fixed parser not handling overlong float literals correctly

also a few other edge cases (see #65)

README.md

@@ -203,7 +203,7 @@ UTF-8 decoding is performed using a state machine based on Bjoern Hoehrmann's '[
 - **[@Reedbeta](https://github.com/Reedbeta)** - Fixed a bug and added additional Visual Studio debugger native visualizers
 - **[@shdnx](https://github.com/shdnx)** - Fixed a bug on GCC 8.2.0 and some meson config issues
 - **[@sobczyk](https://github.com/sobczyk)** - Reported some bugs
-- **[@sneves](https://github.com/sneves)** - Reported a bug
+- **[@sneves](https://github.com/sneves)** - Helped fix a number of parser bugs
 - **[@traversaro](https://github.com/traversaro)** - Added vcpkg support and reported a bunch of bugs
 - **[@ximion](https://github.com/ximion)** - Added support for installation with meson
 - **[@whiterabbit963](https://github.com/whiterabbit963)** - Fixed a bug with value_or conversions

examples/error_printer.cpp

@@ -98,7 +98,10 @@ namespace
 		R"(val = 1_0_ )"sv,
 		R"(val = 999999999999999999999999999999999999 )"sv,
 		R"(val = 9223372036854775808 )"sv,
-		R"(val = 01 )"sv
+		R"(val = 01 )"sv,
+
+		"########## floats"sv,
+		R"(val = 9999999999999999999999999999999999999999999999999999999999999995.0)"sv,
 	};
 }
 

include/toml++/toml_parser.h

@@ -315,6 +315,13 @@ TOML_NAMESPACE_START
 			{
 				return is_err ? const_table_iterator{} : table().cend();
 			}
+
+			/// \brief Prints the held error or table object out to a text stream.
+			template <typename Char>
+			friend std::basic_ostream<Char>& operator << (std::basic_ostream<Char>& os, const parse_result& result)
+			{
+				return result.is_err ? (os << result.error()) : (os << result.table());
+			}
 	};
 
 	#else

include/toml++/toml_parser.hpp

@@ -559,6 +559,7 @@ TOML_IMPL_NAMESPACE_START
 				{
 					if (consume_line_break())
 						return true;
+					return_if_error({});
 
 					if constexpr (TOML_LANG_AT_LEAST(1, 0, 0))
 					{
@@ -896,7 +897,7 @@ TOML_IMPL_NAMESPACE_START
 				std::string str;
 				do
 				{
-					assert_not_error();
+					return_if_error({});
 
 					// handle closing delimiters
 					if (*cp == U'\'')
@@ -957,6 +958,7 @@ TOML_IMPL_NAMESPACE_START
 					if (multi_line && is_line_break(*cp))
 					{
 						consume_line_break();
+						return_if_error({});
 						str += '\n';
 						continue;
 					}
@@ -1194,10 +1196,6 @@ TOML_IMPL_NAMESPACE_START
 						else if (!is_match(*prev, U'e', U'E'))
 							set_error_and_return_default("expected exponent digit, saw '"sv, to_sv(*cp), "'"sv);
 					}
-					else if (length == sizeof(chars))
-						set_error_and_return_default(
-							"exceeds maximum length of "sv, static_cast<uint64_t>(sizeof(chars)), " characters"sv
-						);
 					else if (is_decimal_digit(*cp))
 					{
 						if (!seen_decimal)
@@ -1211,6 +1209,11 @@ TOML_IMPL_NAMESPACE_START
 					else
 						set_error_and_return_default("expected decimal digit, saw '"sv, to_sv(*cp), "'"sv);
 
+					if (length == sizeof(chars))
+						set_error_and_return_default(
+							"exceeds maximum length of "sv, static_cast<uint64_t>(sizeof(chars)), " characters"sv
+						);
+
 					chars[length++] = static_cast<char>(cp->bytes[0]);
 					prev = cp;
 					advance_and_return_if_error({});
@@ -2719,7 +2722,6 @@ TOML_IMPL_NAMESPACE_START
 						|| consume_line_break()
 						|| consume_comment())
 						continue;
-
 					return_if_error();
 
 					// [tables]
@@ -2944,6 +2946,7 @@ TOML_IMPL_NAMESPACE_START
 				while (consume_leading_whitespace())
 					continue;
 			}
+			return_if_error({});
 			set_error_and_return_if_eof({});
 
 			// commas - only legal after a key-value pair

tests/user_feedback.cpp

@@ -92,17 +92,26 @@ TEST_CASE("user feedback")
 	{
 		// see: https://github.com/marzer/tomlplusplus/issues/65
 
-		// this tests two things:
+		// these test a number of things
 		// - a comment at EOF
-		// - a malformed UTF-8 sequence
+		// - a malformed UTF-8 sequence in a comment
-		// 
+		// - a malformed UTF-8 sequence during a KVP
-		// it should fail to parse, but correctly issue an error (not crash!)
+		// - overlong numeric literals
-		parsing_should_fail(FILE_LINE_ARGS, "#\xf1\x63");
+		// all should fail to parse, but correctly issue an error (not crash!)
 		
-		// a malformed UTF-8 sequence during a KVP
+		parsing_should_fail(FILE_LINE_ARGS, "#\xf1\x63");
-		// 
-		// it should fail to parse, but correctly issue an error (not crash!)
 		parsing_should_fail(FILE_LINE_ARGS, "1= 0x6cA#+\xf1");
+		parsing_should_fail(FILE_LINE_ARGS, "p=06:06:06#\x0b\xff");
+		parsing_should_fail(FILE_LINE_ARGS, "''''d' 't' '+o\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c"
+			"\x0c\x0c\x0c\x0c\x0c\r\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c"
+			"\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c"
+			"\x0c\x0c\x0c\x0c\x0c\x0c\x0cop1\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c"
+			"\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c"
+			"\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c\x0c' 'ml'\n\n%\x87"
+		);
+		parsing_should_fail(FILE_LINE_ARGS,
+			R"(t =[ 9, 2, 1,"r", 9999999999999999999999999999999999999999999999999999999999999995.0 ])"
+		);
 	}
 }
 

toml.hpp

@@ -7303,6 +7303,12 @@ TOML_NAMESPACE_START
 			{
 				return is_err ? const_table_iterator{} : table().cend();
 			}
+
+			template <typename Char>
+			friend std::basic_ostream<Char>& operator << (std::basic_ostream<Char>& os, const parse_result& result)
+			{
+				return result.is_err ? (os << result.error()) : (os << result.table());
+			}
 	};
 
 	#else
@@ -9276,6 +9282,7 @@ TOML_IMPL_NAMESPACE_START
 				{
 					if (consume_line_break())
 						return true;
+					return_if_error({});
 
 					if constexpr (TOML_LANG_AT_LEAST(1, 0, 0))
 					{
@@ -9613,7 +9620,7 @@ TOML_IMPL_NAMESPACE_START
 				std::string str;
 				do
 				{
-					assert_not_error();
+					return_if_error({});
 
 					// handle closing delimiters
 					if (*cp == U'\'')
@@ -9674,6 +9681,7 @@ TOML_IMPL_NAMESPACE_START
 					if (multi_line && is_line_break(*cp))
 					{
 						consume_line_break();
+						return_if_error({});
 						str += '\n';
 						continue;
 					}
@@ -9911,10 +9919,6 @@ TOML_IMPL_NAMESPACE_START
 						else if (!is_match(*prev, U'e', U'E'))
 							set_error_and_return_default("expected exponent digit, saw '"sv, to_sv(*cp), "'"sv);
 					}
-					else if (length == sizeof(chars))
-						set_error_and_return_default(
-							"exceeds maximum length of "sv, static_cast<uint64_t>(sizeof(chars)), " characters"sv
-						);
 					else if (is_decimal_digit(*cp))
 					{
 						if (!seen_decimal)
@@ -9928,6 +9932,11 @@ TOML_IMPL_NAMESPACE_START
 					else
 						set_error_and_return_default("expected decimal digit, saw '"sv, to_sv(*cp), "'"sv);
 
+					if (length == sizeof(chars))
+						set_error_and_return_default(
+							"exceeds maximum length of "sv, static_cast<uint64_t>(sizeof(chars)), " characters"sv
+						);
+
 					chars[length++] = static_cast<char>(cp->bytes[0]);
 					prev = cp;
 					advance_and_return_if_error({});
@@ -11433,7 +11442,6 @@ TOML_IMPL_NAMESPACE_START
 						|| consume_line_break()
 						|| consume_comment())
 						continue;
-
 					return_if_error();
 
 					// [tables]
@@ -11657,6 +11665,7 @@ TOML_IMPL_NAMESPACE_START
 				while (consume_leading_whitespace())
 					continue;
 			}
+			return_if_error({});
 			set_error_and_return_if_eof({});
 
 			// commas - only legal after a key-value pair