base/i18n/break_iterator.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/break_iterator.h"
   6
   7 #include "base/logging.h"
   8 #include "third_party/icu/source/common/unicode/ubrk.h"
   9 #include "third_party/icu/source/common/unicode/uchar.h"
  10 #include "third_party/icu/source/common/unicode/ustring.h"
  11
  12 namespace base {
  13 namespace i18n {
  14
  15 const size_t npos = -1;
  16
  17 BreakIterator::BreakIterator(const string16& str, BreakType break_type)
  18     : iter_(NULL),
  19       string_(str),
  20       break_type_(break_type),
  21       prev_(npos),
  22       pos_(0) {
  23 }
  24
  25 BreakIterator::~BreakIterator() {
  26   if (iter_)
  27     ubrk_close(static_cast<UBreakIterator*>(iter_));
  28 }
  29
  30 bool BreakIterator::Init() {
  31   UErrorCode status = U_ZERO_ERROR;
  32   UBreakIteratorType break_type;
  33   switch (break_type_) {
  34     case BREAK_CHARACTER:
  35       break_type = UBRK_CHARACTER;
  36       break;
  37     case BREAK_WORD:
  38       break_type = UBRK_WORD;
  39       break;
  40     case BREAK_LINE:
  41     case BREAK_NEWLINE:
  42       break_type = UBRK_LINE;
  43       break;
  44     default:
  45       NOTREACHED() << "invalid break_type_";
  46       return false;
  47   }
  48   iter_ = ubrk_open(break_type, NULL,
  49                     string_.data(), static_cast<int32_t>(string_.size()),
  50                     &status);
  51   if (U_FAILURE(status)) {
  52     NOTREACHED() << "ubrk_open failed";
  53     return false;
  54   }
  55   // Move the iterator to the beginning of the string.
  56   ubrk_first(static_cast<UBreakIterator*>(iter_));
  57   return true;
  58 }
  59
  60 bool BreakIterator::Advance() {
  61   int32_t pos;
  62   int32_t status;
  63   prev_ = pos_;
  64   switch (break_type_) {
  65     case BREAK_CHARACTER:
  66     case BREAK_WORD:
  67     case BREAK_LINE:
  68       pos = ubrk_next(static_cast<UBreakIterator*>(iter_));
  69       if (pos == UBRK_DONE) {
  70         pos_ = npos;
  71         return false;
  72       }
  73       pos_ = static_cast<size_t>(pos);
  74       return true;
  75     case BREAK_NEWLINE:
  76       do {
  77         pos = ubrk_next(static_cast<UBreakIterator*>(iter_));
  78         if (pos == UBRK_DONE)
  79           break;
  80         pos_ = static_cast<size_t>(pos);
  81         status = ubrk_getRuleStatus(static_cast<UBreakIterator*>(iter_));
  82       } while (status >= UBRK_LINE_SOFT && status < UBRK_LINE_SOFT_LIMIT);
  83       if (pos == UBRK_DONE && prev_ == pos_) {
  84         pos_ = npos;
  85         return false;
  86       }
  87       return true;
  88     default:
  89       NOTREACHED() << "invalid break_type_";
  90       return false;
  91   }
  92 }
  93
  94 bool BreakIterator::IsWord() const {
  95   int32_t status = ubrk_getRuleStatus(static_cast<UBreakIterator*>(iter_));
  96   return (break_type_ == BREAK_WORD && status != UBRK_WORD_NONE);
  97 }
  98
  99 bool BreakIterator::IsEndOfWord(size_t position) const {
 100   if (break_type_ != BREAK_WORD)
 101     return false;
 102
 103   UBreakIterator* iter = static_cast<UBreakIterator*>(iter_);
 104   UBool boundary = ubrk_isBoundary(iter, static_cast<int32_t>(position));
 105   int32_t status = ubrk_getRuleStatus(iter);
 106   return (!!boundary && status != UBRK_WORD_NONE);
 107 }
 108
 109 bool BreakIterator::IsStartOfWord(size_t position) const {
 110   if (break_type_ != BREAK_WORD)
 111     return false;
 112
 113   UBreakIterator* iter = static_cast<UBreakIterator*>(iter_);
 114   UBool boundary = ubrk_isBoundary(iter, static_cast<int32_t>(position));
 115   ubrk_next(iter);
 116   int32_t next_status = ubrk_getRuleStatus(iter);
 117   return (!!boundary && next_status != UBRK_WORD_NONE);
 118 }
 119
 120 string16 BreakIterator::GetString() const {
 121   DCHECK(prev_ != npos && pos_ != npos);
 122   return string_.substr(prev_, pos_ - prev_);
 123 }
 124
 125 }  // namespace i18n
 126 }  // namespace base