Fix crash tokenizing with empty word_to_id (#72)

xhluca · Oct 25, 2024 · f77e90d · f77e90d
1 parent e06ecf4
commit f77e90d
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/bm25s/tokenization.py b/bm25s/tokenization.py
@@ -273,7 +273,7 @@ def streaming_tokenize(
 
             if len(doc_ids) == 0 and allow_empty is True:
                 if update_vocab is True and "" not in self.word_to_id:
-                    self.word_to_id[""] = max(self.word_to_id.values()) + 1
+                    self.word_to_id[""] = max(self.word_to_id.values(), default=0) + 1
 
                 # get the ID for the empty string
                 if "" in self.word_to_id: