fix: corrige is_valid_legal_process para retornar False em entradas malformadas

CHANGELOG.md

@@ -9,6 +9,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed
 
+- `is_valid_legal_process` retornava erro ou aceitava dígitos não ASCII para números de processo malformados; agora retorna `False`, e `format_legal_process` retorna `None` nesses casos [#793](https://github.com/brazilian-utils/python/issues/793)
 - Cálculo do dígito verificador em `is_valid_legal_process` e `generate_legal_process`, que rejeitava números de processo reais e gerava números inválidos [#785](https://github.com/brazilian-utils/python/pull/785)
 
 ## [2.5.0] - 2026-06-30

brutils/legal_process.py

@@ -56,7 +56,7 @@ def format_legal_process(legal_process_id: str) -> str | None:
         None
     """
 
-    if legal_process_id.isdigit() and len(legal_process_id) == 20:
+    if _is_twenty_digits(legal_process_id):
         capture_fields = r"(\d{7})(\d{2})(\d{4})(\d)(\d{2})(\d{4})"
         include_chars = r"\1-\2.\3.\4.\5.\6"
 
@@ -69,6 +69,29 @@ def format_legal_process(legal_process_id: str) -> str | None:
 ############
 
 
+def _is_twenty_digits(legal_process_id: str) -> bool:
+    """
+    Checks whether the string has exactly 20 ASCII digits.
+
+    `str.isdigit` is not enough: it also accepts characters such as "²" and
+    "٣", which `int` either rejects or reads as a different digit.
+
+    Args:
+        legal_process_id (str): The string to be checked.
+
+    Returns:
+        bool: True if the string has exactly 20 ASCII digits, False otherwise.
+
+    Example:
+        >>> _is_twenty_digits("68476506120233030000")
+        True
+        >>> _is_twenty_digits("6847650612023303000٣")
+        False
+    """
+
+    return re.fullmatch(r"\d{20}", legal_process_id, flags=re.ASCII) is not None
+
+
 def is_valid(legal_process_id: str) -> bool:
     """
     Check if a legal process ID is valid.
@@ -92,7 +115,13 @@ def is_valid(legal_process_id: str) -> bool:
         False
     """
 
+    if not isinstance(legal_process_id, str):
+        return False
+
     clean_legal_process_id = remove_symbols(legal_process_id)
+    if not _is_twenty_digits(clean_legal_process_id):
+        return False
+
     DD = clean_legal_process_id[7:9]
     J = clean_legal_process_id[13:14]
     TR = clean_legal_process_id[14:16]

tests/test_legal_process.py

@@ -94,6 +94,25 @@ def test_is_valid(self):
         self.assertIs(is_valid("455323423QQWEQWSsasd&*(()"), False)
         self.assertIsInstance(is_valid("455323423QQWEQWSsasd&*(()"), bool)
 
+    def test_is_valid_returns_false_for_malformed_input(self):
+        # 5046512-94.2016.4.04.7000 is a real, valid legal process ID
+        self.assertIs(is_valid("50465129420164047000"), True)
+        # Truncated: the court (TR) or origin (OOOO) fields are empty
+        self.assertIs(is_valid("50465129420164"), False)
+        self.assertIs(is_valid("5046512942016404"), False)
+        # Letters where digits are expected
+        self.assertIs(is_valid("ABCDEFG9420164047000"), False)
+        # Characters that str.isdigit accepts but are not ASCII digits
+        self.assertIs(is_valid("²0465129420164047000"), False)
+        self.assertIs(is_valid("5046512942016404700٠"), False)
+        # Not a string
+        self.assertIs(is_valid(50465129420164047000), False)
+        self.assertIs(is_valid(None), False)
+
+    def test_format_legal_process_rejects_non_ascii_digits(self):
+        self.assertIsNone(format_legal_process("5046512942016404700٠"))
+        self.assertIsNone(format_legal_process("²" * 20))
+
 
 if __name__ == "__main__":
     main()