🛠️ fix(parser): keep inline markers and ref link variants

Author: philippb

sdiff/parser.py

@@ -11,9 +11,14 @@
 
 _BLOCK_TAGS = {tag.lower() for tag in block_parser.BLOCK_TAGS}
 _HEADING_LINE_RE = re.compile(r'^(\s*)(#{1,6})(?!#)(?=\S)')
-_REF_LINK_OR_IMAGE_RE = re.compile(r'!?\[[^\]]+\]\[[^\]]+\]')
+_REF_LINK_OR_IMAGE_RE = re.compile(r'!?\[[^\]]+\]\s*\[[^\]]*\]')
 _REF_DEF_LINE_RE = re.compile(r'^\s{0,3}\[[^\]]+\]:\s+\S+')
 _FENCE_RE = re.compile(r'^\s*(```|~~~)')
+_INLINE_MARKERS = {
+    'strong': '**',
+    'emphasis': '*',
+    'strikethrough': '~~',
+}
 
 
 class MdParser:
@@ -134,6 +139,14 @@ def flush_buffer():
                 url = attrs.get('url', '')
                 title = attrs.get('title')
                 nodes.append(Image(_format_image_markup(alt, url, title)))
+            elif token_type in _INLINE_MARKERS:
+                flush_buffer()
+                marker = _INLINE_MARKERS[token_type]
+                _append_text(nodes, marker)
+                children = token.get('children', [])
+                if children:
+                    nodes.extend(self._convert_inline_tokens(children))
+                _append_text(nodes, marker)
             else:
                 flush_buffer()
                 children = token.get('children', [])
@@ -155,6 +168,10 @@ def _flatten_inline_text(self, tokens: Iterable[dict]):
                 parts.append(token.get('raw') or token.get('text') or '')
             elif token_type == 'codespan':
                 parts.append(f"`{token.get('raw') or token.get('text') or ''}`")
+            elif token_type in _INLINE_MARKERS:
+                marker = _INLINE_MARKERS[token_type]
+                inner = self._flatten_inline_text(token.get('children', []))
+                parts.append(f'{marker}{inner}{marker}')
             elif token_type in {'linebreak', 'softbreak'}:
                 parts.append(' ')
             else:

tests/test_parser.py

@@ -76,6 +76,10 @@ def test_link_label_with_codespan(self):
         actual = self._parse('[use `foo`](url)')
         self.assertEqual('[use `foo`](url)', actual.nodes[0].nodes[0].text)
 
+    def test_link_label_with_strong_preserves_markers(self):
+        actual = self._parse('[**bold**](url)')
+        self.assertEqual('[**bold**](url)', actual.nodes[0].nodes[0].text)
+
     def test_link_title_preserved(self):
         actual = self._parse('[label](https://example.com "Title Here")')
         self.assertEqual('[label](https://example.com "Title Here")', actual.nodes[0].nodes[0].text)
@@ -91,6 +95,13 @@ def test_reference_definition_preserved(self):
         self.assertEqual('[API][id]', link.text)
         self.assertEqual('[id]: https://example.com', tree.nodes[1].nodes[0].text)
 
+    def test_reference_links_with_whitespace_and_empty_id(self):
+        data = 'See [API][] and [Ref] [id].\n\n[API]: https://example.com\n[id]: https://example.com'
+        tree = self._parse(data)
+        link_texts = [node.text for node in tree.nodes[0].nodes if node.name == 'link']
+        self.assertIn('[API][]', link_texts)
+        self.assertIn('[Ref] [id]', link_texts)
+
     def test_reference_definition_inside_fence_is_text(self):
         data = """```
 [id]: https://example.com