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
  25 //exception
  26 #include <stdexcept>
  27 #include <exception>
  28 #include <cstdlib>
  29
  30
  31 //containers
  32 #include <vector>
  33 #include <set>
  34 #include <map>
  35 #include <string>
  36 #include <list>
  37
  38 //math
  39 #include <cmath>
  40 #include <math.h>
  41 #include <algorithm>
  42
  43 //misc
  44 #include <cerrno>
  45 #include <ctime>
  46 #include <limits>
  47
  48
  49 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
  50         #include <sys/wait.h>
  51         #include <unistd.h>
  52 #endif
  53
  54 using namespace std;
  55
  56 #define exp(x) (exp((double) x))
  57 #define sqrt(x) (sqrt((double) x))
  58 #define log10(x) (log10((double) x))
  59 #define log2(x) (log10(x)/log10(2))
  60 #define isnan(x) ((x) != (x))
  61 #define isinf(x) (fabs(x) == std::numeric_limits<double>::infinity())
  62
  63
  64 typedef unsigned long ull;
  65
  66 struct IntNode {
  67         int lvalue;
  68         int rvalue;
  69         int lcoef;
  70         int rcoef;
  71         IntNode* left;
  72         IntNode* right;
  73 };
  74
  75 struct ThreadNode {
  76         int* pid;
  77         IntNode* left;
  78         IntNode* right;
  79 };
  80
  81 /***********************************************************************/
  82
  83 // snagged from http://www.parashift.com/c++-faq-lite/misc-technical-issues.html#faq-39.2
  84 // works for now, but there should be a way to do it without killing the whole program
  85
  86 class BadConversion : public runtime_error {
  87 public:
  88         BadConversion(const string& s) : runtime_error(s){ }
  89 };
  90
  91 //**********************************************************************************************************************
  92
  93 template<typename T>
  94 inline void convert(const string& s, T& x, bool failIfLeftoverChars = true){
  95         istringstream i(s);
  96         char c;
  97         if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
  98                 throw BadConversion(s);
  99 }
 100
 101 //**********************************************************************************************************************
 102
 103 template<typename T>
 104 inline bool convertTestFloat(const string& s, T& x, bool failIfLeftoverChars = true){
 105         istringstream i(s);
 106         char c;
 107         if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
 108         {
 109                 return false;
 110         }
 111         return true;
 112 }
 113
 114 //**********************************************************************************************************************
 115
 116 template<typename T>
 117 inline bool convertTest(const string& s, T& x, bool failIfLeftoverChars = true){
 118         istringstream i(s);
 119         char c;
 120         if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
 121         {
 122                 cout << "unable to be converted into an integer.\n" << endl;
 123                 return false;
 124         }
 125         return true;
 126 }
 127
 128 //**********************************************************************************************************************
 129
 130 template<typename T>
 131 string toString(const T&x){
 132     stringstream output;
 133     output << x;
 134     return output.str();
 135 }
 136
 137 //**********************************************************************************************************************
 138
 139 template<typename T>
 140 string toHex(const T&x){
 141         stringstream output;
 142
 143         output << hex << x;
 144
 145     return output.str();
 146 }
 147 //**********************************************************************************************************************
 148
 149 template<typename T>
 150 string toString(const T&x, int i){
 151         stringstream output;
 152
 153         output.precision(i);
 154     output << fixed << x;
 155
 156     return output.str();
 157 }
 158 /***********************************************************************/
 159
 160 inline int openOutputFileAppend(string fileName, ofstream& fileHandle){
 161
 162         fileHandle.open(fileName.c_str(), ios::app);
 163         if(!fileHandle) {
 164                 cout << "Error: Could not open " <<  fileName << endl;
 165                 return 1;
 166         }
 167         else {
 168                 return 0;
 169         }
 170
 171 }
 172 /***********************************************************************/
 173
 174 inline void gobble(istream& f){
 175
 176         char d;
 177     while(isspace(d=f.get()))           {;}
 178         f.putback(d);
 179
 180 }
 181
 182 /***********************************************************************/
 183
 184 inline string getline(ifstream& fileHandle) {
 185         try {
 186
 187                 string line = "";
 188
 189                 while (!fileHandle.eof())       {
 190                         //get next character
 191                         char c = fileHandle.get();
 192
 193                         //are you at the end of the line
 194                         if ((c == '\n') || (c == '\r') || (c == '\f')){  break; }
 195                         else {          line += c;              }
 196                 }
 197
 198                 return line;
 199
 200         }
 201         catch(exception& e) {
 202                 cout << "Error in mothur function getline" << endl;
 203                 exit(1);
 204         }
 205 }
 206
 207 /**************************************************************************************************/
 208
 209 inline void mothurOut(string message) {
 210         try{
 211                 ofstream out;
 212                 string logFileName = "mothur.logFile";
 213                 openOutputFileAppend(logFileName, out);
 214
 215                 cout << message;
 216                 out << message;
 217
 218                 out.close();
 219         }
 220         catch(exception& e) {
 221                 cout << "Error in mothur class mothurOut" << endl;
 222                 exit(1);
 223         }
 224 }
 225 /**************************************************************************************************/
 226
 227 inline void mothurOut(string message, string precision) {
 228         try{
 229                 ofstream out;
 230                 string logFileName = "mothur.logFile";
 231                 openOutputFileAppend(logFileName, out);
 232
 233                 cout << precision << message;
 234                 out << precision << message;
 235
 236                 out.close();
 237         }
 238         catch(exception& e) {
 239                 cout << "Error in mothur class mothurOut" << endl;
 240                 exit(1);
 241         }
 242 }
 243
 244 /**************************************************************************************************/
 245
 246 inline void mothurOutEndLine() {
 247         try {
 248                 ofstream out;
 249                 string logFileName = "mothur.logFile";
 250                 openOutputFileAppend(logFileName, out);
 251
 252                 cout << endl;
 253                 out << endl;
 254
 255                 out.close();
 256         }
 257         catch(exception& e) {
 258                 cout << "error in mothur mothurOutEndLine" << endl;
 259                 exit(1);
 260         }
 261 }
 262
 263
 264 /**************************************************************************************************/
 265
 266 inline void errorOut(exception& e, string object, string function) {
 267
 268                 mothurOut("Error: ");
 269                 mothurOut(toString(e.what()));
 270                 mothurOut(" has occurred in the " + object + " class function " + function + ". Please contact Pat Schloss at pschloss@umich.edu, and be sure to include the mothur.logFile with your inquiry.");
 271                 mothurOutEndLine();
 272
 273 }
 274
 275
 276
 277
 278 /***********************************************************************/
 279
 280 inline bool isTrue(string f){
 281
 282         if ((f == "TRUE") || (f == "T") || (f == "true") || (f == "t")) {       return true;    }
 283         else {  return false;  }
 284 }
 285
 286 /***********************************************************************/
 287
 288 inline float roundDist(float dist, int precision){
 289
 290         return int(dist * precision + 0.5)/float(precision);
 291
 292 }
 293
 294 /***********************************************************************/
 295
 296 inline int getNumNames(string names){
 297
 298         int count = 0;
 299
 300         if(names != ""){
 301                 count = 1;
 302                 for(int i=0;i<names.size();i++){
 303                         if(names[i] == ','){
 304                                 count++;
 305                         }
 306                 }
 307         }
 308
 309         return count;
 310
 311 }
 312
 313 /**************************************************************************************************/
 314
 315 inline vector<vector<double> > binomial(int maxOrder){
 316
 317         vector<vector<double> > binomial(maxOrder+1);
 318
 319     for(int i=0;i<=maxOrder;i++){
 320                 binomial[i].resize(maxOrder+1);
 321                 binomial[i][0]=1;
 322                 binomial[0][i]=0;
 323     }
 324     binomial[0][0]=1;
 325
 326     binomial[1][0]=1;
 327     binomial[1][1]=1;
 328
 329     for(int i=2;i<=maxOrder;i++){
 330                 binomial[1][i]=0;
 331     }
 332
 333     for(int i=2;i<=maxOrder;i++){
 334                 for(int j=1;j<=maxOrder;j++){
 335                         if(i==j){       binomial[i][j]=1;                                                                       }
 336                         if(j>i) {       binomial[i][j]=0;                                                                       }
 337                         else    {       binomial[i][j]=binomial[i-1][j-1]+binomial[i-1][j];     }
 338                 }
 339     }
 340
 341         return binomial;
 342 }
 343
 344 /***********************************************************************/
 345
 346 inline string getRootName(string longName){
 347
 348         string rootName = longName;
 349
 350         if(longName.find_last_of(".") != longName.npos){
 351                 int pos = longName.find_last_of('.')+1;
 352                 rootName = longName.substr(0, pos);
 353         }
 354
 355         return rootName;
 356 }
 357 /***********************************************************************/
 358
 359 inline string getSimpleName(string longName){
 360
 361         string simpleName = longName;
 362
 363         if(longName.find_last_of("/") != longName.npos){
 364                 int pos = longName.find_last_of('/')+1;
 365                 simpleName = longName.substr(pos, longName.length());
 366         }
 367
 368         return simpleName;
 369 }
 370
 371 /***********************************************************************/
 372
 373 inline int factorial(int num){
 374         int total = 1;
 375
 376         for (int i = 1; i <= num; i++) {
 377                 total *= i;
 378         }
 379
 380         return total;
 381 }
 382 /**************************************************************************************************
 383
 384 double min(double x, double y)
 385 {
 386     if(x<y){    return x;    }
 387     else   {    return y;    }
 388 }
 389
 390 /***********************************************************************/
 391
 392 inline string getPathName(string longName){
 393
 394         string rootPathName = longName;
 395
 396         if(longName.find_last_of('/') != longName.npos){
 397                 int pos = longName.find_last_of('/')+1;
 398                 rootPathName = longName.substr(0, pos);
 399         }
 400
 401         return rootPathName;
 402 }
 403
 404 /***********************************************************************/
 405
 406 inline string getExtension(string longName){
 407
 408         string extension = longName;
 409
 410         if(longName.find_last_of('.') != longName.npos){
 411                 int pos = longName.find_last_of('.');
 412                 extension = longName.substr(pos, longName.length());
 413         }
 414
 415         return extension;
 416 }
 417
 418 /***********************************************************************/
 419
 420 inline int openInputFile(string fileName, ifstream& fileHandle){
 421
 422         fileHandle.open(fileName.c_str());
 423         if(!fileHandle) {
 424                 mothurOut("Error: Could not open " + fileName);  mothurOutEndLine();
 425                 return 1;
 426         }
 427         else {
 428                 return 0;
 429         }
 430
 431 }
 432
 433 /***********************************************************************/
 434
 435 inline int openOutputFile(string fileName, ofstream& fileHandle){
 436
 437         fileHandle.open(fileName.c_str(), ios::trunc);
 438         if(!fileHandle) {
 439                 mothurOut("Error: Could not open " + fileName);  mothurOutEndLine();
 440                 return 1;
 441         }
 442         else {
 443                 return 0;
 444         }
 445
 446 }
 447
 448 /***********************************************************************/
 449
 450 inline int getNumSeqs(ifstream& file){
 451
 452         int numSeqs = count(istreambuf_iterator<char>(file),istreambuf_iterator<char>(), '>');
 453         file.seekg(0);
 454         return numSeqs;
 455
 456 }
 457
 458 /***********************************************************************/
 459
 460 //This function parses the estimator options and puts them in a vector
 461 inline void splitAtDash(string& estim, vector<string>& container) {
 462         try {
 463                 string individual;
 464
 465                 while (estim.find_first_of('-') != -1) {
 466                         individual = estim.substr(0,estim.find_first_of('-'));
 467                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 468                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 469                                 container.push_back(individual);
 470                         }
 471                 }
 472                 //get last one
 473                 container.push_back(estim);
 474         }
 475         catch(exception& e) {
 476                 errorOut(e, "mothur", "splitAtDash");
 477                 exit(1);
 478         }
 479 }
 480
 481 /***********************************************************************/
 482 //This function parses the label options and puts them in a set
 483 inline void splitAtDash(string& estim, set<string>& container) {
 484         try {
 485                 string individual;
 486
 487                 while (estim.find_first_of('-') != -1) {
 488                         individual = estim.substr(0,estim.find_first_of('-'));
 489                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 490                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 491                                 container.insert(individual);
 492                         }
 493                 }
 494                 //get last one
 495                 container.insert(estim);
 496         }
 497         catch(exception& e) {
 498                 errorOut(e, "mothur", "splitAtDash");
 499                 exit(1);
 500         }
 501 }
 502 /***********************************************************************/
 503 //This function parses the line options and puts them in a set
 504 inline void splitAtDash(string& estim, set<int>& container) {
 505         try {
 506                 string individual;
 507                 int lineNum;
 508
 509                 while (estim.find_first_of('-') != -1) {
 510                         individual = estim.substr(0,estim.find_first_of('-'));
 511                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
 512                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
 513                                 convert(individual, lineNum); //convert the string to int
 514                                 container.insert(lineNum);
 515                         }
 516                 }
 517                 //get last one
 518                 convert(estim, lineNum); //convert the string to int
 519                 container.insert(lineNum);
 520         }
 521         catch(exception& e) {
 522                 errorOut(e, "mothur", "splitAtDash");
 523                 exit(1);
 524         }
 525 }
 526 /***********************************************************************/
 527 //This function parses the a string and puts peices in a vector
 528 inline void splitAtComma(string& estim, vector<string>& container) {
 529         try {
 530                 string individual;
 531
 532                 while (estim.find_first_of(',') != -1) {
 533                         individual = estim.substr(0,estim.find_first_of(','));
 534                         if ((estim.find_first_of(',')+1) <= estim.length()) { //checks to make sure you don't have comma at end of string
 535                                 estim = estim.substr(estim.find_first_of(',')+1, estim.length());
 536                                 container.push_back(individual);
 537                         }
 538                 }
 539                 //get last one
 540                 container.push_back(estim);
 541         }
 542         catch(exception& e) {
 543                 errorOut(e, "mothur", "splitAtComma");
 544                 exit(1);
 545         }
 546 }
 547 /***********************************************************************/
 548
 549 //This function splits up the various option parameters
 550 inline void splitAtComma(string& prefix, string& suffix){
 551         try {
 552                 prefix = suffix.substr(0,suffix.find_first_of(','));
 553                 if ((suffix.find_first_of(',')+2) <= suffix.length()) {  //checks to make sure you don't have comma at end of string
 554                         suffix = suffix.substr(suffix.find_first_of(',')+1, suffix.length());
 555                         string space = " ";
 556                         while(suffix.at(0) == ' ')
 557                                 suffix = suffix.substr(1, suffix.length());
 558                 }
 559
 560         }
 561         catch(exception& e) {
 562                 errorOut(e, "mothur", "splitAtComma");
 563                 exit(1);
 564         }
 565 }
 566 /***********************************************************************/
 567
 568 //This function separates the key value from the option value i.e. dist=96_...
 569 inline void splitAtEquals(string& key, string& value){
 570         try {
 571                 if(value.find_first_of('=') != -1){
 572                         key = value.substr(0,value.find_first_of('='));
 573                         if ((value.find_first_of('=')+1) <= value.length()) {
 574                                 value = value.substr(value.find_first_of('=')+1, value.length());
 575                         }
 576                 }else{
 577                         key = value;
 578                         value = 1;
 579                 }
 580         }
 581         catch(exception& e) {
 582                 errorOut(e, "mothur", "splitAtEquals");
 583                 exit(1);
 584         }
 585 }
 586 /**************************************************************************************************/
 587
 588 inline bool inUsersGroups(string groupname, vector<string> Groups) {
 589         try {
 590                 for (int i = 0; i < Groups.size(); i++) {
 591                         if (groupname == Groups[i]) { return true; }
 592                 }
 593                 return false;
 594         }
 595         catch(exception& e) {
 596                 errorOut(e, "mothur", "inUsersGroups");
 597                 exit(1);
 598         }
 599 }
 600
 601 /**************************************************************************************************/
 602
 603 inline void mothurOutJustToLog(string message) {
 604         try {
 605                 ofstream out;
 606                 string logFileName = "mothur.logFile";
 607                 openOutputFileAppend(logFileName, out);
 608
 609                 out << message;
 610
 611                 out.close();
 612         }
 613         catch(exception& e) {
 614                 errorOut(e, "mothur", "mothurOutJustToLog");
 615                 exit(1);
 616         }
 617 }
 618
 619
 620 /**************************************************************************************************/
 621
 622 inline void mothurOut(float num) {
 623         try {
 624                 ofstream out;
 625                 string logFileName = "mothur.logFile";
 626                 openOutputFileAppend(logFileName, out);
 627
 628                 cout << num;
 629                 out << num;
 630
 631                 out.close();
 632         }
 633         catch(exception& e) {
 634                 cout << "Error in mothur class mothurOut float" << endl;
 635                 exit(1);
 636         }
 637 }
 638 /***********************************************************************/
 639 inline void mothurOut(double value) {
 640         try {
 641                 ofstream out;
 642                 string logFileName = "mothur.logFile";
 643                 openOutputFileAppend(logFileName, out);
 644
 645                 cout << value;
 646                 out << value;
 647
 648                 out.close();
 649         }
 650         catch(exception& e) {
 651                 cout << "Error in mothur class mothurOut double" << endl;
 652                 exit(1);
 653         }
 654 }
 655
 656 /***********************************************************************/
 657 //this function determines if the user has given us labels that are smaller than the given label.
 658 //if so then it returns true so that the calling function can run the previous valid distance.
 659 //it's a "smart" distance function.  It also checks for invalid labels.
 660 inline bool anyLabelsToProcess(string label, set<string>& userLabels, string errorOff) {
 661         try {
 662                 set<string>::iterator it;
 663                 vector<float> orderFloat;
 664                 map<string, float> userMap;  //the conversion process removes trailing 0's which we need to put back
 665                 map<string, float>::iterator it2;
 666                 float labelFloat;
 667                 bool smaller = false;
 668
 669                 //unique is the smallest line
 670                 if (label == "unique") {  return false;  }
 671                 else { convert(label, labelFloat); }
 672
 673                 //go through users set and make them floats
 674                 for(it = userLabels.begin(); it != userLabels.end(); ++it) {
 675
 676                         float temp;
 677                         if ((*it != "unique") && (convertTestFloat(*it, temp) == true)){
 678                                 convert(*it, temp);
 679                                 orderFloat.push_back(temp);
 680                                 userMap[*it] = temp;
 681                         }else if (*it == "unique") {
 682                                 orderFloat.push_back(-1.0);
 683                                 userMap["unique"] = -1.0;
 684                         }else {
 685                                 if (errorOff == "") {  mothurOut(*it + " is not a valid label."); mothurOutEndLine();  }
 686                                 userLabels.erase(*it);
 687                                 it--;
 688                         }
 689                 }
 690
 691                 //sort order
 692                 sort(orderFloat.begin(), orderFloat.end());
 693
 694                 /*************************************************/
 695                 //is this label bigger than any of the users labels
 696                 /*************************************************/
 697
 698                 //loop through order until you find a label greater than label
 699                 for (int i = 0; i < orderFloat.size(); i++) {
 700                         if (orderFloat[i] < labelFloat) {
 701                                 smaller = true;
 702                                 if (orderFloat[i] == -1) {
 703                                         if (errorOff == "") { mothurOut("Your file does not include the label unique."); mothurOutEndLine(); }
 704                                         userLabels.erase("unique");
 705                                 }
 706                                 else {
 707                                         if (errorOff == "") { mothurOut("Your file does not include the label "); mothurOutEndLine(); }
 708                                         string s = "";
 709                                         for (it2 = userMap.begin(); it2!= userMap.end(); it2++) {
 710                                                 if (it2->second == orderFloat[i]) {
 711                                                         s = it2->first;
 712                                                         //remove small labels
 713                                                         userLabels.erase(s);
 714                                                         break;
 715                                                 }
 716                                         }
 717                                         if (errorOff == "") { mothurOut(s + ". I will use the next smallest distance. "); mothurOutEndLine(); }
 718                                 }
 719                         //since they are sorted once you find a bigger one stop looking
 720                         }else { break; }
 721                 }
 722
 723                 return smaller;
 724
 725         }
 726         catch(exception& e) {
 727                 errorOut(e, "mothur", "anyLabelsToProcess");
 728                 exit(1);
 729         }
 730 }
 731
 732 /**************************************************************************************************/
 733 #endif
 734