Get patterns from regex match in ITIN recognizer

presidio-analyzer/presidio_analyzer/pattern_recognizer.py

@@ -187,17 +187,20 @@ def __analyze_patterns(
 
             for match in matches:
                 start, end = match.span()
+
+                pattern_from_match = self.get_pattern_from_match(pattern, match)
+
                 current_match = text[start:end]
 
                 # Skip empty results
                 if current_match == "":
                     continue
 
-                score = pattern.score
+                score = pattern_from_match.score
 
                 validation_result = self.validate_result(current_match)
                 description = self.build_regex_explanation(
-                    self.name, pattern.name, pattern.regex, score, validation_result
+                    self.name, pattern_from_match.name, pattern.regex, score, validation_result
                 )
                 pattern_result = RecognizerResult(
                     entity_type=self.supported_entities[0],
@@ -230,6 +233,18 @@ def __analyze_patterns(
         results = EntityRecognizer.remove_duplicates(results)
         return results
 
+    def get_pattern_from_match(
+        self, pattern: Pattern, match: re.Match
+    ) -> Pattern:
+        """
+        Return a new Pattern based on the matched regex info e.g., the named groups in the regex.
+
+        :param pattern: the pattern initially found.
+        :param math: regex match
+        :return: A new pattern based on the regex match info.
+        """
+        return pattern
+
     def to_dict(self) -> Dict:
         """Serialize instance into a dictionary."""
         return_dict = super().to_dict()

presidio-analyzer/presidio_analyzer/predefined_recognizers/us_itin_recognizer.py

@@ -1,7 +1,7 @@
 from typing import Optional, List
 
 from presidio_analyzer import Pattern, PatternRecognizer
-
+from regex import Match
 
 class UsItinRecognizer(PatternRecognizer):
     """
@@ -14,19 +14,9 @@ class UsItinRecognizer(PatternRecognizer):
     """
 
     PATTERNS = [
-        Pattern(
-            "Itin (very weak)",
-            r"\b9\d{2}[- ](5\d|6[0-5]|7\d|8[0-8]|9([0-2]|[4-9]))\d{4}\b|\b9\d{2}(5\d|6[0-5]|7\d|8[0-8]|9([0-2]|[4-9]))[- ]\d{4}\b",  # noqa: E501
-            0.05,
-        ),
-        Pattern(
-            "Itin (weak)",
-            r"\b9\d{2}(5\d|6[0-5]|7\d|8[0-8]|9([0-2]|[4-9]))\d{4}\b",  # noqa: E501
-            0.3,
-        ),
         Pattern(
             "Itin (medium)",
-            r"\b9\d{2}[- ](5\d|6[0-5]|7\d|8[0-8]|9([0-2]|[4-9]))[- ]\d{4}\b",  # noqa: E501
+            r"\b9\d{2}(?P<firstSeparator>[- ]?)(5\d|6[0-5]|7\d|8[0-8]|9([0-2]|[4-9]))(?P<secondSeparator>[- ]?)\d{4}\b",  # noqa: E501
             0.5,
         ),
     ]
@@ -48,3 +38,25 @@ def __init__(
             context=context,
             supported_language=supported_language,
         )
+
+    def get_pattern_from_match(
+        self, pattern: Pattern, match: Match
+    ) -> Pattern:
+        first_separator = match.group('firstSeparator')
+        second_separator = match.group('secondSeparator')
+
+        if first_separator and second_separator:
+            return pattern
+
+        if not first_separator and not second_separator:
+            return Pattern(
+                "Itin (weak)",
+                pattern.regex,
+                0.3
+            )
+
+        return Pattern(
+            "Itin (very weak)",
+            pattern.regex,
+            0.05
+        )

presidio-analyzer/tests/test_us_itin_recognizer.py

@@ -18,11 +18,17 @@ def entities():
     "text, expected_len, expected_positions, expected_score_ranges",
     [
         # fmt: off
-        ("911-701234 91170-1234", 2, ((0, 10), (11, 21),), ((0.0, 0.3), (0.0, 0.3),),),
+        ("911-701234 91170-1234", 2,
+            ((0, 10), (11, 21),),
+            ((0.0, 0.3), (0.0, 0.3),),),
+        ("911 701234 91170 1234", 2,
+            ((0, 10), (11, 21),),
+            ((0.0, 0.3), (0.0, 0.3),),),
         ("911701234", 1, ((0, 9),), ((0.3, 0.4),),),
         ("911-70-1234", 1, ((0, 11),), ((0.5, 0.6),),),
         ("911-53-1234", 1, ((0, 11),), ((0.5, 0.6),),),
         ("911-64-1234", 1, ((0, 11),), ((0.5, 0.6),),),
+        ("911 63 7534", 1, ((0, 11),), ((0.5, 0.6),),),
         ("911-89-1234", 0, (), (),),
         ("my tax id 911-89-1234", 0, (), (),),
         # fmt: on