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 (pos == 0) { break;  //you are at the end
 465                                 }else if (path[(pos-1)] == '.') { //you want your parent directory ../
 466                                         path = path.substr(0, pos-1);
 467                                         index--;
 468                                         if (index == 0) {  break; }
 469                                 }else if (path[(pos-1)] == '/') { //you want the current working dir ./
 470                                         path = path.substr(0, pos);
 471                                 }else if (pos == 1) { break;  //you are at the end
 472                                 }else { cout << "cannot resolve path for " <<  fileName << endl; return fileName; }
 473                         }
 474
 475                         for (int i = index; i >= 0; i--) {
 476                                 newFileName = dirs[i] +  "/" + newFileName;
 477                         }
 478
 479                         newFileName =  "/" +  newFileName;
 480                         return newFileName;
 481
 482                 #else
 483                         if (path.rfind(".\\") == -1) { return fileName; } //already complete name
 484                         else { newFileName = fileName.substr(fileName.rfind(".\\")+2); } //save the complete part of the name
 485
 486                         char *cwdpath = NULL;
 487                         cwdpath = getcwd(NULL, 0); // or _getcwd
 488                         if ( cwdpath != NULL) { cwd = cwdpath; }
 489                         else { cwd = "";  }
 490
 491                         //break apart the current working directory
 492                         vector<string> dirs;
 493                         while (cwd.find_first_of('\\') != -1) {
 494                                 string dir = cwd.substr(0,cwd.find_first_of('\\'));
 495                                 cwd = cwd.substr(cwd.find_first_of('\\')+1, cwd.length());
 496                                 dirs.push_back(dir);
 497
 498                         }
 499                         //get last one
 500                         dirs.push_back(cwd);  //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
 501
 502                         int index = dirs.size()-1;
 503
 504                         while((pos = path.rfind(".\\")) != -1) { //while you don't have a complete path
 505                                 if (pos == 0) { break;  //you are at the end
 506                                 }else if (path[(pos-1)] == '.') { //you want your parent directory ../
 507                                         path = path.substr(0, pos-1);
 508                                         index--;
 509                                         if (index == 0) {  break; }
 510                                 }else if (path[(pos-1)] == '\\') { //you want the current working dir ./
 511                                         path = path.substr(0, pos);
 512                                 }else if (pos == 1) { break;  //you are at the end
 513                                 }else { cout << "cannot resolve path for " <<  fileName << endl; return fileName; }
 514                         }
 515
 516                         for (int i = index; i >= 0; i--) {
 517                                 newFileName = dirs[i] +  "\\" + newFileName;
 518                         }
 519
 520                         return newFileName;
 521
 522                 #endif
 523         }
 524         }
 525         catch(exception& e) {
 526                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function getFullPathName. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 527                 exit(1);
 528         }
 529 }
 530 /***********************************************************************/
 531
 532 inline int openInputFile(string fileName, ifstream& fileHandle, string m){
 533
 534         //get full path name
 535         string completeFileName = getFullPathName(fileName);
 536
 537         fileHandle.open(completeFileName.c_str());
 538         if(!fileHandle) {
 539                 return 1;
 540         }else {
 541                 //check for blank file
 542                 gobble(fileHandle);
 543                 return 0;
 544         }
 545 }
 546 /***********************************************************************/
 547
 548 inline int openInputFile(string fileName, ifstream& fileHandle){
 549
 550         //get full path name
 551         string completeFileName = getFullPathName(fileName);
 552
 553         fileHandle.open(completeFileName.c_str());
 554         if(!fileHandle) {
 555                 cout << "Error: Could not open " << completeFileName << endl;
 556                 return 1;
 557         }
 558         else {
 559                 //check for blank file
 560                 gobble(fileHandle);
 561                 if (fileHandle.eof()) { cout << completeFileName << " is blank. Please correct." << endl;  return 1;  }
 562
 563                 return 0;
 564         }
 565
 566 }
 567 /***********************************************************************/
 568
 569 inline int renameFile(string oldName, string newName){
 570
 571         ifstream inTest;
 572         int exist = openInputFile(newName, inTest, "");
 573
 574 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 575         if (exist == 0) { //you could open it so you want to delete it
 576                 inTest.close();
 577                 string command = "rm " + newName;
 578                 system(command.c_str());
 579         }
 580
 581         string command = "mv " + oldName + " " + newName;
 582         system(command.c_str());
 583 #else
 584         remove(newName.c_str());
 585         int renameOk = rename(oldName.c_str(), newName.c_str());
 586 #endif
 587         return 0;
 588 }
 589
 590 /***********************************************************************/
 591
 592 inline int openOutputFile(string fileName, ofstream& fileHandle){
 593
 594         string completeFileName = getFullPathName(fileName);
 595
 596         fileHandle.open(completeFileName.c_str(), ios::trunc);
 597         if(!fileHandle) {
 598                 cout << "Error: Could not open " << completeFileName << endl;
 599                 return 1;
 600         }
 601         else {
 602                 return 0;
 603         }
 604
 605 }
 606
 607 /***********************************************************************/
 608
 609 inline int getNumSeqs(ifstream& file){
 610
 611         int numSeqs = count(istreambuf_iterator<char>(file),istreambuf_iterator<char>(), '>');
 612         file.seekg(0);
 613         return numSeqs;
 614
 615 }
 616 /***********************************************************************/
 617
 618 inline bool inVector(string member, vector<string> group){
 619
 620         for (int i = 0; i < group.size(); i++) {
 621                 if (group[i] == member) {  return true;         }
 622         }
 623
 624         return false;
 625 }
 626 /***********************************************************************/
 627
 628 //This function parses the estimator options and puts them in a vector
 629 inline void splitAtDash(string& estim, vector<string>& container) {
 630         try {
 631                 string individual;
 632
 633                 while (estim.find_first_of('-') != -1) {
 634                         individual = estim.substr(0,estim.find_first_of('-'));
 635                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 636                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 637                                 container.push_back(individual);
 638                         }
 639                 }
 640                 //get last one
 641                 container.push_back(estim);
 642         }
 643         catch(exception& e) {
 644                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 645                 exit(1);
 646         }
 647 }
 648
 649 /***********************************************************************/
 650 //This function parses the label options and puts them in a set
 651 inline void splitAtDash(string& estim, set<string>& container) {
 652         try {
 653                 string individual;
 654
 655                 while (estim.find_first_of('-') != -1) {
 656                         individual = estim.substr(0,estim.find_first_of('-'));
 657                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 658                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 659                                 container.insert(individual);
 660                         }
 661                 }
 662                 //get last one
 663                 container.insert(estim);
 664         }
 665         catch(exception& e) {
 666                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 667                 exit(1);
 668         }
 669 }
 670 /***********************************************************************/
 671 //This function parses the line options and puts them in a set
 672 inline void splitAtDash(string& estim, set<int>& container) {
 673         try {
 674                 string individual;
 675                 int lineNum;
 676
 677                 while (estim.find_first_of('-') != -1) {
 678                         individual = estim.substr(0,estim.find_first_of('-'));
 679                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 680                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 681                                 convert(individual, lineNum); //convert the string to int
 682                                 container.insert(lineNum);
 683                         }
 684                 }
 685                 //get last one
 686                 convert(estim, lineNum); //convert the string to int
 687                 container.insert(lineNum);
 688         }
 689         catch(exception& e) {
 690                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 691                 exit(1);
 692         }
 693 }
 694 /***********************************************************************/
 695 //This function parses the a string and puts peices in a vector
 696 inline void splitAtComma(string& estim, vector<string>& container) {
 697         try {
 698                 string individual;
 699
 700                 while (estim.find_first_of(',') != -1) {
 701                         individual = estim.substr(0,estim.find_first_of(','));
 702                         if ((estim.find_first_of(',')+1) <= estim.length()) { //checks to make sure you don't have comma at end of string
 703                                 estim = estim.substr(estim.find_first_of(',')+1, estim.length());
 704                                 container.push_back(individual);
 705                         }
 706                 }
 707                 //get last one
 708                 container.push_back(estim);
 709         }
 710         catch(exception& e) {
 711                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtComma. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 712                 exit(1);
 713         }
 714 }
 715 /***********************************************************************/
 716
 717 //This function splits up the various option parameters
 718 inline void splitAtComma(string& prefix, string& suffix){
 719         try {
 720                 prefix = suffix.substr(0,suffix.find_first_of(','));
 721                 if ((suffix.find_first_of(',')+2) <= suffix.length()) {  //checks to make sure you don't have comma at end of string
 722                         suffix = suffix.substr(suffix.find_first_of(',')+1, suffix.length());
 723                         string space = " ";
 724                         while(suffix.at(0) == ' ')
 725                                 suffix = suffix.substr(1, suffix.length());
 726                 }
 727
 728         }
 729         catch(exception& e) {
 730                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtComma. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 731                 exit(1);
 732         }
 733 }
 734 /***********************************************************************/
 735
 736 //This function separates the key value from the option value i.e. dist=96_...
 737 inline void splitAtEquals(string& key, string& value){
 738         try {
 739                 if(value.find_first_of('=') != -1){
 740                         key = value.substr(0,value.find_first_of('='));
 741                         if ((value.find_first_of('=')+1) <= value.length()) {
 742                                 value = value.substr(value.find_first_of('=')+1, value.length());
 743                         }
 744                 }else{
 745                         key = value;
 746                         value = 1;
 747                 }
 748         }
 749         catch(exception& e) {
 750                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtEquals. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 751                 exit(1);
 752         }
 753 }
 754 /**************************************************************************************************/
 755
 756 inline bool inUsersGroups(string groupname, vector<string> Groups) {
 757         try {
 758                 for (int i = 0; i < Groups.size(); i++) {
 759                         if (groupname == Groups[i]) { return true; }
 760                 }
 761                 return false;
 762         }
 763         catch(exception& e) {
 764                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function inUsersGroups. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 765                 exit(1);
 766         }
 767 }
 768 /**************************************************************************************************/
 769 //returns true if any of the strings in first vector are in second vector
 770 inline bool inUsersGroups(vector<string> groupnames, vector<string> Groups) {
 771         try {
 772
 773                 for (int i = 0; i < groupnames.size(); i++) {
 774                         if (inUsersGroups(groupnames[i], Groups)) { return true; }
 775                 }
 776                 return false;
 777         }
 778         catch(exception& e) {
 779                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function inUsersGroups. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 780                 exit(1);
 781         }
 782 }
 783 /***********************************************************************/
 784 //this function determines if the user has given us labels that are smaller than the given label.
 785 //if so then it returns true so that the calling function can run the previous valid distance.
 786 //it's a "smart" distance function.  It also checks for invalid labels.
 787 inline bool anyLabelsToProcess(string label, set<string>& userLabels, string errorOff) {
 788         try {
 789
 790                 set<string>::iterator it;
 791                 vector<float> orderFloat;
 792                 map<string, float> userMap;  //the conversion process removes trailing 0's which we need to put back
 793                 map<string, float>::iterator it2;
 794                 float labelFloat;
 795                 bool smaller = false;
 796
 797                 //unique is the smallest line
 798                 if (label == "unique") {  return false;  }
 799                 else { convert(label, labelFloat); }
 800
 801                 //go through users set and make them floats
 802                 for(it = userLabels.begin(); it != userLabels.end(); ++it) {
 803
 804                         float temp;
 805                         if ((*it != "unique") && (convertTestFloat(*it, temp) == true)){
 806                                 convert(*it, temp);
 807                                 orderFloat.push_back(temp);
 808                                 userMap[*it] = temp;
 809                         }else if (*it == "unique") {
 810                                 orderFloat.push_back(-1.0);
 811                                 userMap["unique"] = -1.0;
 812                         }else {
 813                                 if (errorOff == "") {  cout << *it << " is not a valid label." << endl;  }
 814                                 userLabels.erase(*it);
 815                                 it--;
 816                         }
 817                 }
 818
 819                 //sort order
 820                 sort(orderFloat.begin(), orderFloat.end());
 821
 822                 /*************************************************/
 823                 //is this label bigger than any of the users labels
 824                 /*************************************************/
 825
 826                 //loop through order until you find a label greater than label
 827                 for (int i = 0; i < orderFloat.size(); i++) {
 828                         if (orderFloat[i] < labelFloat) {
 829                                 smaller = true;
 830                                 if (orderFloat[i] == -1) {
 831                                         if (errorOff == "") { cout << "Your file does not include the label unique." << endl; }
 832                                         userLabels.erase("unique");
 833                                 }
 834                                 else {
 835                                         if (errorOff == "") { cout << "Your file does not include the label " << endl; }
 836                                         string s = "";
 837                                         for (it2 = userMap.begin(); it2!= userMap.end(); it2++) {
 838                                                 if (it2->second == orderFloat[i]) {
 839                                                         s = it2->first;
 840                                                         //remove small labels
 841                                                         userLabels.erase(s);
 842                                                         break;
 843                                                 }
 844                                         }
 845                                         if (errorOff == "") {cout << s <<  ". I will use the next smallest distance. " << endl; }
 846                                 }
 847                         //since they are sorted once you find a bigger one stop looking
 848                         }else { break; }
 849                 }
 850
 851                 return smaller;
 852
 853         }
 854         catch(exception& e) {
 855                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function anyLabelsToProcess. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 856                 exit(1);
 857         }
 858 }
 859
 860 /**************************************************************************************************/
 861 inline void appendFiles(string temp, string filename) {
 862         try{
 863                 ofstream output;
 864                 ifstream input;
 865
 866                 //open output file in append mode
 867                 openOutputFileAppend(filename, output);
 868                 int ableToOpen = openInputFile(temp, input, "no error");
 869
 870                 if (ableToOpen == 0) { //you opened it
 871                         while(char c = input.get()){
 872                                 if(input.eof())         {       break;                  }
 873                                 else                            {       output << c;    }
 874                         }
 875                         input.close();
 876                 }
 877
 878                 output.close();
 879         }
 880         catch(exception& e) {
 881                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function appendFiles. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 882                 exit(1);
 883         }
 884 }
 885
 886 /**************************************************************************************************/
 887 inline string sortFile(string distFile, string outputDir){
 888         try {
 889
 890                 //if (outputDir == "") {  outputDir += hasPath(distFile);  }
 891                 string outfile = getRootName(distFile) + "sorted.dist";
 892
 893
 894                 //if you can, use the unix sort since its been optimized for years
 895                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 896                         string command = "sort -n -k +3 " + distFile + " -o " + outfile;
 897                         system(command.c_str());
 898                 #else //you are stuck with my best attempt...
 899                         //windows sort does not have a way to specify a column, only a character in the line
 900                         //since we cannot assume that the distance will always be at the the same character location on each line
 901                         //due to variable sequence name lengths, I chose to force the distance into first position, then sort and then put it back.
 902
 903                         //read in file line by file and put distance first
 904                         string tempDistFile = distFile + ".temp";
 905                         ifstream input;
 906                         ofstream output;
 907                         openInputFile(distFile, input);
 908                         openOutputFile(tempDistFile, output);
 909
 910                         string firstName, secondName;
 911                         float dist;
 912                         while (input) {
 913                                 input >> firstName >> secondName >> dist;
 914                                 output << dist << '\t' << firstName << '\t' << secondName << endl;
 915                                 gobble(input);
 916                         }
 917                         input.close();
 918                         output.close();
 919
 920
 921                         //sort using windows sort
 922                         string tempOutfile = outfile + ".temp";
 923                         string command = "sort " + tempDistFile + " /O " + tempOutfile;
 924                         system(command.c_str());
 925
 926                         //read in sorted file and put distance at end again
 927                         ifstream input2;
 928                         openInputFile(tempOutfile, input2);
 929                         openOutputFile(outfile, output);
 930
 931                         while (input2) {
 932                                 input2 >> dist >> firstName >> secondName;
 933                                 output << firstName << '\t' << secondName << '\t' << dist << endl;
 934                                 gobble(input2);
 935                         }
 936                         input2.close();
 937                         output.close();
 938
 939                         //remove temp files
 940                         remove(tempDistFile.c_str());
 941                         remove(tempOutfile.c_str());
 942                 #endif
 943
 944                 return outfile;
 945         }
 946         catch(exception& e) {
 947                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function sortfile. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 948                 exit(1);
 949         }
 950 }
 951 /**************************************************************************************************/
 952 inline vector<long> setFilePosFasta(string filename, int& num) {
 953
 954                         vector<long> positions;
 955                         ifstream inFASTA;
 956                         openInputFile(filename, inFASTA);
 957
 958                         string input;
 959                         while(!inFASTA.eof()){
 960                                 input = getline(inFASTA); gobble(inFASTA);
 961                                 if (input.length() != 0) {
 962                                         if(input[0] == '>'){    long pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1);      }
 963                                 }
 964                         }
 965                         inFASTA.close();
 966
 967                         num = positions.size();
 968
 969                         /*FILE * pFile;
 970                         long size;
 971
 972                         //get num bytes in file
 973                         pFile = fopen (filename.c_str(),"rb");
 974                         if (pFile==NULL) perror ("Error opening file");
 975                         else{
 976                                 fseek (pFile, 0, SEEK_END);
 977                                 size=ftell (pFile);
 978                                 fclose (pFile);
 979                         }*/
 980
 981                         long size = positions[(positions.size()-1)];
 982                         ifstream in;
 983                         openInputFile(filename, in);
 984
 985                         in.seekg(size);
 986
 987                         while(char c = in.get()){
 988                                 if(in.eof())            {       break;  }
 989                                 else                            {       size++; }
 990                         }
 991                         in.close();
 992
 993                         positions.push_back(size);
 994
 995                         return positions;
 996 }
 997 /**************************************************************************************************/
 998 inline vector<long> setFilePosEachLine(string filename, int& num) {
 999
1000                         vector<long> positions;
1001                         ifstream in;
1002                         openInputFile(filename, in);
1003
1004                         string input;
1005                         while(!in.eof()){
1006                                 long lastpos = in.tellg();
1007                                 input = getline(in); gobble(in);
1008                                 if (input.length() != 0) {
1009                                         long pos = in.tellg();
1010                                         if (pos != -1) { positions.push_back(pos - input.length() - 1); }
1011                                         else {  positions.push_back(lastpos);  }
1012                                 }
1013                         }
1014                         in.close();
1015
1016                         num = positions.size();
1017
1018                         FILE * pFile;
1019                         long size;
1020
1021                         //get num bytes in file
1022                         pFile = fopen (filename.c_str(),"rb");
1023                         if (pFile==NULL) perror ("Error opening file");
1024                         else{
1025                                 fseek (pFile, 0, SEEK_END);
1026                                 size=ftell (pFile);
1027                                 fclose (pFile);
1028                         }
1029
1030                         positions.push_back(size);
1031
1032                         return positions;
1033 }
1034
1035 /**************************************************************************************************/
1036 #endif
1037