chrome/renderer/spellchecker/spellcheck_language.h

   1 // Copyright (c) 2012 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 CHROME_RENDERER_SPELLCHECKER_SPELLCHECK_LANGUAGE_H_
   6 #define CHROME_RENDERER_SPELLCHECKER_SPELLCHECK_LANGUAGE_H_
   7
   8 #include <queue>
   9 #include <string>
  10 #include <vector>
  11
  12 #include "base/files/file.h"
  13 #include "base/memory/scoped_ptr.h"
  14 #include "base/strings/string16.h"
  15 #include "chrome/renderer/spellchecker/spellcheck_worditerator.h"
  16
  17 class SpellingEngine;
  18
  19 class SpellcheckLanguage {
  20  public:
  21   enum SpellcheckWordResult {
  22     // Denotes that every recognized word is spelled correctly, from the point
  23     // of spellchecking to the end of the text.
  24     IS_CORRECT,
  25     // A sequence of skippable characters, such as punctuation, spaces, or
  26     // characters not recognized by the current spellchecking language.
  27     IS_SKIPPABLE,
  28     // A misspelled word has been found in the text.
  29     IS_MISSPELLED
  30   };
  31
  32   SpellcheckLanguage();
  33   ~SpellcheckLanguage();
  34
  35   void Init(base::File file, const std::string& language);
  36
  37   // Spellcheck a sequence of text.
  38   // |text_length| is the length of the text being spellchecked. The |tag|
  39   // parameter should either be a unique identifier for the document that the
  40   // word came from (if the current platform requires it), or 0.
  41   //
  42   // - Returns IS_CORRECT if every word from |position_in_text| to the end of
  43   //   the text is recognized and spelled correctly. Also, returns IS_CORRECT if
  44   //   the spellchecker failed to initialize.
  45   //
  46   // - Returns IS_SKIPPABLE if a sequence of skippable characters, such as
  47   //   punctuation, spaces, or unrecognized characters, is found.
  48   //   |skip_or_misspelling_start| and |skip_or_misspelling_len| are then set to
  49   //   the position and the length of the sequence of skippable characters.
  50   //
  51   // - Returns IS_MISSPELLED if a misspelled word is found.
  52   //   |skip_or_misspelling_start| and |skip_or_misspelling_len| are then set to
  53   //   the position and length of the misspelled word. In addition, finds the
  54   //   suggested words for a given misspelled word and puts them into
  55   //   |*optional_suggestions|. If optional_suggestions is NULL, suggested words
  56   //   will not be looked up. Note that doing suggest lookups can be slow.
  57   SpellcheckWordResult SpellCheckWord(
  58       const base::char16* text_begin,
  59       int position_in_text,
  60       int text_length,
  61       int tag,
  62       int* skip_or_misspelling_start,
  63       int* skip_or_misspelling_len,
  64       std::vector<base::string16>* optional_suggestions);
  65
  66   // Initialize |spellcheck_| if that hasn't happened yet.
  67   bool InitializeIfNeeded();
  68
  69   // Return true if the underlying spellcheck engine is enabled.
  70   bool IsEnabled();
  71
  72  private:
  73   friend class SpellCheckTest;
  74
  75   // Returns whether or not the given word is a contraction of valid words
  76   // (e.g. "word:word").
  77   bool IsValidContraction(const base::string16& word, int tag);
  78
  79   // Represents character attributes used for filtering out characters which
  80   // are not supported by this SpellCheck object.
  81   SpellcheckCharAttribute character_attributes_;
  82
  83   // Represents word iterators used in this spellchecker. The |text_iterator_|
  84   // splits text provided by WebKit into words, contractions, or concatenated
  85   // words. The |contraction_iterator_| splits a concatenated word extracted by
  86   // |text_iterator_| into word components so we can treat a concatenated word
  87   // consisting only of correct words as a correct word.
  88   SpellcheckWordIterator text_iterator_;
  89   SpellcheckWordIterator contraction_iterator_;
  90
  91   // Pointer to a platform-specific spelling engine, if it is in use. This
  92   // should only be set if hunspell is not used. (I.e. on OSX, for now)
  93   scoped_ptr<SpellingEngine> platform_spelling_engine_;
  94
  95   DISALLOW_COPY_AND_ASSIGN(SpellcheckLanguage);
  96 };
  97
  98 #endif  // CHROME_RENDERER_SPELLCHECKER_SPELLCHECK_LANGUAGE_H_