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 string SplitGroupCommand::getOutputFileNameTag(string type, string inputName=""){
  55         try {
  56         string outputFileName = "";
  57                 map<string, vector<string> >::iterator it;
  58
  59         //is this a type this command creates
  60         it = outputTypes.find(type);
  61         if (it == outputTypes.end()) {  m->mothurOut("[ERROR]: this command doesn't create a " + type + " output file.\n"); }
  62         else {
  63             if (type == "fasta")            {   outputFileName =  "fasta";   }
  64             else if (type == "name")        {   outputFileName =  "names";   }
  65             else { m->mothurOut("[ERROR]: No definition for type " + type + " output file tag.\n"); m->control_pressed = true;  }
  66         }
  67         return outputFileName;
  68         }
  69         catch(exception& e) {
  70                 m->errorOut(e, "SplitGroupCommand", "getOutputFileNameTag");
  71                 exit(1);
  72         }
  73 }
  74 //**********************************************************************************************************************
  75 SplitGroupCommand::SplitGroupCommand(){
  76         try {
  77                 abort = true; calledHelp = true;
  78                 setParameters();
  79                 vector<string> tempOutNames;
  80                 outputTypes["fasta"] = tempOutNames;
  81                 outputTypes["name"] = tempOutNames;
  82         }
  83         catch(exception& e) {
  84                 m->errorOut(e, "SplitGroupCommand", "SplitGroupCommand");
  85                 exit(1);
  86         }
  87 }
  88 //**********************************************************************************************************************
  89 SplitGroupCommand::SplitGroupCommand(string option)  {
  90         try {
  91                 abort = false; calledHelp = false;
  92
  93                 //allow user to run help
  94                 if(option == "help") { help(); abort = true; calledHelp = true; }
  95                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  96
  97                 else {
  98                         vector<string> myArray = setParameters();
  99
 100                         OptionParser parser(option);
 101                         map<string, string> parameters = parser.getParameters();
 102
 103                         ValidParameters validParameter;
 104                         map<string, string>::iterator it;
 105
 106                         //check to make sure all parameters are valid for command
 107                         for (it = parameters.begin(); it != parameters.end(); it++) {
 108                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
 109                         }
 110
 111                         //initialize outputTypes
 112                         vector<string> tempOutNames;
 113                         outputTypes["fasta"] = tempOutNames;
 114                         outputTypes["name"] = tempOutNames;
 115
 116                         //if the user changes the input directory command factory will send this info to us in the output parameter
 117                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 118                         if (inputDir == "not found"){   inputDir = "";          }
 119                         else {
 120                                 string path;
 121                                 it = parameters.find("group");
 122                                 //user has given a template file
 123                                 if(it != parameters.end()){
 124                                         path = m->hasPath(it->second);
 125                                         //if the user has not given a path then, add inputdir. else leave path alone.
 126                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 127                                 }
 128
 129                                 it = parameters.find("fasta");
 130                                 //user has given a template file
 131                                 if(it != parameters.end()){
 132                                         path = m->hasPath(it->second);
 133                                         //if the user has not given a path then, add inputdir. else leave path alone.
 134                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 135                                 }
 136
 137                                 it = parameters.find("name");
 138                                 //user has given a template file
 139                                 if(it != parameters.end()){
 140                                         path = m->hasPath(it->second);
 141                                         //if the user has not given a path then, add inputdir. else leave path alone.
 142                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 143                                 }
 144
 145                         }
 146
 147
 148                         namefile = validParameter.validFile(parameters, "name", true);
 149                         if (namefile == "not open") { namefile = ""; abort = true; }
 150                         else if (namefile == "not found") { namefile = ""; }
 151                         else { m->setNameFile(namefile); }
 152
 153                         fastafile = validParameter.validFile(parameters, "fasta", true);
 154                         if (fastafile == "not open") { abort = true; }
 155                         else if (fastafile == "not found") {
 156                                 fastafile = m->getFastaFile();
 157                                 if (fastafile != "") { m->mothurOut("Using " + fastafile + " as input file for the fasta parameter."); m->mothurOutEndLine(); }
 158                                 else {  m->mothurOut("You have no current fastafile and the fasta parameter is required."); m->mothurOutEndLine(); abort = true; }
 159                         }else { m->setFastaFile(fastafile); }
 160
 161                         groupfile = validParameter.validFile(parameters, "group", true);
 162                         if (groupfile == "not open") {  groupfile = ""; abort = true; }
 163                         else if (groupfile == "not found") {
 164                                 groupfile = m->getGroupFile();
 165                                 if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 166                                 else {  m->mothurOut("You have no current groupfile and the group parameter is required."); m->mothurOutEndLine(); abort = true; }
 167                         }else {  m->setGroupFile(groupfile); }
 168
 169                         groups = validParameter.validFile(parameters, "groups", false);
 170                         if (groups == "not found") { groups = ""; }
 171                         else { m->splitAtDash(groups, Groups);  }
 172
 173                         //if the user changes the output directory command factory will send this info to us in the output parameter
 174                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = m->hasPath(groupfile);      }
 175
 176                         if (namefile == "") {
 177                                 vector<string> files; files.push_back(fastafile);
 178                                 parser.getNameFile(files);
 179                         }
 180                 }
 181
 182         }
 183         catch(exception& e) {
 184                 m->errorOut(e, "SplitGroupCommand", "SplitAbundCommand");
 185                 exit(1);
 186         }
 187 }
 188 //**********************************************************************************************************************
 189 int SplitGroupCommand::execute(){
 190         try {
 191
 192                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 193
 194                 SequenceParser* parser;
 195
 196                 if (namefile == "") {   parser = new SequenceParser(groupfile, fastafile);                              }
 197                 else                            {       parser = new SequenceParser(groupfile, fastafile, namefile);    }
 198
 199                 if (m->control_pressed) { delete parser; return 0; }
 200
 201                 vector<string> namesGroups = parser->getNamesOfGroups();
 202                 SharedUtil util;  util.setGroups(Groups, namesGroups);
 203
 204                 string fastafileRoot = outputDir + m->getRootName(m->getSimpleName(fastafile));
 205                 string namefileRoot = outputDir + m->getRootName(m->getSimpleName(namefile));
 206
 207                 m->mothurOutEndLine();
 208                 for (int i = 0; i < Groups.size(); i++) {
 209
 210                         m->mothurOut("Processing group: " + Groups[i]); m->mothurOutEndLine();
 211
 212                         string newFasta = fastafileRoot + Groups[i] + "." + getOutputFileNameTag("fasta");
 213                         string newName = namefileRoot + Groups[i] + "." + getOutputFileNameTag("name");
 214
 215                         parser->getSeqs(Groups[i], newFasta, false);
 216                         outputNames.push_back(newFasta); outputTypes["fasta"].push_back(newFasta);
 217                         if (m->control_pressed) { delete parser; for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]);        } return 0; }
 218
 219                         if (namefile != "") {
 220                                 parser->getNameMap(Groups[i], newName);
 221                                 outputNames.push_back(newName); outputTypes["name"].push_back(newName);
 222                         }
 223
 224                         if (m->control_pressed) { delete parser; for (int i = 0; i < outputNames.size(); i++) { m->mothurRemove(outputNames[i]);        } return 0; }
 225                 }
 226
 227                 delete parser;
 228
 229                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]);        } return 0; }
 230
 231                 string current = "";
 232                 itTypes = outputTypes.find("fasta");
 233                 if (itTypes != outputTypes.end()) {
 234                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 235                 }
 236
 237                 itTypes = outputTypes.find("name");
 238                 if (itTypes != outputTypes.end()) {
 239                         if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 240                 }
 241
 242                 m->mothurOutEndLine();
 243                 m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 244                 for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 245                 m->mothurOutEndLine();
 246
 247                 return 0;
 248         }
 249         catch(exception& e) {
 250                 m->errorOut(e, "SplitGroupCommand", "execute");
 251                 exit(1);
 252         }
 253 }
 254 //**********************************************************************************************************************
 255
 256