base/i18n/word_iterator.h

   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 #ifndef BASE_I18N_WORD_ITERATOR_H_
   6 #define BASE_I18N_WORD_ITERATOR_H_
   7
   8 #include <string>
   9 #include <vector>
  10
  11 #include "unicode/uchar.h"
  12
  13 #include "base/basictypes.h"
  14
  15 // The WordIterator class iterates through the words and word breaks
  16 // in a string.  (In the string " foo bar! ", the word breaks are at the
  17 // periods in ". .foo. .bar.!. .".)
  18 //
  19 // To extract the words from a string, move a WordIterator through the
  20 // string and test whether IsWord() is true.  E.g.,
  21 //   WordIterator iter(str, WordIterator::BREAK_WORD);
  22 //   if (!iter.Init()) return false;
  23 //   while (iter.Advance()) {
  24 //     if (iter.IsWord()) {
  25 //       // region [iter.prev(),iter.pos()) contains a word.
  26 //       LOG(INFO) << "word: " << iter.GetWord();
  27 //     }
  28 //   }
  29
  30
  31 class WordIterator {
  32  public:
  33   enum BreakType {
  34     BREAK_WORD,
  35     BREAK_LINE
  36   };
  37
  38   // Requires |str| to live as long as the WordIterator does.
  39   WordIterator(const std::wstring& str, BreakType break_type);
  40   ~WordIterator();
  41
  42   // Init() must be called before any of the iterators are valid.
  43   // Returns false if ICU failed to initialize.
  44   bool Init();
  45
  46   // Return the current break position within the string,
  47   // or WordIterator::npos when done.
  48   size_t pos() const { return pos_; }
  49   // Return the value of pos() returned before Advance() was last called.
  50   size_t prev() const { return prev_; }
  51
  52   // Advance to the next break.  Returns false if we've run past the end of
  53   // the string.  (Note that the very last "word break" is after the final
  54   // character in the string, and when we advance to that position it's the
  55   // last time Advance() returns true.)
  56   bool Advance();
  57
  58   // Returns true if the break we just hit is the end of a word.
  59   // (Otherwise, the break iterator just skipped over e.g. whitespace
  60   // or punctuation.)
  61   bool IsWord() const;
  62
  63   // Return the word between prev() and pos().
  64   // Advance() must have been called successfully at least once
  65   // for pos() to have advanced to somewhere useful.
  66   std::wstring GetWord() const;
  67
  68  private:
  69   // ICU iterator.
  70   void* iter_;
  71 #if !defined(WCHAR_T_IS_UTF16)
  72   std::vector<UChar> chars_;
  73 #endif
  74
  75   // The string we're iterating over.
  76   const std::wstring& string_;
  77
  78   // The breaking style (word/line).
  79   BreakType break_type_;
  80
  81   // Previous and current iterator positions.
  82   size_t prev_, pos_;
  83
  84   DISALLOW_COPY_AND_ASSIGN(WordIterator);
  85 };
  86
  87 #endif  // BASE_I18N_WORD_ITERATOR_H__