Merge pull request #8 from PolicyStat/issue_8

If a 100% change of text in TD breaks the table layout
PolicyStat · Jan 19, 2015 · f2f1a68 · f2f1a68
2 parents 432096f + 57aa2f2
commit f2f1a68
Show file tree

Hide file tree

Showing 5 changed files with 399 additions and 4 deletions.
diff --git a/htmltreediff/html.py b/htmltreediff/html.py
@@ -1,11 +1,13 @@
 from htmltreediff.util import (
+    check_text_similarity,
+    is_element,
+    minidom_tostring,
     parse_minidom,
     parse_text,
-    minidom_tostring,
+    remove_node,
     unwrap,
     wrap_inner,
-    remove_node,
-    check_text_similarity,
+    wrap_nodes,
 )
 from htmltreediff.changes import dom_diff, distribute
 
@@ -46,10 +48,50 @@ def diff(old_html, new_html, cutoff=0.0, plaintext=False, pretty=False):
     return minidom_tostring(dom, pretty=pretty)
 
 
+def _internalize_changes_markup(dom, child_tag_names):
+    # Delete tags are always ordered first.
+    for del_tag in list(dom.getElementsByTagName('del')):
+        ins_tag = del_tag.nextSibling
+        # The one child tag of `del_tag` should be child_tag_names
+        if len(del_tag.childNodes) != 1:
+            continue
+        if ins_tag is None or len(ins_tag.childNodes) != 1:
+            continue
+        if ins_tag.tagName != 'ins':
+            continue
+        deleted_tag = del_tag.firstChild
+        if not is_element(deleted_tag):
+            continue
+        if deleted_tag.tagName not in child_tag_names:
+            continue
+        # The one child tag of `ins_tag` should be child_tag_names
+        inserted_tag = ins_tag.firstChild
+        if not is_element(inserted_tag):
+            continue
+        if inserted_tag.tagName not in child_tag_names:
+            continue
+
+        attributes = dict(
+            [key, value] for key, value in
+            inserted_tag.attributes.items()
+        )
+        nodes_to_unwrap = [
+            deleted_tag,
+            inserted_tag,
+        ]
+        for n in nodes_to_unwrap:
+            unwrap(n)
+        new_node = wrap_nodes([del_tag, ins_tag], inserted_tag.tagName)
+        for key, value in attributes.items():
+            new_node.setAttribute(key, value)
+
+
 def fix_lists(dom):
     # <ins> and <del> tags are not allowed within <ul> or <ol> tags.
     # Move them to the nearest li, so that the numbering isn't interrupted.
 
+    _internalize_changes_markup(dom, set(['li']))
+
     # Find all del > li and ins > li sets.
     del_tags = set()
     ins_tags = set()
@@ -73,6 +115,8 @@ def fix_lists(dom):
 
 
 def fix_tables(dom):
+    _internalize_changes_markup(dom, set(['td', 'th']))
+
     # Show table row insertions
     tags = set()
     for node in list(dom.getElementsByTagName('tr')):

diff --git a/htmltreediff/test_html.py b/htmltreediff/test_html.py
@@ -374,6 +374,158 @@ def test_fix_lists():
             </ol>
             '''
         ),
+        (
+            'LI full content change does not add another LI',
+            '''
+            <ol>
+              <del>
+                <li>AAA</li>
+              </del>
+              <ins>
+                <li>BBB</li>
+              </ins>
+            </ol>
+            ''',
+            '''
+            <ol>
+              <li><del>AAA</del><ins>BBB</ins></li>
+            </ol>
+            '''
+        ),
+        (
+            'LI full content change keeps attrs',
+            '''
+            <ol>
+              <del>
+                <li class="old" id="foo">AAA</li>
+              </del>
+              <ins>
+                <li class="new">BBB</li>
+              </ins>
+            </ol>
+            ''',
+            '''
+            <ol>
+              <li class="new"><del>AAA</del><ins>BBB</ins></li>
+            </ol>
+            '''
+        ),
+        (
+            'LI changes markup internalization fix not done if next tag is not an insert',  # noqa
+            '''
+            <ol>
+              <del>
+                <li>AAA</li>
+              </del>
+                <li><strong>BBB</strong></li>
+              <ins>
+                <li>CCC</li>
+              </ins>
+            </ol>
+            ''',
+            '''
+            <ol>
+                <li class="del-li">
+                    <del>AAA</del>
+                </li>
+                <li><strong>BBB</strong></li>
+                <li><ins>CCC</ins></li>
+            </ol>
+            ''',
+        ),
+        (
+            'LI changes markup internalization fix not done if next tag is not an insert',  # noqa
+            '''
+            <ol>
+              <del>
+                <li>AAA</li>
+              </del>
+                <li><strong>BBB</strong></li>
+              <ins>
+                <li>CCC</li>
+              </ins>
+            </ol>
+            ''',
+            '''
+            <ol>
+                <li class="del-li">
+                    <del>AAA</del>
+                </li>
+                <li><strong>BBB</strong></li>
+                <li><ins>CCC</ins></li>
+            </ol>
+            ''',
+        ),
+        (
+            'LI after del must be ins',
+            '''
+            <ol>
+              <del>
+                <li>AAA</li>
+              </del>
+              <del>
+                <li>BBB</li>
+              </del>
+              <ins>
+                <li>CCC</li>
+              </ins>
+            </ol>
+            ''',
+            '''
+            <ol>
+                <li class="del-li">
+                    <del>AAA</del>
+                </li>
+                <li><del>BBB</del><ins>CCC</ins></li>
+            </ol>
+            ''',
+        ),
+        (
+            'LI changes markup internalization fix not performed if next tags child is not li',  # noqa
+            '''
+            <ol>
+              <del>
+                <li>AAA</li>
+              </del>
+              <ins>
+                <foo>BBB</foo>
+              </ins>
+            </ol>
+            ''',
+            '''
+            <ol>
+                <li class="del-li">
+                    <del>AAA</del>
+                </li>
+                <ins>
+                    <foo>BBB</foo>
+                </ins>
+            </ol>
+            ''',
+        ),
+        (
+            'LI changes markup internalization fix not performed if next tags is text',  # noqa
+            '''
+            <ol>
+              <del>
+                <li>AAA</li>
+              </del>
+              <ins>
+                BBB
+              </ins>
+            </ol>
+            ''',
+            '''
+            <ol>
+                <li class="del-li">
+                    <del>AAA</del>
+                </li>
+                <ins>
+                    BBB
+                </ins>
+            </ol>
+            ''',
+        ),
     ]
     for test_name, changes, fixed_changes in cases:
         changes = collapse(changes)