mothur.h

   1 #ifndef MOTHUR_H
   2 #define MOTHUR_H
   3
   4
   5
   6 /*
   7  *  mothur.h
   8  *  Mothur
   9  *
  10  *  Created by Sarah Westcott on 2/19/09.
  11  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
  12  *
  13  */
  14
  15 /* This file contains all the standard incudes we use in the project as well as some common utilities. */
  16
  17 //#include <cstddef>
  18
  19 //io libraries
  20 #include <iostream>
  21 #include <iomanip>
  22 #include <fstream>
  23 #include <sstream>
  24 #include <signal.h>
  25
  26
  27 //exception
  28 #include <stdexcept>
  29 #include <exception>
  30 #include <cstdlib>
  31
  32
  33 //containers
  34 #include <vector>
  35 #include <set>
  36 #include <map>
  37 #include <string>
  38 #include <list>
  39
  40 //math
  41 #include <cmath>
  42 #include <math.h>
  43 #include <algorithm>
  44
  45 //misc
  46 #include <cerrno>
  47 #include <ctime>
  48 #include <limits>
  49
  50 #ifdef USE_MPI
  51         #include "mpi.h"
  52 #endif
  53 /***********************************************************************/
  54
  55 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
  56         #include <sys/wait.h>
  57         #include <unistd.h>
  58
  59         #ifdef USE_READLINE
  60                 #include <readline/readline.h>
  61                 #include <readline/history.h>
  62         #endif
  63
  64 #else
  65         #include <conio.h> //allows unbuffered screen capture from stdin
  66         #include <direct.h> //get cwd
  67 #endif
  68
  69 using namespace std;
  70
  71 #define exp(x) (exp((double) x))
  72 #define sqrt(x) (sqrt((double) x))
  73 #define log10(x) (log10((double) x))
  74 #define log2(x) (log10(x)/log10(2))
  75 #define isnan(x) ((x) != (x))
  76 #define isinf(x) (fabs(x) == std::numeric_limits<double>::infinity())
  77
  78 typedef unsigned long ull;
  79
  80 struct IntNode {
  81         int lvalue;
  82         int rvalue;
  83         int lcoef;
  84         int rcoef;
  85         IntNode* left;
  86         IntNode* right;
  87
  88         IntNode(int lv, int rv, IntNode* l, IntNode* r) : lvalue(lv), rvalue(rv), left(l), right(r) {};
  89         IntNode() {};
  90 };
  91
  92 struct ThreadNode {
  93         int* pid;
  94         IntNode* left;
  95         IntNode* right;
  96 };
  97
  98 /************************************************************/
  99 struct clusterNode {
 100         int numSeq;
 101         int parent;
 102         int smallChild; //used to make linkTable work with list and rabund. represents bin number of this cluster node
 103         clusterNode(int num, int par, int kid) : numSeq(num), parent(par), smallChild(kid) {};
 104 };
 105 /************************************************************/
 106 struct seqDist {
 107         int seq1;
 108         int seq2;
 109         float dist;
 110         seqDist() {}
 111         seqDist(int s1, int s2, float d) : seq1(s1), seq2(s2), dist(d) {}
 112         ~seqDist() {}
 113 };
 114 //********************************************************************************************************************
 115 //sorts lowest to highest
 116 inline bool compareSequenceDistance(seqDist left, seqDist right){
 117         return (left.dist < right.dist);
 118 }
 119 /***********************************************************************/
 120
 121 // snagged from http://www.parashift.com/c++-faq-lite/misc-technical-issues.html#faq-39.2
 122 // works for now, but there should be a way to do it without killing the whole program
 123
 124 class BadConversion : public runtime_error {
 125 public:
 126         BadConversion(const string& s) : runtime_error(s){ }
 127 };
 128
 129 //**********************************************************************************************************************
 130
 131 template<typename T>
 132 inline void convert(const string& s, T& x, bool failIfLeftoverChars = true){
 133         istringstream i(s);
 134         char c;
 135         if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
 136                 throw BadConversion(s);
 137 }
 138
 139 //**********************************************************************************************************************
 140
 141 template<typename T>
 142 inline bool convertTestFloat(const string& s, T& x, bool failIfLeftoverChars = true){
 143         istringstream i(s);
 144         char c;
 145         if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
 146         {
 147                 return false;
 148         }
 149         return true;
 150 }
 151
 152 //**********************************************************************************************************************
 153
 154 template<typename T>
 155 inline bool convertTest(const string& s, T& x, bool failIfLeftoverChars = true){
 156         istringstream i(s);
 157         char c;
 158         if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
 159         {
 160                 cout << "unable to be converted into an integer.\n" << endl;
 161                 return false;
 162         }
 163         return true;
 164 }
 165
 166 //**********************************************************************************************************************
 167
 168 template<typename T>
 169 string toString(const T&x){
 170     stringstream output;
 171     output << x;
 172     return output.str();
 173 }
 174
 175 //**********************************************************************************************************************
 176
 177 template<typename T>
 178 string toHex(const T&x){
 179         stringstream output;
 180
 181         output << hex << x;
 182
 183     return output.str();
 184 }
 185 //**********************************************************************************************************************
 186
 187 template<typename T>
 188 string toString(const T&x, int i){
 189         stringstream output;
 190
 191         output.precision(i);
 192     output << fixed << x;
 193
 194     return output.str();
 195 }
 196 /***********************************************************************/
 197 inline int openOutputFileAppend(string fileName, ofstream& fileHandle){
 198
 199         fileHandle.open(fileName.c_str(), ios::app);
 200         if(!fileHandle) {
 201                 cout << "Error: Could not open " << fileName << endl;
 202                 return 1;
 203         }
 204         else {
 205                 return 0;
 206         }
 207
 208 }
 209 /***********************************************************************/
 210
 211 inline void gobble(istream& f){
 212
 213         char d;
 214     while(isspace(d=f.get()))           {;}
 215         f.putback(d);
 216
 217 }
 218 /***********************************************************************/
 219
 220 inline string getline(ifstream& fileHandle) {
 221         try {
 222
 223                 string line = "";
 224
 225                 while (!fileHandle.eof())       {
 226                         //get next character
 227                         char c = fileHandle.get();
 228
 229                         //are you at the end of the line
 230                         if ((c == '\n') || (c == '\r') || (c == '\f')){  break; }
 231                         else {          line += c;              }
 232                 }
 233
 234                 return line;
 235
 236         }
 237         catch(exception& e) {
 238                 cout << "Error in mothur function getline" << endl;
 239                 exit(1);
 240         }
 241 }
 242
 243 /***********************************************************************/
 244
 245 inline bool isTrue(string f){
 246
 247         if ((f == "TRUE") || (f == "T") || (f == "true") || (f == "t")) {       return true;    }
 248         else {  return false;  }
 249 }
 250
 251 /***********************************************************************/
 252
 253 inline float roundDist(float dist, int precision){
 254
 255         return int(dist * precision + 0.5)/float(precision);
 256
 257 }
 258
 259 /***********************************************************************/
 260
 261 inline int getNumNames(string names){
 262
 263         int count = 0;
 264
 265         if(names != ""){
 266                 count = 1;
 267                 for(int i=0;i<names.size();i++){
 268                         if(names[i] == ','){
 269                                 count++;
 270                         }
 271                 }
 272         }
 273
 274         return count;
 275
 276 }
 277
 278 /**************************************************************************************************/
 279
 280 inline vector<vector<double> > binomial(int maxOrder){
 281
 282         vector<vector<double> > binomial(maxOrder+1);
 283
 284     for(int i=0;i<=maxOrder;i++){
 285                 binomial[i].resize(maxOrder+1);
 286                 binomial[i][0]=1;
 287                 binomial[0][i]=0;
 288     }
 289     binomial[0][0]=1;
 290
 291     binomial[1][0]=1;
 292     binomial[1][1]=1;
 293
 294     for(int i=2;i<=maxOrder;i++){
 295                 binomial[1][i]=0;
 296     }
 297
 298     for(int i=2;i<=maxOrder;i++){
 299                 for(int j=1;j<=maxOrder;j++){
 300                         if(i==j){       binomial[i][j]=1;                                                                       }
 301                         if(j>i) {       binomial[i][j]=0;                                                                       }
 302                         else    {       binomial[i][j]=binomial[i-1][j-1]+binomial[i-1][j];     }
 303                 }
 304     }
 305
 306         return binomial;
 307 }
 308
 309 /***********************************************************************/
 310
 311 inline string getRootName(string longName){
 312
 313         string rootName = longName;
 314
 315         if(longName.find_last_of(".") != longName.npos){
 316                 int pos = longName.find_last_of('.')+1;
 317                 rootName = longName.substr(0, pos);
 318         }
 319
 320         return rootName;
 321 }
 322 /***********************************************************************/
 323
 324 inline string getSimpleName(string longName){
 325
 326         string simpleName = longName;
 327
 328         size_t found;
 329         found=longName.find_last_of("/\\");
 330
 331         if(found != longName.npos){
 332                 simpleName = longName.substr(found+1);
 333         }
 334
 335                 //if(longName.find_last_of("/") != longName.npos){
 336                 //      int pos = longName.find_last_of('/')+1;
 337                 //      simpleName = longName.substr(pos, longName.length());
 338                 //}
 339
 340         return simpleName;
 341 }
 342
 343 /***********************************************************************/
 344
 345 inline int factorial(int num){
 346         int total = 1;
 347
 348         for (int i = 1; i <= num; i++) {
 349                 total *= i;
 350         }
 351
 352         return total;
 353 }
 354 /**************************************************************************************************
 355
 356 double min(double x, double y)
 357 {
 358     if(x<y){    return x;    }
 359     else   {    return y;    }
 360 }
 361
 362 /***********************************************************************/
 363
 364 inline string getPathName(string longName){
 365
 366         string rootPathName = longName;
 367
 368         if(longName.find_last_of("/\\") != longName.npos){
 369                 int pos = longName.find_last_of("/\\")+1;
 370                 rootPathName = longName.substr(0, pos);
 371         }
 372
 373         return rootPathName;
 374 }
 375 /***********************************************************************/
 376
 377 inline string hasPath(string longName){
 378
 379         string path = "";
 380
 381         size_t found;
 382         found=longName.find_last_of("/\\");
 383
 384         if(found != longName.npos){
 385                 path = longName.substr(0, found+1);
 386         }
 387
 388         return path;
 389 }
 390
 391 /***********************************************************************/
 392
 393 inline string getExtension(string longName){
 394
 395         string extension = longName;
 396
 397         if(longName.find_last_of('.') != longName.npos){
 398                 int pos = longName.find_last_of('.');
 399                 extension = longName.substr(pos, longName.length());
 400         }
 401
 402         return extension;
 403 }
 404 /***********************************************************************/
 405 inline bool isBlank(string fileName){
 406
 407         ifstream fileHandle;
 408         fileHandle.open(fileName.c_str());
 409         if(!fileHandle) {
 410                 cout << "Error: Could not open " << fileName << endl;
 411                 return false;
 412         }else {
 413                 //check for blank file
 414                 gobble(fileHandle);
 415                 if (fileHandle.eof()) { fileHandle.close(); return true;  }
 416         }
 417         return false;
 418 }
 419 /***********************************************************************/
 420
 421 inline string getFullPathName(string fileName){
 422         try{
 423
 424         string path = hasPath(fileName);
 425         string newFileName;
 426         int pos;
 427
 428         if (path == "") { return fileName; } //its a simple name
 429         else { //we need to complete the pathname
 430                 // ex. ../../../filename
 431                 // cwd = /user/work/desktop
 432
 433                 string cwd;
 434                 //get current working directory
 435                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 436                         if (path.rfind("./") == -1) { return fileName; } //already complete name
 437                         else { newFileName = fileName.substr(fileName.rfind("./")+2); } //save the complete part of the name
 438
 439                         char* cwdpath = new char[1024];
 440
 441                         size_t size;
 442                         cwdpath=getcwd(cwdpath,size);
 443
 444                         cwd = cwdpath;
 445
 446                         //rip off first '/'
 447                         string simpleCWD;
 448                         if (cwd.length() > 0) { simpleCWD = cwd.substr(1); }
 449
 450                         //break apart the current working directory
 451                         vector<string> dirs;
 452                         while (simpleCWD.find_first_of('/') != -1) {
 453                                 string dir = simpleCWD.substr(0,simpleCWD.find_first_of('/'));
 454                                 simpleCWD = simpleCWD.substr(simpleCWD.find_first_of('/')+1, simpleCWD.length());
 455                                 dirs.push_back(dir);
 456                         }
 457                         //get last one              // ex. ../../../filename = /user/work/desktop/filename
 458                         dirs.push_back(simpleCWD);  //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
 459
 460
 461                         int index = dirs.size()-1;
 462
 463                         while((pos = path.rfind("./")) != -1) { //while you don't have a complete path
 464                                 if (path[(pos-1)] == '.') { //you want your parent directory ../
 465                                         path = path.substr(0, pos-1);
 466                                         index--;
 467                                         if (index == 0) {  break; }
 468                                 }else if (path[(pos-1)] == '/') { //you want the current working dir ./
 469                                         path = path.substr(0, pos);
 470                                 }else if (pos == 1) { break;
 471                                 }else { cout << "cannot resolve path for " <<  fileName << endl; return fileName; }
 472                         }
 473
 474                         for (int i = index; i >= 0; i--) {
 475                                 newFileName = dirs[i] +  "/" + newFileName;
 476                         }
 477
 478                         newFileName =  "/" +  newFileName;
 479                         return newFileName;
 480
 481                 #else
 482                         if (path.rfind(".\\") == -1) { return fileName; } //already complete name
 483                         else { newFileName = fileName.substr(fileName.rfind(".\\")+2); } //save the complete part of the name
 484
 485                         char *cwdpath = NULL;
 486                         cwdpath = getcwd(NULL, 0); // or _getcwd
 487                         if ( cwdpath != NULL) { cwd = cwdpath; }
 488                         else { cwd = "";  }
 489
 490                         //break apart the current working directory
 491                         vector<string> dirs;
 492                         while (cwd.find_first_of('\\') != -1) {
 493                                 string dir = cwd.substr(0,cwd.find_first_of('\\'));
 494                                 cwd = cwd.substr(cwd.find_first_of('\\')+1, cwd.length());
 495                                 dirs.push_back(dir);
 496
 497                         }
 498                         //get last one
 499                         dirs.push_back(cwd);  //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
 500
 501                         int index = dirs.size()-1;
 502
 503                         while((pos = path.rfind(".\\")) != -1) { //while you don't have a complete path
 504                                 if (path[(pos-1)] == '.') { //you want your parent directory ../
 505                                         path = path.substr(0, pos-1);
 506                                         index--;
 507                                         if (index == 0) {  break; }
 508                                 }else if (path[(pos-1)] == '\\') { //you want the current working dir ./
 509                                         path = path.substr(0, pos);
 510                                 }else if (pos == 1) { break;
 511                                 }else {  cout << "cannot resolve path for " <<  fileName << endl;  return fileName; }
 512                         }
 513
 514                         for (int i = index; i >= 0; i--) {
 515                                 newFileName = dirs[i] +  "\\" + newFileName;
 516                         }
 517
 518                         return newFileName;
 519
 520                 #endif
 521         }
 522         }
 523         catch(exception& e) {
 524                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function getFullPathName. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 525                 exit(1);
 526         }
 527 }
 528 /***********************************************************************/
 529
 530 inline int openInputFile(string fileName, ifstream& fileHandle, string m){
 531
 532         //get full path name
 533         string completeFileName = getFullPathName(fileName);
 534
 535         fileHandle.open(completeFileName.c_str());
 536         if(!fileHandle) {
 537                 return 1;
 538         }else {
 539                 //check for blank file
 540                 gobble(fileHandle);
 541                 return 0;
 542         }
 543 }
 544 /***********************************************************************/
 545
 546 inline int openInputFile(string fileName, ifstream& fileHandle){
 547
 548         //get full path name
 549         string completeFileName = getFullPathName(fileName);
 550
 551         fileHandle.open(completeFileName.c_str());
 552         if(!fileHandle) {
 553                 cout << "Error: Could not open " << completeFileName << endl;
 554                 return 1;
 555         }
 556         else {
 557                 //check for blank file
 558                 gobble(fileHandle);
 559                 if (fileHandle.eof()) { cout << completeFileName << " is blank. Please correct." << endl;  return 1;  }
 560
 561                 return 0;
 562         }
 563
 564 }
 565 /***********************************************************************/
 566
 567 inline int renameFile(string oldName, string newName){
 568
 569         ifstream inTest;
 570         int exist = openInputFile(newName, inTest, "");
 571
 572 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 573         if (exist == 0) { //you could open it so you want to delete it
 574                 inTest.close();
 575                 string command = "rm " + newName;
 576                 system(command.c_str());
 577         }
 578
 579         string command = "mv " + oldName + " " + newName;
 580         system(command.c_str());
 581 #else
 582         remove(newName.c_str());
 583         int renameOk = rename(oldName.c_str(), newName.c_str());
 584 #endif
 585         return 0;
 586 }
 587
 588 /***********************************************************************/
 589
 590 inline int openOutputFile(string fileName, ofstream& fileHandle){
 591
 592         string completeFileName = getFullPathName(fileName);
 593
 594         fileHandle.open(completeFileName.c_str(), ios::trunc);
 595         if(!fileHandle) {
 596                 cout << "Error: Could not open " << completeFileName << endl;
 597                 return 1;
 598         }
 599         else {
 600                 return 0;
 601         }
 602
 603 }
 604
 605 /***********************************************************************/
 606
 607 inline int getNumSeqs(ifstream& file){
 608
 609         int numSeqs = count(istreambuf_iterator<char>(file),istreambuf_iterator<char>(), '>');
 610         file.seekg(0);
 611         return numSeqs;
 612
 613 }
 614 /***********************************************************************/
 615
 616 inline bool inVector(string member, vector<string> group){
 617
 618         for (int i = 0; i < group.size(); i++) {
 619                 if (group[i] == member) {  return true;         }
 620         }
 621
 622         return false;
 623 }
 624 /***********************************************************************/
 625
 626 //This function parses the estimator options and puts them in a vector
 627 inline void splitAtDash(string& estim, vector<string>& container) {
 628         try {
 629                 string individual;
 630
 631                 while (estim.find_first_of('-') != -1) {
 632                         individual = estim.substr(0,estim.find_first_of('-'));
 633                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 634                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 635                                 container.push_back(individual);
 636                         }
 637                 }
 638                 //get last one
 639                 container.push_back(estim);
 640         }
 641         catch(exception& e) {
 642                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 643                 exit(1);
 644         }
 645 }
 646
 647 /***********************************************************************/
 648 //This function parses the label options and puts them in a set
 649 inline void splitAtDash(string& estim, set<string>& container) {
 650         try {
 651                 string individual;
 652
 653                 while (estim.find_first_of('-') != -1) {
 654                         individual = estim.substr(0,estim.find_first_of('-'));
 655                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 656                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 657                                 container.insert(individual);
 658                         }
 659                 }
 660                 //get last one
 661                 container.insert(estim);
 662         }
 663         catch(exception& e) {
 664                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 665                 exit(1);
 666         }
 667 }
 668 /***********************************************************************/
 669 //This function parses the line options and puts them in a set
 670 inline void splitAtDash(string& estim, set<int>& container) {
 671         try {
 672                 string individual;
 673                 int lineNum;
 674
 675                 while (estim.find_first_of('-') != -1) {
 676                         individual = estim.substr(0,estim.find_first_of('-'));
 677                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 678                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 679                                 convert(individual, lineNum); //convert the string to int
 680                                 container.insert(lineNum);
 681                         }
 682                 }
 683                 //get last one
 684                 convert(estim, lineNum); //convert the string to int
 685                 container.insert(lineNum);
 686         }
 687         catch(exception& e) {
 688                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 689                 exit(1);
 690         }
 691 }
 692 /***********************************************************************/
 693 //This function parses the a string and puts peices in a vector
 694 inline void splitAtComma(string& estim, vector<string>& container) {
 695         try {
 696                 string individual;
 697
 698                 while (estim.find_first_of(',') != -1) {
 699                         individual = estim.substr(0,estim.find_first_of(','));
 700                         if ((estim.find_first_of(',')+1) <= estim.length()) { //checks to make sure you don't have comma at end of string
 701                                 estim = estim.substr(estim.find_first_of(',')+1, estim.length());
 702                                 container.push_back(individual);
 703                         }
 704                 }
 705                 //get last one
 706                 container.push_back(estim);
 707         }
 708         catch(exception& e) {
 709                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtComma. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 710                 exit(1);
 711         }
 712 }
 713 /***********************************************************************/
 714
 715 //This function splits up the various option parameters
 716 inline void splitAtComma(string& prefix, string& suffix){
 717         try {
 718                 prefix = suffix.substr(0,suffix.find_first_of(','));
 719                 if ((suffix.find_first_of(',')+2) <= suffix.length()) {  //checks to make sure you don't have comma at end of string
 720                         suffix = suffix.substr(suffix.find_first_of(',')+1, suffix.length());
 721                         string space = " ";
 722                         while(suffix.at(0) == ' ')
 723                                 suffix = suffix.substr(1, suffix.length());
 724                 }
 725
 726         }
 727         catch(exception& e) {
 728                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtComma. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 729                 exit(1);
 730         }
 731 }
 732 /***********************************************************************/
 733
 734 //This function separates the key value from the option value i.e. dist=96_...
 735 inline void splitAtEquals(string& key, string& value){
 736         try {
 737                 if(value.find_first_of('=') != -1){
 738                         key = value.substr(0,value.find_first_of('='));
 739                         if ((value.find_first_of('=')+1) <= value.length()) {
 740                                 value = value.substr(value.find_first_of('=')+1, value.length());
 741                         }
 742                 }else{
 743                         key = value;
 744                         value = 1;
 745                 }
 746         }
 747         catch(exception& e) {
 748                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtEquals. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 749                 exit(1);
 750         }
 751 }
 752 /**************************************************************************************************/
 753
 754 inline bool inUsersGroups(string groupname, vector<string> Groups) {
 755         try {
 756                 for (int i = 0; i < Groups.size(); i++) {
 757                         if (groupname == Groups[i]) { return true; }
 758                 }
 759                 return false;
 760         }
 761         catch(exception& e) {
 762                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function inUsersGroups. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 763                 exit(1);
 764         }
 765 }
 766 /**************************************************************************************************/
 767 //returns true if any of the strings in first vector are in second vector
 768 inline bool inUsersGroups(vector<string> groupnames, vector<string> Groups) {
 769         try {
 770
 771                 for (int i = 0; i < groupnames.size(); i++) {
 772                         if (inUsersGroups(groupnames[i], Groups)) { return true; }
 773                 }
 774                 return false;
 775         }
 776         catch(exception& e) {
 777                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function inUsersGroups. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 778                 exit(1);
 779         }
 780 }
 781 /***********************************************************************/
 782 //this function determines if the user has given us labels that are smaller than the given label.
 783 //if so then it returns true so that the calling function can run the previous valid distance.
 784 //it's a "smart" distance function.  It also checks for invalid labels.
 785 inline bool anyLabelsToProcess(string label, set<string>& userLabels, string errorOff) {
 786         try {
 787
 788                 set<string>::iterator it;
 789                 vector<float> orderFloat;
 790                 map<string, float> userMap;  //the conversion process removes trailing 0's which we need to put back
 791                 map<string, float>::iterator it2;
 792                 float labelFloat;
 793                 bool smaller = false;
 794
 795                 //unique is the smallest line
 796                 if (label == "unique") {  return false;  }
 797                 else { convert(label, labelFloat); }
 798
 799                 //go through users set and make them floats
 800                 for(it = userLabels.begin(); it != userLabels.end(); ++it) {
 801
 802                         float temp;
 803                         if ((*it != "unique") && (convertTestFloat(*it, temp) == true)){
 804                                 convert(*it, temp);
 805                                 orderFloat.push_back(temp);
 806                                 userMap[*it] = temp;
 807                         }else if (*it == "unique") {
 808                                 orderFloat.push_back(-1.0);
 809                                 userMap["unique"] = -1.0;
 810                         }else {
 811                                 if (errorOff == "") {  cout << *it << " is not a valid label." << endl;  }
 812                                 userLabels.erase(*it);
 813                                 it--;
 814                         }
 815                 }
 816
 817                 //sort order
 818                 sort(orderFloat.begin(), orderFloat.end());
 819
 820                 /*************************************************/
 821                 //is this label bigger than any of the users labels
 822                 /*************************************************/
 823
 824                 //loop through order until you find a label greater than label
 825                 for (int i = 0; i < orderFloat.size(); i++) {
 826                         if (orderFloat[i] < labelFloat) {
 827                                 smaller = true;
 828                                 if (orderFloat[i] == -1) {
 829                                         if (errorOff == "") { cout << "Your file does not include the label unique." << endl; }
 830                                         userLabels.erase("unique");
 831                                 }
 832                                 else {
 833                                         if (errorOff == "") { cout << "Your file does not include the label " << endl; }
 834                                         string s = "";
 835                                         for (it2 = userMap.begin(); it2!= userMap.end(); it2++) {
 836                                                 if (it2->second == orderFloat[i]) {
 837                                                         s = it2->first;
 838                                                         //remove small labels
 839                                                         userLabels.erase(s);
 840                                                         break;
 841                                                 }
 842                                         }
 843                                         if (errorOff == "") {cout << s <<  ". I will use the next smallest distance. " << endl; }
 844                                 }
 845                         //since they are sorted once you find a bigger one stop looking
 846                         }else { break; }
 847                 }
 848
 849                 return smaller;
 850
 851         }
 852         catch(exception& e) {
 853                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function anyLabelsToProcess. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 854                 exit(1);
 855         }
 856 }
 857
 858 /**************************************************************************************************/
 859 inline void appendFiles(string temp, string filename) {
 860         try{
 861                 ofstream output;
 862                 ifstream input;
 863
 864                 //open output file in append mode
 865                 openOutputFileAppend(filename, output);
 866                 openInputFile(temp, input);
 867
 868                 while(char c = input.get()){
 869                         if(input.eof())         {       break;                  }
 870                         else                            {       output << c;    }
 871                 }
 872
 873                 input.close();
 874                 output.close();
 875         }
 876         catch(exception& e) {
 877                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function appendFiles. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 878                 exit(1);
 879         }
 880 }
 881
 882 /**************************************************************************************************/
 883 inline string sortFile(string distFile){
 884         try {
 885                 string outfile = getRootName(distFile) + "sorted.dist";
 886
 887                 //if you can, use the unix sort since its been optimized for years
 888                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 889                         string command = "sort -n -k +3 " + distFile + " -o " + outfile;
 890                         system(command.c_str());
 891                 #else //you are stuck with my best attempt...
 892                         //windows sort does not have a way to specify a column, only a character in the line
 893                         //since we cannot assume that the distance will always be at the the same character location on each line
 894                         //due to variable sequence name lengths, I chose to force the distance into first position, then sort and then put it back.
 895
 896                         //read in file line by file and put distance first
 897                         string tempDistFile = distFile + ".temp";
 898                         ifstream input;
 899                         ofstream output;
 900                         openInputFile(distFile, input);
 901                         openOutputFile(tempDistFile, output);
 902
 903                         string firstName, secondName;
 904                         float dist;
 905                         while (input) {
 906                                 input >> firstName >> secondName >> dist;
 907                                 output << dist << '\t' << firstName << '\t' << secondName << endl;
 908                                 gobble(input);
 909                         }
 910                         input.close();
 911                         output.close();
 912
 913
 914                         //sort using windows sort
 915                         string tempOutfile = outfile + ".temp";
 916                         string command = "sort " + tempDistFile + " /O " + tempOutfile;
 917                         system(command.c_str());
 918
 919                         //read in sorted file and put distance at end again
 920                         ifstream input2;
 921                         openInputFile(tempOutfile, input2);
 922                         openOutputFile(outfile, output);
 923
 924                         while (input2) {
 925                                 input2 >> dist >> firstName >> secondName;
 926                                 output << firstName << '\t' << secondName << '\t' << dist << endl;
 927                                 gobble(input2);
 928                         }
 929                         input2.close();
 930                         output.close();
 931
 932                         //remove temp files
 933                         remove(tempDistFile.c_str());
 934                         remove(tempOutfile.c_str());
 935                 #endif
 936
 937                 return outfile;
 938         }
 939         catch(exception& e) {
 940                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function sortfile. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 941                 exit(1);
 942         }
 943 }
 944 /**************************************************************************************************/
 945 inline vector<long> setFilePosFasta(string filename, int& num) {
 946
 947                         vector<long> positions;
 948                         ifstream inFASTA;
 949                         openInputFile(filename, inFASTA);
 950
 951                         string input;
 952                         while(!inFASTA.eof()){
 953                                 input = getline(inFASTA); gobble(inFASTA);
 954                                 if (input.length() != 0) {
 955                                         if(input[0] == '>'){    long pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1);      }
 956                                 }
 957                         }
 958                         inFASTA.close();
 959
 960                         num = positions.size();
 961
 962                         FILE * pFile;
 963                         long size;
 964
 965                         //get num bytes in file
 966                         pFile = fopen (filename.c_str(),"rb");
 967                         if (pFile==NULL) perror ("Error opening file");
 968                         else{
 969                                 fseek (pFile, 0, SEEK_END);
 970                                 size=ftell (pFile);
 971                                 fclose (pFile);
 972                         }
 973
 974                         positions.push_back(size);
 975
 976                         return positions;
 977 }
 978 /**************************************************************************************************/
 979 inline vector<long> setFilePosEachLine(string filename, int& num) {
 980
 981                         vector<long> positions;
 982                         ifstream in;
 983                         openInputFile(filename, in);
 984
 985                         string input;
 986                         while(!in.eof()){
 987                                 long lastpos = in.tellg();
 988                                 input = getline(in); gobble(in);
 989                                 if (input.length() != 0) {
 990                                         long pos = in.tellg();
 991                                         if (pos != -1) { positions.push_back(pos - input.length() - 1); }
 992                                         else {  positions.push_back(lastpos);  }
 993                                 }
 994                         }
 995                         in.close();
 996
 997                         num = positions.size();
 998
 999                         FILE * pFile;
1000                         long size;
1001
1002                         //get num bytes in file
1003                         pFile = fopen (filename.c_str(),"rb");
1004                         if (pFile==NULL) perror ("Error opening file");
1005                         else{
1006                                 fseek (pFile, 0, SEEK_END);
1007                                 size=ftell (pFile);
1008                                 fclose (pFile);
1009                         }
1010
1011                         positions.push_back(size);
1012
1013                         return positions;
1014 }
1015
1016 /**************************************************************************************************/
1017 #endif
1018