[web] Use v8BreakIterator where possible (flutter#37317)

* [web] Use v8BreakIterator where possible

* address review comments

lib/web_ui/lib/src/engine/dom.dart

@@ -66,6 +66,9 @@ extension DomWindowExtension on DomWindow {
   /// The Trusted Types API (when available).
   /// See: https://developer.mozilla.org/en-US/docs/Web/API/Trusted_Types_API
   external DomTrustedTypePolicyFactory? get trustedTypes;
+
+  // ignore: non_constant_identifier_names
+  external DomIntl get Intl;
 }
 
 typedef DomRequestAnimationFrameCallback = void Function(num highResTime);
@@ -1659,3 +1662,42 @@ class _DomListWrapper<T> extends Iterable<T> {
 /// `toList` on the `Iterable`.
 Iterable<T> createDomListWrapper<T>(_DomList list) =>
     _DomListWrapper<T>._(list).cast<T>();
+
+@JS()
+@staticInterop
+class DomIntl {}
+
+extension DomIntlExtension on DomIntl {
+  /// This is a V8-only API for segmenting text.
+  ///
+  /// See: https://code.google.com/archive/p/v8-i18n/wikis/BreakIterator.wiki
+  external Object? get v8BreakIterator;
+}
+
+
+@JS()
+@staticInterop
+class DomV8BreakIterator {}
+
+extension DomV8BreakIteratorExtension on DomV8BreakIterator {
+  external void adoptText(String text);
+  external int first();
+  external int next();
+  external int current();
+  external String breakType();
+}
+
+DomV8BreakIterator createV8BreakIterator() {
+  final Object? v8BreakIterator = domWindow.Intl.v8BreakIterator;
+  if (v8BreakIterator == null) {
+    throw UnimplementedError('v8BreakIterator is not supported.');
+  }
+
+  return js_util.callConstructor<DomV8BreakIterator>(
+    v8BreakIterator,
+    <Object?>[
+      js_util.getProperty(domWindow, 'undefined'),
+      js_util.jsify(const <String, String>{'type': 'line'}),
+    ],
+  );
+}

lib/web_ui/lib/src/engine/text/line_breaker.dart

@@ -2,10 +2,25 @@
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
+import '../dom.dart';
 import 'fragmenter.dart';
 import 'line_break_properties.dart';
 import 'unicode_range.dart';
 
+const Set<int> _kNewlines = <int>{
+  0x000A, // LF
+  0x000B, // BK
+  0x000C, // BK
+  0x000D, // CR
+  0x0085, // NL
+  0x2028, // BK
+  0x2029, // BK
+};
+const Set<int> _kSpaces = <int>{
+  0x0020, // SP
+  0x200B, // ZW
+};
+
 /// Various types of line breaks as defined by the Unicode spec.
 enum LineBreakType {
   /// Indicates that a line break is possible but not mandatory.
@@ -25,15 +40,107 @@ enum LineBreakType {
 }
 
 /// Splits [text] into fragments based on line breaks.
-class LineBreakFragmenter extends TextFragmenter {
-  const LineBreakFragmenter(super.text);
+abstract class LineBreakFragmenter extends TextFragmenter {
+  factory LineBreakFragmenter(String text) {
+    if (domWindow.Intl.v8BreakIterator != null) {
+      return V8LineBreakFragmenter(text);
+    }
+    return FWLineBreakFragmenter(text);
+  }
+
+  @override
+  List<LineBreakFragment> fragment();
+}
+
+/// Flutter web's custom implementation of [LineBreakFragmenter].
+class FWLineBreakFragmenter extends TextFragmenter implements LineBreakFragmenter {
+  FWLineBreakFragmenter(super.text);
 
   @override
   List<LineBreakFragment> fragment() {
     return _computeLineBreakFragments(text);
   }
 }
 
+/// An implementation of [LineBreakFragmenter] that uses V8's
+/// `v8BreakIterator` API to find line breaks in the given [text].
+class V8LineBreakFragmenter extends TextFragmenter implements LineBreakFragmenter {
+  V8LineBreakFragmenter(super.text)
+      : assert(domWindow.Intl.v8BreakIterator != null);
+
+  @override
+  List<LineBreakFragment> fragment() {
+    final List<LineBreakFragment> breaks = <LineBreakFragment>[];
+    int fragmentStart = 0;
+
+    final DomV8BreakIterator iterator = createV8BreakIterator();
+
+    iterator.adoptText(text);
+    iterator.first();
+    while (iterator.next() != -1) {
+      final LineBreakType type = _getBreakType(iterator);
+
+      final int fragmentEnd = iterator.current();
+      int trailingNewlines = 0;
+      int trailingSpaces = 0;
+
+      // Calculate trailing newlines and spaces.
+      for (int i = fragmentStart; i < fragmentEnd; i++) {
+        final int codeUnit = text.codeUnitAt(i);
+        if (_kNewlines.contains(codeUnit)) {
+          trailingNewlines++;
+          trailingSpaces++;
+        } else if (_kSpaces.contains(codeUnit)) {
+          trailingSpaces++;
+        } else {
+          // Always break after a sequence of spaces.
+          if (trailingSpaces > 0) {
+            breaks.add(LineBreakFragment(
+              fragmentStart,
+              i,
+              LineBreakType.opportunity,
+              trailingNewlines: trailingNewlines,
+              trailingSpaces: trailingSpaces,
+            ));
+            fragmentStart = i;
+            trailingNewlines = 0;
+            trailingSpaces = 0;
+          }
+        }
+      }
+
+      breaks.add(LineBreakFragment(
+        fragmentStart,
+        fragmentEnd,
+        type,
+        trailingNewlines: trailingNewlines,
+        trailingSpaces: trailingSpaces,
+      ));
+      fragmentStart = fragmentEnd;
+    }
+
+    if (breaks.isEmpty || breaks.last.type == LineBreakType.mandatory) {
+      breaks.add(LineBreakFragment(text.length, text.length, LineBreakType.endOfText, trailingNewlines: 0, trailingSpaces: 0));
+    }
+
+    return breaks;
+  }
+
+  /// Gets break type from v8BreakIterator.
+  LineBreakType _getBreakType(DomV8BreakIterator iterator) {
+    final int fragmentEnd = iterator.current();
+
+    // I don't know why v8BreakIterator uses the type "none" to mean "soft break".
+    if (iterator.breakType() != 'none') {
+      return LineBreakType.mandatory;
+    }
+    if (fragmentEnd == text.length) {
+      return LineBreakType.endOfText;
+    }
+    return LineBreakType.opportunity;
+  }
+}
+
 class LineBreakFragment extends TextFragment {
   const LineBreakFragment(super.start, super.end, this.type, {
     required this.trailingNewlines,

lib/web_ui/test/text/line_breaker_test.dart

@@ -17,7 +17,16 @@ void main() {
 }
 
 void testMain() {
-  group('$LineBreakFragmenter', () {
+  groupForEachFragmenter(({required bool isV8}) {
+    List<Line> split(String text) {
+      final LineBreakFragmenter fragmenter =
+          isV8 ? V8LineBreakFragmenter(text) : FWLineBreakFragmenter(text);
+      return <Line>[
+        for (final LineBreakFragment fragment in fragmenter.fragment())
+          Line.fromLineBreakFragment(text, fragment)
+      ];
+    }
+
     test('empty string', () {
       expect(split(''), <Line>[
         Line('', endOfText),
@@ -316,13 +325,15 @@ void testMain() {
     });
 
     test('comprehensive test', () {
-      final List<TestCase> testCollection =
-          parseRawTestData(rawLineBreakTestData);
+      final List<TestCase> testCollection = parseRawTestData(rawLineBreakTestData, isV8: isV8);
       for (int t = 0; t < testCollection.length; t++) {
         final TestCase testCase = testCollection[t];
 
         final String text = testCase.toText();
-        final List<LineBreakFragment> fragments = LineBreakFragmenter(text).fragment();
+        final LineBreakFragmenter fragmenter = isV8
+            ? V8LineBreakFragmenter(text)
+            : FWLineBreakFragmenter(text);
+        final List<LineBreakFragment> fragments = fragmenter.fragment();
 
         // `f` is the index in the `fragments` list.
         int f = 0;
@@ -401,6 +412,23 @@ void testMain() {
   });
 }
 
+typedef CreateLineBreakFragmenter = LineBreakFragmenter Function(String text);
+typedef GroupBody = void Function({required bool isV8});
+
+void groupForEachFragmenter(GroupBody callback) {
+  group(
+    '$FWLineBreakFragmenter',
+    () => callback(isV8: false),
+  );
+
+  if (domWindow.Intl.v8BreakIterator != null) {
+    group(
+      '$V8LineBreakFragmenter',
+      () => callback(isV8: true),
+    );
+  }
+}
+
 /// Holds information about how a line was split from a string.
 class Line {
   Line(this.text, this.breakType, {this.nl = 0, this.sp = 0});
@@ -447,10 +475,3 @@ class Line {
     return '"$escapedText" ($breakType, nl: $nl, sp: $sp)';
   }
 }
-
-List<Line> split(String text) {
-  return <Line>[
-    for (final LineBreakFragment fragment in LineBreakFragmenter(text).fragment())
-      Line.fromLineBreakFragment(text, fragment)
-  ];
-}