Addition of leniency parameter in predefined PhoneRecognizer (#1311)

microsoft · Feb 24, 2024 · 4c48b92 · 4c48b92
1 parent 173b527
commit 4c48b92
Show file tree

Hide file tree

Showing 2 changed files with 47 additions and 1 deletion.
diff --git a/presidio-analyzer/presidio_analyzer/predefined_recognizers/phone_recognizer.py b/presidio-analyzer/presidio_analyzer/predefined_recognizers/phone_recognizer.py
@@ -18,6 +18,8 @@ class PhoneRecognizer(LocalRecognizer):
     :param context: Base context words for enhancing the assurance scores.
     :param supported_language: Language this recognizer supports
     :param supported_regions: The regions for phone number matching and validation
+    :param leniency: The strictness level of phone number formats.
+    Accepts values from 0 to 3, where 0 is the lenient and 3 is the most strictest.
     """
 
     SCORE = 0.4
@@ -30,9 +32,11 @@ def __init__(
         supported_language: str = "en",
         # For all regions, use phonenumbers.SUPPORTED_REGIONS
         supported_regions=DEFAULT_SUPPORTED_REGIONS,
+        leniency: Optional[int] = 1,
     ):
         context = context if context else self.CONTEXT
         self.supported_regions = supported_regions
+        self.leniency = leniency
         super().__init__(
             supported_entities=self.get_supported_entities(),
             supported_language=supported_language,
@@ -59,7 +63,8 @@ def analyze(
         """
         results = []
         for region in self.supported_regions:
-            for match in phonenumbers.PhoneNumberMatcher(text, region, leniency=1):
+            for match in phonenumbers.PhoneNumberMatcher(text, region,
+                                                         leniency=self.leniency):
                 results += [
                     self._get_recognizer_result(match, text, region, nlp_artifacts)
                 ]

diff --git a/presidio-analyzer/tests/test_phone_recognizer.py b/presidio-analyzer/tests/test_phone_recognizer.py
@@ -41,3 +41,44 @@ def test_when_all_phones_then_succeed(
     assert len(results) == expected_len
     for i, (res, (st_pos, fn_pos)) in enumerate(zip(results, expected_positions)):
         assert_result(res, entities[i], st_pos, fn_pos, score)
+
+
+@pytest.mark.parametrize(
+    "text, expected_len, entities, expected_positions, score, leniency",
+    [
+        # fmt: off
+        ("My US number is (415) 555-0132, and my international one is415-555-0132",
+         1, ["PHONE_NUMBER"], ((16, 30), ), 0.4, 1),
+        ("My US number is (415) 555-0132, and my international one is415-555-0132",
+         2, ["PHONE_NUMBER", "PHONE_NUMBER"], ((16, 30), (59, 71), ), 0.4, 0),
+
+        ("My US number is (415) 555-0132, and my international one is 91-415-555-0132",
+         1, ["PHONE_NUMBER"], ((16, 30), ), 0.4, 2),
+        ("My US number is (415) 555-0132, and my international one is 91-415-555-0132",
+         2, ["PHONE_NUMBER", "PHONE_NUMBER"], ((16, 30), (60, 75), ), 0.4, 1),
+
+        ("My US number is (415) 555-0132, and my international one is +91 4155 550132",
+         1, ["PHONE_NUMBER"], ((16, 30), ), 0.4, 3),
+        ("My US number is (415) 555-0132, and my international one is +91 4155 550132",
+         2, ["PHONE_NUMBER", "PHONE_NUMBER"], ((16, 30), (60, 75), ), 0.4, 2),
+
+        ("My US number is (415) 555-0132, and my international one is +91 4155550132",
+         2, ["PHONE_NUMBER", "PHONE_NUMBER"], ((16, 30), (60, 74), ), 0.4, 3),
+        # fmt: on
+    ],
+)
+def test_when_phone_with_leniency_then_succeed(
+    spacy_nlp_engine,
+    text,
+    expected_len,
+    entities,
+    expected_positions,
+    score,
+    leniency,
+):
+    nlp_artifacts = spacy_nlp_engine.process_text(text, "en")
+    recognizer = PhoneRecognizer(leniency=leniency)
+    results = recognizer.analyze(text, entities, nlp_artifacts=nlp_artifacts)
+    assert len(results) == expected_len
+    for i, (res, (st_pos, fn_pos)) in enumerate(zip(results, expected_positions)):
+        assert_result(res, entities[i], st_pos, fn_pos, score)