base/utf_offset_string_conversions_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 <algorithm>
   6
   7 #include "base/logging.h"
   8 #include "base/string_piece.h"
   9 #include "base/utf_offset_string_conversions.h"
  10 #include "testing/gtest/include/gtest/gtest.h"
  11
  12 namespace base {
  13
  14 namespace {
  15
  16 static const size_t kNpos = string16::npos;
  17
  18 }  // namespace
  19
  20 TEST(UTFOffsetStringConversionsTest, AdjustOffset) {
  21   struct UTF8ToUTF16Case {
  22     const char* utf8;
  23     size_t input_offset;
  24     size_t output_offset;
  25   } utf8_to_utf16_cases[] = {
  26     {"", 0, kNpos},
  27     {"\xe4\xbd\xa0\xe5\xa5\xbd", 1, kNpos},
  28     {"\xe4\xbd\xa0\xe5\xa5\xbd", 3, 1},
  29     {"\xed\xb0\x80z", 3, 1},
  30     {"A\xF0\x90\x8C\x80z", 1, 1},
  31     {"A\xF0\x90\x8C\x80z", 2, kNpos},
  32     {"A\xF0\x90\x8C\x80z", 5, 3},
  33   };
  34   for (size_t i = 0; i < ARRAYSIZE_UNSAFE(utf8_to_utf16_cases); ++i) {
  35     size_t offset = utf8_to_utf16_cases[i].input_offset;
  36     UTF8ToUTF16AndAdjustOffset(utf8_to_utf16_cases[i].utf8, &offset);
  37     EXPECT_EQ(utf8_to_utf16_cases[i].output_offset, offset);
  38   }
  39
  40   struct UTF16ToUTF8Case {
  41     char16 utf16[10];
  42     size_t input_offset;
  43     size_t output_offset;
  44   } utf16_to_utf8_cases[] = {
  45       {{}, 0, kNpos},
  46       // Converted to 3-byte utf-8 sequences
  47       {{0x5909, 0x63DB}, 2, kNpos},
  48       {{0x5909, 0x63DB}, 1, 3},
  49       // Converted to 2-byte utf-8 sequences
  50       {{'A', 0x00bc, 0x00be, 'z'}, 1, 1},
  51       {{'A', 0x00bc, 0x00be, 'z'}, 2, 3},
  52       {{'A', 0x00bc, 0x00be, 'z'}, 3, 5},
  53       // Surrogate pair
  54       {{'A', 0xd800, 0xdf00, 'z'}, 1, 1},
  55       {{'A', 0xd800, 0xdf00, 'z'}, 2, kNpos},
  56       {{'A', 0xd800, 0xdf00, 'z'}, 3, 5},
  57   };
  58   for (size_t i = 0; i < ARRAYSIZE_UNSAFE(utf16_to_utf8_cases); ++i) {
  59     size_t offset = utf16_to_utf8_cases[i].input_offset;
  60     UTF16ToUTF8AndAdjustOffset(utf16_to_utf8_cases[i].utf16, &offset);
  61     EXPECT_EQ(utf16_to_utf8_cases[i].output_offset, offset);
  62   }
  63 }
  64
  65 TEST(UTFOffsetStringConversionsTest, LimitOffsets) {
  66   const size_t kLimit = 10;
  67   const size_t kItems = 20;
  68   std::vector<size_t> size_ts;
  69   for (size_t t = 0; t < kItems; ++t)
  70     size_ts.push_back(t);
  71   std::for_each(size_ts.begin(), size_ts.end(),
  72                 LimitOffset<string16>(kLimit));
  73   size_t unlimited_count = 0;
  74   for (std::vector<size_t>::iterator ti = size_ts.begin(); ti != size_ts.end();
  75        ++ti) {
  76     if (*ti < kLimit && *ti != kNpos)
  77       ++unlimited_count;
  78   }
  79   EXPECT_EQ(10U, unlimited_count);
  80
  81   // Reverse the values in the vector and try again.
  82   size_ts.clear();
  83   for (size_t t = kItems; t > 0; --t)
  84     size_ts.push_back(t - 1);
  85   std::for_each(size_ts.begin(), size_ts.end(),
  86                 LimitOffset<string16>(kLimit));
  87   unlimited_count = 0;
  88   for (std::vector<size_t>::iterator ti = size_ts.begin(); ti != size_ts.end();
  89        ++ti) {
  90     if (*ti < kLimit && *ti != kNpos)
  91       ++unlimited_count;
  92   }
  93   EXPECT_EQ(10U, unlimited_count);
  94 }
  95
  96 TEST(UTFOffsetStringConversionsTest, AdjustOffsets) {
  97   // Imagine we have strings as shown in the following cases where the
  98   // X's represent encoded characters.
  99   // 1: abcXXXdef ==> abcXdef
 100   {
 101     std::vector<size_t> offsets;
 102     for (size_t t = 0; t < 9; ++t)
 103       offsets.push_back(t);
 104     {
 105       OffsetAdjuster offset_adjuster(&offsets);
 106       offset_adjuster.Add(OffsetAdjuster::Adjustment(3, 3, 1));
 107     }
 108     size_t expected_1[] = {0, 1, 2, 3, kNpos, kNpos, 4, 5, 6};
 109     EXPECT_EQ(offsets.size(), arraysize(expected_1));
 110     for (size_t i = 0; i < arraysize(expected_1); ++i)
 111       EXPECT_EQ(expected_1[i], offsets[i]);
 112   }
 113
 114   // 2: XXXaXXXXbcXXXXXXXdefXXX ==> XaXXbcXXXXdefX
 115   {
 116     std::vector<size_t> offsets;
 117     for (size_t t = 0; t < 23; ++t)
 118       offsets.push_back(t);
 119     {
 120       OffsetAdjuster offset_adjuster(&offsets);
 121       offset_adjuster.Add(OffsetAdjuster::Adjustment(0, 3, 1));
 122       offset_adjuster.Add(OffsetAdjuster::Adjustment(4, 4, 2));
 123       offset_adjuster.Add(OffsetAdjuster::Adjustment(10, 7, 4));
 124       offset_adjuster.Add(OffsetAdjuster::Adjustment(20, 3, 1));
 125     }
 126     size_t expected_2[] = {0, kNpos, kNpos, 1, 2, kNpos, kNpos, kNpos, 4, 5, 6,
 127                            kNpos, kNpos, kNpos, kNpos, kNpos, kNpos, 10, 11, 12,
 128                            13, kNpos, kNpos};
 129     EXPECT_EQ(offsets.size(), arraysize(expected_2));
 130     for (size_t i = 0; i < arraysize(expected_2); ++i)
 131       EXPECT_EQ(expected_2[i], offsets[i]);
 132   }
 133
 134   // 3: XXXaXXXXbcdXXXeXX ==> aXXXXbcdXXXe
 135   {
 136     std::vector<size_t> offsets;
 137     for (size_t t = 0; t < 17; ++t)
 138       offsets.push_back(t);
 139     {
 140       OffsetAdjuster offset_adjuster(&offsets);
 141       offset_adjuster.Add(OffsetAdjuster::Adjustment(0, 3, 0));
 142       offset_adjuster.Add(OffsetAdjuster::Adjustment(4, 4, 4));
 143       offset_adjuster.Add(OffsetAdjuster::Adjustment(11, 3, 3));
 144       offset_adjuster.Add(OffsetAdjuster::Adjustment(15, 2, 0));
 145     }
 146     size_t expected_3[] = {kNpos, kNpos, kNpos, 0, 1, kNpos, kNpos, kNpos, 5, 6,
 147                            7, 8, kNpos, kNpos, 11, kNpos, kNpos};
 148     EXPECT_EQ(offsets.size(), arraysize(expected_3));
 149     for (size_t i = 0; i < arraysize(expected_3); ++i)
 150       EXPECT_EQ(expected_3[i], offsets[i]);
 151   }
 152 }
 153
 154 }  // namaspace base