OLD | NEW |
(Empty) | |
| 1 // Copyright (c) 2010 The Chromium Authors. All rights reserved. |
| 2 // Use of this source code is governed by a BSD-style license that can be |
| 3 // found in the LICENSE file. |
| 4 |
| 5 #include "base/i18n/break_iterator.h" |
| 6 |
| 7 #include "base/string_piece.h" |
| 8 #include "base/string_util.h" |
| 9 #include "base/utf_string_conversions.h" |
| 10 #include "testing/gtest/include/gtest/gtest.h" |
| 11 |
| 12 TEST(BreakIteratorTest, BreakWord) { |
| 13 string16 space(UTF8ToUTF16(" ")); |
| 14 |
| 15 string16 str(UTF8ToUTF16(" foo bar! \npouet boom")); |
| 16 base::BreakIterator iter(&str, base::BreakIterator::BREAK_WORD); |
| 17 ASSERT_TRUE(iter.Init()); |
| 18 EXPECT_TRUE(iter.Advance()); |
| 19 EXPECT_FALSE(iter.IsWord()); |
| 20 EXPECT_EQ(space, iter.GetString()); |
| 21 EXPECT_TRUE(iter.Advance()); |
| 22 EXPECT_TRUE(iter.IsWord()); |
| 23 EXPECT_EQ(UTF8ToUTF16("foo"), iter.GetString()); |
| 24 EXPECT_TRUE(iter.Advance()); |
| 25 EXPECT_FALSE(iter.IsWord()); |
| 26 EXPECT_EQ(space, iter.GetString()); |
| 27 EXPECT_TRUE(iter.Advance()); |
| 28 EXPECT_TRUE(iter.IsWord()); |
| 29 EXPECT_EQ(UTF8ToUTF16("bar"), iter.GetString()); |
| 30 EXPECT_TRUE(iter.Advance()); |
| 31 EXPECT_FALSE(iter.IsWord()); |
| 32 EXPECT_EQ(UTF8ToUTF16("!"), iter.GetString()); |
| 33 EXPECT_TRUE(iter.Advance()); |
| 34 EXPECT_FALSE(iter.IsWord()); |
| 35 EXPECT_EQ(space, iter.GetString()); |
| 36 EXPECT_TRUE(iter.Advance()); |
| 37 EXPECT_FALSE(iter.IsWord()); |
| 38 EXPECT_EQ(UTF8ToUTF16("\n"), iter.GetString()); |
| 39 EXPECT_TRUE(iter.Advance()); |
| 40 EXPECT_TRUE(iter.IsWord()); |
| 41 EXPECT_EQ(UTF8ToUTF16("pouet"), iter.GetString()); |
| 42 EXPECT_TRUE(iter.Advance()); |
| 43 EXPECT_FALSE(iter.IsWord()); |
| 44 EXPECT_EQ(space, iter.GetString()); |
| 45 EXPECT_TRUE(iter.Advance()); |
| 46 EXPECT_TRUE(iter.IsWord()); |
| 47 EXPECT_EQ(UTF8ToUTF16("boom"), iter.GetString()); |
| 48 EXPECT_FALSE(iter.Advance()); |
| 49 EXPECT_FALSE(iter.IsWord()); |
| 50 } |
| 51 |
| 52 TEST(BreakIteratorTest, BreakSpace) { |
| 53 string16 str(UTF8ToUTF16(" foo bar! \npouet boom")); |
| 54 base::BreakIterator iter(&str, base::BreakIterator::BREAK_SPACE); |
| 55 ASSERT_TRUE(iter.Init()); |
| 56 EXPECT_TRUE(iter.Advance()); |
| 57 EXPECT_FALSE(iter.IsWord()); |
| 58 EXPECT_EQ(UTF8ToUTF16(" "), iter.GetString()); |
| 59 EXPECT_TRUE(iter.Advance()); |
| 60 EXPECT_FALSE(iter.IsWord()); |
| 61 EXPECT_EQ(UTF8ToUTF16("foo "), iter.GetString()); |
| 62 EXPECT_TRUE(iter.Advance()); |
| 63 EXPECT_FALSE(iter.IsWord()); |
| 64 EXPECT_EQ(UTF8ToUTF16("bar! \n"), iter.GetString()); |
| 65 EXPECT_TRUE(iter.Advance()); |
| 66 EXPECT_FALSE(iter.IsWord()); |
| 67 EXPECT_EQ(UTF8ToUTF16("pouet "), iter.GetString()); |
| 68 EXPECT_TRUE(iter.Advance()); |
| 69 EXPECT_FALSE(iter.IsWord()); |
| 70 EXPECT_EQ(UTF8ToUTF16("boom"), iter.GetString()); |
| 71 EXPECT_FALSE(iter.Advance()); |
| 72 EXPECT_FALSE(iter.IsWord()); |
| 73 } |
| 74 |
| 75 TEST(BreakIteratorTest, BreakWide16) { |
| 76 // "Παγκόσμιος Ιστός" |
| 77 const string16 str(WideToUTF16( |
| 78 L"\x03a0\x03b1\x03b3\x03ba\x03cc\x03c3\x03bc\x03b9" |
| 79 L"\x03bf\x03c2\x0020\x0399\x03c3\x03c4\x03cc\x03c2")); |
| 80 const string16 word1(str.substr(0, 10)); |
| 81 const string16 word2(str.substr(11, 5)); |
| 82 base::BreakIterator iter(&str, base::BreakIterator::BREAK_WORD); |
| 83 ASSERT_TRUE(iter.Init()); |
| 84 EXPECT_TRUE(iter.Advance()); |
| 85 EXPECT_TRUE(iter.IsWord()); |
| 86 EXPECT_EQ(word1, iter.GetString()); |
| 87 EXPECT_TRUE(iter.Advance()); |
| 88 EXPECT_FALSE(iter.IsWord()); |
| 89 EXPECT_EQ(UTF8ToUTF16(" "), iter.GetString()); |
| 90 EXPECT_TRUE(iter.Advance()); |
| 91 EXPECT_TRUE(iter.IsWord()); |
| 92 EXPECT_EQ(word2, iter.GetString()); |
| 93 EXPECT_FALSE(iter.Advance()); |
| 94 EXPECT_FALSE(iter.IsWord()); |
| 95 } |
| 96 |
| 97 TEST(BreakIteratorTest, BreakWide32) { |
| 98 // U+1D49C MATHEMATICAL SCRIPT CAPITAL A |
| 99 const char* very_wide_char = "\xF0\x9D\x92\x9C"; |
| 100 const string16 str( |
| 101 UTF8ToUTF16(StringPrintf("%s a", very_wide_char))); |
| 102 const string16 very_wide_word(str.substr(0, 2)); |
| 103 |
| 104 base::BreakIterator iter(&str, base::BreakIterator::BREAK_WORD); |
| 105 ASSERT_TRUE(iter.Init()); |
| 106 EXPECT_TRUE(iter.Advance()); |
| 107 EXPECT_TRUE(iter.IsWord()); |
| 108 EXPECT_EQ(very_wide_word, iter.GetString()); |
| 109 EXPECT_TRUE(iter.Advance()); |
| 110 EXPECT_FALSE(iter.IsWord()); |
| 111 EXPECT_EQ(UTF8ToUTF16(" "), iter.GetString()); |
| 112 EXPECT_TRUE(iter.Advance()); |
| 113 EXPECT_TRUE(iter.IsWord()); |
| 114 EXPECT_EQ(UTF8ToUTF16("a"), iter.GetString()); |
| 115 EXPECT_FALSE(iter.Advance()); |
| 116 EXPECT_FALSE(iter.IsWord()); |
| 117 } |
OLD | NEW |