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 string getline(ifstream& fileHandle) {
 226         try {
 227
 228                 string line = "";
 229
 230                 while (!fileHandle.eof())       {
 231                         //get next character
 232                         char c = fileHandle.get();
 233
 234                         //are you at the end of the line
 235                         if ((c == '\n') || (c == '\r') || (c == '\f')){  break; }
 236                         else {          line += c;              }
 237                 }
 238
 239                 return line;
 240
 241         }
 242         catch(exception& e) {
 243                 cout << "Error in mothur function getline" << endl;
 244                 exit(1);
 245         }
 246 }
 247
 248 /***********************************************************************/
 249
 250 inline bool isTrue(string f){
 251
 252         if ((f == "TRUE") || (f == "T") || (f == "true") || (f == "t")) {       return true;    }
 253         else {  return false;  }
 254 }
 255
 256 /***********************************************************************/
 257
 258 inline float roundDist(float dist, int precision){
 259
 260         return int(dist * precision + 0.5)/float(precision);
 261
 262 }
 263 /***********************************************************************/
 264
 265 inline float ceilDist(float dist, int precision){
 266
 267         return int(ceil(dist * precision))/float(precision);
 268
 269 }
 270
 271 /***********************************************************************/
 272
 273 inline int getNumNames(string names){
 274
 275         int count = 0;
 276
 277         if(names != ""){
 278                 count = 1;
 279                 for(int i=0;i<names.size();i++){
 280                         if(names[i] == ','){
 281                                 count++;
 282                         }
 283                 }
 284         }
 285
 286         return count;
 287
 288 }
 289
 290 /**************************************************************************************************/
 291
 292 inline vector<vector<double> > binomial(int maxOrder){
 293
 294         vector<vector<double> > binomial(maxOrder+1);
 295
 296     for(int i=0;i<=maxOrder;i++){
 297                 binomial[i].resize(maxOrder+1);
 298                 binomial[i][0]=1;
 299                 binomial[0][i]=0;
 300     }
 301     binomial[0][0]=1;
 302
 303     binomial[1][0]=1;
 304     binomial[1][1]=1;
 305
 306     for(int i=2;i<=maxOrder;i++){
 307                 binomial[1][i]=0;
 308     }
 309
 310     for(int i=2;i<=maxOrder;i++){
 311                 for(int j=1;j<=maxOrder;j++){
 312                         if(i==j){       binomial[i][j]=1;                                                                       }
 313                         if(j>i) {       binomial[i][j]=0;                                                                       }
 314                         else    {       binomial[i][j]=binomial[i-1][j-1]+binomial[i-1][j];     }
 315                 }
 316     }
 317
 318         return binomial;
 319 }
 320
 321 /***********************************************************************/
 322
 323 inline string getRootName(string longName){
 324
 325         string rootName = longName;
 326
 327         if(longName.find_last_of(".") != longName.npos){
 328                 int pos = longName.find_last_of('.')+1;
 329                 rootName = longName.substr(0, pos);
 330         }
 331
 332         return rootName;
 333 }
 334 /***********************************************************************/
 335
 336 inline string getSimpleName(string longName){
 337
 338         string simpleName = longName;
 339
 340         size_t found;
 341         found=longName.find_last_of("/\\");
 342
 343         if(found != longName.npos){
 344                 simpleName = longName.substr(found+1);
 345         }
 346
 347                 //if(longName.find_last_of("/") != longName.npos){
 348                 //      int pos = longName.find_last_of('/')+1;
 349                 //      simpleName = longName.substr(pos, longName.length());
 350                 //}
 351
 352         return simpleName;
 353 }
 354
 355 /***********************************************************************/
 356
 357 inline int factorial(int num){
 358         int total = 1;
 359
 360         for (int i = 1; i <= num; i++) {
 361                 total *= i;
 362         }
 363
 364         return total;
 365 }
 366 /**************************************************************************************************
 367
 368 double min(double x, double y)
 369 {
 370     if(x<y){    return x;    }
 371     else   {    return y;    }
 372 }
 373
 374 /***********************************************************************/
 375
 376 inline string getPathName(string longName){
 377
 378         string rootPathName = longName;
 379
 380         if(longName.find_last_of("/\\") != longName.npos){
 381                 int pos = longName.find_last_of("/\\")+1;
 382                 rootPathName = longName.substr(0, pos);
 383         }
 384
 385         return rootPathName;
 386 }
 387 /***********************************************************************/
 388
 389 inline string hasPath(string longName){
 390
 391         string path = "";
 392
 393         size_t found;
 394         found=longName.find_last_of("~/\\");
 395
 396         if(found != longName.npos){
 397                 path = longName.substr(0, found+1);
 398         }
 399
 400         return path;
 401 }
 402
 403 /***********************************************************************/
 404
 405 inline string getExtension(string longName){
 406
 407         string extension = longName;
 408
 409         if(longName.find_last_of('.') != longName.npos){
 410                 int pos = longName.find_last_of('.');
 411                 extension = longName.substr(pos, longName.length());
 412         }
 413
 414         return extension;
 415 }
 416 /***********************************************************************/
 417 inline bool isBlank(string fileName){
 418
 419         ifstream fileHandle;
 420         fileHandle.open(fileName.c_str());
 421         if(!fileHandle) {
 422                 cout << "Error: Could not open " << fileName << endl;
 423                 return false;
 424         }else {
 425                 //check for blank file
 426                 gobble(fileHandle);
 427                 if (fileHandle.eof()) { fileHandle.close(); return true;  }
 428         }
 429         return false;
 430 }
 431 /***********************************************************************/
 432
 433 inline string getFullPathName(string fileName){
 434         try{
 435
 436         string path = hasPath(fileName);
 437         string newFileName;
 438         int pos;
 439
 440         if (path == "") { return fileName; } //its a simple name
 441         else { //we need to complete the pathname
 442                 // ex. ../../../filename
 443                 // cwd = /user/work/desktop
 444
 445                 string cwd;
 446                 //get current working directory
 447                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 448
 449                         if (path.find("~") != -1) { //go to home directory
 450                                 string homeDir = getenv ("HOME");
 451                                 newFileName = homeDir + fileName.substr(fileName.find("~")+1);
 452                                 return newFileName;
 453                         }else { //find path
 454                                 if (path.rfind("./") == -1) { return fileName; } //already complete name
 455                                 else { newFileName = fileName.substr(fileName.rfind("./")+2); } //save the complete part of the name
 456
 457                                 char* cwdpath = new char[1024];
 458
 459                                 size_t size;
 460                                 cwdpath=getcwd(cwdpath,size);
 461
 462                                 cwd = cwdpath;
 463
 464                                 //rip off first '/'
 465                                 string simpleCWD;
 466                                 if (cwd.length() > 0) { simpleCWD = cwd.substr(1); }
 467
 468                                 //break apart the current working directory
 469                                 vector<string> dirs;
 470                                 while (simpleCWD.find_first_of('/') != -1) {
 471                                         string dir = simpleCWD.substr(0,simpleCWD.find_first_of('/'));
 472                                         simpleCWD = simpleCWD.substr(simpleCWD.find_first_of('/')+1, simpleCWD.length());
 473                                         dirs.push_back(dir);
 474                                 }
 475                                 //get last one              // ex. ../../../filename = /user/work/desktop/filename
 476                                 dirs.push_back(simpleCWD);  //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
 477
 478
 479                                 int index = dirs.size()-1;
 480
 481                                 while((pos = path.rfind("./")) != -1) { //while you don't have a complete path
 482                                         if (pos == 0) { break;  //you are at the end
 483                                         }else if (path[(pos-1)] == '.') { //you want your parent directory ../
 484                                                 path = path.substr(0, pos-1);
 485                                                 index--;
 486                                                 if (index == 0) {  break; }
 487                                         }else if (path[(pos-1)] == '/') { //you want the current working dir ./
 488                                                 path = path.substr(0, pos);
 489                                         }else if (pos == 1) { break;  //you are at the end
 490                                         }else { cout << "cannot resolve path for " <<  fileName << endl; return fileName; }
 491                                 }
 492
 493                                 for (int i = index; i >= 0; i--) {
 494                                         newFileName = dirs[i] +  "/" + newFileName;
 495                                 }
 496
 497                                 newFileName =  "/" +  newFileName;
 498                                 return newFileName;
 499                         }
 500                 #else
 501                         if (path.find("~") != -1) { //go to home directory
 502                                 string homeDir = getenv ("HOMEPATH");
 503                                 newFileName = homeDir + fileName.substr(fileName.find("~")+1);
 504                                 return newFileName;
 505                         }else { //find path
 506                                 if (path.rfind(".\\") == -1) { return fileName; } //already complete name
 507                                 else { newFileName = fileName.substr(fileName.rfind(".\\")+2); } //save the complete part of the name
 508
 509                                 char *cwdpath = NULL;
 510                                 cwdpath = getcwd(NULL, 0); // or _getcwd
 511                                 if ( cwdpath != NULL) { cwd = cwdpath; }
 512                                 else { cwd = "";  }
 513
 514                                 //break apart the current working directory
 515                                 vector<string> dirs;
 516                                 while (cwd.find_first_of('\\') != -1) {
 517                                         string dir = cwd.substr(0,cwd.find_first_of('\\'));
 518                                         cwd = cwd.substr(cwd.find_first_of('\\')+1, cwd.length());
 519                                         dirs.push_back(dir);
 520
 521                                 }
 522                                 //get last one
 523                                 dirs.push_back(cwd);  //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
 524
 525                                 int index = dirs.size()-1;
 526
 527                                 while((pos = path.rfind(".\\")) != -1) { //while you don't have a complete path
 528                                         if (pos == 0) { break;  //you are at the end
 529                                         }else if (path[(pos-1)] == '.') { //you want your parent directory ../
 530                                                 path = path.substr(0, pos-1);
 531                                                 index--;
 532                                                 if (index == 0) {  break; }
 533                                         }else if (path[(pos-1)] == '\\') { //you want the current working dir ./
 534                                                 path = path.substr(0, pos);
 535                                         }else if (pos == 1) { break;  //you are at the end
 536                                         }else { cout << "cannot resolve path for " <<  fileName << endl; return fileName; }
 537                                 }
 538
 539                                 for (int i = index; i >= 0; i--) {
 540                                         newFileName = dirs[i] +  "\\" + newFileName;
 541                                 }
 542
 543                                 return newFileName;
 544                         }
 545
 546                 #endif
 547         }
 548         }
 549         catch(exception& e) {
 550                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function getFullPathName. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 551                 exit(1);
 552         }
 553 }
 554 /***********************************************************************/
 555
 556 inline int openInputFile(string fileName, ifstream& fileHandle, string m){
 557
 558         //get full path name
 559         string completeFileName = getFullPathName(fileName);
 560
 561         fileHandle.open(completeFileName.c_str());
 562         if(!fileHandle) {
 563                 return 1;
 564         }else {
 565                 //check for blank file
 566                 gobble(fileHandle);
 567                 return 0;
 568         }
 569 }
 570 /***********************************************************************/
 571
 572 inline int openInputFile(string fileName, ifstream& fileHandle){
 573
 574         //get full path name
 575         string completeFileName = getFullPathName(fileName);
 576
 577         fileHandle.open(completeFileName.c_str());
 578         if(!fileHandle) {
 579                 cout << "Error: Could not open " << completeFileName << endl;
 580                 return 1;
 581         }
 582         else {
 583                 //check for blank file
 584                 gobble(fileHandle);
 585                 if (fileHandle.eof()) { cout << completeFileName << " is blank. Please correct." << endl;   }
 586
 587                 return 0;
 588         }
 589
 590 }
 591 /***********************************************************************/
 592
 593 inline int renameFile(string oldName, string newName){
 594
 595         ifstream inTest;
 596         int exist = openInputFile(newName, inTest, "");
 597
 598 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 599         if (exist == 0) { //you could open it so you want to delete it
 600                 inTest.close();
 601                 string command = "rm " + newName;
 602                 system(command.c_str());
 603         }
 604
 605         string command = "mv " + oldName + " " + newName;
 606         system(command.c_str());
 607 #else
 608         remove(newName.c_str());
 609         int renameOk = rename(oldName.c_str(), newName.c_str());
 610 #endif
 611         return 0;
 612 }
 613
 614 /***********************************************************************/
 615
 616 inline int openOutputFile(string fileName, ofstream& fileHandle){
 617
 618         string completeFileName = getFullPathName(fileName);
 619
 620         fileHandle.open(completeFileName.c_str(), ios::trunc);
 621         if(!fileHandle) {
 622                 cout << "Error: Could not open " << completeFileName << endl;
 623                 return 1;
 624         }
 625         else {
 626                 return 0;
 627         }
 628
 629 }
 630
 631 /***********************************************************************/
 632
 633 inline int getNumSeqs(ifstream& file){
 634
 635         int numSeqs = count(istreambuf_iterator<char>(file),istreambuf_iterator<char>(), '>');
 636         file.seekg(0);
 637         return numSeqs;
 638
 639 }
 640 /***********************************************************************/
 641 inline void getNumSeqs(ifstream& file, int& numSeqs){
 642
 643         string input;
 644         numSeqs = 0;
 645         while(!file.eof()){
 646                 input = getline(file);
 647                 if (input.length() != 0) {
 648                         if(input[0] == '>'){ numSeqs++; }
 649                 }
 650         }
 651 }
 652
 653 /***********************************************************************/
 654
 655 inline bool inVector(string member, vector<string> group){
 656
 657         for (int i = 0; i < group.size(); i++) {
 658                 if (group[i] == member) {  return true;         }
 659         }
 660
 661         return false;
 662 }
 663 /***********************************************************************/
 664
 665 //This function parses the estimator options and puts them in a vector
 666 inline void splitAtDash(string& estim, vector<string>& container) {
 667         try {
 668                 string individual;
 669
 670                 while (estim.find_first_of('-') != -1) {
 671                         individual = estim.substr(0,estim.find_first_of('-'));
 672                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 673                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 674                                 container.push_back(individual);
 675                         }
 676                 }
 677                 //get last one
 678                 container.push_back(estim);
 679         }
 680         catch(exception& e) {
 681                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 682                 exit(1);
 683         }
 684 }
 685
 686 /***********************************************************************/
 687 //This function parses the label options and puts them in a set
 688 inline void splitAtDash(string& estim, set<string>& container) {
 689         try {
 690                 string individual;
 691
 692                 while (estim.find_first_of('-') != -1) {
 693                         individual = estim.substr(0,estim.find_first_of('-'));
 694                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 695                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 696                                 container.insert(individual);
 697                         }
 698                 }
 699                 //get last one
 700                 container.insert(estim);
 701         }
 702         catch(exception& e) {
 703                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 704                 exit(1);
 705         }
 706 }
 707 /***********************************************************************/
 708 //This function parses the line options and puts them in a set
 709 inline void splitAtDash(string& estim, set<int>& container) {
 710         try {
 711                 string individual;
 712                 int lineNum;
 713
 714                 while (estim.find_first_of('-') != -1) {
 715                         individual = estim.substr(0,estim.find_first_of('-'));
 716                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 717                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 718                                 convert(individual, lineNum); //convert the string to int
 719                                 container.insert(lineNum);
 720                         }
 721                 }
 722                 //get last one
 723                 convert(estim, lineNum); //convert the string to int
 724                 container.insert(lineNum);
 725         }
 726         catch(exception& e) {
 727                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 728                 exit(1);
 729         }
 730 }
 731 /***********************************************************************/
 732 //This function parses the a string and puts peices in a vector
 733 inline void splitAtComma(string& estim, vector<string>& container) {
 734         try {
 735                 string individual;
 736
 737                 while (estim.find_first_of(',') != -1) {
 738                         individual = estim.substr(0,estim.find_first_of(','));
 739                         if ((estim.find_first_of(',')+1) <= estim.length()) { //checks to make sure you don't have comma at end of string
 740                                 estim = estim.substr(estim.find_first_of(',')+1, estim.length());
 741                                 container.push_back(individual);
 742                         }
 743                 }
 744                 //get last one
 745                 container.push_back(estim);
 746         }
 747         catch(exception& e) {
 748                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtComma. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 749                 exit(1);
 750         }
 751 }
 752 /***********************************************************************/
 753
 754 //This function splits up the various option parameters
 755 inline void splitAtComma(string& prefix, string& suffix){
 756         try {
 757                 prefix = suffix.substr(0,suffix.find_first_of(','));
 758                 if ((suffix.find_first_of(',')+2) <= suffix.length()) {  //checks to make sure you don't have comma at end of string
 759                         suffix = suffix.substr(suffix.find_first_of(',')+1, suffix.length());
 760                         string space = " ";
 761                         while(suffix.at(0) == ' ')
 762                                 suffix = suffix.substr(1, suffix.length());
 763                 }
 764
 765         }
 766         catch(exception& e) {
 767                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtComma. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 768                 exit(1);
 769         }
 770 }
 771 /***********************************************************************/
 772
 773 //This function separates the key value from the option value i.e. dist=96_...
 774 inline void splitAtEquals(string& key, string& value){
 775         try {
 776                 if(value.find_first_of('=') != -1){
 777                         key = value.substr(0,value.find_first_of('='));
 778                         if ((value.find_first_of('=')+1) <= value.length()) {
 779                                 value = value.substr(value.find_first_of('=')+1, value.length());
 780                         }
 781                 }else{
 782                         key = value;
 783                         value = 1;
 784                 }
 785         }
 786         catch(exception& e) {
 787                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtEquals. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 788                 exit(1);
 789         }
 790 }
 791 /**************************************************************************************************/
 792
 793 inline bool inUsersGroups(string groupname, vector<string> Groups) {
 794         try {
 795                 for (int i = 0; i < Groups.size(); i++) {
 796                         if (groupname == Groups[i]) { return true; }
 797                 }
 798                 return false;
 799         }
 800         catch(exception& e) {
 801                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function inUsersGroups. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 802                 exit(1);
 803         }
 804 }
 805 /**************************************************************************************************/
 806 //returns true if any of the strings in first vector are in second vector
 807 inline bool inUsersGroups(vector<string> groupnames, vector<string> Groups) {
 808         try {
 809
 810                 for (int i = 0; i < groupnames.size(); i++) {
 811                         if (inUsersGroups(groupnames[i], Groups)) { return true; }
 812                 }
 813                 return false;
 814         }
 815         catch(exception& e) {
 816                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function inUsersGroups. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 817                 exit(1);
 818         }
 819 }
 820 /***********************************************************************/
 821 //this function determines if the user has given us labels that are smaller than the given label.
 822 //if so then it returns true so that the calling function can run the previous valid distance.
 823 //it's a "smart" distance function.  It also checks for invalid labels.
 824 inline bool anyLabelsToProcess(string label, set<string>& userLabels, string errorOff) {
 825         try {
 826
 827                 set<string>::iterator it;
 828                 vector<float> orderFloat;
 829                 map<string, float> userMap;  //the conversion process removes trailing 0's which we need to put back
 830                 map<string, float>::iterator it2;
 831                 float labelFloat;
 832                 bool smaller = false;
 833
 834                 //unique is the smallest line
 835                 if (label == "unique") {  return false;  }
 836                 else {
 837                         if (convertTestFloat(label, labelFloat)) {
 838                                 convert(label, labelFloat);
 839                         }else { //cant convert
 840                                 return false;
 841                         }
 842                 }
 843
 844                 //go through users set and make them floats
 845                 for(it = userLabels.begin(); it != userLabels.end(); ++it) {
 846
 847                         float temp;
 848                         if ((*it != "unique") && (convertTestFloat(*it, temp) == true)){
 849                                 convert(*it, temp);
 850                                 orderFloat.push_back(temp);
 851                                 userMap[*it] = temp;
 852                         }else if (*it == "unique") {
 853                                 orderFloat.push_back(-1.0);
 854                                 userMap["unique"] = -1.0;
 855                         }else {
 856                                 if (errorOff == "") {  cout << *it << " is not a valid label." << endl;  }
 857                                 userLabels.erase(*it);
 858                                 it--;
 859                         }
 860                 }
 861
 862                 //sort order
 863                 sort(orderFloat.begin(), orderFloat.end());
 864
 865                 /*************************************************/
 866                 //is this label bigger than any of the users labels
 867                 /*************************************************/
 868
 869                 //loop through order until you find a label greater than label
 870                 for (int i = 0; i < orderFloat.size(); i++) {
 871                         if (orderFloat[i] < labelFloat) {
 872                                 smaller = true;
 873                                 if (orderFloat[i] == -1) {
 874                                         if (errorOff == "") { cout << "Your file does not include the label unique." << endl; }
 875                                         userLabels.erase("unique");
 876                                 }
 877                                 else {
 878                                         if (errorOff == "") { cout << "Your file does not include the label " << endl; }
 879                                         string s = "";
 880                                         for (it2 = userMap.begin(); it2!= userMap.end(); it2++) {
 881                                                 if (it2->second == orderFloat[i]) {
 882                                                         s = it2->first;
 883                                                         //remove small labels
 884                                                         userLabels.erase(s);
 885                                                         break;
 886                                                 }
 887                                         }
 888                                         if (errorOff == "") {cout << s <<  ". I will use the next smallest distance. " << endl; }
 889                                 }
 890                         //since they are sorted once you find a bigger one stop looking
 891                         }else { break; }
 892                 }
 893
 894                 return smaller;
 895
 896         }
 897         catch(exception& e) {
 898                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function anyLabelsToProcess. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 899                 exit(1);
 900         }
 901 }
 902
 903 /**************************************************************************************************/
 904 inline void appendFiles(string temp, string filename) {
 905         try{
 906                 ofstream output;
 907                 ifstream input;
 908
 909                 //open output file in append mode
 910                 openOutputFileAppend(filename, output);
 911                 int ableToOpen = openInputFile(temp, input, "no error");
 912
 913                 if (ableToOpen == 0) { //you opened it
 914                         while(char c = input.get()){
 915                                 if(input.eof())         {       break;                  }
 916                                 else                            {       output << c;    }
 917                         }
 918                         input.close();
 919                 }
 920
 921                 output.close();
 922         }
 923         catch(exception& e) {
 924                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function appendFiles. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 925                 exit(1);
 926         }
 927 }
 928
 929 /**************************************************************************************************/
 930 inline string sortFile(string distFile, string outputDir){
 931         try {
 932
 933                 //if (outputDir == "") {  outputDir += hasPath(distFile);  }
 934                 string outfile = getRootName(distFile) + "sorted.dist";
 935
 936
 937                 //if you can, use the unix sort since its been optimized for years
 938                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 939                         string command = "sort -n -k +3 " + distFile + " -o " + outfile;
 940                         system(command.c_str());
 941                 #else //you are stuck with my best attempt...
 942                         //windows sort does not have a way to specify a column, only a character in the line
 943                         //since we cannot assume that the distance will always be at the the same character location on each line
 944                         //due to variable sequence name lengths, I chose to force the distance into first position, then sort and then put it back.
 945
 946                         //read in file line by file and put distance first
 947                         string tempDistFile = distFile + ".temp";
 948                         ifstream input;
 949                         ofstream output;
 950                         openInputFile(distFile, input);
 951                         openOutputFile(tempDistFile, output);
 952
 953                         string firstName, secondName;
 954                         float dist;
 955                         while (input) {
 956                                 input >> firstName >> secondName >> dist;
 957                                 output << dist << '\t' << firstName << '\t' << secondName << endl;
 958                                 gobble(input);
 959                         }
 960                         input.close();
 961                         output.close();
 962
 963
 964                         //sort using windows sort
 965                         string tempOutfile = outfile + ".temp";
 966                         string command = "sort " + tempDistFile + " /O " + tempOutfile;
 967                         system(command.c_str());
 968
 969                         //read in sorted file and put distance at end again
 970                         ifstream input2;
 971                         openInputFile(tempOutfile, input2);
 972                         openOutputFile(outfile, output);
 973
 974                         while (input2) {
 975                                 input2 >> dist >> firstName >> secondName;
 976                                 output << firstName << '\t' << secondName << '\t' << dist << endl;
 977                                 gobble(input2);
 978                         }
 979                         input2.close();
 980                         output.close();
 981
 982                         //remove temp files
 983                         remove(tempDistFile.c_str());
 984                         remove(tempOutfile.c_str());
 985                 #endif
 986
 987                 return outfile;
 988         }
 989         catch(exception& e) {
 990                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function sortfile. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 991                 exit(1);
 992         }
 993 }
 994 /**************************************************************************************************/
 995 inline vector<long> setFilePosFasta(string filename, int& num) {
 996
 997                         vector<long> positions;
 998                         ifstream inFASTA;
 999                         openInputFile(filename, inFASTA);
1000
1001                         string input;
1002                         while(!inFASTA.eof()){
1003                                 input = getline(inFASTA); gobble(inFASTA);
1004                                 if (input.length() != 0) {
1005                                         if(input[0] == '>'){    long pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1);      }
1006                                 }
1007                         }
1008                         inFASTA.close();
1009
1010                         num = positions.size();
1011
1012                         /*FILE * pFile;
1013                         long size;
1014
1015                         //get num bytes in file
1016                         pFile = fopen (filename.c_str(),"rb");
1017                         if (pFile==NULL) perror ("Error opening file");
1018                         else{
1019                                 fseek (pFile, 0, SEEK_END);
1020                                 size=ftell (pFile);
1021                                 fclose (pFile);
1022                         }*/
1023
1024                         long size = positions[(positions.size()-1)];
1025                         ifstream in;
1026                         openInputFile(filename, in);
1027
1028                         in.seekg(size);
1029
1030                         while(char c = in.get()){
1031                                 if(in.eof())            {       break;  }
1032                                 else                            {       size++; }
1033                         }
1034                         in.close();
1035
1036                         positions.push_back(size);
1037
1038                         return positions;
1039 }
1040 /**************************************************************************************************/
1041 inline vector<long> setFilePosEachLine(string filename, int& num) {
1042
1043                         vector<long> positions;
1044                         ifstream in;
1045                         openInputFile(filename, in);
1046
1047                         string input;
1048                         while(!in.eof()){
1049                                 long lastpos = in.tellg();
1050                                 input = getline(in); gobble(in);
1051                                 if (input.length() != 0) {
1052                                         long pos = in.tellg();
1053                                         if (pos != -1) { positions.push_back(pos - input.length() - 1); }
1054                                         else {  positions.push_back(lastpos);  }
1055                                 }
1056                         }
1057                         in.close();
1058
1059                         num = positions.size();
1060
1061                         FILE * pFile;
1062                         long size;
1063
1064                         //get num bytes in file
1065                         pFile = fopen (filename.c_str(),"rb");
1066                         if (pFile==NULL) perror ("Error opening file");
1067                         else{
1068                                 fseek (pFile, 0, SEEK_END);
1069                                 size=ftell (pFile);
1070                                 fclose (pFile);
1071                         }
1072
1073                         positions.push_back(size);
1074
1075                         return positions;
1076 }
1077
1078 /**************************************************************************************************/
1079 #endif
1080