RGBCube/serenity
1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-05-31 14:28:12 +00:00
Code Issues Activity Actions

LibUnicode: Generate standalone compile-time array for combining class

Browse source
This commit is contained in:
Timothy Flynn 2021-10-09 19:14:23 -04:00 • committed by Andreas Kling
parent 9f83774913
commit d83b262e64
2 changed files with 31 additions and 24 deletions
Download patch file Download diff file Expand all files Collapse all files

21
Meta/Lagom/Tools/CodeGenerators/LibUnicode/GenerateUnicodeData.cpp
View file

@ -88,6 +88,8 @@ struct CodePointData {
}; };
struct UnicodeData { struct UnicodeData {
u32 code_points_with_non_zero_combining_class { 0 };
u32 simple_uppercase_mapping_size { 0 }; u32 simple_uppercase_mapping_size { 0 };
u32 simple_lowercase_mapping_size { 0 }; u32 simple_lowercase_mapping_size { 0 };
@ -438,6 +440,7 @@ static void parse_unicode_data(Core::File& file, UnicodeData& unicode_data)
} }
} }
unicode_data.code_points_with_non_zero_combining_class += data.canonical_combining_class != 0;
unicode_data.simple_uppercase_mapping_size += data.simple_uppercase_mapping.has_value(); unicode_data.simple_uppercase_mapping_size += data.simple_uppercase_mapping.has_value();
unicode_data.simple_lowercase_mapping_size += data.simple_lowercase_mapping.has_value(); unicode_data.simple_lowercase_mapping_size += data.simple_lowercase_mapping.has_value();
@ -564,6 +567,8 @@ namespace Detail {
Optional<UnicodeData> unicode_data_for_code_point(u32 code_point); Optional<UnicodeData> unicode_data_for_code_point(u32 code_point);
u32 canonical_combining_class(u32 code_point);
u32 simple_uppercase_mapping(u32 code_point); u32 simple_uppercase_mapping(u32 code_point);
u32 simple_lowercase_mapping(u32 code_point); u32 simple_lowercase_mapping(u32 code_point);
Span<SpecialCasing const* const> special_case_mapping(u32 code_point); Span<SpecialCasing const* const> special_case_mapping(u32 code_point);
@ -750,6 +755,12 @@ static constexpr Array<@mapping_type@, @size@> s_@name@_mappings { {
)~~~"); )~~~");
}; };
append_code_point_mappings("combining_class"sv, "CodePointMapping"sv, unicode_data.code_points_with_non_zero_combining_class,
[](auto const& data) -> Optional<u32> {
if (data.canonical_combining_class == 0)
return {};
return data.canonical_combining_class;
});
append_code_point_mappings("uppercase"sv, "CodePointMapping"sv, unicode_data.simple_uppercase_mapping_size, [](auto const& data) { return data.simple_uppercase_mapping; }); append_code_point_mappings("uppercase"sv, "CodePointMapping"sv, unicode_data.simple_uppercase_mapping_size, [](auto const& data) { return data.simple_uppercase_mapping; });
append_code_point_mappings("lowercase"sv, "CodePointMapping"sv, unicode_data.simple_lowercase_mapping_size, [](auto const& data) { return data.simple_lowercase_mapping; }); append_code_point_mappings("lowercase"sv, "CodePointMapping"sv, unicode_data.simple_lowercase_mapping_size, [](auto const& data) { return data.simple_lowercase_mapping; });
append_code_point_mappings("special_case"sv, "SpecialCaseMapping"sv, unicode_data.code_points_with_special_casing, [](auto const& data) { return data.special_casing_indices; }); append_code_point_mappings("special_case"sv, "SpecialCaseMapping"sv, unicode_data.code_points_with_special_casing, [](auto const& data) { return data.special_casing_indices; });
@ -877,20 +888,22 @@ Optional<UnicodeData> unicode_data_for_code_point(u32 code_point)
} }
)~~~"); )~~~");
auto append_code_point_mapping_search = [&](StringView method, StringView mappings) { auto append_code_point_mapping_search = [&](StringView method, StringView mappings, StringView fallback) {
generator.set("method", method); generator.set("method", method);
generator.set("mappings", mappings); generator.set("mappings", mappings);
generator.set("fallback", fallback);
generator.append(R"~~~( generator.append(R"~~~(
u32 @method@(u32 code_point) u32 @method@(u32 code_point)
{ {
auto const* mapping = binary_search(@mappings@, code_point, nullptr, CodePointComparator<CodePointMapping> {}); auto const* mapping = binary_search(@mappings@, code_point, nullptr, CodePointComparator<CodePointMapping> {});
return mapping ? mapping->mapping : code_point; return mapping ? mapping->mapping : @fallback@;
} }
)~~~"); )~~~");
}; };
append_code_point_mapping_search("simple_uppercase_mapping"sv, "s_uppercase_mappings"sv); append_code_point_mapping_search("canonical_combining_class"sv, "s_combining_class_mappings"sv, "0"sv);
append_code_point_mapping_search("simple_lowercase_mapping"sv, "s_lowercase_mappings"sv); append_code_point_mapping_search("simple_uppercase_mapping"sv, "s_uppercase_mappings"sv, "code_point"sv);
append_code_point_mapping_search("simple_lowercase_mapping"sv, "s_lowercase_mappings"sv, "code_point"sv);
generator.append(R"~~~( generator.append(R"~~~(
Span<SpecialCasing const* const> special_case_mapping(u32 code_point) Span<SpecialCasing const* const> special_case_mapping(u32 code_point)

34
Userland/Libraries/LibUnicode/CharacterTypes.cpp
View file

@ -36,13 +36,11 @@ static bool is_after_uppercase_i(Utf8View const& string, size_t index)
continue; continue;
} }
auto unicode_data = Detail::unicode_data_for_code_point(code_point); u32 canonical_combining_class = Detail::canonical_combining_class(code_point);
if (!unicode_data.has_value())
return false;
if (unicode_data->canonical_combining_class == 0) if (canonical_combining_class == 0)
found_uppercase_i = false; found_uppercase_i = false;
else if (unicode_data->canonical_combining_class == 230) else if (canonical_combining_class == 230)
found_uppercase_i = false; found_uppercase_i = false;
} }
@ -62,13 +60,11 @@ static bool is_after_soft_dotted_code_point(Utf8View const& string, size_t index
continue; continue;
} }
auto unicode_data = Detail::unicode_data_for_code_point(code_point); u32 canonical_combining_class = Detail::canonical_combining_class(code_point);
if (!unicode_data.has_value())
return false;
if (unicode_data->canonical_combining_class == 0) if (canonical_combining_class == 0)
found_soft_dotted_code_point = false; found_soft_dotted_code_point = false;
else if (unicode_data->canonical_combining_class == 230) else if (canonical_combining_class == 230)
found_soft_dotted_code_point = false; found_soft_dotted_code_point = false;
} }
@ -123,12 +119,11 @@ static bool is_followed_by_combining_class_above(Utf8View const& string, size_t
: Utf8View {}; : Utf8View {};
for (auto code_point : following_view) { for (auto code_point : following_view) {
auto unicode_data = Detail::unicode_data_for_code_point(code_point); u32 canonical_combining_class = Detail::canonical_combining_class(code_point);
if (!unicode_data.has_value())
if (canonical_combining_class == 0)
return false; return false;
if (unicode_data->canonical_combining_class == 0) if (canonical_combining_class == 230)
return false;
if (unicode_data->canonical_combining_class == 230)
return true; return true;
} }
@ -147,12 +142,11 @@ static bool is_followed_by_combining_dot_above(Utf8View const& string, size_t in
if (code_point == 0x307) if (code_point == 0x307)
return true; return true;
auto unicode_data = Detail::unicode_data_for_code_point(code_point); u32 canonical_combining_class = Detail::canonical_combining_class(code_point);
if (!unicode_data.has_value())
if (canonical_combining_class == 0)
return false; return false;
if (unicode_data->canonical_combining_class == 0) if (canonical_combining_class == 230)
return false;
if (unicode_data->canonical_combining_class == 230)
return false; return false;
} }