source/splitgroupscommand.cpp

   1 /*
   2  *  splitgroupscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 9/20/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "splitgroupscommand.h"
  11 #include "sharedutilities.h"
  12 #include "sequenceparser.h"
  13
  14 //**********************************************************************************************************************
  15 vector<string> SplitGroupCommand::setParameters(){
  16         try {
  17                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pfasta);
  18                 CommandParameter pname("name", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(pname);
  19                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "none", "none",false,true); parameters.push_back(pgroup);
  20                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  21                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  22                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  23
  24                 vector<string> myArray;
  25                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  26                 return myArray;
  27         }
  28         catch(exception& e) {
  29                 m->errorOut(e, "SplitGroupCommand", "setParameters");
  30                 exit(1);
  31         }
  32 }
  33 //**********************************************************************************************************************
  34 string SplitGroupCommand::getHelpString(){
  35         try {
  36                 string helpString = "";
  37                 helpString += "The split.group command reads a group file, and parses your fasta and names files by groups. \n";
  38                 helpString += "The split.group command parameters are fasta, name, group and groups.\n";
  39                 helpString += "The fasta and group parameters are required.\n";
  40                 helpString += "The groups parameter allows you to select groups to create files for.  \n";
  41                 helpString += "For example if you set groups=A-B-C, you will get a .A.fasta, .A.names, .B.fasta, .B.names, .C.fasta, .C.names files.  \n";
  42                 helpString += "If you want .fasta and .names files for all groups, set groups=all.  \n";
  43                 helpString += "The split.group command should be used in the following format: split.group(fasta=yourFasta, group=yourGroupFile).\n";
  44                 helpString += "Example: split.group(fasta=abrecovery.fasta, group=abrecovery.groups).\n";
  45                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  46                 return helpString;
  47         }
  48         catch(exception& e) {
  49                 m->errorOut(e, "SplitGroupCommand", "getHelpString");
  50                 exit(1);
  51         }
  52 }
  53 //**********************************************************************************************************************
  54 SplitGroupCommand::SplitGroupCommand(){
  55         try {
  56                 abort = true; calledHelp = true;
  57                 setParameters();
  58                 vector<string> tempOutNames;
  59                 outputTypes["fasta"] = tempOutNames;
  60                 outputTypes["name"] = tempOutNames;
  61         }
  62         catch(exception& e) {
  63                 m->errorOut(e, "SplitGroupCommand", "SplitGroupCommand");
  64                 exit(1);
  65         }
  66 }
  67 //**********************************************************************************************************************
  68 SplitGroupCommand::SplitGroupCommand(string option)  {
  69         try {
  70                 abort = false; calledHelp = false;
  71
  72                 //allow user to run help
  73                 if(option == "help") { help(); abort = true; calledHelp = true; }
  74                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  75
  76                 else {
  77                         vector<string> myArray = setParameters();
  78
  79                         OptionParser parser(option);
  80                         map<string, string> parameters = parser.getParameters();
  81
  82                         ValidParameters validParameter;
  83                         map<string, string>::iterator it;
  84
  85                         //check to make sure all parameters are valid for command
  86                         for (it = parameters.begin(); it != parameters.end(); it++) {
  87                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  88                         }
  89
  90                         //initialize outputTypes
  91                         vector<string> tempOutNames;
  92                         outputTypes["fasta"] = tempOutNames;
  93                         outputTypes["name"] = tempOutNames;
  94
  95                         //if the user changes the input directory command factory will send this info to us in the output parameter
  96                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
  97                         if (inputDir == "not found"){   inputDir = "";          }
  98                         else {
  99                                 string path;
 100                                 it = parameters.find("group");
 101                                 //user has given a template file
 102                                 if(it != parameters.end()){
 103                                         path = m->hasPath(it->second);
 104                                         //if the user has not given a path then, add inputdir. else leave path alone.
 105                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 106                                 }
 107
 108                                 it = parameters.find("fasta");
 109                                 //user has given a template file
 110                                 if(it != parameters.end()){
 111                                         path = m->hasPath(it->second);
 112                                         //if the user has not given a path then, add inputdir. else leave path alone.
 113                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 114                                 }
 115
 116                                 it = parameters.find("name");
 117                                 //user has given a template file
 118                                 if(it != parameters.end()){
 119                                         path = m->hasPath(it->second);
 120                                         //if the user has not given a path then, add inputdir. else leave path alone.
 121                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 122                                 }
 123
 124                         }
 125
 126
 127                         namefile = validParameter.validFile(parameters, "name", true);
 128                         if (namefile == "not open") { namefile = ""; abort = true; }
 129                         else if (namefile == "not found") { namefile = ""; }
 130                         else { m->setNameFile(namefile); }
 131
 132                         fastafile = validParameter.validFile(parameters, "fasta", true);
 133                         if (fastafile == "not open") { abort = true; }
 134                         else if (fastafile == "not found") {
 135                                 fastafile = m->getFastaFile();
 136                                 if (fastafile != "") { m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 137                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 138                         }else { m->setFastaFile(fastafile); }
 139
 140                         groupfile = validParameter.validFile(parameters, "group", true);
 141                         if (groupfile == "not open") {  groupfile = ""; abort = true; }
 142                         else if (groupfile == "not found") {
 143                                 groupfile = m->getGroupFile();
 144                                 if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 145                                 else {  m->mothurOut("You have no current groupfile and the group parameter is required."); m->mothurOutEndLine(); abort = true; }
 146                         }else {  m->setGroupFile(groupfile); }
 147
 148                         groups = validParameter.validFile(parameters, "groups", false);
 149                         if (groups == "not found") { groups = ""; }
 150                         else { m->splitAtDash(groups, Groups);  }
 151
 152                         //if the user changes the output directory command factory will send this info to us in the output parameter
 153                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(groupfile);      }
 154
 155                         if (namefile == "") {
 156                                 vector<string> files; files.push_back(fastafile);
 157                                 parser.getNameFile(files);
 158                         }
 159                 }
 160
 161         }
 162         catch(exception& e) {
 163                 m->errorOut(e, "SplitGroupCommand", "SplitAbundCommand");
 164                 exit(1);
 165         }
 166 }
 167 //**********************************************************************************************************************
 168 int SplitGroupCommand::execute(){
 169         try {
 170
 171                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 172
 173                 SequenceParser* parser;
 174
 175                 if (namefile == "") {   parser = new SequenceParser(groupfile, fastafile);                              }
 176                 else                            {       parser = new SequenceParser(groupfile, fastafile, namefile);    }
 177
 178                 if (m->control_pressed) { delete parser; return 0; }
 179
 180                 vector<string> namesGroups = parser->getNamesOfGroups();
 181                 SharedUtil util;  util.setGroups(Groups, namesGroups);
 182
 183                 string fastafileRoot = outputDir + m->getRootName(m->getSimpleName(fastafile));
 184                 string namefileRoot = outputDir + m->getRootName(m->getSimpleName(namefile));
 185
 186                 m->mothurOutEndLine();
 187                 for (int i = 0; i < Groups.size(); i++) {
 188
 189                         m->mothurOut("Processing group: " + Groups[i]); m->mothurOutEndLine();
 190
 191                         string newFasta = fastafileRoot + Groups[i] + ".fasta";
 192                         string newName = namefileRoot + Groups[i] + ".names";
 193
 194                         parser->getSeqs(Groups[i], newFasta, false);
 195                         outputNames.push_back(newFasta); outputTypes["fasta"].push_back(newFasta);
 196                         if (m->control_pressed) { delete parser; for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]);        } return 0; }
 197
 198                         if (namefile != "") {
 199                                 parser->getNameMap(Groups[i], newName);
 200                                 outputNames.push_back(newName); outputTypes["name"].push_back(newName);
 201                         }
 202
 203                         if (m->control_pressed) { delete parser; for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]);        } return 0; }
 204                 }
 205
 206                 delete parser;
 207
 208                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);        } return 0; }
 209
 210                 string current = "";
 211                 itTypes = outputTypes.find("fasta");
 212                 if (itTypes != outputTypes.end()) {
 213                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 214                 }
 215
 216                 itTypes = outputTypes.find("name");
 217                 if (itTypes != outputTypes.end()) {
 218                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 219                 }
 220
 221                 m->mothurOutEndLine();
 222                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 223                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 224                 m->mothurOutEndLine();
 225
 226                 return 0;
 227         }
 228         catch(exception& e) {
 229                 m->errorOut(e, "SplitGroupCommand", "execute");
 230                 exit(1);
 231         }
 232 }
 233 //**********************************************************************************************************************
 234
 235