Make python text-format reject octal floats

These are disallowed by the spec, and rejected by other languages (and python implementations). PiperOrigin-RevId: 707222813
protocolbuffers · Dec 17, 2024 · 8f2ad12 · 8f2ad12
1 parent 7004a92
commit 8f2ad12
Show file tree

Hide file tree

Showing 5 changed files with 14 additions and 6 deletions.
diff --git a/conformance/text_format_failure_list_python.txt b/conformance/text_format_failure_list_python.txt
@@ -1,8 +1,6 @@
 # This is the list of text format conformance tests that are known to fail right
 # now.
 # TODO: These should be fixed.
-Required.*.TextFormatInput.FloatFieldNoNegativeOctal                                                               # Should have failed to parse, but didn't.
-Required.*.TextFormatInput.FloatFieldNoOctal                                                                       # Should have failed to parse, but didn't.
 Required.*.TextFormatInput.StringLiteralBasicEscapesBytes.ProtobufOutput                                           # Output was not equivalent to reference message: modified: optional_bytes: "\007\010\014\n\r\t\013?\\\'\"" -> "\007\010\014\n\r\t
 Required.*.TextFormatInput.StringLiteralBasicEscapesBytes.TextFormatOutput                                         # Output was not equivalent to reference message: modified: optional_bytes: "\007\010\014\n\r\t\013?\\\'\"" -> "\007\010\014\n\r\t
 Required.*.TextFormatInput.StringLiteralBasicEscapesString.ProtobufOutput                                          # Output was not equivalent to reference message: modified: optional_string: "\007\010\014\n\r\t\013?\\\'\"" -> "\007\010\014\n\r\

diff --git a/conformance/text_format_failure_list_python_cpp.txt b/conformance/text_format_failure_list_python_cpp.txt
@@ -1,5 +1,3 @@
-Required.*.TextFormatInput.FloatFieldNoNegativeOctal                                                               # Should have failed to parse, but didn't.
-Required.*.TextFormatInput.FloatFieldNoOctal                                                                       # Should have failed to parse, but didn't.
 Required.*.TextFormatInput.StringLiteralBasicEscapesBytes.ProtobufOutput                                           # Output was not equivalent to reference message: modified: optional_bytes: "\007\010\014\n\r\t\013?\\\'\"" -> "\007\010\014\n\r\t
 Required.*.TextFormatInput.StringLiteralBasicEscapesBytes.TextFormatOutput                                         # Output was not equivalent to reference message: modified: optional_bytes: "\007\010\014\n\r\t\013?\\\'\"" -> "\007\010\014\n\r\t
 Required.*.TextFormatInput.StringLiteralBasicEscapesString.ProtobufOutput                                          # Output was not equivalent to reference message: modified: optional_string: "\007\010\014\n\r\t\013?\\\'\"" -> "\007\010\014\n\r\

diff --git a/conformance/text_format_failure_list_python_upb.txt b/conformance/text_format_failure_list_python_upb.txt
@@ -1,8 +1,6 @@
 # This is the list of text format conformance tests that are known to fail right
 # now.
 # TODO: These should be fixed.
-Required.*.TextFormatInput.FloatFieldNoNegativeOctal                                                               # Should have failed to parse, but didn't.
-Required.*.TextFormatInput.FloatFieldNoOctal                                                                       # Should have failed to parse, but didn't.
 Required.*.TextFormatInput.StringLiteralBasicEscapesBytes.ProtobufOutput                                           # Output was not equivalent to reference message: modified: optional_bytes: "\007\010\014\n\r\t\013?\\\'\"" -> "\007\010\014\n\r\t
 Required.*.TextFormatInput.StringLiteralBasicEscapesBytes.TextFormatOutput                                         # Output was not equivalent to reference message: modified: optional_bytes: "\007\010\014\n\r\t\013?\\\'\"" -> "\007\010\014\n\r\t
 Required.*.TextFormatInput.StringLiteralBasicEscapesString.ProtobufOutput                                          # Output was not equivalent to reference message: modified: optional_string: "\007\010\014\n\r\t\013?\\\'\"" -> "\007\010\014\n\r\

diff --git a/python/google/protobuf/internal/text_format_test.py b/python/google/protobuf/internal/text_format_test.py
@@ -2468,6 +2468,17 @@ def testConsumeOctalIntegers(self):
     self.assertEqual(1, tokenizer.ConsumeInteger())
     self.assertTrue(tokenizer.AtEnd())
 
+  @parameterized.parameters('00', '09', '01.123', '-00', '-09', '-01.234')
+  def testConsumeOctalFloats(self, text):
+    """Test rejection of for octal-formatted floats."""
+    tokenizer = text_format.Tokenizer([text])
+
+    self.assertRaisesRegex(
+        text_format.ParseError,
+        'Invalid octal float: %s' % text,
+        tokenizer.ConsumeFloat,
+    )
+
   def testConsumeByteString(self):
     text = '"string1\''
     tokenizer = text_format.Tokenizer(text.splitlines())

diff --git a/python/google/protobuf/text_format.py b/python/google/protobuf/text_format.py
@@ -42,6 +42,7 @@
                      type_checkers.Int64ValueChecker())
 _FLOAT_INFINITY = re.compile('-?inf(?:inity)?f?$', re.IGNORECASE)
 _FLOAT_NAN = re.compile('nanf?$', re.IGNORECASE)
+_FLOAT_OCTAL_PREFIX = re.compile('-?0[0-9]+')
 _QUOTES = frozenset(("'", '"'))
 _ANY_FULL_TYPE_NAME = 'google.protobuf.Any'
 _DEBUG_STRING_SILENT_MARKER = '\t '
@@ -1791,6 +1792,8 @@ def ParseFloat(text):
   Raises:
     ValueError: If a floating point number couldn't be parsed.
   """
+  if _FLOAT_OCTAL_PREFIX.match(text):
+    raise ValueError('Invalid octal float: %s' % text)
   try:
     # Assume Python compatible syntax.
     return float(text)