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 <stddef.h> 9 10 #include "base/i18n/base_i18n_export.h" 11 #include "base/macros.h" 12 #include "base/strings/string16.h" 13 #include "base/strings/string_piece.h" 14 15 // The BreakIterator class iterates through the words, word breaks, and 16 // line breaks in a UTF-16 string. 17 // 18 // It provides several modes, BREAK_WORD, BREAK_LINE, and BREAK_NEWLINE, 19 // which modify how characters are aggregated into the returned string. 20 // 21 // Under BREAK_WORD mode, once a word is encountered any non-word 22 // characters are not included in the returned string (e.g. in the 23 // UTF-16 equivalent of the string " foo bar! ", the word breaks are at 24 // the periods in ". .foo. .bar.!. ."). 25 // Note that Chinese/Japanese/Thai do not use spaces between words so that 26 // boundaries can fall in the middle of a continuous run of non-space / 27 // non-punctuation characters. 28 // 29 // Under BREAK_LINE mode, once a line breaking opportunity is encountered, 30 // any non-word characters are included in the returned string, breaking 31 // only when a space-equivalent character or a line breaking opportunity 32 // is encountered (e.g. in the UTF16-equivalent of the string " foo bar! ", 33 // the breaks are at the periods in ". .foo .bar! ."). 34 // 35 // Note that lines can be broken at any character/syllable/grapheme cluster 36 // boundary in Chinese/Japanese/Korean and at word boundaries in Thai 37 // (Thai does not use spaces between words). Therefore, this is NOT the same 38 // as breaking only at space-equivalent characters where its former 39 // name (BREAK_SPACE) implied. 40 // 41 // Under BREAK_NEWLINE mode, all characters are included in the returned 42 // string, breaking only when a newline-equivalent character is encountered 43 // (eg. in the UTF-16 equivalent of the string "foo\nbar!\n\n", the line 44 // breaks are at the periods in ".foo\n.bar\n.\n."). 45 // 46 // To extract the words from a string, move a BREAK_WORD BreakIterator 47 // through the string and test whether IsWord() is true. E.g., 48 // BreakIterator iter(str, BreakIterator::BREAK_WORD); 49 // if (!iter.Init()) 50 // return false; 51 // while (iter.Advance()) { 52 // if (iter.IsWord()) { 53 // // Region [iter.prev(), iter.pos()) contains a word. 54 // VLOG(1) << "word: " << iter.GetString(); 55 // } 56 // } 57 58 namespace base { 59 namespace i18n { 60 61 class BASE_I18N_EXPORT BreakIterator { 62 public: 63 enum BreakType { 64 BREAK_WORD, 65 BREAK_LINE, 66 // TODO(jshin): Remove this after reviewing call sites. 67 // If call sites really need break only on space-like characters 68 // implement it separately. 69 BREAK_SPACE = BREAK_LINE, 70 BREAK_NEWLINE, 71 BREAK_CHARACTER, 72 // But don't remove this one! 73 RULE_BASED, 74 }; 75 76 enum WordBreakStatus { 77 // The end of text that the iterator recognizes as word characters. 78 // Non-word characters are things like punctuation and spaces. 79 IS_WORD_BREAK, 80 // Characters that the iterator can skip past, such as punctuation, 81 // whitespace, and, if using RULE_BASED mode, characters from another 82 // character set. 83 IS_SKIPPABLE_WORD, 84 // Only used if not in BREAK_WORD or RULE_BASED mode. This is returned for 85 // newlines, line breaks, and character breaks. 86 IS_LINE_OR_CHAR_BREAK 87 }; 88 89 // Requires |str| to live as long as the BreakIterator does. 90 BreakIterator(const StringPiece16& str, BreakType break_type); 91 // Make a rule-based iterator. BreakType == RULE_BASED is implied. 92 // TODO(andrewhayden): This signature could easily be misinterpreted as 93 // "(const string16& str, const string16& locale)". We should do something 94 // better. 95 BreakIterator(const StringPiece16& str, const string16& rules); 96 ~BreakIterator(); 97 98 // Init() must be called before any of the iterators are valid. 99 // Returns false if ICU failed to initialize. 100 bool Init(); 101 102 // Advance to the next break. Returns false if we've run past the end of 103 // the string. (Note that the very last "break" is after the final 104 // character in the string, and when we advance to that position it's the 105 // last time Advance() returns true.) 106 bool Advance(); 107 108 // Updates the text used by the iterator, resetting the iterator as if 109 // if Init() had been called again. Any old state is lost. Returns true 110 // unless there is an error setting the text. 111 bool SetText(const base::char16* text, const size_t length); 112 113 // Under BREAK_WORD mode, returns true if the break we just hit is the 114 // end of a word. (Otherwise, the break iterator just skipped over e.g. 115 // whitespace or punctuation.) Under BREAK_LINE and BREAK_NEWLINE modes, 116 // this distinction doesn't apply and it always returns false. 117 bool IsWord() const; 118 119 // Under BREAK_WORD mode: 120 // - Returns IS_SKIPPABLE_WORD if non-word characters, such as punctuation or 121 // spaces, are found. 122 // - Returns IS_WORD_BREAK if the break we just hit is the end of a sequence 123 // of word characters. 124 // Under RULE_BASED mode: 125 // - Returns IS_SKIPPABLE_WORD if characters outside the rules' character set 126 // or non-word characters, such as punctuation or spaces, are found. 127 // - Returns IS_WORD_BREAK if the break we just hit is the end of a sequence 128 // of word characters that are in the rules' character set. 129 // Not under BREAK_WORD or RULE_BASED mode: 130 // - Returns IS_LINE_OR_CHAR_BREAK. 131 BreakIterator::WordBreakStatus GetWordBreakStatus() const; 132 133 // Under BREAK_WORD mode, returns true if |position| is at the end of word or 134 // at the start of word. It always returns false under BREAK_LINE and 135 // BREAK_NEWLINE modes. 136 bool IsEndOfWord(size_t position) const; 137 bool IsStartOfWord(size_t position) const; 138 139 // Under BREAK_CHARACTER mode, returns whether |position| is a Unicode 140 // grapheme boundary. 141 bool IsGraphemeBoundary(size_t position) const; 142 143 // Returns the string between prev() and pos(). 144 // Advance() must have been called successfully at least once for pos() to 145 // have advanced to somewhere useful. 146 string16 GetString() const; 147 148 StringPiece16 GetStringPiece() const; 149 150 // Returns the value of pos() returned before Advance() was last called. prev()151 size_t prev() const { return prev_; } 152 153 // Returns the current break position within the string, 154 // or BreakIterator::npos when done. pos()155 size_t pos() const { return pos_; } 156 157 private: 158 // ICU iterator, avoiding ICU ubrk.h dependence. 159 // This is actually an ICU UBreakiterator* type, which turns out to be 160 // a typedef for a void* in the ICU headers. Using void* directly prevents 161 // callers from needing access to the ICU public headers directory. 162 void* iter_; 163 164 // The string we're iterating over. Can be changed with SetText(...) 165 StringPiece16 string_; 166 167 // Rules for our iterator. Mutually exclusive with break_type_. 168 const string16 rules_; 169 170 // The breaking style (word/space/newline). Mutually exclusive with rules_ 171 BreakType break_type_; 172 173 // Previous and current iterator positions. 174 size_t prev_, pos_; 175 176 DISALLOW_COPY_AND_ASSIGN(BreakIterator); 177 }; 178 179 } // namespace i18n 180 } // namespace base 181 182 #endif // BASE_I18N_BREAK_ITERATOR_H_ 183