LibJS+LibUnicode: Port Intl.DateTimeFormat to UTF-16 strings

Author: https://github.com/trflynn89
Commit: ee01f857d1
Pull-request: https://github.com/LadybirdBrowser/ladybird/pull/5581

Libraries/LibJS/Runtime/Intl/DateTimeFormat.cpp

@@ -119,11 +119,11 @@ ThrowCompletionOr<Vector<Unicode::DateTimeFormat::Partition>> partition_date_tim
 
 // 11.5.7 FormatDateTime ( dateTimeFormat, x ), https://tc39.es/ecma402/#sec-formatdatetime
 // 15.9.6 FormatDateTime ( dateTimeFormat, x ), https://tc39.es/proposal-temporal/#sec-formatdatetime
-ThrowCompletionOr<String> format_date_time(VM& vm, DateTimeFormat& date_time_format, FormattableDateTime const& time)
+ThrowCompletionOr<Utf16String> format_date_time(VM& vm, DateTimeFormat& date_time_format, FormattableDateTime const& time)
 {
     // 1. Let parts be ? PartitionDateTimePattern(dateTimeFormat, x).
     // 2. Let result be the empty String.
-    String result;
+    Utf16String result;
 
     // NOTE: We short-circuit PartitionDateTimePattern as we do not need individual partitions.
     {
@@ -196,11 +196,11 @@ ThrowCompletionOr<Vector<Unicode::DateTimeFormat::Partition>> partition_date_tim
 
 // 11.5.10 FormatDateTimeRange ( dateTimeFormat, x, y ), https://tc39.es/ecma402/#sec-formatdatetimerange
 // 15.9.9 FormatDateTimeRange ( dateTimeFormat, x, y ), https://tc39.es/proposal-temporal/#sec-formatdatetimerange
-ThrowCompletionOr<String> format_date_time_range(VM& vm, DateTimeFormat& date_time_format, FormattableDateTime const& start, FormattableDateTime const& end)
+ThrowCompletionOr<Utf16String> format_date_time_range(VM& vm, DateTimeFormat& date_time_format, FormattableDateTime const& start, FormattableDateTime const& end)
 {
     // 1. Let parts be ? PartitionDateTimeRangePattern(dateTimeFormat, x, y).
     // 2. Let result be the empty String.
-    String result;
+    Utf16String result;
 
     // NOTE: We short-circuit PartitionDateTimeRangePattern as we do not need individual partitions.
     {

Libraries/LibJS/Runtime/Intl/DateTimeFormat.h

@@ -137,10 +137,10 @@ struct ValueFormat {
 
 Vector<Unicode::DateTimeFormat::Partition> format_date_time_pattern(ValueFormat const&);
 ThrowCompletionOr<Vector<Unicode::DateTimeFormat::Partition>> partition_date_time_pattern(VM&, DateTimeFormat&, FormattableDateTime const&);
-ThrowCompletionOr<String> format_date_time(VM&, DateTimeFormat&, FormattableDateTime const&);
+ThrowCompletionOr<Utf16String> format_date_time(VM&, DateTimeFormat&, FormattableDateTime const&);
 ThrowCompletionOr<GC::Ref<Array>> format_date_time_to_parts(VM&, DateTimeFormat&, FormattableDateTime const&);
 ThrowCompletionOr<Vector<Unicode::DateTimeFormat::Partition>> partition_date_time_range_pattern(VM&, DateTimeFormat&, FormattableDateTime const& start, FormattableDateTime const& end);
-ThrowCompletionOr<String> format_date_time_range(VM&, DateTimeFormat&, FormattableDateTime const& start, FormattableDateTime const& end);
+ThrowCompletionOr<Utf16String> format_date_time_range(VM&, DateTimeFormat&, FormattableDateTime const& start, FormattableDateTime const& end);
 ThrowCompletionOr<GC::Ref<Array>> format_date_time_range_to_parts(VM&, DateTimeFormat&, FormattableDateTime const& start, FormattableDateTime const& end);
 
 Optional<Unicode::CalendarPattern> get_date_time_format(Unicode::CalendarPattern const& options, OptionRequired, OptionDefaults, OptionInherit);

Libraries/LibUnicode/DateTimeFormat.cpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2021-2024, Tim Flynn <trflynn89@serenityos.org>
+ * Copyright (c) 2021-2025, Tim Flynn <trflynn89@ladybird.org>
  *
  * SPDX-License-Identifier: BSD-2-Clause
  */
@@ -672,13 +672,13 @@ public:
 
     virtual CalendarPattern const& chosen_pattern() const override { return m_pattern; }
 
-    virtual String format(double time) const override
+    virtual Utf16String format(double time) const override
     {
         auto formatted_time = format_impl(time);
         if (!formatted_time.has_value())
             return {};
 
-        return icu_string_to_string(*formatted_time);
+        return icu_string_to_utf16_string(*formatted_time);
     }
 
     virtual Vector<Partition> format_to_parts(double time) const override
@@ -694,7 +694,7 @@ public:
         auto create_partition = [&](i32 field, i32 begin, i32 end) {
             Partition partition;
             partition.type = icu_date_time_format_field_to_string(field);
-            partition.value = icu_string_to_string(formatted_time->tempSubStringBetween(begin, end));
+            partition.value = icu_string_to_utf16_string(formatted_time->tempSubStringBetween(begin, end));
             partition.source = "shared"sv;
             result.append(move(partition));
         };
@@ -717,7 +717,7 @@ public:
         return result;
     }
 
-    virtual String format_range(double start, double end) const override
+    virtual Utf16String format_range(double start, double end) const override
     {
         UErrorCode status = U_ZERO_ERROR;
 
@@ -733,7 +733,7 @@ public:
             return {};
 
         normalize_spaces(formatted_time);
-        return icu_string_to_string(formatted_time);
+        return icu_string_to_utf16_string(formatted_time);
     }
 
     virtual Vector<Partition> format_range_to_parts(double start, double end) const override
@@ -763,7 +763,7 @@ public:
         auto create_partition = [&](i32 field, i32 begin, i32 end) {
             Partition partition;
             partition.type = icu_date_time_format_field_to_string(field);
-            partition.value = icu_string_to_string(formatted_time.tempSubStringBetween(begin, end));
+            partition.value = icu_string_to_utf16_string(formatted_time.tempSubStringBetween(begin, end));
 
             if (start_range.has_value() && start_range->contains(begin))
                 partition.source = "startRange"sv;

Libraries/LibUnicode/DateTimeFormat.h

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2021-2024, Tim Flynn <trflynn89@serenityos.org>
+ * Copyright (c) 2021-2025, Tim Flynn <trflynn89@ladybird.org>
  *
  * SPDX-License-Identifier: BSD-2-Clause
  */
@@ -8,10 +8,10 @@
 
 #include <AK/IterationDecision.h>
 #include <AK/Optional.h>
-#include <AK/String.h>
 #include <AK/StringView.h>
 #include <AK/Time.h>
 #include <AK/Types.h>
+#include <AK/Utf16String.h>
 #include <AK/Vector.h>
 #include <LibUnicode/Forward.h>
 
@@ -151,16 +151,16 @@ public:
 
     struct Partition {
         StringView type;
-        String value;
+        Utf16String value;
         StringView source;
     };
 
     virtual CalendarPattern const& chosen_pattern() const = 0;
 
-    virtual String format(double) const = 0;
+    virtual Utf16String format(double) const = 0;
     virtual Vector<Partition> format_to_parts(double) const = 0;
 
-    virtual String format_range(double, double) const = 0;
+    virtual Utf16String format_range(double, double) const = 0;
     virtual Vector<Partition> format_range_to_parts(double, double) const = 0;
 
 protected:

Libraries/LibUnicode/ICU.cpp

@@ -162,6 +162,11 @@ String icu_string_to_string(UChar const* string, i32 length)
     return MUST(Utf16View { string, static_cast<size_t>(length) }.to_utf8());
 }
 
+Utf16String icu_string_to_utf16_string(icu::UnicodeString const& string)
+{
+    return Utf16String::from_utf16_without_validation({ string.getBuffer(), static_cast<size_t>(string.length()) });
+}
+
 UCharIterator icu_string_iterator(Utf16View const& string)
 {
     UCharIterator iterator;

Libraries/LibUnicode/ICU.h

@@ -10,6 +10,7 @@
 #include <AK/OwnPtr.h>
 #include <AK/String.h>
 #include <AK/StringView.h>
+#include <AK/Utf16String.h>
 #include <AK/Vector.h>
 #include <LibUnicode/DurationFormat.h>
 
@@ -103,6 +104,8 @@ Vector<icu::UnicodeString> icu_string_list(ReadonlySpan<String> strings);
 String icu_string_to_string(icu::UnicodeString const& string);
 String icu_string_to_string(UChar const*, i32 length);
 
+Utf16String icu_string_to_utf16_string(icu::UnicodeString const& string);
+
 UCharIterator icu_string_iterator(Utf16View const&);
 
 template<typename Filter>