fixed date parsing when the year has a leading zero (closes #130)

also: - fixed omitting value part from hex/bin/oct being accepted without error (closes #129) - added spec bug github template
marzer · Jan 5, 2022 · de2413e · de2413e
1 parent b41e12f
commit de2413e
Show file tree

Hide file tree

Showing 11 changed files with 125 additions and 25 deletions.
diff --git a/.github/ISSUE_TEMPLATE/bug_report.md b/.github/ISSUE_TEMPLATE/bug_report.md
@@ -1,8 +1,8 @@
 ---
 name: Bug report
-about: Found a bug? Help me squash it.
+about: Regular ol' bugs.
 title: ''
-labels: bug
+labels: [ "bug" ]
 assignees: marzer
 
 ---

diff --git a/.github/ISSUE_TEMPLATE/spec_bug_report.md b/.github/ISSUE_TEMPLATE/spec_bug_report.md
@@ -0,0 +1,55 @@
+---
+name: TOML spec conformance bug
+about: Bugs relating to the library's TOML spec conformance (or lack thereof).
+title: ''
+labels: [ "bug", "TOML spec" ]
+assignees: marzer
+
+---
+
+<!--
+    Replace the HTML/TOML comments below with the requested information.
+    Please don't delete this template and roll your own!
+
+    Thanks for contributing!
+-->
+
+
+## The non-conforming TOML snippet
+```toml
+
+# your TOML here 
+
+```
+
+
+## What you expected
+<!--
+    e.g. a link to, or snippet from, the TOML spec, or some reasonable description thereof
+-->
+
+
+## What you got
+<!--
+    e.g toml-test output
+-->
+
+
+## Environment
+**toml++ version and/or commit hash:**  
+<!--
+    If you're using the single-header version of the library, the version number is right at the top of the file.
+    Otherwise you can find it by opening toml++/impl/version.h; it'll be represented by three defines -
+    TOML_LIB_MAJOR, TOML_LIB_MINOR and TOML_LIB_PATCH.
+    
+    If you're not using any particular release and are instead just living large at HEAD of master, the commit hash
+    would be super helpful too, though it's not critical.
+    
+-->
+
+**Any other useful information:**
+<!--
+    Anything else you think will help me fix the issue. Since this report is for general spec conformance handling
+    you probably don't need to worry about compiler versions, compilation flags, et cetera, though include them if
+    you feel they're relevant.
+-->
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -35,9 +35,8 @@ code changes at callsites or in build systems are indicated with ⚠&#xFE0F;.
 - fixed missing `TOML_API` on interfaces
 - fixed parser not correctly round-tripping the format of binary and octal integers in some cases
 - fixed strong exception guarantee edge-cases in `toml::table` and `toml::array`
-- fixed some incorrect unicode scalar sequence transformations (#125) (@moorereason)
-- fixed extended-precision fractional times causing parse error instead of truncating per the spec (#127) (@moorereason)
-- fixed some non-spec vertical whitespace being accepted as line breaks (#128) (@moorereason)
+- fixed some incorrect unicode scalar sequence transformations (#125)
+- fixed a number of spec conformance issues (#127, #128, #129) (@moorereason)
 
 #### Additions:
 - added `operator->` to `toml::value` for class types

diff --git a/include/toml++/impl/formatter.inl b/include/toml++/impl/formatter.inl
@@ -23,7 +23,7 @@
 
 TOML_IMPL_NAMESPACE_START
 {
-	enum class formatted_string_traits : unsigned
+	enum class TOML_CLOSED_FLAGS_ENUM formatted_string_traits : unsigned
 	{
 		none,
 		line_breaks	  = 1u << 0, // \n

diff --git a/include/toml++/impl/forward_declarations.h b/include/toml++/impl/forward_declarations.h
@@ -264,7 +264,7 @@ TOML_NAMESPACE_START // abi namespace
 	}
 
 	/// \brief Metadata associated with TOML values.
-	enum class TOML_OPEN_FLAGS_ENUM value_flags : uint16_t
+	enum class TOML_OPEN_FLAGS_ENUM value_flags : uint16_t // being an "OPEN" flags enum is not an error
 	{
 		/// \brief None.
 		none,

diff --git a/include/toml++/impl/parser.inl b/include/toml++/impl/parser.inl
@@ -2088,6 +2088,9 @@ TOML_IMPL_NAMESPACE_START
 				if (*cp != traits::prefix_codepoint)
 					set_error_and_return_default("expected '"sv, traits::prefix, "', saw '"sv, to_sv(*cp), "'"sv);
 				advance_and_return_if_error_or_eof({});
+
+				if (!traits::is_digit(*cp))
+					set_error_and_return_default("expected digit, saw '"sv, to_sv(*cp), "'"sv);
 			}
 
 			// consume value chars
@@ -2400,7 +2403,6 @@ TOML_IMPL_NAMESPACE_START
 		node_ptr parse_inline_table();
 
 		TOML_NODISCARD
-		TOML_NEVER_INLINE
 		node_ptr parse_value_known_prefixes()
 		{
 			return_if_error({});
@@ -2505,8 +2507,8 @@ TOML_IMPL_NAMESPACE_START
 					begins_zero	 = 1 << 14,
 
 					signs_msk  = has_plus | has_minus,
-					bzero_msk  = begins_zero | has_digits,
-					bdigit_msk = begins_digit | has_digits,
+					bdigit_msk = has_digits | begins_digit,
+					bzero_msk  = bdigit_msk | begins_zero,
 				};
 				value_traits traits	 = has_nothing;
 				const auto has_any	 = [&](auto t) noexcept { return (traits & t) != has_nothing; };
@@ -2516,7 +2518,11 @@ TOML_IMPL_NAMESPACE_START
 				// examine the first character to get the 'begins with' traits
 				// (good fail-fast opportunity; all the remaining types begin with numeric digits or signs)
 				if (is_decimal_digit(*cp))
-					add_trait(*cp == U'0' ? begins_zero : begins_digit);
+				{
+					add_trait(begins_digit);
+					if (*cp == U'0')
+						add_trait(begins_zero);
+				}
 				else if (is_match(*cp, U'+', U'-'))
 					add_trait(begins_sign);
 				else
@@ -2601,8 +2607,12 @@ TOML_IMPL_NAMESPACE_START
 				return_if_error({});
 
 				// force further scanning if this could have been a date-time with a space instead of a T
-				if (char_count == 10u && traits == (bdigit_msk | has_minus) && chars[4] == U'-' && chars[7] == U'-'
-					&& !is_eof() && *cp == U' ')
+				if (char_count == 10u									 //
+					&& (traits | begins_zero) == (bzero_msk | has_minus) //
+					&& chars[4] == U'-'									 //
+					&& chars[7] == U'-'									 //
+					&& !is_eof()										 //
+					&& *cp == U' ')
 				{
 					const auto pre_advance_count = advance_count;
 					const auto pre_scan_traits	 = traits;
@@ -2646,7 +2656,7 @@ TOML_IMPL_NAMESPACE_START
 				// the only valid value type is an integer.
 				if (char_count == 1u)
 				{
-					if (has_any(begins_zero | begins_digit))
+					if (has_any(begins_digit))
 					{
 						val.reset(new value{ static_cast<int64_t>(chars[0] - U'0') });
 						advance(); // skip the digit
@@ -2692,7 +2702,7 @@ TOML_IMPL_NAMESPACE_START
 					val.reset(new value{ i });
 					val->ref_cast<int64_t>().flags(flags);
 				}
-				else if (has_any(has_e) || (has_any(begins_zero | begins_digit) && chars[1] == U'.'))
+				else if (has_any(has_e) || (has_any(begins_digit) && chars[1] == U'.'))
 					val.reset(new value{ parse_float() });
 				else if (has_any(begins_sign))
 				{

diff --git a/tests/parsing_integers.cpp b/tests/parsing_integers.cpp
@@ -152,6 +152,11 @@ TEST_CASE("parsing - integers (hex, bin, oct)")
 	parsing_should_fail(FILE_LINE_ARGS, "val = 0o1000000000000000000000"sv);
 	parsing_should_fail(FILE_LINE_ARGS, "val = 0b1000000000000000000000000000000000000000000000000000000000000000"sv);
 
+	// missing values after base prefix
+	parsing_should_fail(FILE_LINE_ARGS, "val = 0x "sv);
+	parsing_should_fail(FILE_LINE_ARGS, "val = 0o "sv);
+	parsing_should_fail(FILE_LINE_ARGS, "val = 0b "sv);
+
 	// value tests
 	parse_expected_value(FILE_LINE_ARGS, "0xDEADBEEF"sv, 0xDEADBEEF);
 	parse_expected_value(FILE_LINE_ARGS, "0xdeadbeef"sv, 0xDEADBEEF);

diff --git a/tests/user_feedback.cpp b/tests/user_feedback.cpp
@@ -230,4 +230,21 @@ b = []
 		parsing_should_succeed(FILE_LINE_ARGS, "\t"sv);
 		parsing_should_succeed(FILE_LINE_ARGS, "\n"sv);
 	}
+
+	SECTION("github/issues/129") // https://github.com/marzer/tomlplusplus/issues/129
+	{
+		parsing_should_fail(FILE_LINE_ARGS, R"(
+			hex = 0x
+			oct = 0o
+			bin = 0b
+		)"sv);
+	}
+
+	SECTION("github/issues/130") // https://github.com/marzer/tomlplusplus/issues/130
+	{
+		parse_expected_value(FILE_LINE_ARGS, "0400-01-01 00:00:00"sv, toml::date_time{ { 400, 1, 1 }, { 0, 0, 0 } });
+		parse_expected_value(FILE_LINE_ARGS, "0400-01-01         "sv, toml::date{ 400, 1, 1 });
+		parse_expected_value(FILE_LINE_ARGS, "0400-01-01T00:00:00"sv, toml::date_time{ { 400, 1, 1 }, { 0, 0, 0 } });
+		parse_expected_value(FILE_LINE_ARGS, "1000-01-01 00:00:00"sv, toml::date_time{ { 1000, 1, 1 }, { 0, 0, 0 } });
+	}
 }
diff --git a/toml++.vcxproj b/toml++.vcxproj
@@ -89,6 +89,7 @@
     <None Include=".github\ISSUE_TEMPLATE\bug_report.md" />
     <None Include=".github\ISSUE_TEMPLATE\config.yml" />
     <None Include=".github\ISSUE_TEMPLATE\feature_request.md" />
+    <None Include=".github\ISSUE_TEMPLATE\spec_bug_report.md" />
     <None Include=".github\pull_request_template.md" />
     <None Include=".gitignore" />
     <None Include=".runsettings" />

diff --git a/toml++.vcxproj.filters b/toml++.vcxproj.filters
@@ -217,6 +217,9 @@
     <None Include="include\toml++\impl\unicode.inl">
       <Filter>include\impl</Filter>
     </None>
+    <None Include=".github\ISSUE_TEMPLATE\spec_bug_report.md">
+      <Filter>.github</Filter>
+    </None>
   </ItemGroup>
   <ItemGroup>
     <Filter Include=".circleci">

diff --git a/toml.hpp b/toml.hpp
@@ -1242,7 +1242,7 @@ TOML_NAMESPACE_START // abi namespace
 		}
 	}
 
-	enum class TOML_OPEN_FLAGS_ENUM value_flags : uint16_t
+	enum class TOML_OPEN_FLAGS_ENUM value_flags : uint16_t // being an "OPEN" flags enum is not an error
 	{
 		none,
 		format_as_binary = 1,
@@ -12698,6 +12698,9 @@ TOML_IMPL_NAMESPACE_START
 				if (*cp != traits::prefix_codepoint)
 					set_error_and_return_default("expected '"sv, traits::prefix, "', saw '"sv, to_sv(*cp), "'"sv);
 				advance_and_return_if_error_or_eof({});
+
+				if (!traits::is_digit(*cp))
+					set_error_and_return_default("expected digit, saw '"sv, to_sv(*cp), "'"sv);
 			}
 
 			// consume value chars
@@ -13010,7 +13013,6 @@ TOML_IMPL_NAMESPACE_START
 		node_ptr parse_inline_table();
 
 		TOML_NODISCARD
-		TOML_NEVER_INLINE
 		node_ptr parse_value_known_prefixes()
 		{
 			return_if_error({});
@@ -13114,8 +13116,8 @@ TOML_IMPL_NAMESPACE_START
 					begins_digit = 1 << 13,
 					begins_zero	 = 1 << 14,
 					signs_msk  = has_plus | has_minus,
-					bzero_msk  = begins_zero | has_digits,
-					bdigit_msk = begins_digit | has_digits,
+					bdigit_msk = has_digits | begins_digit,
+					bzero_msk  = bdigit_msk | begins_zero,
 				};
 				value_traits traits	 = has_nothing;
 				const auto has_any	 = [&](auto t) noexcept { return (traits & t) != has_nothing; };
@@ -13125,7 +13127,11 @@ TOML_IMPL_NAMESPACE_START
 				// examine the first character to get the 'begins with' traits
 				// (good fail-fast opportunity; all the remaining types begin with numeric digits or signs)
 				if (is_decimal_digit(*cp))
-					add_trait(*cp == U'0' ? begins_zero : begins_digit);
+				{
+					add_trait(begins_digit);
+					if (*cp == U'0')
+						add_trait(begins_zero);
+				}
 				else if (is_match(*cp, U'+', U'-'))
 					add_trait(begins_sign);
 				else
@@ -13210,8 +13216,12 @@ TOML_IMPL_NAMESPACE_START
 				return_if_error({});
 
 				// force further scanning if this could have been a date-time with a space instead of a T
-				if (char_count == 10u && traits == (bdigit_msk | has_minus) && chars[4] == U'-' && chars[7] == U'-'
-					&& !is_eof() && *cp == U' ')
+				if (char_count == 10u									 //
+					&& (traits | begins_zero) == (bzero_msk | has_minus) //
+					&& chars[4] == U'-'									 //
+					&& chars[7] == U'-'									 //
+					&& !is_eof()										 //
+					&& *cp == U' ')
 				{
 					const auto pre_advance_count = advance_count;
 					const auto pre_scan_traits	 = traits;
@@ -13255,7 +13265,7 @@ TOML_IMPL_NAMESPACE_START
 				// the only valid value type is an integer.
 				if (char_count == 1u)
 				{
-					if (has_any(begins_zero | begins_digit))
+					if (has_any(begins_digit))
 					{
 						val.reset(new value{ static_cast<int64_t>(chars[0] - U'0') });
 						advance(); // skip the digit
@@ -13301,7 +13311,7 @@ TOML_IMPL_NAMESPACE_START
 					val.reset(new value{ i });
 					val->ref_cast<int64_t>().flags(flags);
 				}
-				else if (has_any(has_e) || (has_any(begins_zero | begins_digit) && chars[1] == U'.'))
+				else if (has_any(has_e) || (has_any(begins_digit) && chars[1] == U'.'))
 					val.reset(new value{ parse_float() });
 				else if (has_any(begins_sign))
 				{
@@ -14403,7 +14413,7 @@ TOML_PUSH_WARNINGS;
 
 TOML_IMPL_NAMESPACE_START
 {
-	enum class formatted_string_traits : unsigned
+	enum class TOML_CLOSED_FLAGS_ENUM formatted_string_traits : unsigned
 	{
 		none,
 		line_breaks	  = 1u << 0, // \n