Creates BreakIterator::GetWordBreakStatus.

chrome/renderer/spellchecker/spellcheck_worditerator_unittest.cc

 // Copyright (c) 2012 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include <string>
 #include <vector>
 
 #include "base/format_macros.h"
+#include "base/i18n/break_iterator.h"
 #include "base/strings/string_split.h"
 #include "base/strings/stringprintf.h"
 #include "base/strings/utf_string_conversions.h"
 #include "chrome/renderer/spellchecker/spellcheck_worditerator.h"
 #include "testing/gtest/include/gtest/gtest.h"
 
+using base::i18n::BreakIterator;
+
 namespace {
 
 struct TestCase {
     const char* language;
     bool allow_contraction;
     const wchar_t* expected_words;
 };
 
+base::string16 GetRulesForLanguage(const std::string& language) {
+  SpellcheckCharAttribute attribute;
+  attribute.SetDefaultLanguage(language);
+  return attribute.GetRuleSet(true);
+}
+
 }  // namespace
 
 // Tests whether or not our SpellcheckWordIterator can extract only words used
 // by the specified language from a multi-language text.
 TEST(SpellcheckWordIteratorTest, SplitWord) {
   // An input text. This text includes words of several languages. (Some words
   // are not separated with whitespace characters.) Our SpellcheckWordIterator
   // should extract only the words used by the specified language from this text
   // and normalize them so our spell-checker can check their spellings.
   const wchar_t kTestText[] =
@@ skipping 255 matching lines @@
   }
 
   // Test initialization fails when no default language is set.
   {
     SpellcheckCharAttribute attributes;
 
     SpellcheckWordIterator iterator;
     EXPECT_FALSE(iterator.Initialize(&attributes, true));
   }
 }
+
+// This test uses English rules to check that different character set
+// combinations properly find word breaks and skippable characters.
+TEST(SpellcheckWordIteratorTest, FindSkippableWordsEnglish) {
+  // The string "foo ទេ Can Ми..." which contains English, Khmer, and Russian
+  // characters, in that order.
+  base::string16 text(
+      base::WideToUTF16(L"foo \x1791\x17c1 Can \x041C\x0438..."));
+  BreakIterator iter(text, GetRulesForLanguage("en-US"));
+  ASSERT_TRUE(iter.Init());
+
+  EXPECT_TRUE(iter.Advance());
+  // Finds "foo".
+  EXPECT_EQ(base::UTF8ToUTF16("foo"), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_WORD_BREAK);
+  EXPECT_TRUE(iter.Advance());
+  // Finds the space and then the Khmer characters.
+  EXPECT_EQ(base::UTF8ToUTF16(" "), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_EQ(base::WideToUTF16(L"\x1791\x17c1"), iter.GetString());

[jungshik at Google, 2015/08/11 21:43:50]

Interesting. Even if Khmer is not treated as either ALetter or ALetterPlus,
w-b-iterator still does not break them apart (perhaps because it's a single
grapheme... rules does not have that info?). 

[jungshik at Google, 2015/08/11 22:22:07]

I figured out why Khmer is not split up here. That's because Khmer uses a
dictionary for word (as well as line) breaking. And, it's handled outside
non-dictionary cases. Our custom rules do not change the following line:

# For dictionary-based break
$dictionary $dictionary;

[Julius, 2015/08/12 01:22:21]

Acknowledged.

[Julius, 2015/08/12 01:22:21]

Acknowledged.

+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  // Finds the next space and "Can".
+  EXPECT_EQ(base::UTF8ToUTF16(" "), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_EQ(base::UTF8ToUTF16("Can"), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_WORD_BREAK);
+  EXPECT_TRUE(iter.Advance());
+  // Finds the next space and each Russian character.
+  EXPECT_EQ(base::UTF8ToUTF16(" "), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_EQ(base::WideToUTF16(L"\x041C"), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_EQ(base::WideToUTF16(L"\x0438"), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  // Finds the periods at the end.
+  EXPECT_EQ(base::UTF8ToUTF16("."), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_EQ(base::UTF8ToUTF16("."), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_EQ(base::UTF8ToUTF16("."), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_FALSE(iter.Advance());
+}
+
+// This test uses Russian rules to check that different character set
+// combinations properly find word breaks and skippable characters.
+TEST(SpellcheckWordIteratorTest, FindSkippableWordsRussian) {
+  // The string ".;Ми Can ទេ  " which contains Russian, English, and Khmer
+  // characters, in that order.
+  base::string16 text(base::WideToUTF16(L".;\x041C\x0438 Can \x1791\x17c1  "));
+  BreakIterator iter(text, GetRulesForLanguage("ru-RU"));
+  ASSERT_TRUE(iter.Init());
+
+  EXPECT_TRUE(iter.Advance());
+  // Finds the period and semicolon.
+  EXPECT_EQ(base::UTF8ToUTF16("."), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_EQ(base::UTF8ToUTF16(";"), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  // Finds all the Russian characters.
+  EXPECT_EQ(base::WideToUTF16(L"\x041C\x0438"), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_WORD_BREAK);
+  EXPECT_TRUE(iter.Advance());
+  // Finds the space and each character in "Can".
+  EXPECT_EQ(base::UTF8ToUTF16(" "), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_EQ(base::UTF8ToUTF16("C"), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_EQ(base::UTF8ToUTF16("a"), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_EQ(base::UTF8ToUTF16("n"), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  // Finds the next space, the Khmer characters, and the last two spaces.
+  EXPECT_EQ(base::UTF8ToUTF16(" "), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_EQ(base::WideToUTF16(L"\x1791\x17c1"), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_EQ(base::UTF8ToUTF16(" "), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_EQ(base::UTF8ToUTF16(" "), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_FALSE(iter.Advance());
+}
+
+// This test uses Khmer rules to check that different character set combinations
+// properly find word breaks and skippable characters.
+TEST(SpellcheckWordIteratorTest, FindSkippableWordsKhmer) {
+  // The string "Ми ទេzoo. ," which contains Russian, Khmer, and English
+  // characters, in that order.

[jungshik at Google, 2015/08/11 21:43:50]

A Khmer example can be made more interesting. If you take an example from ICU's
Khmer break iterator tests. 

I took the following example from the first line in the Khmer section of
third_party/icu/source/test/testdata/rbbitst.txt


U+178F U+17BE <word break> U+179B U+17C4 U+1780 <word break> U+1798 U+1780

[Julius, 2015/08/12 01:22:21]

Swapped the Khmer text in this case with your suggested texted.

+  base::string16 text(base::WideToUTF16(L"\x041C\x0438 \x1791\x17c1zoo. ,"));
+  BreakIterator iter(text, GetRulesForLanguage("km"));
+  ASSERT_TRUE(iter.Init());
+
+  EXPECT_TRUE(iter.Advance());
+  // Finds each Russian character and the space.
+  EXPECT_EQ(base::WideToUTF16(L"\x041C"), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_EQ(base::WideToUTF16(L"\x0438"), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_EQ(base::UTF8ToUTF16(" "), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  // Finds all the Khmer characters.
+  EXPECT_EQ(base::WideToUTF16(L"\x1791\x17c1"), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_WORD_BREAK);
+  EXPECT_TRUE(iter.Advance());
+  // Finds each character in "zoo".
+  EXPECT_EQ(base::UTF8ToUTF16("z"), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_EQ(base::UTF8ToUTF16("o"), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_EQ(base::UTF8ToUTF16("o"), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  // Finds the period, space, and comma.
+  EXPECT_EQ(base::UTF8ToUTF16("."), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_EQ(base::UTF8ToUTF16(" "), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_TRUE(iter.Advance());
+  EXPECT_EQ(base::UTF8ToUTF16(","), iter.GetString());
+  EXPECT_EQ(iter.GetWordBreakStatus(), BreakIterator::IS_SKIPPABLE_WORD);
+  EXPECT_FALSE(iter.Advance());
+}