int main(int argc, char **argv)
{

    AjPList idlist;
    AjPList* fieldList = NULL;

    AjBool systemsort;
    AjBool cleanup;

    ajint blastv = 0;
    char dbtype  = '\0';

    ajuint maxindex;
    ajuint maxidlen = 0;
    ajuint maxlen;

    AjPStr version = NULL;
    AjPStr seqtype = NULL;

    AjPFile elistfile  = NULL;
    AjPFile* alistfile = NULL;

    AjPStr dbname   = NULL;
    AjPStr release  = NULL;
    AjPStr datestr  = NULL;
    AjPStr sortopt  = NULL;
    void **entryIds = NULL;

    AjBool usesrc = AJTRUE;

    AjPStr directory;
    AjPStr indexdir;
    AjPStr filename;
    AjPStr exclude;
    AjPStr curfilename = NULL;

    AjPStr idformat = NULL;

    EmbPEntry entry;

    PBlastDb db = NULL;

    ajuint idCount = 0;
    ajuint idDone;
    AjPList listTestFiles = NULL;
    void ** testFiles = NULL;
    ajuint nfiles;
    ajuint ifile;
    ajuint jfile;

    ajuint filesize;
    short recsize;
    ajuint maxfilelen = 20;
    char date[4] =
    {
	0,0,0,0
    };

    AjPStr tmpfname = NULL;
    AjPStr* fields  = NULL;

    AjPFile entFile = NULL;

    AjPStr* divfiles   = NULL;
    ajint* maxFieldLen = NULL;

    ajuint ifield  = 0;
    ajuint nfields = 0;

    AjPFile logfile = NULL;
    ajuint* countField = NULL;
    ajuint* fieldTot = NULL;
    ajuint idCountFile = 0;
    ajuint i = 0;

    embInit("dbiblast", argc, argv);

    idformat = ajStrNewC("NCBI");

    fields     = ajAcdGetList("fields");
    directory  = ajAcdGetDirectoryName("directory");
    indexdir   = ajAcdGetOutdirName("indexoutdir");
    filename   = ajAcdGetString("filenames");
    exclude    = ajAcdGetString("exclude");
    dbname     = ajAcdGetString("dbname");
    release    = ajAcdGetString("release");
    datestr    = ajAcdGetString("date");
    systemsort = ajAcdGetBoolean("systemsort");
    cleanup    = ajAcdGetBoolean("cleanup");
    sortopt    = ajAcdGetString("sortoptions");
    maxindex   = ajAcdGetInt("maxindex");
    version    = ajAcdGetListSingle("blastversion");
    seqtype    = ajAcdGetListSingle("seqtype");
    usesrc     = ajAcdGetBoolean("sourcefile");
    logfile    = ajAcdGetOutfile("outfile");

    while(fields[nfields])		/* array ends with a NULL */
	nfields++;

    if(nfields)
    {
	AJCNEW(maxFieldLen, nfields);
	AJCNEW0(countField, nfields);
	AJCNEW0(fieldTot, nfields);
	for(ifield=0; ifield < nfields; ifield++)
	    maxFieldLen[ifield] = (ajint) maxindex * -1;

	if(systemsort)
	    AJCNEW(alistfile, nfields);
	else
	{
	    AJCNEW(fieldList, nfields);
	    for(ifield=0; ifield < nfields; ifield++)
		fieldList[ifield] = ajListNew();
	}
    }
    
    if(ajStrMatchC(datestr, "00/00/00"))
	ajFmtPrintS(&datestr, "%D", ajTimeRefTodayFmt("dbindex"));

    ajStrRemoveWhite(&dbname);		/* used for temp filenames */
    embDbiDateSet(datestr, date);
    idlist = ajListNew();
    
    if(ajUtilGetBigendian())
	readReverse = ajFalse;
    else
	readReverse = ajTrue;
    
    ajStrToInt(version, &blastv);
    dbtype = ajStrGetCharFirst(seqtype);
    
    ajDebug("reading '%S/%S'\n", directory, filename);
    ajDebug("writing '%S/'\n", indexdir);
    
    listTestFiles = embDbiFileListExc(directory, filename, exclude);
    ajListSort(listTestFiles, ajStrVcmp);
    nfiles = ajListToarray(listTestFiles, &testFiles);
    if(!nfiles)
        ajDie("No input files in '%S' matched filename '%S'",
              directory, filename);
    
    embDbiLogHeader(logfile, dbname, release, datestr,
		     indexdir, maxindex);

    embDbiLogFields(logfile, fields, nfields);
    embDbiLogSource(logfile, directory, filename, exclude,
		    (AjPStr*) testFiles, nfiles);
    embDbiLogCmdline(logfile);

    AJCNEW0(divfiles, nfiles);
    
    /*
    ** process each input file, one at a time
    */
    
    jfile = 0;
    for(ifile=0; ifile < nfiles; ifile++)
    {
	curfilename = (AjPStr) testFiles[ifile];
	if(!dbiblast_blastopenlib(curfilename,
				  usesrc, blastv, dbtype, &db))
	    continue;	 /* could be the wrong file type with "*.*" */

	ajDebug("processing filename '%S' ...\n", curfilename);
	ajDebug("processing file '%S' ...\n", db->TFile->Name);


	ajStrAssignS(&divfiles[jfile], db->TFile->Name);
	ajFilenameTrimPath(&divfiles[jfile]);
	if(ajStrGetLen(divfiles[jfile]) >= maxfilelen)
	    maxfilelen = ajStrGetLen(divfiles[jfile]) + 1;

	if(systemsort)	 /* elistfile for entries, alist for fields */
	    elistfile = embDbiSortOpen(alistfile, jfile,
				       dbname, fields, nfields);

	idCountFile = 0;
	for(i=0;i<nfields;i++)
	    countField[i] = 0;
	while((entry=dbiblast_nextblastentry(db, jfile,
					     idformat, systemsort,
					     fields,
					     maxFieldLen,
					     &maxidlen, countField,
					     elistfile, alistfile)))
	{
	    idCountFile++;
	    if(!systemsort)	    /* save the entry data in lists */
	    {
		embDbiMemEntry(idlist, fieldList, nfields, entry, jfile);
	    }
	}
	idCount += idCountFile;
	if(systemsort)
	{
	    embDbiSortClose(&elistfile, alistfile, nfields);
	    /* lost the entry, so can't free it :-) */
	}

	embDbiLogFile(logfile, curfilename, idCountFile, fields,
		      countField, nfields);
	dbiblast_dbfree(&db);
	jfile++;
    }
    nfiles = jfile;
    
    /*
    ** write the division.lkp file
    */
    
    embDbiWriteDivision(indexdir, dbname, release, date,
			maxfilelen, nfiles, divfiles, NULL);
    
    /*
    ** Write the entryname.idx index
    */
    
    ajStrAssignC(&tmpfname, "entrynam.idx");
    entFile = ajFileNewOutNamePathS(tmpfname, indexdir);
    
    recsize = maxidlen+10;
    filesize = 300 + (idCount*(ajint)recsize);
    embDbiHeader(entFile, filesize, idCount, recsize, dbname, release, date);
    
    if(systemsort)
        idDone = embDbiSortWriteEntry(entFile, maxidlen,
				      dbname, nfiles, cleanup, sortopt);
    else			  /* save entries in entryIds array */
    {
        idDone = embDbiMemWriteEntry(entFile, maxidlen,
				     idlist, &entryIds);
	if(idDone != idCount)
	    ajFatal("Duplicates not allowed for in-memory processing");
    }
    
    embDbiHeaderSize(entFile, 300+(idDone*(ajint)recsize), idDone);
    ajFileClose(&entFile);
    
    /*
    ** Write the fields index files
    */
    
    for(ifield=0; ifield < nfields; ifield++)
    {

        if(maxindex)
	    maxlen = maxindex;
	else
	{
	    if(maxFieldLen[ifield] >= 0)
		maxlen = maxFieldLen[ifield];
	    else
		maxlen = - maxFieldLen[ifield];
	}

        if(systemsort)
	    fieldTot[ifield] = embDbiSortWriteFields(dbname, release,
						     date, indexdir,
						     fields[ifield], maxlen,
						     nfiles, idCount,
						     cleanup, sortopt);
	else
	    fieldTot[ifield] = embDbiMemWriteFields(dbname, release,
						    date, indexdir,
						    fields[ifield], maxlen,
						    fieldList[ifield],
						    entryIds);
    }
    
    embDbiLogFinal(logfile,maxindex, maxFieldLen, fields, fieldTot,
		   nfields, nfiles, idDone, idCount);

    if(systemsort)
	embDbiRmEntryFile(dbname, cleanup);
    
    ajListMap(idlist, embDbiEntryDelMap, NULL);
    ajListFree(&idlist);
    AJFREE(entryIds);

    ajStrDelarray(&fields);

    for(i=0;i<nfields;i++)
    {
	if(systemsort)
	{
	    ajFileClose(&alistfile[i]);
	}
	else
	{
	    ajListMap(fieldList[i], embDbiFieldDelMap, NULL);
	    ajListFree(&fieldList[i]);
	}
    }
    AJFREE(alistfile);
    AJFREE(fieldList);
    ajStrDel(&version);
    ajStrDel(&seqtype);
    ajFileClose(&elistfile);
    for(i=0;i<nfiles;i++)
    {
	ajStrDel(&divfiles[i]);
    }
    AJFREE(countField);
    AJFREE(fieldTot);

    ajStrDel(&dbname);
    ajStrDel(&release);
    ajStrDel(&datestr);
    ajStrDel(&sortopt);
    ajStrDel(&directory);
    ajStrDel(&indexdir);
    ajStrDel(&filename);
    ajStrDel(&exclude);
    ajStrDel(&idformat);
    ajStrDel(&tmpfname);

    AJFREE(maxFieldLen);

    ajFileClose(&logfile);

    ajListstrFreeData(&listTestFiles);

    ajStrDel(&t);
    ajStrDel(&id);
    ajStrDel(&acc);
    ajStrDel(&hline);
    ajStrDel(&tmpdes);
    ajStrDel(&tmpfd);
    ajStrDel(&tmpgi);
    ajStrDel(&tmpdb);
    ajStrDel(&tmpac);
    ajStrDel(&tmpsv);
    ajRegFree(&wrdexp);

    embDbiEntryDel(&dbiblastEntry);

    if(fdl)
    {
        for(i=0; i < nfields; i++)
            ajListFree(&fdl[i]);
        AJFREE(fdl);
    }

    for(i=0;i<nfiles;i++)
    {
        ajStrDel(&divfiles[i]);
    }
    AJFREE(divfiles);
    AJFREE(testFiles);

    embExit();

    return 0;
}
Example #2
0
int main(int argc, char **argv)
{
    AjPList idlist;
    AjPList* fieldList = NULL;

    AjBool systemsort;
    AjBool cleanup;

    ajuint maxindex;
    ajuint maxidlen = 0;
    ajuint maxlen;

    AjPFile elistfile  = NULL;
    AjPFile* alistfile = NULL;

    AjPStr dbname   = NULL;
    AjPStr release  = NULL;
    AjPStr datestr  = NULL;
    AjPStr sortopt  = NULL;
    void **entryIds = NULL;

    AjPStr directory;
    AjPStr indexdir;
    AjPStr filename;
    AjPStr exclude;
    AjPStr curfilename = NULL;

    AjPFile libr=NULL;
    AjPStr idformat = NULL;

    EmbPEntry entry;

    ajuint idtype  = 0;

    ajuint idCount = 0;
    ajuint idDone;
    AjPList listInputFiles = NULL;
    void ** inputFiles = NULL;
    ajuint nfiles;
    ajuint ifile;

    ajuint filesize;
    short recsize;
    ajuint maxfilelen = 20;
    char date[4] =
    {
	0,0,0,0
    };

    AjPStr tmpfname = NULL;
    AjPStr* fields  = NULL;

    AjPFile entFile  = NULL;

    AjPStr* divfiles   = NULL;
    AjPRegexp regIdExp      = NULL;
    ajint* maxFieldLen = NULL;

    ajuint ifield  = 0;
    ajuint nfields = 0;

    AjPFile logfile = NULL;
    ajuint* countField = NULL;
    ajuint* fieldTot = NULL;
    ajuint idCountFile = 0;
    ajuint i;

    embInit("dbifasta", argc, argv);

    idformat   = ajAcdGetListSingle("idformat");
    fields     = ajAcdGetList("fields");
    directory  = ajAcdGetDirectoryName("directory");
    indexdir   = ajAcdGetOutdirName("indexoutdir");
    filename   = ajAcdGetString("filenames");
    exclude    = ajAcdGetString("exclude");
    dbname     = ajAcdGetString("dbname");
    release    = ajAcdGetString("release");
    datestr    = ajAcdGetString("date");
    systemsort = ajAcdGetBoolean("systemsort");
    cleanup    = ajAcdGetBoolean("cleanup");
    sortopt    = ajAcdGetString("sortoptions");
    maxindex   = ajAcdGetInt("maxindex");
    logfile    = ajAcdGetOutfile("outfile");

    while(fields[nfields])		/* array ends with a NULL */
	nfields++;

    if(nfields)
    {
	AJCNEW(maxFieldLen, nfields);
	AJCNEW0(countField, nfields);
	AJCNEW0(fieldTot, nfields);
	for(ifield=0; ifield < nfields; ifield++)
	    maxFieldLen[ifield] = (ajint)maxindex * -1;

	if(systemsort)
	    AJCNEW(alistfile, nfields);
	else
	{
	    AJCNEW(fieldList, nfields);
	    for(ifield=0; ifield < nfields; ifield++)
		fieldList[ifield] = ajListNew();
	}
    }

    if(ajStrMatchC(datestr, "00/00/00"))
	ajFmtPrintS(&datestr, "%D", ajTimeRefTodayFmt("dbindex"));

    ajStrRemoveWhite(&dbname);		/* used for temp filenames */
    embDbiDateSet(datestr, date);
    idlist = ajListNew();

    regIdExp = dbifasta_getExpr(idformat, &idtype);

    ajDebug("reading '%S/%S'\n", directory, filename);
    ajDebug("writing '%S/'\n", indexdir);

    listInputFiles = embDbiFileListExc(directory, filename, exclude);
    ajListSort(listInputFiles, &ajStrVcmp);
    nfiles = (ajuint) ajListToarray(listInputFiles, &inputFiles);
    if(!nfiles)
        ajDie("No input files in '%S' matched filename '%S'",
              directory, filename);

    embDbiLogHeader(logfile, dbname, release, datestr,
		     indexdir, maxindex);

    embDbiLogFields(logfile, fields, nfields);
    embDbiLogSource(logfile, directory, filename, exclude,
		    (AjPStr*) inputFiles, nfiles);
    embDbiLogCmdline(logfile);

    AJCNEW0(divfiles, nfiles);

    /*
    ** process each input file, one at a time
    */

    for(ifile=0; ifile < nfiles; ifile++)
    {
	ajStrAssignS(&curfilename,(AjPStr) inputFiles[ifile]);
	embDbiFlatOpenlib(curfilename, &libr);
	ajFilenameTrimPath(&curfilename);
	if(ajStrGetLen(curfilename) >= maxfilelen)
	    maxfilelen = ajStrGetLen(curfilename) + 1;

	ajDebug("processing filename '%S' ...\n", curfilename);
	ajDebug("processing file '%F' ...\n", libr);
	ajStrAssignS(&divfiles[ifile], curfilename);

	if(systemsort)	 /* elistfile for entries, alist for fields */
	    elistfile = embDbiSortOpen(alistfile, ifile,
				       dbname, fields, nfields);

	idCountFile = 0;
	for(i=0;i<nfields;i++)
	    countField[i] = 0;
	while((entry=dbifasta_NextFlatEntry(libr, ifile,
					    regIdExp, idtype,
					    systemsort, fields, 
					    maxFieldLen, &maxidlen,
					    countField, elistfile,
					    alistfile)))
	{
	    idCountFile++;
	    if(!systemsort)	    /* save the entry data in lists */
		embDbiMemEntry(idlist, fieldList, nfields,
			       entry, ifile);
		entry = NULL;
	}
	idCount += idCountFile;
	if(systemsort)
	{
	    embDbiSortClose(&elistfile, alistfile, nfields);
	    AJFREE(entry);
	}
	else
	{
	    embDbiEntryDel(&dbifastaGEntry);
	}
	embDbiLogFile(logfile, curfilename, idCountFile, fields,
		      countField, nfields);
    }

    /*  write the division.lkp file */
    embDbiWriteDivision(indexdir, dbname, release, date,
			maxfilelen, nfiles, divfiles, NULL);

    /* Write the entryname.idx index */
    ajStrAssignC(&tmpfname, "entrynam.idx");
    entFile = ajFileNewOutNamePathS(tmpfname, indexdir);

    recsize = maxidlen+10;
    filesize = 300 + (idCount*(ajint)recsize);
    embDbiHeader(entFile, filesize, idCount, recsize, dbname, release, date);

    if(systemsort)
        idDone = embDbiSortWriteEntry(entFile, maxidlen,
				      dbname, nfiles, cleanup, sortopt);
    else			  /* save entries in entryIds array */
    {
        idDone = embDbiMemWriteEntry(entFile, maxidlen,
				     idlist, &entryIds);
	if(idDone != idCount)
	    ajFatal("Duplicates not allowed for in-memory processing");
    }

    embDbiHeaderSize(entFile, 300+(idDone*(ajint)recsize), idDone);
    ajFileClose(&entFile);

    /* Write the fields index files */
    for(ifield=0; ifield < nfields; ifield++)
    {
        if(maxindex)
	    maxlen = maxindex;
	else
	{
	    if(maxFieldLen[ifield] >= 0)
		maxlen = maxFieldLen[ifield];
	    else
		maxlen = - maxFieldLen[ifield];
	}

        if(systemsort)
	    fieldTot[ifield] = embDbiSortWriteFields(dbname, release,
						     date, indexdir,
						     fields[ifield], maxlen,
						     nfiles, idCount,
						     cleanup, sortopt);
	else
	    fieldTot[ifield] = embDbiMemWriteFields(dbname, release,
						    date, indexdir,
						    fields[ifield], maxlen,
						    fieldList[ifield],
						    entryIds);
    }

    embDbiLogFinal(logfile,maxindex, maxFieldLen, fields, fieldTot,
		   nfields, nfiles, idDone, idCount);

    if(systemsort)
	embDbiRmEntryFile(dbname, cleanup);

    ajStrDel(&idformat);
    ajStrDelarray(&fields);
    ajStrDel(&filename);
    ajStrDel(&exclude);
    ajStrDel(&dbname);
    ajStrDel(&release);
    ajStrDel(&datestr);
    ajStrDel(&sortopt);
    ajStrDel(&directory);
    ajStrDel(&indexdir);
    ajStrDel(&tmpfname);
    ajFileClose(&libr);
    ajFileClose(&logfile);

    for(i=0;i<nfields;i++)
    {
	if(systemsort)
	{
	    ajFileClose(&alistfile[i]);
	}
	else
	{
	    ajListMap(fieldList[i], &embDbiFieldDelMap, NULL);
	    ajListFree(&fieldList[i]);
	}
    }

    AJFREE(alistfile);
    AJFREE(fieldList);
    AJFREE(maxFieldLen);
    AJFREE(countField);
    AJFREE(fieldTot);

    for(i=0;i<nfiles;i++)
    {
	ajStrDel(&divfiles[i]);
    }

    AJFREE(divfiles);
    AJFREE(inputFiles);

    embDbiEntryDel(&dbifastaGEntry);

    ajStrDel(&dbifastaGRline);
    ajStrDel(&dbifastaGTmpId);

    if(dbifastaGFdl)
    {
	for(i=0; i < nfields; i++)
	    ajListFree(&dbifastaGFdl[i]);
	AJFREE(dbifastaGFdl);
    }

    ajListMap(idlist, &embDbiEntryDelMap, NULL);
    ajListFree(&idlist);
    ajListstrFreeData(&listInputFiles);
    AJFREE(entryIds);
    ajRegFree(&dbifastaGIdexp);
    ajRegFree(&dbifastaGWrdexp);
    ajRegFree(&regIdExp);

    ajStrDel(&dbifastaGTmpAc);
    ajStrDel(&dbifastaGTmpSv);
    ajStrDel(&dbifastaGTmpGi);
    ajStrDel(&dbifastaGTmpDb);
    ajStrDel(&dbifastaGTmpDes);
    ajStrDel(&dbifastaGTmpFd);
    ajStrDel(&curfilename);

    embExit();

    return 0;
}