lexicaltree.h

这是一个从音频信号里提取特征参量的程序

// file: $isip/class/search/LexicalTree/LexicalTree.h
// version: $Id: LexicalTree.h,v 1.10 2002/12/05 21:42:54 alphonso Exp $
//

// make sure definitions are only made once
//
#ifndef ISIP_LEXICAL_TREE
#define ISIP_LEXICAL_TREE

#ifndef ISIP_DI_GRAPH
#include <DiGraph.h>
#endif

#ifndef ISIP_SEARCH_NODE
#include <SearchNode.h>
#endif

// LexicalTree: A class implement a lexical tree using Digraph data structure
// 
class LexicalTree : public DiGraph<SearchNode> {

  //---------------------------------------------------------------------------
  //
  // public constants
  //
  //---------------------------------------------------------------------------
public:

  // define the class name
  //
  static const String CLASS_NAME;

  //----------------------------------------
  //
  // i/o related constants
  //
  //----------------------------------------  

  static const String DEF_PARAM;

  //----------------------------------------
  //
  // other important constants
  //
  //----------------------------------------  

  // define algorithm choices
  //
  enum ALGORITHM { UNFACTORED = 0, DI_GRAPH, NGRAM, NGRAM_CACHED,
		   DEF_ALGORITHM = UNFACTORED};

  // define implementation choices
  //
  enum IMPLEMENTATION { ALWAYS_MAX=0, UPPER_BOUND, SUM, 
			DEF_IMPLEMENTATION = ALWAYS_MAX };

  //----------------------------------------
  //
  // default values and arguments
  //
  //----------------------------------------  
  
  //---------------------------------------
  //
  // error codes
  //
  //---------------------------------------
  
  static const long ERR = (long)90600;
  
  //---------------------------------------------------------------------------
  //
  // protected data
  //
  //---------------------------------------------------------------------------
protected:

  // type definition
  // 
  typedef SearchNode LexicalNode;
  
  // type definition
  // 
  typedef SearchSymbol LexSymbol;
  
  // type definition
  // 
  typedef GraphVertex<LexicalNode> GVLexicalNode;
  
  // type definition
  // 
  typedef DiGraph<LexicalNode> Pronunciation;
  
  // type definition
  // 
  typedef GraphArc<LexicalNode> GALexicalNode;
  
  // define a static debug level
  //
  static Integral::DEBUG debug_level_d;
  
  // define a static memory manager
  //
  static MemoryManager mgr_d;

  //---------------------------------------------------------------------------
  //
  // private data
  //
  //---------------------------------------------------------------------------
private:

  // If we put the codes on spreading ngrams in the search process,
  // these definitions can be removed in the future. 
  // 
  
  // algorithm name
  //
  ALGORITHM algorithm_d;

  // implementation type
  //
  IMPLEMENTATION implementation_d;

  // the start node of the LexicalTree
  //  
  GVLexicalNode* root_vert_d;

  // the weight to term
  //
  float term_weight_d;
  
  //---------------------------------------------------------------------------
  //
  // required public methods
  //
  //---------------------------------------------------------------------------
public:

  // method: name
  //
  static const String& name() {
    return CLASS_NAME;
  }

  // method: diagnose
  //
  static boolean diagnose(Integral::DEBUG debug_level);

  // method: debug
  //
  boolean debug(const unichar* message_a) ;

  // method: setDebug
  //
  static boolean setDebug(Integral::DEBUG debug_level) {
    debug_level_d = debug_level;
    return true;
  }
  
  // constructor(s)
  //
  LexicalTree();
  LexicalTree(const LexicalTree& copy_tree);
  
  // method: destructor
  //
  ~LexicalTree() {}

  // assign methods
  //
  boolean LexicalTree::assign(const LexicalTree& copy_tree_a);
  
  // method: sofSize
  //
  long sofSize() const {
    return Error::handle(name(), L"sofSize", Error::NOT_IMPLEM, __FILE__,
			 __LINE__);
  }
  
  // method: read
  //
  boolean read(Sof& sof, long tag, const String& cname = CLASS_NAME) {
    return Error::handle(name(), L"read", Error::NOT_IMPLEM, __FILE__,
			 __LINE__);
  }

  // method: write
  //
  boolean write(Sof& sof, long tag, const String& cname = CLASS_NAME) const {
    return Error::handle(name(), L"write", Error::NOT_IMPLEM, __FILE__,
			 __LINE__);
  }

  // method: readData
  //
  boolean readData(Sof& sof, const String& pname = DEF_PARAM,
		   long size = SofParser::FULL_OBJECT, boolean param = true,
                   boolean nested = false) {
    return Error::handle(name(), L"readData", Error::NOT_IMPLEM, __FILE__,
			 __LINE__);
  }

  // method: writeData
  //
  boolean writeData(Sof& sof, const String& pname = DEF_PARAM) const {
    return Error::handle(name(), L"writeData", Error::NOT_IMPLEM, __FILE__,
			 __LINE__);
  }

  // equality method
  //
  boolean eq(const LexicalTree& compare_tree_a) const;
  
  // method: new
  //
  static void* operator new(size_t size) {
    return mgr_d.get();
  }

  // method: new[]
  //
  static void* operator new[](size_t size) {
    return mgr_d.getBlock(size);
  }

  // method: delete
  //
  static void operator delete(void* ptr) {
    mgr_d.release(ptr);
  }

  // method: delete[]
  //
  static void operator delete[](void* ptr) {
    mgr_d.releaseBlock(ptr);
  }

  // method: setGrowSize
  //
  static boolean setGrowSize(long grow_size) {
    return mgr_d.setGrow(grow_size);
  }
  
  // clear methods
  //
  boolean clear(Integral::CMODE ctype = Integral::DEF_CMODE);

  //---------------------------------------------------------------------------
  //
  // class-specific public methods
  //
  //---------------------------------------------------------------------------


  // set the algorithm
  //
  boolean setAlgorithm(LexicalTree::ALGORITHM algorithm){
    return (algorithm_d = algorithm);
  }

  // get the algorithm
  //
  ALGORITHM getAlgorithm() {
    return algorithm_d;
  }

  // set the implementation
  //
  boolean setImplementation(IMPLEMENTATION implementation) {
    return (implementation_d = implementation);
  }

  // get the implementation
  //
  IMPLEMENTATION getImplementation() {
    return implementation_d;
  }
  
  // expand subgraphs into lexical trees
  // 
  static boolean 
  LexicalTree::expandLexicalTree(DiGraph<LexicalNode>& word_graph_a, 
			       const Vector<DiGraph<LexicalNode> >& pron_vec_a,
			       long level_a);

  static
  boolean expandLexicalTree(GVLexicalNode*& root_node_a, 
			    const Vector<DiGraph<LexicalNode> >& pron_vec_a,
			    long level_a);
  
  //---------------------------------------------------------------------------
  //
  // private methods
  //
  //---------------------------------------------------------------------------
private:


  // some of these methods are reserved for future changes. If we move
  // the ngram spreading in the search process, many of thsee
  // methods will be useless
  //

  // debug functions
  //
  boolean debugTree(const unichar* message_a);
  
  boolean debugVertex (GVLexicalNode*& lex_vert_a,
		       boolean recursive_a = false,
		       GALexicalNode*  arc_a = NULL);
  
  // factor the lexical tree from a vertex
  //
  float factorLexicalTree( GVLexicalNode*,
			   const HashTable<Long, Float>* ngram_a = NULL,
			   long history_a  = -1 );

  // factor the LexicalTree (whole tree )
  //
  float factorLexicalTree( const HashTable<Long, Float>* ngram_a = NULL,
			   long history_a  = -1 ){
    return factorLexicalTree(this->getStart(), ngram_a, history_a);
  }

  // build the lexical tree
  // 
  boolean
  LexicalTree::buildLexicalTree(DiGraph<LexicalNode>& word_graph_a, 
			       const Vector<DiGraph<LexicalNode> >& pron_vec_a,
				SingleLinkedList<LexicalTree>& tree_vec_a);
    
  LexicalTree* buildLexicalTree(GVLexicalNode*& root_node_a, 
			   const Vector<DiGraph<LexicalNode> >& pron_vec_a);
  
  // insert a new pronication into the lexical tree
  //
  boolean insertPron(GVLexicalNode*& word_vert_a,
		     const DiGraph<LexicalNode>& pron_a,
		     const float & weight_a);
  //insert a subgraph of a pronication into the lexical tree
  //
  boolean insertSubgraph(GVLexicalNode*& word_vert_a,
			 GVLexicalNode*& curr_lex_vert_a,
			 GVLexicalNode*& curr_pron_vert_a,
			 const float & pron_prob_a,
			 boolean insert_mode_a = true);
  
  //expand a subgraph of a pronication into the lexical tree structure
  //
  static
  boolean expandSubgraph(GVLexicalNode*& word_vert_a,
			 GVLexicalNode*& curr_lex_vert_a,
			 GVLexicalNode*& curr_pron_vert_a,
			 const float & pron_prob_a,
			 boolean insert_mode_a = true);
  
  // this method find the successor with given symbol ID.
  // If it is not exsited, return a NULL.
  //
  static
  GVLexicalNode* findSuccVert(GVLexicalNode*& lex_vert_a,
			      long symbol_id_a );

  // overload the parent's insertArc function
  //
  boolean insertArc(GVLexicalNode* start_vertex_a,
		    GVLexicalNode* end_vertex_a,
		    boolean is_epsilon = GALexicalNode::DEF_EPSILON,
		    float weight = GALexicalNode::DEF_WEIGHT);
  
};

// end of include file
//
#endif