1 // Copyright (c) 2009 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/word_iterator.h"
6
7 #include "base/logging.h"
8 #include "unicode/ubrk.h"
9 #include "unicode/ustring.h"
10
11 const size_t npos = -1;
12
WordIterator(const std::wstring & str,BreakType break_type)13 WordIterator::WordIterator(const std::wstring& str, BreakType break_type)
14 : iter_(NULL),
15 string_(str),
16 break_type_(break_type),
17 prev_(npos),
18 pos_(0) {
19 }
20
~WordIterator()21 WordIterator::~WordIterator() {
22 if (iter_)
23 ubrk_close(iter_);
24 }
25
Init()26 bool WordIterator::Init() {
27 UErrorCode status = U_ZERO_ERROR;
28 UBreakIteratorType break_type;
29 switch (break_type_) {
30 case BREAK_WORD:
31 break_type = UBRK_WORD;
32 break;
33 case BREAK_LINE:
34 break_type = UBRK_LINE;
35 break;
36 default:
37 NOTREACHED();
38 break_type = UBRK_LINE;
39 }
40 #if defined(WCHAR_T_IS_UTF16)
41 iter_ = ubrk_open(break_type, NULL,
42 string_.data(), static_cast<int32_t>(string_.size()),
43 &status);
44 #else // WCHAR_T_IS_UTF16
45 // When wchar_t is wider than UChar (16 bits), transform |string_| into a
46 // UChar* string. Size the UChar* buffer to be large enough to hold twice
47 // as many UTF-16 code points as there are UCS-4 characters, in case each
48 // character translates to a UTF-16 surrogate pair, and leave room for a NUL
49 // terminator.
50 // TODO(avi): avoid this alloc
51 chars_.resize(string_.length() * sizeof(UChar) + 1);
52
53 UErrorCode error = U_ZERO_ERROR;
54 int32_t destLength;
55 u_strFromWCS(&chars_[0], chars_.size(), &destLength, string_.data(),
56 string_.length(), &error);
57
58 iter_ = ubrk_open(break_type, NULL, &chars_[0], destLength, &status);
59 #endif
60 if (U_FAILURE(status)) {
61 NOTREACHED() << "ubrk_open failed";
62 return false;
63 }
64 ubrk_first(iter_); // Move the iterator to the beginning of the string.
65 return true;
66 }
67
Advance()68 bool WordIterator::Advance() {
69 prev_ = pos_;
70 const int32_t pos = ubrk_next(iter_);
71 if (pos == UBRK_DONE) {
72 pos_ = npos;
73 return false;
74 } else {
75 pos_ = static_cast<size_t>(pos);
76 return true;
77 }
78 }
79
IsWord() const80 bool WordIterator::IsWord() const {
81 return (ubrk_getRuleStatus(iter_) != UBRK_WORD_NONE);
82 }
83
GetWord() const84 std::wstring WordIterator::GetWord() const {
85 DCHECK(prev_ != npos && pos_ != npos);
86 #if defined(WCHAR_T_IS_UTF16)
87 return string_.substr(prev_, pos_ - prev_);
88 #else // WCHAR_T_IS_UTF16
89 // See comment in Init(). If there are no surrogate pairs,
90 // |out_length| will be exactly |in_length|, if there are surrogate
91 // pairs it will be less than |in_length|.
92 int32_t out_length;
93 UErrorCode error = U_ZERO_ERROR;
94 const int32_t in_length = pos_ - prev_;
95 std::vector<std::wstring::value_type> out_buffer(in_length);
96 u_strToWCS(&out_buffer[0], in_length, &out_length,
97 &chars_[prev_], in_length, &error);
98 DCHECK_LE(out_length, in_length);
99 return std::wstring(&out_buffer[0], out_length);
100 #endif
101 }
102