main page namespaces classes files
BIU home

src/biu/Alphabet.hh

Go to the documentation of this file.

#ifndef ALPHABET_HH_
#define ALPHABET_HH_


#include <vector>
#include <string>

#include "biu/HashMap.hh"

namespace biu
{
    class Alphabet
    {
    public:
        typedef size_t  AlphElem;
        typedef unsigned char   CAlphElem;
        
        typedef std::vector<AlphElem> Sequence;
        
        typedef std::vector<CAlphElem> CSequence;
        
    private:

#ifdef HAVE_GNU_HASH_MAP

        class hash_string {
        public:

            size_t operator()(const std::string& str) const
            {
                size_t hash = 5381;
    
                for (size_t i = 0; i < str.size(); i++) {
                    hash = ((hash << 5) + hash) + (size_t)str[i]; // hash * 33 + str[i]
                }
    
                return hash;
            }
                 
        };
    
        typedef __gnu_cxx::hash_map< std::string, AlphElem, hash_string > STR2ALPH_MAP;
#else
        typedef std::map< std::string, AlphElem > STR2ALPH_MAP;
#endif

        STR2ALPH_MAP string2alph;   
        
        std::vector<std::string> alph2string;   
        
        size_t elementLength;
        
        std::vector<int> compressBase;
        
    public:
    
        Alphabet(   const std::string& alphabetString, 
                    const size_t elementLength);
        
        Alphabet(const std::vector<std::string> & alphabetStrings);
        
        virtual ~Alphabet();

        bool operator== (const Alphabet& alph2) const;
        bool operator!= (const Alphabet& alph2) const;

        size_t getAlphabetSize() const;
        
        size_t getElementLength() const;
    
        // string to sequence to string
    
        Sequence getSequence(const std::string& seqString) const;
        
        AlphElem getElement(const std::string& alphElemStr) const;
        
        std::string getString(const Alphabet::Sequence& sequence) const;
        
        std::string getString(const Alphabet::AlphElem& elem) const;
        
        CSequence compress(const Alphabet::Sequence& sequence) const;
        
        CSequence compressS(const std::string& sequence) const;
        
        Sequence decompress(const CSequence& sequence, const size_t seqLength) const;
        
        std::string decompressS(const CSequence& sequence, const size_t seqLength) const;
        
    
        // miscellaneous
    
        bool isAlphabetString(const std::string& str) const;
        
        bool isAlphabetSequence(const Sequence& seq) const;
        
        size_t getIndex(const AlphElem& elem) const;
        
        size_t getIndex(const std::string& elemStr) const;
        
        AlphElem getElement(const size_t index) const;
        
    };

} // namespace biu

  // include definitions
#include "biu/Alphabet.icc"

#endif /*ALPHABET_HH_*/