ j1:j2 ] )

      if len( open_tags ) == 0:
-        result_a.append( ''.join( open_del_items ) )
-        result_b.append( ''.join( open_ins_items ) )
+        if len( open_del_items ) > 0:
+          result_a.append( ''.join( open_del_items ) )
+        if len( open_ins_items ) > 0:
+          result_b.append( ''.join( open_ins_items ) )
        open_del_items = []
        open_ins_items = []

    return ( result_a, result_b )

-  def __diff_lists( self, a, b ):
+  def diff_lists( self, a, b ):
    """
    Diff two prepared lists and return the result as an HTML string.
+
+    @type a: [ unicode, ... ]
+    @type b: [ unicode, ... ]
+    @rtype: unicode
+    @return: composite HTML diff
    """
    matcher = SequenceMatcher( None, a, b )
    result = []
--- a/controller/test/Test_html_differ.py
+++ b/controller/test/Test_html_differ.py
@ -0,0 +1,201 @@
+from controller.Html_differ import Html_differ
+
+
+class Test_html_differ( object ):
+  def setUp( self ):
+    self.differ = Html_differ()
+
+  def test_convert_html_to_list( self ):
+    result = self.differ.convert_html_to_list( u"foo <i>bar baz</i> quux" )
+
+    assert len( result ) == 7
+    assert result[ 0 ] == u"foo "
+    assert result[ 1 ] == u"<i>"
+    assert result[ 2 ] == u"bar "
+    assert result[ 3 ] == u"baz"
+    assert result[ 4 ] == u"</i>"
+    assert result[ 5 ] == u" "
+    assert result[ 6 ] == u"quux"
+
+  def test_convert_html_to_list_with_character_ref( self ):
+    result = self.differ.convert_html_to_list( u"foo &#35; quux" )
+
+    assert len( result ) == 4
+    assert result[ 0 ] == u"foo "
+    assert result[ 1 ] == u"&#35;"
+    assert result[ 2 ] == u" "
+    assert result[ 3 ] == u"quux"
+
+  def test_convert_html_to_list_with_entity_ref( self ):
+    result = self.differ.convert_html_to_list( u"foo &nbsp; quux" )
+
+    assert len( result ) == 4
+    assert result[ 0 ] == u"foo "
+    assert result[ 1 ] == u"&nbsp;"
+    assert result[ 2 ] == u" "
+    assert result[ 3 ] == u"quux"
+
+  def test_diff_with_insert( self ):
+    a = 'foo bar baz quux'
+    b = 'foo bar whee baz quux'
+
+    result = self.differ.diff( a, b )
+
+    assert result == 'foo bar <ins class="diff">whee </ins>baz quux'
+
+  def test_diff_with_delete( self ):
+    a = 'foo bar baz quux'
+    b = 'foo bar quux'
+
+    result = self.differ.diff( a, b )
+
+    assert result == 'foo bar <del class="diff">baz </del>quux'
+
+  def test_diff_with_replace( self ):
+    a = 'foo bar baz quux'
+    b = 'foo bar whee quux'
+
+    result = self.differ.diff( a, b )
+
+    assert result == 'foo bar <del class="diff modified">baz </del><ins class="diff modified">whee </ins>quux'
+
+  def test_diff_with_italics( self ):
+    a = 'foo bar baz quux'
+    b = 'foo <i>bar baz</i> quux'
+
+    result = self.differ.diff( a, b )
+
+    assert result == 'foo <del class="diff modified">bar baz </del><ins class="diff modified"><i>bar baz</i> </ins>quux'
+
+  def test_diff_with_italics_and_insert( self ):
+    a = 'foo bar baz quux'
+    b = 'foo <i>bar whee baz</i> quux'
+
+    result = self.differ.diff( a, b )
+
+    assert result == 'foo <del class="diff modified">bar baz </del><ins class="diff modified"><i>bar whee baz</i> </ins>quux'
+
+  def test_track_open_tags( self ):
+    open_tags = []
+
+    self.differ.track_open_tags( u"foo ", open_tags )
+    assert open_tags == []
+    self.differ.track_open_tags( u"<i>", open_tags )
+    assert open_tags == [ u"i" ]
+    self.differ.track_open_tags( u"bar ", open_tags )
+    assert open_tags == [ u"i" ]
+    self.differ.track_open_tags( u"<b>", open_tags )
+    assert open_tags == [ u"i", u"b" ]
+    self.differ.track_open_tags( u"baz", open_tags )
+    assert open_tags == [ u"i", u"b" ]
+    self.differ.track_open_tags( u"</b>", open_tags )
+    assert open_tags == [ u"i" ]
+    self.differ.track_open_tags( u"</i>", open_tags )
+    assert open_tags == []
+    self.differ.track_open_tags( u"quux", open_tags )
+    assert open_tags == []
+
+  def test_prepare_lists_with_insert( self ):
+    a = [ 'foo ', 'bar ', 'baz ', 'quux' ]
+    b = [ 'foo ', 'bar ', 'whee ', 'baz ', 'quux' ]
+
+    result = self.differ.prepare_lists( a, b )
+
+    assert len( result ) == 2
+    ( new_a, new_b ) = result
+
+    # there should be no change
+    assert new_a == a
+    assert new_b == b
+
+  def test_prepare_lists_with_delete( self ):
+    a = [ 'foo ', 'bar ', 'baz ', 'quux' ]
+    b = [ 'foo ', 'bar ', 'quux' ]
+
+    result = self.differ.prepare_lists( a, b )
+
+    assert len( result ) == 2
+    ( new_a, new_b ) = result
+
+    # there should be no change
+    assert new_a == a
+    assert new_b == b
+
+  def test_prepare_lists_with_replace( self ):
+    a = [ 'foo ', 'bar ', 'baz ', 'quux' ]
+    b = [ 'foo ', 'bar ', 'whee ', 'quux' ]
+
+    result = self.differ.prepare_lists( a, b )
+
+    assert len( result ) == 2
+    ( new_a, new_b ) = result
+
+    # there should be no change
+    assert new_a == a
+    assert new_b == b
+
+  def test_prepare_lists_with_italics( self ):
+    a = [ 'foo ', 'bar ', 'baz ', 'quux' ]
+    b = [ 'foo ', '<i>', 'bar ', 'baz', '</i> ', 'quux' ]
+
+    result = self.differ.prepare_lists( a, b )
+
+    assert len( result ) == 2
+    ( new_a, new_b ) = result
+
+    # the elements within italics should be merged
+    assert new_a == [ 'foo ', 'bar baz ', 'quux' ]
+    assert new_b == [ 'foo ', '<i>bar baz</i> ', 'quux' ]
+
+  def test_prepare_lists_with_italics_and_insert( self ):
+    a = [ 'foo ', 'bar ', 'baz ', 'quux' ]
+    b = [ 'foo ', '<i>', 'bar ', 'whee ', 'baz', '</i> ', 'quux' ]
+
+    result = self.differ.prepare_lists( a, b )
+
+    assert len( result ) == 2
+    ( new_a, new_b ) = result
+
+    # the elements within italics should be merged
+    assert new_a == [ 'foo ', 'bar baz ', 'quux' ]
+    assert new_b == [ 'foo ', '<i>bar whee baz</i> ', 'quux' ]
+
+  def test_diff_lists_with_insert( self ):
+    a = [ 'foo ', 'bar ', 'baz ', 'quux' ]
+    b = [ 'foo ', 'bar ', 'whee ', 'baz ', 'quux' ]
+
+    result = self.differ.diff_lists( a, b )
+
+    assert result == 'foo bar <ins class="diff">whee </ins>baz quux'
+
+  def test_diff_lists_with_delete( self ):
+    a = [ 'foo ', 'bar ', 'baz ', 'quux' ]
+    b = [ 'foo ', 'bar ', 'quux' ]
+
+    result = self.differ.diff_lists( a, b )
+
+    assert result == 'foo bar <del class="diff">baz </del>quux'
+
+  def test_diff_lists_with_replace( self ):
+    a = [ 'foo ', 'bar ', 'baz ', 'quux' ]
+    b = [ 'foo ', 'bar ', 'whee ', 'quux' ]
+
+    result = self.differ.diff_lists( a, b )
+
+    assert result == 'foo bar <del class="diff modified">baz </del><ins class="diff modified">whee </ins>quux'
+
+  def test_diff_lists_with_italics( self ):
+    a = [ 'foo ', 'bar baz ', 'quux' ]
+    b = [ 'foo ', '<i>bar baz</i> ', 'quux' ]
+
+    result = self.differ.diff_lists( a, b )
+
+    assert result == 'foo <del class="diff modified">bar baz </del><ins class="diff modified"><i>bar baz</i> </ins>quux'
+
+  def test_diff_lists_with_italics_and_insert( self ):
+    a = [ 'foo ', 'bar baz ', 'quux' ]
+    b = [ 'foo ', '<i>bar whee baz</i> ', 'quux' ]
+
+    result = self.differ.diff_lists( a, b )
+
+    assert result == 'foo <del class="diff modified">bar baz </del><ins class="diff modified"><i>bar whee baz</i> </ins>quux'