Merge pull request #501 from Crozzers/fix-link-patterns-issue287

Fix link patterns extra matching against internal hashes

CHANGES.md

@@ -3,6 +3,7 @@
 ## python-markdown2 2.4.9 (not yet released)
 
 - [pull #500] Add `<thead>` tag to html-classes extra
+- [pull #501] Fix link patterns extra matching against internal hashes
 - [pull #502] Replace deprecated `optparse` with `argparse`
 - [pull #506] Fix `_uniform_outdent` failing with empty strings (issue #505)
 

lib/markdown2.py

@@ -2533,6 +2533,9 @@ def _do_link_patterns(self, text):
         for regex, repl in self.link_patterns:
             replacements = []
             for match in regex.finditer(text):
+                if any(self._match_overlaps_substr(text, match, h) for h in link_from_hash):
+                    continue
+
                 if hasattr(repl, "__call__"):
                     href = repl(match)
                 else:
@@ -2642,6 +2645,19 @@ def _uniform_indent(self, text, indent, include_empty_lines=False):
             for line in text.splitlines(True)
         )
 
+    @staticmethod
+    def _match_overlaps_substr(text, match, substr):
+        '''
+        Checks if a regex match overlaps with a substring in the given text.
+        '''
+        for instance in re.finditer(re.escape(substr), text):
+            start, end = instance.span()
+            if start <= match.start() <= end:
+                return True
+            if start <= match.end() <= end:
+                return True
+        return False
+
 
 class MarkdownWithExtras(Markdown):
     """A markdowner class that enables most extras:

test/tm-cases/link_patterns_hash_matching_issue287.html

@@ -0,0 +1 @@
+<p>this is a test issue <a href="https://github.com/pyfa-org/Pyfa/issues/1234">#1234</a> with a test commit (<a href="https://github.com/pyfa-org/Pyfa/commit/addeddd">addeddd</a>) made by test <a href="https://github.com/username">@username</a> more text</p>

test/tm-cases/link_patterns_hash_matching_issue287.opts

@@ -0,0 +1,7 @@
+{"extras": ["link-patterns"],
+ "link_patterns": [
+    (re.compile("#(\d+)", re.I), r"https://github.com/pyfa-org/Pyfa/issues/\1"),
+    (re.compile("@(\w+)", re.I), r"https://github.com/\1"),
+    (re.compile("([0-9a-f]{6,40})", re.I), r"https://github.com/pyfa-org/Pyfa/commit/\1")
+    ]
+}

test/tm-cases/link_patterns_hash_matching_issue287.text

@@ -0,0 +1 @@
+this is a test issue #1234 with a test commit (addeddd) made by test @username more text