Improve parsing of malformed decimals (#1042)

Signed-off-by: Olunusi Best <[email protected]>
Co-authored-by: Aarni Koskela <[email protected]>

Author: olunusib

babel/numbers.py

@@ -1099,7 +1099,7 @@ def parse_decimal(
         raise NumberFormatError(f"{string!r} is not a valid decimal number") from exc
     if strict and group_symbol in string:
         proper = format_decimal(parsed, locale=locale, decimal_quantization=False, numbering_system=numbering_system)
-        if string != proper and string.rstrip('0') != (proper + decimal_symbol):
+        if string != proper and proper != _remove_trailing_zeros_after_decimal(string, decimal_symbol):
             try:
                 parsed_alt = decimal.Decimal(string.replace(decimal_symbol, '')
                                                    .replace(group_symbol, '.'))
@@ -1131,6 +1131,41 @@ def parse_decimal(
     return parsed
 
 
+def _remove_trailing_zeros_after_decimal(string: str, decimal_symbol: str) -> str:
+    """
+    Remove trailing zeros from the decimal part of a numeric string.
+
+    This function takes a string representing a numeric value and a decimal symbol.
+    It removes any trailing zeros that appear after the decimal symbol in the number.
+    If the decimal part becomes empty after removing trailing zeros, the decimal symbol
+    is also removed. If the string does not contain the decimal symbol, it is returned unchanged.
+
+    :param string: The numeric string from which to remove trailing zeros.
+    :type string: str
+    :param decimal_symbol: The symbol used to denote the decimal point.
+    :type decimal_symbol: str
+    :return: The numeric string with trailing zeros removed from its decimal part.
+    :rtype: str
+
+    Example:
+    >>> _remove_trailing_zeros_after_decimal("123.4500", ".")
+    '123.45'
+    >>> _remove_trailing_zeros_after_decimal("100.000", ".")
+    '100'
+    >>> _remove_trailing_zeros_after_decimal("100", ".")
+    '100'
+    """
+    integer_part, _, decimal_part = string.partition(decimal_symbol)
+
+    if decimal_part:
+        decimal_part = decimal_part.rstrip("0")
+        if decimal_part:
+            return integer_part + decimal_symbol + decimal_part
+        return integer_part
+
+    return string
+
+
 PREFIX_END = r'[^0-9@#.,]'
 NUMBER_TOKEN = r'[0-9@#.,E+]'
 

tests/test_numbers.py

@@ -195,7 +195,6 @@ def test_can_parse_decimals(self):
         with pytest.raises(numbers.UnsupportedNumberingSystemError):
             numbers.parse_decimal('2,109,998', locale='de', numbering_system="unknown")
 
-
     def test_parse_decimal_strict_mode(self):
         # Numbers with a misplaced grouping symbol should be rejected
         with pytest.raises(numbers.NumberFormatError) as info:
@@ -221,8 +220,19 @@ def test_parse_decimal_strict_mode(self):
         assert str(numbers.parse_decimal('1.001', locale='de', strict=True)) == '1001'
         # Trailing zeroes should be accepted
         assert str(numbers.parse_decimal('3.00', locale='en_US', strict=True)) == '3.00'
+        # Numbers with a grouping symbol and no trailing zeroes should be accepted
+        assert str(numbers.parse_decimal('3,400.6', locale='en_US', strict=True)) == '3400.6'
+        # Numbers with a grouping symbol and trailing zeroes (not all zeroes after decimal) should be accepted
+        assert str(numbers.parse_decimal('3,400.60', locale='en_US', strict=True)) == '3400.60'
+        # Numbers with a grouping symbol and trailing zeroes (all zeroes after decimal) should be accepted
+        assert str(numbers.parse_decimal('3,400.00', locale='en_US', strict=True)) == '3400.00'
+        assert str(numbers.parse_decimal('3,400.0000', locale='en_US', strict=True)) == '3400.0000'
+        # Numbers with a grouping symbol and no decimal part should be accepted
+        assert str(numbers.parse_decimal('3,800', locale='en_US', strict=True)) == '3800'
         # Numbers without any grouping symbol should be accepted
         assert str(numbers.parse_decimal('2000.1', locale='en_US', strict=True)) == '2000.1'
+        # Numbers without any grouping symbol and no decimal should be accepted
+        assert str(numbers.parse_decimal('2580', locale='en_US', strict=True)) == '2580'
         # High precision numbers should be accepted
         assert str(numbers.parse_decimal('5,000001', locale='fr', strict=True)) == '5.000001'