started work on classify.seqs command. changed the database class so that it does...

[mothur.git] / removeseqscommand.cpp

diff --git a/removeseqscommand.cpp b/removeseqscommand.cpp
index d3d12cc321f9cc6d1021bf7bcab4c13636d1ce80..1b9edb0e98e20f7752fb9b8aab3584c1c05bccc0 100644 (file)
--- a/removeseqscommand.cpp
+++ b/removeseqscommand.cpp
@@ -21,7 +21,7 @@ RemoveSeqsCommand::RemoveSeqsCommand(string option){
                
                else {
                        //valid paramters for this command
-                       string Array[] =  {"fasta","name", "group", "align", "accnos" };
+                       string Array[] =  {"fasta","name", "group", "alignreport", "accnos" };
                        vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
                        
                        OptionParser parser(option);
@@ -51,13 +51,13 @@ RemoveSeqsCommand::RemoveSeqsCommand(string option){
                        if (groupfile == "not open") { abort = true; }
                        else if (groupfile == "not found") {  groupfile = "";  }        
                        
-                       alignfile = validParameter.validFile(parameters, "align", true);
+                       alignfile = validParameter.validFile(parameters, "alignreport", true);
                        if (alignfile == "not open") { abort = true; }
                        else if (alignfile == "not found") {  alignfile = "";  }
                        
-                       if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == ""))  { mothurOut("You must provide one of the following: fasta, name, group, align."); mothurOutEndLine(); abort = true; }
+                       if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == ""))  { mothurOut("You must provide one of the following: fasta, name, group, alignreport."); mothurOutEndLine(); abort = true; }
                        
-                       if (parameters.size() > 2) { mothurOut("You may only enter one of the following: fasta, name, group, align."); mothurOutEndLine(); abort = true;  }
+                       if (parameters.size() > 2) { mothurOut("You may only enter one of the following: fasta, name, group, alignreport."); mothurOutEndLine(); abort = true;  }
                }
 
        }
@@ -72,7 +72,7 @@ void RemoveSeqsCommand::help(){
        try {
                mothurOut("The remove.seqs command reads an .accnos file and one of the following file types: fasta, name, group or alignreport file.\n");
                mothurOut("It outputs a file containing the sequences NOT in the .accnos file.\n");
-               mothurOut("The remove.seqs command parameters are accnos, fasta, name, group and align.  You must provide accnos and one of the other parameters.\n");
+               mothurOut("The remove.seqs command parameters are accnos, fasta, name, group and alignreport.  You must provide accnos and one of the other parameters.\n");
                mothurOut("The remove.seqs command should be in the following format: remove.seqs(accnos=yourAccnos, fasta=yourFasta).\n");
                mothurOut("Example remove.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n");
                mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
@@ -111,7 +111,7 @@ int RemoveSeqsCommand::execute(){
 //**********************************************************************************************************************
 void RemoveSeqsCommand::readFasta(){
        try {
-               string outputFileName = getRootName(fastafile) + "pick";
+               string outputFileName = getRootName(fastafile) + "pick" + getExtension(fastafile);
                ofstream out;
                openOutputFile(outputFileName, out);
                
@@ -153,7 +153,8 @@ void RemoveSeqsCommand::readFasta(){
 void RemoveSeqsCommand::readName(){
        try {
        
-               string outputFileName = getRootName(namefile) + "pick";
+               string outputFileName = getRootName(namefile) + "pick" + getExtension(namefile);
+
                ofstream out;
                openOutputFile(outputFileName, out);
 
@@ -167,29 +168,29 @@ void RemoveSeqsCommand::readName(){
 
                        in >> firstCol;                         
                        in >> secondCol;                        
-                       
+
                        vector<string> parsedNames;
                        //parse second column saving each name
                        while (secondCol.find_first_of(',') != -1) { 
                                name = secondCol.substr(0,secondCol.find_first_of(','));
                                secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
                                parsedNames.push_back(name);
+
                        }
                        
                        //get name after last ,
                        parsedNames.push_back(secondCol);
-                       
-                       vector<string> validSecond;
+
+                       vector<string> validSecond;  validSecond.clear();
                        for (int i = 0; i < parsedNames.size(); i++) {
                                if (names.count(parsedNames[i]) == 0) {
                                        validSecond.push_back(parsedNames[i]);
-                               }else { names.erase(parsedNames[i]); }
+                               }
                        }
-
                        
                        //if the name in the first column is in the set then print it and any other names in second column also in set
                        if (names.count(firstCol) == 0) {
-                               
+                       
                                wroteSomething = true;
                                
                                out << firstCol << '\t';
@@ -200,8 +201,7 @@ void RemoveSeqsCommand::readName(){
                        
                        //make first name in set you come to first column and then add the remaining names to second column
                        }else {
-                               names.erase(firstCol);  
-                                       
+                               
                                //you want part of this row
                                if (validSecond.size() != 0) {
                                
@@ -236,7 +236,7 @@ void RemoveSeqsCommand::readName(){
 void RemoveSeqsCommand::readGroup(){
        try {
        
-               string outputFileName = getRootName(groupfile) + "pick";
+               string outputFileName = getRootName(groupfile) + "pick" + getExtension(groupfile);
                ofstream out;
                openOutputFile(outputFileName, out);
 
@@ -278,7 +278,7 @@ void RemoveSeqsCommand::readGroup(){
 //alignreport file has a column header line then all other lines contain 16 columns.  we just want the first column since that contains the name
 void RemoveSeqsCommand::readAlign(){
        try {
-               string outputFileName = getRootName(alignfile) + "pick";
+               string outputFileName = getRootName(getRootName(alignfile)) + "pick.align.report";
                ofstream out;
                openOutputFile(outputFileName, out);