[Intl] Add more tests for Intl.Segmenter

Bug: v8:6891
Cq-Include-Trybots: luci.v8.try:v8_linux_noi18n_rel_ng
Change-Id: I0b82b194cb7089aeaa322ed4e45008db6890e7a1
Reviewed-on: https://chromium-review.googlesource.com/c/1266995Reviewed-by: Sathya Gunasekaran <gsathya@chromium.org>
Commit-Queue: Frank Tang <ftang@chromium.org>
Cr-Commit-Position: refs/heads/master@{#56704}

test/intl/intl.status

@@ -32,10 +32,33 @@
 
   # https://bugs.chromium.org/p/v8/issues/detail?id=6891
   'segmenter/segment': [FAIL],
+  'segmenter/segment-grapheme': [FAIL],
+  'segmenter/segment-grapheme-following': [FAIL],
+  'segmenter/segment-grapheme-iterable': [FAIL],
+  'segmenter/segment-grapheme-next': [FAIL],
+  'segmenter/segment-grapheme-preceding': [FAIL],
   'segmenter/segment-iterator': [FAIL],
+  'segmenter/segment-iterator-breakType': [FAIL],
   'segmenter/segment-iterator-following': [FAIL],
   'segmenter/segment-iterator-next': [FAIL],
+  'segmenter/segment-iterator-position': [FAIL],
   'segmenter/segment-iterator-preceding': [FAIL],
+  'segmenter/segment-line': [FAIL],
+  'segmenter/segment-line-following': [FAIL],
+  'segmenter/segment-line-following-modes': [FAIL],
+  'segmenter/segment-line-iterable': [FAIL],
+  'segmenter/segment-line-next': [FAIL],
+  'segmenter/segment-line-preceding': [FAIL],
+  'segmenter/segment-sentence': [FAIL],
+  'segmenter/segment-sentence-following': [FAIL],
+  'segmenter/segment-sentence-iterable': [FAIL],
+  'segmenter/segment-sentence-next': [FAIL],
+  'segmenter/segment-sentence-preceding': [FAIL],
+  'segmenter/segment-word': [FAIL],
+  'segmenter/segment-word-following': [FAIL],
+  'segmenter/segment-word-iterable': [FAIL],
+  'segmenter/segment-word-next': [FAIL],
+  'segmenter/segment-word-preceding': [FAIL],
 }],  # ALWAYS
 
 ['variant == no_wasm_traps', {

test/intl/segmenter/segment-grapheme-following.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const seg = new Intl.Segmenter([], {granularity: "grapheme"})
+for (const text of [
+    "Hello world!", // English
+    " Hello world! ",  // English with space before/after
+    " Hello world? Foo bar!", // English
+    "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech
+    "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?",  // Vietnamese
+    "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek
+    "Решение Индии о покупке российских С-400 расценили как вызов США",  // Russian
+    "הרופא שהציל נשים והנערה ששועבדה ע",  // Hebrew,
+    "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic
+    "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर",  //  Hindi
+    "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil
+    "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu
+    "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese
+    "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว",  // Thai
+    "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp",  // Japanese
+    "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean
+    ]) {
+  const iter = seg.segment(text);
+  let prev = 0;
+  let segments = [];
+  while (!iter.following()) {
+    assertEquals(undefined, iter.breakType);
+    assertTrue(iter.position >= 0);
+    assertTrue(iter.position <= text.length);
+    assertTrue(iter.position > prev);
+    segments.push(text.substring(prev, iter.position));
+    prev = iter.position;
+  }
+  assertEquals(text, segments.join(""));
+}

test/intl/segmenter/segment-grapheme-iterable.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const seg = new Intl.Segmenter([], {granularity: "grapheme"})
+for (const text of [
+    "Hello world!", // English
+    " Hello world! ",  // English with space before/after
+    " Hello world? Foo bar!", // English
+    "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech
+    "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?",  // Vietnamese
+    "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek
+    "Решение Индии о покупке российских С-400 расценили как вызов США",  // Russian
+    "הרופא שהציל נשים והנערה ששועבדה ע",  // Hebrew,
+    "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic
+    "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर",  //  Hindi
+    "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil
+    "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu
+    "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese
+    "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว",  // Thai
+    "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp",  // Japanese
+    "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean
+    ]) {
+  let segments = [];
+  // Create another %SegmentIterator% to compare with result from the one that
+  // created in the for of loop.
+  let iter = seg.segment(text);
+  let prev = 0;
+  for (const v of seg.segment(text)) {
+    assertEquals(undefined, v.breakType);
+    assertEquals("string", typeof v.segment);
+    assertTrue(v.segment.length > 0);
+    segments.push(v.segment);
+
+    // manually advance the iter.
+    assertFalse(iter.following());
+    assertEquals(iter.breakType, v.breakType);
+    assertEquals(text.substring(prev, iter.position), v.segment);
+    prev = iter.position;
+  }
+  assertTrue(iter.following());
+  assertEquals(text, segments.join(''));
+}

test/intl/segmenter/segment-grapheme-next.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const seg = new Intl.Segmenter([], {granularity: "grapheme"})
+for (const text of [
+    "Hello world!", // English
+    " Hello world! ",  // English with space before/after
+    " Hello world? Foo bar!", // English
+    "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech
+    "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?",  // Vietnamese
+    "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek
+    "Решение Индии о покупке российских С-400 расценили как вызов США",  // Russian
+    "הרופא שהציל נשים והנערה ששועבדה ע",  // Hebrew,
+    "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic
+    "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर",  //  Hindi
+    "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil
+    "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu
+    "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese
+    "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว",  // Thai
+    "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp",  // Japanese
+    "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean
+    ]) {
+  const iter = seg.segment(text);
+  let segments = [];
+  for (let result = iter.next(); !result.done; result = iter.next()) {
+    const v = result.value;
+    assertEquals(undefined, v.breakType);
+    assertEquals("string", typeof v.segment);
+    assertTrue(v.segment.length > 0);
+    segments.push(v.segment);
+  }
+  assertEquals(text, segments.join(''));
+}

test/intl/segmenter/segment-grapheme-preceding.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const seg = new Intl.Segmenter([], {granularity: "grapheme"})
+for (const text of [
+    "Hello world!", // English
+    " Hello world! ",  // English with space before/after
+    " Hello world? Foo bar!", // English
+    "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech
+    "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?",  // Vietnamese
+    "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek
+    "Решение Индии о покупке российских С-400 расценили как вызов США",  // Russian
+    "הרופא שהציל נשים והנערה ששועבדה ע",  // Hebrew,
+    "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic
+    "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर",  //  Hindi
+    "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil
+    "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu
+    "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese
+    "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว",  // Thai
+    "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp",  // Japanese
+    "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean
+    ]) {
+  const iter = seg.segment(text);
+  let prev = text.length;
+  let segments = [];
+  iter.preceding(prev)
+  assertEquals(undefined, iter.breakType)
+  assertTrue(iter.position >= 0);
+  assertTrue(iter.position < prev);
+  segments.push(text.substring(iter.position, prev));
+  prev = iter.position;
+  while (!iter.preceding()) {
+    assertEquals(undefined, iter.breakType);
+    assertTrue(iter.position >= 0);
+    assertTrue(iter.position <= text.length);
+    assertTrue(iter.position < prev);
+    segments.push(text.substring(iter.position, prev));
+    prev = iter.position;
+  }
+  assertEquals(text, segments.reverse().join(""));
+}

test/intl/segmenter/segment-grapheme.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const seg = new Intl.Segmenter([], {granularity: "grapheme"})
+for (const text of [
+    "Hello world!", // English
+    " Hello world! ",  // English with space before/after
+    " Hello world? Foo bar!", // English
+    "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech
+    "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?",  // Vietnamese
+    "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek
+    "Решение Индии о покупке российских С-400 расценили как вызов США",  // Russian
+    "הרופא שהציל נשים והנערה ששועבדה ע",  // Hebrew,
+    "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic
+    "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर",  //  Hindi
+    "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil
+    "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu
+    "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese
+    "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว",  // Thai
+    "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp",  // Japanese
+    "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean
+    ]) {
+  const iter = seg.segment(text);
+  assertEquals(undefined, iter.breakType);
+  assertEquals(0, iter.position);
+}

test/intl/segmenter/segment-iterator-breakType.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const segmenter = new Intl.Segmenter();
+const text = "Hello World, Test 123! Foo Bar. How are you?";
+const iter = segmenter.segment(text);
+
+assertEquals(undefined, iter.breakType);

test/intl/segmenter/segment-iterator-position.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const segmenter = new Intl.Segmenter();
+const text = "Hello World, Test 123! Foo Bar. How are you?";
+const iter = segmenter.segment(text);
+
+assertEquals(0, iter.position);

test/intl/segmenter/segment-iterator.js

@@ -4,10 +4,17 @@
 
 // Flags: --harmony-intl-segmenter
 
-const segmenter = new Intl.Segmenter();
+let descriptor;
 const text = "Hello World, Test 123! Foo Bar. How are you?";
-const iter = segmenter.segment(text);
+for (const granularity of ["grapheme", "word", "sentence", "line"]) {
+  const segmenter = new Intl.Segmenter("en", { granularity });
+  const iter = segmenter.segment(text);
 
-assertEquals("number", typeof iter.position);
-assertEquals(0, iter.position);
-assertEquals("strig", typeof iter.breakType);
+  assertEquals("number", typeof iter.position);
+  assertEquals(0, iter.position);
+  if (granularity === "grapheme") {
+    assertEquals(undefined, iter.breakType);
+  } else {
+    assertEquals("string", typeof iter.breakType);
+  }
+}

test/intl/segmenter/segment-line-following-modes.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+let breakCounts = {};
+for (const locale of ["en", "fr", "ja", "zh", "ko"]) {
+  for (const lineBreakStyle of ["strict", "normal", "loose"]) {
+    const seg = new Intl.Segmenter(
+        [locale], {granularity: "line", lineBreakStyle: lineBreakStyle});
+    let opportunity = 0;
+    for (const text of [
+      // We know the following data caused different line break results between
+      // different modes.
+      // https://www.w3.org/TR/css-text-3/#propdef-line-break
+      // Japanese small kana or the Katakana-Hiragana prolonged sound mark
+      "あぁーぃーあーいーぁーぃー",
+      // hyphens:
+      // ‐ U+2010, – U+2013, 〜 U+301C, ゠ U+30A0
+      "ABC‐DEF–GHI〜JKL゠MNO",
+      // iteration marks:
+      // 々 U+3005, 〻 U+303B, ゝ U+309D, ゞ U+309E, ヽ U+30FD, ヾ U+30FE
+      "あ々あ〻あゝあゞあヽあヾあ",
+      // centered punctuation marks:
+      // ・ U+30FB, ： U+FF1A, ； U+FF1B, ･ U+FF65, ‼ U+203C
+      "ABC・DEF：GHI；JKL･MNO‼PQR",
+      // centered punctuation marks:
+      // ⁇ U+2047, ⁈ U+2048, ⁉ U+2049, ！ U+FF01, ？ U+FF1F
+      "ABC⁇DEF⁈GHI⁉JKL！MNO？PQR",
+      ]) {
+      const iter = seg.segment(text);
+      while (!iter.following()) {
+        opportunity++;
+      }
+    }
+    breakCounts[locale + "-" + lineBreakStyle] = opportunity;
+  }
+}
+// In Japanese
+// Just test the break count in loose mode is greater than normal mode.
+assertTrue(breakCounts["ja-loose"] > breakCounts["ja-normal"]);
+// and test the break count in normal mode is greater than strict mode.
+assertTrue(breakCounts["ja-normal"] > breakCounts["ja-strict"]);
+// In Chinese
+// Just test the break count in loose mode is greater than normal mode.
+assertTrue(breakCounts["zh-loose"] > breakCounts["zh-normal"]);
+// and test the break count in normal mode is greater than strict mode.
+assertTrue(breakCounts["zh-normal"] > breakCounts["zh-strict"]);
+// In English, French and Korean
+assertEquals(breakCounts["en-loose"], breakCounts["en-normal"]);
+assertEquals(breakCounts["fr-loose"], breakCounts["fr-normal"]);
+assertEquals(breakCounts["ko-loose"], breakCounts["ko-normal"]);
+// and test the break count in normal mode is greater than strict mode.
+assertTrue(breakCounts["en-normal"] > breakCounts["en-strict"]);
+assertTrue(breakCounts["fr-normal"] > breakCounts["fr-strict"]);
+assertTrue(breakCounts["ko-normal"] > breakCounts["ko-strict"]);

test/intl/segmenter/segment-line-following.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const seg = new Intl.Segmenter([], {granularity: "line"})
+for (const text of [
+    "Hello world!", // English
+    " Hello world! ",  // English with space before/after
+    " Hello world? Foo bar!", // English
+    "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech
+    "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?",  // Vietnamese
+    "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek
+    "Решение Индии о покупке российских С-400 расценили как вызов США",  // Russian
+    "הרופא שהציל נשים והנערה ששועבדה ע",  // Hebrew,
+    "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic
+    "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर",  //  Hindi
+    "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil
+    "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu
+    "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese
+    "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว",  // Thai
+    "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp",  // Japanese
+    "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean
+    ]) {
+  const iter = seg.segment(text);
+  let prev = 0;
+  let segments = [];
+  while (!iter.following()) {
+    assertTrue(["soft", "hard"].includes(iter.breakType), iter.breakType);
+    assertTrue(iter.position >= 0);
+    assertTrue(iter.position <= text.length);
+    assertTrue(iter.position > prev);
+    segments.push(text.substring(prev, iter.position));
+    prev = iter.position;
+  }
+  assertEquals(text, segments.join(""));
+}

test/intl/segmenter/segment-line-iterable.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const seg = new Intl.Segmenter([], {granularity: "line"})
+for (const text of [
+    "Hello world!", // English
+    " Hello world! ",  // English with space before/after
+    " Hello world? Foo bar!", // English
+    "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech
+    "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?",  // Vietnamese
+    "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek
+    "Решение Индии о покупке российских С-400 расценили как вызов США",  // Russian
+    "הרופא שהציל נשים והנערה ששועבדה ע",  // Hebrew,
+    "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic
+    "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर",  //  Hindi
+    "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil
+    "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu
+    "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese
+    "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว",  // Thai
+    "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp",  // Japanese
+    "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean
+    ]) {
+  let segments = [];
+  // Create another %SegmentIterator% to compare with result from the one that
+  // created in the for of loop.
+  let iter = seg.segment(text);
+  let prev = 0;
+  for (const v of seg.segment(text)) {
+    assertTrue(["soft", "hard"].includes(v.breakType), v.breakType);
+    assertEquals("string", typeof v.segment);
+    assertTrue(v.segment.length > 0);
+    segments.push(v.segment);
+
+    // manually advance the iter.
+    assertFalse(iter.following());
+    assertEquals(iter.breakType, v.breakType);
+    assertEquals(text.substring(prev, iter.position), v.segment);
+    prev = iter.position;
+  }
+  assertTrue(iter.following());
+  assertEquals(text, segments.join(''));
+}

test/intl/segmenter/segment-line-next.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const seg = new Intl.Segmenter([], {granularity: "line"})
+for (const text of [
+    "Hello world!", // English
+    " Hello world! ",  // English with space before/after
+    " Hello world? Foo bar!", // English
+    "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech
+    "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?",  // Vietnamese
+    "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek
+    "Решение Индии о покупке российских С-400 расценили как вызов США",  // Russian
+    "הרופא שהציל נשים והנערה ששועבדה ע",  // Hebrew,
+    "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic
+    "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर",  //  Hindi
+    "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil
+    "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu
+    "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese
+    "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว",  // Thai
+    "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp",  // Japanese
+    "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean
+    ]) {
+  const iter = seg.segment(text);
+  let segments = [];
+  for (let result = iter.next(); !result.done; result = iter.next()) {
+    const v = result.value;
+    assertTrue(["soft", "hard"].includes(iter.breakType), iter.breakType);
+    assertEquals("string", typeof v.segment);
+    assertTrue(v.segment.length > 0);
+    segments.push(v.segment);
+   }
+   assertEquals(text, segments.join(''));
+}

test/intl/segmenter/segment-line-preceding.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const seg = new Intl.Segmenter([], {granularity: "line"})
+for (const text of [
+    "Hello world!", // English
+    " Hello world! ",  // English with space before/after
+    " Hello world? Foo bar!", // English
+    "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech
+    "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?",  // Vietnamese
+    "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek
+    "Решение Индии о покупке российских С-400 расценили как вызов США",  // Russian
+    "הרופא שהציל נשים והנערה ששועבדה ע",  // Hebrew,
+    "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic
+    "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर",  //  Hindi
+    "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil
+    "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu
+    "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese
+    "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว",  // Thai
+    "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp",  // Japanese
+    "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean
+    ]) {
+  const iter = seg.segment(text);
+  let prev = text.length;
+  let segments = [];
+  iter.preceding(prev)
+  assertTrue(["soft", "hard"].includes(iter.breakType), iter.breakType);
+  assertTrue(iter.position >= 0);
+  assertTrue(iter.position < prev);
+  segments.push(text.substring(iter.position, prev));
+  prev = iter.position;
+  while (!iter.preceding()) {
+    assertTrue(["soft", "hard"].includes(iter.breakType), iter.breakType);
+    assertTrue(iter.position >= 0);
+    assertTrue(iter.position <= text.length);
+    assertTrue(iter.position < prev);
+    segments.push(text.substring(iter.position, prev));
+    prev = iter.position;
+  }
+  assertEquals(text, segments.reverse().join(""));
+}

test/intl/segmenter/segment-line.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const seg = new Intl.Segmenter([], {granularity: "line"})
+for (const text of [
+    "Hello world!", // English
+    " Hello world! ",  // English with space before/after
+    " Hello world? Foo bar!", // English
+    "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech
+    "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?",  // Vietnamese
+    "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek
+    "Решение Индии о покупке российских С-400 расценили как вызов США",  // Russian
+    "הרופא שהציל נשים והנערה ששועבדה ע",  // Hebrew,
+    "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic
+    "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर",  //  Hindi
+    "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil
+    "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu
+    "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese
+    "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว",  // Thai
+    "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp",  // Japanese
+    "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean
+    ]) {
+  const iter = seg.segment(text);
+  assertTrue(["soft", "hard"].includes(iter.breakType), iter.breakType);
+  assertEquals(0, iter.position);
+}

test/intl/segmenter/segment-sentence-following.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const seg = new Intl.Segmenter([], {granularity: "sentence"})
+for (const text of [
+    "Hello world!", // English
+    " Hello world! ",  // English with space before/after
+    " Hello world? Foo bar!", // English
+    "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech
+    "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?",  // Vietnamese
+    "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek
+    "Решение Индии о покупке российских С-400 расценили как вызов США",  // Russian
+    "הרופא שהציל נשים והנערה ששועבדה ע",  // Hebrew,
+    "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic
+    "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर",  //  Hindi
+    "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil
+    "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu
+    "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese
+    "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว",  // Thai
+    "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp",  // Japanese
+    "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean
+    ]) {
+  const iter = seg.segment(text);
+  let prev = 0;
+  let segments = [];
+  while (!iter.following()) {
+    assertTrue(["sep", "term"].includes(iter.breakType), iter.breakType);
+    assertTrue(iter.position >= 0);
+    assertTrue(iter.position <= text.length);
+    assertTrue(iter.position > prev);
+    segments.push(text.substring(prev, iter.position));
+    prev = iter.position;
+  }
+  assertEquals(text, segments.join(""));
+}

test/intl/segmenter/segment-sentence-iterable.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const seg = new Intl.Segmenter([], {granularity: "sentence"})
+for (const text of [
+    "Hello world!", // English
+    " Hello world! ",  // English with space before/after
+    " Hello world? Foo bar!", // English
+    "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech
+    "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?",  // Vietnamese
+    "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek
+    "Решение Индии о покупке российских С-400 расценили как вызов США",  // Russian
+    "הרופא שהציל נשים והנערה ששועבדה ע",  // Hebrew,
+    "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic
+    "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर",  //  Hindi
+    "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil
+    "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu
+    "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese
+    "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว",  // Thai
+    "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp",  // Japanese
+    "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean
+    ]) {
+  let segments = [];
+  // Create another %SegmentIterator% to compare with result from the one that
+  // created in the for of loop.
+  let iter = seg.segment(text);
+  let prev = 0;
+  for (const v of seg.segment(text)) {
+    assertTrue(["sep", "term"].includes(v.breakType), v.breakType);
+    assertEquals("string", typeof v.segment);
+    assertTrue(v.segment.length > 0);
+    segments.push(v.segment);
+
+    // manually advance the iter.
+    assertFalse(iter.following());
+    assertEquals(iter.breakType, v.breakType);
+    assertEquals(text.substring(prev, iter.position), v.segment);
+    prev = iter.position;
+  }
+  assertTrue(iter.following());
+  assertEquals(text, segments.join(''));
+}

test/intl/segmenter/segment-sentence-next.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const seg = new Intl.Segmenter([], {granularity: "sentence"})
+for (const text of [
+    "Hello world!", // English
+    " Hello world! ",  // English with space before/after
+    " Hello world? Foo bar!", // English
+    "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech
+    "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?",  // Vietnamese
+    "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek
+    "Решение Индии о покупке российских С-400 расценили как вызов США",  // Russian
+    "הרופא שהציל נשים והנערה ששועבדה ע",  // Hebrew,
+    "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic
+    "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर",  //  Hindi
+    "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil
+    "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu
+    "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese
+    "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว",  // Thai
+    "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp",  // Japanese
+    "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean
+    ]) {
+  const iter = seg.segment(text);
+  let segments = [];
+  for (let result = iter.next(); !result.done; result = iter.next()) {
+    const v = result.value;
+    assertTrue(["sep", "term"].includes(iter.breakType), iter.breakType);
+    assertEquals("string", typeof v.segment);
+    assertTrue(v.segment.length > 0);
+    segments.push(v.segment);
+  }
+  assertEquals(text, segments.join(''));
+}

test/intl/segmenter/segment-sentence-preceding.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const seg = new Intl.Segmenter([], {granularity: "sentence"})
+for (const text of [
+    "Hello world!", // English
+    " Hello world! ",  // English with space before/after
+    " Hello world? Foo bar!", // English
+    "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech
+    "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?",  // Vietnamese
+    "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek
+    "Решение Индии о покупке российских С-400 расценили как вызов США",  // Russian
+    "הרופא שהציל נשים והנערה ששועבדה ע",  // Hebrew,
+    "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic
+    "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर",  //  Hindi
+    "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil
+    "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu
+    "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese
+    "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว",  // Thai
+    "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp",  // Japanese
+    "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean
+    ]) {
+  const iter = seg.segment(text);
+  let prev = text.length;
+  let segments = [];
+  iter.preceding(prev);
+  assertTrue(["sep", "term"].includes(iter.breakType), iter.breakType);
+  assertTrue(iter.position >= 0);
+  assertTrue(iter.position < prev);
+  segments.push(text.substring(iter.position, prev));
+  prev = iter.position;
+  while (!iter.preceding()) {
+    assertTrue(["sep", "term"].includes(iter.breakType), iter.breakType);
+    assertTrue(iter.position >= 0);
+    assertTrue(iter.position <= text.length);
+    assertTrue(iter.position < prev);
+    segments.push(text.substring(iter.position, prev));
+    prev = iter.position;
+  }
+  assertEquals(text, segments.reverse().join(""));
+}

test/intl/segmenter/segment-sentence.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const seg = new Intl.Segmenter([], {granularity: "sentence"})
+for (const text of [
+    "Hello world!", // English
+    " Hello world! ",  // English with space before/after
+    " Hello world? Foo bar!", // English
+    "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech
+    "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?",  // Vietnamese
+    "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek
+    "Решение Индии о покупке российских С-400 расценили как вызов США",  // Russian
+    "הרופא שהציל נשים והנערה ששועבדה ע",  // Hebrew,
+    "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic
+    "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर",  //  Hindi
+    "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil
+    "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu
+    "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese
+    "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว",  // Thai
+    "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp",  // Japanese
+    "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean
+    ]) {
+  const iter = seg.segment(text);
+  assertTrue(["sep", "term"].includes(iter.breakType), iter.breakType);
+  assertEquals(0, iter.position);
+}

test/intl/segmenter/segment-word-following.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const seg = new Intl.Segmenter([], {granularity: "word"})
+for (const text of [
+    "Hello world!", // English
+    " Hello world! ",  // English with space before/after
+    " Hello world? Foo bar!", // English
+    "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech
+    "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?",  // Vietnamese
+    "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek
+    "Решение Индии о покупке российских С-400 расценили как вызов США",  // Russian
+    "הרופא שהציל נשים והנערה ששועבדה ע",  // Hebrew,
+    "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic
+    "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर",  //  Hindi
+    "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil
+    "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu
+    "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese
+    "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว",  // Thai
+    "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp",  // Japanese
+    "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean
+    ]) {
+  const iter = seg.segment(text);
+  let prev = 0;
+  let segments = [];
+  while (!iter.following()) {
+    assertTrue(["word", "none"].includes(iter.breakType), iter.breakType);
+    assertTrue(iter.position >= 0);
+    assertTrue(iter.position <= text.length);
+    assertTrue(iter.position > prev);
+    segments.push(text.substring(prev, iter.position));
+    prev = iter.position;
+  }
+  assertEquals(text, segments.join(""));
+}

test/intl/segmenter/segment-word-iterable.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const seg = new Intl.Segmenter([], {granularity: "word"})
+for (const text of [
+    "Hello world!", // English
+    " Hello world! ",  // English with space before/after
+    " Hello world? Foo bar!", // English
+    "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech
+    "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?",  // Vietnamese
+    "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek
+    "Решение Индии о покупке российских С-400 расценили как вызов США",  // Russian
+    "הרופא שהציל נשים והנערה ששועבדה ע",  // Hebrew,
+    "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic
+    "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर",  //  Hindi
+    "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil
+    "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu
+    "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese
+    "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว",  // Thai
+    "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp",  // Japanese
+    "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean
+    ]) {
+  let segments = [];
+  // Create another %SegmentIterator% to compare with result from the one that
+  // created in the for of loop.
+  let iter = seg.segment(text);
+  let prev = 0;
+  for (const v of seg.segment(text)) {
+    assertTrue(["word", "none"].includes(v.breakType), v.breakType);
+    assertEquals("string", typeof v.segment);
+    assertTrue(v.segment.length > 0);
+    segments.push(v.segment);
+
+    // manually advance the iter.
+    assertFalse(iter.following());
+    assertEquals(iter.breakType, v.breakType);
+    assertEquals(text.substring(prev, iter.position), v.segment);
+    prev = iter.position;
+  }
+  assertTrue(iter.following());
+  assertEquals(text, segments.join(''));
+}

test/intl/segmenter/segment-word-next.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const seg = new Intl.Segmenter([], {granularity: "word"})
+for (const text of [
+    "Hello world!", // English
+    " Hello world! ",  // English with space before/after
+    " Hello world? Foo bar!", // English
+    "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech
+    "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?",  // Vietnamese
+    "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek
+    "Решение Индии о покупке российских С-400 расценили как вызов США",  // Russian
+    "הרופא שהציל נשים והנערה ששועבדה ע",  // Hebrew,
+    "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic
+    "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर",  //  Hindi
+    "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil
+    "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu
+    "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese
+    "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว",  // Thai
+    "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp",  // Japanese
+    "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean
+    ]) {
+  const iter = seg.segment(text);
+  let segments = [];
+  for (let result = iter.next(); !result.done; result = iter.next()) {
+    const v = result.value;
+    assertTrue(["word", "none"].includes(iter.breakType), iter.breakType);
+    assertEquals("string", typeof v.segment);
+    assertTrue(v.segment.length > 0);
+    segments.push(v.segment);
+  }
+  assertEquals(text, segments.join(''));
+}

test/intl/segmenter/segment-word-preceding.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const seg = new Intl.Segmenter([], {granularity: "word"})
+for (const text of [
+    "Hello world!", // English
+    " Hello world! ",  // English with space before/after
+    " Hello world? Foo bar!", // English
+    "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech
+    "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?",  // Vietnamese
+    "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek
+    "Решение Индии о покупке российских С-400 расценили как вызов США",  // Russian
+    "הרופא שהציל נשים והנערה ששועבדה ע",  // Hebrew,
+    "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic
+    "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर",  //  Hindi
+    "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil
+    "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu
+    "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese
+    "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว",  // Thai
+    "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp",  // Japanese
+    "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean
+    ]) {
+  const iter = seg.segment(text);
+  let prev = text.length;
+  let segments = [];
+  iter.preceding(prev);
+  assertTrue(["word", "none"].includes(iter.breakType), iter.breakType);
+  assertTrue(iter.position >= 0);
+  assertTrue(iter.position < prev);
+  segments.push(text.substring(iter.position, prev));
+  prev = iter.position;
+  while (!iter.preceding()) {
+    assertTrue(["word", "none"].includes(iter.breakType), iter.breakType);
+    assertTrue(iter.position >= 0);
+    assertTrue(iter.position <= text.length);
+    assertTrue(iter.position < prev);
+    segments.push(text.substring(iter.position, prev));
+    prev = iter.position;
+  }
+  assertEquals(text, segments.reverse().join(""));
+}

test/intl/segmenter/segment-word.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+const seg = new Intl.Segmenter([], {granularity: "word"})
+for (const text of [
+    "Hello world!", // English
+    " Hello world! ",  // English with space before/after
+    " Hello world? Foo bar!", // English
+    "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech
+    "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?",  // Vietnamese
+    "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek
+    "Решение Индии о покупке российских С-400 расценили как вызов США",  // Russian
+    "הרופא שהציל נשים והנערה ששועבדה ע",  // Hebrew,
+    "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic
+    "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर",  //  Hindi
+    "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil
+    "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu
+    "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese
+    "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว",  // Thai
+    "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp",  // Japanese
+    "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean
+    ]) {
+  const iter = seg.segment(text);
+  assertTrue(["word", "none"].includes(iter.breakType), iter.breakType);
+  assertEquals(0, iter.position);
+}

test/intl/segmenter/subclassing.js

+// Copyright 2018 the V8 project authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+// Flags: --harmony-intl-segmenter
+
+// Test subclassing of Segmenter
+class CustomSegmenter extends Intl.Segmenter {
+  constructor(locales, options) {
+    super(locales, options);
+    this.isCustom = true;
+  }
+}
+
+const seg = new CustomSegmenter("zh");
+assertEquals(true, seg.isCustom, "Custom property");
+assertEquals(Object.getPrototypeOf(seg), CustomSegmenter.prototype, "Prototype");

test/intl/segmenter/supported-locale.js

@@ -9,14 +9,14 @@ assertEquals(
     "Intl.Segmenter.supportedLocalesOf should be a function"
 );
 
-var undef = Intl.Segmenter.supportedLocalesOf();
+const undef = Intl.Segmenter.supportedLocalesOf();
 assertEquals([], undef);
 
-var empty = Intl.Segmenter.supportedLocalesOf([]);
+const empty = Intl.Segmenter.supportedLocalesOf([]);
 assertEquals([], empty);
 
-var strLocale = Intl.Segmenter.supportedLocalesOf("sr");
+const strLocale = Intl.Segmenter.supportedLocalesOf("sr");
 assertEquals("sr", strLocale[0]);
 
-var multiLocale = ["sr-Thai-RS", "de", "zh-CN"];
+const multiLocale = ["sr-Thai-RS", "de", "zh-CN"];
 assertEquals(multiLocale, Intl.Segmenter.supportedLocalesOf(multiLocale));