added check to make sure shhh.flows child processes finish properly. added subsamplin...

[mothur.git] / seqerrorcommand.cpp

diff --git a/seqerrorcommand.cpp b/seqerrorcommand.cpp
index fc027b9261822880ac0afa1cc99929791a309df4..0dc15f59ee98d0ee2d93659c4c3ad1f033f8a193 100644 (file)
--- a/seqerrorcommand.cpp
+++ b/seqerrorcommand.cpp
@@ -176,9 +176,13 @@ SeqErrorCommand::SeqErrorCommand(string option)  {
                                if (queryFileName != "") { m->mothurOut("Using " + queryFileName + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
                                else {  m->mothurOut("You have no current fasta file and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
                        }
-                       else if (queryFileName == "not open") { abort = true; } 
+                       else if (queryFileName == "not open") { queryFileName = ""; abort = true; }     
                        else { m->setFastaFile(queryFileName); }
-
+                       
+                       referenceFileName = validParameter.validFile(parameters, "reference", true);
+                       if (referenceFileName == "not found") { m->mothurOut("reference is a required parameter for the seq.error command."); m->mothurOutEndLine(); abort = true; }
+                       else if (referenceFileName == "not open") { abort = true; }     
+                       
                        //check for optional parameters
                        namesFileName = validParameter.validFile(parameters, "name", true);
                        if(namesFileName == "not found"){       namesFileName = "";     }
@@ -209,7 +213,7 @@ SeqErrorCommand::SeqErrorCommand(string option)  {
                        //check for optional parameter and set defaults
                        // ...at some point should added some additional type checking...
                        temp = validParameter.validFile(parameters, "threshold", false);        if (temp == "not found") { temp = "1.00"; }
-                       convert(temp, threshold);  
+                       m->mothurConvert(temp, threshold);  
                        
                        temp = validParameter.validFile(parameters, "save", false);                     if (temp == "not found"){       temp = "f";                             }
                        save = m->isTrue(temp); 
@@ -238,10 +242,15 @@ SeqErrorCommand::SeqErrorCommand(string option)  {
                        
                        temp = validParameter.validFile(parameters, "processors", false);       if (temp == "not found"){       temp = m->getProcessors();      }
                        m->setProcessors(temp);
-                       convert(temp, processors); 
+                       m->mothurConvert(temp, processors); 
 
                        substitutionMatrix.resize(6);
                        for(int i=0;i<6;i++){   substitutionMatrix[i].resize(6,0);      }
+                       
+                       if ((namesFileName == "") && (queryFileName != "")){
+                               vector<string> files; files.push_back(queryFileName); 
+                               parser.getNameFile(files);
+                       }
                }
        }
        catch(exception& e) {
@@ -273,9 +282,9 @@ int SeqErrorCommand::execute(){
 
                if(namesFileName != ""){        weights = getWeights(); }
                
-               vector<unsigned long int> fastaFilePos;
-               vector<unsigned long int> qFilePos;
-               vector<unsigned long int> reportFilePos;
+               vector<unsigned long long> fastaFilePos;
+               vector<unsigned long long> qFilePos;
+               vector<unsigned long long> reportFilePos;
                
                setLines(queryFileName, qualFileName, reportFileName, fastaFilePos, qFilePos, reportFilePos);
                
@@ -289,7 +298,7 @@ int SeqErrorCommand::execute(){
                if(qualFileName == "")  {       qLines = lines; rLines = lines; } //fills with duds
                
                int numSeqs = 0;
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                if(processors == 1){
                        numSeqs = driver(queryFileName, qualFileName, reportFileName, errorSummaryFileName, errorSeqFileName, errorChimeraFileName, lines[0], qLines[0], rLines[0]);
                }else{
@@ -306,7 +315,7 @@ int SeqErrorCommand::execute(){
                
                printErrorFRFile(errorForward, errorReverse);
                
-               if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
+               if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; }
 
                string errorCountFileName = queryFileName.substr(0,queryFileName.find_last_of('.')) + ".error.count";
                ofstream errorCountFile;
@@ -321,7 +330,7 @@ int SeqErrorCommand::execute(){
                }
                errorCountFile.close();
                
-               if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
+               if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]); } return 0; }
 
                printSubMatrix();
                                
@@ -357,7 +366,7 @@ int SeqErrorCommand::createProcesses(string filename, string qFileName, string r
                processIDS.clear();
                map<char, vector<int> >::iterator it;
                int num = 0;
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                
                //loop through and create all the processes you want
                while (process != processors) {
@@ -478,11 +487,11 @@ int SeqErrorCommand::createProcesses(string filename, string qFileName, string r
                        m->mothurOut("Appending files from process " + toString(processIDS[i])); m->mothurOutEndLine();
                        
                        m->appendFiles((summaryFileName + toString(processIDS[i]) + ".temp"), summaryFileName);
-                       remove((summaryFileName + toString(processIDS[i]) + ".temp").c_str());
+                       m->mothurRemove((summaryFileName + toString(processIDS[i]) + ".temp"));
                        m->appendFiles((errorOutputFileName + toString(processIDS[i]) + ".temp"), errorOutputFileName);
-                       remove((errorOutputFileName + toString(processIDS[i]) + ".temp").c_str());
+                       m->mothurRemove((errorOutputFileName + toString(processIDS[i]) + ".temp"));
                        m->appendFiles((chimeraOutputFileName + toString(processIDS[i]) + ".temp"), chimeraOutputFileName);
-                       remove((chimeraOutputFileName + toString(processIDS[i]) + ".temp").c_str());
+                       m->mothurRemove((chimeraOutputFileName + toString(processIDS[i]) + ".temp"));
                        
                        ifstream in;
                        string tempFile =  filename + toString(processIDS[i]) + ".info.temp";
@@ -577,7 +586,7 @@ int SeqErrorCommand::createProcesses(string filename, string qFileName, string r
                        }
                        m->gobble(in);
                        
-                       in.close(); remove(tempFile.c_str());
+                       in.close(); m->mothurRemove(tempFile);
                        
                }
 #endif         
@@ -673,7 +682,8 @@ int SeqErrorCommand::driver(string filename, string qFileName, string rFileName,
                        if(numParentSeqs > 1 && ignoreChimeras == 1)    {       ignoreSeq = 1;  }
                        else                                                                                    {       ignoreSeq = 0;  }
                        
-                       Compare minCompare = getErrors(query, referenceSeqs[closestRefIndex]);
+                       Compare minCompare;
+            getErrors(query, referenceSeqs[closestRefIndex], minCompare);
                        
                        if(namesFileName != ""){
                                it = weights.find(query.getName());
@@ -730,8 +740,8 @@ int SeqErrorCommand::driver(string filename, string qFileName, string rFileName,
                        
                        index++;
                        
-                       #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
-                               unsigned long int pos = queryFile.tellg();
+                       #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
+                               unsigned long long pos = queryFile.tellg();
                                if ((pos == -1) || (pos >= line.end)) { break; }
                        #else
                                if (queryFile.eof()) { break; }
@@ -830,7 +840,7 @@ void SeqErrorCommand::getReferences(){
 
 //***************************************************************************************************************
 
-Compare SeqErrorCommand::getErrors(Sequence query, Sequence reference){
+int SeqErrorCommand::getErrors(Sequence query, Sequence reference, Compare& errors){
        try {
                if(query.getAlignLength() != reference.getAlignLength()){
                        m->mothurOut("Warning: " + toString(query.getName()) + " and " + toString(reference.getName()) + " are different lengths\n");
@@ -841,7 +851,7 @@ Compare SeqErrorCommand::getErrors(Sequence query, Sequence reference){
                string r = reference.getAligned();
 
                int started = 0;
-               Compare errors;
+               //Compare errors;
 
                for(int i=0;i<alignLength;i++){
 //                     cout << r[i] << '\t' << q[i] << '\t';
@@ -920,7 +930,8 @@ Compare SeqErrorCommand::getErrors(Sequence query, Sequence reference){
                errors.queryName = query.getName();
                errors.refName = reference.getName();
                
-               return errors;
+               //return errors;
+        return 0;
        }
        catch(exception& e) {
                m->errorOut(e, "SeqErrorCommand", "getErrors");
@@ -1204,9 +1215,9 @@ void SeqErrorCommand::printQualityFR(vector<vector<int> > qualForwardMap, vector
 }
 /**************************************************************************************************/
 
-int SeqErrorCommand::setLines(string filename, string qfilename, string rfilename, vector<unsigned long int>& fastaFilePos, vector<unsigned long int>& qfileFilePos, vector<unsigned long int>& rfileFilePos) {
+int SeqErrorCommand::setLines(string filename, string qfilename, string rfilename, vector<unsigned long long>& fastaFilePos, vector<unsigned long long>& qfileFilePos, vector<unsigned long long>& rfileFilePos) {
        try {
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                //set file positions for fasta file
                fastaFilePos = m->divideFile(filename, processors);
                
@@ -1246,7 +1257,7 @@ int SeqErrorCommand::setLines(string filename, string qfilename, string rfilenam
                                        map<string, int>::iterator it = firstSeqNames.find(sname);
                                        
                                        if(it != firstSeqNames.end()) { //this is the start of a new chunk
-                                               unsigned long int pos = inQual.tellg(); 
+                                               unsigned long long pos = inQual.tellg(); 
                                                qfileFilePos.push_back(pos - input.length() - 1);       
                                                firstSeqNames.erase(it);
                                        }
@@ -1267,7 +1278,7 @@ int SeqErrorCommand::setLines(string filename, string qfilename, string rfilenam
                
                //get last file position of qfile
                FILE * pFile;
-               unsigned long int size;
+               unsigned long long size;
                
                //get num bytes in file
                pFile = fopen (qfilename.c_str(),"rb");
@@ -1305,7 +1316,7 @@ int SeqErrorCommand::setLines(string filename, string qfilename, string rfilenam
                                map<string, int>::iterator it = firstSeqNamesReport.find(sname);
                        
                                if(it != firstSeqNamesReport.end()) { //this is the start of a new chunk
-                                       unsigned long int pos = inR.tellg(); 
+                                       unsigned long long pos = inR.tellg(); 
                                        rfileFilePos.push_back(pos - input.length() - 1);       
                                        firstSeqNamesReport.erase(it);
                                }
@@ -1326,7 +1337,7 @@ int SeqErrorCommand::setLines(string filename, string qfilename, string rfilenam
                
                //get last file position of qfile
                FILE * rFile;
-               unsigned long int sizeR;
+               unsigned long long sizeR;
                
                //get num bytes in file
                rFile = fopen (rfilename.c_str(),"rb");
@@ -1346,7 +1357,7 @@ int SeqErrorCommand::setLines(string filename, string qfilename, string rfilenam
                fastaFilePos.push_back(0); qfileFilePos.push_back(0);
                //get last file position of fastafile
                FILE * pFile;
-               unsigned long int size;
+               unsigned long long size;
                
                //get num bytes in file
                pFile = fopen (filename.c_str(),"rb");