change handling of words with less than 3 characters

2022-05-28 15:19:00 +02:00 · 2022-05-28 15:19:00 +02:00 · 0dc6fd60d5
parent f09d42433f
commit 0dc6fd60d5
1 changed files with 3 additions and 2 deletions
--- a/tagger.py
+++ b/tagger.py
@ -46,8 +46,9 @@ class FileScanner(HTMLParser):
                continue
            if word in title_words:
                score *= 4
-            if len(word) <= 3:
-                score //= 2
+            word_length = len(word)
+            if word_length <= 3:
+                score = int(score * word_length / 4)
            upper_letters_count = len(UPPER_CHECK.findall(tag_name))
            score += upper_letters_count * 5
            if word not in words_with_usage: