汉字字索引

#include  " ../DiskBuf.h "
#pragma  once


#ifndef IsHz
     #define  IsHz(x) (((x)>=0x81 && (x)<=0xA0)||((x)>=0xAA && (x)<=0xFE))
     #define  HzCode(x) ((((x)&0x7F)<<8)+((x)>>8))
#endif

#ifdef _DEBUG
     #define  _IndexBuffer_Size 0x80
#else
     #define  _IndexBuffer_Size 0x1000
#endif

class  IndexMaker
... {
private:
    struct IndexMakerBuf    //词库缓冲数组
    ...{
        char nWords[3];        //[WordMaxLen];
        unsigned long _Last_FileNum;
        unsigned long * _Data;
        unsigned long * DataPtr;
        unsigned long * DataEpr;
    } _WordsIndex[0x8000];

    unsigned long *_Diskbuf;
    unsigned long _filenum;            //文件号
    char _mOutput_Path[512];        //索引输出路径
    char* _mOutPath_EndPtr;            //指向索引路径末端
public:
    long __Error;
    //function
    ~IndexMaker(void)
    ...{
        if (_Diskbuf)
        ...{
            free(_Diskbuf);
        }
    }

    IndexMaker(char *OutPutDir)//cache set:default=840, unsigned long CacheSet
    ...{
        _filenum=1;
        __Error=NULL;

        strcpy(_mOutput_Path,OutPutDir);
        _mOutPath_EndPtr=_mOutput_Path+strlen(_mOutput_Path);

        unsigned long *tbuf=_Diskbuf=(unsigned long *)calloc(0x8000*_IndexBuffer_Size,4);
        if (!_Diskbuf) ...{ __Error=_Error_Malloc;return ;}
        
        memset(_WordsIndex,0,sizeof(_WordsIndex));
        for(unsigned long i=0;i<0x8000;i++)
        ...{
            IndexMakerBuf *tWi=_WordsIndex+i;
            *(unsigned short *)(tWi->nWords)=(unsigned short)(((i>>8)|0x80)+((i&0xFF)<<8));
            tWi->nWords[2]=NULL;
            tWi->_Last_FileNum=NULL;
            tWi->DataPtr=tWi->_Data=tbuf;
            tbuf+=_IndexBuffer_Size;
            tWi->DataEpr=tbuf;
        }
    }        

    long _MakeIndex(DiskBuffer *m_Buffer)
    ...{
        char *m_FileText=NULL;
        while (m_Buffer->_PopData(&m_FileText,&_filenum)) _fenci(m_FileText);
        for(unsigned long i=0;i<0x8000;i++)
        ...{
            if (_WordsIndex[i].DataPtr!=_WordsIndex[i]._Data) _Write2File(_WordsIndex+i);
        }
        return __Error;
    }
private:
    void _Write2File(IndexMakerBuf *nDataCache)
    ...{
        strcpy(_mOutPath_EndPtr,nDataCache->nWords);//char name
        FILE *outFile;
        if (NULL==(outFile=fopen(_mOutput_Path,"ab+")))
        ...{
            printf("error write to file %s %lx ",_mOutput_Path,*_mOutPath_EndPtr);
        }
        else
        ...{
            fwrite(nDataCache->_Data,4,(nDataCache->DataPtr)-(nDataCache->_Data),outFile);
            fclose(outFile);
        }
        nDataCache->DataPtr=nDataCache->_Data;
    }

    void _fenci(char *mTextCharPtr)
    ...{
        mTextCharPtr--;//起始位置减一，使第一个位置为一而不是零
        char * bptChar=mTextCharPtr;
        while (*(bptChar)) 
        ...{
            if ((*bptChar)<0)
            ...{
                if (IsHz(*(unsigned char *)(bptChar)))
                ...{
                    //_Write2Cache(_WordsIndex+HzCode(*(unsigned short *)(bptChar)),(unsigned long)(bptChar-mTextCharPtr));
                    IndexMakerBuf *tDataCache=_WordsIndex+HzCode(*(unsigned short *)(bptChar));
                    unsigned long pushData=(unsigned long)(bptChar-mTextCharPtr);
                    //void _Write2Cache(IndexMakerBuf *tDataCache,unsigned long pushData)
                    //{
                        if (tDataCache->DataPtr==tDataCache->DataEpr) _Write2File(tDataCache);
                        if (tDataCache->_Last_FileNum!=_filenum)
                        ...{
                            *(tDataCache->DataPtr)=_filenum;
                            tDataCache->_Last_FileNum=_filenum;
                            tDataCache->DataPtr++;
                            if (tDataCache->DataPtr==tDataCache->DataEpr) _Write2File(tDataCache);
                        }
                        *(tDataCache->DataPtr)=pushData;
                        tDataCache->DataPtr++;
                    //}
                    //end function _Write2Cache
                }
                bptChar+=2;
            }
            else bptChar++;
        }
        //*//_WriteZero();
        //for(unsigned long i=1;i<0x7F00;i++) //Powered by barenx
        IndexMakerBuf *nDataCacheEpr=_WordsIndex+0x7F00;
        for(IndexMakerBuf *nDataCache=_WordsIndex+1;nDataCache<nDataCacheEpr;nDataCache++)
        ...{
            if (nDataCache->_Last_FileNum==_filenum)
            ...{
                if (nDataCache->DataPtr==nDataCache->DataEpr) _Write2File(nDataCache);
                *(nDataCache->DataPtr)=NULL;
                (nDataCache->DataPtr)++;
            }
        }
        //end _WriteZero();*/
    }
} ;

Powered by barenx