RGBCube/serenity
1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-07-25 19:37:35 +00:00
Code Issues Activity Actions

LibUnicode: Convert UnicodeData to link with weak symbols

Browse source
This commit is contained in:
Timothy Flynn 2022-01-04 13:26:43 -05:00 • committed by Linus Groh
parent f576142fe8
commit 437b9fe204
3 changed files with 25 additions and 75 deletions
Download patch file Download diff file Expand all files Collapse all files

13
Meta/Lagom/Tools/CodeGenerators/LibUnicode/GenerateUnicodeData.cpp
View file

@ -625,9 +625,10 @@ static void generate_unicode_data_implementation(Core::File& file, UnicodeData c
#include <AK/Span.h> #include <AK/Span.h>
#include <AK/String.h> #include <AK/String.h>
#include <AK/StringView.h> #include <AK/StringView.h>
#include <LibUnicode/CharacterTypes.h>
#include <LibUnicode/UnicodeData.h> #include <LibUnicode/UnicodeData.h>
namespace Unicode::Detail { namespace Unicode {
)~~~"); )~~~");
auto append_list_and_size = [&](auto const& list, StringView format) { auto append_list_and_size = [&](auto const& list, StringView format) {
@ -851,7 +852,6 @@ static constexpr Array<CodePointName, @code_point_display_names_size@> s_code_po
)~~~"); )~~~");
generator.append(R"~~~( generator.append(R"~~~(
Optional<String> code_point_display_name(u32 code_point) asm("unicode_code_point_display_name");
Optional<String> code_point_display_name(u32 code_point) Optional<String> code_point_display_name(u32 code_point)
{ {
if (auto const* entry = binary_search(s_code_point_display_names, code_point, nullptr, CodePointNameComparator {})) { if (auto const* entry = binary_search(s_code_point_display_names, code_point, nullptr, CodePointNameComparator {})) {
@ -870,7 +870,6 @@ Optional<String> code_point_display_name(u32 code_point)
generator.set("mappings", mappings); generator.set("mappings", mappings);
generator.set("fallback", fallback); generator.set("fallback", fallback);
generator.append(R"~~~( generator.append(R"~~~(
u32 @method@(u32 code_point) asm("unicode_@method@");
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> {});
@ -880,11 +879,10 @@ u32 @method@(u32 code_point)
}; };
append_code_point_mapping_search("canonical_combining_class"sv, "s_combining_class_mappings"sv, "0"sv); append_code_point_mapping_search("canonical_combining_class"sv, "s_combining_class_mappings"sv, "0"sv);
append_code_point_mapping_search("simple_uppercase_mapping"sv, "s_uppercase_mappings"sv, "code_point"sv); append_code_point_mapping_search("to_unicode_uppercase"sv, "s_uppercase_mappings"sv, "code_point"sv);
append_code_point_mapping_search("simple_lowercase_mapping"sv, "s_lowercase_mappings"sv, "code_point"sv); append_code_point_mapping_search("to_unicode_lowercase"sv, "s_lowercase_mappings"sv, "code_point"sv);
generator.append(R"~~~( generator.append(R"~~~(
Span<SpecialCasing const* const> special_case_mapping(u32 code_point) asm("unicode_special_case_mapping");
Span<SpecialCasing const* const> special_case_mapping(u32 code_point) Span<SpecialCasing const* const> special_case_mapping(u32 code_point)
{ {
auto const* mapping = binary_search(s_special_case_mappings, code_point, nullptr, CodePointComparator<SpecialCaseMapping> {}); auto const* mapping = binary_search(s_special_case_mappings, code_point, nullptr, CodePointComparator<SpecialCaseMapping> {});
@ -900,7 +898,6 @@ Span<SpecialCasing const* const> special_case_mapping(u32 code_point)
generator.set("enum_snake", enum_snake); generator.set("enum_snake", enum_snake);
generator.set("collection_name", collection_name); generator.set("collection_name", collection_name);
generator.append(R"~~~( generator.append(R"~~~(
bool code_point_has_@enum_snake@(u32 code_point, @enum_title@ @enum_snake@) asm("unicode_code_point_has_@enum_snake@");
bool code_point_has_@enum_snake@(u32 code_point, @enum_title@ @enum_snake@) bool code_point_has_@enum_snake@(u32 code_point, @enum_title@ @enum_snake@)
{ {
auto index = static_cast<@enum_title@UnderlyingType>(@enum_snake@); auto index = static_cast<@enum_title@UnderlyingType>(@enum_snake@);
@ -921,7 +918,7 @@ bool code_point_has_@enum_snake@(u32 code_point, @enum_title@ @enum_snake@)
for (auto const& alias : aliases) for (auto const& alias : aliases)
hashes.set(alias.alias.hash(), alias.alias); hashes.set(alias.alias.hash(), alias.alias);
generate_value_from_string_for_dynamic_loading(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes)); generate_value_from_string(generator, "{}_from_string"sv, enum_title, enum_snake, move(hashes));
}; };
append_prop_search("GeneralCategory"sv, "general_category"sv, "s_general_categories"sv); append_prop_search("GeneralCategory"sv, "general_category"sv, "s_general_categories"sv);

80
Userland/Libraries/LibUnicode/CharacterTypes.cpp
View file

@ -11,7 +11,6 @@
#include <AK/Utf8View.h> #include <AK/Utf8View.h>
#include <LibUnicode/CharacterTypes.h> #include <LibUnicode/CharacterTypes.h>
#include <LibUnicode/Locale.h> #include <LibUnicode/Locale.h>
#include <LibUnicode/UnicodeSymbols.h>
#if ENABLE_UNICODE_DATA #if ENABLE_UNICODE_DATA
# include <LibUnicode/UnicodeData.h> # include <LibUnicode/UnicodeData.h>
@ -22,20 +21,12 @@
namespace Unicode { namespace Unicode {
Optional<String> __attribute__((weak)) code_point_display_name(u32) { return {}; }
u32 __attribute__((weak)) canonical_combining_class(u32) { return {}; }
Span<SpecialCasing const* const> __attribute__((weak)) special_case_mapping(u32) { return {}; }
#if ENABLE_UNICODE_DATA #if ENABLE_UNICODE_DATA
static u32 canonical_combining_class(u32 code_point)
{
static auto const& symbols = Detail::Symbols::ensure_loaded();
return symbols.canonical_combining_class(code_point);
}
static Span<Unicode::SpecialCasing const* const> special_case_mapping(u32 code_point)
{
static auto const& symbols = Detail::Symbols::ensure_loaded();
return symbols.special_case_mapping(code_point);
}
static bool is_after_uppercase_i(Utf8View const& string, size_t index) static bool is_after_uppercase_i(Utf8View const& string, size_t index)
{ {
// There is an uppercase I before C, and there is no intervening combining character class 230 (Above) or 0. // There is an uppercase I before C, and there is no intervening combining character class 230 (Above) or 0.
@ -217,22 +208,14 @@ static SpecialCasing const* find_matching_special_case(u32 code_point, Utf8View
#endif #endif
u32 to_unicode_lowercase(u32 code_point) u32 __attribute__((weak)) to_unicode_lowercase(u32 code_point)
{ {
static auto const& symbols = Detail::Symbols::ensure_loaded(); return to_ascii_lowercase(code_point);
return symbols.simple_lowercase_mapping(code_point);
} }
u32 to_unicode_uppercase(u32 code_point) u32 __attribute__((weak)) to_unicode_uppercase(u32 code_point)
{ {
static auto const& symbols = Detail::Symbols::ensure_loaded(); return to_ascii_uppercase(code_point);
return symbols.simple_uppercase_mapping(code_point);
}
Optional<String> code_point_display_name(u32 code_point)
{
static auto const& symbols = Detail::Symbols::ensure_loaded();
return symbols.code_point_display_name(code_point);
} }
String to_unicode_lowercase_full(StringView string, [[maybe_unused]] Optional<StringView> locale) String to_unicode_lowercase_full(StringView string, [[maybe_unused]] Optional<StringView> locale)
@ -293,29 +276,10 @@ String to_unicode_uppercase_full(StringView string, [[maybe_unused]] Optional<St
#endif #endif
} }
Optional<GeneralCategory> general_category_from_string(StringView general_category) Optional<GeneralCategory> __attribute__((weak)) general_category_from_string(StringView) { return {}; }
{ bool __attribute__((weak)) code_point_has_general_category(u32, GeneralCategory) { return {}; }
static auto const& symbols = Detail::Symbols::ensure_loaded(); Optional<Property> __attribute__((weak)) property_from_string(StringView) { return {}; }
return symbols.general_category_from_string(general_category); bool __attribute__((weak)) code_point_has_property(u32, Property) { return {}; }
}
bool code_point_has_general_category(u32 code_point, GeneralCategory general_category)
{
static auto const& symbols = Detail::Symbols::ensure_loaded();
return symbols.code_point_has_general_category(code_point, general_category);
}
Optional<Property> property_from_string(StringView property)
{
static auto const& symbols = Detail::Symbols::ensure_loaded();
return symbols.property_from_string(property);
}
bool code_point_has_property(u32 code_point, Property property)
{
static auto const& symbols = Detail::Symbols::ensure_loaded();
return symbols.code_point_has_property(code_point, property);
}
bool is_ecma262_property([[maybe_unused]] Property property) bool is_ecma262_property([[maybe_unused]] Property property)
{ {
@ -384,22 +348,8 @@ bool is_ecma262_property([[maybe_unused]] Property property)
#endif #endif
} }
Optional<Script> script_from_string(StringView script) Optional<Script> __attribute__((weak)) script_from_string(StringView) { return {}; }
{ bool __attribute__((weak)) code_point_has_script(u32, Script) { return {}; }
static auto const& symbols = Detail::Symbols::ensure_loaded(); bool __attribute__((weak)) code_point_has_script_extension(u32, Script) { return {}; }
return symbols.script_from_string(script);
}
bool code_point_has_script(u32 code_point, Script script)
{
static auto const& symbols = Detail::Symbols::ensure_loaded();
return symbols.code_point_has_script(code_point, script);
}
bool code_point_has_script_extension(u32 code_point, Script script)
{
static auto const& symbols = Detail::Symbols::ensure_loaded();
return symbols.code_point_has_script_extension(code_point, script);
}
} }

7
Userland/Libraries/LibUnicode/CharacterTypes.h
View file

@ -8,19 +8,22 @@
#include <AK/Forward.h> #include <AK/Forward.h>
#include <AK/Optional.h> #include <AK/Optional.h>
#include <AK/Span.h>
#include <AK/String.h> #include <AK/String.h>
#include <AK/Types.h> #include <AK/Types.h>
#include <LibUnicode/Forward.h> #include <LibUnicode/Forward.h>
namespace Unicode { namespace Unicode {
Optional<String> code_point_display_name(u32 code_point);
u32 canonical_combining_class(u32 code_point);
Span<SpecialCasing const* const> special_case_mapping(u32 code_point);
// Note: The single code point case conversions only perform simple case folding. // Note: The single code point case conversions only perform simple case folding.
// Use the full-string transformations for full case folding. // Use the full-string transformations for full case folding.
u32 to_unicode_lowercase(u32 code_point); u32 to_unicode_lowercase(u32 code_point);
u32 to_unicode_uppercase(u32 code_point); u32 to_unicode_uppercase(u32 code_point);
Optional<String> code_point_display_name(u32 code_point);
String to_unicode_lowercase_full(StringView, Optional<StringView> locale = {}); String to_unicode_lowercase_full(StringView, Optional<StringView> locale = {});
String to_unicode_uppercase_full(StringView, Optional<StringView> locale = {}); String to_unicode_uppercase_full(StringView, Optional<StringView> locale = {});