Revert "don't include links anymore"

This reverts commit fba3f789bb.
2022-05-28 15:24:19 +02:00 · 2022-05-28 15:24:19 +02:00 · c9f435f0f5
parent 0dc6fd60d5
commit c9f435f0f5
1 changed files with 8 additions and 0 deletions
--- a/tagger.py
+++ b/tagger.py
@ -57,6 +57,14 @@ class FileScanner(HTMLParser):
                words_with_usage[word].score += score
        return sorted(words_with_usage.values(), key=lambda tag: tag.score, reverse=True)
    
+    def handle_starttag(self, tag, attrs):
+        if tag != "a":
+            return
+        for attr_name, attr_value in attrs:
+            if attr_name == "href":
+                self.texte.append(attr_value)
+                break
+    
    def handle_data(self, data):
        self.texte.append(data)