bool GroupBy::init()
{
Tokenizer groupColsTokens;
groupColsTokens.tokenize(upCast(_context)->getGroupCols(), ',');
int numElems = groupColsTokens.getNumValidElems();
for (int i=0; i < numElems; i++) {
//if the item is a range, such as 3-5,
//must split that as well.
const QuickString &elem = groupColsTokens.getElem(i);
if (strchr(elem.c_str(), '-')) {
Tokenizer rangeElems;
rangeElems.tokenize(elem, '-');
int startNum = str2chrPos(rangeElems.getElem(0));
int endNum = str2chrPos(rangeElems.getElem(1));
for (int i=startNum; i <= endNum; i++) {
_groupCols.push_back(i);
}
} else {
_groupCols.push_back(str2chrPos(elem));
}
}
_queryFRM = _context->getFile(0);
_prevFields.resize(_groupCols.size());
_prevRecord = getNextRecord();
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;
string chrName;
while (!genFile.eof()) {
sLine.clear();
chrSize = 0;
chrName.clear();
getline(genFile, sLine);
int numFields = fieldTokens.tokenize(sLine.c_str());
// allow use of .fai files.
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();
}
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);
}
_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 KeyListOps::isValidColumnOps(FileRecordMgr *dbFile) {
//get the strings from context containing the comma-delimited lists of columns
//and operations. Split both of these into vectors. Get the operation code
//for each operation string. Finally, make a vector of pairs, where the first
//member of each pair is a column number, and the second member is the code for the
//operation to perform on that column.
Tokenizer colTokens;
Tokenizer opsTokens;
int numCols = colTokens.tokenize(_columns, ',');
int numOps = opsTokens.tokenize(_operations, ',');
if (numOps < 1 || numCols < 1) {
cerr << endl << "*****" << endl
<< "***** ERROR: There must be at least one column and at least one operation named." << endl;
return false;
}
if (numOps > 1 && numCols > 1 && numCols != numOps) {
cerr << endl << "*****" << endl
<< "***** ERROR: There are " << numCols <<" columns given, but there are " << numOps << " operations." << endl;
cerr << "\tPlease provide either a single operation that will be applied to all listed columns, " << endl;
cerr << "\ta single column to which all operations will be applied," << endl;
cerr << "\tor an operation for each column." << endl;
return false;
}
int loop = max(numCols, numOps);
// If there is only one column, all ops are performed on it.
// Otherwise, if there is only op, it is performed on all columns.
// Besides that, ops are performed on columns in their respective
// ordering.
for (int i=0; i < loop; i++) {
int col = str2chrPos(colTokens.getElem(numCols > 1 ? i : 0));
//check that the column number is valid
if (col < 1 || col > dbFile->getNumFields()) {
cerr << endl << "*****" << endl << "***** ERROR: Requested column " << col << ", but database file "
<< dbFile->getFileName() << " only has fields 1 - " << dbFile->getNumFields() << "." << endl;
return false;
}
const QuickString &operation = opsTokens.getElem(numOps > 1 ? i : 0);
OP_TYPES opCode = getOpCode(operation);
if (opCode == INVALID) {
cerr << endl << "*****" << endl
<< "***** ERROR: " << operation << " is not a valid operation. " << endl;
return false;
}
_colOps.push_back(pair<int, OP_TYPES>(col, opCode));
}
//lastly, if the file is BAM, and they asked for column 2, which is the
//flags field, then for now we have to throw an error, as the flag field
//is currently not supported.
if (_dbFileType == FileRecordTypeChecker::BAM_FILE_TYPE) {
//also, tell the methods class we're dealing with BAM.
_methods.setIsBam(true);
for (size_t i = 0; i < _colOps.size(); i++) {
if (_colOps[i].first == 2) {
cerr << endl << "*****" << endl << "***** ERROR: Requested column 2 of a BAM file, which is the Flags field." << endl;
cerr << " We currently do not support this, but may in future versions." << endl;
return false;
}
}
}
return true;
}
