/Users/dbr/ma/src/bas/lml/cmd/cmdGenerateSuffixes.c

Go to the documentation of this file.

#include "../base.h"

typedef struct tCntStarts{
    VoidSequence * cnt;
    UINTSequence * starts;
    void * zero;
} CntStarts;

static SINT Cmp( void * ptr, UINT s1, UINT s2 ){
    CntStarts * cntStarts;
    void * str1;
    void * str2;
    UINT symbolSize;

    cntStarts = (CntStarts *)(ptr);
    symbolSize = cntStarts->cnt->elementSize;
    str1 = mVoidSequenceElement( cntStarts->cnt, cntStarts->starts->seq[s1] );
    str2 = mVoidSequenceElement( cntStarts->cnt, cntStarts->starts->seq[s2] );
    while( CmpSymbols( str1, str2, symbolSize ) == 0 && SymbolToUINT( str1, symbolSize ) != 0 ){
        str1 = str1 + symbolSize;
        str2 = str2 + symbolSize;
    }
    return CmpSymbols( str1, str2, symbolSize );
}

static void Swap( void * ptr, UINT s1, UINT s2 ){
    CntStarts * cntStarts;
    UINT tmp;

    cntStarts = (CntStarts *)(ptr);
    mSwapVariables( cntStarts->starts->seq[s1], cntStarts->starts->seq[s2], tmp );
}

static void Print( CntStarts * cntStarts, UINT s, FILE * fp, UINT encoding ){
    UINT symbolSize;
    void * str;

    symbolSize = cntStarts->cnt->elementSize;
    str = mVoidSequenceElement( cntStarts->cnt, cntStarts->starts->seq[s] );
    while( SymbolToUINT( str, symbolSize ) != 0 ){
        PrintSymbol( str, fp, symbolSize, encoding );
        str = str + symbolSize;
    }
    PrintLine( fp, symbolSize, encoding );
}

void CmdGenerateSuffixes( const S8 * fileNameInput, const S8 * fileNameOutput, UINT bitsPerSymbol, UINT encoding ){
    FILE * fpInput;
    FILE * fpOutput;
    VoidSequence * line = NULL;
    VoidSequence * cnt;
    UINTSequence * starts;
    CntStarts cntStarts;
    UINT i, nLine, symbolSize;
    mSymbolAndVariables( zero )

    mValidateBitsPerSymbol( bitsPerSymbol )
    symbolSize = bitsPerSymbol/8;
    mSymbolAssignValue( zero, 0, symbolSize )
    cnt = VoidSequenceInit( symbolSize );
    fpInput = Fopen( fileNameInput, "rb" );
    nLine = 0;
    while( bTRUE ){
        if( line != NULL ){
            VoidSequenceFree( line );
        }
        line = VoidSequenceReadLine( fpInput, symbolSize, encoding );
        if( line == NULL ){
            break;
        }
        nLine++;
        for( i = 0; i < line->seqStored; i++ ){
            if( SymbolToUINT( mVoidSequenceElement(line, i), symbolSize ) == 0 ){
                S8 msg[MAX_INPUT_STRING_SIZE];

                sprintf( msg, "Cannot process line %llu - 0 is reserved as a new symbol.", (U64)(nLine) );
                Throw( msg );
            }
        }
        VoidSequenceAdd( line, zero );
        VoidSequenceAppend( cnt, line );
    }
    Fclose( fpInput );
    starts = UINTSequenceInit2( cnt->seqStored, 1 );
    for( i = 0; i < cnt->seqStored; i++ ){
        UINTSequenceAdd( starts, i );
    }
    cntStarts.cnt = cnt;
    cntStarts.starts = starts;
    cntStarts.zero = zero;
    Sort( &cntStarts, starts->seqStored, Cmp, Swap );
    fpOutput = Fopen( fileNameOutput, "wb" );
    if( starts->seqStored > 0 ){
        Print( &cntStarts, 0, fpOutput, encoding );
        for( i = 1; i < starts->seqStored; i++ ){
            if( Cmp( &cntStarts, i-1, i ) != 0 ){
                Print( &cntStarts, i, fpOutput, encoding );
            }
        }
    }
    Fclose( fpOutput );
    VoidSequenceFree( cnt );
    UINTSequenceFree( starts );
}