Fix: headers with tag

md_dead_link_check/preprocess.py

@@ -10,8 +10,9 @@
 
 RE_HEADER = r"^(?:\s*[-+*]\s+|)[#]{1,6}\s*(.*)"
 RE_LINK = r"([!]{0,1})\[([^\]!]*)\]\(([^()\s]+(?:\([^()\s]*\))*)\s*(.*?)\)"
-RE_HTML_A_TAG_ID = r"<\w+\s+(?:[^>]*?\s+)?(?:id|name)=([\"'])(.*?)\1"
-RE_HTML_A_TAG_HREF = r"<\w+\s+(?:[^>]*?\s+)?href=([\"'])(.*?)\1"
+RE_HTML_TAG = r"</?\w+[^>]*>"
+RE_HTML_TAG_ID = r"<\w+\s+(?:[^>]*?\s+)?(?:id|name)=([\"'])(.*?)\1"
+RE_HTML_TAG_HREF = r"<\w+\s+(?:[^>]*?\s+)?href=([\"'])(.*?)\1"
 RE_SUB = r"[$`][^`]+?[$`]"
 
 
@@ -57,6 +58,7 @@ def process_header_to_fragment(header: str) -> str:
     """
 
     fragment = header.strip()
+    fragment = re.sub(RE_HTML_TAG, "", fragment)
     while True:
         res = re.search(RE_LINK, fragment)
         if not res:
@@ -103,7 +105,6 @@ def process_md_file(path: Path, root_dir: Path) -> MarkdownInfo:
                     repeat += 1
                     fragment = f"{_fragment}-{repeat}"
                 fragments.append(fragment)
-                continue
 
             # Skip $ and ` tags
             line = re.sub(RE_SUB, "", line)
@@ -121,12 +122,12 @@ def process_md_file(path: Path, root_dir: Path) -> MarkdownInfo:
                     links.append(LinkInfo(link, path, line_num))
 
             # Detect id under a tag <a id="introduction"></a>
-            matches = re.findall(RE_HTML_A_TAG_ID, line)
+            matches = re.findall(RE_HTML_TAG_ID, line)
             for _, id in matches:
                 fragments.append(id.lower())
 
             # Detect links under a tag <a href="introduction"></a>
-            matches = re.findall(RE_HTML_A_TAG_HREF, line)
+            matches = re.findall(RE_HTML_TAG_HREF, line)
             for _, link in matches:
                 links.append(LinkInfo(link, path, line_num))
     return MarkdownInfo(path=path, fragments=fragments, links=links)

tests/test_md_files/a.md

@@ -49,3 +49,7 @@ Some text
 [0](#badge)
 [1](#badge-1)
 [1](#badge-2)
+
+#### Some [link](b.md) [link2](b.md)
+
+#### Some tag<a id="id"><a id="id2"></a> asd <a id="id3"></a>

tests/test_preprocess.py

@@ -38,6 +38,7 @@ def test_find_all_markdowns_in_repo():
         ("🙀 header with icon", "-header-with-icon"),
         ("דוגמא", "דוגמא"),
         ("例子", "例子"),
+        ("text (br)", "text"),
     ),
 )
 def test_process_header_to_fragment(header, fragment):
@@ -59,6 +60,11 @@ def test_process_md_file():
         "badge",
         "badge-1",
         "badge-2",
+        "some-link-link2",
+        "some-tag-asd-",
+        "id",
+        "id2",
+        "id3",
     ]
 
     ref_links = [
@@ -127,5 +133,15 @@ def test_process_md_file():
             location=Path("tests/test_md_files/a.md"),
             line_num=51,
         ),
+        LinkInfo(
+            link="b.md",
+            location=Path("tests/test_md_files/a.md"),
+            line_num=53,
+        ),
+        LinkInfo(
+            link="b.md",
+            location=Path("tests/test_md_files/a.md"),
+            line_num=53,
+        ),
     ]
     assert md_info.links == ref_links