getgroupscommand.cpp

   1 /*
   2  *  getgroupscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 11/10/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "getgroupscommand.h"
  11 #include "sequence.hpp"
  12 #include "listvector.hpp"
  13 #include "sharedutilities.h"
  14 #include "inputdata.h"
  15
  16 //**********************************************************************************************************************
  17 vector<string> GetGroupsCommand::setParameters(){
  18         try {
  19                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pfasta);
  20                 CommandParameter pshared("shared", "InputTypes", "", "", "none", "FNGLT-sharedGroup", "none",false,false); parameters.push_back(pshared);
  21                 CommandParameter pname("name", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pname);
  22                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "FNGLT-sharedGroup", "none",false,false); parameters.push_back(pgroup);
  23                 CommandParameter plist("list", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(plist);
  24                 CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(ptaxonomy);
  25                 CommandParameter paccnos("accnos", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(paccnos);
  26                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  27                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  28                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  29
  30                 vector<string> myArray;
  31                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  32                 return myArray;
  33         }
  34         catch(exception& e) {
  35                 m->errorOut(e, "GetGroupsCommand", "setParameters");
  36                 exit(1);
  37         }
  38 }
  39 //**********************************************************************************************************************
  40 string GetGroupsCommand::getHelpString(){
  41         try {
  42                 string helpString = "";
  43                 helpString += "The get.groups command selects sequences from a specfic group or set of groups from the following file types: fasta, name, group, list, taxonomy or shared file.\n";
  44                 helpString += "It outputs a file containing the sequences in the those specified groups, or a sharedfile containing only those groups.\n";
  45                 helpString += "The get.groups command parameters are accnos, fasta, name, group, list, taxonomy, shared and groups. The group parameter is required, unless you have a current group file, or are using a shared file.\n";
  46                 helpString += "You must also provide an accnos containing the list of groups to get or set the groups parameter to the groups you wish to select.\n";
  47                 helpString += "The groups parameter allows you to specify which of the groups in your groupfile you would like.  You can separate group names with dashes.\n";
  48                 helpString += "The get.groups command should be in the following format: get.groups(accnos=yourAccnos, fasta=yourFasta, group=yourGroupFile).\n";
  49                 helpString += "Example get.groups(accnos=amazon.accnos, fasta=amazon.fasta, group=amazon.groups).\n";
  50                 helpString += "or get.groups(groups=pasture, fasta=amazon.fasta, group=amazon.groups).\n";
  51                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  52                 return helpString;
  53         }
  54         catch(exception& e) {
  55                 m->errorOut(e, "GetGroupsCommand", "getHelpString");
  56                 exit(1);
  57         }
  58 }
  59 //**********************************************************************************************************************
  60 GetGroupsCommand::GetGroupsCommand(){
  61         try {
  62                 abort = true; calledHelp = true;
  63                 setParameters();
  64                 vector<string> tempOutNames;
  65                 outputTypes["fasta"] = tempOutNames;
  66                 outputTypes["taxonomy"] = tempOutNames;
  67                 outputTypes["name"] = tempOutNames;
  68                 outputTypes["group"] = tempOutNames;
  69                 outputTypes["list"] = tempOutNames;
  70                 outputTypes["shared"] = tempOutNames;
  71         }
  72         catch(exception& e) {
  73                 m->errorOut(e, "GetGroupsCommand", "GetGroupsCommand");
  74                 exit(1);
  75         }
  76 }
  77 //**********************************************************************************************************************
  78 GetGroupsCommand::GetGroupsCommand(string option)  {
  79         try {
  80                 abort = false; calledHelp = false;
  81
  82                 //allow user to run help
  83                 if(option == "help") { help(); abort = true; calledHelp = true; }
  84                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  85
  86                 else {
  87                         vector<string> myArray = setParameters();
  88
  89                         OptionParser parser(option);
  90                         map<string,string> parameters = parser.getParameters();
  91
  92                         ValidParameters validParameter;
  93                         map<string,string>::iterator it;
  94
  95                         //check to make sure all parameters are valid for command
  96                         for (it = parameters.begin(); it != parameters.end(); it++) {
  97                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  98                         }
  99
 100                         //initialize outputTypes
 101                         vector<string> tempOutNames;
 102                         outputTypes["fasta"] = tempOutNames;
 103                         outputTypes["taxonomy"] = tempOutNames;
 104                         outputTypes["name"] = tempOutNames;
 105                         outputTypes["group"] = tempOutNames;
 106                         outputTypes["list"] = tempOutNames;
 107                         outputTypes["shared"] = tempOutNames;
 108
 109
 110                         //if the user changes the output directory command factory will send this info to us in the output parameter
 111                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 112
 113                         //if the user changes the input directory command factory will send this info to us in the output parameter
 114                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 115                         if (inputDir == "not found"){   inputDir = "";          }
 116                         else {
 117                                 string path;
 118                                 it = parameters.find("fasta");
 119                                 //user has given a template file
 120                                 if(it != parameters.end()){
 121                                         path = m->hasPath(it->second);
 122                                         //if the user has not given a path then, add inputdir. else leave path alone.
 123                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 124                                 }
 125
 126                                 it = parameters.find("accnos");
 127                                 //user has given a template file
 128                                 if(it != parameters.end()){
 129                                         path = m->hasPath(it->second);
 130                                         //if the user has not given a path then, add inputdir. else leave path alone.
 131                                         if (path == "") {       parameters["accnos"] = inputDir + it->second;           }
 132                                 }
 133
 134                                 it = parameters.find("list");
 135                                 //user has given a template file
 136                                 if(it != parameters.end()){
 137                                         path = m->hasPath(it->second);
 138                                         //if the user has not given a path then, add inputdir. else leave path alone.
 139                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 140                                 }
 141
 142                                 it = parameters.find("name");
 143                                 //user has given a template file
 144                                 if(it != parameters.end()){
 145                                         path = m->hasPath(it->second);
 146                                         //if the user has not given a path then, add inputdir. else leave path alone.
 147                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 148                                 }
 149
 150                                 it = parameters.find("group");
 151                                 //user has given a template file
 152                                 if(it != parameters.end()){
 153                                         path = m->hasPath(it->second);
 154                                         //if the user has not given a path then, add inputdir. else leave path alone.
 155                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 156                                 }
 157
 158                                 it = parameters.find("taxonomy");
 159                                 //user has given a template file
 160                                 if(it != parameters.end()){
 161                                         path = m->hasPath(it->second);
 162                                         //if the user has not given a path then, add inputdir. else leave path alone.
 163                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 164                                 }
 165
 166                                 it = parameters.find("shared");
 167                                 //user has given a template file
 168                                 if(it != parameters.end()){
 169                                         path = m->hasPath(it->second);
 170                                         //if the user has not given a path then, add inputdir. else leave path alone.
 171                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 172                                 }
 173                         }
 174
 175
 176                         //check for required parameters
 177                         accnosfile = validParameter.validFile(parameters, "accnos", true);
 178                         if (accnosfile == "not open") { abort = true; }
 179                         else if (accnosfile == "not found") {  accnosfile = ""; }
 180                         else { m->setAccnosFile(accnosfile); }
 181
 182                         fastafile = validParameter.validFile(parameters, "fasta", true);
 183                         if (fastafile == "not open") { abort = true; }
 184                         else if (fastafile == "not found") {  fastafile = "";  }
 185                         else { m->setFastaFile(fastafile); }
 186
 187                         namefile = validParameter.validFile(parameters, "name", true);
 188                         if (namefile == "not open") { abort = true; }
 189                         else if (namefile == "not found") {  namefile = "";  }
 190                         else { m->setNameFile(namefile); }
 191
 192                         groupfile = validParameter.validFile(parameters, "group", true);
 193                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 194                         else if (groupfile == "not found") {  groupfile = "";                   }
 195                         else { m->setGroupFile(groupfile); }
 196
 197                         listfile = validParameter.validFile(parameters, "list", true);
 198                         if (listfile == "not open") { abort = true; }
 199                         else if (listfile == "not found") {  listfile = "";  }
 200                         else { m->setListFile(listfile); }
 201
 202                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 203                         if (taxfile == "not open") { abort = true; }
 204                         else if (taxfile == "not found") {  taxfile = "";  }
 205                         else { m->setTaxonomyFile(taxfile); }
 206
 207                         groups = validParameter.validFile(parameters, "groups", false);
 208                         if (groups == "not found") { groups = ""; }
 209                         else {
 210                                 m->splitAtDash(groups, Groups);
 211                                 m->setGroups(Groups);
 212                         }
 213
 214                         sharedfile = validParameter.validFile(parameters, "shared", true);
 215                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 216                         else if (sharedfile == "not found") {  sharedfile = "";  }
 217                         else { m->setSharedFile(sharedfile); }
 218
 219                         groupfile = validParameter.validFile(parameters, "group", true);
 220                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 221                         else if (groupfile == "not found") {    groupfile = ""; }
 222                         else { m->setGroupFile(groupfile); }
 223
 224                         if ((sharedfile == "") && (groupfile == "")) {
 225                                 //is there are current file available for any of these?
 226                                 if ((namefile != "") || (fastafile != "") || (listfile != "") || (taxfile != "")) {
 227                                         //give priority to group, then shared
 228                                         groupfile = m->getGroupFile();
 229                                         if (groupfile != "") {  m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 230                                         else {
 231                                                 sharedfile = m->getSharedFile();
 232                                                 if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 233                                                 else {
 234                                                         m->mothurOut("You have no current groupfile or sharedfile and one is required."); m->mothurOutEndLine(); abort = true;
 235                                                 }
 236                                         }
 237                                 }else {
 238                                         //give priority to shared, then group
 239                                         sharedfile = m->getSharedFile();
 240                                         if (sharedfile != "") {  m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 241                                         else {
 242                                                 groupfile = m->getGroupFile();
 243                                                 if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 244                                                 else {
 245                                                         m->mothurOut("You have no current groupfile or sharedfile and one is required."); m->mothurOutEndLine(); abort = true;
 246                                                 }
 247                                         }
 248                                 }
 249                         }
 250
 251                         if ((accnosfile == "") && (Groups.size() == 0)) { m->mothurOut("You must provide an accnos file or specify groups using the groups parameter."); m->mothurOutEndLine(); abort = true; }
 252
 253                         if ((fastafile == "") && (namefile == "") && (groupfile == "")  && (sharedfile == "") && (listfile == "") && (taxfile == ""))  { m->mothurOut("You must provide at least one of the following: fasta, name, taxonomy, group, shared or list."); m->mothurOutEndLine(); abort = true; }
 254                         if ((groupfile == "") && ((namefile != "") || (fastafile != "") || (listfile != "") || (taxfile != "")))  { m->mothurOut("If using a fasta, name, taxonomy, group or list, then you must provide a group file."); m->mothurOutEndLine(); abort = true; }
 255
 256                 }
 257
 258         }
 259         catch(exception& e) {
 260                 m->errorOut(e, "GetGroupsCommand", "GetGroupsCommand");
 261                 exit(1);
 262         }
 263 }
 264 //**********************************************************************************************************************
 265
 266 int GetGroupsCommand::execute(){
 267         try {
 268
 269                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 270
 271                 //get groups you want to remove
 272                 if (accnosfile != "") { readAccnos(); }
 273
 274                 if (groupfile != "") {
 275                         groupMap = new GroupMap(groupfile);
 276                         groupMap->readMap();
 277
 278                         //make sure groups are valid
 279                         //takes care of user setting groupNames that are invalid or setting groups=all
 280                         SharedUtil* util = new SharedUtil();
 281                         vector<string> gNamesOfGroups = groupMap->getNamesOfGroups();
 282                         util->setGroups(Groups, gNamesOfGroups);
 283                         groupMap->setNamesOfGroups(gNamesOfGroups);
 284                         delete util;
 285
 286                         //fill names with names of sequences that are from the groups we want to remove
 287                         fillNames();
 288
 289                         delete groupMap;
 290                 }
 291
 292                 if (m->control_pressed) { return 0; }
 293
 294                 //read through the correct file and output lines you want to keep
 295                 if (namefile != "")                     {               readName();             }
 296                 if (fastafile != "")            {               readFasta();    }
 297                 if (groupfile != "")            {               readGroup();    }
 298                 if (listfile != "")                     {               readList();             }
 299                 if (taxfile != "")                      {               readTax();              }
 300                 if (sharedfile != "")           {               readShared();   }
 301
 302                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); } return 0; }
 303
 304
 305                 if (outputNames.size() != 0) {
 306                         m->mothurOutEndLine();
 307                         m->mothurOut("Output File names: "); m->mothurOutEndLine();
 308                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 309                         m->mothurOutEndLine();
 310
 311                         //set fasta file as new current fastafile
 312                         string current = "";
 313                         itTypes = outputTypes.find("fasta");
 314                         if (itTypes != outputTypes.end()) {
 315                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 316                         }
 317
 318                         itTypes = outputTypes.find("name");
 319                         if (itTypes != outputTypes.end()) {
 320                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 321                         }
 322
 323                         itTypes = outputTypes.find("group");
 324                         if (itTypes != outputTypes.end()) {
 325                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 326                         }
 327
 328                         itTypes = outputTypes.find("list");
 329                         if (itTypes != outputTypes.end()) {
 330                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 331                         }
 332
 333                         itTypes = outputTypes.find("taxonomy");
 334                         if (itTypes != outputTypes.end()) {
 335                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 336                         }
 337
 338                         itTypes = outputTypes.find("shared");
 339                         if (itTypes != outputTypes.end()) {
 340                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSharedFile(current); }
 341                         }
 342                 }
 343
 344                 return 0;
 345         }
 346
 347         catch(exception& e) {
 348                 m->errorOut(e, "GetGroupsCommand", "execute");
 349                 exit(1);
 350         }
 351 }
 352
 353 //**********************************************************************************************************************
 354 int GetGroupsCommand::readFasta(){
 355         try {
 356                 string thisOutputDir = outputDir;
 357                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 358                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
 359
 360                 ofstream out;
 361                 m->openOutputFile(outputFileName, out);
 362
 363                 ifstream in;
 364                 m->openInputFile(fastafile, in);
 365                 string name;
 366
 367                 bool wroteSomething = false;
 368                 int selectedCount = 0;
 369
 370                 while(!in.eof()){
 371                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 372
 373                         Sequence currSeq(in);
 374                         name = currSeq.getName();
 375
 376                         if (name != "") {
 377                                 //if this name is in the accnos file
 378                                 if (names.count(name) != 0) {
 379                                         wroteSomething = true;
 380
 381                                         currSeq.printSequence(out);
 382                                         selectedCount++;
 383                                 }else{
 384                                         //if you are not in the accnos file check if you are a name that needs to be changed
 385                                         map<string, string>::iterator it = uniqueToRedundant.find(name);
 386                                         if (it != uniqueToRedundant.end()) {
 387                                                 wroteSomething = true;
 388                                                 currSeq.setName(it->second);
 389                                                 currSeq.printSequence(out);
 390                                                 selectedCount++;
 391                                         }
 392                                 }
 393                         }
 394                         m->gobble(in);
 395                 }
 396                 in.close();
 397                 out.close();
 398
 399                 if (wroteSomething == false) {  m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine();  }
 400                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 401
 402                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
 403
 404
 405                 return 0;
 406
 407         }
 408         catch(exception& e) {
 409                 m->errorOut(e, "GetGroupsCommand", "readFasta");
 410                 exit(1);
 411         }
 412 }
 413 //**********************************************************************************************************************
 414 int GetGroupsCommand::readShared(){
 415         try {
 416                 string thisOutputDir = outputDir;
 417                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 418
 419                 InputData input(sharedfile, "sharedfile");
 420                 vector<SharedRAbundVector*> lookup = input.getSharedRAbundVectors();
 421
 422                 bool wroteSomething = false;
 423
 424                 while(lookup[0] != NULL) {
 425
 426                         string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + lookup[0]->getLabel() + ".pick" + m->getExtension(sharedfile);
 427                         ofstream out;
 428                         m->openOutputFile(outputFileName, out);
 429                         outputTypes["shared"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 430
 431                         if (m->control_pressed) { out.close();  m->mothurRemove(outputFileName);  for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } return 0; }
 432
 433                         lookup[0]->printHeaders(out);
 434
 435                         for (int i = 0; i < lookup.size(); i++) {
 436                                 out << lookup[i]->getLabel() << '\t' << lookup[i]->getGroup() << '\t';
 437                                 lookup[i]->print(out);
 438                                 wroteSomething = true;
 439
 440                         }
 441
 442                         //get next line to process
 443                         //prevent memory leak
 444                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 445                         lookup = input.getSharedRAbundVectors();
 446
 447                         out.close();
 448                 }
 449
 450                 if (wroteSomething == false) {  m->mothurOut("Your file contains only the groups you wish to remove."); m->mothurOutEndLine();  }
 451
 452                 string groupsString = "";
 453                 for (int i = 0; i < Groups.size()-1; i++) {     groupsString += Groups[i] + ", "; }
 454                 groupsString += Groups[Groups.size()-1];
 455
 456                 m->mothurOut("Selected groups: " + groupsString + " from your shared file."); m->mothurOutEndLine();
 457
 458                 return 0;
 459
 460         }
 461         catch(exception& e) {
 462                 m->errorOut(e, "GetGroupsCommand", "readShared");
 463                 exit(1);
 464         }
 465 }
 466 //**********************************************************************************************************************
 467 int GetGroupsCommand::readList(){
 468         try {
 469                 string thisOutputDir = outputDir;
 470                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 471                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 472
 473                 ofstream out;
 474                 m->openOutputFile(outputFileName, out);
 475
 476                 ifstream in;
 477                 m->openInputFile(listfile, in);
 478
 479                 bool wroteSomething = false;
 480                 int selectedCount = 0;
 481
 482                 while(!in.eof()){
 483
 484                         selectedCount = 0;
 485
 486                         //read in list vector
 487                         ListVector list(in);
 488
 489                         //make a new list vector
 490                         ListVector newList;
 491                         newList.setLabel(list.getLabel());
 492
 493                         //for each bin
 494                         for (int i = 0; i < list.getNumBins(); i++) {
 495                                 if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 496
 497                                 //parse out names that are in accnos file
 498                                 string binnames = list.get(i);
 499
 500                                 string newNames = "";
 501                                 while (binnames.find_first_of(',') != -1) {
 502                                         string name = binnames.substr(0,binnames.find_first_of(','));
 503                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 504
 505                                         //if that name is in the .accnos file, add it
 506                                         if (names.count(name) != 0) {  newNames += name + ",";  selectedCount++;  }
 507                                         else{
 508                                                 //if you are not in the accnos file check if you are a name that needs to be changed
 509                                                 map<string, string>::iterator it = uniqueToRedundant.find(name);
 510                                                 if (it != uniqueToRedundant.end()) {
 511                                                         newNames += it->second + ",";
 512                                                         selectedCount++;
 513                                                 }
 514                                         }
 515                                 }
 516
 517                                 //get last name
 518                                 if (names.count(binnames) != 0) {  newNames += binnames + ",";  selectedCount++;  }
 519                                 else{
 520                                         //if you are not in the accnos file check if you are a name that needs to be changed
 521                                         map<string, string>::iterator it = uniqueToRedundant.find(binnames);
 522                                         if (it != uniqueToRedundant.end()) {
 523                                                 newNames += it->second + ",";
 524                                                 selectedCount++;
 525                                         }
 526                                 }
 527
 528                                 //if there are names in this bin add to new list
 529                                 if (newNames != "") {
 530                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 531                                         newList.push_back(newNames);
 532                                 }
 533                         }
 534
 535                         //print new listvector
 536                         if (newList.getNumBins() != 0) {
 537                                 wroteSomething = true;
 538                                 newList.print(out);
 539                         }
 540
 541                         m->gobble(in);
 542                 }
 543                 in.close();
 544                 out.close();
 545
 546                 if (wroteSomething == false) {  m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine();  }
 547                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 548
 549                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your list file."); m->mothurOutEndLine();
 550
 551                 return 0;
 552
 553         }
 554         catch(exception& e) {
 555                 m->errorOut(e, "GetGroupsCommand", "readList");
 556                 exit(1);
 557         }
 558 }
 559 //**********************************************************************************************************************
 560 int GetGroupsCommand::readName(){
 561         try {
 562                 string thisOutputDir = outputDir;
 563                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 564                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
 565
 566                 ofstream out;
 567                 m->openOutputFile(outputFileName, out);
 568
 569                 ifstream in;
 570                 m->openInputFile(namefile, in);
 571                 string name, firstCol, secondCol;
 572
 573                 bool wroteSomething = false;
 574                 int selectedCount = 0;
 575
 576                 while(!in.eof()){
 577                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 578
 579                         in >> firstCol;         m->gobble(in);
 580                         in >> secondCol;
 581
 582                         vector<string> parsedNames;
 583                         m->splitAtComma(secondCol, parsedNames);
 584
 585                         vector<string> validSecond;  validSecond.clear();
 586                         for (int i = 0; i < parsedNames.size(); i++) {
 587                                 if (names.count(parsedNames[i]) != 0) {
 588                                         validSecond.push_back(parsedNames[i]);
 589                                 }
 590                         }
 591
 592                         selectedCount += validSecond.size();
 593
 594                         //if the name in the first column is in the set then print it and any other names in second column also in set
 595                         if (names.count(firstCol) != 0) {
 596
 597                                 wroteSomething = true;
 598
 599                                 out << firstCol << '\t';
 600
 601                                 //you know you have at least one valid second since first column is valid
 602                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 603                                 out << validSecond[validSecond.size()-1] << endl;
 604
 605                                 //make first name in set you come to first column and then add the remaining names to second column
 606                         }else {
 607
 608                                 //you want part of this row
 609                                 if (validSecond.size() != 0) {
 610
 611                                         wroteSomething = true;
 612
 613                                         out << validSecond[0] << '\t';
 614
 615                                         //you know you have at least one valid second since first column is valid
 616                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 617                                         out << validSecond[validSecond.size()-1] << endl;
 618                                         uniqueToRedundant[firstCol] = validSecond[0];
 619                                 }
 620                         }
 621
 622                         m->gobble(in);
 623                 }
 624                 in.close();
 625                 out.close();
 626
 627                 if (wroteSomething == false) {  m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine();  }
 628                 outputTypes["name"].push_back(outputFileName); outputNames.push_back(outputFileName);
 629
 630                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your name file."); m->mothurOutEndLine();
 631
 632                 return 0;
 633         }
 634         catch(exception& e) {
 635                 m->errorOut(e, "GetGroupsCommand", "readName");
 636                 exit(1);
 637         }
 638 }
 639
 640 //**********************************************************************************************************************
 641 int GetGroupsCommand::readGroup(){
 642         try {
 643                 string thisOutputDir = outputDir;
 644                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 645                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 646
 647                 ofstream out;
 648                 m->openOutputFile(outputFileName, out);
 649
 650                 ifstream in;
 651                 m->openInputFile(groupfile, in);
 652                 string name, group;
 653
 654                 bool wroteSomething = false;
 655                 int selectedCount = 0;
 656
 657                 while(!in.eof()){
 658                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 659
 660                         in >> name;                             //read from first column
 661                         in >> group;                    //read from second column
 662
 663                         //if this name is in the accnos file
 664                         if (names.count(name) != 0) {
 665                                 wroteSomething = true;
 666                                 out << name << '\t' << group << endl;
 667                                 selectedCount++;
 668                         }
 669
 670                         m->gobble(in);
 671                 }
 672                 in.close();
 673                 out.close();
 674
 675                 if (wroteSomething == false) {  m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine();  }
 676                 outputTypes["group"].push_back(outputFileName); outputNames.push_back(outputFileName);
 677
 678                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your group file."); m->mothurOutEndLine();
 679
 680                 return 0;
 681         }
 682         catch(exception& e) {
 683                 m->errorOut(e, "GetGroupsCommand", "readGroup");
 684                 exit(1);
 685         }
 686 }
 687 //**********************************************************************************************************************
 688 int GetGroupsCommand::readTax(){
 689         try {
 690                 string thisOutputDir = outputDir;
 691                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 692                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 693                 ofstream out;
 694                 m->openOutputFile(outputFileName, out);
 695
 696                 ifstream in;
 697                 m->openInputFile(taxfile, in);
 698                 string name, tax;
 699
 700                 bool wroteSomething = false;
 701
 702                 while(!in.eof()){
 703                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 704
 705                         in >> name;                             //read from first column
 706                         in >> tax;                      //read from second column
 707
 708                         //if this name is in the accnos file
 709                         if (names.count(name) != 0) {
 710                                 wroteSomething = true;
 711                                 out << name << '\t' << tax << endl;
 712                         }else{
 713                                 //if you are not in the accnos file check if you are a name that needs to be changed
 714                                 map<string, string>::iterator it = uniqueToRedundant.find(name);
 715                                 if (it != uniqueToRedundant.end()) {
 716                                         wroteSomething = true;
 717                                         out << it->second << '\t' << tax << endl;
 718                                 }
 719                         }
 720
 721                         m->gobble(in);
 722                 }
 723                 in.close();
 724                 out.close();
 725
 726                 if (wroteSomething == false) {  m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine();  }
 727                 outputTypes["taxonomy"].push_back(outputFileName); outputNames.push_back(outputFileName);
 728
 729                 return 0;
 730         }
 731         catch(exception& e) {
 732                 m->errorOut(e, "GetGroupsCommand", "readTax");
 733                 exit(1);
 734         }
 735 }
 736 //**********************************************************************************************************************
 737 void GetGroupsCommand::readAccnos(){
 738         try {
 739                 Groups.clear();
 740
 741                 ifstream in;
 742                 m->openInputFile(accnosfile, in);
 743                 string name;
 744
 745                 while(!in.eof()){
 746                         in >> name;
 747
 748                         Groups.push_back(name);
 749
 750                         m->gobble(in);
 751                 }
 752                 in.close();
 753
 754                 m->setGroups(Groups);
 755
 756         }
 757         catch(exception& e) {
 758                 m->errorOut(e, "GetGroupsCommand", "readAccnos");
 759                 exit(1);
 760         }
 761 }
 762 //**********************************************************************************************************************
 763 int GetGroupsCommand::fillNames(){
 764         try {
 765                 vector<string> seqs = groupMap->getNamesSeqs();
 766
 767                 for (int i = 0; i < seqs.size(); i++) {
 768
 769                         if (m->control_pressed) { return 0; }
 770
 771                         string group = groupMap->getGroup(seqs[i]);
 772
 773                         if (m->inUsersGroups(group, Groups)) {
 774                                 names.insert(seqs[i]);
 775                         }
 776                 }
 777
 778                 return 0;
 779         }
 780         catch(exception& e) {
 781                 m->errorOut(e, "GetGroupsCommand", "fillNames");
 782                 exit(1);
 783         }
 784 }
 785
 786 //**********************************************************************************************************************
 787
 788