bool BufferedStreamMgr::getLine(QuickString &line)
{
line.clear();
if (_mainBufCurrStartPos >= _mainBufCurrLen) {
if (!readFileChunk()) {
_eof = true;
return false;
}
}
while (1) {
int searchPos = _mainBufCurrStartPos;
while (searchPos < _mainBufCurrLen && _mainBuf[searchPos] != '\n') {
searchPos++;
}
line.append((char *)_mainBuf + _mainBufCurrStartPos, searchPos - _mainBufCurrStartPos);
_mainBufCurrStartPos = searchPos +1;
if (searchPos == _mainBufCurrLen) { //hit end of buffer, but no newline yet
if (!readFileChunk()) { //hit eof
return true;
}
} else if (_mainBuf[searchPos] == '\n') {
return true;
}
}
}
bool BufferedStreamMgr::getTypeData()
{
QuickString currScanBuffer;
_inputStreamMgr->getSavedData(currScanBuffer);
do {
if (!_typeChecker.scanBuffer(currScanBuffer.c_str(), currScanBuffer.size()) && !_typeChecker.needsMoreData()) {
return false;
} else if (_typeChecker.needsMoreData()) {
_inputStreamMgr->populateScanBuffer();
currScanBuffer.clear();
_inputStreamMgr->getSavedData(currScanBuffer);
}
} while (_typeChecker.needsMoreData());
_inputStreamMgr->reset();
return true;
}
void NewGenomeFile::loadGenomeFileIntoMap() {
ifstream genFile(_genomeFileName.c_str());
if (!genFile.good()) {
cerr << "Error: Can't open genome file" << _genomeFileName << "Exiting..." << endl;
exit(1);
}
string sLine;
Tokenizer fieldTokens;
CHRPOS chrSize = 0;
QuickString chrName;
while (!genFile.eof()) {
sLine.clear();
chrSize = 0;
chrName.clear();
getline(genFile, sLine);
int numFields = fieldTokens.tokenize(sLine.c_str());
if (numFields != 2) {
continue;
}
chrName = fieldTokens.getElem(0);
chrSize = str2chrPos(fieldTokens.getElem(1));
_maxId++;
_chromSizeIds[chrName] = pair<CHRPOS, int>(chrSize, _maxId);
_startOffsets.push_back(_genomeLength);
_genomeLength += chrSize;
_chromList.push_back(chrName);
}
if (_maxId == -1) {
cerr << "Error: The genome file " << _genomeFileName << " has no valid entries. Exiting." << endl;
exit(1);
}
// Special: BAM files can have unmapped reads, which show as no chromosome, or an empty chrom string.
// Add in an empty chrom so these don't error.
_maxId++;
_chromSizeIds[""] = pair<CHRPOS, int>(0, _maxId);
_chromList.push_back("");
_startOffsets.push_back(_genomeLength); //insert the final length as the last element
//to help with the lower_bound call in the projectOnGenome method.
genFile.close();
}
bool BufferedStreamMgr::getLine(QuickString &line)
{
line.clear();
if (_mainBufCurrStartPos >= _mainBufCurrLen) {
if (!readFileChunk()) {
_eof = true;
return false;
}
}
bool retVal = true;
while (1) {
int searchPos = _mainBufCurrStartPos;
while (searchPos < _mainBufCurrLen && _mainBuf[searchPos] != '\n') {
searchPos++;
}
line.append((char *)_mainBuf + _mainBufCurrStartPos, searchPos - _mainBufCurrStartPos);
_mainBufCurrStartPos = searchPos +1;
if (searchPos == _mainBufCurrLen) { //hit end of buffer, but no newline yet
if (!readFileChunk()) { //hit eof
retVal = true;
break;
}
} else if (_mainBuf[searchPos] == '\n') {
retVal = true;
break;
}
}
//strip any whitespace characters, such as DOS newline characters or extra tabs,
//from the end of the line
int lastPos = line.size();
while (isspace(line[lastPos-1])) lastPos--;
line.resize(lastPos);
return retVal;
}