base/i18n/char_iterator_unittest.cc

   1 // Copyright (c) 2011 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/char_iterator.h"
   6
   7 #include "base/strings/utf_string_conversions.h"
   8 #include "testing/gtest/include/gtest/gtest.h"
   9
  10 namespace base {
  11 namespace i18n {
  12
  13 TEST(CharIteratorsTest, TestUTF8) {
  14   std::string empty;
  15   UTF8CharIterator empty_iter(&empty);
  16   ASSERT_TRUE(empty_iter.end());
  17   ASSERT_EQ(0, empty_iter.array_pos());
  18   ASSERT_EQ(0, empty_iter.char_pos());
  19   ASSERT_FALSE(empty_iter.Advance());
  20
  21   std::string str("s\303\273r");  // [u with circumflex]
  22   UTF8CharIterator iter(&str);
  23   ASSERT_FALSE(iter.end());
  24   ASSERT_EQ(0, iter.array_pos());
  25   ASSERT_EQ(0, iter.char_pos());
  26   ASSERT_EQ('s', iter.get());
  27   ASSERT_TRUE(iter.Advance());
  28
  29   ASSERT_FALSE(iter.end());
  30   ASSERT_EQ(1, iter.array_pos());
  31   ASSERT_EQ(1, iter.char_pos());
  32   ASSERT_EQ(251, iter.get());
  33   ASSERT_TRUE(iter.Advance());
  34
  35   ASSERT_FALSE(iter.end());
  36   ASSERT_EQ(3, iter.array_pos());
  37   ASSERT_EQ(2, iter.char_pos());
  38   ASSERT_EQ('r', iter.get());
  39   ASSERT_TRUE(iter.Advance());
  40
  41   ASSERT_TRUE(iter.end());
  42   ASSERT_EQ(4, iter.array_pos());
  43   ASSERT_EQ(3, iter.char_pos());
  44
  45   // Don't care what it returns, but this shouldn't crash
  46   iter.get();
  47
  48   ASSERT_FALSE(iter.Advance());
  49 }
  50
  51 TEST(CharIteratorsTest, TestUTF16) {
  52   string16 empty = UTF8ToUTF16("");
  53   UTF16CharIterator empty_iter(&empty);
  54   ASSERT_TRUE(empty_iter.end());
  55   ASSERT_EQ(0, empty_iter.array_pos());
  56   ASSERT_EQ(0, empty_iter.char_pos());
  57   ASSERT_FALSE(empty_iter.Advance());
  58
  59   // This test string contains 4 characters:
  60   //   x
  61   //   u with circumflex - 2 bytes in UTF8, 1 codeword in UTF16
  62   //   math double-struck A - 4 bytes in UTF8, 2 codewords in UTF16
  63   //   z
  64   string16 str = UTF8ToUTF16("x\303\273\360\235\224\270z");
  65   UTF16CharIterator iter(&str);
  66   ASSERT_FALSE(iter.end());
  67   ASSERT_EQ(0, iter.array_pos());
  68   ASSERT_EQ(0, iter.char_pos());
  69   ASSERT_EQ('x', iter.get());
  70   ASSERT_TRUE(iter.Advance());
  71
  72   ASSERT_FALSE(iter.end());
  73   ASSERT_EQ(1, iter.array_pos());
  74   ASSERT_EQ(1, iter.char_pos());
  75   ASSERT_EQ(251, iter.get());
  76   ASSERT_TRUE(iter.Advance());
  77
  78   ASSERT_FALSE(iter.end());
  79   ASSERT_EQ(2, iter.array_pos());
  80   ASSERT_EQ(2, iter.char_pos());
  81   ASSERT_EQ(120120, iter.get());
  82   ASSERT_TRUE(iter.Advance());
  83
  84   ASSERT_FALSE(iter.end());
  85   ASSERT_EQ(4, iter.array_pos());
  86   ASSERT_EQ(3, iter.char_pos());
  87   ASSERT_EQ('z', iter.get());
  88   ASSERT_TRUE(iter.Advance());
  89
  90   ASSERT_TRUE(iter.end());
  91   ASSERT_EQ(5, iter.array_pos());
  92   ASSERT_EQ(4, iter.char_pos());
  93
  94   // Don't care what it returns, but this shouldn't crash
  95   iter.get();
  96
  97   ASSERT_FALSE(iter.Advance());
  98 }
  99
 100 }  // namespace i18n
 101 }  // namespace base