ui/app_list/search/term_break_iterator.cc

   1 // Copyright 2013 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 "ui/app_list/search/term_break_iterator.h"
   6
   7 #include "base/i18n/char_iterator.h"
   8 #include "base/logging.h"
   9 #include "base/strings/string_util.h"
  10 #include "third_party/icu/source/common/unicode/uchar.h"
  11
  12 namespace app_list {
  13
  14 TermBreakIterator::TermBreakIterator(const base::string16& word)
  15     : word_(word),
  16       prev_(npos),
  17       pos_(0),
  18       iter_(new base::i18n::UTF16CharIterator(&word)),
  19       state_(STATE_START) {
  20 }
  21
  22 TermBreakIterator::~TermBreakIterator() {}
  23
  24 bool TermBreakIterator::Advance() {
  25   // 2D matrix that defines term boundaries. Each row represents current state.
  26   // Each col represents new state from input char. Cells with true value
  27   // represents a term boundary.
  28   const bool kBoundary[][STATE_LAST] = {
  29     // START  NUMBER UPPER  LOWER  CHAR
  30     {  false, false, false, false, false },  // START
  31     {  false, false, true,  true,  true },   // NUMBER
  32     {  false, true,  false, false, true },   // UPPER
  33     {  false, true,  true,  false, true },   // LOWER
  34     {  false, true,  true,  true,  false },  // CHAR
  35   };
  36
  37   while (iter_->Advance()) {
  38     const State new_state = GetNewState(word_[iter_->array_pos()]);
  39     const bool is_boundary = kBoundary[state_][new_state];
  40     state_ = new_state;
  41     if (is_boundary)
  42       break;
  43   }
  44
  45   prev_ = pos_;
  46   pos_ = iter_->array_pos();
  47
  48   return prev_ != pos_ || !iter_->end();
  49 }
  50
  51 const base::string16 TermBreakIterator::GetCurrentTerm() const {
  52   DCHECK(prev_ != npos && pos_ != npos);
  53   return word_.substr(prev_, pos_ - prev_);
  54 }
  55
  56 TermBreakIterator::State TermBreakIterator::GetNewState(base::char16 ch) {
  57   if (IsAsciiDigit(ch) || ch == '.' || ch == ',')
  58     return STATE_NUMBER;
  59
  60   const bool is_upper = !!u_isUUppercase(ch);
  61   const bool is_lower = !!u_isULowercase(ch);
  62
  63   if (is_upper && is_lower) {
  64     NOTREACHED() << "Invalid state for ch=" << ch;
  65     return STATE_CHAR;
  66   }
  67
  68   if (is_upper)
  69     return STATE_UPPER;
  70   if (is_lower)
  71     return STATE_LOWER;
  72
  73   return STATE_CHAR;
  74 }
  75
  76 }  // namespace app_list