More diff fixes. This time for "<br/>" tags.

NEWS

@@ -1,3 +1,7 @@
+1.3.11: May 2, 2008
+ * Fixed another bug that prevented diffs from working. This time, those with
+   <br/> tags didn't parse correctly.
+
 1.3.10: May 2, 2008
  * Fixed a bug that prevented diffs from working for notes with links.
 

controller/Html_differ.py

@@ -82,8 +82,9 @@ class Html_differ( HTMLParser ):
     ( a, b ) = self.prepare_lists( a, b )
     return self.diff_lists( a, b )
 
-  START_TAG_PATTERN = re.compile( "<(\w+)(\s+[^>]*)*>" )
-  END_TAG_PATTERN = re.compile( "</(\w+)>" )
+  SINGLE_TAG_PATTERN = re.compile( "<(\w+)(\s+[^>]*)*\s*/>" ) # e.g. '<br/>' or '<br />' or '<img src="foo" />'
+  START_TAG_PATTERN = re.compile( "<(\w+)(\s+[^>]*)*>" )      # e.g. '<i>' or '<a href="foo">'
+  END_TAG_PATTERN = re.compile( "</(\w+)>" )                  # e.g. '</i>' or '</a>'
 
   @staticmethod
   def track_open_tags( item, open_tags ):
@@ -96,6 +97,9 @@ class Html_differ( HTMLParser ):
     @type open_tags: [ unicode, ... ]
     @param open_tags: list of open tags
     """
+    match = Html_differ.SINGLE_TAG_PATTERN.search( item )
+    if match: return
+
     match = Html_differ.START_TAG_PATTERN.search( item )
     if match:
       open_tags.append( match.group( 1 ) )
@@ -148,11 +152,14 @@ class Html_differ( HTMLParser ):
         continue
 
       # go through the altered items looking for start and end tags
+      orig_len_open_tags = len( open_tags )
       for i in range( i1, i2 ):
         Html_differ.track_open_tags( a[ i ], open_tags )
       for j in range( j1, j2 ):
         Html_differ.track_open_tags( b[ j ], open_tags )
 
+      all_tags_got_closed = ( orig_len_open_tags > 0 and len( open_tags ) == 0 )
+
       if change_type == "replace":
         open_del_items.extend( a[ i1:i2 ] )
         open_ins_items.extend( b[ j1:j2 ] )
@@ -161,13 +168,19 @@ class Html_differ( HTMLParser ):
       elif change_type == "insert":
         open_ins_items.extend( b[ j1:j2 ] )
 
-      if len( open_tags ) == 0:
+      if all_tags_got_closed:
+        # if all tags were just closed, then merge the items that were in those tags
         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 = []
+      elif len( open_tags ) == 0:
+        result_a.extend( open_del_items )
+        result_b.extend( open_ins_items )
+        open_del_items = []
+        open_ins_items = []
 
     return ( result_a, result_b )
 

controller/test/Test_html_differ.py

@@ -83,11 +83,22 @@ class Test_html_differ( object ):
 
     assert result == 'foo bar <del class="diff modified">baz </del><ins class="diff modified"><a href="whee">baz</a> </ins>quux'
 
+  def test_diff_with_br( self ):
+    a = 'foo bar baz quux'
+    b = 'foo bar <br/><br />baz quux'
+
+    result = self.differ.diff( a, b )
+
+    print result
+    assert result == 'foo bar <ins class="diff"><br /><br /></ins>baz 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"<br/>", 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 )
@@ -96,6 +107,8 @@ class Test_html_differ( object ):
     assert open_tags == [ u"i", u"a" ]
     self.differ.track_open_tags( u"baz", open_tags )
     assert open_tags == [ u"i", u"a" ]
+    self.differ.track_open_tags( u"<br />", open_tags )
+    assert open_tags == [ u"i", u"a" ]
     self.differ.track_open_tags( u"</a>", open_tags )
     assert open_tags == [ u"i" ]
     self.differ.track_open_tags( u"</i>", open_tags )
@@ -181,6 +194,19 @@ class Test_html_differ( object ):
     assert new_a == [ 'foo ', 'bar baz ', 'quux' ]
     assert new_b == [ 'foo ', '<a href="whee">bar baz</a> ', 'quux' ]
 
+  def test_prepare_lists_with_br( self ):
+    a = [ 'foo ', 'bar ', 'baz ', 'quux' ]
+    b = [ 'foo ', 'bar ', '<br/>', '<br />', '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_diff_lists_with_insert( self ):
     a = [ 'foo ', 'bar ', 'baz ', 'quux' ]
     b = [ 'foo ', 'bar ', 'whee ', 'baz ', 'quux' ]
@@ -229,3 +255,11 @@ class Test_html_differ( object ):
 
     assert result == 'foo <del class="diff modified">bar baz </del><ins class="diff modified"><a href="whee">bar baz</a> </ins>quux'
 
+  def test_diff_lists_with_br( self ):
+    a = [ 'foo ', 'bar ', 'baz ', 'quux' ]
+    b = [ 'foo ', 'bar ', '<br/>', '<br />', 'baz ', 'quux' ]
+
+    result = self.differ.diff_lists( a, b )
+
+    print result
+    assert result == 'foo bar <ins class="diff"><br/><br /></ins>baz quux'