void parseSeqs(string filename, SeqSet &data) {
ifstream input;
SeqStream input_stream(openSeqFile(filename, input));
if(!input_stream.good()) {
throw("File stream is in a bad state");
}
string format = guessFormat(input_stream);
if(format == "fasta") {
try {
parseFasta(input_stream, filename, data);
} catch(...) {
input.close();
throw;
}
} else if(format == "fastq") {
try {
parseFastq(input_stream, filename, data);
} catch(...) {
input.close();
throw;
}
} else {
input.close();
throw("Unrecognized format");
}
input.close();
}
seqCache::seqCache(const char *filename, uint32 cachesize, bool verbose) {
_fb = openSeqFile(filename);
_idToGetNext = 0;
_allSequencesLoaded = false;
_reportLoading = verbose;
_cacheMap = 0L;
_cacheSize = 0;
_cacheNext = 0;
_cache = 0L;
setCacheSize(cachesize);
}
* 'Celera Assembler' (http://wgs-assembler.sourceforge.net)
* the 'kmer package' (http://kmer.sourceforge.net)
* both originally distributed by Applera Corporation under the GNU General
* Public License, version 2.
*
* Canu branched from Celera Assembler at its revision 4587.
* Canu branched from the kmer project at its revision 1994.
*
* Modifications by:
*
* File 'README.licenses' in the root directory of this distribution contains
* full conditions and disclaimers for each license.
*/
{
seqFile *SF = openSeqFile(argv[1]);
fprintf(stdout, "source '%s' of type '%s' has "uint32FMT" sequences.\n",
SF->getSourceName(), SF->getFileTypeName(), SF->getNumberOfSequences());
fprintf(stdout, "getSequenceLength() vs getSequence(full)\n");
{
char *h = 0L;
char *s = 0L;
uint32 hLen=0, hMax=0;
uint32 sLen=0, sMax=0;
for (uint32 sid=0; sid<SF->getNumberOfSequences(); sid++) {
SF->getSequence(sid, h, hLen, hMax, s, sLen, sMax);
if ((strlen(s) != SF->getSequenceLength(sid)) ||
