Added new -iobuf option and unit tests.

src/utils/Contexts/ContextBase.cpp

@@ -8,6 +8,7 @@
 #include "ContextBase.h"
 #include <unistd.h>
 #include <sys/types.h>
+#include <cctype>
 
 ContextBase::ContextBase()
 :
@@ -21,6 +22,7 @@ ContextBase::ContextBase()
   _obeySplits(false),
   _uncompressedBam(false),
   _useBufferedOutput(true),
+  _ioBufSize(0),
   _anyHit(false),
   _noHit(false),
   _writeA(false),
@@ -167,6 +169,9 @@ bool ContextBase::parseCmdArgs(int argc, char **argv, int skipFirstArgs) {
         else if (strcmp(_argv[_i], "-nobuf") == 0) {
 			if (!handle_nobuf()) return false;
         }
+        else if (strcmp(_argv[_i], "-iobuf") == 0) {
+			if (!handle_iobuf()) return false;
+        }
         else if (strcmp(_argv[_i], "-header") == 0) {
 			if (!handle_header()) return false;
         }
@@ -257,6 +262,7 @@ bool ContextBase::openFiles() {
 		}
 		frm->setFullBamFlags(_useFullBamTags);
 		frm->setIsSorted(_sortedInput);
+		frm->setIoBufSize(_ioBufSize);
 		if (!frm->open()) {
 			return false;
 		}
@@ -369,6 +375,19 @@ bool ContextBase::handle_nobuf()
 	return true;
 }
 
+bool ContextBase::handle_iobuf()
+{
+	if (_argc <= _i+1) {
+		_errorMsg = "\n***** ERROR: -iobuf option given, but size of input buffer not specified. *****";
+		return false;
+	}
+	if (!parseIoBufSize(_argv[_i + 1])) return false;
+	markUsed(_i - _skipFirstArgs);
+	_i++;
+	markUsed(_i - _skipFirstArgs);
+	return true;
+}
+
 bool ContextBase::handle_seed()
 {
 	if (_argc <= _i+1) {
@@ -499,3 +518,40 @@ FileRecordMgr *ContextBase::getNewFRM(const QuickString &filename) {
 		return frm;
 	}
 }
+
+bool ContextBase::parseIoBufSize(QuickString bufStr)
+{
+	char lastChar = bufStr[bufStr.size()-1];
+	int multiplier = 1;
+	if (!isdigit(lastChar)) {
+		switch (lastChar) {
+		case 'K':
+			multiplier = 1 << 10;
+			break;
+		case 'M':
+			multiplier = 1 << 20;
+			break;
+		case 'G':
+			multiplier = 1 << 30;
+			break;
+		default:
+			_errorMsg = "\n***** ERROR: Unrecognized memory buffer size suffix \'";
+			_errorMsg += lastChar;
+			_errorMsg += "\' given. *****";
+			return false;
+			break;
+		}
+		//lop off suffix character
+		bufStr.resize(bufStr.size()-1);
+	}
+	if (!isNumeric(bufStr)) {
+		_errorMsg = "\n***** ERROR: argument passed to -iobuf is not numeric. *****";
+		return false;
+	}
+	_ioBufSize = str2chrPos(bufStr) * multiplier;
+	if (_ioBufSize < MIN_ALLOWED_BUF_SIZE) {
+		_errorMsg = "\n***** ERROR: specified buffer size is too small. *****";
+		return false;
+	}
+	return true;
+}

src/utils/Contexts/ContextBase.h

@@ -176,6 +176,7 @@ protected:
     bool _obeySplits;
     bool _uncompressedBam;
     bool _useBufferedOutput;
+    int _ioBufSize;
 
 	bool _anyHit;
     bool _noHit;
@@ -234,6 +235,8 @@ protected:
 	char **_argv;
 	int _i;
 
+	static const int MIN_ALLOWED_BUF_SIZE = 8;
+
 	virtual bool handle_bed();
 	virtual bool handle_fbam();
 	virtual bool handle_g();
@@ -242,6 +245,8 @@ protected:
 	virtual bool handle_i();
 	virtual bool handle_n();
 	virtual bool handle_nobuf();
+	virtual bool handle_iobuf();
+
 	virtual bool handle_seed();
 	virtual bool handle_split();
 	virtual bool handle_sorted();
@@ -251,6 +256,9 @@ protected:
 	virtual bool handle_o();
 	virtual bool handle_null();
 	virtual bool handle_delim();
+
+	bool parseIoBufSize(QuickString bufStr);
+
 };
 
 #endif /* CONTEXTBASE_H_ */

src/utils/FileRecordTools/FileReaders/BufferedStreamMgr.cpp

@@ -17,7 +17,7 @@ BufferedStreamMgr::BufferedStreamMgr(const QuickString &filename)
   	_mainBufCurrStartPos(0),
   	_mainBufCurrLen(0),
   	_eof(false),
-  	_useBufSize(0),
+  	_useBufSize(DEFAULT_MAIN_BUF_READ_SIZE),
   	_streamFinished(false)
 {
 
@@ -51,8 +51,6 @@ bool BufferedStreamMgr::init()
 	}
 	if (_inputStreamMgr->isGzipped()) {
 		_useBufSize = GZIP_LINE_BUF_SIZE;
-	} else {
-		_useBufSize =  MAIN_BUF_READ_SIZE;
 	}
 
 	size_t trueBufSize = max(_useBufSize, (int)_currScanBuffer.size());

src/utils/FileRecordTools/FileReaders/BufferedStreamMgr.h

@@ -27,7 +27,8 @@ public:
 	bool eof() const { return _eof; }
 	bool getLine(QuickString &line);
 	BamTools::BamReader *getBamReader() { return _inputStreamMgr->getBamReader(); }
-
+	static const int DEFAULT_MAIN_BUF_READ_SIZE = 1023;
+	void setIoBufSize(int val) { _useBufSize = val; }
 private:
 	InputStreamMgr *_inputStreamMgr;
 	typedef unsigned char bufType;
@@ -42,8 +43,8 @@ private:
 	int _useBufSize;
 	bool _streamFinished;
 	QuickString _currScanBuffer;
+
 	//The minus ones in these constants are for leaving room for a null terminator after reading into buffers.
-	static const int MAIN_BUF_READ_SIZE = 1023; // 2 Kb minus 1
 	static const int GZIP_LINE_BUF_SIZE = 8191; // 8K
 	bool readFileChunk();
 	bool getTypeData();

src/utils/FileRecordTools/FileRecordMgr.cpp

@@ -24,7 +24,8 @@ FileRecordMgr::FileRecordMgr(const QuickString &filename)
   _blockMgr(NULL),
   _bamReader(NULL),
   _hasGenomeFile(false),
-  _genomeFile(NULL)
+  _genomeFile(NULL),
+  _ioBufSize(0)
 {
 }
 
@@ -43,6 +44,7 @@ FileRecordMgr::~FileRecordMgr(){
 
 bool FileRecordMgr::open(){
 	_bufStreamMgr = new BufferedStreamMgr(_filename);
+	if (_ioBufSize > 0) _bufStreamMgr->setIoBufSize(_ioBufSize);
 	if (!_bufStreamMgr->init()) {
 		cerr << "Error: unable to open file or unable to determine types for file " << _filename << endl;
 		delete _bufStreamMgr;

src/utils/FileRecordTools/FileRecordMgr.h

@@ -100,6 +100,7 @@ public:
 	}
 
 	void setIsSorted(bool val) { _isSortedInput = val; }
+	void setIoBufSize(int val) { _ioBufSize = val; }
 
 protected:
 	QuickString _filename;
@@ -131,6 +132,7 @@ protected:
 	BamTools::BamReader *_bamReader;
 	bool _hasGenomeFile;
 	NewGenomeFile *_genomeFile;
+	int _ioBufSize;
 
 	void allocateFileReader();
 	void testInputSortOrder(Record *record);

test/intersect/new_test-intersect.sh

@@ -751,7 +751,7 @@ echo \
 "chr1	11323785	11617177
 chr1	12645605	13926923
 chr1	14750216	15119039" >exp
-~/mergeBugSpace/bt2-merge-debug/bin/bedtools intersect -a dosLineChar_a.bed -b dosLineCharWithExtraTab_b.bed -v > obs
+$BT intersect -a dosLineChar_a.bed -b dosLineCharWithExtraTab_b.bed -v > obs
 check exp obs
 rm exp obs
 

test/merge/test-merge.sh

@@ -449,3 +449,72 @@ check exp obs
 rm obs exp
 
 
+
+
+###########################################################
+#
+#  Test new -iobuf option
+#  
+###########################################################
+
+
+###########################################################
+#  Test -iobuf expects an argument
+###########################################################
+echo "    merge.t37...\c"
+echo "***** ERROR: -iobuf option given, but size of input buffer not specified. *****" >exp
+$BT merge -i a.bed -iobuf 2>&1 > /dev/null | head -2 | tail -1 > obs
+check obs exp
+rm obs exp
+
+###########################################################
+#  Test -iobuf allows only suffixes K/M/G
+###########################################################
+echo "    merge.t38...\c"
+echo \
+"***** ERROR: Unrecognized memory buffer size suffix 'L' given. *****" > exp
+$BT merge -i a.bed -iobuf 20L 2>&1 > /dev/null | head -2 | tail -1 > obs
+check obs exp
+rm obs exp
+
+###########################################################
+#  Test -iobuf doesn't allow a buffer size below 8 bytes.
+###########################################################
+echo "    merge.t39...\c"
+echo \
+"***** ERROR: specified buffer size is too small. *****" > exp
+$BT merge -i a.bed -iobuf 7 2>&1 > /dev/null | head -2 | tail -1 > obs
+check exp obs
+rm exp obs
+
+###########################################################
+#  Test -iobuf doesn't allow non-numeric arguments
+###########################################################
+echo "    merge.t40...\c"
+echo \
+"***** ERROR: argument passed to -iobuf is not numeric. *****" > exp
+$BT merge -i a.bed -iobuf beerM 2>&1 > /dev/null | head -2 | tail -1 > obs
+check exp obs
+rm exp obs
+
+###########################################################
+#  Test -iobuf allows correct argument with suffix
+###########################################################
+echo "    merge.t41...\c"
+echo \
+"chr1	10	20
+chr1	30	100" > exp
+$BT merge -i a.bed -iobuf 128M > obs
+check exp obs
+rm exp obs
+
+###########################################################
+#  Test -iobuf allows correct argument without suffix
+###########################################################
+echo "    merge.t42...\c"
+echo \
+"chr1	10	20
+chr1	30	100" > exp
+$BT merge -i a.bed -iobuf 8192 > obs
+check exp obs
+rm exp obs