#include "ReadAlignChunk.h"
#include "GlobalVariables.h"
#include "ThreadControl.h"
#include "ErrorWarning.h"

void ReadAlignChunk::processChunks() {//read-map-write chunks  
    noReadsLeft=false; //true if there no more reads left in the file
    
    while (!noReadsLeft) {//continue until the input EOF
            //////////////read a chunk from input files and store in memory
        if (P->outFilterBySJoutStage<2) {//read chunks from input file

            if (P->runThreadN>1) pthread_mutex_lock(&g_threadChunks.mutexInRead);

            uint chunkInSizeBytesTotal[2]={0,0};
            while (chunkInSizeBytesTotal[0] < P->chunkInSizeBytes && chunkInSizeBytesTotal[1] < P->chunkInSizeBytes && !P->inOut->readIn[0].eof() && !P->inOut->readIn[1].eof()) {
                char nextChar=P->inOut->readIn[0].peek();
                if (nextChar=='@') {//fastq, not multi-line
                    P->iReadAll++; //increment read number
                    for (uint imate=0; imate<P->readNmates; imate++) {//for all mates           
                        uint32 iline=0;
                        if (P->outFilterBySJoutStage!=2) {//not the 2nd stage of the 2-stage mapping                        
                            
                            //read or skip the 1st field of the read name line
                            if (P->outSAMreadID=="Number") {
                                chunkInSizeBytesTotal[imate] += sprintf(chunkIn[imate] + chunkInSizeBytesTotal[imate], "@%llu", P->iReadAll);
                                string dummy1;
                                P->inOut->readIn[imate] >> dummy1; //skip the first field of the read name
                            } else {
                                P->inOut->readIn[imate] >> (chunkIn[imate] + chunkInSizeBytesTotal[imate]);
                                chunkInSizeBytesTotal[imate] += strlen(chunkIn[imate] + chunkInSizeBytesTotal[imate]);
                            };
                            
                            //read the second field of the read name line
                            char passFilterIllumina='N';
                            if (P->inOut->readIn[imate].peek()!='\n') {//2nd field exists
                                string field2;
                                P->inOut->readIn[imate] >> field2;
                                passFilterIllumina=field2.at(2);
                                if (passFilterIllumina!='Y') passFilterIllumina='N';
                            };
                            
                            //ignore the rest of the read name
                            P->inOut->readIn[imate].ignore(DEF_readNameSeqLengthMax,'\n');

                            chunkInSizeBytesTotal[imate] += sprintf(chunkIn[imate] + chunkInSizeBytesTotal[imate], " %llu %c %i \n", P->iReadAll, passFilterIllumina, P->readFilesIndex);
                            
                            iline=1;
                        };
//                         else {//2nd stage of 2-stage mapping
//                         read index and file index are already recorded with the read name, simply copy it
//                         P->inOut->readIn[imate].getline(chunkIn[imate] + chunkInSizeBytesTotal[imate], DEF_readNameSeqLengthMax+1 );                            
//                         };
                        for (;iline<4;iline++) {//TODO ignore the 3rd line of fastq
                            P->inOut->readIn[imate].getline(chunkIn[imate] + chunkInSizeBytesTotal[imate], DEF_readNameSeqLengthMax+1 );
                            chunkInSizeBytesTotal[imate] += P->inOut->readIn[imate].gcount();
                            chunkIn[imate][chunkInSizeBytesTotal[imate]-1]='\n';                        
                        };
                    };
                } else if (nextChar=='>') {//fasta, can be multiline, which is converted to single line
                    for (uint imate=0; imate<P->readNmates; imate++) {
                        if (P->outFilterBySJoutStage!=2) {//not the 2nd stage of the 2-stage mapping                        

                            if (P->outSAMreadID=="Number") {
                                chunkInSizeBytesTotal[imate] += sprintf(chunkIn[imate] + chunkInSizeBytesTotal[imate], ">%llu", P->iReadAll);
                            } else {
                                P->inOut->readIn[imate] >> (chunkIn[imate] + chunkInSizeBytesTotal[imate]);
                                chunkInSizeBytesTotal[imate] += strlen(chunkIn[imate] + chunkInSizeBytesTotal[imate]);
                            };
                            
                            P->inOut->readIn[imate].ignore(DEF_readNameSeqLengthMax,'\n');

                            chunkInSizeBytesTotal[imate] += sprintf(chunkIn[imate] + chunkInSizeBytesTotal[imate], " %llu %c %i \n", P->iReadAll, 'N', P->readFilesIndex);
                            
                            
                        };
//                         else {//2nd stage of 2-stage mapping
//                         read index and file index are already recorded with the read name, simply copy it
//                         P->inOut->readIn[imate].getline(chunkIn[imate] + chunkInSizeBytesTotal[imate], DEF_readNameSeqLengthMax+1 );                            
//                         };                        
                        nextChar=P->inOut->readIn[imate].peek();                        
                        while (nextChar!='@' && nextChar!='>' && nextChar!=' ' && nextChar!='\n' && !P->inOut->readIn[0].eof()) {//read multi-line fasta
                            P->inOut->readIn[imate].getline(chunkIn[imate] + chunkInSizeBytesTotal[imate], DEF_readSeqLengthMax + 1 );
                            if (P->inOut->readIn[imate].gcount()<2) break; //no more input
                            chunkInSizeBytesTotal[imate] += P->inOut->readIn[imate].gcount()-1;   
                            nextChar=P->inOut->readIn[imate].peek();
                        };
                        chunkIn[imate][chunkInSizeBytesTotal[imate]]='\n'; 
                        chunkInSizeBytesTotal[imate] ++;   
                    };                    
                } else if (nextChar==' ' || nextChar=='\n' || P->inOut->readIn[0].eof()) {//end of stream
                    break;
                } else {
                    string word1;
                    P->inOut->readIn[0] >> word1;
                    if (word1=="FILE") {//new file marker
                        P->inOut->readIn[0] >> P->readFilesIndex;
                        P->inOut->logMain << "Starting to map file # " << P->readFilesIndex<<"\n";
                        for (uint imate=0; imate<P->readNmates; imate++) {
                            P->inOut->logMain << "mate " <<imate+1 <<":   "<<P->readFilesNames.at(imate).at(P->readFilesIndex) <<"\n";
                            P->inOut->readIn[imate].ignore(numeric_limits<streamsize>::max(),'\n');
                        };
                        P->inOut->logMain<<flush;
//                         if (P->readNmates==2) {//skip the FILE line for the second read
//                             getline(P->inOut->readIn[1],word1);
//                         };
                    } else {//error
                        ostringstream errOut;
                        errOut << "ERROR_0001: EXITING because of FATAL ERROR in input reads: unknown file format: the read ID should start with @ or > \n";
                        exitWithError(errOut.str(),std::cerr, P->inOut->logMain, EXIT_CODE_INPUT_FILES, *P);       
                    };
                };
            };
            //TODO: check here that both mates are zero or non-zero
            if (chunkInSizeBytesTotal[0]==0) {
                noReadsLeft=true; //true if there no more reads left in the file
                iChunkIn=g_threadChunks.chunkInN;//to keep things consistent
                g_threadChunks.chunkInN++;                
            } else {
                noReadsLeft=false; 
                iChunkIn=g_threadChunks.chunkInN;
                g_threadChunks.chunkInN++;
            };

            for (uint imate=0; imate<P->readNmates; imate++) chunkIn[imate][chunkInSizeBytesTotal[imate]]='\n';//extra empty line at the end of the chunks

            if (P->runThreadN>1) pthread_mutex_unlock(&g_threadChunks.mutexInRead);
            
        } else {//read from one file per thread
            noReadsLeft=true;
            for (uint imate=0; imate<P->readNmates; imate++) {
                RA->chunkOutFilterBySJoutFiles[imate].flush();
                RA->chunkOutFilterBySJoutFiles[imate].seekg(0,ios::beg);
                RA->readInStream[imate]=& RA->chunkOutFilterBySJoutFiles[imate];
            };
        };
        
        mapChunk();
                
        if (iThread==0 && P->runThreadN>1 && P->outSAMorder=="PairedKeepInputOrder") {//concatenate Aligned.* files
            chunkFilesCat(P->inOut->outSAM, P->outFileTmp + "/Aligned.out.sam.chunk", g_threadChunks.chunkOutN);
        };
    };//cycle over input chunks
    
    if (P->outFilterBySJoutStage!=1) {//not the first stage of the 2-stage mapping 
        if (P->outBAMunsorted) chunkOutBAMunsorted->unsortedFlush();
        if (P->outBAMcoord) chunkOutBAMcoord->coordFlush();
        if (chunkOutBAMquant!=NULL) chunkOutBAMquant->unsortedFlush();
        
        //the thread is finished mapping reads, concatenate the temp files into output files
        if (P->chimSegmentMin>0) {    
            chunkFstreamCat (RA->chunkOutChimSAM, P->inOut->outChimSAM, P->runThreadN>1, g_threadChunks.mutexOutChimSAM);
            chunkFstreamCat (RA->chunkOutChimJunction, P->inOut->outChimJunction, P->runThreadN>1, g_threadChunks.mutexOutChimJunction);
        };
        if (P->outReadsUnmapped=="Fastx" ) {
            for (uint ii=0;ii<P->readNmates;ii++) {    
                chunkFstreamCat (RA->chunkOutUnmappedReadsStream[ii],P->inOut->outUnmappedReadsStream[ii], P->runThreadN>1, g_threadChunks.mutexOutUnmappedFastx);
            };
        };
    };
    P->inOut->logMain << "Completed: thread #" <<iThread <<endl;
};