add max text length

analyze.py

@@ -10,6 +10,7 @@ from presidio_analyzer import AnalyzerEngine, BatchAnalyzerEngine, RecognizerRes
 Row = dict[str, Any]
 T = TypeVar("T")
 BATCH_SIZE = 1
+MAX_TEXT_LENGTH = 3000
 batch_analyzer: Optional[BatchAnalyzerEngine] = None
 
 
@@ -124,7 +125,7 @@ def presidio_scan_entities(
     if batch_analyzer is None:
         batch_analyser = BatchAnalyzerEngine(AnalyzerEngine())
     rows_with_scanned_columns_only = (
-        {column_name: get_strings(row[column_name]) for column_name in scanned_columns} for row in rows
+        {column_name: get_strings(row[column_name])[:MAX_TEXT_LENGTH] for column_name in scanned_columns} for row in rows
     )
     for indices, batch in batched(rows_with_scanned_columns_only, BATCH_SIZE, with_indices=True):
         yield from analyze(