mothur.h

   1 #ifndef MOTHUR_H
   2 #define MOTHUR_H
   3
   4
   5
   6 /*
   7  *  mothur.h
   8  *  Mothur
   9  *
  10  *  Created by Sarah Westcott on 2/19/09.
  11  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
  12  *
  13  */
  14
  15 /* This file contains all the standard incudes we use in the project as well as some common utilities. */
  16
  17 //#include <cstddef>
  18
  19 //io libraries
  20 #include <iostream>
  21 #include <iomanip>
  22 #include <fstream>
  23 #include <sstream>
  24 #include <signal.h>
  25
  26
  27 //exception
  28 #include <stdexcept>
  29 #include <exception>
  30 #include <cstdlib>
  31
  32
  33 //containers
  34 #include <vector>
  35 #include <set>
  36 #include <map>
  37 #include <string>
  38 #include <list>
  39
  40 //math
  41 #include <cmath>
  42 #include <math.h>
  43 #include <algorithm>
  44
  45 //misc
  46 #include <cerrno>
  47 #include <ctime>
  48 #include <limits>
  49
  50 #ifdef USE_MPI
  51         #include "mpi.h"
  52 #endif
  53 /***********************************************************************/
  54
  55 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
  56         #include <sys/wait.h>
  57         #include <unistd.h>
  58
  59         #ifdef USE_READLINE
  60                 #include <readline/readline.h>
  61                 #include <readline/history.h>
  62         #endif
  63
  64 #else
  65         #include <conio.h> //allows unbuffered screen capture from stdin
  66         #include <direct.h> //get cwd
  67 #endif
  68
  69 using namespace std;
  70
  71 #define exp(x) (exp((double) x))
  72 #define sqrt(x) (sqrt((double) x))
  73 #define log10(x) (log10((double) x))
  74 #define log2(x) (log10(x)/log10(2))
  75 #define isnan(x) ((x) != (x))
  76 #define isinf(x) (fabs(x) == std::numeric_limits<double>::infinity())
  77
  78 typedef unsigned long ull;
  79
  80 struct IntNode {
  81         int lvalue;
  82         int rvalue;
  83         int lcoef;
  84         int rcoef;
  85         IntNode* left;
  86         IntNode* right;
  87
  88         IntNode(int lv, int rv, IntNode* l, IntNode* r) : lvalue(lv), rvalue(rv), left(l), right(r) {};
  89         IntNode() {};
  90 };
  91
  92 struct ThreadNode {
  93         int* pid;
  94         IntNode* left;
  95         IntNode* right;
  96 };
  97
  98 /************************************************************/
  99 struct clusterNode {
 100         int numSeq;
 101         int parent;
 102         int smallChild; //used to make linkTable work with list and rabund. represents bin number of this cluster node
 103         clusterNode(int num, int par, int kid) : numSeq(num), parent(par), smallChild(kid) {};
 104 };
 105 /************************************************************/
 106 struct seqDist {
 107         int seq1;
 108         int seq2;
 109         float dist;
 110         seqDist() {}
 111         seqDist(int s1, int s2, float d) : seq1(s1), seq2(s2), dist(d) {}
 112         ~seqDist() {}
 113 };
 114 //********************************************************************************************************************
 115 //sorts lowest to highest
 116 inline bool compareSequenceDistance(seqDist left, seqDist right){
 117         return (left.dist < right.dist);
 118 }
 119 /***********************************************************************/
 120
 121 // snagged from http://www.parashift.com/c++-faq-lite/misc-technical-issues.html#faq-39.2
 122 // works for now, but there should be a way to do it without killing the whole program
 123
 124 class BadConversion : public runtime_error {
 125 public:
 126         BadConversion(const string& s) : runtime_error(s){ }
 127 };
 128
 129 //**********************************************************************************************************************
 130
 131 template<typename T>
 132 inline void convert(const string& s, T& x, bool failIfLeftoverChars = true){
 133         istringstream i(s);
 134         char c;
 135         if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
 136                 throw BadConversion(s);
 137 }
 138
 139 //**********************************************************************************************************************
 140
 141 template<typename T>
 142 inline bool convertTestFloat(const string& s, T& x, bool failIfLeftoverChars = true){
 143         istringstream i(s);
 144         char c;
 145         if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
 146         {
 147                 return false;
 148         }
 149         return true;
 150 }
 151
 152 //**********************************************************************************************************************
 153
 154 template<typename T>
 155 inline bool convertTest(const string& s, T& x, bool failIfLeftoverChars = true){
 156         istringstream i(s);
 157         char c;
 158         if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
 159         {
 160                 cout << "unable to be converted into an integer.\n" << endl;
 161                 return false;
 162         }
 163         return true;
 164 }
 165
 166 //**********************************************************************************************************************
 167
 168 template<typename T>
 169 string toString(const T&x){
 170     stringstream output;
 171     output << x;
 172     return output.str();
 173 }
 174
 175 //**********************************************************************************************************************
 176
 177 template<typename T>
 178 string toHex(const T&x){
 179         stringstream output;
 180
 181         output << hex << x;
 182
 183     return output.str();
 184 }
 185 //**********************************************************************************************************************
 186
 187 template<typename T>
 188 string toString(const T&x, int i){
 189         stringstream output;
 190
 191         output.precision(i);
 192     output << fixed << x;
 193
 194     return output.str();
 195 }
 196 /***********************************************************************/
 197 inline int openOutputFileAppend(string fileName, ofstream& fileHandle){
 198
 199         fileHandle.open(fileName.c_str(), ios::app);
 200         if(!fileHandle) {
 201                 cout << "Error: Could not open " << fileName << endl;
 202                 return 1;
 203         }
 204         else {
 205                 return 0;
 206         }
 207
 208 }
 209 /***********************************************************************/
 210
 211 inline void gobble(istream& f){
 212
 213         char d;
 214     while(isspace(d=f.get()))           {;}
 215         f.putback(d);
 216
 217 }
 218 /***********************************************************************/
 219
 220 inline string getline(ifstream& fileHandle) {
 221         try {
 222
 223                 string line = "";
 224
 225                 while (!fileHandle.eof())       {
 226                         //get next character
 227                         char c = fileHandle.get();
 228
 229                         //are you at the end of the line
 230                         if ((c == '\n') || (c == '\r') || (c == '\f')){  break; }
 231                         else {          line += c;              }
 232                 }
 233
 234                 return line;
 235
 236         }
 237         catch(exception& e) {
 238                 cout << "Error in mothur function getline" << endl;
 239                 exit(1);
 240         }
 241 }
 242
 243 /***********************************************************************/
 244
 245 inline bool isTrue(string f){
 246
 247         if ((f == "TRUE") || (f == "T") || (f == "true") || (f == "t")) {       return true;    }
 248         else {  return false;  }
 249 }
 250
 251 /***********************************************************************/
 252
 253 inline float roundDist(float dist, int precision){
 254
 255         return int(dist * precision + 0.5)/float(precision);
 256
 257 }
 258 /***********************************************************************/
 259
 260 inline float ceilDist(float dist, int precision){
 261
 262         return int(ceil(dist * precision))/float(precision);
 263
 264 }
 265
 266 /***********************************************************************/
 267
 268 inline int getNumNames(string names){
 269
 270         int count = 0;
 271
 272         if(names != ""){
 273                 count = 1;
 274                 for(int i=0;i<names.size();i++){
 275                         if(names[i] == ','){
 276                                 count++;
 277                         }
 278                 }
 279         }
 280
 281         return count;
 282
 283 }
 284
 285 /**************************************************************************************************/
 286
 287 inline vector<vector<double> > binomial(int maxOrder){
 288
 289         vector<vector<double> > binomial(maxOrder+1);
 290
 291     for(int i=0;i<=maxOrder;i++){
 292                 binomial[i].resize(maxOrder+1);
 293                 binomial[i][0]=1;
 294                 binomial[0][i]=0;
 295     }
 296     binomial[0][0]=1;
 297
 298     binomial[1][0]=1;
 299     binomial[1][1]=1;
 300
 301     for(int i=2;i<=maxOrder;i++){
 302                 binomial[1][i]=0;
 303     }
 304
 305     for(int i=2;i<=maxOrder;i++){
 306                 for(int j=1;j<=maxOrder;j++){
 307                         if(i==j){       binomial[i][j]=1;                                                                       }
 308                         if(j>i) {       binomial[i][j]=0;                                                                       }
 309                         else    {       binomial[i][j]=binomial[i-1][j-1]+binomial[i-1][j];     }
 310                 }
 311     }
 312
 313         return binomial;
 314 }
 315
 316 /***********************************************************************/
 317
 318 inline string getRootName(string longName){
 319
 320         string rootName = longName;
 321
 322         if(longName.find_last_of(".") != longName.npos){
 323                 int pos = longName.find_last_of('.')+1;
 324                 rootName = longName.substr(0, pos);
 325         }
 326
 327         return rootName;
 328 }
 329 /***********************************************************************/
 330
 331 inline string getSimpleName(string longName){
 332
 333         string simpleName = longName;
 334
 335         size_t found;
 336         found=longName.find_last_of("/\\");
 337
 338         if(found != longName.npos){
 339                 simpleName = longName.substr(found+1);
 340         }
 341
 342                 //if(longName.find_last_of("/") != longName.npos){
 343                 //      int pos = longName.find_last_of('/')+1;
 344                 //      simpleName = longName.substr(pos, longName.length());
 345                 //}
 346
 347         return simpleName;
 348 }
 349
 350 /***********************************************************************/
 351
 352 inline int factorial(int num){
 353         int total = 1;
 354
 355         for (int i = 1; i <= num; i++) {
 356                 total *= i;
 357         }
 358
 359         return total;
 360 }
 361 /**************************************************************************************************
 362
 363 double min(double x, double y)
 364 {
 365     if(x<y){    return x;    }
 366     else   {    return y;    }
 367 }
 368
 369 /***********************************************************************/
 370
 371 inline string getPathName(string longName){
 372
 373         string rootPathName = longName;
 374
 375         if(longName.find_last_of("/\\") != longName.npos){
 376                 int pos = longName.find_last_of("/\\")+1;
 377                 rootPathName = longName.substr(0, pos);
 378         }
 379
 380         return rootPathName;
 381 }
 382 /***********************************************************************/
 383
 384 inline string hasPath(string longName){
 385
 386         string path = "";
 387
 388         size_t found;
 389         found=longName.find_last_of("~/\\");
 390
 391         if(found != longName.npos){
 392                 path = longName.substr(0, found+1);
 393         }
 394
 395         return path;
 396 }
 397
 398 /***********************************************************************/
 399
 400 inline string getExtension(string longName){
 401
 402         string extension = longName;
 403
 404         if(longName.find_last_of('.') != longName.npos){
 405                 int pos = longName.find_last_of('.');
 406                 extension = longName.substr(pos, longName.length());
 407         }
 408
 409         return extension;
 410 }
 411 /***********************************************************************/
 412 inline bool isBlank(string fileName){
 413
 414         ifstream fileHandle;
 415         fileHandle.open(fileName.c_str());
 416         if(!fileHandle) {
 417                 cout << "Error: Could not open " << fileName << endl;
 418                 return false;
 419         }else {
 420                 //check for blank file
 421                 gobble(fileHandle);
 422                 if (fileHandle.eof()) { fileHandle.close(); return true;  }
 423         }
 424         return false;
 425 }
 426 /***********************************************************************/
 427
 428 inline string getFullPathName(string fileName){
 429         try{
 430
 431         string path = hasPath(fileName);
 432         string newFileName;
 433         int pos;
 434
 435         if (path == "") { return fileName; } //its a simple name
 436         else { //we need to complete the pathname
 437                 // ex. ../../../filename
 438                 // cwd = /user/work/desktop
 439
 440                 string cwd;
 441                 //get current working directory
 442                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 443
 444                         if (path.find("~") != -1) { //go to home directory
 445                                 string homeDir = getenv ("HOME");
 446                                 newFileName = homeDir + fileName.substr(fileName.find("~")+1);
 447                                 return newFileName;
 448                         }else { //find path
 449                                 if (path.rfind("./") == -1) { return fileName; } //already complete name
 450                                 else { newFileName = fileName.substr(fileName.rfind("./")+2); } //save the complete part of the name
 451
 452                                 char* cwdpath = new char[1024];
 453
 454                                 size_t size;
 455                                 cwdpath=getcwd(cwdpath,size);
 456
 457                                 cwd = cwdpath;
 458
 459                                 //rip off first '/'
 460                                 string simpleCWD;
 461                                 if (cwd.length() > 0) { simpleCWD = cwd.substr(1); }
 462
 463                                 //break apart the current working directory
 464                                 vector<string> dirs;
 465                                 while (simpleCWD.find_first_of('/') != -1) {
 466                                         string dir = simpleCWD.substr(0,simpleCWD.find_first_of('/'));
 467                                         simpleCWD = simpleCWD.substr(simpleCWD.find_first_of('/')+1, simpleCWD.length());
 468                                         dirs.push_back(dir);
 469                                 }
 470                                 //get last one              // ex. ../../../filename = /user/work/desktop/filename
 471                                 dirs.push_back(simpleCWD);  //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
 472
 473
 474                                 int index = dirs.size()-1;
 475
 476                                 while((pos = path.rfind("./")) != -1) { //while you don't have a complete path
 477                                         if (pos == 0) { break;  //you are at the end
 478                                         }else if (path[(pos-1)] == '.') { //you want your parent directory ../
 479                                                 path = path.substr(0, pos-1);
 480                                                 index--;
 481                                                 if (index == 0) {  break; }
 482                                         }else if (path[(pos-1)] == '/') { //you want the current working dir ./
 483                                                 path = path.substr(0, pos);
 484                                         }else if (pos == 1) { break;  //you are at the end
 485                                         }else { cout << "cannot resolve path for " <<  fileName << endl; return fileName; }
 486                                 }
 487
 488                                 for (int i = index; i >= 0; i--) {
 489                                         newFileName = dirs[i] +  "/" + newFileName;
 490                                 }
 491
 492                                 newFileName =  "/" +  newFileName;
 493                                 return newFileName;
 494                         }
 495                 #else
 496                         if (path.find("~") != -1) { //go to home directory
 497                                 string homeDir = getenv ("HOMEPATH");
 498                                 newFileName = homeDir + fileName.substr(fileName.find("~")+1);
 499                                 return newFileName;
 500                         }else { //find path
 501                                 if (path.rfind(".\\") == -1) { return fileName; } //already complete name
 502                                 else { newFileName = fileName.substr(fileName.rfind(".\\")+2); } //save the complete part of the name
 503
 504                                 char *cwdpath = NULL;
 505                                 cwdpath = getcwd(NULL, 0); // or _getcwd
 506                                 if ( cwdpath != NULL) { cwd = cwdpath; }
 507                                 else { cwd = "";  }
 508
 509                                 //break apart the current working directory
 510                                 vector<string> dirs;
 511                                 while (cwd.find_first_of('\\') != -1) {
 512                                         string dir = cwd.substr(0,cwd.find_first_of('\\'));
 513                                         cwd = cwd.substr(cwd.find_first_of('\\')+1, cwd.length());
 514                                         dirs.push_back(dir);
 515
 516                                 }
 517                                 //get last one
 518                                 dirs.push_back(cwd);  //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
 519
 520                                 int index = dirs.size()-1;
 521
 522                                 while((pos = path.rfind(".\\")) != -1) { //while you don't have a complete path
 523                                         if (pos == 0) { break;  //you are at the end
 524                                         }else if (path[(pos-1)] == '.') { //you want your parent directory ../
 525                                                 path = path.substr(0, pos-1);
 526                                                 index--;
 527                                                 if (index == 0) {  break; }
 528                                         }else if (path[(pos-1)] == '\\') { //you want the current working dir ./
 529                                                 path = path.substr(0, pos);
 530                                         }else if (pos == 1) { break;  //you are at the end
 531                                         }else { cout << "cannot resolve path for " <<  fileName << endl; return fileName; }
 532                                 }
 533
 534                                 for (int i = index; i >= 0; i--) {
 535                                         newFileName = dirs[i] +  "\\" + newFileName;
 536                                 }
 537
 538                                 return newFileName;
 539                         }
 540
 541                 #endif
 542         }
 543         }
 544         catch(exception& e) {
 545                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function getFullPathName. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 546                 exit(1);
 547         }
 548 }
 549 /***********************************************************************/
 550
 551 inline int openInputFile(string fileName, ifstream& fileHandle, string m){
 552
 553         //get full path name
 554         string completeFileName = getFullPathName(fileName);
 555
 556         fileHandle.open(completeFileName.c_str());
 557         if(!fileHandle) {
 558                 return 1;
 559         }else {
 560                 //check for blank file
 561                 gobble(fileHandle);
 562                 return 0;
 563         }
 564 }
 565 /***********************************************************************/
 566
 567 inline int openInputFile(string fileName, ifstream& fileHandle){
 568
 569         //get full path name
 570         string completeFileName = getFullPathName(fileName);
 571
 572         fileHandle.open(completeFileName.c_str());
 573         if(!fileHandle) {
 574                 cout << "Error: Could not open " << completeFileName << endl;
 575                 return 1;
 576         }
 577         else {
 578                 //check for blank file
 579                 gobble(fileHandle);
 580                 if (fileHandle.eof()) { cout << completeFileName << " is blank. Please correct." << endl;  return 1;  }
 581
 582                 return 0;
 583         }
 584
 585 }
 586 /***********************************************************************/
 587
 588 inline int renameFile(string oldName, string newName){
 589
 590         ifstream inTest;
 591         int exist = openInputFile(newName, inTest, "");
 592
 593 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 594         if (exist == 0) { //you could open it so you want to delete it
 595                 inTest.close();
 596                 string command = "rm " + newName;
 597                 system(command.c_str());
 598         }
 599
 600         string command = "mv " + oldName + " " + newName;
 601         system(command.c_str());
 602 #else
 603         remove(newName.c_str());
 604         int renameOk = rename(oldName.c_str(), newName.c_str());
 605 #endif
 606         return 0;
 607 }
 608
 609 /***********************************************************************/
 610
 611 inline int openOutputFile(string fileName, ofstream& fileHandle){
 612
 613         string completeFileName = getFullPathName(fileName);
 614
 615         fileHandle.open(completeFileName.c_str(), ios::trunc);
 616         if(!fileHandle) {
 617                 cout << "Error: Could not open " << completeFileName << endl;
 618                 return 1;
 619         }
 620         else {
 621                 return 0;
 622         }
 623
 624 }
 625
 626 /***********************************************************************/
 627
 628 inline int getNumSeqs(ifstream& file){
 629
 630         int numSeqs = count(istreambuf_iterator<char>(file),istreambuf_iterator<char>(), '>');
 631         file.seekg(0);
 632         return numSeqs;
 633
 634 }
 635 /***********************************************************************/
 636
 637 inline bool inVector(string member, vector<string> group){
 638
 639         for (int i = 0; i < group.size(); i++) {
 640                 if (group[i] == member) {  return true;         }
 641         }
 642
 643         return false;
 644 }
 645 /***********************************************************************/
 646
 647 //This function parses the estimator options and puts them in a vector
 648 inline void splitAtDash(string& estim, vector<string>& container) {
 649         try {
 650                 string individual;
 651
 652                 while (estim.find_first_of('-') != -1) {
 653                         individual = estim.substr(0,estim.find_first_of('-'));
 654                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 655                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 656                                 container.push_back(individual);
 657                         }
 658                 }
 659                 //get last one
 660                 container.push_back(estim);
 661         }
 662         catch(exception& e) {
 663                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 664                 exit(1);
 665         }
 666 }
 667
 668 /***********************************************************************/
 669 //This function parses the label options and puts them in a set
 670 inline void splitAtDash(string& estim, set<string>& container) {
 671         try {
 672                 string individual;
 673
 674                 while (estim.find_first_of('-') != -1) {
 675                         individual = estim.substr(0,estim.find_first_of('-'));
 676                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 677                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 678                                 container.insert(individual);
 679                         }
 680                 }
 681                 //get last one
 682                 container.insert(estim);
 683         }
 684         catch(exception& e) {
 685                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 686                 exit(1);
 687         }
 688 }
 689 /***********************************************************************/
 690 //This function parses the line options and puts them in a set
 691 inline void splitAtDash(string& estim, set<int>& container) {
 692         try {
 693                 string individual;
 694                 int lineNum;
 695
 696                 while (estim.find_first_of('-') != -1) {
 697                         individual = estim.substr(0,estim.find_first_of('-'));
 698                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 699                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 700                                 convert(individual, lineNum); //convert the string to int
 701                                 container.insert(lineNum);
 702                         }
 703                 }
 704                 //get last one
 705                 convert(estim, lineNum); //convert the string to int
 706                 container.insert(lineNum);
 707         }
 708         catch(exception& e) {
 709                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 710                 exit(1);
 711         }
 712 }
 713 /***********************************************************************/
 714 //This function parses the a string and puts peices in a vector
 715 inline void splitAtComma(string& estim, vector<string>& container) {
 716         try {
 717                 string individual;
 718
 719                 while (estim.find_first_of(',') != -1) {
 720                         individual = estim.substr(0,estim.find_first_of(','));
 721                         if ((estim.find_first_of(',')+1) <= estim.length()) { //checks to make sure you don't have comma at end of string
 722                                 estim = estim.substr(estim.find_first_of(',')+1, estim.length());
 723                                 container.push_back(individual);
 724                         }
 725                 }
 726                 //get last one
 727                 container.push_back(estim);
 728         }
 729         catch(exception& e) {
 730                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtComma. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 731                 exit(1);
 732         }
 733 }
 734 /***********************************************************************/
 735
 736 //This function splits up the various option parameters
 737 inline void splitAtComma(string& prefix, string& suffix){
 738         try {
 739                 prefix = suffix.substr(0,suffix.find_first_of(','));
 740                 if ((suffix.find_first_of(',')+2) <= suffix.length()) {  //checks to make sure you don't have comma at end of string
 741                         suffix = suffix.substr(suffix.find_first_of(',')+1, suffix.length());
 742                         string space = " ";
 743                         while(suffix.at(0) == ' ')
 744                                 suffix = suffix.substr(1, suffix.length());
 745                 }
 746
 747         }
 748         catch(exception& e) {
 749                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtComma. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 750                 exit(1);
 751         }
 752 }
 753 /***********************************************************************/
 754
 755 //This function separates the key value from the option value i.e. dist=96_...
 756 inline void splitAtEquals(string& key, string& value){
 757         try {
 758                 if(value.find_first_of('=') != -1){
 759                         key = value.substr(0,value.find_first_of('='));
 760                         if ((value.find_first_of('=')+1) <= value.length()) {
 761                                 value = value.substr(value.find_first_of('=')+1, value.length());
 762                         }
 763                 }else{
 764                         key = value;
 765                         value = 1;
 766                 }
 767         }
 768         catch(exception& e) {
 769                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtEquals. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 770                 exit(1);
 771         }
 772 }
 773 /**************************************************************************************************/
 774
 775 inline bool inUsersGroups(string groupname, vector<string> Groups) {
 776         try {
 777                 for (int i = 0; i < Groups.size(); i++) {
 778                         if (groupname == Groups[i]) { return true; }
 779                 }
 780                 return false;
 781         }
 782         catch(exception& e) {
 783                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function inUsersGroups. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 784                 exit(1);
 785         }
 786 }
 787 /**************************************************************************************************/
 788 //returns true if any of the strings in first vector are in second vector
 789 inline bool inUsersGroups(vector<string> groupnames, vector<string> Groups) {
 790         try {
 791
 792                 for (int i = 0; i < groupnames.size(); i++) {
 793                         if (inUsersGroups(groupnames[i], Groups)) { return true; }
 794                 }
 795                 return false;
 796         }
 797         catch(exception& e) {
 798                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function inUsersGroups. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 799                 exit(1);
 800         }
 801 }
 802 /***********************************************************************/
 803 //this function determines if the user has given us labels that are smaller than the given label.
 804 //if so then it returns true so that the calling function can run the previous valid distance.
 805 //it's a "smart" distance function.  It also checks for invalid labels.
 806 inline bool anyLabelsToProcess(string label, set<string>& userLabels, string errorOff) {
 807         try {
 808
 809                 set<string>::iterator it;
 810                 vector<float> orderFloat;
 811                 map<string, float> userMap;  //the conversion process removes trailing 0's which we need to put back
 812                 map<string, float>::iterator it2;
 813                 float labelFloat;
 814                 bool smaller = false;
 815
 816                 //unique is the smallest line
 817                 if (label == "unique") {  return false;  }
 818                 else { convert(label, labelFloat); }
 819
 820                 //go through users set and make them floats
 821                 for(it = userLabels.begin(); it != userLabels.end(); ++it) {
 822
 823                         float temp;
 824                         if ((*it != "unique") && (convertTestFloat(*it, temp) == true)){
 825                                 convert(*it, temp);
 826                                 orderFloat.push_back(temp);
 827                                 userMap[*it] = temp;
 828                         }else if (*it == "unique") {
 829                                 orderFloat.push_back(-1.0);
 830                                 userMap["unique"] = -1.0;
 831                         }else {
 832                                 if (errorOff == "") {  cout << *it << " is not a valid label." << endl;  }
 833                                 userLabels.erase(*it);
 834                                 it--;
 835                         }
 836                 }
 837
 838                 //sort order
 839                 sort(orderFloat.begin(), orderFloat.end());
 840
 841                 /*************************************************/
 842                 //is this label bigger than any of the users labels
 843                 /*************************************************/
 844
 845                 //loop through order until you find a label greater than label
 846                 for (int i = 0; i < orderFloat.size(); i++) {
 847                         if (orderFloat[i] < labelFloat) {
 848                                 smaller = true;
 849                                 if (orderFloat[i] == -1) {
 850                                         if (errorOff == "") { cout << "Your file does not include the label unique." << endl; }
 851                                         userLabels.erase("unique");
 852                                 }
 853                                 else {
 854                                         if (errorOff == "") { cout << "Your file does not include the label " << endl; }
 855                                         string s = "";
 856                                         for (it2 = userMap.begin(); it2!= userMap.end(); it2++) {
 857                                                 if (it2->second == orderFloat[i]) {
 858                                                         s = it2->first;
 859                                                         //remove small labels
 860                                                         userLabels.erase(s);
 861                                                         break;
 862                                                 }
 863                                         }
 864                                         if (errorOff == "") {cout << s <<  ". I will use the next smallest distance. " << endl; }
 865                                 }
 866                         //since they are sorted once you find a bigger one stop looking
 867                         }else { break; }
 868                 }
 869
 870                 return smaller;
 871
 872         }
 873         catch(exception& e) {
 874                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function anyLabelsToProcess. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 875                 exit(1);
 876         }
 877 }
 878
 879 /**************************************************************************************************/
 880 inline void appendFiles(string temp, string filename) {
 881         try{
 882                 ofstream output;
 883                 ifstream input;
 884
 885                 //open output file in append mode
 886                 openOutputFileAppend(filename, output);
 887                 int ableToOpen = openInputFile(temp, input, "no error");
 888
 889                 if (ableToOpen == 0) { //you opened it
 890                         while(char c = input.get()){
 891                                 if(input.eof())         {       break;                  }
 892                                 else                            {       output << c;    }
 893                         }
 894                         input.close();
 895                 }
 896
 897                 output.close();
 898         }
 899         catch(exception& e) {
 900                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function appendFiles. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 901                 exit(1);
 902         }
 903 }
 904
 905 /**************************************************************************************************/
 906 inline string sortFile(string distFile, string outputDir){
 907         try {
 908
 909                 //if (outputDir == "") {  outputDir += hasPath(distFile);  }
 910                 string outfile = getRootName(distFile) + "sorted.dist";
 911
 912
 913                 //if you can, use the unix sort since its been optimized for years
 914                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 915                         string command = "sort -n -k +3 " + distFile + " -o " + outfile;
 916                         system(command.c_str());
 917                 #else //you are stuck with my best attempt...
 918                         //windows sort does not have a way to specify a column, only a character in the line
 919                         //since we cannot assume that the distance will always be at the the same character location on each line
 920                         //due to variable sequence name lengths, I chose to force the distance into first position, then sort and then put it back.
 921
 922                         //read in file line by file and put distance first
 923                         string tempDistFile = distFile + ".temp";
 924                         ifstream input;
 925                         ofstream output;
 926                         openInputFile(distFile, input);
 927                         openOutputFile(tempDistFile, output);
 928
 929                         string firstName, secondName;
 930                         float dist;
 931                         while (input) {
 932                                 input >> firstName >> secondName >> dist;
 933                                 output << dist << '\t' << firstName << '\t' << secondName << endl;
 934                                 gobble(input);
 935                         }
 936                         input.close();
 937                         output.close();
 938
 939
 940                         //sort using windows sort
 941                         string tempOutfile = outfile + ".temp";
 942                         string command = "sort " + tempDistFile + " /O " + tempOutfile;
 943                         system(command.c_str());
 944
 945                         //read in sorted file and put distance at end again
 946                         ifstream input2;
 947                         openInputFile(tempOutfile, input2);
 948                         openOutputFile(outfile, output);
 949
 950                         while (input2) {
 951                                 input2 >> dist >> firstName >> secondName;
 952                                 output << firstName << '\t' << secondName << '\t' << dist << endl;
 953                                 gobble(input2);
 954                         }
 955                         input2.close();
 956                         output.close();
 957
 958                         //remove temp files
 959                         remove(tempDistFile.c_str());
 960                         remove(tempOutfile.c_str());
 961                 #endif
 962
 963                 return outfile;
 964         }
 965         catch(exception& e) {
 966                 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function sortfile. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
 967                 exit(1);
 968         }
 969 }
 970 /**************************************************************************************************/
 971 inline vector<long> setFilePosFasta(string filename, int& num) {
 972
 973                         vector<long> positions;
 974                         ifstream inFASTA;
 975                         openInputFile(filename, inFASTA);
 976
 977                         string input;
 978                         while(!inFASTA.eof()){
 979                                 input = getline(inFASTA); gobble(inFASTA);
 980                                 if (input.length() != 0) {
 981                                         if(input[0] == '>'){    long pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1);      }
 982                                 }
 983                         }
 984                         inFASTA.close();
 985
 986                         num = positions.size();
 987
 988                         /*FILE * pFile;
 989                         long size;
 990
 991                         //get num bytes in file
 992                         pFile = fopen (filename.c_str(),"rb");
 993                         if (pFile==NULL) perror ("Error opening file");
 994                         else{
 995                                 fseek (pFile, 0, SEEK_END);
 996                                 size=ftell (pFile);
 997                                 fclose (pFile);
 998                         }*/
 999
1000                         long size = positions[(positions.size()-1)];
1001                         ifstream in;
1002                         openInputFile(filename, in);
1003
1004                         in.seekg(size);
1005
1006                         while(char c = in.get()){
1007                                 if(in.eof())            {       break;  }
1008                                 else                            {       size++; }
1009                         }
1010                         in.close();
1011
1012                         positions.push_back(size);
1013
1014                         return positions;
1015 }
1016 /**************************************************************************************************/
1017 inline vector<long> setFilePosEachLine(string filename, int& num) {
1018
1019                         vector<long> positions;
1020                         ifstream in;
1021                         openInputFile(filename, in);
1022
1023                         string input;
1024                         while(!in.eof()){
1025                                 long lastpos = in.tellg();
1026                                 input = getline(in); gobble(in);
1027                                 if (input.length() != 0) {
1028                                         long pos = in.tellg();
1029                                         if (pos != -1) { positions.push_back(pos - input.length() - 1); }
1030                                         else {  positions.push_back(lastpos);  }
1031                                 }
1032                         }
1033                         in.close();
1034
1035                         num = positions.size();
1036
1037                         FILE * pFile;
1038                         long size;
1039
1040                         //get num bytes in file
1041                         pFile = fopen (filename.c_str(),"rb");
1042                         if (pFile==NULL) perror ("Error opening file");
1043                         else{
1044                                 fseek (pFile, 0, SEEK_END);
1045                                 size=ftell (pFile);
1046                                 fclose (pFile);
1047                         }
1048
1049                         positions.push_back(size);
1050
1051                         return positions;
1052 }
1053
1054 /**************************************************************************************************/
1055 #endif
1056