debug

analyze.py

@@ -72,12 +72,15 @@ def _simple_analyze_iterator_cache(
     score_threshold: float,
     cache: dict[str, list[RecognizerResult]],
 ) -> list[list[RecognizerResult]]:
+    print(cache)
+    print(texts)
     not_cached_results = iter(
         batch_analyzer.analyze_iterator(
             (text for text in texts if text not in cache), language=language, score_threshold=score_threshold
         )
     )
     results = [cache[text] if text in cache else next(not_cached_results) for text in texts]
+    print(results)
     # cache the last results
     cache.clear()
     cache.update(dict(zip(texts, results)))
@@ -98,7 +101,6 @@ def analyze(
         for example in batch
         for column_name, columns_description in zip(scanned_columns, columns_descriptions)
     ]
-    print(texts)
     return [
         PresidioEntity(
             text=mask(texts[i][recognizer_result.start : recognizer_result.end]),