1 // Copyright 2011 The Chromium Authors
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/check_op.h"
8 #include "base/third_party/icu/icu_utf.h"
9
10 namespace base {
11 namespace i18n {
12
13 // UTF8CharIterator ------------------------------------------------------------
14
UTF8CharIterator(base::StringPiece str)15 UTF8CharIterator::UTF8CharIterator(base::StringPiece str)
16 : str_(str), array_pos_(0), next_pos_(0), char_pos_(0), char_(0) {
17 if (!str_.empty())
18 CBU8_NEXT(str_.data(), next_pos_, str_.length(), char_);
19 }
20
21 UTF8CharIterator::~UTF8CharIterator() = default;
22
Advance()23 bool UTF8CharIterator::Advance() {
24 if (array_pos_ >= str_.length())
25 return false;
26
27 array_pos_ = next_pos_;
28 char_pos_++;
29 if (next_pos_ < str_.length())
30 CBU8_NEXT(str_.data(), next_pos_, str_.length(), char_);
31
32 return true;
33 }
34
35 // UTF16CharIterator -----------------------------------------------------------
36
UTF16CharIterator(StringPiece16 str)37 UTF16CharIterator::UTF16CharIterator(StringPiece16 str)
38 : UTF16CharIterator(str, 0) {}
39
40 UTF16CharIterator::UTF16CharIterator(UTF16CharIterator&& to_move) = default;
41
42 UTF16CharIterator::~UTF16CharIterator() = default;
43
44 UTF16CharIterator& UTF16CharIterator::operator=(UTF16CharIterator&& to_move) =
45 default;
46
47 // static
LowerBound(StringPiece16 str,size_t array_index)48 UTF16CharIterator UTF16CharIterator::LowerBound(StringPiece16 str,
49 size_t array_index) {
50 DCHECK_LE(array_index, str.length());
51 CBU16_SET_CP_START(str.data(), 0, array_index);
52 return UTF16CharIterator(str, array_index);
53 }
54
55 // static
UpperBound(StringPiece16 str,size_t array_index)56 UTF16CharIterator UTF16CharIterator::UpperBound(StringPiece16 str,
57 size_t array_index) {
58 DCHECK_LE(array_index, str.length());
59 CBU16_SET_CP_LIMIT(str.data(), 0, array_index, str.length());
60 return UTF16CharIterator(str, array_index);
61 }
62
NextCodePoint() const63 int32_t UTF16CharIterator::NextCodePoint() const {
64 if (next_pos_ >= str_.length())
65 return 0;
66
67 base_icu::UChar32 c;
68 CBU16_GET(str_.data(), 0, next_pos_, str_.length(), c);
69 return c;
70 }
71
PreviousCodePoint() const72 int32_t UTF16CharIterator::PreviousCodePoint() const {
73 if (array_pos_ == 0)
74 return 0;
75
76 uint32_t pos = array_pos_;
77 base_icu::UChar32 c;
78 CBU16_PREV(str_.data(), 0, pos, c);
79 return c;
80 }
81
Advance()82 bool UTF16CharIterator::Advance() {
83 if (array_pos_ >= str_.length())
84 return false;
85
86 array_pos_ = next_pos_;
87 char_offset_++;
88 if (next_pos_ < str_.length())
89 ReadChar();
90
91 return true;
92 }
93
Rewind()94 bool UTF16CharIterator::Rewind() {
95 if (array_pos_ == 0)
96 return false;
97
98 next_pos_ = array_pos_;
99 char_offset_--;
100 CBU16_PREV(str_.data(), 0, array_pos_, char_);
101 return true;
102 }
103
UTF16CharIterator(StringPiece16 str,size_t initial_pos)104 UTF16CharIterator::UTF16CharIterator(StringPiece16 str, size_t initial_pos)
105 : str_(str),
106 array_pos_(initial_pos),
107 next_pos_(initial_pos),
108 char_offset_(0),
109 char_(0) {
110 // This has the side-effect of advancing |next_pos_|.
111 if (array_pos_ < str_.length())
112 ReadChar();
113 }
114
ReadChar()115 void UTF16CharIterator::ReadChar() {
116 // This is actually a huge macro, so is worth having in a separate function.
117 CBU16_NEXT(str_.data(), next_pos_, str_.length(), char_);
118 }
119
120 } // namespace i18n
121 } // namespace base
122