getseqscommand.cpp

   1 /*
   2  *  getseqscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 7/8/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "getseqscommand.h"
  11 #include "sequence.hpp"
  12 #include "listvector.hpp"
  13
  14 //**********************************************************************************************************************
  15 vector<string> GetSeqsCommand::getValidParameters(){
  16         try {
  17                 string Array[] =  {"fasta","name", "group", "qfile","alignreport", "accnos", "dups", "list","taxonomy","outputdir","inputdir"};
  18                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  19                 return myArray;
  20         }
  21         catch(exception& e) {
  22                 m->errorOut(e, "GetSeqsCommand", "getValidParameters");
  23                 exit(1);
  24         }
  25 }
  26 //**********************************************************************************************************************
  27 GetSeqsCommand::GetSeqsCommand(){
  28         try {
  29                 abort = true;
  30                 //initialize outputTypes
  31                 vector<string> tempOutNames;
  32                 outputTypes["fasta"] = tempOutNames;
  33                 outputTypes["taxonomy"] = tempOutNames;
  34                 outputTypes["name"] = tempOutNames;
  35                 outputTypes["group"] = tempOutNames;
  36                 outputTypes["alignreport"] = tempOutNames;
  37                 outputTypes["list"] = tempOutNames;
  38                 outputTypes["qfile"] = tempOutNames;
  39         }
  40         catch(exception& e) {
  41                 m->errorOut(e, "GetSeqsCommand", "GetSeqsCommand");
  42                 exit(1);
  43         }
  44 }
  45 //**********************************************************************************************************************
  46 vector<string> GetSeqsCommand::getRequiredParameters(){
  47         try {
  48                 string Array[] =  {"accnos"};
  49                 vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  50                 return myArray;
  51         }
  52         catch(exception& e) {
  53                 m->errorOut(e, "GetSeqsCommand", "getRequiredParameters");
  54                 exit(1);
  55         }
  56 }
  57 //**********************************************************************************************************************
  58 vector<string> GetSeqsCommand::getRequiredFiles(){
  59         try {
  60                 vector<string> myArray;
  61                 return myArray;
  62         }
  63         catch(exception& e) {
  64                 m->errorOut(e, "GetSeqsCommand", "getRequiredFiles");
  65                 exit(1);
  66         }
  67 }
  68 //**********************************************************************************************************************
  69 GetSeqsCommand::GetSeqsCommand(string option)  {
  70         try {
  71                 abort = false;
  72
  73                 //allow user to run help
  74                 if(option == "help") { help(); abort = true; }
  75
  76                 else {
  77                         //valid paramters for this command
  78                         string Array[] =  {"fasta","name", "group", "alignreport", "qfile", "accnos", "dups", "list","taxonomy","outputdir","inputdir"};
  79                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  80
  81                         OptionParser parser(option);
  82                         map<string,string> parameters = parser.getParameters();
  83
  84                         ValidParameters validParameter;
  85                         map<string,string>::iterator it;
  86
  87                         //check to make sure all parameters are valid for command
  88                         for (it = parameters.begin(); it != parameters.end(); it++) {
  89                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  90                         }
  91
  92                         //initialize outputTypes
  93                         vector<string> tempOutNames;
  94                         outputTypes["fasta"] = tempOutNames;
  95                         outputTypes["taxonomy"] = tempOutNames;
  96                         outputTypes["name"] = tempOutNames;
  97                         outputTypes["group"] = tempOutNames;
  98                         outputTypes["alignreport"] = tempOutNames;
  99                         outputTypes["list"] = tempOutNames;
 100                         outputTypes["qfile"] = tempOutNames;
 101
 102                         //if the user changes the output directory command factory will send this info to us in the output parameter
 103                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 104
 105                         //if the user changes the input directory command factory will send this info to us in the output parameter
 106                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 107                         if (inputDir == "not found"){   inputDir = "";          }
 108                         else {
 109                                 string path;
 110                                 it = parameters.find("alignreport");
 111                                 //user has given a template file
 112                                 if(it != parameters.end()){
 113                                         path = m->hasPath(it->second);
 114                                         //if the user has not given a path then, add inputdir. else leave path alone.
 115                                         if (path == "") {       parameters["alignreport"] = inputDir + it->second;              }
 116                                 }
 117
 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("qfile");
 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["qfile"] = 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 = "";  m->mothurOut("You must provide an accnos file."); m->mothurOutEndLine(); abort = true; }
 180
 181                         fastafile = validParameter.validFile(parameters, "fasta", true);
 182                         if (fastafile == "not open") { abort = true; }
 183                         else if (fastafile == "not found") {  fastafile = "";  }
 184
 185                         namefile = validParameter.validFile(parameters, "name", true);
 186                         if (namefile == "not open") { abort = true; }
 187                         else if (namefile == "not found") {  namefile = "";  }
 188
 189                         groupfile = validParameter.validFile(parameters, "group", true);
 190                         if (groupfile == "not open") { abort = true; }
 191                         else if (groupfile == "not found") {  groupfile = "";  }
 192
 193                         alignfile = validParameter.validFile(parameters, "alignreport", true);
 194                         if (alignfile == "not open") { abort = true; }
 195                         else if (alignfile == "not found") {  alignfile = "";  }
 196
 197                         listfile = validParameter.validFile(parameters, "list", true);
 198                         if (listfile == "not open") { abort = true; }
 199                         else if (listfile == "not found") {  listfile = "";  }
 200
 201                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 202                         if (taxfile == "not open") { abort = true; }
 203                         else if (taxfile == "not found") {  taxfile = "";  }
 204
 205                         qualfile = validParameter.validFile(parameters, "qfile", true);
 206                         if (qualfile == "not open") { abort = true; }
 207                         else if (qualfile == "not found") {  qualfile = "";  }
 208
 209                         string usedDups = "true";
 210                         string temp = validParameter.validFile(parameters, "dups", false);      if (temp == "not found") { temp = "false"; usedDups = ""; }
 211                         dups = m->isTrue(temp);
 212
 213                         if ((fastafile == "") && (namefile == "") && (groupfile == "") && (alignfile == "") && (listfile == "") && (taxfile == "") && (qualfile == ""))  { m->mothurOut("You must provide one of the following: fasta, name, group, alignreport, taxonomy, quality or listfile."); m->mothurOutEndLine(); abort = true; }
 214
 215                         if ((usedDups != "") && (namefile == "")) {  m->mothurOut("You may only use dups with the name option."); m->mothurOutEndLine();  abort = true; }
 216
 217                 }
 218
 219         }
 220         catch(exception& e) {
 221                 m->errorOut(e, "GetSeqsCommand", "GetSeqsCommand");
 222                 exit(1);
 223         }
 224 }
 225 //**********************************************************************************************************************
 226
 227 void GetSeqsCommand::help(){
 228         try {
 229                 m->mothurOut("The get.seqs command reads an .accnos file and any of the following file types: fasta, name, group, list, taxonomy, quality or alignreport file.\n");
 230                 m->mothurOut("It outputs a file containing only the sequences in the .accnos file.\n");
 231                 m->mothurOut("The get.seqs command parameters are accnos, fasta, name, group, list, taxonomy, qfile, alignreport and dups.  You must provide accnos and at least one of the other parameters.\n");
 232                 m->mothurOut("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");
 233                 m->mothurOut("The get.seqs command should be in the following format: get.seqs(accnos=yourAccnos, fasta=yourFasta).\n");
 234                 m->mothurOut("Example get.seqs(accnos=amazon.accnos, fasta=amazon.fasta).\n");
 235                 m->mothurOut("Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n\n");
 236         }
 237         catch(exception& e) {
 238                 m->errorOut(e, "GetSeqsCommand", "help");
 239                 exit(1);
 240         }
 241 }
 242
 243 //**********************************************************************************************************************
 244
 245 int GetSeqsCommand::execute(){
 246         try {
 247
 248                 if (abort == true) { return 0; }
 249
 250                 //get names you want to keep
 251                 readAccnos();
 252
 253                 if (m->control_pressed) { return 0; }
 254
 255                 //read through the correct file and output lines you want to keep
 256                 if (namefile != "")                     {               readName();             }
 257                 if (fastafile != "")            {               readFasta();    }
 258                 if (groupfile != "")            {               readGroup();    }
 259                 if (alignfile != "")            {               readAlign();    }
 260                 if (listfile != "")                     {               readList();             }
 261                 if (taxfile != "")                      {               readTax();              }
 262                 if (qualfile != "")                     {               readQual();             }
 263
 264                 if (m->control_pressed) { outputTypes.clear(); for (int i = 0; i < outputNames.size(); i++) {   remove(outputNames[i].c_str());  } return 0; }
 265
 266                 if (outputNames.size() != 0) {
 267                         m->mothurOutEndLine();
 268                         m->mothurOut("Output File Names: "); m->mothurOutEndLine();
 269                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 270                         m->mothurOutEndLine();
 271                 }
 272
 273                 return 0;
 274         }
 275
 276         catch(exception& e) {
 277                 m->errorOut(e, "GetSeqsCommand", "execute");
 278                 exit(1);
 279         }
 280 }
 281
 282 //**********************************************************************************************************************
 283 int GetSeqsCommand::readFasta(){
 284         try {
 285                 string thisOutputDir = outputDir;
 286                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 287                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" +  m->getExtension(fastafile);
 288                 ofstream out;
 289                 m->openOutputFile(outputFileName, out);
 290
 291
 292                 ifstream in;
 293                 m->openInputFile(fastafile, in);
 294                 string name;
 295
 296                 bool wroteSomething = false;
 297
 298                 while(!in.eof()){
 299
 300                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 301
 302                         Sequence currSeq(in);
 303                         name = currSeq.getName();
 304
 305                         if (name != "") {
 306                                 //if this name is in the accnos file
 307                                 if (names.count(name) != 0) {
 308                                         wroteSomething = true;
 309
 310                                         currSeq.printSequence(out);
 311                                 }
 312                         }
 313                         m->gobble(in);
 314                 }
 315                 in.close();
 316                 out.close();
 317
 318
 319                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 320                 outputNames.push_back(outputFileName);  outputTypes["fasta"].push_back(outputFileName);
 321
 322                 return 0;
 323
 324         }
 325         catch(exception& e) {
 326                 m->errorOut(e, "GetSeqsCommand", "readFasta");
 327                 exit(1);
 328         }
 329 }
 330 //**********************************************************************************************************************
 331 int GetSeqsCommand::readQual(){
 332         try {
 333                 string thisOutputDir = outputDir;
 334                 if (outputDir == "") {  thisOutputDir += m->hasPath(qualfile);  }
 335                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(qualfile)) + "pick" +  m->getExtension(qualfile);
 336                 ofstream out;
 337                 m->openOutputFile(outputFileName, out);
 338
 339
 340                 ifstream in;
 341                 m->openInputFile(qualfile, in);
 342                 string name;
 343
 344                 bool wroteSomething = false;
 345
 346
 347                 while(!in.eof()){
 348                         string saveName = "";
 349                         string name = "";
 350                         string scores = "";
 351
 352                         in >> name;
 353
 354                         if (name.length() != 0) {
 355                                 saveName = name.substr(1);
 356                                 while (!in.eof())       {
 357                                         char c = in.get();
 358                                         if (c == 10 || c == 13){        break;  }
 359                                         else { name += c; }
 360                                 }
 361                                 m->gobble(in);
 362                         }
 363
 364                         while(in){
 365                                 char letter= in.get();
 366                                 if(letter == '>'){      in.putback(letter);     break;  }
 367                                 else{ scores += letter; }
 368                         }
 369
 370                         m->gobble(in);
 371
 372                         if (names.count(saveName) != 0) {
 373                                 wroteSomething = true;
 374
 375                                 out << name << endl << scores;
 376                         }
 377
 378                         m->gobble(in);
 379                 }
 380                 in.close();
 381                 out.close();
 382
 383
 384                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 385                 outputNames.push_back(outputFileName);  outputTypes["qfile"].push_back(outputFileName);
 386
 387                 return 0;
 388
 389         }
 390         catch(exception& e) {
 391                 m->errorOut(e, "GetSeqsCommand", "readQual");
 392                 exit(1);
 393         }
 394 }
 395 //**********************************************************************************************************************
 396 int GetSeqsCommand::readList(){
 397         try {
 398                 string thisOutputDir = outputDir;
 399                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 400                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 401                 ofstream out;
 402                 m->openOutputFile(outputFileName, out);
 403
 404                 ifstream in;
 405                 m->openInputFile(listfile, in);
 406
 407                 bool wroteSomething = false;
 408
 409                 while(!in.eof()){
 410
 411                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 412
 413                         //read in list vector
 414                         ListVector list(in);
 415
 416                         //make a new list vector
 417                         ListVector newList;
 418                         newList.setLabel(list.getLabel());
 419
 420                         //for each bin
 421                         for (int i = 0; i < list.getNumBins(); i++) {
 422
 423                                 //parse out names that are in accnos file
 424                                 string binnames = list.get(i);
 425
 426                                 string newNames = "";
 427                                 while (binnames.find_first_of(',') != -1) {
 428                                         string name = binnames.substr(0,binnames.find_first_of(','));
 429                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 430
 431                                         //if that name is in the .accnos file, add it
 432                                         if (names.count(name) != 0) {  newNames += name + ",";  }
 433                                 }
 434
 435                                 //get last name
 436                                 if (names.count(binnames) != 0) {  newNames += binnames + ",";  }
 437
 438                                 //if there are names in this bin add to new list
 439                                 if (newNames != "") {
 440                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 441                                         newList.push_back(newNames);
 442                                 }
 443                         }
 444
 445                         //print new listvector
 446                         if (newList.getNumBins() != 0) {
 447                                 wroteSomething = true;
 448                                 newList.print(out);
 449                         }
 450
 451                         m->gobble(in);
 452                 }
 453                 in.close();
 454                 out.close();
 455
 456                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 457                 outputNames.push_back(outputFileName); outputTypes["list"].push_back(outputFileName);
 458
 459                 return 0;
 460
 461         }
 462         catch(exception& e) {
 463                 m->errorOut(e, "GetSeqsCommand", "readList");
 464                 exit(1);
 465         }
 466 }
 467 //**********************************************************************************************************************
 468 int GetSeqsCommand::readName(){
 469         try {
 470                 string thisOutputDir = outputDir;
 471                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 472                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" +  m->getExtension(namefile);
 473                 ofstream out;
 474                 m->openOutputFile(outputFileName, out);
 475
 476
 477                 ifstream in;
 478                 m->openInputFile(namefile, in);
 479                 string name, firstCol, secondCol;
 480
 481                 bool wroteSomething = false;
 482
 483
 484                 while(!in.eof()){
 485
 486                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 487
 488                         in >> firstCol;
 489                         in >> secondCol;
 490
 491                         string hold = "";
 492                         if (dups) { hold = secondCol; }
 493
 494                         vector<string> parsedNames;
 495                         //parse second column saving each name
 496                         while (secondCol.find_first_of(',') != -1) {
 497                                 name = secondCol.substr(0,secondCol.find_first_of(','));
 498                                 secondCol = secondCol.substr(secondCol.find_first_of(',')+1, secondCol.length());
 499                                 parsedNames.push_back(name);
 500                         }
 501
 502                         //get name after last ,
 503                         parsedNames.push_back(secondCol);
 504
 505                         vector<string> validSecond;
 506                         for (int i = 0; i < parsedNames.size(); i++) {
 507                                 if (names.count(parsedNames[i]) != 0) {
 508                                         validSecond.push_back(parsedNames[i]);
 509                                 }
 510                         }
 511
 512                         if ((dups) && (validSecond.size() != 0)) { //dups = true and we want to add someone, then add everyone
 513                                 for (int i = 0; i < parsedNames.size(); i++) {  names.insert(parsedNames[i]);  }
 514                                 out << firstCol << '\t' << hold << endl;
 515                                 wroteSomething = true;
 516                         }else {
 517                                 //if the name in the first column is in the set then print it and any other names in second column also in set
 518                                 if (names.count(firstCol) != 0) {
 519
 520                                         wroteSomething = true;
 521
 522                                         out << firstCol << '\t';
 523
 524                                         //you know you have at least one valid second since first column is valid
 525                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 526                                         out << validSecond[validSecond.size()-1] << endl;
 527
 528
 529                                 //make first name in set you come to first column and then add the remaining names to second column
 530                                 }else {
 531                                         //you want part of this row
 532                                         if (validSecond.size() != 0) {
 533
 534                                                 wroteSomething = true;
 535
 536                                                 out << validSecond[0] << '\t';
 537
 538                                                 //you know you have at least one valid second since first column is valid
 539                                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 540                                                 out << validSecond[validSecond.size()-1] << endl;
 541                                         }
 542                                 }
 543                         }
 544                         m->gobble(in);
 545                 }
 546                 in.close();
 547                 out.close();
 548
 549                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 550                 outputNames.push_back(outputFileName); outputTypes["name"].push_back(outputFileName);
 551
 552                 return 0;
 553
 554         }
 555         catch(exception& e) {
 556                 m->errorOut(e, "GetSeqsCommand", "readName");
 557                 exit(1);
 558         }
 559 }
 560
 561 //**********************************************************************************************************************
 562 int GetSeqsCommand::readGroup(){
 563         try {
 564                 string thisOutputDir = outputDir;
 565                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 566                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 567                 ofstream out;
 568                 m->openOutputFile(outputFileName, out);
 569
 570
 571                 ifstream in;
 572                 m->openInputFile(groupfile, in);
 573                 string name, group;
 574
 575                 bool wroteSomething = false;
 576
 577                 while(!in.eof()){
 578
 579                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 580
 581
 582                         in >> name;                             //read from first column
 583                         in >> group;                    //read from second column
 584
 585                         //if this name is in the accnos file
 586                         if (names.count(name) != 0) {
 587                                 wroteSomething = true;
 588
 589                                 out << name << '\t' << group << endl;
 590                         }
 591
 592                         m->gobble(in);
 593                 }
 594                 in.close();
 595                 out.close();
 596
 597                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 598                 outputNames.push_back(outputFileName);  outputTypes["group"].push_back(outputFileName);
 599
 600                 return 0;
 601
 602         }
 603         catch(exception& e) {
 604                 m->errorOut(e, "GetSeqsCommand", "readGroup");
 605                 exit(1);
 606         }
 607 }
 608 //**********************************************************************************************************************
 609 int GetSeqsCommand::readTax(){
 610         try {
 611                 string thisOutputDir = outputDir;
 612                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 613                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 614                 ofstream out;
 615                 m->openOutputFile(outputFileName, out);
 616
 617                 ifstream in;
 618                 m->openInputFile(taxfile, in);
 619                 string name, tax;
 620
 621                 bool wroteSomething = false;
 622
 623                 while(!in.eof()){
 624
 625                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 626
 627                         in >> name;                             //read from first column
 628                         in >> tax;                      //read from second column
 629
 630                         //if this name is in the accnos file
 631                         if (names.count(name) != 0) {
 632                                 wroteSomething = true;
 633
 634                                 out << name << '\t' << tax << endl;
 635                         }
 636
 637                         m->gobble(in);
 638                 }
 639                 in.close();
 640                 out.close();
 641
 642                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 643                 outputNames.push_back(outputFileName);  outputTypes["taxonomy"].push_back(outputFileName);
 644
 645                 return 0;
 646
 647         }
 648         catch(exception& e) {
 649                 m->errorOut(e, "GetSeqsCommand", "readTax");
 650                 exit(1);
 651         }
 652 }
 653 //**********************************************************************************************************************
 654 //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
 655 int GetSeqsCommand::readAlign(){
 656         try {
 657                 string thisOutputDir = outputDir;
 658                 if (outputDir == "") {  thisOutputDir += m->hasPath(alignfile);  }
 659                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(alignfile)) + "pick.align.report";
 660                 ofstream out;
 661                 m->openOutputFile(outputFileName, out);
 662
 663
 664                 ifstream in;
 665                 m->openInputFile(alignfile, in);
 666                 string name, junk;
 667
 668                 bool wroteSomething = false;
 669
 670                 //read column headers
 671                 for (int i = 0; i < 16; i++) {
 672                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 673                         else                    {       break;                  }
 674                 }
 675                 out << endl;
 676
 677                 while(!in.eof()){
 678
 679                         if (m->control_pressed) { in.close(); out.close(); remove(outputFileName.c_str());  return 0; }
 680
 681
 682                         in >> name;                             //read from first column
 683
 684                         //if this name is in the accnos file
 685                         if (names.count(name) != 0) {
 686                                 wroteSomething = true;
 687
 688                                 out << name << '\t';
 689
 690                                 //read rest
 691                                 for (int i = 0; i < 15; i++) {
 692                                         if (!in.eof())  {       in >> junk;      out << junk << '\t';   }
 693                                         else                    {       break;                  }
 694                                 }
 695                                 out << endl;
 696
 697                         }else {//still read just don't do anything with it
 698                                 //read rest
 699                                 for (int i = 0; i < 15; i++) {
 700                                         if (!in.eof())  {       in >> junk;             }
 701                                         else                    {       break;                  }
 702                                 }
 703                         }
 704
 705                         m->gobble(in);
 706                 }
 707                 in.close();
 708                 out.close();
 709
 710                 if (wroteSomething == false) { m->mothurOut("Your file does not contain any sequence from the .accnos file."); m->mothurOutEndLine();  }
 711                 outputNames.push_back(outputFileName);  outputTypes["alignreport"].push_back(outputFileName);
 712
 713                 return 0;
 714
 715         }
 716         catch(exception& e) {
 717                 m->errorOut(e, "GetSeqsCommand", "readAlign");
 718                 exit(1);
 719         }
 720 }
 721 //**********************************************************************************************************************
 722
 723 int GetSeqsCommand::readAccnos(){
 724         try {
 725
 726                 ifstream in;
 727                 m->openInputFile(accnosfile, in);
 728                 string name;
 729
 730                 while(!in.eof()){
 731                         in >> name;
 732
 733                         names.insert(name);
 734
 735                         m->gobble(in);
 736                 }
 737                 in.close();
 738
 739                 return 0;
 740
 741         }
 742         catch(exception& e) {
 743                 m->errorOut(e, "GetSeqsCommand", "readAccnos");
 744                 exit(1);
 745         }
 746 }
 747
 748 //**********************************************************************************************************************
 749