fix(es): upper-case NIF/NIE before mod-23 checksum so valid lowercase IDs are detected

presidio-analyzer/presidio_analyzer/predefined_recognizers/country_specific/spain/es_nie_recognizer.py

@@ -60,7 +60,7 @@ def validate_result(self, pattern_text: str) -> bool:
 
         pattern_text = EntityRecognizer.sanitize_value(
             pattern_text, self.replacement_pairs
-        )
+        ).upper()
 
         letters = "TRWAGMYFPDXBNJZSQVHLCKE"
         letter = pattern_text[-1]

presidio-analyzer/presidio_analyzer/predefined_recognizers/country_specific/spain/es_nif_recognizer.py

@@ -53,7 +53,7 @@ def __init__(
     def validate_result(self, pattern_text: str) -> bool:  # noqa: D102
         pattern_text = EntityRecognizer.sanitize_value(
             pattern_text, self.replacement_pairs
-        )
+        ).upper()
         letter = pattern_text[-1]
         number = int("".join(filter(str.isdigit, pattern_text)))
         letters = "TRWAGMYFPDXBNJZSQVHLCKE"

presidio-analyzer/tests/test_es_nie_recognizer.py

@@ -28,11 +28,19 @@ def entities():
         ("Z8078221M en mi NIE", 1, ((0, 9),),),
         ("Mi Número de identificación de extranjero es Y8063915-Z", 1, \
             ((45, 55),),),
+        # valid NIE in lowercase (prefix and control letter are upper-cased
+        # before checksum)
+        ("x9613851n", 1, ((0, 9),),),
+        ("z8078221m", 1, ((0, 9),),),
+        # uppercase still detected
+        ("X9613851N", 1, ((0, 9),),),
         # invalid NIE scores
         ("Y8063915Q", 0, ()),
         ("Y806391Q", 0, ()),
         ("58063915Q", 0, ()),
         ("W8063915Q", 0, ()),
+        # invalid checksum stays rejected regardless of case
+        ("x9613851q", 0, ()),
     ],
 )
 def test_when_all_es_nie_then_succeed(

presidio-analyzer/tests/test_es_nif_recognizer.py

@@ -24,8 +24,15 @@ def entities():
         ("1111111-G", 1, ((0, 9),),),
         ("1111111G", 1, ((0, 8),),),
         ("01111111G", 1, ((0, 9),),),
+        # valid NIF in lowercase (control letter is upper-cased before checksum)
+        ("55555555k", 1, ((0, 9),),),
+        ("12345678z", 1, ((0, 9),),),
+        # uppercase still detected
+        ("12345678Z", 1, ((0, 9),),),
         # invalid NIF scores
         ("401-023-2138", 0, ()),
+        # invalid checksum stays rejected regardless of case
+        ("12345678a", 0, ()),
         # fmt: on
     ],
 )