adding labels to list file.

[mothur.git] / getseqscommand.cpp

diff --git a/getseqscommand.cpp b/getseqscommand.cpp
index 142023e7a40881c1c6397a761ba350132490b75c..7f0f3ddb51da8496c292367b5dfc728b3915d53f 100644 (file)
--- a/getseqscommand.cpp
+++ b/getseqscommand.cpp
@@ -45,7 +45,7 @@ string GetSeqsCommand::getHelpString(){
                helpString += "The get.seqs command reads an .accnos file and any of the following file types: fasta, name, group, count, list, taxonomy, quality or alignreport file.\n";
                helpString += "It outputs a file containing only the sequences in the .accnos file.\n";
                helpString += "The get.seqs command parameters are accnos, fasta, name, group, list, taxonomy, qfile, alignreport and dups.  You must provide accnos unless you have a valid current accnos file, and at least one of the other parameters.\n";
-               helpString += "The dups parameter allows you to add the entire line from a name file if you add any name from the line. default=false. \n";
+               helpString += "The dups parameter allows you to add the entire line from a name file if you add any name from the line. default=true. \n";
                helpString += "The get.seqs command should be in the following format: get.seqs(accnos=yourAccnos, fasta=yourFasta).\n";
                helpString += "Example get.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n";
                helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
@@ -88,7 +88,7 @@ string GetSeqsCommand::getOutputPattern(string type) {
         else if (type == "name")        {   pattern = "[filename],pick,[extension]";    }
         else if (type == "group")       {   pattern = "[filename],pick,[extension]";    }
         else if (type == "count")       {   pattern = "[filename],pick,[extension]";    }
-        else if (type == "list")        {   pattern = "[filename],pick,[extension]";    }
+        else if (type == "list")        {   pattern = "[filename],[distance],pick,[extension]";    }
         else if (type == "qfile")       {   pattern = "[filename],pick,[extension]";    }
         else if (type == "accnosreport")      {   pattern = "[filename],pick.accnos.report";    }
         else if (type == "alignreport")      {   pattern = "[filename],pick.align.report";    }
@@ -430,9 +430,11 @@ int GetSeqsCommand::readFasta(){
             
             if (!dups) {//adjust name if needed
                 map<string, string>::iterator it = uniqueMap.find(name);
-                if (it != uniqueMap.end()) { name = it->second; }
+                if (it != uniqueMap.end()) { currSeq.setName(it->second); }
             }
                        
+            name = currSeq.getName();
+            
                        if (name != "") {
                                //if this name is in the accnos file
                                if (names.count(name) != 0) {
@@ -587,7 +589,7 @@ int GetSeqsCommand::readCount(){
         //check for groups that have been eliminated
         CountTable ct;
         if (ct.testGroups(outputFileName)) {
-            ct.readTable(outputFileName, true);
+            ct.readTable(outputFileName, true, false);
             ct.printTable(outputFileName);
         }
                
@@ -612,9 +614,6 @@ int GetSeqsCommand::readList(){
         map<string, string> variables; 
                variables["[filename]"] = thisOutputDir + m->getRootName(m->getSimpleName(listfile));
         variables["[extension]"] = m->getExtension(listfile);
-               string outputFileName = getOutputFileName("list", variables);
-               ofstream out;
-               m->openOutputFile(outputFileName, out);
                
                ifstream in;
                m->openInputFile(listfile, in);
@@ -627,8 +626,6 @@ int GetSeqsCommand::readList(){
                while(!in.eof()){
                        
                        selectedCount = 0;
-                       
-                       if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 
                        //read in list vector
                        ListVector list(in);
@@ -636,6 +633,18 @@ int GetSeqsCommand::readList(){
                        //make a new list vector
                        ListVector newList;
                        newList.setLabel(list.getLabel());
+            
+            variables["[distance]"] = list.getLabel();
+            string outputFileName = getOutputFileName("list", variables);
+                       
+                       ofstream out;
+                       m->openOutputFile(outputFileName, out);
+                       outputTypes["list"].push_back(outputFileName);  outputNames.push_back(outputFileName);
+            
+            vector<string> binLabels = list.getLabels();
+            vector<string> newBinLabels;
+            
+            if (m->control_pressed) { in.close(); out.close();  return 0; }
                        
                        //for each bin
                        for (int i = 0; i < list.getNumBins(); i++) {
@@ -655,23 +664,26 @@ int GetSeqsCommand::readList(){
                                //if there are names in this bin add to new list
                                if (newNames != "") { 
                                        newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
-                                       newList.push_back(newNames);    
+                                       newList.push_back(newNames);
+                    newBinLabels.push_back(binLabels[i]);
                                }
                        }
                                
                        //print new listvector
                        if (newList.getNumBins() != 0) {
                                wroteSomething = true;
+                               newList.setLabels(newBinLabels);
+                newList.printHeaders(out);
                                newList.print(out);
                        }
                        
                        m->gobble(in);
+            out.close();
                }
                in.close();     
-               out.close();
+               
                
                if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
-               outputNames.push_back(outputFileName); outputTypes["list"].push_back(outputFileName);
                
                m->mothurOut("Selected " + toString(selectedCount) + " sequences from your list file."); m->mothurOutEndLine();
                
@@ -710,7 +722,7 @@ int GetSeqsCommand::readName(){
                
                        if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(outputFileName);  return 0; }
 
-                       in >> firstCol;                         
+                       in >> firstCol;                 m->gobble(in);
                        in >> secondCol;
                        
                        string hold = "";
@@ -734,6 +746,7 @@ int GetSeqsCommand::readName(){
                                selectedCount += parsedNames.size();
                 if (m->debug) { sanity["name"].insert(firstCol); }
                        }else {
+                
                                selectedCount += validSecond.size();
                                
                                //if the name in the first column is in the set then print it and any other names in second column also in set
@@ -752,6 +765,7 @@ int GetSeqsCommand::readName(){
                                
                                //make first name in set you come to first column and then add the remaining names to second column
                                }else {
+                    
                                        //you want part of this row
                                        if (validSecond.size() != 0) {