trunk/Src/DasherCore/LanguageModelling/DictLanguageModel.h

   1 // PPMLanguageModel.h
   2 //
   3 /////////////////////////////////////////////////////////////////////////////
   4 //
   5 // Copyright (c) 1999-2004 David Ward
   6 //
   7 /////////////////////////////////////////////////////////////////////////////
   8
   9 #ifndef __DictLanguageModel_h__
  10 #define __DictLanguageModel_h__
  11
  12 #include "../../Common/NoClones.h"
  13 #include "../../Common/Allocators/PooledAlloc.h"
  14 #include "LanguageModel.h"
  15 #include "PPMLanguageModel.h"
  16 #include "../DasherTypes.h"
  17
  18 #include <vector>
  19 #include <map>
  20 #include <string>
  21 #include <stdio.h>
  22
  23 //static char dumpTrieStr[40000];
  24 //const int maxcont =200;
  25
  26 namespace Dasher {
  27   /// \ingroup LM
  28   /// \{
  29   class CDictLanguageModel:public CLanguageModel {
  30   public:
  31     CDictLanguageModel(Dasher::CEventHandler * pEventHandler, CSettingsStore * pSettingsStore, const CSymbolAlphabet & Alphabet);
  32     virtual ~CDictLanguageModel();
  33
  34     Context CreateEmptyContext();
  35     void ReleaseContext(Context context);
  36     Context CloneContext(Context context);
  37
  38     virtual void GetProbs(Context Context, std::vector < unsigned int >&Probs, int iNorm) const;
  39
  40     virtual void EnterSymbol(Context context, int Symbol);
  41     virtual void LearnSymbol(Context context, int Symbol) {
  42       EnterSymbol(context, Symbol);
  43     };                          // Never learn in this model
  44
  45     virtual int GetMemory() {
  46       return NodesAllocated;
  47   } private:
  48
  49     void MyLearnSymbol(Context context, int Symbol);
  50
  51     class CDictnode {
  52     public:
  53       CDictnode * find_symbol(int sym) const;
  54       CDictnode *child;
  55       CDictnode *next;
  56       CDictnode *vine;
  57       unsigned short int count;
  58       int sbl;
  59
  60         CDictnode(int sym);
  61         CDictnode();
  62     };
  63
  64     class CDictContext {
  65     public:
  66       CDictContext(CDictContext const &input) {
  67         head = input.head;
  68         word_head = input.word_head;
  69         current_word = input.current_word;
  70         order = input.order;
  71         word_order = input.word_order;
  72     } CDictContext(CDictnode * _head = 0, int _order = 0):head(_head), order(_order), word_head(_head), word_order(0) {
  73       };                        // FIXME - doesn't work if we're trying to create a non-empty context
  74       ~CDictContext() {
  75       };
  76       void dump();
  77       CDictnode *head;
  78       int order;
  79
  80       std::string current_word;
  81       CDictnode *word_head;
  82       int word_order;
  83
  84     };
  85
  86     CDictnode *AddSymbolToNode(CDictnode * pNode, symbol sym, int *update);
  87
  88     void AddSymbol(CDictContext & context, symbol sym);
  89
  90     void CollapseContext(CDictContext & context) const;
  91
  92     int lookup_word(const std::string & w);
  93     int lookup_word_const(const std::string & w) const;
  94
  95     CDictContext *m_rootcontext;
  96     CDictnode *m_pRoot;
  97
  98     std::map < std::string, int >dict;  // Dictionary
  99     int nextid;
 100
 101     int NodesAllocated;
 102
 103     int max_order;
 104
 105     mutable CSimplePooledAlloc < CDictnode > m_NodeAlloc;
 106     CPooledAlloc < CDictContext > m_ContextAlloc;
 107   };
 108   /// \}
 109
 110 ////////////////////////////////////////////////////////////////////////
 111 // Inline functions
 112 ////////////////////////////////////////////////////////////////////////
 113
 114 ////////////////////////////////////////////////////////////////////////
 115
 116   inline Dasher::CDictLanguageModel::CDictnode::CDictnode(symbol sym):sbl(sym) {
 117     child = next = vine = 0;
 118     count = 1;
 119   }
 120
 121 ////////////////////////////////////////////////////////////////////////
 122
 123   inline CDictLanguageModel::CDictnode::CDictnode() {
 124     child = next = vine = 0;
 125     count = 1;
 126   }
 127
 128 ///////////////////////////////////////////////////////////////////
 129
 130   inline CLanguageModel::Context CDictLanguageModel::CreateEmptyContext() {
 131     CDictContext *pCont = m_ContextAlloc.Alloc();
 132     *pCont = *m_rootcontext;
 133     return (Context) pCont;
 134   }
 135
 136 ///////////////////////////////////////////////////////////////////
 137
 138   inline CLanguageModel::Context CDictLanguageModel::CloneContext(Context Copy) {
 139     CDictContext *pCont = m_ContextAlloc.Alloc();
 140     CDictContext *pCopy = (CDictContext *) Copy;
 141     *pCont = *pCopy;
 142     return (Context) pCont;
 143   }
 144
 145 ///////////////////////////////////////////////////////////////////
 146
 147   inline void CDictLanguageModel::ReleaseContext(Context release) {
 148     m_ContextAlloc.Free((CDictContext *) release);
 149   }
 150
 151 ///////////////////////////////////////////////////////////////////
 152
 153 }                               // end namespace Dasher
 154
 155 #endif /* #ifndef __DictLanguageModel_H__ */