Remove hash links

sethblack · sethblack · commit a0641a74cf4d · 2021-01-18T14:28:15.000-06:00
diff --git a/seoanalyzer/page.py b/seoanalyzer/page.py
@@ -368,9 +368,14 @@ def analyze_a_tags(self, bs):
 
             url_filename, url_file_extension = os.path.splitext(modified_url)
 
+            # ignore links to images
             if url_file_extension in IMAGE_EXTENSIONS:
                 continue
 
+            # remove hash links to all urls
+            if '#' in modified_url:
+                modified_url = modified_url[:modified_url.rindex('#')]
+
             self.links.append(modified_url)
 
     def rel_to_abs_url(self, link):
