bool MergeFile::merge()
{
RecordKeyVector hitSet;
FileRecordMgr *frm = _context->getFile(0);
while (!frm->eof()) {
Record *key = frm->getNextRecord(&hitSet);
if (key == NULL) continue;
_recordOutputMgr->printRecord(hitSet.getKey(), _context->getColumnOpsVal(hitSet));
}
return true;
}
FileRecordMgr *ContextBase::getNewFRM(const string &filename, int fileIdx) {
if (_useMergedIntervals) {
FileRecordMergeMgr *frm = new FileRecordMergeMgr(filename);
frm->setStrandType(_desiredStrand);
frm->setMaxDistance(_maxDistance);
frm->setFileIdx(fileIdx);
return frm;
} else {
FileRecordMgr *frm = new FileRecordMgr(filename);
frm->setFileIdx(fileIdx);
return frm;
}
}
bool ContextBase::isValidState()
{
if (!openFiles()) {
return false;
}
if (!cmdArgsValid()) {
return false;
}
if (!determineOutputType()) {
return false;
}
if (_program != GROUP_BY &&
_files[0]->getRecordType() == FileRecordTypeChecker::NO_POS_PLUS_RECORD_TYPE)
{
_errorMsg = "ERROR: file ";
_errorMsg.append(_files[0]->getFileName());
_errorMsg.append(" has non positional records, which are only valid for \n");
_errorMsg.append(" the groupBy tool. Perhaps you are using a header");
_errorMsg.append(" line(s) that starts with \n");
_errorMsg.append(" something other than \"#\", \"chrom\", or \"chr\" (any case)?");
return false;
}
if (getObeySplits()) {
_splitBlockInfo = new BlockMgr(_overlapFractionA, _reciprocalFraction);
}
if (hasColumnOpsMethods()) {
if (hasIntersectMethods()) {
for (int i=0; i < (int)_dbFileIdxs.size(); i++) {
FileRecordMgr *dbFile = getFile(_dbFileIdxs[i]);
_keyListOps->setDBfileType(dbFile->getFileType());
if (!_keyListOps->isValidColumnOps(dbFile)) {
return false;
}
}
} else {
FileRecordMgr *dbFile = getFile(0);
_keyListOps->setDBfileType(dbFile->getFileType());
if (!_keyListOps->isValidColumnOps(dbFile)) {
return false;
}
}
//if user specified a precision, pass it to
//keyList ops
if (_reportPrecision != -1) {
_keyListOps->setPrecision(_reportPrecision);
}
}
return true;
}
const string &ContextGroupBy::getDefaultHeader() {
//groupBy does not support multiple databases.
FileRecordMgr *frm = _files[0];
int numFields = frm->getNumFields();
_defaultHeader.clear();
ostringstream s;
for (int i=1; i <= numFields; i++) {
s << "col_";
s << i;
s << "\t";
}
_defaultHeader.append(s.str());
//change last tab into newline
_defaultHeader[_defaultHeader.size()-1] = '\n';
return _defaultHeader;
}
bool ContextBase::isValidState()
{
if (!openFiles()) {
return false;
}
if (!cmdArgsValid()) {
return false;
}
if (!determineOutputType()) {
return false;
}
if (hasColumnOpsMethods()) {
FileRecordMgr *dbFile = getFile(hasIntersectMethods() ? _databaseFileIdx : 0);
_keyListOps->setDBfileType(dbFile->getFileType());
if (!_keyListOps->isValidColumnOps(dbFile)) {
return false;
}
}
return true;
}
bool ContextBase::isValidState()
{
if (!openFiles()) {
return false;
}
if (!cmdArgsValid()) {
return false;
}
if (!determineOutputType()) {
return false;
}
if (getObeySplits()) {
_splitBlockInfo = new BlockMgr(_overlapFraction, _reciprocal);
}
if (hasColumnOpsMethods()) {
if (hasIntersectMethods()) {
for (int i=0; i < (int)_dbFileIdxs.size(); i++) {
FileRecordMgr *dbFile = getFile(_dbFileIdxs[i]);
_keyListOps->setDBfileType(dbFile->getFileType());
if (!_keyListOps->isValidColumnOps(dbFile)) {
return false;
}
}
} else {
FileRecordMgr *dbFile = getFile(0);
_keyListOps->setDBfileType(dbFile->getFileType());
if (!_keyListOps->isValidColumnOps(dbFile)) {
return false;
}
}
//if user specified a precision, pass it to
//keyList ops
if (_reportPrecision != -1) {
_keyListOps->setPrecision(_reportPrecision);
}
}
return true;
}
bool ContextBase::openFiles() {
//Make a vector of FileRecordMgr objects by going through the vector
//of filenames and opening each one.
if (_allFilesOpened) {
return true;
}
_files.resize(_fileNames.size());
for (int i = 0; i < (int)_fileNames.size(); i++) {
FileRecordMgr *frm = new FileRecordMgr(_fileNames[i], _sortedInput);
if (hasGenomeFile()) {
frm->setGenomeFile(_genomeFile);
}
frm->setFullBamFlags(_useFullBamTags);
if (!frm->open()) {
return false;
}
_files[i] = frm;
}
_allFilesOpened = true;
return true;
}
bool ContextBase::openFiles() {
//Make a vector of FileRecordMgr objects by going through the vector
//of filenames and opening each one.
if (_allFilesOpened) {
return true;
}
if (_fileNames.size() == 0) {
//No input was specified. Error and exit.
_errorMsg += "\n***** ERROR: No input file given. Exiting. *****";
return false;
}
_files.resize(_fileNames.size());
for (int i = 0; i < (int)_fileNames.size(); i++) {
FileRecordMgr *frm = getNewFRM(_fileNames[i], i);
if (hasGenomeFile()) {
frm->setGenomeFile(_genomeFile);
}
//If we're going to do column operations, and an input file
// is BAM, we'll need the full flags.
if (hasColumnOpsMethods()) {
setUseFullBamTags(true);
}
frm->setFullBamFlags(_useFullBamTags);
frm->setIsSorted(_sortedInput);
frm->setIoBufSize(_ioBufSize);
frm->setIsGroupBy(_program == GROUP_BY);
if (!frm->open(_inheader)) {
return false;
}
if (_noEnforceCoordSort) {
frm->setNoEnforceCoodSort(true);
}
_files[i] = frm;
}
_allFilesOpened = true;
return true;
}
bool FileIntersect::processUnsortedFiles()
{
const QuickString &databaseFilename = _context->getDatabaseFileName();
BinTree *binTree = new BinTree(_context->getDatabaseFileIdx(), _context);
FileRecordMgr *queryFRM = new FileRecordMgr(_context->getQueryFileIdx(), _context);
if (!queryFRM->open()) {
return false;
}
if (!binTree->loadDB()) {
fprintf(stderr, "Error: Unable to load database file %s.\n", databaseFilename.c_str());
delete binTree;
exit(1);
}
_context->determineOutputType();
_recordOutputMgr->init(_context);
while (!queryFRM->eof()) {
Record *queryRecord = queryFRM->allocateAndGetNextRecord();
if (queryRecord == NULL) {
continue;
}
RecordKeyList hitSet(queryRecord);
binTree->getHits(queryRecord, hitSet);
if (_context->getObeySplits()) {
RecordKeyList keySet(hitSet.getKey());
RecordKeyList resultSet;
_blockMgr->findBlockedOverlaps(keySet, hitSet, resultSet);
processHits(resultSet);
} else {
processHits(hitSet);
}
queryFRM->deleteRecord(queryRecord);
}
queryFRM->close();
//clean up.
delete queryFRM;
delete binTree;
return true;
}
