Add in a check to remove trailing commas or slash characters from a URL

Skptak · Nov 8, 2023 · c48d7f8 · c48d7f8
1 parent d8787aa
commit c48d7f8
Showing 1 changed file with 4 additions and 1 deletion.
diff --git a/link-verifier/verify-links.py b/link-verifier/verify-links.py
@@ -14,7 +14,7 @@
 import traceback
 from collections import defaultdict
 
-MARKDOWN_SEARCH_TERM = r'\.md$'
+MARKDOWN_SEARCH_TERM = r"\.md$"
 # Regex to find a URL
 URL_SEARCH_TERM = r'(\b(https?)://[^\s\)\]\\"<>]+[^\s\)\.\]\\"<>])'
 HTTP_URL_SEARCH_TERM = r'https?://'
@@ -354,6 +354,9 @@ def main():
                         text = f.read()
                         urls = re.findall(URL_SEARCH_TERM, text)
                         for url in urls:
+                            # Remove trailing comma or slash
+                            if(url.endswith(',') or url.endswith('/')):
+                                url[0] = url[0][:-1]
                             link_set.add(url[0])
                             link_to_files[url[0]].add(f_path)