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 <sys/time.h>
  58         #include <sys/resource.h>
  59         #include <unistd.h>
  60
  61         #ifdef USE_READLINE
  62                 #include <readline/readline.h>
  63                 #include <readline/history.h>
  64         #endif
  65
  66 #else
  67         #include <conio.h> //allows unbuffered screen capture from stdin
  68         #include <direct.h> //get cwd
  69         #include <windows.h>
  70         #include <psapi.h>
  71
  72 #endif
  73
  74 using namespace std;
  75
  76 #define exp(x) (exp((double) x))
  77 #define sqrt(x) (sqrt((double) x))
  78 #define log10(x) (log10((double) x))
  79 #define log2(x) (log10(x)/log10(2))
  80 #define isnan(x) ((x) != (x))
  81 #define isinf(x) (fabs(x) == std::numeric_limits<double>::infinity())
  82
  83 typedef unsigned long ull;
  84
  85 struct IntNode {
  86         int lvalue;
  87         int rvalue;
  88         int lcoef;
  89         int rcoef;
  90         IntNode* left;
  91         IntNode* right;
  92
  93         IntNode(int lv, int rv, IntNode* l, IntNode* r) : lvalue(lv), rvalue(rv), left(l), right(r) {};
  94         IntNode() {};
  95 };
  96
  97 struct ThreadNode {
  98         int* pid;
  99         IntNode* left;
 100         IntNode* right;
 101 };
 102
 103 /************************************************************/
 104 struct clusterNode {
 105         int numSeq;
 106         int parent;
 107         int smallChild; //used to make linkTable work with list and rabund. represents bin number of this cluster node
 108         clusterNode(int num, int par, int kid) : numSeq(num), parent(par), smallChild(kid) {};
 109 };
 110 /************************************************************/
 111 struct seqDist {
 112         int seq1;
 113         int seq2;
 114         float dist;
 115         seqDist() {}
 116         seqDist(int s1, int s2, float d) : seq1(s1), seq2(s2), dist(d) {}
 117         ~seqDist() {}
 118 };
 119 //********************************************************************************************************************
 120 //sorts lowest to highest
 121 inline bool compareSequenceDistance(seqDist left, seqDist right){
 122         return (left.dist < right.dist);
 123 }
 124 /***********************************************************************/
 125
 126 // snagged from http://www.parashift.com/c++-faq-lite/misc-technical-issues.html#faq-39.2
 127 // works for now, but there should be a way to do it without killing the whole program
 128
 129 class BadConversion : public runtime_error {
 130 public:
 131         BadConversion(const string& s) : runtime_error(s){ }
 132 };
 133
 134 //**********************************************************************************************************************
 135
 136 template<typename T>
 137 inline void convert(const string& s, T& x, bool failIfLeftoverChars = true){
 138         istringstream i(s);
 139         char c;
 140         if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
 141                 throw BadConversion(s);
 142 }
 143
 144 //**********************************************************************************************************************
 145
 146 template<typename T>
 147 inline bool convertTestFloat(const string& s, T& x, bool failIfLeftoverChars = true){
 148         istringstream i(s);
 149         char c;
 150         if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
 151         {
 152                 return false;
 153         }
 154         return true;
 155 }
 156
 157 //**********************************************************************************************************************
 158
 159 template<typename T>
 160 inline bool convertTest(const string& s, T& x, bool failIfLeftoverChars = true){
 161         istringstream i(s);
 162         char c;
 163         if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
 164         {
 165                 cout << "unable to be converted into an integer.\n" << endl;
 166                 return false;
 167         }
 168         return true;
 169 }
 170
 171 //**********************************************************************************************************************
 172
 173 template<typename T>
 174 string toString(const T&x){
 175     stringstream output;
 176     output << x;
 177     return output.str();
 178 }
 179
 180 //**********************************************************************************************************************
 181
 182 template<typename T>
 183 string toHex(const T&x){
 184         stringstream output;
 185
 186         output << hex << x;
 187
 188     return output.str();
 189 }
 190 //**********************************************************************************************************************
 191
 192 template<typename T>
 193 string toString(const T&x, int i){
 194         stringstream output;
 195
 196         output.precision(i);
 197     output << fixed << x;
 198
 199     return output.str();
 200 }
 201 /***********************************************************************/
 202 inline int openOutputFileAppend(string fileName, ofstream& fileHandle){
 203
 204         fileHandle.open(fileName.c_str(), ios::app);
 205         if(!fileHandle) {
 206                 cout << "Error: Could not open " << fileName << endl;
 207                 return 1;
 208         }
 209         else {
 210                 return 0;
 211         }
 212
 213 }
 214 /***********************************************************************/
 215
 216 inline void gobble(istream& f){
 217
 218         char d;
 219     while(isspace(d=f.get()))           {;}
 220         f.putback(d);
 221
 222 }
 223 /***********************************************************************/
 224
 225 inline void gobble(istringstream& f){
 226
 227         char d;
 228     while(isspace(d=f.get()))           {;}
 229         f.putback(d);
 230
 231 }
 232
 233 /***********************************************************************/
 234
 235 inline string getline(istringstream& fileHandle) {
 236         try {
 237
 238                 string line = "";
 239
 240                 while (!fileHandle.eof())       {
 241                         //get next character
 242                         char c = fileHandle.get();
 243
 244                         //are you at the end of the line
 245                         if ((c == '\n') || (c == '\r') || (c == '\f')){  break; }
 246                         else {          line += c;              }
 247                 }
 248
 249                 return line;
 250
 251         }
 252         catch(exception& e) {
 253                 cout << "Error in mothur function getline" << endl;
 254                 exit(1);
 255         }
 256 }
 257 /***********************************************************************/
 258
 259 inline string getline(ifstream& fileHandle) {
 260         try {
 261
 262                 string line = "";
 263
 264                 while (!fileHandle.eof())       {
 265                         //get next character
 266                         char c = fileHandle.get();
 267
 268                         //are you at the end of the line
 269                         if ((c == '\n') || (c == '\r') || (c == '\f')){  break; }
 270                         else {          line += c;              }
 271                 }
 272
 273                 return line;
 274
 275         }
 276         catch(exception& e) {
 277                 cout << "Error in mothur function getline" << endl;
 278                 exit(1);
 279         }
 280 }
 281 /***********************************************************************/
 282
 283 inline bool isTrue(string f){
 284
 285         if ((f == "TRUE") || (f == "T") || (f == "true") || (f == "t")) {       return true;    }
 286         else {  return false;  }
 287 }
 288
 289 /***********************************************************************/
 290
 291 inline float roundDist(float dist, int precision){
 292
 293         return int(dist * precision + 0.5)/float(precision);
 294
 295 }
 296 /***********************************************************************/
 297
 298 inline float ceilDist(float dist, int precision){
 299
 300         return int(ceil(dist * precision))/float(precision);
 301
 302 }
 303
 304 /***********************************************************************/
 305
 306 inline int getNumNames(string names){
 307
 308         int count = 0;
 309
 310         if(names != ""){
 311                 count = 1;
 312                 for(int i=0;i<names.size();i++){
 313                         if(names[i] == ','){
 314                                 count++;
 315                         }
 316                 }
 317         }
 318
 319         return count;
 320
 321 }
 322
 323 /**************************************************************************************************/
 324
 325 inline vector<vector<double> > binomial(int maxOrder){
 326
 327         vector<vector<double> > binomial(maxOrder+1);
 328
 329     for(int i=0;i<=maxOrder;i++){
 330                 binomial[i].resize(maxOrder+1);
 331                 binomial[i][0]=1;
 332                 binomial[0][i]=0;
 333     }
 334     binomial[0][0]=1;
 335
 336     binomial[1][0]=1;
 337     binomial[1][1]=1;
 338
 339     for(int i=2;i<=maxOrder;i++){
 340                 binomial[1][i]=0;
 341     }
 342
 343     for(int i=2;i<=maxOrder;i++){
 344                 for(int j=1;j<=maxOrder;j++){
 345                         if(i==j){       binomial[i][j]=1;                                                                       }
 346                         if(j>i) {       binomial[i][j]=0;                                                                       }
 347                         else    {       binomial[i][j]=binomial[i-1][j-1]+binomial[i-1][j];     }
 348                 }
 349     }
 350
 351         return binomial;
 352 }
 353
 354 /***********************************************************************/
 355
 356 inline string getRootName(string longName){
 357
 358         string rootName = longName;
 359
 360         if(longName.find_last_of(".") != longName.npos){
 361                 int pos = longName.find_last_of('.')+1;
 362                 rootName = longName.substr(0, pos);
 363         }
 364
 365         return rootName;
 366 }
 367 /***********************************************************************/
 368
 369 inline string getSimpleName(string longName){
 370
 371         string simpleName = longName;
 372
 373         size_t found;
 374         found=longName.find_last_of("/\\");
 375
 376         if(found != longName.npos){
 377                 simpleName = longName.substr(found+1);
 378         }
 379
 380                 //if(longName.find_last_of("/") != longName.npos){
 381                 //      int pos = longName.find_last_of('/')+1;
 382                 //      simpleName = longName.substr(pos, longName.length());
 383                 //}
 384
 385         return simpleName;
 386 }
 387
 388 /***********************************************************************/
 389
 390 inline int factorial(int num){
 391         int total = 1;
 392
 393         for (int i = 1; i <= num; i++) {
 394                 total *= i;
 395         }
 396
 397         return total;
 398 }
 399 /**************************************************************************************************
 400
 401 double min(double x, double y)
 402 {
 403     if(x<y){    return x;    }
 404     else   {    return y;    }
 405 }
 406
 407 /***********************************************************************/
 408
 409 inline string getPathName(string longName){
 410
 411         string rootPathName = longName;
 412
 413         if(longName.find_last_of("/\\") != longName.npos){
 414                 int pos = longName.find_last_of("/\\")+1;
 415                 rootPathName = longName.substr(0, pos);
 416         }
 417
 418         return rootPathName;
 419 }
 420 /***********************************************************************/
 421
 422 inline string hasPath(string longName){
 423
 424         string path = "";
 425
 426         size_t found;
 427         found=longName.find_last_of("~/\\");
 428
 429         if(found != longName.npos){
 430                 path = longName.substr(0, found+1);
 431         }
 432
 433         return path;
 434 }
 435
 436 /***********************************************************************/
 437
 438 inline string getExtension(string longName){
 439
 440         string extension = longName;
 441
 442         if(longName.find_last_of('.') != longName.npos){
 443                 int pos = longName.find_last_of('.');
 444                 extension = longName.substr(pos, longName.length());
 445         }
 446
 447         return extension;
 448 }
 449 /***********************************************************************/
 450 inline bool isBlank(string fileName){
 451
 452         ifstream fileHandle;
 453         fileHandle.open(fileName.c_str());
 454         if(!fileHandle) {
 455                 cout << "Error: Could not open " << fileName << endl;
 456                 return false;
 457         }else {
 458                 //check for blank file
 459                 gobble(fileHandle);
 460                 if (fileHandle.eof()) { fileHandle.close(); return true;  }
 461         }
 462         return false;
 463 }
 464 /***********************************************************************/
 465
 466 inline string getFullPathName(string fileName){
 467         try{
 468
 469         string path = hasPath(fileName);
 470         string newFileName;
 471         int pos;
 472
 473         if (path == "") { return fileName; } //its a simple name
 474         else { //we need to complete the pathname
 475                 // ex. ../../../filename
 476                 // cwd = /user/work/desktop
 477
 478                 string cwd;
 479                 //get current working directory
 480                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 481
 482                         if (path.find("~") != -1) { //go to home directory
 483                                 string homeDir = getenv ("HOME");
 484                                 newFileName = homeDir + fileName.substr(fileName.find("~")+1);
 485                                 return newFileName;
 486                         }else { //find path
 487                                 if (path.rfind("./") == -1) { return fileName; } //already complete name
 488                                 else { newFileName = fileName.substr(fileName.rfind("./")+2); } //save the complete part of the name
 489
 490                                 char* cwdpath = new char[1024];
 491
 492                                 size_t size;
 493                                 cwdpath=getcwd(cwdpath,size);
 494
 495                                 cwd = cwdpath;
 496
 497                                 //rip off first '/'
 498                                 string simpleCWD;
 499                                 if (cwd.length() > 0) { simpleCWD = cwd.substr(1); }
 500
 501                                 //break apart the current working directory
 502                                 vector<string> dirs;
 503                                 while (simpleCWD.find_first_of('/') != -1) {
 504                                         string dir = simpleCWD.substr(0,simpleCWD.find_first_of('/'));
 505                                         simpleCWD = simpleCWD.substr(simpleCWD.find_first_of('/')+1, simpleCWD.length());
 506                                         dirs.push_back(dir);
 507                                 }
 508                                 //get last one              // ex. ../../../filename = /user/work/desktop/filename
 509                                 dirs.push_back(simpleCWD);  //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
 510
 511
 512                                 int index = dirs.size()-1;
 513
 514                                 while((pos = path.rfind("./")) != -1) { //while you don't have a complete path
 515                                         if (pos == 0) { break;  //you are at the end
 516                                         }else if (path[(pos-1)] == '.') { //you want your parent directory ../
 517                                                 path = path.substr(0, pos-1);
 518                                                 index--;
 519                                                 if (index == 0) {  break; }
 520                                         }else if (path[(pos-1)] == '/') { //you want the current working dir ./
 521                                                 path = path.substr(0, pos);
 522                                         }else if (pos == 1) { break;  //you are at the end
 523                                         }else { cout << "cannot resolve path for " <<  fileName << endl; return fileName; }
 524                                 }
 525
 526                                 for (int i = index; i >= 0; i--) {
 527                                         newFileName = dirs[i] +  "/" + newFileName;
 528                                 }
 529
 530                                 newFileName =  "/" +  newFileName;
 531                                 return newFileName;
 532                         }
 533                 #else
 534                         if (path.find("~") != -1) { //go to home directory
 535                                 string homeDir = getenv ("HOMEPATH");
 536                                 newFileName = homeDir + fileName.substr(fileName.find("~")+1);
 537                                 return newFileName;
 538                         }else { //find path
 539                                 if (path.rfind(".\\") == -1) { return fileName; } //already complete name
 540                                 else { newFileName = fileName.substr(fileName.rfind(".\\")+2); } //save the complete part of the name
 541
 542                                 char *cwdpath = NULL;
 543                                 cwdpath = getcwd(NULL, 0); // or _getcwd
 544                                 if ( cwdpath != NULL) { cwd = cwdpath; }
 545                                 else { cwd = "";  }
 546
 547                                 //break apart the current working directory
 548                                 vector<string> dirs;
 549                                 while (cwd.find_first_of('\\') != -1) {
 550                                         string dir = cwd.substr(0,cwd.find_first_of('\\'));
 551                                         cwd = cwd.substr(cwd.find_first_of('\\')+1, cwd.length());
 552                                         dirs.push_back(dir);
 553
 554                                 }
 555                                 //get last one
 556                                 dirs.push_back(cwd);  //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
 557
 558                                 int index = dirs.size()-1;
 559
 560                                 while((pos = path.rfind(".\\")) != -1) { //while you don't have a complete path
 561                                         if (pos == 0) { break;  //you are at the end
 562                                         }else if (path[(pos-1)] == '.') { //you want your parent directory ../
 563                                                 path = path.substr(0, pos-1);
 564                                                 index--;
 565                                                 if (index == 0) {  break; }
 566                                         }else if (path[(pos-1)] == '\\') { //you want the current working dir ./
 567                                                 path = path.substr(0, pos);
 568                                         }else if (pos == 1) { break;  //you are at the end
 569                                         }else { cout << "cannot resolve path for " <<  fileName << endl; return fileName; }
 570                                 }
 571
 572                                 for (int i = index; i >= 0; i--) {
 573                                         newFileName = dirs[i] +  "\\" + newFileName;
 574                                 }
 575
 576                                 return newFileName;
 577                         }
 578
 579                 #endif
 580         }
 581         }
 582         catch(exception& e) {
 583                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function getFullPathName. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 584                 exit(1);
 585         }
 586 }
 587 /***********************************************************************/
 588
 589 inline int openInputFile(string fileName, ifstream& fileHandle, string m){
 590
 591         //get full path name
 592         string completeFileName = getFullPathName(fileName);
 593
 594         fileHandle.open(completeFileName.c_str());
 595         if(!fileHandle) {
 596                 return 1;
 597         }else {
 598                 //check for blank file
 599                 gobble(fileHandle);
 600                 return 0;
 601         }
 602 }
 603 /***********************************************************************/
 604
 605 inline int openInputFile(string fileName, ifstream& fileHandle){
 606
 607         //get full path name
 608         string completeFileName = getFullPathName(fileName);
 609
 610         fileHandle.open(completeFileName.c_str());
 611         if(!fileHandle) {
 612                 cout << "Error: Could not open " << completeFileName << endl;
 613                 return 1;
 614         }
 615         else {
 616                 //check for blank file
 617                 gobble(fileHandle);
 618                 if (fileHandle.eof()) { cout << completeFileName << " is blank. Please correct." << endl;   }
 619
 620                 return 0;
 621         }
 622
 623 }
 624 /***********************************************************************/
 625
 626 inline int renameFile(string oldName, string newName){
 627
 628         ifstream inTest;
 629         int exist = openInputFile(newName, inTest, "");
 630
 631 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 632         if (exist == 0) { //you could open it so you want to delete it
 633                 inTest.close();
 634                 string command = "rm " + newName;
 635                 system(command.c_str());
 636         }
 637
 638         string command = "mv " + oldName + " " + newName;
 639         system(command.c_str());
 640 #else
 641         remove(newName.c_str());
 642         int renameOk = rename(oldName.c_str(), newName.c_str());
 643 #endif
 644         return 0;
 645 }
 646
 647 /***********************************************************************/
 648
 649 inline int openOutputFile(string fileName, ofstream& fileHandle){
 650
 651         string completeFileName = getFullPathName(fileName);
 652
 653         fileHandle.open(completeFileName.c_str(), ios::trunc);
 654         if(!fileHandle) {
 655                 cout << "Error: Could not open " << completeFileName << endl;
 656                 return 1;
 657         }
 658         else {
 659                 return 0;
 660         }
 661
 662 }
 663
 664 /***********************************************************************/
 665
 666 inline int getNumSeqs(ifstream& file){
 667
 668         int numSeqs = count(istreambuf_iterator<char>(file),istreambuf_iterator<char>(), '>');
 669         file.seekg(0);
 670         return numSeqs;
 671
 672 }
 673 /***********************************************************************/
 674 inline void getNumSeqs(ifstream& file, int& numSeqs){
 675
 676         string input;
 677         numSeqs = 0;
 678         while(!file.eof()){
 679                 input = getline(file);
 680                 if (input.length() != 0) {
 681                         if(input[0] == '>'){ numSeqs++; }
 682                 }
 683         }
 684 }
 685
 686 /***********************************************************************/
 687
 688 inline bool inVector(string member, vector<string> group){
 689
 690         for (int i = 0; i < group.size(); i++) {
 691                 if (group[i] == member) {  return true;         }
 692         }
 693
 694         return false;
 695 }
 696 /***********************************************************************/
 697
 698 //This function parses the estimator options and puts them in a vector
 699 inline void splitAtChar(string& estim, vector<string>& container, char symbol) {
 700         try {
 701                 string individual;
 702
 703                 while (estim.find_first_of(symbol) != -1) {
 704                         individual = estim.substr(0,estim.find_first_of(symbol));
 705                         if ((estim.find_first_of(symbol)+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 706                                 estim = estim.substr(estim.find_first_of(symbol)+1, estim.length());
 707                                 container.push_back(individual);
 708                         }
 709                 }
 710                 //get last one
 711                 container.push_back(estim);
 712         }
 713         catch(exception& e) {
 714                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 715                 exit(1);
 716         }
 717 }
 718
 719 /***********************************************************************/
 720
 721 //This function parses the estimator options and puts them in a vector
 722 inline void splitAtDash(string& estim, vector<string>& container) {
 723         try {
 724                 string individual;
 725
 726                 while (estim.find_first_of('-') != -1) {
 727                         individual = estim.substr(0,estim.find_first_of('-'));
 728                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 729                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 730                                 container.push_back(individual);
 731                         }
 732                 }
 733                 //get last one
 734                 container.push_back(estim);
 735         }
 736         catch(exception& e) {
 737                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 738                 exit(1);
 739         }
 740 }
 741
 742 /***********************************************************************/
 743 //This function parses the label options and puts them in a set
 744 inline void splitAtDash(string& estim, set<string>& container) {
 745         try {
 746                 string individual;
 747
 748                 while (estim.find_first_of('-') != -1) {
 749                         individual = estim.substr(0,estim.find_first_of('-'));
 750                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 751                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 752                                 container.insert(individual);
 753                         }
 754                 }
 755                 //get last one
 756                 container.insert(estim);
 757         }
 758         catch(exception& e) {
 759                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 760                 exit(1);
 761         }
 762 }
 763 /***********************************************************************/
 764 //This function parses the line options and puts them in a set
 765 inline void splitAtDash(string& estim, set<int>& container) {
 766         try {
 767                 string individual;
 768                 int lineNum;
 769
 770                 while (estim.find_first_of('-') != -1) {
 771                         individual = estim.substr(0,estim.find_first_of('-'));
 772                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 773                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 774                                 convert(individual, lineNum); //convert the string to int
 775                                 container.insert(lineNum);
 776                         }
 777                 }
 778                 //get last one
 779                 convert(estim, lineNum); //convert the string to int
 780                 container.insert(lineNum);
 781         }
 782         catch(exception& e) {
 783                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 784                 exit(1);
 785         }
 786 }
 787 /***********************************************************************/
 788 //This function parses the a string and puts peices in a vector
 789 inline void splitAtComma(string& estim, vector<string>& container) {
 790         try {
 791                 string individual;
 792
 793                 while (estim.find_first_of(',') != -1) {
 794                         individual = estim.substr(0,estim.find_first_of(','));
 795                         if ((estim.find_first_of(',')+1) <= estim.length()) { //checks to make sure you don't have comma at end of string
 796                                 estim = estim.substr(estim.find_first_of(',')+1, estim.length());
 797                                 container.push_back(individual);
 798                         }
 799                 }
 800                 //get last one
 801                 container.push_back(estim);
 802         }
 803         catch(exception& e) {
 804                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtComma. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 805                 exit(1);
 806         }
 807 }
 808 /***********************************************************************/
 809
 810 //This function splits up the various option parameters
 811 inline void splitAtComma(string& prefix, string& suffix){
 812         try {
 813                 prefix = suffix.substr(0,suffix.find_first_of(','));
 814                 if ((suffix.find_first_of(',')+2) <= suffix.length()) {  //checks to make sure you don't have comma at end of string
 815                         suffix = suffix.substr(suffix.find_first_of(',')+1, suffix.length());
 816                         string space = " ";
 817                         while(suffix.at(0) == ' ')
 818                                 suffix = suffix.substr(1, suffix.length());
 819                 }
 820
 821         }
 822         catch(exception& e) {
 823                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtComma. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 824                 exit(1);
 825         }
 826 }
 827 /***********************************************************************/
 828
 829 //This function separates the key value from the option value i.e. dist=96_...
 830 inline void splitAtEquals(string& key, string& value){
 831         try {
 832                 if(value.find_first_of('=') != -1){
 833                         key = value.substr(0,value.find_first_of('='));
 834                         if ((value.find_first_of('=')+1) <= value.length()) {
 835                                 value = value.substr(value.find_first_of('=')+1, value.length());
 836                         }
 837                 }else{
 838                         key = value;
 839                         value = 1;
 840                 }
 841         }
 842         catch(exception& e) {
 843                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtEquals. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 844                 exit(1);
 845         }
 846 }
 847
 848 /**************************************************************************************************/
 849
 850 inline bool inUsersGroups(string groupname, vector<string> Groups) {
 851         try {
 852                 for (int i = 0; i < Groups.size(); i++) {
 853                         if (groupname == Groups[i]) { return true; }
 854                 }
 855                 return false;
 856         }
 857         catch(exception& e) {
 858                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function inUsersGroups. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 859                 exit(1);
 860         }
 861 }
 862 /**************************************************************************************************/
 863 //returns true if any of the strings in first vector are in second vector
 864 inline bool inUsersGroups(vector<string> groupnames, vector<string> Groups) {
 865         try {
 866
 867                 for (int i = 0; i < groupnames.size(); i++) {
 868                         if (inUsersGroups(groupnames[i], Groups)) { return true; }
 869                 }
 870                 return false;
 871         }
 872         catch(exception& e) {
 873                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function inUsersGroups. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 874                 exit(1);
 875         }
 876 }
 877 /***********************************************************************/
 878 //this function determines if the user has given us labels that are smaller than the given label.
 879 //if so then it returns true so that the calling function can run the previous valid distance.
 880 //it's a "smart" distance function.  It also checks for invalid labels.
 881 inline bool anyLabelsToProcess(string label, set<string>& userLabels, string errorOff) {
 882         try {
 883
 884                 set<string>::iterator it;
 885                 vector<float> orderFloat;
 886                 map<string, float> userMap;  //the conversion process removes trailing 0's which we need to put back
 887                 map<string, float>::iterator it2;
 888                 float labelFloat;
 889                 bool smaller = false;
 890
 891                 //unique is the smallest line
 892                 if (label == "unique") {  return false;  }
 893                 else {
 894                         if (convertTestFloat(label, labelFloat)) {
 895                                 convert(label, labelFloat);
 896                         }else { //cant convert
 897                                 return false;
 898                         }
 899                 }
 900
 901                 //go through users set and make them floats
 902                 for(it = userLabels.begin(); it != userLabels.end(); ++it) {
 903
 904                         float temp;
 905                         if ((*it != "unique") && (convertTestFloat(*it, temp) == true)){
 906                                 convert(*it, temp);
 907                                 orderFloat.push_back(temp);
 908                                 userMap[*it] = temp;
 909                         }else if (*it == "unique") {
 910                                 orderFloat.push_back(-1.0);
 911                                 userMap["unique"] = -1.0;
 912                         }else {
 913                                 if (errorOff == "") {  cout << *it << " is not a valid label." << endl;  }
 914                                 userLabels.erase(*it);
 915                                 it--;
 916                         }
 917                 }
 918
 919                 //sort order
 920                 sort(orderFloat.begin(), orderFloat.end());
 921
 922                 /*************************************************/
 923                 //is this label bigger than any of the users labels
 924                 /*************************************************/
 925
 926                 //loop through order until you find a label greater than label
 927                 for (int i = 0; i < orderFloat.size(); i++) {
 928                         if (orderFloat[i] < labelFloat) {
 929                                 smaller = true;
 930                                 if (orderFloat[i] == -1) {
 931                                         if (errorOff == "") { cout << "Your file does not include the label unique." << endl; }
 932                                         userLabels.erase("unique");
 933                                 }
 934                                 else {
 935                                         if (errorOff == "") { cout << "Your file does not include the label " << endl; }
 936                                         string s = "";
 937                                         for (it2 = userMap.begin(); it2!= userMap.end(); it2++) {
 938                                                 if (it2->second == orderFloat[i]) {
 939                                                         s = it2->first;
 940                                                         //remove small labels
 941                                                         userLabels.erase(s);
 942                                                         break;
 943                                                 }
 944                                         }
 945                                         if (errorOff == "") {cout << s <<  ". I will use the next smallest distance. " << endl; }
 946                                 }
 947                         //since they are sorted once you find a bigger one stop looking
 948                         }else { break; }
 949                 }
 950
 951                 return smaller;
 952
 953         }
 954         catch(exception& e) {
 955                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function anyLabelsToProcess. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 956                 exit(1);
 957         }
 958 }
 959
 960 /**************************************************************************************************/
 961 inline void appendFiles(string temp, string filename) {
 962         try{
 963                 ofstream output;
 964                 ifstream input;
 965
 966                 //open output file in append mode
 967                 openOutputFileAppend(filename, output);
 968                 int ableToOpen = openInputFile(temp, input, "no error");
 969
 970                 if (ableToOpen == 0) { //you opened it
 971                         while(char c = input.get()){
 972                                 if(input.eof())         {       break;                  }
 973                                 else                            {       output << c;    }
 974                         }
 975                         input.close();
 976                 }
 977
 978                 output.close();
 979         }
 980         catch(exception& e) {
 981                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function appendFiles. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 982                 exit(1);
 983         }
 984 }
 985
 986 /**************************************************************************************************/
 987 inline string sortFile(string distFile, string outputDir){
 988         try {
 989
 990                 //if (outputDir == "") {  outputDir += hasPath(distFile);  }
 991                 string outfile = getRootName(distFile) + "sorted.dist";
 992
 993
 994                 //if you can, use the unix sort since its been optimized for years
 995                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 996                         string command = "sort -n -k +3 " + distFile + " -o " + outfile;
 997                         system(command.c_str());
 998                 #else //you are stuck with my best attempt...
 999                         //windows sort does not have a way to specify a column, only a character in the line
1000                         //since we cannot assume that the distance will always be at the the same character location on each line
1001                         //due to variable sequence name lengths, I chose to force the distance into first position, then sort and then put it back.
1002
1003                         //read in file line by file and put distance first
1004                         string tempDistFile = distFile + ".temp";
1005                         ifstream input;
1006                         ofstream output;
1007                         openInputFile(distFile, input);
1008                         openOutputFile(tempDistFile, output);
1009
1010                         string firstName, secondName;
1011                         float dist;
1012                         while (input) {
1013                                 input >> firstName >> secondName >> dist;
1014                                 output << dist << '\t' << firstName << '\t' << secondName << endl;
1015                                 gobble(input);
1016                         }
1017                         input.close();
1018                         output.close();
1019
1020
1021                         //sort using windows sort
1022                         string tempOutfile = outfile + ".temp";
1023                         string command = "sort " + tempDistFile + " /O " + tempOutfile;
1024                         system(command.c_str());
1025
1026                         //read in sorted file and put distance at end again
1027                         ifstream input2;
1028                         openInputFile(tempOutfile, input2);
1029                         openOutputFile(outfile, output);
1030
1031                         while (input2) {
1032                                 input2 >> dist >> firstName >> secondName;
1033                                 output << firstName << '\t' << secondName << '\t' << dist << endl;
1034                                 gobble(input2);
1035                         }
1036                         input2.close();
1037                         output.close();
1038
1039                         //remove temp files
1040                         remove(tempDistFile.c_str());
1041                         remove(tempOutfile.c_str());
1042                 #endif
1043
1044                 return outfile;
1045         }
1046         catch(exception& e) {
1047                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function sortfile. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
1048                 exit(1);
1049         }
1050 }
1051 /**************************************************************************************************/
1052 inline vector<unsigned long int> setFilePosFasta(string filename, int& num) {
1053
1054                         vector<unsigned long int> positions;
1055                         ifstream inFASTA;
1056                         openInputFile(filename, inFASTA);
1057
1058                         string input;
1059                         while(!inFASTA.eof()){
1060                                 input = getline(inFASTA);
1061                                 if (input.length() != 0) {
1062                                         if(input[0] == '>'){    unsigned long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); }
1063                                 }
1064                                 gobble(inFASTA); //has to be here since windows line endings are 2 characters and mess up the positions
1065                         }
1066                         inFASTA.close();
1067
1068                         num = positions.size();
1069
1070                         /*FILE * pFile;
1071                         long size;
1072
1073                         //get num bytes in file
1074                         pFile = fopen (filename.c_str(),"rb");
1075                         if (pFile==NULL) perror ("Error opening file");
1076                         else{
1077                                 fseek (pFile, 0, SEEK_END);
1078                                 size=ftell (pFile);
1079                                 fclose (pFile);
1080                         }*/
1081
1082                         unsigned long int size = positions[(positions.size()-1)];
1083                         ifstream in;
1084                         openInputFile(filename, in);
1085
1086                         in.seekg(size);
1087
1088                         while(char c = in.get()){
1089                                 if(in.eof())            {       break;  }
1090                                 else                            {       size++; }
1091                         }
1092                         in.close();
1093
1094                         positions.push_back(size);
1095
1096                         return positions;
1097 }
1098 /**************************************************************************************************/
1099 inline vector<unsigned long int> setFilePosEachLine(string filename, int& num) {
1100
1101                         vector<unsigned long int> positions;
1102                         ifstream in;
1103                         openInputFile(filename, in);
1104
1105                         string input;
1106                         while(!in.eof()){
1107                                 unsigned long int lastpos = in.tellg();
1108                                 input = getline(in);
1109                                 if (input.length() != 0) {
1110                                         unsigned long int pos = in.tellg();
1111                                         if (pos != -1) { positions.push_back(pos - input.length() - 1); }
1112                                         else {  positions.push_back(lastpos);  }
1113                                 }
1114                                 gobble(in); //has to be here since windows line endings are 2 characters and mess up the positions
1115                         }
1116                         in.close();
1117
1118                         num = positions.size();
1119
1120                         FILE * pFile;
1121                         unsigned long int size;
1122
1123                         //get num bytes in file
1124                         pFile = fopen (filename.c_str(),"rb");
1125                         if (pFile==NULL) perror ("Error opening file");
1126                         else{
1127                                 fseek (pFile, 0, SEEK_END);
1128                                 size=ftell (pFile);
1129                                 fclose (pFile);
1130                         }
1131
1132                         positions.push_back(size);
1133
1134                         return positions;
1135 }
1136 /**************************************************************************************************/
1137
1138 inline vector<unsigned long int> divideFile(string filename, int& proc) {
1139         try{
1140
1141                 vector<unsigned long int> filePos;
1142                 filePos.push_back(0);
1143
1144                 FILE * pFile;
1145                 unsigned long int size;
1146
1147                 //get num bytes in file
1148                 pFile = fopen (filename.c_str(),"rb");
1149                 if (pFile==NULL) perror ("Error opening file");
1150                 else{
1151                         fseek (pFile, 0, SEEK_END);
1152                         size=ftell (pFile);
1153                         fclose (pFile);
1154                 }
1155
1156                 //estimate file breaks
1157                 unsigned long int chunkSize = 0;
1158                 chunkSize = size / proc;
1159
1160                 //file to small to divide by processors
1161                 if (chunkSize == 0)  {  proc = 1;       filePos.push_back(size); return filePos;        }
1162
1163                 //for each process seekg to closest file break and search for next '>' char. make that the filebreak
1164                 for (int i = 0; i < proc; i++) {
1165                         unsigned long int spot = (i+1) * chunkSize;
1166
1167                         ifstream in;
1168                         openInputFile(filename, in);
1169                         in.seekg(spot);
1170
1171                         //look for next '>'
1172                         unsigned long int newSpot = spot;
1173                         while (!in.eof()) {
1174                            char c = in.get();
1175                            if (c == '>') {   in.putback(c); newSpot = in.tellg(); break;  }
1176                         }
1177
1178                         //there was not another sequence before the end of the file
1179                         unsigned long int sanityPos = in.tellg();
1180                         if (sanityPos == -1) {  break;  }
1181                         else {   filePos.push_back(newSpot);  }
1182
1183                         in.close();
1184                 }
1185
1186                 //save end pos
1187                 filePos.push_back(size);
1188
1189                 //sanity check filePos
1190                 for (int i = 0; i < (filePos.size()-1); i++) {
1191                         if (filePos[(i+1)] <= filePos[i]) {  filePos.erase(filePos.begin()+(i+1)); i--; }
1192                 }
1193
1194                 proc = (filePos.size() - 1);
1195
1196                 return filePos;
1197         }
1198         catch(exception& e) {
1199                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function divideFile. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
1200                 exit(1);
1201         }
1202 }
1203 /**************************************************************************************************/
1204 inline bool checkReleaseVersion(ifstream& file, string version) {
1205         try {
1206
1207                 bool good = true;
1208
1209                 string line = getline(file);
1210
1211                 //before we added this check
1212                 if (line[0] != '#') {  good = false;  }
1213                 else {
1214                         //rip off #
1215                         line = line.substr(1);
1216
1217                         vector<string> versionVector;
1218                         splitAtChar(version, versionVector, '.');
1219
1220                         //check file version
1221                         vector<string> linesVector;
1222                         splitAtChar(line, linesVector, '.');
1223
1224                         if (versionVector.size() != linesVector.size()) { good = false; }
1225                         else {
1226                                 for (int j = 0; j < versionVector.size(); j++) {
1227                                         int num1, num2;
1228                                         convert(versionVector[j], num1);
1229                                         convert(linesVector[j], num2);
1230
1231                                         //if mothurs version is newer than this files version, then we want to remake it
1232                                         if (num1 > num2) {  good = false; break;  }
1233                                 }
1234                         }
1235
1236                 }
1237
1238                 if (!good) {  file.close();  }
1239                 else { file.seekg(0);  }
1240
1241                 return good;
1242         }
1243         catch(exception& e) {
1244                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function checkReleaseVersion. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
1245                 exit(1);
1246         }
1247 }
1248 /**************************************************************************************************/
1249 #endif
1250