base/i18n/bidi_line_iterator_unittest.cc

   1 // Copyright 2017 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/bidi_line_iterator.h"
   6
   7 #include "base/macros.h"
   8 #include "base/strings/utf_string_conversions.h"
   9 #include "testing/gtest/include/gtest/gtest.h"
  10
  11 namespace base {
  12 namespace i18n {
  13 namespace {
  14
  15 class BiDiLineIteratorTest : public testing::TestWithParam<TextDirection> {
  16  public:
  17   BiDiLineIteratorTest() = default;
  18
  19   BiDiLineIterator* iterator() { return &iterator_; }
  20
  21  private:
  22   BiDiLineIterator iterator_;
  23
  24   DISALLOW_COPY_AND_ASSIGN(BiDiLineIteratorTest);
  25 };
  26
  27 TEST_P(BiDiLineIteratorTest, OnlyLTR) {
  28   iterator()->Open(UTF8ToUTF16("abc 😁 测试"), GetParam(),
  29                    BiDiLineIterator::CustomBehavior::NONE);
  30   ASSERT_EQ(1, iterator()->CountRuns());
  31
  32   int start, length;
  33   EXPECT_EQ(UBIDI_LTR, iterator()->GetVisualRun(0, &start, &length));
  34   EXPECT_EQ(0, start);
  35   EXPECT_EQ(9, length);
  36
  37   int end;
  38   UBiDiLevel level;
  39   iterator()->GetLogicalRun(0, &end, &level);
  40   EXPECT_EQ(9, end);
  41   if (GetParam() == TextDirection::RIGHT_TO_LEFT)
  42     EXPECT_EQ(2, level);
  43   else
  44     EXPECT_EQ(0, level);
  45 }
  46
  47 TEST_P(BiDiLineIteratorTest, OnlyRTL) {
  48   iterator()->Open(UTF8ToUTF16("מה השעה"), GetParam(),
  49                    BiDiLineIterator::CustomBehavior::NONE);
  50   ASSERT_EQ(1, iterator()->CountRuns());
  51
  52   int start, length;
  53   EXPECT_EQ(UBIDI_RTL, iterator()->GetVisualRun(0, &start, &length));
  54   EXPECT_EQ(0, start);
  55   EXPECT_EQ(7, length);
  56
  57   int end;
  58   UBiDiLevel level;
  59   iterator()->GetLogicalRun(0, &end, &level);
  60   EXPECT_EQ(7, end);
  61   EXPECT_EQ(1, level);
  62 }
  63
  64 TEST_P(BiDiLineIteratorTest, Mixed) {
  65   iterator()->Open(UTF8ToUTF16("אני משתמש ב- Chrome כדפדפן האינטרנט שלי"),
  66                    GetParam(), BiDiLineIterator::CustomBehavior::NONE);
  67   ASSERT_EQ(3, iterator()->CountRuns());
  68
  69   // We'll get completely different results depending on the top-level paragraph
  70   // direction.
  71   if (GetParam() == TextDirection::RIGHT_TO_LEFT) {
  72     // If para direction is RTL, expect the LTR substring "Chrome" to be nested
  73     // within the surrounding RTL text.
  74     int start, length;
  75     EXPECT_EQ(UBIDI_RTL, iterator()->GetVisualRun(0, &start, &length));
  76     EXPECT_EQ(19, start);
  77     EXPECT_EQ(20, length);
  78     EXPECT_EQ(UBIDI_LTR, iterator()->GetVisualRun(1, &start, &length));
  79     EXPECT_EQ(13, start);
  80     EXPECT_EQ(6, length);
  81     EXPECT_EQ(UBIDI_RTL, iterator()->GetVisualRun(2, &start, &length));
  82     EXPECT_EQ(0, start);
  83     EXPECT_EQ(13, length);
  84
  85     int end;
  86     UBiDiLevel level;
  87     iterator()->GetLogicalRun(0, &end, &level);
  88     EXPECT_EQ(13, end);
  89     EXPECT_EQ(1, level);
  90     iterator()->GetLogicalRun(13, &end, &level);
  91     EXPECT_EQ(19, end);
  92     EXPECT_EQ(2, level);
  93     iterator()->GetLogicalRun(19, &end, &level);
  94     EXPECT_EQ(39, end);
  95     EXPECT_EQ(1, level);
  96   } else {
  97     // If the para direction is LTR, expect the LTR substring "- Chrome " to be
  98     // at the top level, with two nested RTL runs on either side.
  99     int start, length;
 100     EXPECT_EQ(UBIDI_RTL, iterator()->GetVisualRun(0, &start, &length));
 101     EXPECT_EQ(0, start);
 102     EXPECT_EQ(11, length);
 103     EXPECT_EQ(UBIDI_LTR, iterator()->GetVisualRun(1, &start, &length));
 104     EXPECT_EQ(11, start);
 105     EXPECT_EQ(9, length);
 106     EXPECT_EQ(UBIDI_RTL, iterator()->GetVisualRun(2, &start, &length));
 107     EXPECT_EQ(20, start);
 108     EXPECT_EQ(19, length);
 109
 110     int end;
 111     UBiDiLevel level;
 112     iterator()->GetLogicalRun(0, &end, &level);
 113     EXPECT_EQ(11, end);
 114     EXPECT_EQ(1, level);
 115     iterator()->GetLogicalRun(11, &end, &level);
 116     EXPECT_EQ(20, end);
 117     EXPECT_EQ(0, level);
 118     iterator()->GetLogicalRun(20, &end, &level);
 119     EXPECT_EQ(39, end);
 120     EXPECT_EQ(1, level);
 121   }
 122 }
 123
 124 TEST_P(BiDiLineIteratorTest, RTLPunctuationNoCustomBehavior) {
 125   // This string features Hebrew characters interleaved with ASCII punctuation.
 126   iterator()->Open(UTF8ToUTF16("א!ב\"ג#ד$ה%ו&ז'ח(ט)י*ך+כ,ל-ם.מ/"
 127                                "ן:נ;ס<ע=ף>פ?ץ@צ[ק\\ר]ש^ת_א`ב{ג|ד}ה~ו"),
 128                    GetParam(), BiDiLineIterator::CustomBehavior::NONE);
 129
 130   // Expect a single RTL run.
 131   ASSERT_EQ(1, iterator()->CountRuns());
 132
 133   int start, length;
 134   EXPECT_EQ(UBIDI_RTL, iterator()->GetVisualRun(0, &start, &length));
 135   EXPECT_EQ(0, start);
 136   EXPECT_EQ(65, length);
 137
 138   int end;
 139   UBiDiLevel level;
 140   iterator()->GetLogicalRun(0, &end, &level);
 141   EXPECT_EQ(65, end);
 142   EXPECT_EQ(1, level);
 143 }
 144
 145 TEST_P(BiDiLineIteratorTest, RTLPunctuationAsURL) {
 146   // This string features Hebrew characters interleaved with ASCII punctuation.
 147   iterator()->Open(UTF8ToUTF16("א!ב\"ג#ד$ה%ו&ז'ח(ט)י*ך+כ,ל-ם.מ/"
 148                                "ן:נ;ס<ע=ף>פ?ץ@צ[ק\\ר]ש^ת_א`ב{ג|ד}ה~ו"),
 149                    GetParam(), BiDiLineIterator::CustomBehavior::AS_URL);
 150
 151   const int kStringSize = 65;
 152
 153   // Expect a primary RTL run, broken up by each of the 8 punctuation marks that
 154   // are considered strong LTR (17 runs total).
 155   struct {
 156     int start;
 157     UBiDiDirection dir;
 158   } expected_runs[] = {
 159       {0, UBIDI_RTL},  {5, UBIDI_LTR},   // '#'
 160       {6, UBIDI_RTL},  {11, UBIDI_LTR},  // '&'
 161       {12, UBIDI_RTL}, {27, UBIDI_LTR},  // '.'
 162       {28, UBIDI_RTL}, {29, UBIDI_LTR},  // '/'
 163       {30, UBIDI_RTL}, {31, UBIDI_LTR},  // ':'
 164       {32, UBIDI_RTL}, {37, UBIDI_LTR},  // '='
 165       {38, UBIDI_RTL}, {41, UBIDI_LTR},  // '?'
 166       {42, UBIDI_RTL}, {43, UBIDI_LTR},  // '@'
 167       {44, UBIDI_RTL},
 168   };
 169
 170   ASSERT_EQ(arraysize(expected_runs),
 171             static_cast<size_t>(iterator()->CountRuns()));
 172
 173   for (size_t i = 0; i < arraysize(expected_runs); ++i) {
 174     const auto& expected_run = expected_runs[i];
 175     int expected_run_end = i >= arraysize(expected_runs) - 1
 176                                ? kStringSize
 177                                : expected_runs[i + 1].start;
 178
 179     size_t visual_index = GetParam() == TextDirection::RIGHT_TO_LEFT
 180                               ? arraysize(expected_runs) - 1 - i
 181                               : i;
 182     int start, length;
 183     EXPECT_EQ(expected_run.dir,
 184               iterator()->GetVisualRun(visual_index, &start, &length))
 185         << "(i = " << i << ")";
 186     EXPECT_EQ(expected_run.start, start) << "(i = " << i << ")";
 187     EXPECT_EQ(expected_run_end - expected_run.start, length)
 188         << "(i = " << i << ")";
 189
 190     int expected_level =
 191         expected_run.dir == UBIDI_RTL
 192             ? 1
 193             : (GetParam() == TextDirection::RIGHT_TO_LEFT ? 2 : 0);
 194     int end;
 195     UBiDiLevel level;
 196     iterator()->GetLogicalRun(expected_run.start, &end, &level);
 197     EXPECT_EQ(expected_run_end, end) << "(i = " << i << ")";
 198     EXPECT_EQ(expected_level, level) << "(i = " << i << ")";
 199   }
 200 }
 201
 202 INSTANTIATE_TEST_CASE_P(,
 203                         BiDiLineIteratorTest,
 204                         ::testing::Values(TextDirection::LEFT_TO_RIGHT,
 205                                           TextDirection::RIGHT_TO_LEFT));
 206
 207 }  // namespace
 208 }  // namespace i18n
 209 }  // namespace base