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
 437                         if (path.find("~") != -1) { //go to home directory
 438                                 string homeDir = getenv ("HOME");
 439                                 newFileName = homeDir + fileName.substr(fileName.find("~")+1);
 440                                 return newFileName;
 441                         }else { //find path
 442                                 if (path.rfind("./") == -1) { return fileName; } //already complete name
 443                                 else { newFileName = fileName.substr(fileName.rfind("./")+2); } //save the complete part of the name
 444
 445                                 char* cwdpath = new char[1024];
 446
 447                                 size_t size;
 448                                 cwdpath=getcwd(cwdpath,size);
 449
 450                                 cwd = cwdpath;
 451
 452                                 //rip off first '/'
 453                                 string simpleCWD;
 454                                 if (cwd.length() > 0) { simpleCWD = cwd.substr(1); }
 455
 456                                 //break apart the current working directory
 457                                 vector<string> dirs;
 458                                 while (simpleCWD.find_first_of('/') != -1) {
 459                                         string dir = simpleCWD.substr(0,simpleCWD.find_first_of('/'));
 460                                         simpleCWD = simpleCWD.substr(simpleCWD.find_first_of('/')+1, simpleCWD.length());
 461                                         dirs.push_back(dir);
 462                                 }
 463                                 //get last one              // ex. ../../../filename = /user/work/desktop/filename
 464                                 dirs.push_back(simpleCWD);  //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
 465
 466
 467                                 int index = dirs.size()-1;
 468
 469                                 while((pos = path.rfind("./")) != -1) { //while you don't have a complete path
 470                                         if (pos == 0) { break;  //you are at the end
 471                                         }else if (path[(pos-1)] == '.') { //you want your parent directory ../
 472                                                 path = path.substr(0, pos-1);
 473                                                 index--;
 474                                                 if (index == 0) {  break; }
 475                                         }else if (path[(pos-1)] == '/') { //you want the current working dir ./
 476                                                 path = path.substr(0, pos);
 477                                         }else if (pos == 1) { break;  //you are at the end
 478                                         }else { cout << "cannot resolve path for " <<  fileName << endl; return fileName; }
 479                                 }
 480
 481                                 for (int i = index; i >= 0; i--) {
 482                                         newFileName = dirs[i] +  "/" + newFileName;
 483                                 }
 484
 485                                 newFileName =  "/" +  newFileName;
 486                                 return newFileName;
 487                         }
 488                 #else
 489                         if (path.find("~") != -1) { //go to home directory
 490                                 string homeDir = getenv ("HOMEPATH");
 491                                 newFileName = homeDir + fileName.substr(fileName.find("~")+1);
 492                                 return newFileName;
 493                         }else { //find path
 494                                 if (path.rfind(".\\") == -1) { return fileName; } //already complete name
 495                                 else { newFileName = fileName.substr(fileName.rfind(".\\")+2); } //save the complete part of the name
 496
 497                                 char *cwdpath = NULL;
 498                                 cwdpath = getcwd(NULL, 0); // or _getcwd
 499                                 if ( cwdpath != NULL) { cwd = cwdpath; }
 500                                 else { cwd = "";  }
 501
 502                                 //break apart the current working directory
 503                                 vector<string> dirs;
 504                                 while (cwd.find_first_of('\\') != -1) {
 505                                         string dir = cwd.substr(0,cwd.find_first_of('\\'));
 506                                         cwd = cwd.substr(cwd.find_first_of('\\')+1, cwd.length());
 507                                         dirs.push_back(dir);
 508
 509                                 }
 510                                 //get last one
 511                                 dirs.push_back(cwd);  //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
 512
 513                                 int index = dirs.size()-1;
 514
 515                                 while((pos = path.rfind(".\\")) != -1) { //while you don't have a complete path
 516                                         if (pos == 0) { break;  //you are at the end
 517                                         }else if (path[(pos-1)] == '.') { //you want your parent directory ../
 518                                                 path = path.substr(0, pos-1);
 519                                                 index--;
 520                                                 if (index == 0) {  break; }
 521                                         }else if (path[(pos-1)] == '\\') { //you want the current working dir ./
 522                                                 path = path.substr(0, pos);
 523                                         }else if (pos == 1) { break;  //you are at the end
 524                                         }else { cout << "cannot resolve path for " <<  fileName << endl; return fileName; }
 525                                 }
 526
 527                                 for (int i = index; i >= 0; i--) {
 528                                         newFileName = dirs[i] +  "\\" + newFileName;
 529                                 }
 530
 531                                 return newFileName;
 532                         }
 533
 534                 #endif
 535         }
 536         }
 537         catch(exception& e) {
 538                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function getFullPathName. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 539                 exit(1);
 540         }
 541 }
 542 /***********************************************************************/
 543
 544 inline int openInputFile(string fileName, ifstream& fileHandle, string m){
 545
 546         //get full path name
 547         string completeFileName = getFullPathName(fileName);
 548
 549         fileHandle.open(completeFileName.c_str());
 550         if(!fileHandle) {
 551                 return 1;
 552         }else {
 553                 //check for blank file
 554                 gobble(fileHandle);
 555                 return 0;
 556         }
 557 }
 558 /***********************************************************************/
 559
 560 inline int openInputFile(string fileName, ifstream& fileHandle){
 561
 562         //get full path name
 563         string completeFileName = getFullPathName(fileName);
 564
 565         fileHandle.open(completeFileName.c_str());
 566         if(!fileHandle) {
 567                 cout << "Error: Could not open " << completeFileName << endl;
 568                 return 1;
 569         }
 570         else {
 571                 //check for blank file
 572                 gobble(fileHandle);
 573                 if (fileHandle.eof()) { cout << completeFileName << " is blank. Please correct." << endl;  return 1;  }
 574
 575                 return 0;
 576         }
 577
 578 }
 579 /***********************************************************************/
 580
 581 inline int renameFile(string oldName, string newName){
 582
 583         ifstream inTest;
 584         int exist = openInputFile(newName, inTest, "");
 585
 586 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 587         if (exist == 0) { //you could open it so you want to delete it
 588                 inTest.close();
 589                 string command = "rm " + newName;
 590                 system(command.c_str());
 591         }
 592
 593         string command = "mv " + oldName + " " + newName;
 594         system(command.c_str());
 595 #else
 596         remove(newName.c_str());
 597         int renameOk = rename(oldName.c_str(), newName.c_str());
 598 #endif
 599         return 0;
 600 }
 601
 602 /***********************************************************************/
 603
 604 inline int openOutputFile(string fileName, ofstream& fileHandle){
 605
 606         string completeFileName = getFullPathName(fileName);
 607
 608         fileHandle.open(completeFileName.c_str(), ios::trunc);
 609         if(!fileHandle) {
 610                 cout << "Error: Could not open " << completeFileName << endl;
 611                 return 1;
 612         }
 613         else {
 614                 return 0;
 615         }
 616
 617 }
 618
 619 /***********************************************************************/
 620
 621 inline int getNumSeqs(ifstream& file){
 622
 623         int numSeqs = count(istreambuf_iterator<char>(file),istreambuf_iterator<char>(), '>');
 624         file.seekg(0);
 625         return numSeqs;
 626
 627 }
 628 /***********************************************************************/
 629
 630 inline bool inVector(string member, vector<string> group){
 631
 632         for (int i = 0; i < group.size(); i++) {
 633                 if (group[i] == member) {  return true;         }
 634         }
 635
 636         return false;
 637 }
 638 /***********************************************************************/
 639
 640 //This function parses the estimator options and puts them in a vector
 641 inline void splitAtDash(string& estim, vector<string>& container) {
 642         try {
 643                 string individual;
 644
 645                 while (estim.find_first_of('-') != -1) {
 646                         individual = estim.substr(0,estim.find_first_of('-'));
 647                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 648                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 649                                 container.push_back(individual);
 650                         }
 651                 }
 652                 //get last one
 653                 container.push_back(estim);
 654         }
 655         catch(exception& e) {
 656                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 657                 exit(1);
 658         }
 659 }
 660
 661 /***********************************************************************/
 662 //This function parses the label options and puts them in a set
 663 inline void splitAtDash(string& estim, set<string>& container) {
 664         try {
 665                 string individual;
 666
 667                 while (estim.find_first_of('-') != -1) {
 668                         individual = estim.substr(0,estim.find_first_of('-'));
 669                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 670                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 671                                 container.insert(individual);
 672                         }
 673                 }
 674                 //get last one
 675                 container.insert(estim);
 676         }
 677         catch(exception& e) {
 678                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 679                 exit(1);
 680         }
 681 }
 682 /***********************************************************************/
 683 //This function parses the line options and puts them in a set
 684 inline void splitAtDash(string& estim, set<int>& container) {
 685         try {
 686                 string individual;
 687                 int lineNum;
 688
 689                 while (estim.find_first_of('-') != -1) {
 690                         individual = estim.substr(0,estim.find_first_of('-'));
 691                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 692                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 693                                 convert(individual, lineNum); //convert the string to int
 694                                 container.insert(lineNum);
 695                         }
 696                 }
 697                 //get last one
 698                 convert(estim, lineNum); //convert the string to int
 699                 container.insert(lineNum);
 700         }
 701         catch(exception& e) {
 702                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 703                 exit(1);
 704         }
 705 }
 706 /***********************************************************************/
 707 //This function parses the a string and puts peices in a vector
 708 inline void splitAtComma(string& estim, vector<string>& container) {
 709         try {
 710                 string individual;
 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 comma at end of string
 715                                 estim = estim.substr(estim.find_first_of(',')+1, estim.length());
 716                                 container.push_back(individual);
 717                         }
 718                 }
 719                 //get last one
 720                 container.push_back(estim);
 721         }
 722         catch(exception& e) {
 723                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtComma. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 724                 exit(1);
 725         }
 726 }
 727 /***********************************************************************/
 728
 729 //This function splits up the various option parameters
 730 inline void splitAtComma(string& prefix, string& suffix){
 731         try {
 732                 prefix = suffix.substr(0,suffix.find_first_of(','));
 733                 if ((suffix.find_first_of(',')+2) <= suffix.length()) {  //checks to make sure you don't have comma at end of string
 734                         suffix = suffix.substr(suffix.find_first_of(',')+1, suffix.length());
 735                         string space = " ";
 736                         while(suffix.at(0) == ' ')
 737                                 suffix = suffix.substr(1, suffix.length());
 738                 }
 739
 740         }
 741         catch(exception& e) {
 742                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtComma. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 743                 exit(1);
 744         }
 745 }
 746 /***********************************************************************/
 747
 748 //This function separates the key value from the option value i.e. dist=96_...
 749 inline void splitAtEquals(string& key, string& value){
 750         try {
 751                 if(value.find_first_of('=') != -1){
 752                         key = value.substr(0,value.find_first_of('='));
 753                         if ((value.find_first_of('=')+1) <= value.length()) {
 754                                 value = value.substr(value.find_first_of('=')+1, value.length());
 755                         }
 756                 }else{
 757                         key = value;
 758                         value = 1;
 759                 }
 760         }
 761         catch(exception& e) {
 762                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtEquals. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 763                 exit(1);
 764         }
 765 }
 766 /**************************************************************************************************/
 767
 768 inline bool inUsersGroups(string groupname, vector<string> Groups) {
 769         try {
 770                 for (int i = 0; i < Groups.size(); i++) {
 771                         if (groupname == Groups[i]) { return true; }
 772                 }
 773                 return false;
 774         }
 775         catch(exception& e) {
 776                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function inUsersGroups. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 777                 exit(1);
 778         }
 779 }
 780 /**************************************************************************************************/
 781 //returns true if any of the strings in first vector are in second vector
 782 inline bool inUsersGroups(vector<string> groupnames, vector<string> Groups) {
 783         try {
 784
 785                 for (int i = 0; i < groupnames.size(); i++) {
 786                         if (inUsersGroups(groupnames[i], Groups)) { return true; }
 787                 }
 788                 return false;
 789         }
 790         catch(exception& e) {
 791                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function inUsersGroups. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 792                 exit(1);
 793         }
 794 }
 795 /***********************************************************************/
 796 //this function determines if the user has given us labels that are smaller than the given label.
 797 //if so then it returns true so that the calling function can run the previous valid distance.
 798 //it's a "smart" distance function.  It also checks for invalid labels.
 799 inline bool anyLabelsToProcess(string label, set<string>& userLabels, string errorOff) {
 800         try {
 801
 802                 set<string>::iterator it;
 803                 vector<float> orderFloat;
 804                 map<string, float> userMap;  //the conversion process removes trailing 0's which we need to put back
 805                 map<string, float>::iterator it2;
 806                 float labelFloat;
 807                 bool smaller = false;
 808
 809                 //unique is the smallest line
 810                 if (label == "unique") {  return false;  }
 811                 else { convert(label, labelFloat); }
 812
 813                 //go through users set and make them floats
 814                 for(it = userLabels.begin(); it != userLabels.end(); ++it) {
 815
 816                         float temp;
 817                         if ((*it != "unique") && (convertTestFloat(*it, temp) == true)){
 818                                 convert(*it, temp);
 819                                 orderFloat.push_back(temp);
 820                                 userMap[*it] = temp;
 821                         }else if (*it == "unique") {
 822                                 orderFloat.push_back(-1.0);
 823                                 userMap["unique"] = -1.0;
 824                         }else {
 825                                 if (errorOff == "") {  cout << *it << " is not a valid label." << endl;  }
 826                                 userLabels.erase(*it);
 827                                 it--;
 828                         }
 829                 }
 830
 831                 //sort order
 832                 sort(orderFloat.begin(), orderFloat.end());
 833
 834                 /*************************************************/
 835                 //is this label bigger than any of the users labels
 836                 /*************************************************/
 837
 838                 //loop through order until you find a label greater than label
 839                 for (int i = 0; i < orderFloat.size(); i++) {
 840                         if (orderFloat[i] < labelFloat) {
 841                                 smaller = true;
 842                                 if (orderFloat[i] == -1) {
 843                                         if (errorOff == "") { cout << "Your file does not include the label unique." << endl; }
 844                                         userLabels.erase("unique");
 845                                 }
 846                                 else {
 847                                         if (errorOff == "") { cout << "Your file does not include the label " << endl; }
 848                                         string s = "";
 849                                         for (it2 = userMap.begin(); it2!= userMap.end(); it2++) {
 850                                                 if (it2->second == orderFloat[i]) {
 851                                                         s = it2->first;
 852                                                         //remove small labels
 853                                                         userLabels.erase(s);
 854                                                         break;
 855                                                 }
 856                                         }
 857                                         if (errorOff == "") {cout << s <<  ". I will use the next smallest distance. " << endl; }
 858                                 }
 859                         //since they are sorted once you find a bigger one stop looking
 860                         }else { break; }
 861                 }
 862
 863                 return smaller;
 864
 865         }
 866         catch(exception& e) {
 867                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function anyLabelsToProcess. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 868                 exit(1);
 869         }
 870 }
 871
 872 /**************************************************************************************************/
 873 inline void appendFiles(string temp, string filename) {
 874         try{
 875                 ofstream output;
 876                 ifstream input;
 877
 878                 //open output file in append mode
 879                 openOutputFileAppend(filename, output);
 880                 int ableToOpen = openInputFile(temp, input, "no error");
 881
 882                 if (ableToOpen == 0) { //you opened it
 883                         while(char c = input.get()){
 884                                 if(input.eof())         {       break;                  }
 885                                 else                            {       output << c;    }
 886                         }
 887                         input.close();
 888                 }
 889
 890                 output.close();
 891         }
 892         catch(exception& e) {
 893                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function appendFiles. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 894                 exit(1);
 895         }
 896 }
 897
 898 /**************************************************************************************************/
 899 inline string sortFile(string distFile, string outputDir){
 900         try {
 901
 902                 //if (outputDir == "") {  outputDir += hasPath(distFile);  }
 903                 string outfile = getRootName(distFile) + "sorted.dist";
 904
 905
 906                 //if you can, use the unix sort since its been optimized for years
 907                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 908                         string command = "sort -n -k +3 " + distFile + " -o " + outfile;
 909                         system(command.c_str());
 910                 #else //you are stuck with my best attempt...
 911                         //windows sort does not have a way to specify a column, only a character in the line
 912                         //since we cannot assume that the distance will always be at the the same character location on each line
 913                         //due to variable sequence name lengths, I chose to force the distance into first position, then sort and then put it back.
 914
 915                         //read in file line by file and put distance first
 916                         string tempDistFile = distFile + ".temp";
 917                         ifstream input;
 918                         ofstream output;
 919                         openInputFile(distFile, input);
 920                         openOutputFile(tempDistFile, output);
 921
 922                         string firstName, secondName;
 923                         float dist;
 924                         while (input) {
 925                                 input >> firstName >> secondName >> dist;
 926                                 output << dist << '\t' << firstName << '\t' << secondName << endl;
 927                                 gobble(input);
 928                         }
 929                         input.close();
 930                         output.close();
 931
 932
 933                         //sort using windows sort
 934                         string tempOutfile = outfile + ".temp";
 935                         string command = "sort " + tempDistFile + " /O " + tempOutfile;
 936                         system(command.c_str());
 937
 938                         //read in sorted file and put distance at end again
 939                         ifstream input2;
 940                         openInputFile(tempOutfile, input2);
 941                         openOutputFile(outfile, output);
 942
 943                         while (input2) {
 944                                 input2 >> dist >> firstName >> secondName;
 945                                 output << firstName << '\t' << secondName << '\t' << dist << endl;
 946                                 gobble(input2);
 947                         }
 948                         input2.close();
 949                         output.close();
 950
 951                         //remove temp files
 952                         remove(tempDistFile.c_str());
 953                         remove(tempOutfile.c_str());
 954                 #endif
 955
 956                 return outfile;
 957         }
 958         catch(exception& e) {
 959                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function sortfile. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 960                 exit(1);
 961         }
 962 }
 963 /**************************************************************************************************/
 964 inline vector<long> setFilePosFasta(string filename, int& num) {
 965
 966                         vector<long> positions;
 967                         ifstream inFASTA;
 968                         openInputFile(filename, inFASTA);
 969
 970                         string input;
 971                         while(!inFASTA.eof()){
 972                                 input = getline(inFASTA); gobble(inFASTA);
 973                                 if (input.length() != 0) {
 974                                         if(input[0] == '>'){    long pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1);      }
 975                                 }
 976                         }
 977                         inFASTA.close();
 978
 979                         num = positions.size();
 980
 981                         /*FILE * pFile;
 982                         long size;
 983
 984                         //get num bytes in file
 985                         pFile = fopen (filename.c_str(),"rb");
 986                         if (pFile==NULL) perror ("Error opening file");
 987                         else{
 988                                 fseek (pFile, 0, SEEK_END);
 989                                 size=ftell (pFile);
 990                                 fclose (pFile);
 991                         }*/
 992
 993                         long size = positions[(positions.size()-1)];
 994                         ifstream in;
 995                         openInputFile(filename, in);
 996
 997                         in.seekg(size);
 998
 999                         while(char c = in.get()){
1000                                 if(in.eof())            {       break;  }
1001                                 else                            {       size++; }
1002                         }
1003                         in.close();
1004
1005                         positions.push_back(size);
1006
1007                         return positions;
1008 }
1009 /**************************************************************************************************/
1010 inline vector<long> setFilePosEachLine(string filename, int& num) {
1011
1012                         vector<long> positions;
1013                         ifstream in;
1014                         openInputFile(filename, in);
1015
1016                         string input;
1017                         while(!in.eof()){
1018                                 long lastpos = in.tellg();
1019                                 input = getline(in); gobble(in);
1020                                 if (input.length() != 0) {
1021                                         long pos = in.tellg();
1022                                         if (pos != -1) { positions.push_back(pos - input.length() - 1); }
1023                                         else {  positions.push_back(lastpos);  }
1024                                 }
1025                         }
1026                         in.close();
1027
1028                         num = positions.size();
1029
1030                         FILE * pFile;
1031                         long size;
1032
1033                         //get num bytes in file
1034                         pFile = fopen (filename.c_str(),"rb");
1035                         if (pFile==NULL) perror ("Error opening file");
1036                         else{
1037                                 fseek (pFile, 0, SEEK_END);
1038                                 size=ftell (pFile);
1039                                 fclose (pFile);
1040                         }
1041
1042                         positions.push_back(size);
1043
1044                         return positions;
1045 }
1046
1047 /**************************************************************************************************/
1048 #endif
1049