Src/DasherCore/Trainer.h

   1 #ifndef __trainer_h__
   2 #define __trainer_h__
   3
   4 #include "LanguageModelling/PPMPYLanguageModel.h"
   5 #include "Alphabet/AlphInfo.h"
   6 #include "AbstractXMLParser.h"
   7
   8 namespace Dasher {
   9   class CTrainer : public AbstractParser {
  10
  11   public:
  12     CTrainer(CMessageDisplay *pMsgs, CLanguageModel *pLanguageModel, const CAlphInfo *pInfo, const CAlphabetMap *pAlphabet);
  13     virtual ~CTrainer();
  14     class ProgressIndicator {
  15     public:
  16       virtual void bytesRead(off_t)=0;
  17     };
  18
  19     void SetProgressIndicator(ProgressIndicator *pProg) {m_pProg = pProg;}
  20
  21     ///Parses a text file; bUser ignored.
  22     bool Parse(const std::string &strDesc, std::istream &in, bool bUser);
  23
  24   protected:
  25
  26     virtual void Train(CAlphabetMap::SymbolStream &syms);
  27
  28     ///Try to read a context-switch escape sequence from the symbolstream.
  29     /// \param sContext context to be reinitialized if a context-switch command is found
  30     /// \param sym symbol that was most recently read from stream, i.e. that
  31     /// might be the start of the context-switch command
  32     /// \syms symbolstream to read body of command from, if necessary; should be positioned
  33     /// just after the first occurrence of the escape character.
  34     /// \return true if a context-switch command was found (=> sContext reinitialized);
  35     ///  false if the character was not a context-switch; specifically, if there was a
  36     ///  double context-switch character, meaning that an actual occurrence of that
  37     ///  character is desired to be fed into the LanguageModel, this method returns false
  38     ///  with the stream positioned just after the second ctx-switch character
  39     ///  (ready to continue reading as per normal)
  40     bool readEscape(CLanguageModel::Context &sContext, symbol sym, CAlphabetMap::SymbolStream &syms);
  41
  42     ///Returns the description of the file as passed to Parse()
  43     /// (usually a filename)
  44     const std::string &GetDesc() {return m_strDesc;}
  45     const CAlphabetMap * const m_pAlphabet;
  46     CLanguageModel * const m_pLanguageModel;
  47     const CAlphInfo * const m_pInfo;
  48     // symbol number in alphabet of the context-switch character (maybe 0 if not in alphabet!)
  49     int m_iCtxEsc;
  50   private:
  51     ProgressIndicator *m_pProg;
  52     std::string m_strDesc;
  53   };
  54
  55 }
  56
  57 #endif