QVariantMap MSAEditorState::saveState(MSAEditor* v) {
MSAEditorState ss;
ss.stateData[VIEW_ID]=MsaEditorFactory::ID;
MultipleSequenceAlignmentObject* msaObj = v->getMaObject();
if (msaObj) {
ss.setMSAObjectRef(GObjectReference(msaObj));
}
ss.setFont(v->getFont());
ss.setFirstPos(v->getFirstVisibleBase());
return ss.stateData;
}
QVariantMap TreeViewerState::saveState(TreeViewer* v) {
TreeViewerState ss;
ss.stateData[VIEW_ID] = TreeViewerFactory::ID;
PhyTreeObject* phyObj = v->getPhyObject();
if (phyObj) {
ss.setPhyObject(GObjectReference(phyObj));
}
ss.setHorizontalZoom(v->getHorizontalZoom());
ss.setVerticalZoom(v->getVerticalZoom());
ss.setTransform(v->getTransform());
ss.stateData.unite(v->getSettingsState());
return ss.stateData;
}
Task* ClustalOWorker::tick() {
if (input->hasMessage()) {
Message inputMessage = getMessageAndSetupScriptValues(input);
if (inputMessage.isEmpty()) {
output->transit();
return NULL;
}
cfg.numIterations=actor->getParameter(NUM_ITERATIONS)->getAttributeValue<int>(context);
cfg.maxGuidetreeIterations=actor->getParameter(MAX_GT_ITERATIONS)->getAttributeValue<int>(context);
cfg.maxHMMIterations=actor->getParameter(MAX_HMM_ITERATIONS)->getAttributeValue<int>(context);
cfg.setAutoOptions=actor->getParameter(SET_AUTO)->getAttributeValue<bool>(context);
cfg.numberOfProcessors=AppContext::getAppSettings()->getAppResourcePool()->getIdealThreadCount();
QString path=actor->getParameter(EXT_TOOL_PATH)->getAttributeValue<QString>(context);
if(QString::compare(path, "default", Qt::CaseInsensitive) != 0){
AppContext::getExternalToolRegistry()->getByName(ET_CLUSTALO)->setPath(path);
}
path=actor->getParameter(TMP_DIR_PATH)->getAttributeValue<QString>(context);
if(QString::compare(path, "default", Qt::CaseInsensitive) != 0){
AppContext::getAppSettings()->getUserAppsSettings()->setUserTemporaryDirPath(path);
}
QVariantMap qm = inputMessage.getData().toMap();
SharedDbiDataHandler msaId = qm.value(BaseSlots::MULTIPLE_ALIGNMENT_SLOT().getId()).value<SharedDbiDataHandler>();
QScopedPointer<MultipleSequenceAlignmentObject> msaObj(StorageUtils::getMsaObject(context->getDataStorage(), msaId));
SAFE_POINT(!msaObj.isNull(), "NULL MSA Object!", NULL);
const MultipleSequenceAlignment msa = msaObj->getMultipleAlignment();
if (msa->isEmpty()) {
algoLog.error(tr("An empty MSA '%1' has been supplied to ClustalO.").arg(msa->getName()));
return NULL;
}
ClustalOSupportTask* supportTask = new ClustalOSupportTask(msa, GObjectReference(), cfg);
supportTask->addListeners(createLogListeners());
Task *t = new NoFailTaskWrapper(supportTask);
connect(t, SIGNAL(si_stateChanged()), SLOT(sl_taskFinished()));
return t;
} else if (input->isEnded()) {
setDone();
output->setEnded();
}
return NULL;
}
SendSelectionDialog::SendSelectionDialog(const U2SequenceObject* dnaso, bool _isAminoSeq, QWidget *p)
: QDialog(p), translateToAmino(false), isAminoSeq(_isAminoSeq), extImported(false)
{
CreateAnnotationModel ca_m;
ca_m.hideAnnotationType = true;
ca_m.hideAnnotationName = true;
ca_m.hideLocation = true;
ca_m.sequenceObjectRef = GObjectReference(dnaso);
ca_m.sequenceLen = dnaso->getSequenceLength();
ca_c = new CreateAnnotationWidgetController(ca_m, this);
setupUi(this);
new HelpButton(this, buttonBox, "17467726");
buttonBox->button(QDialogButtonBox::Ok)->setText(tr("Search"));
buttonBox->button(QDialogButtonBox::Cancel)->setText(tr("Cancel"));
optionsTab->setCurrentIndex(0);
int idx = 2;
QWidget *wdgt;
wdgt = ca_c->getWidget();
layoutAnnotations->insertWidget(idx, wdgt);
matrixComboBox->addItems(ParametersLists::blastp_matrix);
setupDataBaseList();
setUpSettings();
megablastCheckBox->setEnabled(false);
alignComboBoxes();
connect( dataBase, SIGNAL(currentIndexChanged(int)), SLOT(sl_scriptSelected(int)) );
QPushButton* okButton = buttonBox->button(QDialogButtonBox::Ok);
QPushButton* cancelButton = buttonBox->button(QDialogButtonBox::Cancel);
connect( okButton, SIGNAL(clicked()), SLOT(sl_OK()) );
connect( cancelButton, SIGNAL(clicked()), SLOT(sl_Cancel()) );
connect( megablastCheckBox, SIGNAL(stateChanged(int)),SLOT(sl_megablastChecked(int)) );
connect( serviceComboBox,SIGNAL(currentIndexChanged(int)),SLOT(sl_serviceChanged(int)) );
sl_scriptSelected( 0 );
}
GObjectReference MSAEditorState::getMSAObjectRef() const {
return stateData.contains(MSA_OBJ) ? stateData[MSA_OBJ].value<GObjectReference>() : GObjectReference();
}
Task* ClustalWWorker::tick() {
if (input->hasMessage()) {
Message inputMessage = getMessageAndSetupScriptValues(input);
if (inputMessage.isEmpty()) {
output->transit();
return NULL;
}
cfg.gapOpenPenalty=actor->getParameter(GAP_OPEN_PENALTY)->getAttributeValue<float>(context);
cfg.gapExtenstionPenalty=actor->getParameter(GAP_EXT_PENALTY)->getAttributeValue<float>(context);
cfg.gapDist=actor->getParameter(GAP_DIST)->getAttributeValue<float>(context);
cfg.endGaps=actor->getParameter(END_GAPS)->getAttributeValue<bool>(context);
cfg.noHGaps=actor->getParameter(NO_HGAPS)->getAttributeValue<bool>(context);
cfg.noPGaps=actor->getParameter(NO_PGAPS)->getAttributeValue<bool>(context);
if(actor->getParameter(ITERATION)->getAttributeValue<int>(context) != 0){
if(actor->getParameter(ITERATION)->getAttributeValue<int>(context) == 1){
cfg.iterationType="TREE";
}else if(actor->getParameter(ITERATION)->getAttributeValue<int>(context) == 2){
cfg.iterationType="ALIGNMENT";
}
if(actor->getParameter(NUM_ITERATIONS)->getAttributeValue<int>(context) != 3){
cfg.numIterations=actor->getParameter(NUM_ITERATIONS)->getAttributeValue<int>(context);
}
}
if(actor->getParameter(MATRIX)->getAttributeValue<int>(context) == -1){
if(actor->getParameter(MATRIX)->getAttributeValue<int>(context) == 0){
cfg.matrix="IUB";
}else if(actor->getParameter(MATRIX)->getAttributeValue<int>(context) == 1){
cfg.matrix="CLUSTALW";
}else if(actor->getParameter(MATRIX)->getAttributeValue<int>(context) == 2){
cfg.matrix="BLOSUM";
}else if(actor->getParameter(MATRIX)->getAttributeValue<int>(context) == 3){
cfg.matrix="PAM";
}else if(actor->getParameter(MATRIX)->getAttributeValue<int>(context) == 4){
cfg.matrix="GONNET";
}else if(actor->getParameter(MATRIX)->getAttributeValue<int>(context) == 5){
cfg.matrix="ID";
}
}
QString path=actor->getParameter(EXT_TOOL_PATH)->getAttributeValue<QString>(context);
if(QString::compare(path, "default", Qt::CaseInsensitive) != 0){
AppContext::getExternalToolRegistry()->getByName(ET_CLUSTAL)->setPath(path);
}
path=actor->getParameter(TMP_DIR_PATH)->getAttributeValue<QString>(context);
if(QString::compare(path, "default", Qt::CaseInsensitive) != 0){
AppContext::getAppSettings()->getUserAppsSettings()->setUserTemporaryDirPath(path);
}
QVariantMap qm = inputMessage.getData().toMap();
SharedDbiDataHandler msaId = qm.value(BaseSlots::MULTIPLE_ALIGNMENT_SLOT().getId()).value<SharedDbiDataHandler>();
QScopedPointer<MultipleSequenceAlignmentObject> msaObj(StorageUtils::getMsaObject(context->getDataStorage(), msaId));
SAFE_POINT(!msaObj.isNull(), "NULL MSA Object!", NULL);
const MultipleSequenceAlignment msa = msaObj->getMultipleAlignment();
if (msa->isEmpty()) {
algoLog.error(tr("An empty MSA '%1' has been supplied to ClustalW.").arg(msa->getName()));
return NULL;
}
ClustalWSupportTask* supportTask = new ClustalWSupportTask(msa, GObjectReference(), cfg);
supportTask->addListeners(createLogListeners());
Task *t = new NoFailTaskWrapper(supportTask);
connect(t, SIGNAL(si_stateChanged()), SLOT(sl_taskFinished()));
return t;
} else if (input->isEnded()) {
setDone();
output->setEnded();
}
return NULL;
}
/**
* FASTQ format specification: http://maq.sourceforge.net/fastq.shtml
*/
static void load(IOAdapter* io, const U2DbiRef& dbiRef, const QVariantMap& hints, const GUrl& docUrl, QList<GObject*>& objects, U2OpStatus& os,
int gapSize, int predictedSize, QString& writeLockReason) {
DbiOperationsBlock opBlock(dbiRef, os);
CHECK_OP(os, );
Q_UNUSED(opBlock);
writeLockReason.clear();
bool merge = gapSize!=-1;
QByteArray sequence;
QByteArray qualityScores;
QStringList headers;
QSet<QString> uniqueNames;
QVector<U2Region> mergedMapping;
QByteArray gapSequence((merge ? gapSize : 0), 0);
sequence.reserve(predictedSize);
qualityScores.reserve(predictedSize);
// for lower case annotations
GObjectReference sequenceRef;
qint64 sequenceStart = 0;
U2SequenceImporter seqImporter(hints, true);
const QString folder = hints.value(DocumentFormat::DBI_FOLDER_HINT, U2ObjectDbi::ROOT_FOLDER).toString();
int seqNumber = 0;
int progressUpNum = 0;
const int objectsCountLimit = hints.contains(DocumentReadingMode_MaxObjectsInDoc) ? hints[DocumentReadingMode_MaxObjectsInDoc].toInt() : -1;
const bool settingsMakeUniqueName = !hints.value(DocumentReadingMode_DontMakeUniqueNames, false).toBool();
while (!os.isCoR()) {
//read header
QString sequenceName = readSequenceName(os, io, '@');
// check for eof while trying to read another FASTQ block
if (io->isEof()) {
break;
}
CHECK_OP_BREAK(os);
if(sequenceName.isEmpty()){
sequenceName = "Sequence";
}
if ((merge == false) || (seqNumber == 0)) {
QString objName = sequenceName;
if (settingsMakeUniqueName) {
objName = (merge) ? "Sequence" : TextUtils::variate(sequenceName, "_", uniqueNames);
objName.squeeze();
uniqueNames.insert(objName);
}
seqImporter.startSequence(dbiRef, folder, objName, false, os);
CHECK_OP_BREAK(os);
}
//read sequence
if (merge && sequence.length() > 0) {
seqImporter.addDefaultSymbolsBlock(gapSize,os);
sequenceStart += sequence.length();
sequenceStart+=gapSize;
CHECK_OP_BREAK(os);
}
sequence.clear();
readSequence(os, io, sequence);
MemoryLocker lSequence(os, qCeil(sequence.size()/(1000*1000)));
CHECK_OP_BREAK(os);
Q_UNUSED(lSequence);
seqImporter.addBlock(sequence.data(),sequence.length(),os);
CHECK_OP_BREAK(os);
QString qualSequenceName = readSequenceName(os, io, '+');
if (!qualSequenceName.isEmpty()) {
static const QString err = U2::FastqFormat::tr("Not a valid FASTQ file: %1, sequence name differs from quality scores name: %2 and %3");
CHECK_EXT_BREAK(sequenceName == qualSequenceName,
os.setError(err.arg(docUrl.getURLString()).arg(sequenceName).arg(qualSequenceName)));
}
// read qualities
qualityScores.clear();
readQuality(os, io, qualityScores, sequence.size());
CHECK_OP_BREAK(os);
static const QString err = U2::FastqFormat::tr("Not a valid FASTQ file: %1. Bad quality scores: inconsistent size.").arg(docUrl.getURLString());
CHECK_EXT_BREAK(sequence.length() == qualityScores.length(), os.setError(err));
seqNumber++;
progressUpNum++;
if (merge) {
headers.append(sequenceName);
mergedMapping.append(U2Region(sequenceStart, sequence.length() ));
}
else {
if (objectsCountLimit > 0 && objects.size() >= objectsCountLimit) {
os.setError(FastqFormat::tr("File \"%1\" contains too many sequences to be displayed. "
"However, you can process these data using instruments from the menu <i>Tools -> NGS data analysis</i> "
"or pipelines built with Workflow Designer.")
.arg(io->getURL().getURLString()));
break;
}
U2Sequence u2seq = seqImporter.finalizeSequenceAndValidate(os);
CHECK_OP_BREAK(os);
sequenceRef = GObjectReference(io->getURL().getURLString(), u2seq.visualName, GObjectTypes::SEQUENCE, U2EntityRef(dbiRef, u2seq.id));
U2SequenceObject* seqObj = new U2SequenceObject(u2seq.visualName, U2EntityRef(dbiRef, u2seq.id));
CHECK_EXT_BREAK(seqObj != NULL, os.setError("U2SequenceObject is NULL"));
seqObj->setQuality(DNAQuality(qualityScores));
objects << seqObj;
U1AnnotationUtils::addAnnotations(objects, seqImporter.getCaseAnnotations(), sequenceRef, NULL, hints);
}
if (PROGRESS_UPDATE_STEP == progressUpNum) {
progressUpNum = 0;
os.setProgress(io->getProgress());
}
}
CHECK_OP_EXT(os, qDeleteAll(objects); objects.clear(), );
bool emptyObjects = objects.isEmpty();
CHECK_EXT(!emptyObjects || merge, os.setError(Document::tr("Document is empty.")), );
SAFE_POINT(headers.size() == mergedMapping.size(), "headers <-> regions mapping failed!", );
if (!merge) {
return;
}
U2Sequence u2seq = seqImporter.finalizeSequenceAndValidate(os);
CHECK_OP(os,);
sequenceRef = GObjectReference(io->getURL().getURLString(), u2seq.visualName, GObjectTypes::SEQUENCE, U2EntityRef(dbiRef, u2seq.id));
U1AnnotationUtils::addAnnotations(objects, seqImporter.getCaseAnnotations(), sequenceRef, NULL, hints);
objects << new U2SequenceObject(u2seq.visualName, U2EntityRef(dbiRef, u2seq.id));
objects << DocumentFormatUtils::addAnnotationsForMergedU2Sequence(sequenceRef, dbiRef, headers, mergedMapping, hints);
if (headers.size() > 1) {
writeLockReason = DocumentFormat::MERGED_SEQ_LOCK;
}
}
void EMBLGenbankAbstractDocument::load(const U2DbiRef& dbiRef, IOAdapter* io, QList<GObject*>& objects, QVariantMap& fs, U2OpStatus& os, QString& writeLockReason) {
DbiOperationsBlock opBlock(dbiRef, os);
CHECK_OP(os, );
Q_UNUSED(opBlock);
writeLockReason.clear();
//get settings
int gapSize = qBound(-1, DocumentFormatUtils::getMergeGap(fs), 1000*1000);
bool merge = gapSize!=-1;
QScopedPointer<AnnotationTableObject> mergedAnnotations(NULL);
QStringList contigs;
QVector<U2Region> mergedMapping;
// Sequence loading is 'lazy', so, if there is no sequence, it won't be created and there is no need to remove it.
U2SequenceImporter seqImporter(fs, true);
const QString folder = fs.value(DBI_FOLDER_HINT, U2ObjectDbi::ROOT_FOLDER).toString();
QSet<QString> usedNames;
GObjectReference sequenceRef(GObjectReference(io->getURL().getURLString(), "", GObjectTypes::SEQUENCE));
QByteArray readBuffer(ParserState::LOCAL_READ_BUFFER_SIZE, '\0');
ParserState st(isNcbiLikeFormat() ? 12 : 5, io, NULL, os);
st.buff = readBuffer.data();
TmpDbiObjects dbiObjects(dbiRef, os);
int num_sequence = 0;
qint64 sequenceStart = 0;
int sequenceSize = 0;
int fullSequenceSize = 0;
const int objectsCountLimit = fs.contains(DocumentReadingMode_MaxObjectsInDoc) ? fs[DocumentReadingMode_MaxObjectsInDoc].toInt() : -1;
for (int i=0; !os.isCoR(); i++, ++num_sequence) {
if (objectsCountLimit > 0 && objects.size() >= objectsCountLimit) {
os.setError(EMBLGenbankAbstractDocument::tr("File \"%1\" contains too many sequences to be displayed. "
"However, you can process these data using instruments from the menu <i>Tools -> NGS data analysis</i> "
"or pipelines built with Workflow Designer.")
.arg(io->getURL().getURLString()));
break;
}
//TODO: reference to a local variable??? Such a pointer will become invalid
EMBLGenbankDataEntry data;
st.entry = &data;
if (num_sequence == 0 || merge == false){
seqImporter.startSequence(dbiRef, folder, "default sequence name", false, os); //change name and circularity after finalize method
CHECK_OP(os, );
}
sequenceSize = 0;
os.setDescription(tr("Reading entry header"));
int offset = 0;
if (merge && num_sequence > 0) {
offset = gapSize;
}
if (!readEntry(&st,seqImporter,sequenceSize,fullSequenceSize,merge,offset, os)) {
break;
}
if (merge && sequenceSize > 0 && num_sequence > 0) {
sequenceStart = fullSequenceSize - sequenceSize;
sequenceStart += gapSize;
fullSequenceSize += gapSize;
}
// tolerate blank lines between records
char ch;
bool b;
while ((b = st.io->getChar(&ch)) && (ch == '\n' || ch == '\r')){}
if (b) {
st.io->skip(-1);
}
AnnotationTableObject *annotationsObject = NULL;
if (data.hasAnnotationObjectFlag) {
QString annotationName = genObjectName(usedNames, data.name, data.tags, i+1, GObjectTypes::ANNOTATION_TABLE);
QVariantMap hints;
hints.insert(DBI_FOLDER_HINT, fs.value(DBI_FOLDER_HINT, U2ObjectDbi::ROOT_FOLDER));
if (Q_UNLIKELY(merge && NULL == mergedAnnotations)) {
mergedAnnotations.reset(new AnnotationTableObject(annotationName, dbiRef, hints));
}
annotationsObject = merge ? mergedAnnotations.data() : new AnnotationTableObject(annotationName, dbiRef, hints);
QStringList groupNames;
QMap<QString, QList<SharedAnnotationData> > groupName2Annotations;
for (int i = 0, n = data.features.size(); i < n; ++i) {
SharedAnnotationData &d = data.features[i];
if (!d->location->regions.isEmpty()) {
for (int i = 0, n = d->location->regions.size(); i < n; ++i) {
// for some reason larger numbers cannot be stored within rtree SQLite tables
if (d->location->regions[i].endPos() > 9223371036854775807LL) {
d->location->regions[i].length = 9223371036854775807LL - d->location->regions[i].startPos;
}
}
}
groupNames.clear();
d->removeAllQualifiers(GBFeatureUtils::QUALIFIER_GROUP, groupNames);
if (groupNames.isEmpty()) {
groupName2Annotations[""].append(d);
} else {
foreach(const QString &gName, groupNames) {
groupName2Annotations[gName].append(d);
}
}
CHECK_OP(os, );
}
GObjectReference TreeViewerState::getPhyObject() const {
return stateData.contains(PHY_OBJ) ? stateData[PHY_OBJ].value<GObjectReference>() : GObjectReference();
}
Task* RepeatWorker::tick() {
if (input->hasMessage()) {
Message inputMessage = getMessageAndSetupScriptValues(input);
if (inputMessage.isEmpty()) {
output->transit();
return NULL;
}
cfg.algo = RFAlgorithm(actor->getParameter(ALGO_ATTR)->getAttributeValue<int>(context));
cfg.minLen = actor->getParameter(LEN_ATTR)->getAttributeValue<int>(context);
if(actor->getParameter(USE_MIN_DISTANCE_ATTR)->getAttributeValue<bool>(context)){
cfg.minDist = actor->getParameter(MIN_DIST_ATTR)->getAttributeValue<int>(context);
}else{
cfg.minDist = 0;
}
if(actor->getParameter(USE_MAX_DISTANCE_ATTR)->getAttributeValue<bool>(context)){
cfg.maxDist = actor->getParameter(MAX_DIST_ATTR)->getAttributeValue<int>(context);
}else{
cfg.maxDist = INT_MAX;
}
int identity = actor->getParameter(IDENTITY_ATTR)->getAttributeValue<int>(context);
cfg.setIdentity(identity);
cfg.nThreads = actor->getParameter(THREADS_ATTR)->getAttributeValue<int>(context);
cfg.inverted = actor->getParameter(INVERT_ATTR)->getAttributeValue<bool>(context);
cfg.filter = RepeatsFilterAlgorithm(actor->getParameter(NESTED_ATTR)->getAttributeValue<int>(context));
cfg.excludeTandems = actor->getParameter(TANMEDS_ATTR)->getAttributeValue<bool>(context);
resultName = actor->getParameter(NAME_ATTR)->getAttributeValue<QString>(context);
if(resultName.isEmpty()){
resultName = "repeat_unit";
algoLog.error(tr("result name is empty, default name used"));
}
if(identity > 100 || identity < 0){
algoLog.error(tr("Incorrect value: identity value must be between 0 and 100"));
return new FailTask(tr("Incorrect value: identity value must be between 0 and 100"));
}
QVariantMap map = inputMessage.getData().toMap();
SharedDbiDataHandler seqId = map.value(BaseSlots::DNA_SEQUENCE_SLOT().getId()).value<SharedDbiDataHandler>();
QScopedPointer<U2SequenceObject> seqObj(StorageUtils::getSequenceObject(context->getDataStorage(), seqId));
if (seqObj.isNull()) {
return NULL;
}
U2OpStatusImpl os;
DNASequence seq = seqObj->getWholeSequence(os);
CHECK_OP(os, new FailTask(os.getError()));
if(cfg.minDist < 0){
algoLog.error(tr("Incorrect value: minimal distance must be greater then zero"));
return new FailTask(tr("Incorrect value: minimal distance must be greater then zero"));
}
if (!seq.alphabet->isNucleic()) {
QString err = tr("Sequence alphabet is not nucleic!");
return new FailTask(err);
}
Task* t = new FindRepeatsToAnnotationsTask(cfg, seq, resultName, QString(), "", GObjectReference());
connect(t, SIGNAL(si_stateChanged()), SLOT(sl_taskFinished()));
return t;
} else if (input->isEnded()) {
setDone();
output->setEnded();
}
return NULL;
}
static void load(IOAdapter* io, const U2DbiRef& dbiRef, const QVariantMap& fs, QList<GObject*>& objects,
int gapSize, QString& writeLockReason, U2OpStatus& os)
{
DbiOperationsBlock opBlock(dbiRef, os);
CHECK_OP(os, );
Q_UNUSED(opBlock);
static char fastaCommentStartChar = FastaFormat::FASTA_COMMENT_START_SYMBOL;
MemoryLocker memoryLocker(os, 1);
CHECK_OP(os, );
writeLockReason.clear();
QByteArray readBuff(DocumentFormat::READ_BUFF_SIZE + 1, 0);
char* buff = readBuff.data();
qint64 len = 0;
bool merge = gapSize != -1;
QStringList headers;
QSet<QString> uniqueNames;
QVector<U2Region> mergedMapping;
// for lower case annotations
GObjectReference sequenceRef;
//skip leading whites if present
bool lineOk = true;
static QBitArray nonWhites = ~TextUtils::WHITES;
io->readUntil(buff, DocumentFormat::READ_BUFF_SIZE, nonWhites, IOAdapter::Term_Exclude, &lineOk);
CHECK_EXT(!io->hasError(), os.setError(io->errorString()), );
U2SequenceImporter seqImporter(fs, true);
const QString folder = fs.value(DocumentFormat::DBI_FOLDER_HINT, U2ObjectDbi::ROOT_FOLDER).toString();
qint64 sequenceStart = 0;
int sequenceNumber = 0;
DbiConnection con(dbiRef, os);
bool headerReaded = false;
QStringList emptySeqNames;
const int objectsCountLimit = fs.contains(DocumentReadingMode_MaxObjectsInDoc) ? fs[DocumentReadingMode_MaxObjectsInDoc].toInt() : -1;
const bool settingsMakeUniqueName = !fs.value(DocumentReadingMode_DontMakeUniqueNames, false).toBool();
while (!os.isCoR()) {
//skip start comments and read header
if(!headerReaded){
do{
len = io->readLine(buff, DocumentFormat::READ_BUFF_SIZE);
CHECK_EXT(!io->hasError(), os.setError(io->errorString()), );
}while(buff[0] == fastaCommentStartChar && len > 0);
}
if (len == 0 && io->isEof()) { //end if stream
break;
}
CHECK_EXT(!io->hasError(), os.setError(io->errorString()), );
CHECK_EXT_BREAK(lineOk, os.setError(FastaFormat::tr("Line is too long")));
QString headerLine = QString(QByteArray(buff+1, len-1)).trimmed();
CHECK_EXT_BREAK(buff[0] == FastaFormat::FASTA_HEADER_START_SYMBOL, os.setError(FastaFormat::tr("First line is not a FASTA header")));
//read sequence
if (sequenceNumber == 0 || !merge) {
QString objName = headerLine;
if(objName.isEmpty()){
objName = "Sequence";
}
if (settingsMakeUniqueName) {
objName = (merge) ? "Sequence" : TextUtils::variate(objName, "_", uniqueNames);
objName.squeeze();
memoryLocker.tryAcquire(2*objName.size());
CHECK_OP_BREAK(os);
uniqueNames.insert(objName);
}
seqImporter.startSequence(os, dbiRef, folder, objName, false);
CHECK_OP_BREAK(os);
sequenceRef = GObjectReference(io->getURL().getURLString(), objName, GObjectTypes::SEQUENCE);
}
if (sequenceNumber >= 1 && merge) {
seqImporter.addDefaultSymbolsBlock(gapSize, os);
sequenceStart += gapSize;
CHECK_OP_BREAK(os);
}
int sequenceLen = 0;
while (!os.isCoR()) {
do {
len = io->readLine(buff, DocumentFormat::READ_BUFF_SIZE);
CHECK_EXT(!io->hasError(), os.setError(io->errorString()), );
} while (len <= 0 && !io->isEof());
CHECK_EXT(!io->hasError(), os.setError(io->errorString()), );
if (len <= 0 && io->isEof()) {
break;
}
CHECK_EXT(!io->hasError(), os.setError(io->errorString()), );
buff[len] = 0;
if(buff[0] != fastaCommentStartChar && buff[0] != FastaFormat::FASTA_HEADER_START_SYMBOL){
len = TextUtils::remove(buff, len, TextUtils::WHITES);
if(len > 0){
seqImporter.addBlock(buff, len, os);
sequenceLen += len;
}
}else if( buff[0] == FastaFormat::FASTA_HEADER_START_SYMBOL){
headerReaded = true;
break;
}
CHECK_OP_BREAK(os);
os.setProgress(io->getProgress());
}
if (merge) {
memoryLocker.tryAcquire(headerLine.size());
CHECK_OP_BREAK(os);
headers.append(headerLine);
mergedMapping.append(U2Region(sequenceStart, sequenceLen));
} else {
if (objectsCountLimit > 0 && objects.size() >= objectsCountLimit) {
os.setError(FastaFormat::tr("File \"%1\" contains too many sequences to be displayed. "
"However, you can process these data using instruments from the menu <i>Tools -> NGS data analysis</i> "
"or pipelines built with Workflow Designer.")
.arg(io->getURL().getURLString()));
break;
}
memoryLocker.tryAcquire(800);
CHECK_OP_BREAK(os);
U2Sequence seq = seqImporter.finalizeSequenceAndValidate(os);
if (os.hasError() && os.getError() == U2SequenceImporter::EMPTY_SEQUENCE_ERROR) {
os.setError("");
emptySeqNames << headerLine;
continue;
}
sequenceRef.entityRef = U2EntityRef(dbiRef, seq.id);
//TODO parse header
U2StringAttribute attr(seq.id, DNAInfo::FASTA_HDR, headerLine);
con.dbi->getAttributeDbi()->createStringAttribute(attr, os);
CHECK_OP_BREAK(os);
objects << new U2SequenceObject(seq.visualName, U2EntityRef(dbiRef, seq.id));
CHECK_OP_BREAK(os);
U1AnnotationUtils::addAnnotations(objects, seqImporter.getCaseAnnotations(), sequenceRef, NULL, fs);
}
sequenceStart += sequenceLen;
sequenceNumber++;
ioLog.trace(QString("Sequence #%1 is processed").arg(sequenceNumber));
}
CHECK_OP_EXT(os, qDeleteAll(objects); objects.clear(), );
CHECK_EXT(!objects.isEmpty() || merge, os.setError(Document::tr("Document is empty.")), );
SAFE_POINT(headers.size() == mergedMapping.size(), "headers <-> regions mapping failed!", );
ioLog.trace("All sequences are processed");
if (!emptySeqNames.isEmpty()) {
QString warningMessage;
warningMessage.append(FastaFormat::tr("Loaded sequences: %1.\n").arg(sequenceNumber));
warningMessage.append(FastaFormat::tr("Skipped sequences: %1.\n").arg(emptySeqNames.size()));
warningMessage.append(FastaFormat::tr("The following sequences are empty:\n%1").arg(emptySeqNames.join(",\n")));
os.addWarning(warningMessage);
}
if (!merge) {
return;
}
U2Sequence seq = seqImporter.finalizeSequenceAndValidate(os);
CHECK_OP(os, );
sequenceRef.entityRef = U2EntityRef(dbiRef, seq.id);
U1AnnotationUtils::addAnnotations(objects, seqImporter.getCaseAnnotations(), sequenceRef, NULL, fs);
objects << new U2SequenceObject(seq.visualName, U2EntityRef(dbiRef, seq.id));
objects << DocumentFormatUtils::addAnnotationsForMergedU2Sequence( sequenceRef, dbiRef, headers, mergedMapping, fs );
if (headers.size() > 1) {
writeLockReason = QObject::tr("Document sequences were merged");
}
}
