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         #include <windows.h>
  68         #include <psapi.h>
  69
  70 #endif
  71
  72 using namespace std;
  73
  74 #define exp(x) (exp((double) x))
  75 #define sqrt(x) (sqrt((double) x))
  76 #define log10(x) (log10((double) x))
  77 #define log2(x) (log10(x)/log10(2))
  78 #define isnan(x) ((x) != (x))
  79 #define isinf(x) (fabs(x) == std::numeric_limits<double>::infinity())
  80
  81 typedef unsigned long ull;
  82
  83 struct IntNode {
  84         int lvalue;
  85         int rvalue;
  86         int lcoef;
  87         int rcoef;
  88         IntNode* left;
  89         IntNode* right;
  90
  91         IntNode(int lv, int rv, IntNode* l, IntNode* r) : lvalue(lv), rvalue(rv), left(l), right(r) {};
  92         IntNode() {};
  93 };
  94
  95 struct ThreadNode {
  96         int* pid;
  97         IntNode* left;
  98         IntNode* right;
  99 };
 100
 101 /************************************************************/
 102 struct clusterNode {
 103         int numSeq;
 104         int parent;
 105         int smallChild; //used to make linkTable work with list and rabund. represents bin number of this cluster node
 106         clusterNode(int num, int par, int kid) : numSeq(num), parent(par), smallChild(kid) {};
 107 };
 108 /************************************************************/
 109 struct seqDist {
 110         int seq1;
 111         int seq2;
 112         float dist;
 113         seqDist() {}
 114         seqDist(int s1, int s2, float d) : seq1(s1), seq2(s2), dist(d) {}
 115         ~seqDist() {}
 116 };
 117 //********************************************************************************************************************
 118 //sorts lowest to highest
 119 inline bool compareSequenceDistance(seqDist left, seqDist right){
 120         return (left.dist < right.dist);
 121 }
 122 /***********************************************************************/
 123
 124 // snagged from http://www.parashift.com/c++-faq-lite/misc-technical-issues.html#faq-39.2
 125 // works for now, but there should be a way to do it without killing the whole program
 126
 127 class BadConversion : public runtime_error {
 128 public:
 129         BadConversion(const string& s) : runtime_error(s){ }
 130 };
 131
 132 //**********************************************************************************************************************
 133
 134 template<typename T>
 135 inline void convert(const string& s, T& x, bool failIfLeftoverChars = true){
 136         istringstream i(s);
 137         char c;
 138         if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
 139                 throw BadConversion(s);
 140 }
 141
 142 //**********************************************************************************************************************
 143
 144 template<typename T>
 145 inline bool convertTestFloat(const string& s, T& x, bool failIfLeftoverChars = true){
 146         istringstream i(s);
 147         char c;
 148         if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
 149         {
 150                 return false;
 151         }
 152         return true;
 153 }
 154
 155 //**********************************************************************************************************************
 156
 157 template<typename T>
 158 inline bool convertTest(const string& s, T& x, bool failIfLeftoverChars = true){
 159         istringstream i(s);
 160         char c;
 161         if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
 162         {
 163                 cout << "unable to be converted into an integer.\n" << endl;
 164                 return false;
 165         }
 166         return true;
 167 }
 168
 169 //**********************************************************************************************************************
 170
 171 template<typename T>
 172 string toString(const T&x){
 173     stringstream output;
 174     output << x;
 175     return output.str();
 176 }
 177
 178 //**********************************************************************************************************************
 179
 180 template<typename T>
 181 string toHex(const T&x){
 182         stringstream output;
 183
 184         output << hex << x;
 185
 186     return output.str();
 187 }
 188 //**********************************************************************************************************************
 189
 190 template<typename T>
 191 string toString(const T&x, int i){
 192         stringstream output;
 193
 194         output.precision(i);
 195     output << fixed << x;
 196
 197     return output.str();
 198 }
 199 /***********************************************************************/
 200 inline int openOutputFileAppend(string fileName, ofstream& fileHandle){
 201
 202         fileHandle.open(fileName.c_str(), ios::app);
 203         if(!fileHandle) {
 204                 cout << "Error: Could not open " << fileName << endl;
 205                 return 1;
 206         }
 207         else {
 208                 return 0;
 209         }
 210
 211 }
 212 /***********************************************************************/
 213
 214 inline void gobble(istream& f){
 215
 216         char d;
 217     while(isspace(d=f.get()))           {;}
 218         f.putback(d);
 219
 220 }
 221 /***********************************************************************/
 222
 223 inline string getline(ifstream& fileHandle) {
 224         try {
 225
 226                 string line = "";
 227
 228                 while (!fileHandle.eof())       {
 229                         //get next character
 230                         char c = fileHandle.get();
 231
 232                         //are you at the end of the line
 233                         if ((c == '\n') || (c == '\r') || (c == '\f')){  break; }
 234                         else {          line += c;              }
 235                 }
 236
 237                 return line;
 238
 239         }
 240         catch(exception& e) {
 241                 cout << "Error in mothur function getline" << endl;
 242                 exit(1);
 243         }
 244 }
 245
 246 /***********************************************************************/
 247
 248 inline bool isTrue(string f){
 249
 250         if ((f == "TRUE") || (f == "T") || (f == "true") || (f == "t")) {       return true;    }
 251         else {  return false;  }
 252 }
 253
 254 /***********************************************************************/
 255
 256 inline float roundDist(float dist, int precision){
 257
 258         return int(dist * precision + 0.5)/float(precision);
 259
 260 }
 261 /***********************************************************************/
 262
 263 inline float ceilDist(float dist, int precision){
 264
 265         return int(ceil(dist * precision))/float(precision);
 266
 267 }
 268
 269 /***********************************************************************/
 270
 271 inline int getNumNames(string names){
 272
 273         int count = 0;
 274
 275         if(names != ""){
 276                 count = 1;
 277                 for(int i=0;i<names.size();i++){
 278                         if(names[i] == ','){
 279                                 count++;
 280                         }
 281                 }
 282         }
 283
 284         return count;
 285
 286 }
 287
 288 /**************************************************************************************************/
 289
 290 inline vector<vector<double> > binomial(int maxOrder){
 291
 292         vector<vector<double> > binomial(maxOrder+1);
 293
 294     for(int i=0;i<=maxOrder;i++){
 295                 binomial[i].resize(maxOrder+1);
 296                 binomial[i][0]=1;
 297                 binomial[0][i]=0;
 298     }
 299     binomial[0][0]=1;
 300
 301     binomial[1][0]=1;
 302     binomial[1][1]=1;
 303
 304     for(int i=2;i<=maxOrder;i++){
 305                 binomial[1][i]=0;
 306     }
 307
 308     for(int i=2;i<=maxOrder;i++){
 309                 for(int j=1;j<=maxOrder;j++){
 310                         if(i==j){       binomial[i][j]=1;                                                                       }
 311                         if(j>i) {       binomial[i][j]=0;                                                                       }
 312                         else    {       binomial[i][j]=binomial[i-1][j-1]+binomial[i-1][j];     }
 313                 }
 314     }
 315
 316         return binomial;
 317 }
 318
 319 /***********************************************************************/
 320
 321 inline string getRootName(string longName){
 322
 323         string rootName = longName;
 324
 325         if(longName.find_last_of(".") != longName.npos){
 326                 int pos = longName.find_last_of('.')+1;
 327                 rootName = longName.substr(0, pos);
 328         }
 329
 330         return rootName;
 331 }
 332 /***********************************************************************/
 333
 334 inline string getSimpleName(string longName){
 335
 336         string simpleName = longName;
 337
 338         size_t found;
 339         found=longName.find_last_of("/\\");
 340
 341         if(found != longName.npos){
 342                 simpleName = longName.substr(found+1);
 343         }
 344
 345                 //if(longName.find_last_of("/") != longName.npos){
 346                 //      int pos = longName.find_last_of('/')+1;
 347                 //      simpleName = longName.substr(pos, longName.length());
 348                 //}
 349
 350         return simpleName;
 351 }
 352
 353 /***********************************************************************/
 354
 355 inline int factorial(int num){
 356         int total = 1;
 357
 358         for (int i = 1; i <= num; i++) {
 359                 total *= i;
 360         }
 361
 362         return total;
 363 }
 364 /**************************************************************************************************
 365
 366 double min(double x, double y)
 367 {
 368     if(x<y){    return x;    }
 369     else   {    return y;    }
 370 }
 371
 372 /***********************************************************************/
 373
 374 inline string getPathName(string longName){
 375
 376         string rootPathName = longName;
 377
 378         if(longName.find_last_of("/\\") != longName.npos){
 379                 int pos = longName.find_last_of("/\\")+1;
 380                 rootPathName = longName.substr(0, pos);
 381         }
 382
 383         return rootPathName;
 384 }
 385 /***********************************************************************/
 386
 387 inline string hasPath(string longName){
 388
 389         string path = "";
 390
 391         size_t found;
 392         found=longName.find_last_of("~/\\");
 393
 394         if(found != longName.npos){
 395                 path = longName.substr(0, found+1);
 396         }
 397
 398         return path;
 399 }
 400
 401 /***********************************************************************/
 402
 403 inline string getExtension(string longName){
 404
 405         string extension = longName;
 406
 407         if(longName.find_last_of('.') != longName.npos){
 408                 int pos = longName.find_last_of('.');
 409                 extension = longName.substr(pos, longName.length());
 410         }
 411
 412         return extension;
 413 }
 414 /***********************************************************************/
 415 inline bool isBlank(string fileName){
 416
 417         ifstream fileHandle;
 418         fileHandle.open(fileName.c_str());
 419         if(!fileHandle) {
 420                 cout << "Error: Could not open " << fileName << endl;
 421                 return false;
 422         }else {
 423                 //check for blank file
 424                 gobble(fileHandle);
 425                 if (fileHandle.eof()) { fileHandle.close(); return true;  }
 426         }
 427         return false;
 428 }
 429 /***********************************************************************/
 430
 431 inline string getFullPathName(string fileName){
 432         try{
 433
 434         string path = hasPath(fileName);
 435         string newFileName;
 436         int pos;
 437
 438         if (path == "") { return fileName; } //its a simple name
 439         else { //we need to complete the pathname
 440                 // ex. ../../../filename
 441                 // cwd = /user/work/desktop
 442
 443                 string cwd;
 444                 //get current working directory
 445                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 446
 447                         if (path.find("~") != -1) { //go to home directory
 448                                 string homeDir = getenv ("HOME");
 449                                 newFileName = homeDir + fileName.substr(fileName.find("~")+1);
 450                                 return newFileName;
 451                         }else { //find path
 452                                 if (path.rfind("./") == -1) { return fileName; } //already complete name
 453                                 else { newFileName = fileName.substr(fileName.rfind("./")+2); } //save the complete part of the name
 454
 455                                 char* cwdpath = new char[1024];
 456
 457                                 size_t size;
 458                                 cwdpath=getcwd(cwdpath,size);
 459
 460                                 cwd = cwdpath;
 461
 462                                 //rip off first '/'
 463                                 string simpleCWD;
 464                                 if (cwd.length() > 0) { simpleCWD = cwd.substr(1); }
 465
 466                                 //break apart the current working directory
 467                                 vector<string> dirs;
 468                                 while (simpleCWD.find_first_of('/') != -1) {
 469                                         string dir = simpleCWD.substr(0,simpleCWD.find_first_of('/'));
 470                                         simpleCWD = simpleCWD.substr(simpleCWD.find_first_of('/')+1, simpleCWD.length());
 471                                         dirs.push_back(dir);
 472                                 }
 473                                 //get last one              // ex. ../../../filename = /user/work/desktop/filename
 474                                 dirs.push_back(simpleCWD);  //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
 475
 476
 477                                 int index = dirs.size()-1;
 478
 479                                 while((pos = path.rfind("./")) != -1) { //while you don't have a complete path
 480                                         if (pos == 0) { break;  //you are at the end
 481                                         }else if (path[(pos-1)] == '.') { //you want your parent directory ../
 482                                                 path = path.substr(0, pos-1);
 483                                                 index--;
 484                                                 if (index == 0) {  break; }
 485                                         }else if (path[(pos-1)] == '/') { //you want the current working dir ./
 486                                                 path = path.substr(0, pos);
 487                                         }else if (pos == 1) { break;  //you are at the end
 488                                         }else { cout << "cannot resolve path for " <<  fileName << endl; return fileName; }
 489                                 }
 490
 491                                 for (int i = index; i >= 0; i--) {
 492                                         newFileName = dirs[i] +  "/" + newFileName;
 493                                 }
 494
 495                                 newFileName =  "/" +  newFileName;
 496                                 return newFileName;
 497                         }
 498                 #else
 499                         if (path.find("~") != -1) { //go to home directory
 500                                 string homeDir = getenv ("HOMEPATH");
 501                                 newFileName = homeDir + fileName.substr(fileName.find("~")+1);
 502                                 return newFileName;
 503                         }else { //find path
 504                                 if (path.rfind(".\\") == -1) { return fileName; } //already complete name
 505                                 else { newFileName = fileName.substr(fileName.rfind(".\\")+2); } //save the complete part of the name
 506
 507                                 char *cwdpath = NULL;
 508                                 cwdpath = getcwd(NULL, 0); // or _getcwd
 509                                 if ( cwdpath != NULL) { cwd = cwdpath; }
 510                                 else { cwd = "";  }
 511
 512                                 //break apart the current working directory
 513                                 vector<string> dirs;
 514                                 while (cwd.find_first_of('\\') != -1) {
 515                                         string dir = cwd.substr(0,cwd.find_first_of('\\'));
 516                                         cwd = cwd.substr(cwd.find_first_of('\\')+1, cwd.length());
 517                                         dirs.push_back(dir);
 518
 519                                 }
 520                                 //get last one
 521                                 dirs.push_back(cwd);  //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
 522
 523                                 int index = dirs.size()-1;
 524
 525                                 while((pos = path.rfind(".\\")) != -1) { //while you don't have a complete path
 526                                         if (pos == 0) { break;  //you are at the end
 527                                         }else if (path[(pos-1)] == '.') { //you want your parent directory ../
 528                                                 path = path.substr(0, pos-1);
 529                                                 index--;
 530                                                 if (index == 0) {  break; }
 531                                         }else if (path[(pos-1)] == '\\') { //you want the current working dir ./
 532                                                 path = path.substr(0, pos);
 533                                         }else if (pos == 1) { break;  //you are at the end
 534                                         }else { cout << "cannot resolve path for " <<  fileName << endl; return fileName; }
 535                                 }
 536
 537                                 for (int i = index; i >= 0; i--) {
 538                                         newFileName = dirs[i] +  "\\" + newFileName;
 539                                 }
 540
 541                                 return newFileName;
 542                         }
 543
 544                 #endif
 545         }
 546         }
 547         catch(exception& e) {
 548                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function getFullPathName. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 549                 exit(1);
 550         }
 551 }
 552 /***********************************************************************/
 553
 554 inline int openInputFile(string fileName, ifstream& fileHandle, string m){
 555
 556         //get full path name
 557         string completeFileName = getFullPathName(fileName);
 558
 559         fileHandle.open(completeFileName.c_str());
 560         if(!fileHandle) {
 561                 return 1;
 562         }else {
 563                 //check for blank file
 564                 gobble(fileHandle);
 565                 return 0;
 566         }
 567 }
 568 /***********************************************************************/
 569
 570 inline int openInputFile(string fileName, ifstream& fileHandle){
 571
 572         //get full path name
 573         string completeFileName = getFullPathName(fileName);
 574
 575         fileHandle.open(completeFileName.c_str());
 576         if(!fileHandle) {
 577                 cout << "Error: Could not open " << completeFileName << endl;
 578                 return 1;
 579         }
 580         else {
 581                 //check for blank file
 582                 gobble(fileHandle);
 583                 if (fileHandle.eof()) { cout << completeFileName << " is blank. Please correct." << endl;   }
 584
 585                 return 0;
 586         }
 587
 588 }
 589 /***********************************************************************/
 590
 591 inline int renameFile(string oldName, string newName){
 592
 593         ifstream inTest;
 594         int exist = openInputFile(newName, inTest, "");
 595
 596 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 597         if (exist == 0) { //you could open it so you want to delete it
 598                 inTest.close();
 599                 string command = "rm " + newName;
 600                 system(command.c_str());
 601         }
 602
 603         string command = "mv " + oldName + " " + newName;
 604         system(command.c_str());
 605 #else
 606         remove(newName.c_str());
 607         int renameOk = rename(oldName.c_str(), newName.c_str());
 608 #endif
 609         return 0;
 610 }
 611
 612 /***********************************************************************/
 613
 614 inline int openOutputFile(string fileName, ofstream& fileHandle){
 615
 616         string completeFileName = getFullPathName(fileName);
 617
 618         fileHandle.open(completeFileName.c_str(), ios::trunc);
 619         if(!fileHandle) {
 620                 cout << "Error: Could not open " << completeFileName << endl;
 621                 return 1;
 622         }
 623         else {
 624                 return 0;
 625         }
 626
 627 }
 628
 629 /***********************************************************************/
 630
 631 inline int getNumSeqs(ifstream& file){
 632
 633         int numSeqs = count(istreambuf_iterator<char>(file),istreambuf_iterator<char>(), '>');
 634         file.seekg(0);
 635         return numSeqs;
 636
 637 }
 638 /***********************************************************************/
 639 inline void getNumSeqs(ifstream& file, int& numSeqs){
 640
 641         string input;
 642         numSeqs = 0;
 643         while(!file.eof()){
 644                 input = getline(file);
 645                 if (input.length() != 0) {
 646                         if(input[0] == '>'){ numSeqs++; }
 647                 }
 648         }
 649 }
 650
 651 /***********************************************************************/
 652
 653 inline bool inVector(string member, vector<string> group){
 654
 655         for (int i = 0; i < group.size(); i++) {
 656                 if (group[i] == member) {  return true;         }
 657         }
 658
 659         return false;
 660 }
 661 /***********************************************************************/
 662
 663 //This function parses the estimator options and puts them in a vector
 664 inline void splitAtDash(string& estim, vector<string>& container) {
 665         try {
 666                 string individual;
 667
 668                 while (estim.find_first_of('-') != -1) {
 669                         individual = estim.substr(0,estim.find_first_of('-'));
 670                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 671                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 672                                 container.push_back(individual);
 673                         }
 674                 }
 675                 //get last one
 676                 container.push_back(estim);
 677         }
 678         catch(exception& e) {
 679                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 680                 exit(1);
 681         }
 682 }
 683
 684 /***********************************************************************/
 685 //This function parses the label options and puts them in a set
 686 inline void splitAtDash(string& estim, set<string>& container) {
 687         try {
 688                 string individual;
 689
 690                 while (estim.find_first_of('-') != -1) {
 691                         individual = estim.substr(0,estim.find_first_of('-'));
 692                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 693                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 694                                 container.insert(individual);
 695                         }
 696                 }
 697                 //get last one
 698                 container.insert(estim);
 699         }
 700         catch(exception& e) {
 701                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 702                 exit(1);
 703         }
 704 }
 705 /***********************************************************************/
 706 //This function parses the line options and puts them in a set
 707 inline void splitAtDash(string& estim, set<int>& container) {
 708         try {
 709                 string individual;
 710                 int lineNum;
 711
 712                 while (estim.find_first_of('-') != -1) {
 713                         individual = estim.substr(0,estim.find_first_of('-'));
 714                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 715                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 716                                 convert(individual, lineNum); //convert the string to int
 717                                 container.insert(lineNum);
 718                         }
 719                 }
 720                 //get last one
 721                 convert(estim, lineNum); //convert the string to int
 722                 container.insert(lineNum);
 723         }
 724         catch(exception& e) {
 725                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 726                 exit(1);
 727         }
 728 }
 729 /***********************************************************************/
 730 //This function parses the a string and puts peices in a vector
 731 inline void splitAtComma(string& estim, vector<string>& container) {
 732         try {
 733                 string individual;
 734
 735                 while (estim.find_first_of(',') != -1) {
 736                         individual = estim.substr(0,estim.find_first_of(','));
 737                         if ((estim.find_first_of(',')+1) <= estim.length()) { //checks to make sure you don't have comma at end of string
 738                                 estim = estim.substr(estim.find_first_of(',')+1, estim.length());
 739                                 container.push_back(individual);
 740                         }
 741                 }
 742                 //get last one
 743                 container.push_back(estim);
 744         }
 745         catch(exception& e) {
 746                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtComma. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 747                 exit(1);
 748         }
 749 }
 750 /***********************************************************************/
 751
 752 //This function splits up the various option parameters
 753 inline void splitAtComma(string& prefix, string& suffix){
 754         try {
 755                 prefix = suffix.substr(0,suffix.find_first_of(','));
 756                 if ((suffix.find_first_of(',')+2) <= suffix.length()) {  //checks to make sure you don't have comma at end of string
 757                         suffix = suffix.substr(suffix.find_first_of(',')+1, suffix.length());
 758                         string space = " ";
 759                         while(suffix.at(0) == ' ')
 760                                 suffix = suffix.substr(1, suffix.length());
 761                 }
 762
 763         }
 764         catch(exception& e) {
 765                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtComma. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 766                 exit(1);
 767         }
 768 }
 769 /***********************************************************************/
 770
 771 //This function separates the key value from the option value i.e. dist=96_...
 772 inline void splitAtEquals(string& key, string& value){
 773         try {
 774                 if(value.find_first_of('=') != -1){
 775                         key = value.substr(0,value.find_first_of('='));
 776                         if ((value.find_first_of('=')+1) <= value.length()) {
 777                                 value = value.substr(value.find_first_of('=')+1, value.length());
 778                         }
 779                 }else{
 780                         key = value;
 781                         value = 1;
 782                 }
 783         }
 784         catch(exception& e) {
 785                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtEquals. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 786                 exit(1);
 787         }
 788 }
 789 /**************************************************************************************************/
 790
 791 inline bool inUsersGroups(string groupname, vector<string> Groups) {
 792         try {
 793                 for (int i = 0; i < Groups.size(); i++) {
 794                         if (groupname == Groups[i]) { return true; }
 795                 }
 796                 return false;
 797         }
 798         catch(exception& e) {
 799                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function inUsersGroups. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 800                 exit(1);
 801         }
 802 }
 803 /**************************************************************************************************/
 804 //returns true if any of the strings in first vector are in second vector
 805 inline bool inUsersGroups(vector<string> groupnames, vector<string> Groups) {
 806         try {
 807
 808                 for (int i = 0; i < groupnames.size(); i++) {
 809                         if (inUsersGroups(groupnames[i], Groups)) { return true; }
 810                 }
 811                 return false;
 812         }
 813         catch(exception& e) {
 814                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function inUsersGroups. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 815                 exit(1);
 816         }
 817 }
 818 /***********************************************************************/
 819 //this function determines if the user has given us labels that are smaller than the given label.
 820 //if so then it returns true so that the calling function can run the previous valid distance.
 821 //it's a "smart" distance function.  It also checks for invalid labels.
 822 inline bool anyLabelsToProcess(string label, set<string>& userLabels, string errorOff) {
 823         try {
 824
 825                 set<string>::iterator it;
 826                 vector<float> orderFloat;
 827                 map<string, float> userMap;  //the conversion process removes trailing 0's which we need to put back
 828                 map<string, float>::iterator it2;
 829                 float labelFloat;
 830                 bool smaller = false;
 831
 832                 //unique is the smallest line
 833                 if (label == "unique") {  return false;  }
 834                 else {
 835                         if (convertTestFloat(label, labelFloat)) {
 836                                 convert(label, labelFloat);
 837                         }else { //cant convert
 838                                 return false;
 839                         }
 840                 }
 841
 842                 //go through users set and make them floats
 843                 for(it = userLabels.begin(); it != userLabels.end(); ++it) {
 844
 845                         float temp;
 846                         if ((*it != "unique") && (convertTestFloat(*it, temp) == true)){
 847                                 convert(*it, temp);
 848                                 orderFloat.push_back(temp);
 849                                 userMap[*it] = temp;
 850                         }else if (*it == "unique") {
 851                                 orderFloat.push_back(-1.0);
 852                                 userMap["unique"] = -1.0;
 853                         }else {
 854                                 if (errorOff == "") {  cout << *it << " is not a valid label." << endl;  }
 855                                 userLabels.erase(*it);
 856                                 it--;
 857                         }
 858                 }
 859
 860                 //sort order
 861                 sort(orderFloat.begin(), orderFloat.end());
 862
 863                 /*************************************************/
 864                 //is this label bigger than any of the users labels
 865                 /*************************************************/
 866
 867                 //loop through order until you find a label greater than label
 868                 for (int i = 0; i < orderFloat.size(); i++) {
 869                         if (orderFloat[i] < labelFloat) {
 870                                 smaller = true;
 871                                 if (orderFloat[i] == -1) {
 872                                         if (errorOff == "") { cout << "Your file does not include the label unique." << endl; }
 873                                         userLabels.erase("unique");
 874                                 }
 875                                 else {
 876                                         if (errorOff == "") { cout << "Your file does not include the label " << endl; }
 877                                         string s = "";
 878                                         for (it2 = userMap.begin(); it2!= userMap.end(); it2++) {
 879                                                 if (it2->second == orderFloat[i]) {
 880                                                         s = it2->first;
 881                                                         //remove small labels
 882                                                         userLabels.erase(s);
 883                                                         break;
 884                                                 }
 885                                         }
 886                                         if (errorOff == "") {cout << s <<  ". I will use the next smallest distance. " << endl; }
 887                                 }
 888                         //since they are sorted once you find a bigger one stop looking
 889                         }else { break; }
 890                 }
 891
 892                 return smaller;
 893
 894         }
 895         catch(exception& e) {
 896                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function anyLabelsToProcess. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 897                 exit(1);
 898         }
 899 }
 900
 901 /**************************************************************************************************/
 902 inline void appendFiles(string temp, string filename) {
 903         try{
 904                 ofstream output;
 905                 ifstream input;
 906
 907                 //open output file in append mode
 908                 openOutputFileAppend(filename, output);
 909                 int ableToOpen = openInputFile(temp, input, "no error");
 910
 911                 if (ableToOpen == 0) { //you opened it
 912                         while(char c = input.get()){
 913                                 if(input.eof())         {       break;                  }
 914                                 else                            {       output << c;    }
 915                         }
 916                         input.close();
 917                 }
 918
 919                 output.close();
 920         }
 921         catch(exception& e) {
 922                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function appendFiles. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 923                 exit(1);
 924         }
 925 }
 926
 927 /**************************************************************************************************/
 928 inline string sortFile(string distFile, string outputDir){
 929         try {
 930
 931                 //if (outputDir == "") {  outputDir += hasPath(distFile);  }
 932                 string outfile = getRootName(distFile) + "sorted.dist";
 933
 934
 935                 //if you can, use the unix sort since its been optimized for years
 936                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 937                         string command = "sort -n -k +3 " + distFile + " -o " + outfile;
 938                         system(command.c_str());
 939                 #else //you are stuck with my best attempt...
 940                         //windows sort does not have a way to specify a column, only a character in the line
 941                         //since we cannot assume that the distance will always be at the the same character location on each line
 942                         //due to variable sequence name lengths, I chose to force the distance into first position, then sort and then put it back.
 943
 944                         //read in file line by file and put distance first
 945                         string tempDistFile = distFile + ".temp";
 946                         ifstream input;
 947                         ofstream output;
 948                         openInputFile(distFile, input);
 949                         openOutputFile(tempDistFile, output);
 950
 951                         string firstName, secondName;
 952                         float dist;
 953                         while (input) {
 954                                 input >> firstName >> secondName >> dist;
 955                                 output << dist << '\t' << firstName << '\t' << secondName << endl;
 956                                 gobble(input);
 957                         }
 958                         input.close();
 959                         output.close();
 960
 961
 962                         //sort using windows sort
 963                         string tempOutfile = outfile + ".temp";
 964                         string command = "sort " + tempDistFile + " /O " + tempOutfile;
 965                         system(command.c_str());
 966
 967                         //read in sorted file and put distance at end again
 968                         ifstream input2;
 969                         openInputFile(tempOutfile, input2);
 970                         openOutputFile(outfile, output);
 971
 972                         while (input2) {
 973                                 input2 >> dist >> firstName >> secondName;
 974                                 output << firstName << '\t' << secondName << '\t' << dist << endl;
 975                                 gobble(input2);
 976                         }
 977                         input2.close();
 978                         output.close();
 979
 980                         //remove temp files
 981                         remove(tempDistFile.c_str());
 982                         remove(tempOutfile.c_str());
 983                 #endif
 984
 985                 return outfile;
 986         }
 987         catch(exception& e) {
 988                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function sortfile. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 989                 exit(1);
 990         }
 991 }
 992 /**************************************************************************************************/
 993 inline vector<long> setFilePosFasta(string filename, int& num) {
 994
 995                         vector<long> positions;
 996                         ifstream inFASTA;
 997                         openInputFile(filename, inFASTA);
 998
 999                         string input;
1000                         while(!inFASTA.eof()){
1001                                 input = getline(inFASTA); gobble(inFASTA);
1002                                 if (input.length() != 0) {
1003                                         if(input[0] == '>'){    long pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1);      }
1004                                 }
1005                         }
1006                         inFASTA.close();
1007
1008                         num = positions.size();
1009
1010                         /*FILE * pFile;
1011                         long size;
1012
1013                         //get num bytes in file
1014                         pFile = fopen (filename.c_str(),"rb");
1015                         if (pFile==NULL) perror ("Error opening file");
1016                         else{
1017                                 fseek (pFile, 0, SEEK_END);
1018                                 size=ftell (pFile);
1019                                 fclose (pFile);
1020                         }*/
1021
1022                         long size = positions[(positions.size()-1)];
1023                         ifstream in;
1024                         openInputFile(filename, in);
1025
1026                         in.seekg(size);
1027
1028                         while(char c = in.get()){
1029                                 if(in.eof())            {       break;  }
1030                                 else                            {       size++; }
1031                         }
1032                         in.close();
1033
1034                         positions.push_back(size);
1035
1036                         return positions;
1037 }
1038 /**************************************************************************************************/
1039 inline vector<long> setFilePosEachLine(string filename, int& num) {
1040
1041                         vector<long> positions;
1042                         ifstream in;
1043                         openInputFile(filename, in);
1044
1045                         string input;
1046                         while(!in.eof()){
1047                                 long lastpos = in.tellg();
1048                                 input = getline(in); gobble(in);
1049                                 if (input.length() != 0) {
1050                                         long pos = in.tellg();
1051                                         if (pos != -1) { positions.push_back(pos - input.length() - 1); }
1052                                         else {  positions.push_back(lastpos);  }
1053                                 }
1054                         }
1055                         in.close();
1056
1057                         num = positions.size();
1058
1059                         FILE * pFile;
1060                         long size;
1061
1062                         //get num bytes in file
1063                         pFile = fopen (filename.c_str(),"rb");
1064                         if (pFile==NULL) perror ("Error opening file");
1065                         else{
1066                                 fseek (pFile, 0, SEEK_END);
1067                                 size=ftell (pFile);
1068                                 fclose (pFile);
1069                         }
1070
1071                         positions.push_back(size);
1072
1073                         return positions;
1074 }
1075
1076 /**************************************************************************************************/
1077 #endif
1078