diff --git a/src/word_stats.py b/src/word_stats.py index ae21618..46a4ca1 100644 --- a/src/word_stats.py +++ b/src/word_stats.py @@ -110,7 +110,8 @@ class WordStats: statement = """SELECT msd, text, frequency FROM UniqWords WHERE lemma=:lemma ORDER BY frequency DESC""" for msd, text, _f in self.db.execute(statement, {'lemma': lemma}): - yield (msd, text) + if (msd, text) not in counted_texts: + yield (msd, text) def num_words(self, lemma, msd0): statement = "SELECT frequency FROM WordCount WHERE lemma=? AND msd0=? LIMIT 1"