base/i18n/break_iterator.h

   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 #ifndef BASE_I18N_BREAK_ITERATOR_H_
   6 #define BASE_I18N_BREAK_ITERATOR_H_
   7
   8 #include "base/basictypes.h"
   9 #include "base/i18n/base_i18n_export.h"
  10 #include "base/strings/string16.h"
  11 #include "base/strings/string_piece.h"
  12
  13 // The BreakIterator class iterates through the words, word breaks, and
  14 // line breaks in a UTF-16 string.
  15 //
  16 // It provides several modes, BREAK_WORD, BREAK_LINE, and BREAK_NEWLINE,
  17 // which modify how characters are aggregated into the returned string.
  18 //
  19 // Under BREAK_WORD mode, once a word is encountered any non-word
  20 // characters are not included in the returned string (e.g. in the
  21 // UTF-16 equivalent of the string " foo bar! ", the word breaks are at
  22 // the periods in ". .foo. .bar.!. .").
  23 // Note that Chinese/Japanese/Thai do not use spaces between words so that
  24 // boundaries can fall in the middle of a continuous run of non-space /
  25 // non-punctuation characters.
  26 //
  27 // Under BREAK_LINE mode, once a line breaking opportunity is encountered,
  28 // any non-word  characters are included in the returned string, breaking
  29 // only when a space-equivalent character or a line breaking opportunity
  30 // is encountered (e.g. in the UTF16-equivalent of the string " foo bar! ",
  31 // the breaks are at the periods in ". .foo .bar! .").
  32 //
  33 // Note that lines can be broken at any character/syllable/grapheme cluster
  34 // boundary in Chinese/Japanese/Korean and at word boundaries in Thai
  35 // (Thai does not use spaces between words). Therefore, this is NOT the same
  36 // as breaking only at space-equivalent characters where its former
  37 // name (BREAK_SPACE) implied.
  38 //
  39 // Under BREAK_NEWLINE mode, all characters are included in the returned
  40 // string, breaking only when a newline-equivalent character is encountered
  41 // (eg. in the UTF-16 equivalent of the string "foo\nbar!\n\n", the line
  42 // breaks are at the periods in ".foo\n.bar\n.\n.").
  43 //
  44 // To extract the words from a string, move a BREAK_WORD BreakIterator
  45 // through the string and test whether IsWord() is true. E.g.,
  46 //   BreakIterator iter(str, BreakIterator::BREAK_WORD);
  47 //   if (!iter.Init())
  48 //     return false;
  49 //   while (iter.Advance()) {
  50 //     if (iter.IsWord()) {
  51 //       // Region [iter.prev(), iter.pos()) contains a word.
  52 //       VLOG(1) << "word: " << iter.GetString();
  53 //     }
  54 //   }
  55
  56 namespace base {
  57 namespace i18n {
  58
  59 class BASE_I18N_EXPORT BreakIterator {
  60  public:
  61   enum BreakType {
  62     BREAK_WORD,
  63     BREAK_LINE,
  64     // TODO(jshin): Remove this after reviewing call sites.
  65     // If call sites really need break only on space-like characters
  66     // implement it separately.
  67     BREAK_SPACE = BREAK_LINE,
  68     BREAK_NEWLINE,
  69     BREAK_CHARACTER,
  70     // But don't remove this one!
  71     RULE_BASED,
  72   };
  73
  74   enum WordBreakStatus {
  75     // The end of text that the iterator recognizes as word characters.
  76     // Non-word characters are things like punctuation and spaces.
  77     IS_WORD_BREAK,
  78     // Characters that the iterator can skip past, such as punctuation,
  79     // whitespace, and, if using RULE_BASED mode, characters from another
  80     // character set.
  81     IS_SKIPPABLE_WORD,
  82     // Only used if not in BREAK_WORD or RULE_BASED mode. This is returned for
  83     // newlines, line breaks, and character breaks.
  84     IS_LINE_OR_CHAR_BREAK
  85   };
  86
  87   // Requires |str| to live as long as the BreakIterator does.
  88   BreakIterator(const StringPiece16& str, BreakType break_type);
  89   // Make a rule-based iterator. BreakType == RULE_BASED is implied.
  90   // TODO(andrewhayden): This signature could easily be misinterpreted as
  91   // "(const string16& str, const string16& locale)". We should do something
  92   // better.
  93   BreakIterator(const StringPiece16& str, const string16& rules);
  94   ~BreakIterator();
  95
  96   // Init() must be called before any of the iterators are valid.
  97   // Returns false if ICU failed to initialize.
  98   bool Init();
  99
 100   // Advance to the next break.  Returns false if we've run past the end of
 101   // the string.  (Note that the very last "break" is after the final
 102   // character in the string, and when we advance to that position it's the
 103   // last time Advance() returns true.)
 104   bool Advance();
 105
 106   // Updates the text used by the iterator, resetting the iterator as if
 107   // if Init() had been called again. Any old state is lost. Returns true
 108   // unless there is an error setting the text.
 109   bool SetText(const base::char16* text, const size_t length);
 110
 111   // Under BREAK_WORD mode, returns true if the break we just hit is the
 112   // end of a word. (Otherwise, the break iterator just skipped over e.g.
 113   // whitespace or punctuation.)  Under BREAK_LINE and BREAK_NEWLINE modes,
 114   // this distinction doesn't apply and it always returns false.
 115   bool IsWord() const;
 116
 117   // Under BREAK_WORD mode:
 118   //  - Returns IS_SKIPPABLE_WORD if non-word characters, such as punctuation or
 119   //    spaces, are found.
 120   //  - Returns IS_WORD_BREAK if the break we just hit is the end of a sequence
 121   //    of word characters.
 122   // Under RULE_BASED mode:
 123   //  - Returns IS_SKIPPABLE_WORD if characters outside the rules' character set
 124   //    or non-word characters, such as punctuation or spaces, are found.
 125   //  - Returns IS_WORD_BREAK if the break we just hit is the end of a sequence
 126   //    of word characters that are in the rules' character set.
 127   // Not under BREAK_WORD or RULE_BASED mode:
 128   //  - Returns IS_LINE_OR_CHAR_BREAK.
 129   BreakIterator::WordBreakStatus GetWordBreakStatus() const;
 130
 131   // Under BREAK_WORD mode, returns true if |position| is at the end of word or
 132   // at the start of word. It always returns false under BREAK_LINE and
 133   // BREAK_NEWLINE modes.
 134   bool IsEndOfWord(size_t position) const;
 135   bool IsStartOfWord(size_t position) const;
 136
 137   // Under BREAK_CHARACTER mode, returns whether |position| is a Unicode
 138   // grapheme boundary.
 139   bool IsGraphemeBoundary(size_t position) const;
 140
 141   // Returns the string between prev() and pos().
 142   // Advance() must have been called successfully at least once for pos() to
 143   // have advanced to somewhere useful.
 144   string16 GetString() const;
 145
 146   StringPiece16 GetStringPiece() const;
 147
 148   // Returns the value of pos() returned before Advance() was last called.
 149   size_t prev() const { return prev_; }
 150
 151   // Returns the current break position within the string,
 152   // or BreakIterator::npos when done.
 153   size_t pos() const { return pos_; }
 154
 155  private:
 156   // ICU iterator, avoiding ICU ubrk.h dependence.
 157   // This is actually an ICU UBreakiterator* type, which turns out to be
 158   // a typedef for a void* in the ICU headers. Using void* directly prevents
 159   // callers from needing access to the ICU public headers directory.
 160   void* iter_;
 161
 162   // The string we're iterating over. Can be changed with SetText(...)
 163   StringPiece16 string_;
 164
 165   // Rules for our iterator. Mutually exclusive with break_type_.
 166   const string16 rules_;
 167
 168   // The breaking style (word/space/newline). Mutually exclusive with rules_
 169   BreakType break_type_;
 170
 171   // Previous and current iterator positions.
 172   size_t prev_, pos_;
 173
 174   DISALLOW_COPY_AND_ASSIGN(BreakIterator);
 175 };
 176
 177 }  // namespace i18n
 178 }  // namespace base
 179
 180 #endif  // BASE_I18N_BREAK_ITERATOR_H_