fixed negative coords, zero len intervals, Bed4 format error

src/utils/FileRecordTools/FileRecordMgr.cpp

@@ -44,10 +44,10 @@ FileRecordMgr::~FileRecordMgr(){
 
 bool FileRecordMgr::open(){
 
-	const QuickString &filename = _context->getInputFileName(_contextFileIdx);
-	_bufStreamMgr = new BufferedStreamMgr(filename);
+	_filename = _context->getInputFileName(_contextFileIdx);
+	_bufStreamMgr = new BufferedStreamMgr(_filename);
 	if (!_bufStreamMgr->init()) {
-		fprintf(stderr, "Error: unable to open file or unable to determine types for file %s.\n", filename.c_str());
+		cerr << "Error: unable to open file or unable to determine types for file " << _filename << endl;
 		delete _bufStreamMgr;
 		_bufStreamMgr = NULL;
 		exit(1);
@@ -56,7 +56,7 @@ bool FileRecordMgr::open(){
 	_fileType = _bufStreamMgr->getTypeChecker().getFileType();
 	_recordType = _bufStreamMgr->getTypeChecker().getRecordType();
 	if (_fileType == FileRecordTypeChecker::UNKNOWN_FILE_TYPE || _recordType == FileRecordTypeChecker::UNKNOWN_RECORD_TYPE) {
-		fprintf(stderr, "Error: Unable to determine type for file %s.\n", filename.c_str());
+		cerr << "Error: Unable to determine type for file " << _filename << endl;
 		delete _bufStreamMgr;
 		_bufStreamMgr = NULL;
 		exit(1);
@@ -64,11 +64,11 @@ bool FileRecordMgr::open(){
 	allocateFileReader();
 	_recordMgr = new RecordMgr(_recordType, _freeListBlockSize);
 
-	_fileReader->setFileName(filename.c_str());
+	_fileReader->setFileName(_filename.c_str());
 	_fileReader->setInputStream(_bufStreamMgr);
 	_fileReader->setContext(_context);
 	if (!_fileReader->open()) {
-		fprintf(stderr, "Error: Types determined but can't open file %s.\n", filename.c_str());
+		cerr << "Error: Types determined but can't open file " << _filename << endl;
 		delete _bufStreamMgr;
 		_bufStreamMgr = NULL;
 		exit(1);
@@ -125,6 +125,11 @@ Record *FileRecordMgr::allocateAndGetNextRecord()
 		return NULL;
 	}
 
+	if (!record->coordsValid()) {
+		cerr << "Error: Invalid record in file " << _filename << ". Record is " << endl << *record << endl;
+		exit(1);
+	}
+
 	//test for sorted order, if necessary.
 	if (_context->getSortedInput()) {
 		testInputSortOrder(record);
@@ -184,17 +189,16 @@ void FileRecordMgr::testInputSortOrder(Record *record)
 void FileRecordMgr::sortError(const Record *record, bool genomeFileError)
 {
 	if (genomeFileError) {
-		fprintf(stderr, "Error: Sorted input specified, but the file %s has the following record with a different sort order than the genomeFile %s:\n",
-				_context->getInputFileName(_contextFileIdx).c_str(), _context->getGenomeFile()->getGenomeFileName().c_str());
+		cerr << "Error: Sorted input specified, but the file " << _filename << " has the following record with a different sort order than the genomeFile " <<
+				_context->getGenomeFile()->getGenomeFileName() << endl;
 	} else {
-		fprintf(stderr, "Error: Sorted input specified, but the file %s has the following out of order record:\n", _context->getInputFileName(_contextFileIdx).c_str());
+		cerr << "Error: Sorted input specified, but the file " << _filename << " has the following out of order record" << endl;
 	}
-	QuickString errBuf;
-	record->print(errBuf);
-	fprintf(stderr, "%s\n", errBuf.c_str());
+	cerr << *record << endl;
 	exit(1);
-
 }
+
+
 void FileRecordMgr::deleteRecord(const Record *record) {
 	_recordMgr->deleteRecord(record);
 }

src/utils/FileRecordTools/FileRecordMgr.h

@@ -141,6 +141,7 @@ public:
 	void setFullBamFlags(bool flag) { _useFullBamTags = flag; }
 
 private:
+	QuickString _filename;
 	BufferedStreamMgr *_bufStreamMgr;
 	int _contextFileIdx;
 

src/utils/FileRecordTools/RecordOutputMgr.cpp

@@ -113,6 +113,7 @@ void RecordOutputMgr::printRecord(RecordKeyList &keyList, RecordKeyList *blockLi
 		_outBuf.append(_context->getHeader(_context->getQueryFileIdx()));
 		_context->setPrintHeader(false);
 	}
+	const_cast<Record *>(keyList.getKey())->undoZeroLength();
 
 	_currBlockList = blockList;
 
@@ -157,15 +158,67 @@ void RecordOutputMgr::printRecord(RecordKeyList &keyList, RecordKeyList *blockLi
 void RecordOutputMgr::reportOverlapDetail(const Record *keyRecord, const Record *hitRecord)
 {
 	//get the max start and min end as strings.
-	const QuickString &startStr = keyRecord->getStartPos() > hitRecord->getStartPos() ? keyRecord->getStartPosStr() : hitRecord->getStartPosStr();
-	const QuickString &endStr = keyRecord->getEndPos() < hitRecord->getEndPos() ? keyRecord->getEndPosStr() : hitRecord->getEndPosStr();
+	const_cast<Record *>(hitRecord)->undoZeroLength();
 
-	int maxStart = max(keyRecord->getStartPos(), hitRecord->getStartPos());
-	int minEnd = min(keyRecord->getEndPos(), hitRecord->getEndPos());
 
+	const QuickString *startStr = NULL;
+	const QuickString *endStr = NULL;
+	int maxStart = 0;
+	int minEnd = 0;
+
+	int keyStart = keyRecord->getStartPos();
+	int keyEnd = keyRecord->getEndPos();
+	int hitStart = hitRecord->getStartPos();
+	int hitEnd = hitRecord->getEndPos();
+
+	if (  keyStart>= hitStart) {
+		//the key start is after the hit start, but we need to check and make sure the hit end is at least after the keyStart.
+		//The reason for this is that, in some rare cases, such as both the key and hit having been zero length intervals,
+		//the normal process for intersection that allows us to simply report the maxStart and minEnd do not necessarily apply.
+		if (hitEnd >= keyStart) {
+			//this is ok. We have a normal intersection where the key comes after the hit.
+
+			maxStart = keyStart;
+			startStr = &(keyRecord->getStartPosStr());
+
+			minEnd = min(keyEnd, hitEnd);
+			endStr = keyRecord->getEndPos() < hitRecord->getEndPos() ? &(keyRecord->getEndPosStr()) : &(hitRecord->getEndPosStr());
+
+		} else {
+			//this is the weird case of not a "real" intersection. The keyStart is greater than the hitEnd. So just report the key as is.
+			maxStart = keyStart;
+			minEnd = keyEnd;
+			startStr = &(keyRecord->getStartPosStr());
+			endStr = &(keyRecord->getEndPosStr());
+		}
+
+	} else {
+		//all of the above, but backwards. keyStart is before hitStart.
+		if (keyEnd >= hitStart) {
+			//normal intersection, key first
+			maxStart = hitStart;
+			startStr = &(hitRecord->getStartPosStr());
+			minEnd = min(keyEnd, hitEnd);
+			endStr = keyRecord->getEndPos() < hitRecord->getEndPos() ? &(keyRecord->getEndPosStr()) : &(hitRecord->getEndPosStr());
+		} else {
+			//this is the weird case of not a "real" intersection. The hitStart is greater than the keyEnd. So just report the hit as is.
+			maxStart = hitStart;
+			minEnd = hitEnd;
+			startStr = &(hitRecord->getStartPosStr());
+			endStr = &(hitRecord->getEndPosStr());
+
+		}
+	}
+
+//	const QuickString &startStr = keyRecord->getStartPos() > hitRecord->getStartPos() ? keyRecord->getStartPosStr() : hitRecord->getStartPosStr();
+//	const QuickString &endStr = keyRecord->getEndPos() < hitRecord->getEndPos() ? keyRecord->getEndPosStr() : hitRecord->getEndPosStr();
+//
+//	int maxStart = max(keyRecord->getStartPos(), hitRecord->getStartPos());
+//	int minEnd = min(keyRecord->getEndPos(), hitRecord->getEndPos());
+//
 
 	if (!_context->getWriteA() && !_context->getWriteB() && !_context->getWriteOverlap() && !_context->getLeftJoin()) {
-		printKey(keyRecord, startStr, endStr);
+		printKey(keyRecord, *startStr, *endStr);
 		newline();
 	}
 	else if ((_context->getWriteA() && _context->getWriteB()) || _context->getLeftJoin()) {
@@ -179,7 +232,7 @@ void RecordOutputMgr::reportOverlapDetail(const Record *keyRecord, const Record
 		newline();
 	}
 	else if (_context->getWriteB()) {
-		printKey(keyRecord, startStr, endStr);
+		printKey(keyRecord, *startStr, *endStr);
 		tab();
 		hitRecord->print(_outBuf);
 		newline();

src/utils/FileRecordTools/Records/Bed4Interval.cpp

@@ -23,6 +23,7 @@ bool Bed4Interval::initFromFile(SingleLineDelimTextFileReader *fileReader)
 void Bed4Interval::print(QuickString &outBuf) const
 {
 	Bed3Interval::print(outBuf);
+	outBuf.append('\t');
 	outBuf.append(_name);
 }
 

src/utils/FileRecordTools/Records/Record.cpp

@@ -6,7 +6,8 @@ Record::Record()
 : _chrId(-1),
   _startPos(-1),
   _endPos(-1),
-  _strand(UNKNOWN)
+  _strand(UNKNOWN),
+  _zeroLength(false)
 {
 }
 
@@ -177,3 +178,37 @@ bool Record::sameChromIntersects(const Record *record,
 
 	return false;
 }
+
+bool Record::coordsValid() {
+	if (_startPos < 0 || _endPos < 0 || _endPos < _startPos) {
+		return false;
+	}
+	adjustZeroLength();
+	return true;
+}
+
+void Record::adjustZeroLength()
+{
+	if (_startPos == _endPos) {
+		_zeroLength = true;
+		_startPos--;
+		_endPos++;
+	}
+}
+
+void Record::undoZeroLength()
+{
+	if (_zeroLength) {
+		_startPos++;
+		_endPos--;
+		_zeroLength = false;
+	}
+}
+
+ostream &operator << (ostream &out, const Record &record)
+{
+	QuickString errBuf;
+	record.print(errBuf);
+	out << errBuf;
+	return out;
+}

src/utils/FileRecordTools/Records/Record.h

@@ -33,6 +33,8 @@ public:
 	virtual void print(QuickString &outBuf, int start, int end) const {}
 	virtual void print(QuickString &outBuf, const QuickString & start, const QuickString & end) const {}
 	virtual void printNull(QuickString &outBuf) const {}
+	friend ostream &operator << (ostream &out, const Record &record);
+
 	virtual const Record & operator=(const Record &);
 
 	virtual const QuickString &getChrName() const { return _chrName; }
@@ -54,13 +56,14 @@ public:
 	virtual const QuickString &getEndPosStr() const { return _endPosStr; }
 	virtual void setEndPosStr(const QuickString &str) { _endPosStr = str; }
 
+	virtual bool getZeroLength() const { return _zeroLength; }
+	virtual void setZeroLength(bool val) { _zeroLength = val; }
 
 	virtual strandType getStrand() const { return _strand; }
 	virtual void setStrand(strandType val) { _strand = val; }
 	virtual void setStrand(char val);
 	virtual char getStrandChar() const;
 
-	//And we have a similar problem with name and score
 	virtual const QuickString &getName() const { return _name; }
 	virtual void setName(const QuickString &chr) { _name = chr; }
 	virtual void setName(const string &chr) { _name = chr; }
@@ -73,6 +76,18 @@ public:
 
 	virtual FileRecordTypeChecker::RECORD_TYPE getType() const { return FileRecordTypeChecker::UNKNOWN_RECORD_TYPE; }
 
+	virtual bool coordsValid(); //test that no coords negative, end not less than start, check zeroLength (see below).
+
+	//Some files can have insertions of the form 2,2. If found this should translate to cover the base before and after,
+	//thus meaning the startPos is decremented and the endPos is incremented. This method will find and handle that case.
+	//Don't adjust the startPosStr and endPosStr strings because they aren't used in
+	//calculation. They're only used in output, and it would be slower to change them
+	//and then change them back.
+	virtual void adjustZeroLength();
+	virtual void undoZeroLength(); //change it back just before output;
+	virtual bool isZeroLength() const { return _zeroLength; }
+
+
 	virtual bool operator < (const Record &other) const;
 	virtual bool operator > (const Record &other) const;
 
@@ -110,6 +125,7 @@ protected:
 	QuickString _name;
 	QuickString _score;
 	strandType _strand;
+	bool _zeroLength;
 };