From 930ddf38f069b40a5306857446e373a64417f1bc Mon Sep 17 00:00:00 2001
From: nkindlon <nek3d@virginia.edu>
Date: Wed, 18 Dec 2013 22:24:04 -0500
Subject: [PATCH] Added -nobuf option, and changed flush to occur after every
record if -nobuf is set.
---
src/utils/Contexts/Context.cpp | 5 +++++
src/utils/Contexts/Context.h | 4 ++++
.../FileRecordTools/FileReaders/BufferedStreamMgr.h | 2 +-
src/utils/FileRecordTools/RecordOutputMgr.cpp | 12 ++++++++++++
src/utils/FileRecordTools/RecordOutputMgr.h | 10 ++++++++--
5 files changed, 30 insertions(+), 3 deletions(-)
diff --git a/src/utils/Contexts/Context.cpp b/src/utils/Contexts/Context.cpp
index 22bbf126..439c1fef 100644
--- a/src/utils/Contexts/Context.cpp
+++ b/src/utils/Contexts/Context.cpp
@@ -20,6 +20,7 @@ Context::Context()
_showHelp(false),
_obeySplits(false),
_uncompressedBam(false),
+ _useBufferedOutput(true),
_anyHit(false),
_noHit(false),
_writeA(false),
@@ -288,6 +289,10 @@ bool Context::parseCmdArgs(int argc, char **argv, int skipFirstArgs) {
setSortedInput(true);
markUsed(i - skipFirstArgs);
}
+ else if(strcmp(argv[i], "-nobuf") == 0) {
+ setUseBufferedOutput(false);
+ markUsed(i - skipFirstArgs);
+ }
else if(strcmp(argv[i], "-header") == 0) {
setPrintHeader(true);
markUsed(i - skipFirstArgs);
diff --git a/src/utils/Contexts/Context.h b/src/utils/Contexts/Context.h
index 7f26fc0c..19160e95 100644
--- a/src/utils/Contexts/Context.h
+++ b/src/utils/Contexts/Context.h
@@ -105,6 +105,9 @@ public:
bool getUncompressedBam() const { return _uncompressedBam; }
void setUncompressedBam(bool val) { _uncompressedBam = val; }
+
+ bool getUseBufferedOutput() const { return _useBufferedOutput; }
+ void setUseBufferedOutput(bool val) { _useBufferedOutput = val; }
//
// INTERSECT METOHDS
//
@@ -241,6 +244,7 @@ protected:
bool _showHelp;
bool _obeySplits;
bool _uncompressedBam;
+ bool _useBufferedOutput;
bool _anyHit;
bool _noHit;
diff --git a/src/utils/FileRecordTools/FileReaders/BufferedStreamMgr.h b/src/utils/FileRecordTools/FileReaders/BufferedStreamMgr.h
index 963b5ce3..2a77854a 100644
--- a/src/utils/FileRecordTools/FileReaders/BufferedStreamMgr.h
+++ b/src/utils/FileRecordTools/FileReaders/BufferedStreamMgr.h
@@ -43,7 +43,7 @@ private:
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 = 2047; // 2 Kb minus 1
+ 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();
diff --git a/src/utils/FileRecordTools/RecordOutputMgr.cpp b/src/utils/FileRecordTools/RecordOutputMgr.cpp
index c405a435..b9b1a0a3 100644
--- a/src/utils/FileRecordTools/RecordOutputMgr.cpp
+++ b/src/utils/FileRecordTools/RecordOutputMgr.cpp
@@ -82,6 +82,7 @@ bool RecordOutputMgr::printKeyAndTerminate(RecordKeyList &keyList) {
} else if (bamCode == NOT_BAM) {
keyList.getKey()->print(_outBuf);
return false;
+
}
//otherwise, it was BAM_AS_BED, and the key was printed.
return false;
@@ -157,6 +158,7 @@ void RecordOutputMgr::printRecord(RecordKeyList &keyList, RecordKeyList *blockLi
tab();
_outBuf.append('0');
newline();
+ if (needsFlush()) flush();
}
else if (_context->getLeftJoin()) {
if (printKeyAndTerminate(keyList)) {
@@ -166,6 +168,7 @@ void RecordOutputMgr::printRecord(RecordKeyList &keyList, RecordKeyList *blockLi
tab();
null(false, true);
newline();
+ if (needsFlush()) flush();
_currBlockList = NULL;
return;
}
@@ -202,6 +205,7 @@ void RecordOutputMgr::checkForHeader() {
}
}
_context->setPrintHeader(false);
+ if (needsFlush()) flush();
}
void RecordOutputMgr::reportOverlapDetail(const Record *keyRecord, const Record *hitRecord)
@@ -269,22 +273,26 @@ void RecordOutputMgr::reportOverlapDetail(const Record *keyRecord, const Record
if (!_context->getWriteA() && !_context->getWriteB() && !_context->getWriteOverlap() && !_context->getLeftJoin()) {
printKey(keyRecord, *startStr, *endStr);
newline();
+ if (needsFlush()) flush();
}
else if ((_context->getWriteA() && _context->getWriteB()) || _context->getLeftJoin()) {
printKey(keyRecord);
tab();
hitRecord->print(_outBuf);
newline();
+ if (needsFlush()) flush();
}
else if (_context->getWriteA()) {
printKey(keyRecord);
newline();
+ if (needsFlush()) flush();
}
else if (_context->getWriteB()) {
printKey(keyRecord, *startStr, *endStr);
tab();
hitRecord->print(_outBuf);
newline();
+ if (needsFlush()) flush();
}
else if (_context->getWriteOverlap()) {
int printOverlapBases = max(0, minEnd-maxStart);
@@ -294,6 +302,7 @@ void RecordOutputMgr::reportOverlapDetail(const Record *keyRecord, const Record
tab();
int2str(printOverlapBases, _outBuf, true);
newline();
+ if (needsFlush()) flush();
}
}
@@ -305,6 +314,7 @@ void RecordOutputMgr::reportOverlapSummary(RecordKeyList &keyList)
return;
}
newline();
+ if (needsFlush()) flush();
} else if (_context->getWriteCount()) {
if (printKeyAndTerminate(keyList)) {
return;
@@ -312,11 +322,13 @@ void RecordOutputMgr::reportOverlapSummary(RecordKeyList &keyList)
tab();
int2str(numOverlapsFound, _outBuf, true);
newline();
+ if (needsFlush()) flush();
} else if (_context->getNoHit() && numOverlapsFound == 0) {
if (printKeyAndTerminate(keyList)) {
return;
}
newline();
+ if (needsFlush()) flush();
}
}
diff --git a/src/utils/FileRecordTools/RecordOutputMgr.h b/src/utils/FileRecordTools/RecordOutputMgr.h
index 1bb8bbfc..de626116 100644
--- a/src/utils/FileRecordTools/RecordOutputMgr.h
+++ b/src/utils/FileRecordTools/RecordOutputMgr.h
@@ -10,10 +10,10 @@
using namespace std;
+#include "Context.h"
#include "RecordKeyList.h"
#include "api/BamWriter.h"
-class Context;
class BlockMgr;
class RecordOutputMgr {
@@ -53,7 +53,13 @@ private:
void reportOverlapSummary(RecordKeyList &keyList);
static const unsigned int MAX_OUTBUF_SIZE = 16384; //16 K
- bool needsFlush() const { return _outBuf.size() >= MAX_OUTBUF_SIZE *.9; }
+
+ // If we are using buffered output, only flush the output buffer if it's least
+ // 90% full. If we're not using buffered output, flush if it's not empty
+ bool needsFlush() const {
+ return ((_context->getUseBufferedOutput() &&_outBuf.size() >= MAX_OUTBUF_SIZE *.9) ||
+ (!_context->getUseBufferedOutput() && !_outBuf.empty()));
+ }
void flush();
};
--
GitLab