third_party/hunspell/google/bdict_reader.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 CHROME_THIRD_PARTY_HUNSPELL_GOOGLE_DIC_READER_H__
   6 #define CHROME_THIRD_PARTY_HUNSPELL_GOOGLE_DIC_READER_H__
   7
   8 #include <string>
   9 #include <vector>
  10
  11 #include "base/basictypes.h"
  12 #include "third_party/hunspell/google/bdict.h"
  13
  14 namespace hunspell {
  15
  16 class BDictReader;
  17 class NodeReader;
  18
  19 // Iterators -------------------------------------------------------------------
  20
  21 // Iterates through all words in the dictionary. It will fill the word into
  22 // a caller-specified buffer.
  23 class WordIterator {
  24  public:
  25   WordIterator(const WordIterator& other);
  26   ~WordIterator();
  27
  28   // This must be explicitly declared and implemneted in the .cc file so it will
  29   // compile without knowing the size of NodeInfo.
  30   WordIterator& operator=(const WordIterator&);
  31
  32   // Fills the buffer with the next word and the affixes for it into the given
  33   // array. Returns the number of affixes. A return value of 0 means there are
  34   // no more words.
  35   int Advance(char* output_buffer, size_t output_len,
  36               int affix_ids[BDict::MAX_AFFIXES_PER_WORD]);
  37
  38  private:
  39   friend class BDictReader;
  40   struct NodeInfo;
  41
  42   WordIterator(const NodeReader& reader);
  43
  44   // Called by Advance when a leaf is found to generate the word, affix list,
  45   // and return value.
  46   int FoundLeaf(const NodeReader& node, char cur_char,
  47                 char* output_buffer, size_t output_len,
  48                 int affix_ids[BDict::MAX_AFFIXES_PER_WORD]);
  49
  50   std::vector<NodeInfo> stack_;
  51 };
  52
  53 // Will iterate over a list of lines separated by NULLs.
  54 class LineIterator {
  55  public:
  56   // Returns the next word in the sequence or NULL if there are no mode.
  57   const char* Advance();
  58
  59   // Advances to the next word in the sequence and copies it into the given
  60   // buffer, of the given length. If it doesn't fit, it will be truncated.
  61   // Returns true on success.
  62   bool AdvanceAndCopy(char* buf, size_t buf_len);
  63
  64   // Returns true when all data has been read. We're done when we reach a
  65   // double-NULL or a the end of the input (shouldn't happen).
  66   bool IsDone() const;
  67
  68  protected:
  69   friend class BDictReader;
  70
  71   LineIterator(const unsigned char* bdict_data, size_t bdict_length,
  72                size_t first_offset);
  73
  74   const unsigned char* bdict_data_;
  75   size_t bdict_length_;
  76
  77   // Current offset within bdict_data of the next string to read.
  78   size_t cur_offset_;
  79 };
  80
  81 // Created by GetReplacementIterator to iterate over all replacement pairs.
  82 class ReplacementIterator : public LineIterator {
  83  public:
  84   // Fills pointers to NULL terminated strings into the given output params.
  85   // Returns false if there are no more pairs and nothing was filled in.
  86   bool GetNext(const char** first, const char** second);
  87
  88  private:
  89   friend class BDictReader;
  90
  91   ReplacementIterator(const unsigned char* bdict_data, size_t bdict_length,
  92                       size_t first_offset)
  93       : LineIterator(bdict_data, bdict_length, first_offset) {
  94   }
  95 };
  96
  97 // Reads a BDict file mapped into memory.
  98 class BDictReader {
  99  public:
 100   // You must call Init and it must succeed before calling any other functions.
 101   BDictReader();
 102
 103   // Initializes the reader with the given data. The data does not transfer
 104   // ownership, and the caller must keep it valid until the reader is destroyed.
 105   // Returns true on success.
 106   bool Init(const unsigned char* bdic_data, size_t bdic_length);
 107
 108   // Returns true if Init() succeeded and other functions can be called.
 109   bool IsValid() const { return !!bdict_data_; }
 110
 111   // Locates the given word in the dictionary. There may be multiple matches if
 112   // the word is listed multiple times in the dictionary with different affix
 113   // rules.
 114   //
 115   // The number of matches is returned, and that number of corresponding affix
 116   // group IDs are filled into |*affix_indices|. These IDs may be 0 to indicate
 117   // there is no affix for that particular match. A return valuf of 0 means that
 118   // there are no matches.
 119   int FindWord(const char* word,
 120                int affix_indices[BDict::MAX_AFFIXES_PER_WORD]) const;
 121
 122   // Returns an iterator that will go over all AF lines ("affix groups").
 123   LineIterator GetAfLineIterator() const;
 124
 125   // Returns an iterator that will go over all SFX/PFX lines ("affix rules").
 126   LineIterator GetAffixLineIterator() const;
 127
 128   // Returns an iterator that will go over all "other" lines.
 129   LineIterator GetOtherLineIterator() const;
 130
 131   // Returns an iterator that can be used to iterate all replacements.
 132   ReplacementIterator GetReplacementIterator() const;
 133
 134   // Used for testing, returns an iterator for all words in the dictionary.
 135   WordIterator GetAllWordIterator() const;
 136
 137  private:
 138   // Non-NULL indicates Init succeeded.
 139   const unsigned char* bdict_data_;
 140   size_t bdict_length_;
 141
 142   // Pointer not owned by this class. It will point into the data. It will be
 143   // NULL if the data is invalid.
 144   const BDict::Header* header_;
 145
 146   const BDict::AffHeader* aff_header_;
 147
 148   DISALLOW_EVIL_CONSTRUCTORS(BDictReader);
 149 };
 150
 151 }  // namespace hunspell
 152
 153 #endif  // CHROME_THIRD_PARTY_HUNSPELL_GOOGLE_DIC_READER_H__