LibJS+LibLocale: Replace relative time formatting with ICU

This uses ICU for all of the Intl.RelativeTimeFormat prototypes, which
lets us remove all data from our relative-time format generator.
This commit is contained in:
Timothy Flynn 2024-06-14 11:54:37 -04:00 committed by Andreas Kling
parent f7a83e5755
commit 7f9ccd39f5
Notes: sideshowbarker 2024-07-17 05:01:20 +09:00
9 changed files with 353 additions and 631 deletions

View File

@ -44,9 +44,6 @@ if (ENABLE_UNICODE_DATABASE_DOWNLOAD)
set(PLURAL_RULES_DATA_HEADER PluralRulesData.h)
set(PLURAL_RULES_DATA_IMPLEMENTATION PluralRulesData.cpp)
set(RELATIVE_TIME_FORMAT_DATA_HEADER RelativeTimeFormatData.h)
set(RELATIVE_TIME_FORMAT_DATA_IMPLEMENTATION RelativeTimeFormatData.cpp)
invoke_generator(
"LocaleData"
Lagom::GenerateLocaleData
@ -63,21 +60,11 @@ if (ENABLE_UNICODE_DATABASE_DOWNLOAD)
"${PLURAL_RULES_DATA_IMPLEMENTATION}"
arguments -r "${CLDR_CORE_PATH}" -l "${CLDR_LOCALES_PATH}"
)
invoke_generator(
"RelativeTimeFormatData"
Lagom::GenerateRelativeTimeFormatData
"${CLDR_VERSION_FILE}"
"${RELATIVE_TIME_FORMAT_DATA_HEADER}"
"${RELATIVE_TIME_FORMAT_DATA_IMPLEMENTATION}"
arguments -d "${CLDR_DATES_PATH}"
)
set(LOCALE_DATA_SOURCES
${LOCALE_DATA_HEADER}
${LOCALE_DATA_IMPLEMENTATION}
${PLURAL_RULES_DATA_HEADER}
${PLURAL_RULES_DATA_IMPLEMENTATION}
${RELATIVE_TIME_FORMAT_DATA_HEADER}
${RELATIVE_TIME_FORMAT_DATA_IMPLEMENTATION}
)
endif()

View File

@ -1,3 +1,2 @@
lagom_tool(GenerateLocaleData SOURCES GenerateLocaleData.cpp LIBS LibMain)
lagom_tool(GeneratePluralRulesData SOURCES GeneratePluralRulesData.cpp LIBS LibMain)
lagom_tool(GenerateRelativeTimeFormatData SOURCES GenerateRelativeTimeFormatData.cpp LIBS LibMain)

View File

@ -1,304 +0,0 @@
/*
* Copyright (c) 2022-2023, Tim Flynn <trflynn89@serenityos.org>
*
* SPDX-License-Identifier: BSD-2-Clause
*/
#include "../LibUnicode/GeneratorUtil.h" // FIXME: Move this somewhere common.
#include <AK/ByteString.h>
#include <AK/Format.h>
#include <AK/HashMap.h>
#include <AK/JsonObject.h>
#include <AK/JsonParser.h>
#include <AK/JsonValue.h>
#include <AK/LexicalPath.h>
#include <AK/SourceGenerator.h>
#include <AK/StringBuilder.h>
#include <LibCore/ArgsParser.h>
#include <LibCore/Directory.h>
#include <LibLocale/Locale.h>
#include <LibLocale/RelativeTimeFormat.h>
struct RelativeTimeFormat {
unsigned hash() const
{
auto hash = time_unit.hash();
hash = pair_int_hash(hash, style.hash());
hash = pair_int_hash(hash, plurality.hash());
hash = pair_int_hash(hash, tense_or_number);
hash = pair_int_hash(hash, pattern);
return hash;
}
bool operator==(RelativeTimeFormat const& other) const
{
return (time_unit == other.time_unit)
&& (plurality == other.plurality)
&& (style == other.style)
&& (tense_or_number == other.tense_or_number)
&& (pattern == other.pattern);
}
ByteString time_unit;
ByteString style;
ByteString plurality;
size_t tense_or_number { 0 };
size_t pattern { 0 };
};
template<>
struct AK::Formatter<RelativeTimeFormat> : Formatter<FormatString> {
ErrorOr<void> format(FormatBuilder& builder, RelativeTimeFormat const& format)
{
return Formatter<FormatString>::format(builder,
"{{ TimeUnit::{}, Style::{}, PluralCategory::{}, {}, {} }}"sv,
format.time_unit,
format.style,
format.plurality,
format.tense_or_number,
format.pattern);
}
};
template<>
struct AK::Traits<RelativeTimeFormat> : public DefaultTraits<RelativeTimeFormat> {
static unsigned hash(RelativeTimeFormat const& format) { return format.hash(); }
};
struct LocaleData {
Vector<size_t> time_units;
};
struct CLDR {
UniqueStringStorage unique_strings;
UniqueStorage<RelativeTimeFormat> unique_formats;
HashMap<ByteString, LocaleData> locales;
};
static ErrorOr<void> parse_date_fields(ByteString locale_dates_path, CLDR& cldr, LocaleData& locale)
{
LexicalPath date_fields_path(move(locale_dates_path));
date_fields_path = date_fields_path.append("dateFields.json"sv);
auto date_fields = TRY(read_json_file(date_fields_path.string()));
auto const& main_object = date_fields.as_object().get_object("main"sv).value();
auto const& locale_object = main_object.get_object(date_fields_path.parent().basename()).value();
auto const& dates_object = locale_object.get_object("dates"sv).value();
auto const& fields_object = dates_object.get_object("fields"sv).value();
auto is_sanctioned_unit = [](auto unit) {
// This is a copy of the time units sanctioned for use within ECMA-402.
// https://tc39.es/ecma402/#sec-singularrelativetimeunit
return unit.is_one_of("second"sv, "minute"sv, "hour"sv, "day"sv, "week"sv, "month"sv, "quarter"sv, "year"sv);
};
auto parse_pattern = [&](auto unit, auto style, auto plurality, auto tense_or_number, auto const& pattern) {
RelativeTimeFormat format {};
format.time_unit = unit.to_titlecase_string();
format.style = style.to_titlecase_string();
format.plurality = plurality.to_titlecase_string();
format.tense_or_number = cldr.unique_strings.ensure(tense_or_number);
format.pattern = cldr.unique_strings.ensure(pattern.as_string());
locale.time_units.append(cldr.unique_formats.ensure(move(format)));
};
fields_object.for_each_member([&](auto const& unit_and_style, auto const& patterns) {
auto segments = unit_and_style.split_view('-');
auto unit = segments[0];
auto style = (segments.size() > 1) ? segments[1] : "long"sv;
if (!is_sanctioned_unit(unit))
return;
patterns.as_object().for_each_member([&](auto const& type, auto const& pattern_value) {
constexpr auto number_key = "relative-type-"sv;
constexpr auto tense_key = "relativeTime-type-"sv;
constexpr auto plurality_key = "relativeTimePattern-count-"sv;
if (type.starts_with(number_key)) {
auto number = type.substring_view(number_key.length());
parse_pattern(unit, style, "Other"sv, number, pattern_value);
} else if (type.starts_with(tense_key)) {
pattern_value.as_object().for_each_member([&](auto const& key, auto const& pattern) {
VERIFY(key.starts_with(plurality_key));
auto plurality = key.substring_view(plurality_key.length());
auto tense = type.substring_view(tense_key.length());
parse_pattern(unit, style, plurality, tense, pattern);
});
}
});
});
return {};
}
static ErrorOr<void> parse_all_locales(ByteString dates_path, CLDR& cldr)
{
auto remove_variants_from_path = [&](ByteString path) -> ErrorOr<ByteString> {
auto parsed_locale = TRY(CanonicalLanguageID::parse(cldr.unique_strings, LexicalPath::basename(path)));
StringBuilder builder;
builder.append(cldr.unique_strings.get(parsed_locale.language));
if (auto script = cldr.unique_strings.get(parsed_locale.script); !script.is_empty())
builder.appendff("-{}", script);
if (auto region = cldr.unique_strings.get(parsed_locale.region); !region.is_empty())
builder.appendff("-{}", region);
return builder.to_byte_string();
};
TRY(Core::Directory::for_each_entry(TRY(String::formatted("{}/main", dates_path)), Core::DirIterator::SkipParentAndBaseDir, [&](auto& entry, auto& directory) -> ErrorOr<IterationDecision> {
auto dates_path = LexicalPath::join(directory.path().string(), entry.name).string();
auto language = TRY(remove_variants_from_path(dates_path));
auto& locale = cldr.locales.ensure(language);
TRY(parse_date_fields(move(dates_path), cldr, locale));
return IterationDecision::Continue;
}));
return {};
}
static ErrorOr<void> generate_unicode_locale_header(Core::InputBufferedFile& file, CLDR&)
{
StringBuilder builder;
SourceGenerator generator { builder };
generator.append(R"~~~(
#pragma once
#include <LibLocale/Forward.h>
namespace Locale {
)~~~");
generator.append(R"~~~(
}
)~~~");
TRY(file.write_until_depleted(generator.as_string_view().bytes()));
return {};
}
static ErrorOr<void> generate_unicode_locale_implementation(Core::InputBufferedFile& file, CLDR& cldr)
{
StringBuilder builder;
SourceGenerator generator { builder };
generator.set("string_index_type"sv, cldr.unique_strings.type_that_fits());
generator.set("relative_time_format_index_type"sv, cldr.unique_formats.type_that_fits());
generator.append(R"~~~(
#include <AK/Array.h>
#include <AK/StringView.h>
#include <AK/Vector.h>
#include <LibLocale/Locale.h>
#include <LibLocale/PluralRules.h>
#include <LibLocale/RelativeTimeFormat.h>
#include <LibLocale/RelativeTimeFormatData.h>
namespace Locale {
)~~~");
cldr.unique_strings.generate(generator);
generator.append(R"~~~(
struct RelativeTimeFormatImpl {
RelativeTimeFormat to_relative_time_format() const
{
RelativeTimeFormat relative_time_format {};
relative_time_format.plurality = plurality;
relative_time_format.pattern = decode_string(pattern);
return relative_time_format;
}
TimeUnit time_unit;
Style style;
PluralCategory plurality;
@string_index_type@ tense_or_number { 0 };
@string_index_type@ pattern { 0 };
};
)~~~");
cldr.unique_formats.generate(generator, "RelativeTimeFormatImpl"sv, "s_relative_time_formats"sv, 10);
auto append_list = [&](ByteString name, auto const& list) {
generator.set("name", name);
generator.set("size", ByteString::number(list.size()));
generator.append(R"~~~(
static constexpr Array<@relative_time_format_index_type@, @size@> @name@ { {)~~~");
bool first = true;
for (auto index : list) {
generator.append(first ? " "sv : ", "sv);
generator.append(ByteString::number(index));
first = false;
}
generator.append(" } };");
};
generate_mapping(generator, cldr.locales, cldr.unique_formats.type_that_fits(), "s_locale_relative_time_formats"sv, "s_number_systems_digits_{}"sv, nullptr, [&](auto const& name, auto const& value) { append_list(name, value.time_units); });
generator.append(R"~~~(
Vector<RelativeTimeFormat> get_relative_time_format_patterns(StringView locale, TimeUnit time_unit, StringView tense_or_number, Style style)
{
Vector<RelativeTimeFormat> formats;
auto locale_value = locale_from_string(locale);
if (!locale_value.has_value())
return formats;
auto locale_index = to_underlying(*locale_value) - 1; // Subtract 1 because 0 == Locale::None.
auto const& locale_formats = s_locale_relative_time_formats.at(locale_index);
for (auto const& locale_format_index : locale_formats) {
auto const& locale_format = s_relative_time_formats.at(locale_format_index);
if (locale_format.time_unit != time_unit)
continue;
if (locale_format.style != style)
continue;
if (decode_string(locale_format.tense_or_number) != tense_or_number)
continue;
formats.append(locale_format.to_relative_time_format());
}
return formats;
}
}
)~~~");
TRY(file.write_until_depleted(generator.as_string_view().bytes()));
return {};
}
ErrorOr<int> serenity_main(Main::Arguments arguments)
{
StringView generated_header_path;
StringView generated_implementation_path;
StringView dates_path;
Core::ArgsParser args_parser;
args_parser.add_option(generated_header_path, "Path to the Unicode locale header file to generate", "generated-header-path", 'h', "generated-header-path");
args_parser.add_option(generated_implementation_path, "Path to the Unicode locale implementation file to generate", "generated-implementation-path", 'c', "generated-implementation-path");
args_parser.add_option(dates_path, "Path to cldr-dates directory", "dates-path", 'd', "dates-path");
args_parser.parse(arguments);
auto generated_header_file = TRY(open_file(generated_header_path, Core::File::OpenMode::Write));
auto generated_implementation_file = TRY(open_file(generated_implementation_path, Core::File::OpenMode::Write));
CLDR cldr;
TRY(parse_all_locales(dates_path, cldr));
TRY(generate_unicode_locale_header(*generated_header_file, cldr));
TRY(generate_unicode_locale_implementation(*generated_implementation_file, cldr));
return 0;
}

View File

@ -4,6 +4,7 @@
* SPDX-License-Identifier: BSD-2-Clause
*/
#include <AK/Enumerate.h>
#include <AK/StringBuilder.h>
#include <LibJS/Runtime/AbstractOperations.h>
#include <LibJS/Runtime/Array.h>
@ -23,223 +24,75 @@ RelativeTimeFormat::RelativeTimeFormat(Object& prototype)
{
}
void RelativeTimeFormat::visit_edges(Cell::Visitor& visitor)
{
Base::visit_edges(visitor);
if (m_number_format)
visitor.visit(m_number_format);
if (m_plural_rules)
visitor.visit(m_plural_rules);
}
void RelativeTimeFormat::set_numeric(StringView numeric)
{
if (numeric == "always"sv) {
m_numeric = Numeric::Always;
} else if (numeric == "auto"sv) {
m_numeric = Numeric::Auto;
} else {
VERIFY_NOT_REACHED();
}
}
StringView RelativeTimeFormat::numeric_string() const
{
switch (m_numeric) {
case Numeric::Always:
return "always"sv;
case Numeric::Auto:
return "auto"sv;
default:
VERIFY_NOT_REACHED();
}
}
// 17.5.1 SingularRelativeTimeUnit ( unit ), https://tc39.es/ecma402/#sec-singularrelativetimeunit
ThrowCompletionOr<::Locale::TimeUnit> singular_relative_time_unit(VM& vm, StringView unit)
{
// 1. Assert: Type(unit) is String.
// 2. If unit is "seconds", return "second".
// 1. If unit is "seconds", return "second".
if (unit == "seconds"sv)
return ::Locale::TimeUnit::Second;
// 3. If unit is "minutes", return "minute".
// 2. If unit is "minutes", return "minute".
if (unit == "minutes"sv)
return ::Locale::TimeUnit::Minute;
// 4. If unit is "hours", return "hour".
// 3. If unit is "hours", return "hour".
if (unit == "hours"sv)
return ::Locale::TimeUnit::Hour;
// 5. If unit is "days", return "day".
// 4. If unit is "days", return "day".
if (unit == "days"sv)
return ::Locale::TimeUnit::Day;
// 6. If unit is "weeks", return "week".
// 5. If unit is "weeks", return "week".
if (unit == "weeks"sv)
return ::Locale::TimeUnit::Week;
// 7. If unit is "months", return "month".
// 6. If unit is "months", return "month".
if (unit == "months"sv)
return ::Locale::TimeUnit::Month;
// 8. If unit is "quarters", return "quarter".
// 7. If unit is "quarters", return "quarter".
if (unit == "quarters"sv)
return ::Locale::TimeUnit::Quarter;
// 9. If unit is "years", return "year".
// 8. If unit is "years", return "year".
if (unit == "years"sv)
return ::Locale::TimeUnit::Year;
// 10. If unit is not one of "second", "minute", "hour", "day", "week", "month", "quarter", or "year", throw a RangeError exception.
// 11. Return unit.
// 9. If unit is not one of "second", "minute", "hour", "day", "week", "month", "quarter", or "year", throw a RangeError exception.
// 10. Return unit.
if (auto time_unit = ::Locale::time_unit_from_string(unit); time_unit.has_value())
return *time_unit;
return vm.throw_completion<RangeError>(ErrorType::IntlInvalidUnit, unit);
}
// 17.5.2 PartitionRelativeTimePattern ( relativeTimeFormat, value, unit ), https://tc39.es/ecma402/#sec-PartitionRelativeTimePattern
ThrowCompletionOr<Vector<PatternPartitionWithUnit>> partition_relative_time_pattern(VM& vm, RelativeTimeFormat& relative_time_format, double value, StringView unit)
ThrowCompletionOr<Vector<::Locale::RelativeTimeFormat::Partition>> partition_relative_time_pattern(VM& vm, RelativeTimeFormat& relative_time_format, double value, StringView unit)
{
// 1. Assert: relativeTimeFormat has an [[InitializedRelativeTimeFormat]] internal slot.
// 2. Assert: Type(value) is Number.
// 3. Assert: Type(unit) is String.
// 4. If value is NaN, +∞𝔽, or -∞𝔽, throw a RangeError exception.
// 1. If value is NaN, +∞𝔽, or -∞𝔽, throw a RangeError exception.
if (!Value(value).is_finite_number())
return vm.throw_completion<RangeError>(ErrorType::NumberIsNaNOrInfinity);
// 5. Let unit be ? SingularRelativeTimeUnit(unit).
// 2. Let unit be ? SingularRelativeTimeUnit(unit).
auto time_unit = TRY(singular_relative_time_unit(vm, unit));
// 6. Let localeData be %RelativeTimeFormat%.[[LocaleData]].
// 7. Let dataLocale be relativeTimeFormat.[[DataLocale]].
auto const& data_locale = relative_time_format.data_locale();
// 8. Let fields be localeData.[[<dataLocale>]].
// 9. Let style be relativeTimeFormat.[[Style]].
auto style = relative_time_format.style();
// NOTE: The next steps form a "key" based on combining various formatting options into a string,
// then filtering the large set of locale data down to the pattern we are looking for. Instead,
// LibUnicode expects the individual options as enumeration values, and returns the couple of
// patterns that match those options.
auto find_patterns_for_tense_or_number = [&](StringView tense_or_number) {
// 10. If style is equal to "short", then
// a. Let entry be the string-concatenation of unit and "-short".
// 11. Else if style is equal to "narrow", then
// a. Let entry be the string-concatenation of unit and "-narrow".
// 12. Else,
// a. Let entry be unit.
auto patterns = ::Locale::get_relative_time_format_patterns(data_locale, time_unit, tense_or_number, style);
// 13. If fields doesn't have a field [[<entry>]], then
if (patterns.is_empty()) {
// a. Let entry be unit.
// NOTE: In the CLDR, the lack of "short" or "narrow" in the key implies "long".
patterns = ::Locale::get_relative_time_format_patterns(data_locale, time_unit, tense_or_number, ::Locale::Style::Long);
}
// 14. Let patterns be fields.[[<entry>]].
return patterns;
};
// 15. Let numeric be relativeTimeFormat.[[Numeric]].
// 16. If numeric is equal to "auto", then
if (relative_time_format.numeric() == RelativeTimeFormat::Numeric::Auto) {
// a. Let valueString be ToString(value).
auto value_string = MUST(Value(value).to_string(vm));
// b. If patterns has a field [[<valueString>]], then
if (auto patterns = find_patterns_for_tense_or_number(value_string); !patterns.is_empty()) {
VERIFY(patterns.size() == 1);
// i. Let result be patterns.[[<valueString>]].
auto result = MUST(String::from_utf8(patterns[0].pattern));
// ii. Return a List containing the Record { [[Type]]: "literal", [[Value]]: result }.
return Vector<PatternPartitionWithUnit> { { "literal"sv, move(result) } };
}
}
// 17. If value is -0𝔽 or if value is less than 0, then
StringView tense;
if (Value(value).is_negative_zero() || (value < 0)) {
// a. Let tl be "past".
tense = "past"sv;
// FIXME: The spec does not say to do this, but nothing makes sense after this with a negative value.
value = fabs(value);
}
// 18. Else,
else {
// a. Let tl be "future".
tense = "future"sv;
}
// 19. Let po be patterns.[[<tl>]].
auto patterns = find_patterns_for_tense_or_number(tense);
// 20. Let fv be ! PartitionNumberPattern(relativeTimeFormat.[[NumberFormat]], value).
auto value_partitions = partition_number_pattern(relative_time_format.number_format(), Value(value));
// 21. Let pr be ! ResolvePlural(relativeTimeFormat.[[PluralRules]], value).[[PluralCategory]].
auto plurality = resolve_plural(relative_time_format.plural_rules(), Value(value));
// 22. Let pattern be po.[[<pr>]].
auto pattern = patterns.find_if([&](auto& p) { return p.plurality == plurality.plural_category; });
if (pattern == patterns.end())
return Vector<PatternPartitionWithUnit> {};
// 23. Return ! MakePartsList(pattern, unit, fv).
return make_parts_list(pattern->pattern, ::Locale::time_unit_to_string(time_unit), move(value_partitions));
}
// 17.5.3 MakePartsList ( pattern, unit, parts ), https://tc39.es/ecma402/#sec-makepartslist
Vector<PatternPartitionWithUnit> make_parts_list(StringView pattern, StringView unit, Vector<::Locale::NumberFormat::Partition> parts)
{
// 1. Let patternParts be PartitionPattern(pattern).
auto pattern_parts = partition_pattern(pattern);
// 2. Let result be a new empty List.
Vector<PatternPartitionWithUnit> result;
// 3. For each Record { [[Type]], [[Value]] } patternPart in patternParts, do
for (auto& pattern_part : pattern_parts) {
// a. If patternPart.[[Type]] is "literal", then
if (pattern_part.type == "literal"sv) {
// i. Append Record { [[Type]]: "literal", [[Value]]: patternPart.[[Value]], [[Unit]]: empty } to result.
result.empend("literal"sv, move(pattern_part.value));
}
// b. Else,
else {
// i. Assert: patternPart.[[Type]] is "0".
VERIFY(pattern_part.type == "0"sv);
// ii. For each Record { [[Type]], [[Value]] } part in parts, do
for (auto& part : parts) {
// 1. Append Record { [[Type]]: part.[[Type]], [[Value]]: part.[[Value]], [[Unit]]: unit } to result.
result.empend(part.type, move(part.value), unit);
}
}
}
// 4. Return result.
return result;
return relative_time_format.formatter().format_to_parts(value, time_unit, relative_time_format.numeric());
}
// 17.5.4 FormatRelativeTime ( relativeTimeFormat, value, unit ), https://tc39.es/ecma402/#sec-FormatRelativeTime
ThrowCompletionOr<String> format_relative_time(VM& vm, RelativeTimeFormat& relative_time_format, double value, StringView unit)
{
// 1. Let parts be ? PartitionRelativeTimePattern(relativeTimeFormat, value, unit).
auto parts = TRY(partition_relative_time_pattern(vm, relative_time_format, value, unit));
auto time_unit = TRY([&]() -> ThrowCompletionOr<::Locale::TimeUnit> {
// NOTE: We short-circuit PartitionRelativeTimePattern as we do not need individual partitions. But we must still
// perform the NaN/Infinity sanity checks and unit parsing from its first steps.
// 1. If value is NaN, +∞𝔽, or -∞𝔽, throw a RangeError exception.
if (!Value(value).is_finite_number())
return vm.throw_completion<RangeError>(ErrorType::NumberIsNaNOrInfinity);
// 2. Let unit be ? SingularRelativeTimeUnit(unit).
return TRY(singular_relative_time_unit(vm, unit));
}());
// 2. Let result be an empty String.
StringBuilder result;
// 3. For each Record { [[Type]], [[Value]], [[Unit]] } part in parts, do
for (auto& part : parts) {
// a. Set result to the string-concatenation of result and part.[[Value]].
result.append(part.value);
}
// a. Set result to the string-concatenation of result and part.[[Value]].
// 4. Return result.
return MUST(result.to_string());
return relative_time_format.formatter().format(value, time_unit, relative_time_format.numeric());
}
// 17.5.5 FormatRelativeTimeToParts ( relativeTimeFormat, value, unit ), https://tc39.es/ecma402/#sec-FormatRelativeTimeToParts
@ -254,10 +107,8 @@ ThrowCompletionOr<NonnullGCPtr<Array>> format_relative_time_to_parts(VM& vm, Rel
auto result = MUST(Array::create(realm, 0));
// 3. Let n be 0.
size_t n = 0;
// 4. For each Record { [[Type]], [[Value]], [[Unit]] } part in parts, do
for (auto& part : parts) {
for (auto [n, part] : enumerate(parts)) {
// a. Let O be OrdinaryObjectCreate(%Object.prototype%).
auto object = Object::create(realm, realm.intrinsics().object_prototype());
@ -277,7 +128,6 @@ ThrowCompletionOr<NonnullGCPtr<Array>> format_relative_time_to_parts(VM& vm, Rel
MUST(result->create_data_property_or_throw(n, object));
// f. Increment n by 1.
++n;
}
// 5. Return result.

View File

@ -23,11 +23,6 @@ class RelativeTimeFormat final : public Object {
JS_DECLARE_ALLOCATOR(RelativeTimeFormat);
public:
enum class Numeric {
Always,
Auto,
};
static constexpr auto relevant_extension_keys()
{
// 17.2.3 Internal slots, https://tc39.es/ecma402/#sec-Intl.RelativeTimeFormat-internal-slots
@ -50,43 +45,28 @@ public:
void set_style(StringView style) { m_style = ::Locale::style_from_string(style); }
StringView style_string() const { return ::Locale::style_to_string(m_style); }
Numeric numeric() const { return m_numeric; }
void set_numeric(StringView numeric);
StringView numeric_string() const;
::Locale::NumericDisplay numeric() const { return m_numeric; }
void set_numeric(StringView numeric) { m_numeric = ::Locale::numeric_display_from_string(numeric); }
StringView numeric_string() const { return ::Locale::numeric_display_to_string(m_numeric); }
NumberFormat& number_format() const { return *m_number_format; }
void set_number_format(NumberFormat* number_format) { m_number_format = number_format; }
PluralRules& plural_rules() const { return *m_plural_rules; }
void set_plural_rules(PluralRules* plural_rules) { m_plural_rules = plural_rules; }
::Locale::RelativeTimeFormat const& formatter() const { return *m_formatter; }
void set_formatter(NonnullOwnPtr<::Locale::RelativeTimeFormat> formatter) { m_formatter = move(formatter); }
private:
explicit RelativeTimeFormat(Object& prototype);
virtual void visit_edges(Cell::Visitor&) override;
String m_locale; // [[Locale]]
String m_data_locale; // [[DataLocale]]
String m_numbering_system; // [[NumberingSystem]]
::Locale::Style m_style { ::Locale::Style::Long }; // [[Style]]
::Locale::NumericDisplay m_numeric { ::Locale::NumericDisplay::Always }; // [[Numeric]]
String m_locale; // [[Locale]]
String m_data_locale; // [[DataLocale]]
String m_numbering_system; // [[NumberingSystem]]
::Locale::Style m_style { ::Locale::Style::Long }; // [[Style]]
Numeric m_numeric { Numeric::Always }; // [[Numeric]]
GCPtr<NumberFormat> m_number_format; // [[NumberFormat]]
GCPtr<PluralRules> m_plural_rules; // [[PluralRules]]
};
struct PatternPartitionWithUnit : public PatternPartition {
PatternPartitionWithUnit(StringView type, String value, StringView unit_string = {})
: PatternPartition(type, move(value))
, unit(unit_string)
{
}
StringView unit;
// Non-standard. Stores the ICU relative-time formatter for the Intl object's formatting options.
OwnPtr<::Locale::RelativeTimeFormat> m_formatter;
};
ThrowCompletionOr<::Locale::TimeUnit> singular_relative_time_unit(VM&, StringView unit);
ThrowCompletionOr<Vector<PatternPartitionWithUnit>> partition_relative_time_pattern(VM&, RelativeTimeFormat&, double value, StringView unit);
Vector<PatternPartitionWithUnit> make_parts_list(StringView pattern, StringView unit, Vector<::Locale::NumberFormat::Partition> parts);
ThrowCompletionOr<Vector<::Locale::RelativeTimeFormat::Partition>> partition_relative_time_pattern(VM&, RelativeTimeFormat&, double value, StringView unit);
ThrowCompletionOr<String> format_relative_time(VM&, RelativeTimeFormat&, double value, StringView unit);
ThrowCompletionOr<NonnullGCPtr<Array>> format_relative_time_to_parts(VM&, RelativeTimeFormat&, double value, StringView unit);

View File

@ -1,5 +1,5 @@
/*
* Copyright (c) 2022-2023, Tim Flynn <trflynn89@serenityos.org>
* Copyright (c) 2022-2024, Tim Flynn <trflynn89@serenityos.org>
*
* SPDX-License-Identifier: BSD-2-Clause
*/
@ -8,10 +8,6 @@
#include <LibJS/Runtime/Array.h>
#include <LibJS/Runtime/GlobalObject.h>
#include <LibJS/Runtime/Intl/AbstractOperations.h>
#include <LibJS/Runtime/Intl/NumberFormat.h>
#include <LibJS/Runtime/Intl/NumberFormatConstructor.h>
#include <LibJS/Runtime/Intl/PluralRules.h>
#include <LibJS/Runtime/Intl/PluralRulesConstructor.h>
#include <LibJS/Runtime/Intl/RelativeTimeFormat.h>
#include <LibJS/Runtime/Intl/RelativeTimeFormatConstructor.h>
#include <LibLocale/Locale.h>
@ -52,14 +48,77 @@ ThrowCompletionOr<NonnullGCPtr<Object>> RelativeTimeFormatConstructor::construct
{
auto& vm = this->vm();
auto locales = vm.argument(0);
auto options = vm.argument(1);
auto locales_value = vm.argument(0);
auto options_value = vm.argument(1);
// 2. Let relativeTimeFormat be ? OrdinaryCreateFromConstructor(NewTarget, "%RelativeTimeFormat.prototype%", « [[InitializedRelativeTimeFormat]], [[Locale]], [[DataLocale]], [[Style]], [[Numeric]], [[NumberFormat]], [[NumberingSystem]], [[PluralRules]] »).
// 2. Let relativeTimeFormat be ? OrdinaryCreateFromConstructor(NewTarget, "%RelativeTimeFormat.prototype%", « [[InitializedRelativeTimeFormat]], [[Locale]], [[LocaleData]], [[Style]], [[Numeric]], [[NumberFormat]], [[NumberingSystem]], [[PluralRules]] »).
auto relative_time_format = TRY(ordinary_create_from_constructor<RelativeTimeFormat>(vm, new_target, &Intrinsics::intl_relative_time_format_prototype));
// 3. Return ? InitializeRelativeTimeFormat(relativeTimeFormat, locales, options).
return TRY(initialize_relative_time_format(vm, relative_time_format, locales, options));
// 3. Let requestedLocales be ? CanonicalizeLocaleList(locales).
auto requested_locales = TRY(canonicalize_locale_list(vm, locales_value));
// 4. Set options to ? CoerceOptionsToObject(options).
auto* options = TRY(coerce_options_to_object(vm, options_value));
// 5. Let opt be a new Record.
LocaleOptions opt {};
// 6. Let matcher be ? GetOption(options, "localeMatcher", string, « "lookup", "best fit" », "best fit").
auto matcher = TRY(get_option(vm, *options, vm.names.localeMatcher, OptionType::String, AK::Array { "lookup"sv, "best fit"sv }, "best fit"sv));
// 7. Set opt.[[LocaleMatcher]] to matcher.
opt.locale_matcher = matcher;
// 8. Let numberingSystem be ? GetOption(options, "numberingSystem", string, empty, undefined).
auto numbering_system = TRY(get_option(vm, *options, vm.names.numberingSystem, OptionType::String, {}, Empty {}));
// 9. If numberingSystem is not undefined, then
if (!numbering_system.is_undefined()) {
// a. If numberingSystem cannot be matched by the type Unicode locale nonterminal, throw a RangeError exception.
if (!::Locale::is_type_identifier(numbering_system.as_string().utf8_string_view()))
return vm.throw_completion<RangeError>(ErrorType::OptionIsNotValidValue, numbering_system, "numberingSystem"sv);
// 10. Set opt.[[nu]] to numberingSystem.
opt.nu = numbering_system.as_string().utf8_string();
}
// 11. Let r be ResolveLocale(%Intl.RelativeTimeFormat%.[[AvailableLocales]], requestedLocales, opt, %Intl.RelativeTimeFormat%.[[RelevantExtensionKeys]], %Intl.RelativeTimeFormat%.[[LocaleData]]).
auto result = resolve_locale(requested_locales, opt, RelativeTimeFormat::relevant_extension_keys());
// 12. Let locale be r.[[Locale]].
auto locale = move(result.locale);
// 13. Set relativeTimeFormat.[[Locale]] to locale.
relative_time_format->set_locale(locale);
// 14. Set relativeTimeFormat.[[LocaleData]] to r.[[LocaleData]].
relative_time_format->set_data_locale(move(result.data_locale));
// 15. Set relativeTimeFormat.[[NumberingSystem]] to r.[[nu]].
if (result.nu.has_value())
relative_time_format->set_numbering_system(result.nu.release_value());
// 16. Let style be ? GetOption(options, "style", string, « "long", "short", "narrow" », "long").
auto style = TRY(get_option(vm, *options, vm.names.style, OptionType::String, { "long"sv, "short"sv, "narrow"sv }, "long"sv));
// 17. Set relativeTimeFormat.[[Style]] to style.
relative_time_format->set_style(style.as_string().utf8_string_view());
// 18. Let numeric be ? GetOption(options, "numeric", string, « "always", "auto" », "always").
auto numeric = TRY(get_option(vm, *options, vm.names.numeric, OptionType::String, { "always"sv, "auto"sv }, "always"sv));
// 19. Set relativeTimeFormat.[[Numeric]] to numeric.
relative_time_format->set_numeric(numeric.as_string().utf8_string_view());
// 20. Let relativeTimeFormat.[[NumberFormat]] be ! Construct(%Intl.NumberFormat%, « locale »).
// 21. Let relativeTimeFormat.[[PluralRules]] be ! Construct(%Intl.PluralRules%, « locale »).
auto formatter = ::Locale::RelativeTimeFormat::create(
relative_time_format->locale(),
relative_time_format->style());
relative_time_format->set_formatter(move(formatter));
// 22. Return relativeTimeFormat.
return relative_time_format;
}
// 17.2.2 Intl.RelativeTimeFormat.supportedLocalesOf ( locales [ , options ] ), https://tc39.es/ecma402/#sec-Intl.RelativeTimeFormat.supportedLocalesOf
@ -77,78 +136,4 @@ JS_DEFINE_NATIVE_FUNCTION(RelativeTimeFormatConstructor::supported_locales_of)
return TRY(supported_locales(vm, requested_locales, options));
}
// 17.1.2 InitializeRelativeTimeFormat ( relativeTimeFormat, locales, options ), https://tc39.es/ecma402/#sec-InitializeRelativeTimeFormat
ThrowCompletionOr<NonnullGCPtr<RelativeTimeFormat>> initialize_relative_time_format(VM& vm, RelativeTimeFormat& relative_time_format, Value locales_value, Value options_value)
{
auto& realm = *vm.current_realm();
// 1. Let requestedLocales be ? CanonicalizeLocaleList(locales).
auto requested_locales = TRY(canonicalize_locale_list(vm, locales_value));
// 2. Set options to ? CoerceOptionsToObject(options).
auto* options = TRY(coerce_options_to_object(vm, options_value));
// 3. Let opt be a new Record.
LocaleOptions opt {};
// 4. Let matcher be ? GetOption(options, "localeMatcher", string, « "lookup", "best fit" », "best fit").
auto matcher = TRY(get_option(vm, *options, vm.names.localeMatcher, OptionType::String, AK::Array { "lookup"sv, "best fit"sv }, "best fit"sv));
// 5. Set opt.[[LocaleMatcher]] to matcher.
opt.locale_matcher = matcher;
// 6. Let numberingSystem be ? GetOption(options, "numberingSystem", string, empty, undefined).
auto numbering_system = TRY(get_option(vm, *options, vm.names.numberingSystem, OptionType::String, {}, Empty {}));
// 7. If numberingSystem is not undefined, then
if (!numbering_system.is_undefined()) {
// a. If numberingSystem does not match the Unicode Locale Identifier type nonterminal, throw a RangeError exception.
if (!::Locale::is_type_identifier(numbering_system.as_string().utf8_string_view()))
return vm.throw_completion<RangeError>(ErrorType::OptionIsNotValidValue, numbering_system, "numberingSystem"sv);
// 8. Set opt.[[nu]] to numberingSystem.
opt.nu = numbering_system.as_string().utf8_string();
}
// 9. Let localeData be %RelativeTimeFormat%.[[LocaleData]].
// 10. Let r be ResolveLocale(%RelativeTimeFormat%.[[AvailableLocales]], requestedLocales, opt, %RelativeTimeFormat%.[[RelevantExtensionKeys]], localeData).
auto result = resolve_locale(requested_locales, opt, RelativeTimeFormat::relevant_extension_keys());
// 11. Let locale be r.[[locale]].
auto locale = move(result.locale);
// 12. Set relativeTimeFormat.[[Locale]] to locale.
relative_time_format.set_locale(locale);
// 13. Set relativeTimeFormat.[[DataLocale]] to r.[[dataLocale]].
relative_time_format.set_data_locale(move(result.data_locale));
// 14. Set relativeTimeFormat.[[NumberingSystem]] to r.[[nu]].
if (result.nu.has_value())
relative_time_format.set_numbering_system(result.nu.release_value());
// 15. Let style be ? GetOption(options, "style", string, « "long", "short", "narrow" », "long").
auto style = TRY(get_option(vm, *options, vm.names.style, OptionType::String, { "long"sv, "short"sv, "narrow"sv }, "long"sv));
// 16. Set relativeTimeFormat.[[Style]] to style.
relative_time_format.set_style(style.as_string().utf8_string_view());
// 17. Let numeric be ? GetOption(options, "numeric", string, « "always", "auto" », "always").
auto numeric = TRY(get_option(vm, *options, vm.names.numeric, OptionType::String, { "always"sv, "auto"sv }, "always"sv));
// 18. Set relativeTimeFormat.[[Numeric]] to numeric.
relative_time_format.set_numeric(numeric.as_string().utf8_string_view());
// 19. Let relativeTimeFormat.[[NumberFormat]] be ! Construct(%NumberFormat%, « locale »).
auto number_format = MUST(construct(vm, realm.intrinsics().intl_number_format_constructor(), PrimitiveString::create(vm, locale)));
relative_time_format.set_number_format(static_cast<NumberFormat*>(number_format.ptr()));
// 20. Let relativeTimeFormat.[[PluralRules]] be ! Construct(%PluralRules%, « locale »).
auto plural_rules = MUST(construct(vm, realm.intrinsics().intl_plural_rules_constructor(), PrimitiveString::create(vm, locale)));
relative_time_format.set_plural_rules(static_cast<PluralRules*>(plural_rules.ptr()));
// 21. Return relativeTimeFormat.
return relative_time_format;
}
}

View File

@ -1,5 +1,5 @@
/*
* Copyright (c) 2022, Tim Flynn <trflynn89@serenityos.org>
* Copyright (c) 2022-2024, Tim Flynn <trflynn89@serenityos.org>
*
* SPDX-License-Identifier: BSD-2-Clause
*/
@ -29,6 +29,4 @@ private:
JS_DECLARE_NATIVE_FUNCTION(supported_locales_of);
};
ThrowCompletionOr<NonnullGCPtr<RelativeTimeFormat>> initialize_relative_time_format(VM& vm, RelativeTimeFormat& relative_time_format, Value locales_value, Value options_value);
}

View File

@ -1,11 +1,20 @@
/*
* Copyright (c) 2022-2023, Tim Flynn <trflynn89@serenityos.org>
* Copyright (c) 2022-2024, Tim Flynn <trflynn89@serenityos.org>
*
* SPDX-License-Identifier: BSD-2-Clause
*/
#define AK_DONT_REPLACE_STD
#include <LibLocale/ICU.h>
#include <LibLocale/Locale.h>
#include <LibLocale/NumberFormat.h>
#include <LibLocale/RelativeTimeFormat.h>
#include <unicode/decimfmt.h>
#include <unicode/numfmt.h>
#include <unicode/reldatefmt.h>
namespace Locale {
Optional<TimeUnit> time_unit_from_string(StringView time_unit)
@ -48,11 +57,211 @@ StringView time_unit_to_string(TimeUnit time_unit)
return "quarter"sv;
case TimeUnit::Year:
return "year"sv;
default:
VERIFY_NOT_REACHED();
}
VERIFY_NOT_REACHED();
}
Vector<RelativeTimeFormat> __attribute__((weak)) get_relative_time_format_patterns(StringView, TimeUnit, StringView, Style) { return {}; }
static constexpr URelativeDateTimeUnit icu_time_unit(TimeUnit unit)
{
switch (unit) {
case TimeUnit::Second:
return URelativeDateTimeUnit::UDAT_REL_UNIT_SECOND;
case TimeUnit::Minute:
return URelativeDateTimeUnit::UDAT_REL_UNIT_MINUTE;
case TimeUnit::Hour:
return URelativeDateTimeUnit::UDAT_REL_UNIT_HOUR;
case TimeUnit::Day:
return URelativeDateTimeUnit::UDAT_REL_UNIT_DAY;
case TimeUnit::Week:
return URelativeDateTimeUnit::UDAT_REL_UNIT_WEEK;
case TimeUnit::Month:
return URelativeDateTimeUnit::UDAT_REL_UNIT_MONTH;
case TimeUnit::Quarter:
return URelativeDateTimeUnit::UDAT_REL_UNIT_QUARTER;
case TimeUnit::Year:
return URelativeDateTimeUnit::UDAT_REL_UNIT_YEAR;
}
VERIFY_NOT_REACHED();
}
NumericDisplay numeric_display_from_string(StringView numeric_display)
{
if (numeric_display == "always"sv)
return NumericDisplay::Always;
if (numeric_display == "auto"sv)
return NumericDisplay::Auto;
VERIFY_NOT_REACHED();
}
StringView numeric_display_to_string(NumericDisplay numeric_display)
{
switch (numeric_display) {
case NumericDisplay::Always:
return "always"sv;
case NumericDisplay::Auto:
return "auto"sv;
}
VERIFY_NOT_REACHED();
}
static constexpr UDateRelativeDateTimeFormatterStyle icu_relative_date_time_style(Style unit_display)
{
switch (unit_display) {
case Style::Long:
return UDAT_STYLE_LONG;
case Style::Short:
return UDAT_STYLE_SHORT;
case Style::Narrow:
return UDAT_STYLE_NARROW;
}
VERIFY_NOT_REACHED();
}
// ICU does not contain a field enumeration for "literal" partitions. Define a custom field so that we may provide a
// type for those partitions.
static constexpr i32 LITERAL_FIELD = -1;
static constexpr StringView icu_relative_time_format_field_to_string(i32 field)
{
switch (field) {
case LITERAL_FIELD:
return "literal"sv;
case UNUM_INTEGER_FIELD:
return "integer"sv;
case UNUM_FRACTION_FIELD:
return "fraction"sv;
case UNUM_DECIMAL_SEPARATOR_FIELD:
return "decimal"sv;
case UNUM_GROUPING_SEPARATOR_FIELD:
return "group"sv;
}
VERIFY_NOT_REACHED();
}
struct Range {
i32 field { 0 };
i32 start { 0 };
i32 end { 0 };
};
class RelativeTimeFormatImpl : public RelativeTimeFormat {
public:
explicit RelativeTimeFormatImpl(NonnullOwnPtr<icu::RelativeDateTimeFormatter> formatter)
: m_formatter(move(formatter))
{
}
virtual ~RelativeTimeFormatImpl() override = default;
virtual String format(double time, TimeUnit unit, NumericDisplay numeric_display) const override
{
UErrorCode status = U_ZERO_ERROR;
auto formatted = format_impl(time, unit, numeric_display);
auto formatted_time = formatted->toTempString(status);
if (icu_failure(status))
return {};
return icu_string_to_string(formatted_time);
}
virtual Vector<Partition> format_to_parts(double time, TimeUnit unit, NumericDisplay numeric_display) const override
{
UErrorCode status = U_ZERO_ERROR;
auto formatted = format_impl(time, unit, numeric_display);
auto unit_string = time_unit_to_string(unit);
auto formatted_time = formatted->toTempString(status);
if (icu_failure(status))
return {};
Vector<Partition> result;
Vector<Range> separators;
auto create_partition = [&](i32 field, i32 begin, i32 end, bool is_unit) {
Partition partition;
partition.type = icu_relative_time_format_field_to_string(field);
partition.value = icu_string_to_string(formatted_time.tempSubStringBetween(begin, end));
if (is_unit)
partition.unit = unit_string;
result.append(move(partition));
};
icu::ConstrainedFieldPosition position;
position.constrainCategory(UFIELD_CATEGORY_NUMBER);
i32 previous_end_index = 0;
while (static_cast<bool>(formatted->nextPosition(position, status)) && icu_success(status)) {
if (position.getField() == UNUM_GROUPING_SEPARATOR_FIELD) {
separators.empend(position.getField(), position.getStart(), position.getLimit());
continue;
}
if (previous_end_index < position.getStart())
create_partition(LITERAL_FIELD, previous_end_index, position.getStart(), false);
auto start = position.getStart();
if (position.getField() == UNUM_INTEGER_FIELD) {
for (auto const& separator : separators) {
if (start >= separator.start)
continue;
create_partition(position.getField(), start, separator.start, true);
create_partition(separator.field, separator.start, separator.end, true);
start = separator.end;
break;
}
}
create_partition(position.getField(), start, position.getLimit(), true);
previous_end_index = position.getLimit();
}
if (previous_end_index < formatted_time.length())
create_partition(LITERAL_FIELD, previous_end_index, formatted_time.length(), false);
return result;
}
private:
Optional<icu::FormattedRelativeDateTime> format_impl(double time, TimeUnit unit, NumericDisplay numeric_display) const
{
UErrorCode status = U_ZERO_ERROR;
auto formatted = numeric_display == NumericDisplay::Always
? m_formatter->formatNumericToValue(time, icu_time_unit(unit), status)
: m_formatter->formatToValue(time, icu_time_unit(unit), status);
if (icu_failure(status))
return {};
return formatted;
}
NonnullOwnPtr<icu::RelativeDateTimeFormatter> m_formatter;
};
NonnullOwnPtr<RelativeTimeFormat> RelativeTimeFormat::create(StringView locale, Style style)
{
UErrorCode status = U_ZERO_ERROR;
auto locale_data = LocaleData::for_locale(locale);
VERIFY(locale_data.has_value());
auto* number_formatter = icu::NumberFormat::createInstance(locale_data->locale(), UNUM_DECIMAL, status);
VERIFY(locale_data.has_value());
if (number_formatter->getDynamicClassID() == icu::DecimalFormat::getStaticClassID())
static_cast<icu::DecimalFormat&>(*number_formatter).setMinimumGroupingDigits(UNUM_MINIMUM_GROUPING_DIGITS_AUTO);
auto formatter = make<icu::RelativeDateTimeFormatter>(locale_data->locale(), number_formatter, icu_relative_date_time_style(style), UDISPCTX_CAPITALIZATION_NONE, status);
VERIFY(icu_success(status));
return make<RelativeTimeFormatImpl>(move(formatter));
}
}

View File

@ -1,16 +1,17 @@
/*
* Copyright (c) 2022-2023, Tim Flynn <trflynn89@serenityos.org>
* Copyright (c) 2022-2024, Tim Flynn <trflynn89@serenityos.org>
*
* SPDX-License-Identifier: BSD-2-Clause
*/
#pragma once
#include <AK/NonnullOwnPtr.h>
#include <AK/Optional.h>
#include <AK/String.h>
#include <AK/StringView.h>
#include <AK/Vector.h>
#include <LibLocale/Forward.h>
#include <LibLocale/Locale.h>
namespace Locale {
@ -25,15 +26,32 @@ enum class TimeUnit {
Quarter,
Year,
};
Optional<TimeUnit> time_unit_from_string(StringView);
StringView time_unit_to_string(TimeUnit);
struct RelativeTimeFormat {
PluralCategory plurality;
StringView pattern;
enum class NumericDisplay {
Always,
Auto,
};
NumericDisplay numeric_display_from_string(StringView);
StringView numeric_display_to_string(NumericDisplay);
class RelativeTimeFormat {
public:
static NonnullOwnPtr<RelativeTimeFormat> create(StringView locale, Style style);
virtual ~RelativeTimeFormat() = default;
struct Partition {
StringView type;
String value;
StringView unit;
};
virtual String format(double, TimeUnit, NumericDisplay) const = 0;
virtual Vector<Partition> format_to_parts(double, TimeUnit, NumericDisplay) const = 0;
protected:
RelativeTimeFormat() = default;
};
Optional<TimeUnit> time_unit_from_string(StringView time_unit);
StringView time_unit_to_string(TimeUnit time_unit);
Vector<RelativeTimeFormat> get_relative_time_format_patterns(StringView locale, TimeUnit time_unit, StringView tense_or_number, Style style);
}