Merge pull request #30 from OpenScienceFramework/issue_30

Do cleanup on insert/delete/sdt tags

Author: jlward

pydocx/DocxParser.py

@@ -576,12 +576,22 @@ def parse_p(self, el, text):
     def _should_append_break_tag(self, next_el):
         paragraph_like_tags = [
             'p',
-            'sdt',
+        ]
+        inline_like_tags = [
+            'smartTag',
+            'ins',
+            'delText',
         ]
         if next_el.is_list_item:
             return False
         if next_el.previous is None:
             return False
+        tag_is_inline_like = any(
+            next_el.has_descendant_with_tag(tag) for
+            tag in inline_like_tags
+        )
+        if tag_is_inline_like:
+            return False
         if next_el.previous.is_last_list_item_in_root:
             return False
         if next_el.previous.tag not in paragraph_like_tags:

pydocx/parsers/Docx2Html.py

@@ -54,8 +54,7 @@ def heading(self, text, heading_value):
 
     def insertion(self, text, author, date):
         return (
-            "<span class='pydocx-insert' author='%(author)s' "
-            "date='%(date)s'>%(text)s</span>"
+            "<span class='pydocx-insert'>%(text)s</span>"
         ) % {
             'author': author,
             'date': date,
@@ -88,8 +87,7 @@ def image(self, path, x, y):
 
     def deletion(self, text, author, date):
         return (
-            "<span class='pydocx-delete' author='%(author)s' "
-            "date='%(date)s'>%(text)s</span>"
+            "<span class='pydocx-delete'>%(text)s</span>"
         ) % {
             'author': author,
             'date': date,

pydocx/tests/test_xml.py

@@ -641,7 +641,7 @@ def test_performance(self):
 
 class NonStandardTextTagsTestCase(_TranslationTestCase):
     expected_output = '''
-        <p><span class='pydocx-insert' author='' date=''>insert </span>
+        <p><span class='pydocx-insert'>insert </span>
         smarttag</p>
     '''
 
@@ -674,8 +674,8 @@ def get_xml(self):
 class DeleteTagInList(_TranslationTestCase):
     expected_output = '''
         <ol list-style-type="decimal">
-            <li>AAA<br />
-                <span class='pydocx-delete' author='' date=''>BBB</span>
+            <li>AAA
+                <span class='pydocx-delete'>BBB</span>
             </li>
             <li>CCC</li>
         </ol>
@@ -696,8 +696,7 @@ def get_xml(self):
 class InsertTagInList(_TranslationTestCase):
     expected_output = '''
         <ol list-style-type="decimal">
-            <li>AAA<br />
-                <span class='pydocx-insert' author='' date=''>BBB</span>
+            <li>AAA<span class='pydocx-insert'>BBB</span>
             </li>
             <li>CCC</li>
         </ol>
@@ -719,8 +718,7 @@ def get_xml(self):
 class SmartTagInList(_TranslationTestCase):
     expected_output = '''
         <ol list-style-type="decimal">
-            <li>AAA<br />
-                BBB
+            <li>AAABBB
             </li>
             <li>CCC</li>
         </ol>
@@ -875,8 +873,7 @@ def get_xml(self):
 class SDTTestCase(_TranslationTestCase):
     expected_output = '''
         <ol list-style-type="decimal">
-            <li>AAA<br />
-                BBB
+            <li>AAABBB
             </li>
             <li>CCC</li>
         </ol>