Adds ability to write compressed files in get.seqs

#792 #804
mothur · Dec 14, 2021 · 457517b · 457517b
1 parent 07713e0
commit 457517b
Show file tree

Hide file tree

Showing 5 changed files with 19 additions and 22 deletions.
diff --git a/source/commands/getseqscommand.cpp b/source/commands/getseqscommand.cpp
@@ -349,14 +349,14 @@ void GetSeqsCommand::readGZFastq(string fastqfile){
         if (outputdir == "") {  thisOutputDir += util.hasPath(fastqfile);  }
         map<string, string> variables;
         variables["[filename]"] = thisOutputDir + util.getRootName(util.getSimpleName(fastqfile));
-        variables["[extension]"] = util.getExtension(fastqfile);
+        variables["[extension]"] = ".fastq" + util.getExtension(fastqfile);
         string outputFileName = getOutputFileName("fastq", variables);
 
         ifstream in; boost::iostreams::filtering_istream inBoost;
         util.openInputFileBinary(fastqfile, in, inBoost);
 
-        ofstream out; boost::iostreams::filtering_ostream outBoost;
-        util.openOutputFileBinary(fastqfile, out, outBoost);
+        ofstream file; ostream* out; boost::iostreams::filtering_streambuf<boost::iostreams::output> outBoost;
+        util.openOutputFileBinary(outputFileName, file, out, outBoost);
 
         bool wroteSomething = false; int selectedCount = 0; set<string> uniqueNames;
 
@@ -375,7 +375,7 @@ void GetSeqsCommand::readGZFastq(string fastqfile){
                     if (uniqueNames.count(name) == 0) { //this name hasn't been seen yet
                         wroteSomething = true;
                         selectedCount++;
-                        fread.printFastq(out);
+                        fread.printFastq(*out);
                         uniqueNames.insert(name);
                     }else {
                         m->mothurOut("[WARNING]: " + name + " is in your fastq file more than once.  Mothur requires sequence names to be unique. I will only add it once.\n");
@@ -384,10 +384,9 @@ void GetSeqsCommand::readGZFastq(string fastqfile){
             }
             util.gobble(inBoost);
         }
-        in.close();
-        out.close();
-        inBoost.pop();
-        outBoost.pop();
+        in.close(); inBoost.pop();
+        boost::iostreams::close(outBoost);
+        file.close(); delete out;
 
         if (m->getControl_pressed()) { util.mothurRemove(outputFileName); return; }
 

diff --git a/source/datastructures/fastqread.h b/source/datastructures/fastqread.h
@@ -56,7 +56,6 @@ class FastqRead {
     Sequence getSequence();
     QualityScores getQuality();
 
-
 private:
     MothurOut* m;
     Utils util;

diff --git a/source/refchimeratest.cpp b/source/refchimeratest.cpp
@@ -255,8 +255,6 @@ int RefChimeraTest::analyzeUnalignedQuery(string queryName, string querySeq, str
 /**************************************************************************************************/
 
 double RefChimeraTest::alignQueryToReferences(string query, string reference, string& qAlign, string& rAlign, double& length){
-
-
     try {
 		double GAP = -5;
 		double MATCH = 1;
@@ -414,7 +412,7 @@ double RefChimeraTest::alignQueryToReferences(string query, string reference, st
             rAlign = reference.substr(0, j) + rAlign;
         }
 
-        if (length == 0) { diffs = MAXINT; }
+        if (length == 0) { diffs = MOTHURMAX; }
 
 		return diffs;
 	}
@@ -423,7 +421,6 @@ double RefChimeraTest::alignQueryToReferences(string query, string reference, st
 		exit(1);
 	}
 }
-
 /**************************************************************************************************/
 
 int RefChimeraTest::getUnalignedDiffs(string qAlign, string rAlign, vector<int>& leftDiffs, vector<int>& leftMap, vector<int>& rightDiffs, vector<int>& rightMap){
@@ -491,7 +488,7 @@ int RefChimeraTest::getUnalignedDiffs(string qAlign, string rAlign, vector<int>&
 
 int RefChimeraTest::getAlignedMismatches(string& querySeq, vector<vector<int> >& left, vector<vector<int> >& right, int& bestRefSeq){
 
-	int bestSequenceMismatch = MAXINT;
+	int bestSequenceMismatch = MOTHURMAX;
 
 	for(int i=0;i<numRefSeqs;i++){
 
@@ -525,7 +522,7 @@ int RefChimeraTest::getAlignedMismatches(string& querySeq, vector<vector<int> >&
 
 int RefChimeraTest::getChimera(vector<vector<int> >& left, vector<vector<int> >& right, int& leftParent, int& rightParent, int& breakPoint, vector<int>& singleLeft, vector<int>& bestLeft, vector<int>& singleRight, vector<int>& bestRight){
 
-	singleLeft.resize(alignLength, MAXINT);
+	singleLeft.resize(alignLength, MOTHURMAX);
 	bestLeft.resize(alignLength, -1);
 
 	for(int l=0;l<alignLength;l++){
@@ -537,7 +534,7 @@ int RefChimeraTest::getChimera(vector<vector<int> >& left, vector<vector<int> >&
 		}
 	}
 
-	singleRight.resize(alignLength, MAXINT);
+	singleRight.resize(alignLength, MOTHURMAX);
 	bestRight.resize(alignLength, -1);
 
 	for(int l=0;l<alignLength;l++){
@@ -549,7 +546,7 @@ int RefChimeraTest::getChimera(vector<vector<int> >& left, vector<vector<int> >&
 		}
 	}
 
-	int bestChimeraMismatches = MAXINT;
+	int bestChimeraMismatches = MOTHURMAX;
 	leftParent = -1;
 	rightParent = -1;
 	breakPoint = -1;
@@ -571,7 +568,7 @@ int RefChimeraTest::getChimera(vector<vector<int> >& left, vector<vector<int> >&
 
 int RefChimeraTest::getTrimera(vector<vector<int> >& left, vector<vector<int> >& right, int& leftParent, int& middleParent, int& rightParent, int& breakPointA, int& breakPointB, vector<int>& singleLeft, vector<int>& bestLeft, vector<int>& singleRight, vector<int>& bestRight){
 
-	int bestTrimeraMismatches = MAXINT;
+	int bestTrimeraMismatches = MOTHURMAX;
 
 	leftParent = -1;
 	middleParent = -1;
@@ -584,7 +581,7 @@ int RefChimeraTest::getTrimera(vector<vector<int> >& left, vector<vector<int> >&
     vector<vector<int> > minDeltaSeq; minDeltaSeq.resize(alignLength);
 
 	for(int i=0;i<alignLength;i++){
-		minDelta[i].assign(alignLength, MAXINT);
+		minDelta[i].assign(alignLength, MOTHURMAX);
 		minDeltaSeq[i].assign(alignLength, -1);
 	}
 

diff --git a/source/utils.cpp b/source/utils.cpp
@@ -1114,16 +1114,18 @@ bool Utils::openInputFileBinary(string fileName, ifstream& file, boost::iostream
     }
 }
 /***********************************************************************/
-bool Utils::openOutputFileBinary(string fileName, ofstream& file, boost::iostreams::filtering_ostream& outBoost){
+bool Utils::openOutputFileBinary(string fileName, ofstream& file, ostream*& out, boost::iostreams::filtering_streambuf<boost::iostreams::output>& outBoost){
     try {
         string completeFileName = getFullPathName(fileName);
 
         file.open(completeFileName, ios_base::out | ios_base::binary | ios_base::trunc);
 
         if(!file) { return false; }
         else { //check for blank file
-            outBoost.push(boost::iostreams::gzip_compressor());
+            outBoost.push(boost::iostreams::gzip_compressor(boost::iostreams::gzip_params(9)));
             outBoost.push(file);
+
+            out = new ostream(&outBoost);
             return true;
         }
     }

diff --git a/source/utils.hpp b/source/utils.hpp
@@ -118,7 +118,7 @@ class Utils {
 #ifdef USE_BOOST
     bool openInputFileBinary(string, ifstream&, boost::iostreams::filtering_istream&);
     bool openInputFileBinary(string, ifstream&, boost::iostreams::filtering_istream&, string);
-    bool openOutputFileBinary(string, ofstream&, boost::iostreams::filtering_ostream&);
+    bool openOutputFileBinary(string fileName, ofstream& file, ostream*& out, boost::iostreams::filtering_streambuf<boost::iostreams::output>& outBoost);
     string getline(boost::iostreams::filtering_istream& fileHandle);
 #endif