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

Go to the documentation of this file.

#include "../base.h"

void CmdBuildSuffixTree( const S8 * fileNameInput, const S8 * fileNameOutput, UINT bitsPerSymbol, UINT encoding, boolean tarjanTable ){
    FILE * fpInput;
    FILE * fpOutput;
    VoidSequence * line = NULL;
    CompressedAutomaton * aut;
    SuffixTreeBuildHelp * help;
    UINT i, nLine, symbolSize;
    mSymbolAndVariables( zero )

    mValidateBitsPerSymbol( bitsPerSymbol )
    symbolSize = bitsPerSymbol/8;
    mSymbolAssignValue( zero, 0, symbolSize )
    aut = CompressedAutomatonInit( symbolSize );
    help = SuffixTreeBuildHelpInit( aut );
    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 );
        SuffixTreeAdd( help, line );
    }
    Fclose( fpInput );
    CompressedAutomatonSortTransitions( aut, help->transitionsFrom );
    CompressedAutomatonShrink( aut );
    SuffixTreeBuildHelpFree( help );
    if( tarjanTable ){
        CompressedAutomatonAddTarjanTable( aut );
    }
    fpOutput = Fopen( fileNameOutput, "wb" );
    CompressedAutomatonWrite( aut, fpOutput );
    Fclose( fpOutput );
    CompressedAutomatonFree( aut );
}

void CmdBuildCDAWG( const S8 * fileNameInput, const S8 * fileNameOutput, UINT bitsPerSymbol, UINT encoding, boolean tarjanTable ){
    FILE * fpInput;
    FILE * fpOutput;
    VoidSequence * line = NULL;
    CompressedAutomaton * aut;
    CDAWGBuildHelp * help;
    UINT i, nLine, symbolSize;
    mSymbolAndVariables( zero )

    mValidateBitsPerSymbol( bitsPerSymbol )
    symbolSize = bitsPerSymbol/8;
    mSymbolAssignValue( zero, 0, symbolSize )
    aut = CompressedAutomatonInit( symbolSize );
    help = CDAWGBuildHelpInit( aut );
    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 );
        CDAWGAdd( help, line );
    }
    Fclose( fpInput );
    CompressedAutomatonSortTransitions( aut, help->transitionsFrom );
    CompressedAutomatonShrink( aut );
    CDAWGBuildHelpFree( help );
    if( tarjanTable ){
        CompressedAutomatonAddTarjanTable( aut );
    }
    fpOutput = Fopen( fileNameOutput, "wb" );
    CompressedAutomatonWrite( aut, fpOutput );
    Fclose( fpOutput );
    CompressedAutomatonFree( aut );
}

void CmdBuildSCDAWG( const S8 * fileNameInput, const S8 * fileNameOutput, UINT bitsPerSymbol, UINT encoding, boolean tarjanTable ){
    FILE * fpInput;
    FILE * fpOutput;
    VoidSequence * line = NULL;
    VoidSequence * sharpDocumentDollar = NULL;
    SCDAWG * aut;
    SCDAWGBuildHelp * help;
    UINT i, n, nLine, symbolSize;
    mSymbolAndVariables( zero )
    mSymbolAndVariables( one )

    mValidateBitsPerSymbol( bitsPerSymbol )
    symbolSize = bitsPerSymbol/8;
    mSymbolAssignValue( zero, 0, symbolSize )
    mSymbolAssignValue( one, 1, symbolSize )
    aut = SCDAWGInit( symbolSize );
    help = SCDAWGBuildHelpInit( aut );
    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++ ){
            n = SymbolToUINT( mVoidSequenceElement(line, i), symbolSize );
            if( n == 0 || n == 1 ){
                S8 msg[MAX_INPUT_STRING_SIZE];

                sprintf( msg, "Cannot process line %llu - 0 and 1 are reserved as new symbols. (pos is %d, sign is %d)", (U64)(nLine), i, n );
                Throw( msg );
            }
        }
        sharpDocumentDollar = VoidSequenceInit2( symbolSize, line->seqStored + 2, 1 );
        VoidSequenceAdd( sharpDocumentDollar, one );
        VoidSequenceAppend( sharpDocumentDollar, line );
        VoidSequenceAdd( sharpDocumentDollar, zero );
        SCDAWGAdd( help, sharpDocumentDollar );
        VoidSequenceFree( sharpDocumentDollar );
    }
    Fclose( fpInput );
    SCDAWGClose( help );
    SCDAWGShrink( aut );
    SCDAWGBuildHelpFree( help );
    if( tarjanTable ){
        SCDAWGAddTarjanTable( aut );
    }
    fpOutput = Fopen( fileNameOutput, "wb" );
    SCDAWGWrite( aut, fpOutput );
    Fclose( fpOutput );
    SCDAWGFree( aut );
}