Add filter func

Author: hupe1980

aisploit/classifiers/presidio/presidio_analyser.py

@@ -1,19 +1,20 @@
 from dataclasses import dataclass, field
-from typing import List
+from typing import Callable, List, Optional
 
 from presidio_analyzer import AnalyzerEngine, EntityRecognizer, RecognizerResult
 
 from ...core import BaseTextClassifier, Score
 
 
-@dataclass
+@dataclass(kw_only=True)
 class PresidioAnalyserClassifier(BaseTextClassifier[List[RecognizerResult]]):
     """A text classifier using the Presidio Analyzer for detecting Personally Identifiable Information (PII)."""
 
     language: str = "en"
     entities: List[str] | None = None
     threshold: float = 0.7
     additional_recognizers: List[EntityRecognizer] = field(default_factory=list)
+    filter_func: Optional[Callable[[str, RecognizerResult], bool]] = None
     tags: List[str] = field(default_factory=lambda: ["leakage"], init=False)
 
     def __post_init__(self) -> None:
@@ -36,6 +37,9 @@ def score(self, input: str, _: List[str] | None = None) -> Score[List[Recognizer
         """
         results = self._analyzer.analyze(text=input, entities=self.entities, language=self.language)
 
+        if self.filter_func:
+            results = [result for result in results if self.filter_func(input, result)]
+
         return Score[List[RecognizerResult]](
             flagged=len(results) > 0,
             value=results,

examples/classifier.ipynb

@@ -85,7 +85,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 2,
+   "execution_count": 4,
    "metadata": {},
    "outputs": [
     {
@@ -94,7 +94,7 @@
        "Score(flagged=True, value=[type: PERSON, start: 11, end: 19, score: 0.85, type: PHONE_NUMBER, start: 43, end: 55, score: 0.75], description='Returns True if entities are found in the input', explanation='Found 2 entities in input')"
       ]
      },
-     "execution_count": 2,
+     "execution_count": 4,
      "metadata": {},
      "output_type": "execute_result"
     }