1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-05-25 22:25:08 +00:00

LibUnicode: Generate and use unique subtag and complex alias strings

This commit is contained in:
Timothy Flynn 2021-10-10 13:56:33 -04:00 committed by Andreas Kling
parent 3d67f6bd29
commit acb7bd917f

View file

@ -43,10 +43,10 @@ struct Locale {
}; };
struct CanonicalLanguageID { struct CanonicalLanguageID {
String language {}; size_t language { 0 };
String script {}; size_t script { 0 };
String region {}; size_t region { 0 };
Vector<String> variants {}; Vector<size_t> variants {};
}; };
struct LanguageMapping { struct LanguageMapping {
@ -95,7 +95,16 @@ static size_t ensure_unique_string(UnicodeLocaleData& locale_data, String string
return index; return index;
} }
static Optional<CanonicalLanguageID> parse_language(StringView language) static StringView get_unique_string(UnicodeLocaleData& locale_data, size_t index)
{
if (index == 0)
return {};
VERIFY(index <= locale_data.unique_strings.size());
return locale_data.unique_strings.at(index - 1);
}
static Optional<CanonicalLanguageID> parse_language(UnicodeLocaleData& locale_data, StringView language)
{ {
CanonicalLanguageID language_id {}; CanonicalLanguageID language_id {};
@ -104,7 +113,7 @@ static Optional<CanonicalLanguageID> parse_language(StringView language)
size_t index = 0; size_t index = 0;
if (Unicode::is_unicode_language_subtag(segments[index])) { if (Unicode::is_unicode_language_subtag(segments[index])) {
language_id.language = segments[index]; language_id.language = ensure_unique_string(locale_data, segments[index]);
if (segments.size() == ++index) if (segments.size() == ++index)
return language_id; return language_id;
} else { } else {
@ -112,13 +121,13 @@ static Optional<CanonicalLanguageID> parse_language(StringView language)
} }
if (Unicode::is_unicode_script_subtag(segments[index])) { if (Unicode::is_unicode_script_subtag(segments[index])) {
language_id.script = segments[index]; language_id.script = ensure_unique_string(locale_data, segments[index]);
if (segments.size() == ++index) if (segments.size() == ++index)
return language_id; return language_id;
} }
if (Unicode::is_unicode_region_subtag(segments[index])) { if (Unicode::is_unicode_region_subtag(segments[index])) {
language_id.region = segments[index]; language_id.region = ensure_unique_string(locale_data, segments[index]);
if (segments.size() == ++index) if (segments.size() == ++index)
return language_id; return language_id;
} }
@ -126,19 +135,19 @@ static Optional<CanonicalLanguageID> parse_language(StringView language)
while (index < segments.size()) { while (index < segments.size()) {
if (!Unicode::is_unicode_variant_subtag(segments[index])) if (!Unicode::is_unicode_variant_subtag(segments[index]))
return {}; return {};
language_id.variants.append(segments[index++]); language_id.variants.append(ensure_unique_string(locale_data, segments[index++]));
} }
return language_id; return language_id;
} }
static Optional<LanguageMapping> parse_language_mapping(StringView key, StringView alias) static Optional<LanguageMapping> parse_language_mapping(UnicodeLocaleData& locale_data, StringView key, StringView alias)
{ {
auto parsed_key = parse_language(key); auto parsed_key = parse_language(locale_data, key);
if (!parsed_key.has_value()) if (!parsed_key.has_value())
return {}; return {};
auto parsed_alias = parse_language(alias); auto parsed_alias = parse_language(locale_data, alias);
if (!parsed_alias.has_value()) if (!parsed_alias.has_value())
return {}; return {};
@ -166,7 +175,7 @@ static void parse_core_aliases(String core_supplemental_path, UnicodeLocaleData&
auto alias = value.as_object().get("_replacement"sv).as_string(); auto alias = value.as_object().get("_replacement"sv).as_string();
if (key.contains('-')) { if (key.contains('-')) {
auto mapping = parse_language_mapping(key, alias); auto mapping = parse_language_mapping(locale_data, key, alias);
if (!mapping.has_value()) if (!mapping.has_value())
return; return;
@ -202,7 +211,7 @@ static void parse_likely_subtags(String core_supplemental_path, UnicodeLocaleDat
auto const& likely_subtags_object = supplemental_object.as_object().get("likelySubtags"sv); auto const& likely_subtags_object = supplemental_object.as_object().get("likelySubtags"sv);
likely_subtags_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) { likely_subtags_object.as_object().for_each_member([&](auto const& key, JsonValue const& value) {
auto mapping = parse_language_mapping(key, value.as_string()); auto mapping = parse_language_mapping(locale_data, key, value.as_string());
if (!mapping.has_value()) if (!mapping.has_value())
return; return;
@ -474,17 +483,17 @@ static void parse_all_locales(String core_path, String locale_names_path, String
parse_core_aliases(core_supplemental_path.string(), locale_data); parse_core_aliases(core_supplemental_path.string(), locale_data);
parse_likely_subtags(core_supplemental_path.string(), locale_data); parse_likely_subtags(core_supplemental_path.string(), locale_data);
auto remove_variants_from_path = [](String path) -> Optional<String> { auto remove_variants_from_path = [&](String path) -> Optional<String> {
auto parsed_locale = parse_language(LexicalPath::basename(path)); auto parsed_locale = parse_language(locale_data, LexicalPath::basename(path));
if (!parsed_locale.has_value()) if (!parsed_locale.has_value())
return {}; return {};
StringBuilder builder; StringBuilder builder;
builder.append(parsed_locale->language); builder.append(get_unique_string(locale_data, parsed_locale->language));
if (!parsed_locale->script.is_empty()) if (auto script = get_unique_string(locale_data, parsed_locale->script); !script.is_empty())
builder.appendff("-{}", parsed_locale->script); builder.appendff("-{}", script);
if (!parsed_locale->region.is_empty()) if (auto region = get_unique_string(locale_data, parsed_locale->region); !region.is_empty())
builder.appendff("-{}", parsed_locale->region); builder.appendff("-{}", region);
return builder.build(); return builder.build();
}; };
@ -702,11 +711,8 @@ static constexpr Array<StringView, @strings_size@ + 1> s_string_list { {
return String::formatted(format, mapping_name); return String::formatted(format, mapping_name);
}; };
auto append_string = [&](StringView value) { auto append_index = [&](size_t index) {
if (value.is_empty()) generator.append(String::formatted(", {}", index));
generator.append(", {}"sv);
else
generator.append(String::formatted(", \"{}\"sv", value));
}; };
auto append_list_and_size = [&](auto const& list) { auto append_list_and_size = [&](auto const& list) {
@ -719,7 +725,7 @@ static constexpr Array<StringView, @strings_size@ + 1> s_string_list { {
generator.append(", {"); generator.append(", {");
for (auto const& item : list) { for (auto const& item : list) {
generator.append(first ? " " : ", "); generator.append(first ? " " : ", ");
generator.append(String::formatted("\"{}\"sv", item)); generator.append(String::number(item));
first = false; first = false;
} }
generator.append(String::formatted(" }}, {}", list.size())); generator.append(String::formatted(" }}, {}", list.size()));
@ -834,13 +840,13 @@ struct CanonicalLanguageID {
Unicode::LanguageID language_id {}; Unicode::LanguageID language_id {};
language_id.variants.ensure_capacity(variants_size); language_id.variants.ensure_capacity(variants_size);
language_id.language = language.to_string(); language_id.language = s_string_list[language];
if (!script.is_empty()) if (script != 0)
language_id.script = script.to_string(); language_id.script = s_string_list[script];
if (!region.is_empty()) if (region != 0)
language_id.region = region.to_string(); language_id.region = s_string_list[region];
for (size_t i = 0; i < variants_size; ++i) for (size_t i = 0; i < variants_size; ++i)
language_id.variants.append(variants[i].to_string()); language_id.variants.append(s_string_list[variants[i]]);
return language_id; return language_id;
} }
@ -852,17 +858,17 @@ struct CanonicalLanguageID {
return false; return false;
for (size_t i = 0; i < variants_size; ++i) { for (size_t i = 0; i < variants_size; ++i) {
if (variants[i] != other_variants[i]) if (s_string_list[variants[i]] != other_variants[i])
return false; return false;
} }
return true; return true;
}; };
StringView language {}; size_t language { 0 };
StringView script {}; size_t script { 0 };
StringView region {}; size_t region { 0 };
Array<StringView, @variants_size@> variants {}; Array<size_t, @variants_size@> variants {};
size_t variants_size { 0 }; size_t variants_size { 0 };
}; };
@ -881,9 +887,9 @@ struct LanguageMapping {
static constexpr Array<LanguageMapping, @size@> s_@name@ { { static constexpr Array<LanguageMapping, @size@> s_@name@ { {
)~~~"); )~~~");
quick_sort(mappings, [](auto const& lhs, auto const& rhs) { quick_sort(mappings, [&](auto const& lhs, auto const& rhs) {
auto const& lhs_language = lhs.key.language; auto const& lhs_language = get_unique_string(locale_data, lhs.key.language);
auto const& rhs_language = rhs.key.language; auto const& rhs_language = get_unique_string(locale_data, rhs.key.language);
// Sort the keys such that "und" language tags are at the end, as those are less specific. // Sort the keys such that "und" language tags are at the end, as those are less specific.
if (lhs_language.starts_with("und"sv) && !rhs_language.starts_with("und"sv)) if (lhs_language.starts_with("und"sv) && !rhs_language.starts_with("und"sv))
@ -894,18 +900,18 @@ static constexpr Array<LanguageMapping, @size@> s_@name@ { {
}); });
for (auto const& mapping : mappings) { for (auto const& mapping : mappings) {
generator.set("language"sv, mapping.key.language); generator.set("language"sv, String::number(mapping.key.language));
generator.append(" { { \"@language@\"sv"); generator.append(" { { @language@");
append_string(mapping.key.script); append_index(mapping.key.script);
append_string(mapping.key.region); append_index(mapping.key.region);
append_list_and_size(mapping.key.variants); append_list_and_size(mapping.key.variants);
generator.set("language"sv, mapping.alias.language); generator.set("language"sv, String::number(mapping.alias.language));
generator.append(" }, { \"@language@\"sv"); generator.append(" }, { @language@");
append_string(mapping.alias.script); append_index(mapping.alias.script);
append_string(mapping.alias.region); append_index(mapping.alias.region);
append_list_and_size(mapping.alias.variants); append_list_and_size(mapping.alias.variants);
generator.append(" } },\n"); generator.append(" } },\n");
@ -933,7 +939,7 @@ static LanguageMapping const* resolve_likely_subtag(Unicode::LanguageID const& l
auto state = State::LanguageScriptRegion; auto state = State::LanguageScriptRegion;
while (state != State::Done) { while (state != State::Done) {
CanonicalLanguageID search_key; Unicode::LanguageID search_key;
switch (state) { switch (state) {
case State::LanguageScriptRegion: case State::LanguageScriptRegion:
@ -983,12 +989,20 @@ static LanguageMapping const* resolve_likely_subtag(Unicode::LanguageID const& l
} }
for (auto const& map : s_likely_subtags) { for (auto const& map : s_likely_subtags) {
if (map.key.language != search_key.language) auto const& key_language = s_string_list[map.key.language];
continue; auto const& key_script = s_string_list[map.key.script];
if (map.key.script != search_key.script) auto const& key_region = s_string_list[map.key.region];
continue;
if (map.key.region != search_key.region) if (key_language != search_key.language)
continue; continue;
if (!key_script.is_empty() || search_key.script.has_value()) {
if (key_script != search_key.script)
continue;
}
if (!key_region.is_empty() || search_key.region.has_value()) {
if (key_region != search_key.region)
continue;
}
return &map; return &map;
} }
@ -1153,11 +1167,15 @@ Optional<ListPatterns> get_locale_list_pattern_mapping(StringView locale, String
void resolve_complex_language_aliases(Unicode::LanguageID& language_id) void resolve_complex_language_aliases(Unicode::LanguageID& language_id)
{ {
for (auto const& map : s_complex_alias) { for (auto const& map : s_complex_alias) {
if ((map.key.language != language_id.language) && (map.key.language != "und"sv)) auto const& key_language = s_string_list[map.key.language];
auto const& key_script = s_string_list[map.key.script];
auto const& key_region = s_string_list[map.key.region];
if ((key_language != language_id.language) && (key_language != "und"sv))
continue; continue;
if (!map.key.script.is_empty() && (map.key.script != language_id.script)) if (!key_script.is_empty() && (key_script != language_id.script))
continue; continue;
if (!map.key.region.is_empty() && (map.key.region != language_id.region)) if (!key_region.is_empty() && (key_region != language_id.region))
continue; continue;
if (!map.key.matches_variants(language_id.variants)) if (!map.key.matches_variants(language_id.variants))
continue; continue;
@ -1166,9 +1184,9 @@ void resolve_complex_language_aliases(Unicode::LanguageID& language_id)
if (alias.language == "und"sv) if (alias.language == "und"sv)
alias.language = move(language_id.language); alias.language = move(language_id.language);
if (map.key.script.is_empty() && !alias.script.has_value()) if (key_script.is_empty() && !alias.script.has_value())
alias.script = move(language_id.script); alias.script = move(language_id.script);
if (map.key.region.is_empty() && !alias.region.has_value()) if (key_region.is_empty() && !alias.region.has_value())
alias.region = move(language_id.region); alias.region = move(language_id.region);
if (map.key.variants_size == 0 && alias.variants.is_empty()) if (map.key.variants_size == 0 && alias.variants.is_empty())
alias.variants = move(language_id.variants); alias.variants = move(language_id.variants);
@ -1186,15 +1204,20 @@ Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& lang
return {}; return {};
auto maximized = language_id; auto maximized = language_id;
auto const& key = likely_subtag->key;
auto const& alias = likely_subtag->alias; auto const& key_script = s_string_list[likely_subtag->key.script];
auto const& key_region = s_string_list[likely_subtag->key.region];
auto const& alias_language = s_string_list[likely_subtag->alias.language];
auto const& alias_script = s_string_list[likely_subtag->alias.script];
auto const& alias_region = s_string_list[likely_subtag->alias.region];
if (maximized.language == "und"sv) if (maximized.language == "und"sv)
maximized.language = alias.language; maximized.language = alias_language;
if (!maximized.script.has_value() || (!key.script.is_empty() && !alias.script.is_empty())) if (!maximized.script.has_value() || (!key_script.is_empty() && !alias_script.is_empty()))
maximized.script = alias.script; maximized.script = alias_script;
if (!maximized.region.has_value() || (!key.region.is_empty() && !alias.region.is_empty())) if (!maximized.region.has_value() || (!key_region.is_empty() && !alias_region.is_empty()))
maximized.region = alias.region; maximized.region = alias_region;
return maximized; return maximized;
} }
@ -1202,7 +1225,7 @@ Optional<Unicode::LanguageID> add_likely_subtags(Unicode::LanguageID const& lang
Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id) Optional<String> resolve_most_likely_territory(Unicode::LanguageID const& language_id)
{ {
if (auto const* likely_subtag = resolve_likely_subtag(language_id); likely_subtag != nullptr) if (auto const* likely_subtag = resolve_likely_subtag(language_id); likely_subtag != nullptr)
return likely_subtag->alias.region; return s_string_list[likely_subtag->alias.region];
return {}; return {};
} }