FixedUserSerachAndSupplierPizza9

extractors/supplier_extractor.py

@@ -130,7 +130,7 @@ class ExpenseSupplierExtractor:
         self.sup_norm = [normalize_text(s) for s in suppliers]
         self.sup_tokens = [s.split() for s in self.sup_norm]
         self.sup_num_sets = [self.numeric_tokens(s) for s in self.sup_norm]
-        self.sup_number_tokens = {token for supplier in self.sup_tokens for token in supplier if token.isdigit()}
+        self.sup_number_tokens = {num for nums in self.sup_num_sets for num in nums}
         self.supplier_lexicon = [
             token
             for token in sorted({tok for tokens in self.sup_tokens for tok in tokens})

extractors/user_extractor.py

@@ -16,7 +16,7 @@ class ExpenseUserExtractor:
         users: list[str],
         suppliers: list[str],
         model: Any = None,
-        threshold: float = 0.5,
+        threshold: float = 0.33,
     ) -> None:
         self.users = users
         self.threshold = threshold