mothurout.cpp

   1 /*
   2  *  mothurOut.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 2/25/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "mothurout.h"
  11
  12 /******************************************************/
  13 MothurOut* MothurOut::getInstance() {
  14         if( _uniqueInstance == 0) {
  15                 _uniqueInstance = new MothurOut();
  16         }
  17         return _uniqueInstance;
  18 }
  19 /*********************************************************************************************/
  20 void MothurOut::setFileName(string filename)  {
  21         try {
  22                 logFileName = filename;
  23
  24                 #ifdef USE_MPI
  25                         int pid;
  26                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
  27
  28                         if (pid == 0) { //only one process should output to screen
  29                 #endif
  30
  31                 openOutputFile(filename, out);
  32
  33                 #ifdef USE_MPI
  34                         }
  35                 #endif
  36         }
  37         catch(exception& e) {
  38                 errorOut(e, "MothurOut", "setFileName");
  39                 exit(1);
  40         }
  41 }
  42 /*********************************************************************************************/
  43 void MothurOut::setDefaultPath(string pathname)  {
  44         try {
  45
  46                 //add / to name if needed
  47                 string lastChar = pathname.substr(pathname.length()-1);
  48                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
  49                         if (lastChar != "/") { pathname += "/"; }
  50                 #else
  51                         if (lastChar != "\\") { pathname += "\\"; }
  52                 #endif
  53
  54                 defaultPath = pathname;
  55
  56         }
  57         catch(exception& e) {
  58                 errorOut(e, "MothurOut", "setDefaultPath");
  59                 exit(1);
  60         }
  61 }
  62 /*********************************************************************************************/
  63 void MothurOut::closeLog()  {
  64         try {
  65
  66                 #ifdef USE_MPI
  67                         int pid;
  68                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
  69
  70                         if (pid == 0) { //only one process should output to screen
  71                 #endif
  72
  73                 out.close();
  74
  75                 #ifdef USE_MPI
  76                         }
  77                 #endif
  78         }
  79         catch(exception& e) {
  80                 errorOut(e, "MothurOut", "closeLog");
  81                 exit(1);
  82         }
  83 }
  84
  85 /*********************************************************************************************/
  86 MothurOut::~MothurOut() {
  87         try {
  88                 _uniqueInstance = 0;
  89
  90         }
  91         catch(exception& e) {
  92                 errorOut(e, "MothurOut", "MothurOut");
  93                 exit(1);
  94         }
  95 }
  96 /*********************************************************************************************/
  97 void MothurOut::mothurOut(string output) {
  98         try {
  99
 100                 #ifdef USE_MPI
 101                         int pid;
 102                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 103
 104                         if (pid == 0) { //only one process should output to screen
 105                 #endif
 106
 107                 cout << output;
 108                 out << output;
 109
 110                 #ifdef USE_MPI
 111                         }
 112                 #endif
 113         }
 114         catch(exception& e) {
 115                 errorOut(e, "MothurOut", "MothurOut");
 116                 exit(1);
 117         }
 118 }
 119 /*********************************************************************************************/
 120 void MothurOut::mothurOutEndLine() {
 121         try {
 122                 #ifdef USE_MPI
 123                         int pid;
 124                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 125
 126                         if (pid == 0) { //only one process should output to screen
 127                 #endif
 128
 129                 cout << endl;
 130                 out << endl;
 131
 132                 #ifdef USE_MPI
 133                         }
 134                 #endif
 135         }
 136         catch(exception& e) {
 137                 errorOut(e, "MothurOut", "MothurOutEndLine");
 138                 exit(1);
 139         }
 140 }
 141 /*********************************************************************************************/
 142 void MothurOut::mothurOutJustToLog(string output) {
 143         try {
 144                 #ifdef USE_MPI
 145                         int pid;
 146                         MPI_Comm_rank(MPI_COMM_WORLD, &pid);
 147
 148                         if (pid == 0) { //only one process should output to screen
 149                 #endif
 150
 151                 out << output;
 152
 153                 #ifdef USE_MPI
 154                         }
 155                 #endif
 156         }
 157         catch(exception& e) {
 158                 errorOut(e, "MothurOut", "MothurOutJustToLog");
 159                 exit(1);
 160         }
 161 }
 162 /*********************************************************************************************/
 163 void MothurOut::errorOut(exception& e, string object, string function) {
 164         //double vm, rss;
 165         //mem_usage(vm, rss);
 166
 167         mothurOut("[ERROR]: ");
 168         mothurOut(toString(e.what()));
 169         mothurOut(" has occurred in the " + object + " class function " + function + ". Please contact Pat Schloss at mothur.bugs@gmail.com, and be sure to include the mothur.logFile with your inquiry.");
 170         mothurOutEndLine();
 171 }
 172 /*********************************************************************************************/
 173 //The following was originally from http://stackoverflow.com/questions/669438/how-to-get-memory-usage-at-run-time-in-c
 174 // process_mem_usage(double &, double &) - takes two doubles by reference,
 175 // attempts to read the system-dependent data for a process' virtual memory
 176 // size and resident set size, and return the results in KB.
 177 //
 178 // On failure, returns 0.0, 0.0
 179 int MothurOut::mem_usage(double& vm_usage, double& resident_set) {
 180   #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 181
 182            vm_usage     = 0.0;
 183            resident_set = 0.0;
 184
 185            // 'file' stat seems to give the most reliable results
 186            //
 187            ifstream stat_stream("/proc/self/stat",ios_base::in);
 188
 189            // dummy vars for leading entries in stat that we don't care about
 190            //
 191            string pid, comm, state, ppid, pgrp, session, tty_nr;
 192            string tpgid, flags, minflt, cminflt, majflt, cmajflt;
 193            string utime, stime, cutime, cstime, priority, nice;
 194            string O, itrealvalue, starttime;
 195
 196            // the two fields we want
 197            //
 198            unsigned long vsize;
 199            long rss;
 200
 201            stat_stream >> pid >> comm >> state >> ppid >> pgrp >> session >> tty_nr
 202                                    >> tpgid >> flags >> minflt >> cminflt >> majflt >> cmajflt
 203                                    >> utime >> stime >> cutime >> cstime >> priority >> nice
 204                                    >> O >> itrealvalue >> starttime >> vsize >> rss; // don't care about the rest
 205
 206            long page_size_kb = sysconf(_SC_PAGE_SIZE) / 1024; // in case x86-64 is configured to use 2MB pages
 207            vm_usage     = vsize / 1024.0;
 208            resident_set = rss * page_size_kb;
 209
 210            mothurOut("Memory Usage: vm = " + toString(vm_usage) + " rss = " + toString(resident_set) + "\n");
 211                 return 0;
 212
 213         #else
 214 /*              //windows memory usage
 215                 // Get the list of process identifiers.
 216                 DWORD aProcesses[1024], cbNeeded, cProcesses;
 217
 218                 if ( !EnumProcesses( aProcesses, sizeof(aProcesses), &cbNeeded ) ){ return 1; }
 219
 220                 // Calculate how many process identifiers were returned.
 221                 cProcesses = cbNeeded / sizeof(DWORD);
 222
 223                 // Print the memory usage for each process
 224                 for (int i = 0; i < cProcesses; i++ ) {
 225                         DWORD processID = aProcesses[i];
 226
 227                         PROCESS_MEMORY_COUNTERS pmc;
 228
 229                         HANDLE hProcess = OpenProcess((PROCESS_QUERY_INFORMATION | PROCESS_VM_READ), FALSE, processID);
 230
 231                         // Print the process identifier.
 232                         printf( "\nProcess ID: %u\n", processID);
 233
 234                         if (NULL != hProcess) {
 235
 236                                 if ( GetProcessMemoryInfo( hProcess, &pmc, sizeof(pmc)) ) {
 237                                         printf( "\tPageFaultCount: 0x%08X\n", pmc.PageFaultCount );
 238                                         printf( "\tPeakWorkingSetSize: 0x%08X\n", pmc.PeakWorkingSetSize );
 239                                         printf( "\tWorkingSetSize: 0x%08X\n", pmc.WorkingSetSize );
 240                                         printf( "\tQuotaPeakPagedPoolUsage: 0x%08X\n", pmc.QuotaPeakPagedPoolUsage );
 241                                         printf( "\tQuotaPagedPoolUsage: 0x%08X\n", pmc.QuotaPagedPoolUsage );
 242                                         printf( "\tQuotaPeakNonPagedPoolUsage: 0x%08X\n", pmc.QuotaPeakNonPagedPoolUsage );
 243                                         printf( "\tQuotaNonPagedPoolUsage: 0x%08X\n", pmc.QuotaNonPagedPoolUsage );
 244                                         printf( "\tPagefileUsage: 0x%08X\n", pmc.PagefileUsage );
 245                                         printf( "\tPeakPagefileUsage: 0x%08X\n", pmc.PeakPagefileUsage );
 246                                 }
 247                                 CloseHandle(hProcess);
 248                         }
 249                 }
 250 */
 251                         return 0;
 252
 253         #endif
 254 }
 255
 256
 257 /***********************************************************************/
 258 int MothurOut::openOutputFileAppend(string fileName, ofstream& fileHandle){
 259         try {
 260                 fileName = getFullPathName(fileName);
 261
 262                 fileHandle.open(fileName.c_str(), ios::app);
 263                 if(!fileHandle) {
 264                         mothurOut("[ERROR]: Could not open " + fileName); mothurOutEndLine();
 265                         return 1;
 266                 }
 267                 else {
 268                         return 0;
 269                 }
 270         }
 271         catch(exception& e) {
 272                 errorOut(e, "MothurOut", "openOutputFileAppend");
 273                 exit(1);
 274         }
 275 }
 276 /***********************************************************************/
 277 void MothurOut::gobble(istream& f){
 278         try {
 279                 char d;
 280                 while(isspace(d=f.get()))               {;}
 281                 f.putback(d);
 282         }
 283         catch(exception& e) {
 284                 errorOut(e, "MothurOut", "gobble");
 285                 exit(1);
 286         }
 287 }
 288 /***********************************************************************/
 289 void MothurOut::gobble(istringstream& f){
 290         try {
 291                 char d;
 292                 while(isspace(d=f.get()))               {;}
 293                 f.putback(d);
 294         }
 295         catch(exception& e) {
 296                 errorOut(e, "MothurOut", "gobble");
 297                 exit(1);
 298         }
 299 }
 300
 301 /***********************************************************************/
 302
 303 string MothurOut::getline(istringstream& fileHandle) {
 304         try {
 305
 306                 string line = "";
 307
 308                 while (!fileHandle.eof())       {
 309                         //get next character
 310                         char c = fileHandle.get();
 311
 312                         //are you at the end of the line
 313                         if ((c == '\n') || (c == '\r') || (c == '\f')){  break; }
 314                         else {          line += c;              }
 315                 }
 316
 317                 return line;
 318
 319         }
 320         catch(exception& e) {
 321                 errorOut(e, "MothurOut", "getline");
 322                 exit(1);
 323         }
 324 }
 325 /***********************************************************************/
 326
 327 string MothurOut::getline(ifstream& fileHandle) {
 328         try {
 329
 330                 string line = "";
 331
 332                 while (!fileHandle.eof())       {
 333                         //get next character
 334                         char c = fileHandle.get();
 335
 336                         //are you at the end of the line
 337                         if ((c == '\n') || (c == '\r') || (c == '\f')){  break; }
 338                         else {          line += c;              }
 339                 }
 340
 341                 return line;
 342
 343         }
 344         catch(exception& e) {
 345                 errorOut(e, "MothurOut", "getline");
 346                 exit(1);
 347         }
 348 }
 349 /***********************************************************************/
 350
 351 string MothurOut::getRootName(string longName){
 352         try {
 353
 354                 string rootName = longName;
 355
 356                 if(longName.find_last_of(".") != longName.npos){
 357                         int pos = longName.find_last_of('.')+1;
 358                         rootName = longName.substr(0, pos);
 359                 }
 360
 361                 return rootName;
 362         }
 363         catch(exception& e) {
 364                 errorOut(e, "MothurOut", "getRootName");
 365                 exit(1);
 366         }
 367 }
 368 /***********************************************************************/
 369
 370 string MothurOut::getSimpleName(string longName){
 371         try {
 372                 string simpleName = longName;
 373
 374                 size_t found;
 375                 found=longName.find_last_of("/\\");
 376
 377                 if(found != longName.npos){
 378                         simpleName = longName.substr(found+1);
 379                 }
 380
 381                 return simpleName;
 382         }
 383         catch(exception& e) {
 384                 errorOut(e, "MothurOut", "getSimpleName");
 385                 exit(1);
 386         }
 387 }
 388
 389 /***********************************************************************/
 390
 391 string MothurOut::getPathName(string longName){
 392         try {
 393                 string rootPathName = longName;
 394
 395                 if(longName.find_last_of("/\\") != longName.npos){
 396                         int pos = longName.find_last_of("/\\")+1;
 397                         rootPathName = longName.substr(0, pos);
 398                 }
 399
 400                 return rootPathName;
 401         }
 402         catch(exception& e) {
 403                 errorOut(e, "MothurOut", "getPathName");
 404                 exit(1);
 405         }
 406
 407 }
 408 /***********************************************************************/
 409
 410 string MothurOut::hasPath(string longName){
 411         try {
 412                 string path = "";
 413
 414                 size_t found;
 415                 found=longName.find_last_of("~/\\");
 416
 417                 if(found != longName.npos){
 418                         path = longName.substr(0, found+1);
 419                 }
 420
 421                 return path;
 422         }
 423         catch(exception& e) {
 424                 errorOut(e, "MothurOut", "hasPath");
 425                 exit(1);
 426         }
 427 }
 428
 429 /***********************************************************************/
 430
 431 string MothurOut::getExtension(string longName){
 432         try {
 433                 string extension = longName;
 434
 435                 if(longName.find_last_of('.') != longName.npos){
 436                         int pos = longName.find_last_of('.');
 437                         extension = longName.substr(pos, longName.length());
 438                 }
 439
 440                 return extension;
 441         }
 442         catch(exception& e) {
 443                 errorOut(e, "MothurOut", "getExtension");
 444                 exit(1);
 445         }
 446 }
 447 /***********************************************************************/
 448 bool MothurOut::isBlank(string fileName){
 449         try {
 450
 451                 fileName = getFullPathName(fileName);
 452
 453                 ifstream fileHandle;
 454                 fileHandle.open(fileName.c_str());
 455                 if(!fileHandle) {
 456                         mothurOut("[ERROR]: Could not open " + fileName); mothurOutEndLine();
 457                         return false;
 458                 }else {
 459                         //check for blank file
 460                         gobble(fileHandle);
 461                         if (fileHandle.eof()) { fileHandle.close(); return true;  }
 462                 }
 463                 return false;
 464         }
 465         catch(exception& e) {
 466                 errorOut(e, "MothurOut", "isBlank");
 467                 exit(1);
 468         }
 469 }
 470 /***********************************************************************/
 471
 472 string MothurOut::getFullPathName(string fileName){
 473         try{
 474
 475         string path = hasPath(fileName);
 476         string newFileName;
 477         int pos;
 478
 479         if (path == "") { return fileName; } //its a simple name
 480         else { //we need to complete the pathname
 481                 // ex. ../../../filename
 482                 // cwd = /user/work/desktop
 483
 484                 string cwd;
 485                 //get current working directory
 486                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 487
 488                         if (path.find("~") != -1) { //go to home directory
 489                                 string homeDir = getenv ("HOME");
 490                                 newFileName = homeDir + fileName.substr(fileName.find("~")+1);
 491                                 return newFileName;
 492                         }else { //find path
 493                                 if (path.rfind("./") == -1) { return fileName; } //already complete name
 494                                 else { newFileName = fileName.substr(fileName.rfind("./")+2); } //save the complete part of the name
 495
 496                                 //char* cwdpath = new char[1024];
 497
 498                                 //size_t size;
 499                                 //cwdpath=getcwd(cwdpath,size);
 500                                 //cwd = cwdpath;
 501
 502                                 char *cwdpath = NULL;
 503                                 cwdpath = getcwd(NULL, 0); // or _getcwd
 504                                 if ( cwdpath != NULL) { cwd = cwdpath; }
 505                                 else { cwd = "";  }
 506
 507                                 //rip off first '/'
 508                                 string simpleCWD;
 509                                 if (cwd.length() > 0) { simpleCWD = cwd.substr(1); }
 510
 511                                 //break apart the current working directory
 512                                 vector<string> dirs;
 513                                 while (simpleCWD.find_first_of('/') != -1) {
 514                                         string dir = simpleCWD.substr(0,simpleCWD.find_first_of('/'));
 515                                         simpleCWD = simpleCWD.substr(simpleCWD.find_first_of('/')+1, simpleCWD.length());
 516                                         dirs.push_back(dir);
 517                                 }
 518                                 //get last one              // ex. ../../../filename = /user/work/desktop/filename
 519                                 dirs.push_back(simpleCWD);  //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
 520
 521
 522                                 int index = dirs.size()-1;
 523
 524                                 while((pos = path.rfind("./")) != -1) { //while you don't have a complete path
 525                                         if (pos == 0) { break;  //you are at the end
 526                                         }else if (path[(pos-1)] == '.') { //you want your parent directory ../
 527                                                 path = path.substr(0, pos-1);
 528                                                 index--;
 529                                                 if (index == 0) {  break; }
 530                                         }else if (path[(pos-1)] == '/') { //you want the current working dir ./
 531                                                 path = path.substr(0, pos);
 532                                         }else if (pos == 1) { break;  //you are at the end
 533                                         }else { cout << "cannot resolve path for " <<  fileName << endl; return fileName; }
 534                                 }
 535
 536                                 for (int i = index; i >= 0; i--) {
 537                                         newFileName = dirs[i] +  "/" + newFileName;
 538                                 }
 539
 540                                 newFileName =  "/" +  newFileName;
 541                                 return newFileName;
 542
 543                         }
 544                 #else
 545                         if (path.find("~") != -1) { //go to home directory
 546                                 string homeDir = getenv ("HOMEPATH");
 547                                 newFileName = homeDir + fileName.substr(fileName.find("~")+1);
 548                                 return newFileName;
 549                         }else { //find path
 550                                 if (path.rfind(".\\") == -1) { return fileName; } //already complete name
 551                                 else { newFileName = fileName.substr(fileName.rfind(".\\")+2); } //save the complete part of the name
 552
 553                                 char *cwdpath = NULL;
 554                                 cwdpath = getcwd(NULL, 0); // or _getcwd
 555                                 if ( cwdpath != NULL) { cwd = cwdpath; }
 556                                 else { cwd = "";  }
 557
 558                                 //break apart the current working directory
 559                                 vector<string> dirs;
 560                                 while (cwd.find_first_of('\\') != -1) {
 561                                         string dir = cwd.substr(0,cwd.find_first_of('\\'));
 562                                         cwd = cwd.substr(cwd.find_first_of('\\')+1, cwd.length());
 563                                         dirs.push_back(dir);
 564
 565                                 }
 566                                 //get last one
 567                                 dirs.push_back(cwd);  //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
 568
 569                                 int index = dirs.size()-1;
 570
 571                                 while((pos = path.rfind(".\\")) != -1) { //while you don't have a complete path
 572                                         if (pos == 0) { break;  //you are at the end
 573                                         }else if (path[(pos-1)] == '.') { //you want your parent directory ../
 574                                                 path = path.substr(0, pos-1);
 575                                                 index--;
 576                                                 if (index == 0) {  break; }
 577                                         }else if (path[(pos-1)] == '\\') { //you want the current working dir ./
 578                                                 path = path.substr(0, pos);
 579                                         }else if (pos == 1) { break;  //you are at the end
 580                                         }else { cout << "cannot resolve path for " <<  fileName << endl; return fileName; }
 581                                 }
 582
 583                                 for (int i = index; i >= 0; i--) {
 584                                         newFileName = dirs[i] +  "\\" + newFileName;
 585                                 }
 586
 587                                 return newFileName;
 588                         }
 589
 590                 #endif
 591         }
 592         }
 593         catch(exception& e) {
 594                 errorOut(e, "MothurOut", "getFullPathName");
 595                 exit(1);
 596         }
 597 }
 598 /***********************************************************************/
 599 //no error open
 600 int MothurOut::openInputFile(string fileName, ifstream& fileHandle, string m){
 601         try {
 602                         //get full path name
 603                         string completeFileName = getFullPathName(fileName);
 604
 605                         fileHandle.open(completeFileName.c_str());
 606                         if(!fileHandle) {
 607                                 return 1;
 608                         }else {
 609                                 //check for blank file
 610                                 gobble(fileHandle);
 611                                 return 0;
 612                         }
 613         }
 614         catch(exception& e) {
 615                 errorOut(e, "MothurOut", "openInputFile - no Error");
 616                 exit(1);
 617         }
 618 }
 619 /***********************************************************************/
 620
 621 int MothurOut::openInputFile(string fileName, ifstream& fileHandle){
 622         try {
 623                 //get full path name
 624                 string completeFileName = getFullPathName(fileName);
 625
 626                 fileHandle.open(completeFileName.c_str());
 627                 if(!fileHandle) {
 628                         mothurOut("[ERROR]: Could not open " + completeFileName); mothurOutEndLine();
 629                         return 1;
 630                 }
 631                 else {
 632                         //check for blank file
 633                         gobble(fileHandle);
 634                         if (fileHandle.eof()) { mothurOut("[ERROR]: " + completeFileName + " is blank. Please correct."); mothurOutEndLine();  }
 635
 636                         return 0;
 637                 }
 638         }
 639         catch(exception& e) {
 640                 errorOut(e, "MothurOut", "openInputFile");
 641                 exit(1);
 642         }
 643 }
 644 /***********************************************************************/
 645
 646 int MothurOut::renameFile(string oldName, string newName){
 647         try {
 648                 ifstream inTest;
 649                 int exist = openInputFile(newName, inTest, "");
 650
 651         #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 652                 if (exist == 0) { //you could open it so you want to delete it
 653                         inTest.close();
 654                         string command = "rm " + newName;
 655                         system(command.c_str());
 656                 }
 657
 658                 string command = "mv " + oldName + " " + newName;
 659                 system(command.c_str());
 660         #else
 661                 remove(newName.c_str());
 662                 int renameOk = rename(oldName.c_str(), newName.c_str());
 663         #endif
 664                 return 0;
 665
 666         }
 667         catch(exception& e) {
 668                 errorOut(e, "MothurOut", "renameFile");
 669                 exit(1);
 670         }
 671 }
 672
 673 /***********************************************************************/
 674
 675 int MothurOut::openOutputFile(string fileName, ofstream& fileHandle){
 676         try {
 677
 678                 string completeFileName = getFullPathName(fileName);
 679
 680                 fileHandle.open(completeFileName.c_str(), ios::trunc);
 681                 if(!fileHandle) {
 682                         mothurOut("[ERROR]: Could not open " + completeFileName); mothurOutEndLine();
 683                         return 1;
 684                 }
 685                 else {
 686                         return 0;
 687                 }
 688         }
 689         catch(exception& e) {
 690                 errorOut(e, "MothurOut", "openOutputFile");
 691                 exit(1);
 692         }
 693
 694 }
 695
 696 /**************************************************************************************************/
 697 void MothurOut::appendFiles(string temp, string filename) {
 698         try{
 699                 ofstream output;
 700                 ifstream input;
 701
 702                 //open output file in append mode
 703                 openOutputFileAppend(filename, output);
 704                 int ableToOpen = openInputFile(temp, input, "no error");
 705
 706                 if (ableToOpen == 0) { //you opened it
 707                         while(char c = input.get()){
 708                                 if(input.eof())         {       break;                  }
 709                                 else                            {       output << c;    }
 710                         }
 711                         input.close();
 712                 }
 713
 714                 output.close();
 715         }
 716         catch(exception& e) {
 717                 errorOut(e, "MothurOut", "appendFiles");
 718                 exit(1);
 719         }
 720 }
 721
 722 /**************************************************************************************************/
 723 string MothurOut::sortFile(string distFile, string outputDir){
 724         try {
 725
 726                 //if (outputDir == "") {  outputDir += hasPath(distFile);  }
 727                 string outfile = getRootName(distFile) + "sorted.dist";
 728
 729
 730                 //if you can, use the unix sort since its been optimized for years
 731                 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 732                         string command = "sort -n -k +3 " + distFile + " -o " + outfile;
 733                         system(command.c_str());
 734                 #else //you are stuck with my best attempt...
 735                         //windows sort does not have a way to specify a column, only a character in the line
 736                         //since we cannot assume that the distance will always be at the the same character location on each line
 737                         //due to variable sequence name lengths, I chose to force the distance into first position, then sort and then put it back.
 738
 739                         //read in file line by file and put distance first
 740                         string tempDistFile = distFile + ".temp";
 741                         ifstream input;
 742                         ofstream output;
 743                         openInputFile(distFile, input);
 744                         openOutputFile(tempDistFile, output);
 745
 746                         string firstName, secondName;
 747                         float dist;
 748                         while (input) {
 749                                 input >> firstName >> secondName >> dist;
 750                                 output << dist << '\t' << firstName << '\t' << secondName << endl;
 751                                 gobble(input);
 752                         }
 753                         input.close();
 754                         output.close();
 755
 756
 757                         //sort using windows sort
 758                         string tempOutfile = outfile + ".temp";
 759                         string command = "sort " + tempDistFile + " /O " + tempOutfile;
 760                         system(command.c_str());
 761
 762                         //read in sorted file and put distance at end again
 763                         ifstream input2;
 764                         openInputFile(tempOutfile, input2);
 765                         openOutputFile(outfile, output);
 766
 767                         while (input2) {
 768                                 input2 >> dist >> firstName >> secondName;
 769                                 output << firstName << '\t' << secondName << '\t' << dist << endl;
 770                                 gobble(input2);
 771                         }
 772                         input2.close();
 773                         output.close();
 774
 775                         //remove temp files
 776                         remove(tempDistFile.c_str());
 777                         remove(tempOutfile.c_str());
 778                 #endif
 779
 780                 return outfile;
 781         }
 782         catch(exception& e) {
 783                 errorOut(e, "MothurOut", "sortFile");
 784                 exit(1);
 785         }
 786 }
 787 /**************************************************************************************************/
 788 vector<unsigned long int> MothurOut::setFilePosFasta(string filename, int& num) {
 789         try {
 790                         vector<unsigned long int> positions;
 791                         ifstream inFASTA;
 792                         openInputFile(filename, inFASTA);
 793
 794                         string input;
 795                         while(!inFASTA.eof()){
 796                                 input = getline(inFASTA);
 797                                 if (input.length() != 0) {
 798                                         if(input[0] == '>'){    unsigned long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); }
 799                                 }
 800                                 gobble(inFASTA); //has to be here since windows line endings are 2 characters and mess up the positions
 801                         }
 802                         inFASTA.close();
 803
 804                         num = positions.size();
 805
 806                         /*FILE * pFile;
 807                         long size;
 808
 809                         //get num bytes in file
 810                         pFile = fopen (filename.c_str(),"rb");
 811                         if (pFile==NULL) perror ("Error opening file");
 812                         else{
 813                                 fseek (pFile, 0, SEEK_END);
 814                                 size=ftell (pFile);
 815                                 fclose (pFile);
 816                         }*/
 817
 818                         unsigned long int size = positions[(positions.size()-1)];
 819                         ifstream in;
 820                         openInputFile(filename, in);
 821
 822                         in.seekg(size);
 823
 824                         while(char c = in.get()){
 825                                 if(in.eof())            {       break;  }
 826                                 else                            {       size++; }
 827                         }
 828                         in.close();
 829
 830                         positions.push_back(size);
 831
 832                         return positions;
 833         }
 834         catch(exception& e) {
 835                 errorOut(e, "MothurOut", "setFilePosFasta");
 836                 exit(1);
 837         }
 838 }
 839 /**************************************************************************************************/
 840 vector<unsigned long int> MothurOut::setFilePosEachLine(string filename, int& num) {
 841         try {
 842                         filename = getFullPathName(filename);
 843
 844                         vector<unsigned long int> positions;
 845                         ifstream in;
 846                         openInputFile(filename, in);
 847
 848                         string input;
 849                         while(!in.eof()){
 850                                 unsigned long int lastpos = in.tellg();
 851                                 input = getline(in);
 852                                 if (input.length() != 0) {
 853                                         unsigned long int pos = in.tellg();
 854                                         if (pos != -1) { positions.push_back(pos - input.length() - 1); }
 855                                         else {  positions.push_back(lastpos);  }
 856                                 }
 857                                 gobble(in); //has to be here since windows line endings are 2 characters and mess up the positions
 858                         }
 859                         in.close();
 860
 861                         num = positions.size();
 862
 863                         FILE * pFile;
 864                         unsigned long int size;
 865
 866                         //get num bytes in file
 867                         pFile = fopen (filename.c_str(),"rb");
 868                         if (pFile==NULL) perror ("Error opening file");
 869                         else{
 870                                 fseek (pFile, 0, SEEK_END);
 871                                 size=ftell (pFile);
 872                                 fclose (pFile);
 873                         }
 874
 875                         positions.push_back(size);
 876
 877                         return positions;
 878         }
 879         catch(exception& e) {
 880                 errorOut(e, "MothurOut", "setFilePosEachLine");
 881                 exit(1);
 882         }
 883 }
 884 /**************************************************************************************************/
 885
 886 vector<unsigned long int> MothurOut::divideFile(string filename, int& proc) {
 887         try{
 888
 889                 vector<unsigned long int> filePos;
 890                 filePos.push_back(0);
 891
 892                 FILE * pFile;
 893                 unsigned long int size;
 894
 895                 filename = getFullPathName(filename);
 896
 897                 //get num bytes in file
 898                 pFile = fopen (filename.c_str(),"rb");
 899                 if (pFile==NULL) perror ("Error opening file");
 900                 else{
 901                         fseek (pFile, 0, SEEK_END);
 902                         size=ftell (pFile);
 903                         fclose (pFile);
 904                 }
 905
 906                 //estimate file breaks
 907                 unsigned long int chunkSize = 0;
 908                 chunkSize = size / proc;
 909
 910                 //file to small to divide by processors
 911                 if (chunkSize == 0)  {  proc = 1;       filePos.push_back(size); return filePos;        }
 912
 913                 //for each process seekg to closest file break and search for next '>' char. make that the filebreak
 914                 for (int i = 0; i < proc; i++) {
 915                         unsigned long int spot = (i+1) * chunkSize;
 916
 917                         ifstream in;
 918                         openInputFile(filename, in);
 919                         in.seekg(spot);
 920
 921                         //look for next '>'
 922                         unsigned long int newSpot = spot;
 923                         while (!in.eof()) {
 924                            char c = in.get();
 925                            if (c == '>') {   in.putback(c); newSpot = in.tellg(); break;  }
 926                         }
 927
 928                         //there was not another sequence before the end of the file
 929                         unsigned long int sanityPos = in.tellg();
 930
 931                         if (sanityPos == -1) {  break;  }
 932                         else {  filePos.push_back(newSpot);  }
 933
 934                         in.close();
 935                 }
 936
 937                 //save end pos
 938                 filePos.push_back(size);
 939
 940                 //sanity check filePos
 941                 for (int i = 0; i < (filePos.size()-1); i++) {
 942                         if (filePos[(i+1)] <= filePos[i]) {  filePos.erase(filePos.begin()+(i+1)); i--; }
 943                 }
 944
 945                 proc = (filePos.size() - 1);
 946
 947                 return filePos;
 948         }
 949         catch(exception& e) {
 950                 errorOut(e, "MothurOut", "divideFile");
 951                 exit(1);
 952         }
 953 }
 954
 955 /***********************************************************************/
 956
 957 bool MothurOut::isTrue(string f){
 958         try {
 959
 960                 for (int i = 0; i < f.length(); i++) { f[i] = toupper(f[i]); }
 961
 962                 if ((f == "TRUE") || (f == "T")) {      return true;    }
 963                 else {  return false;  }
 964         }
 965         catch(exception& e) {
 966                 errorOut(e, "MothurOut", "isTrue");
 967                 exit(1);
 968         }
 969 }
 970
 971 /***********************************************************************/
 972
 973 float MothurOut::roundDist(float dist, int precision){
 974         try {
 975                 return int(dist * precision + 0.5)/float(precision);
 976         }
 977         catch(exception& e) {
 978                 errorOut(e, "MothurOut", "roundDist");
 979                 exit(1);
 980         }
 981 }
 982 /***********************************************************************/
 983
 984 float MothurOut::ceilDist(float dist, int precision){
 985         try {
 986                 return int(ceil(dist * precision))/float(precision);
 987         }
 988         catch(exception& e) {
 989                 errorOut(e, "MothurOut", "ceilDist");
 990                 exit(1);
 991         }
 992 }
 993
 994 /***********************************************************************/
 995
 996 int MothurOut::getNumNames(string names){
 997         try {
 998                 int count = 0;
 999
1000                 if(names != ""){
1001                         count = 1;
1002                         for(int i=0;i<names.size();i++){
1003                                 if(names[i] == ','){
1004                                         count++;
1005                                 }
1006                         }
1007                 }
1008
1009                 return count;
1010         }
1011         catch(exception& e) {
1012                 errorOut(e, "MothurOut", "getNumNames");
1013                 exit(1);
1014         }
1015 }
1016
1017 /**************************************************************************************************/
1018
1019 vector<vector<double> > MothurOut::binomial(int maxOrder){
1020         try {
1021         vector<vector<double> > binomial(maxOrder+1);
1022
1023     for(int i=0;i<=maxOrder;i++){
1024                 binomial[i].resize(maxOrder+1);
1025                 binomial[i][0]=1;
1026                 binomial[0][i]=0;
1027     }
1028     binomial[0][0]=1;
1029
1030     binomial[1][0]=1;
1031     binomial[1][1]=1;
1032
1033     for(int i=2;i<=maxOrder;i++){
1034                 binomial[1][i]=0;
1035     }
1036
1037     for(int i=2;i<=maxOrder;i++){
1038                 for(int j=1;j<=maxOrder;j++){
1039                         if(i==j){       binomial[i][j]=1;                                                                       }
1040                         if(j>i) {       binomial[i][j]=0;                                                                       }
1041                         else    {       binomial[i][j]=binomial[i-1][j-1]+binomial[i-1][j];     }
1042                 }
1043     }
1044
1045         return binomial;
1046
1047         }
1048         catch(exception& e) {
1049                 errorOut(e, "MothurOut", "binomial");
1050                 exit(1);
1051         }
1052 }
1053
1054 /***********************************************************************/
1055
1056 int MothurOut::factorial(int num){
1057         try {
1058                 int total = 1;
1059
1060                 for (int i = 1; i <= num; i++) {
1061                         total *= i;
1062                 }
1063
1064                 return total;
1065         }
1066         catch(exception& e) {
1067                 errorOut(e, "MothurOut", "factorial");
1068                 exit(1);
1069         }
1070 }
1071 /***********************************************************************/
1072
1073 int MothurOut::getNumSeqs(ifstream& file){
1074         try {
1075                 int numSeqs = count(istreambuf_iterator<char>(file),istreambuf_iterator<char>(), '>');
1076                 file.seekg(0);
1077                 return numSeqs;
1078         }
1079         catch(exception& e) {
1080                 errorOut(e, "MothurOut", "getNumSeqs");
1081                 exit(1);
1082         }
1083 }
1084 /***********************************************************************/
1085 void MothurOut::getNumSeqs(ifstream& file, int& numSeqs){
1086         try {
1087                 string input;
1088                 numSeqs = 0;
1089                 while(!file.eof()){
1090                         input = getline(file);
1091                         if (input.length() != 0) {
1092                                 if(input[0] == '>'){ numSeqs++; }
1093                         }
1094                 }
1095         }
1096         catch(exception& e) {
1097                 errorOut(e, "MothurOut", "getNumSeqs");
1098                 exit(1);
1099         }
1100 }
1101 /***********************************************************************/
1102
1103 //This function parses the estimator options and puts them in a vector
1104 void MothurOut::splitAtChar(string& estim, vector<string>& container, char symbol) {
1105         try {
1106                 string individual;
1107
1108                 while (estim.find_first_of(symbol) != -1) {
1109                         individual = estim.substr(0,estim.find_first_of(symbol));
1110                         if ((estim.find_first_of(symbol)+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
1111                                 estim = estim.substr(estim.find_first_of(symbol)+1, estim.length());
1112                                 container.push_back(individual);
1113                         }
1114                 }
1115                 //get last one
1116                 container.push_back(estim);
1117         }
1118         catch(exception& e) {
1119                 errorOut(e, "MothurOut", "splitAtChar");
1120                 exit(1);
1121         }
1122 }
1123
1124 /***********************************************************************/
1125
1126 //This function parses the estimator options and puts them in a vector
1127 void MothurOut::splitAtDash(string& estim, vector<string>& container) {
1128         try {
1129                 string individual = "";
1130                 int estimLength = estim.size();
1131                 for(int i=0;i<estimLength;i++){
1132                         if(estim[i] == '-'){
1133                                 container.push_back(individual);
1134                                 individual = "";
1135                         }
1136                         else{
1137                                 individual += estim[i];
1138                         }
1139                 }
1140                 container.push_back(individual);
1141
1142
1143         /*      string individual;
1144
1145                 while (estim.find_first_of('-') != -1) {
1146                         individual = estim.substr(0,estim.find_first_of('-'));
1147                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
1148                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
1149                                 container.push_back(individual);
1150                         }
1151                 }
1152                 //get last one
1153                 container.push_back(estim); */
1154         }
1155         catch(exception& e) {
1156                 errorOut(e, "MothurOut", "splitAtDash");
1157                 exit(1);
1158         }
1159 }
1160
1161 /***********************************************************************/
1162 //This function parses the label options and puts them in a set
1163 void MothurOut::splitAtDash(string& estim, set<string>& container) {
1164         try {
1165                 string individual = "";
1166                 int estimLength = estim.size();
1167                 for(int i=0;i<estimLength;i++){
1168                         if(estim[i] == '-'){
1169                                 container.insert(individual);
1170                                 individual = "";
1171                         }
1172                         else{
1173                                 individual += estim[i];
1174                         }
1175                 }
1176                 container.insert(individual);
1177
1178         //      string individual;
1179
1180         //      while (estim.find_first_of('-') != -1) {
1181         //              individual = estim.substr(0,estim.find_first_of('-'));
1182         //              if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
1183         //                      estim = estim.substr(estim.find_first_of('-')+1, estim.length());
1184         //                      container.insert(individual);
1185         //              }
1186         //      }
1187                 //get last one
1188         //      container.insert(estim);
1189
1190         }
1191         catch(exception& e) {
1192                 errorOut(e, "MothurOut", "splitAtDash");
1193                 exit(1);
1194         }
1195 }
1196 /***********************************************************************/
1197 //This function parses the line options and puts them in a set
1198 void MothurOut::splitAtDash(string& estim, set<int>& container) {
1199         try {
1200                 string individual;
1201                 int lineNum;
1202
1203                 while (estim.find_first_of('-') != -1) {
1204                         individual = estim.substr(0,estim.find_first_of('-'));
1205                         if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
1206                                 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
1207                                 convert(individual, lineNum); //convert the string to int
1208                                 container.insert(lineNum);
1209                         }
1210                 }
1211                 //get last one
1212                 convert(estim, lineNum); //convert the string to int
1213                 container.insert(lineNum);
1214         }
1215         catch(exception& e) {
1216                 errorOut(e, "MothurOut", "splitAtDash");
1217                 exit(1);
1218         }
1219 }
1220 /***********************************************************************/
1221 //This function parses the a string and puts peices in a vector
1222 void MothurOut::splitAtComma(string& estim, vector<string>& container) {
1223         try {
1224                 string individual = "";
1225                 int estimLength = estim.size();
1226                 for(int i=0;i<estimLength;i++){
1227                         if(estim[i] == ','){
1228                                 container.push_back(individual);
1229                                 individual = "";
1230                         }
1231                         else{
1232                                 individual += estim[i];
1233                         }
1234                 }
1235                 container.push_back(individual);
1236
1237
1238
1239
1240 //              string individual;
1241 //
1242 //              while (estim.find_first_of(',') != -1) {
1243 //                      individual = estim.substr(0,estim.find_first_of(','));
1244 //                      if ((estim.find_first_of(',')+1) <= estim.length()) { //checks to make sure you don't have comma at end of string
1245 //                              estim = estim.substr(estim.find_first_of(',')+1, estim.length());
1246 //                              container.push_back(individual);
1247 //                      }
1248 //              }
1249 //              //get last one
1250 //              container.push_back(estim);
1251         }
1252         catch(exception& e) {
1253                 errorOut(e, "MothurOut", "splitAtComma");
1254                 exit(1);
1255         }
1256 }
1257 /***********************************************************************/
1258
1259 //This function splits up the various option parameters
1260 void MothurOut::splitAtComma(string& prefix, string& suffix){
1261         try {
1262                 prefix = suffix.substr(0,suffix.find_first_of(','));
1263                 if ((suffix.find_first_of(',')+2) <= suffix.length()) {  //checks to make sure you don't have comma at end of string
1264                         suffix = suffix.substr(suffix.find_first_of(',')+1, suffix.length());
1265                         string space = " ";
1266                         while(suffix.at(0) == ' ')
1267                                 suffix = suffix.substr(1, suffix.length());
1268                 }
1269
1270         }
1271         catch(exception& e) {
1272                 errorOut(e, "MothurOut", "splitAtComma");
1273                 exit(1);
1274         }
1275 }
1276 /***********************************************************************/
1277
1278 //This function separates the key value from the option value i.e. dist=96_...
1279 void MothurOut::splitAtEquals(string& key, string& value){
1280         try {
1281                 if(value.find_first_of('=') != -1){
1282                         key = value.substr(0,value.find_first_of('='));
1283                         if ((value.find_first_of('=')+1) <= value.length()) {
1284                                 value = value.substr(value.find_first_of('=')+1, value.length());
1285                         }
1286                 }else{
1287                         key = value;
1288                         value = 1;
1289                 }
1290         }
1291         catch(exception& e) {
1292                 errorOut(e, "MothurOut", "splitAtEquals");
1293                 exit(1);
1294         }
1295 }
1296
1297 /**************************************************************************************************/
1298
1299 bool MothurOut::inUsersGroups(string groupname, vector<string> Groups) {
1300         try {
1301                 for (int i = 0; i < Groups.size(); i++) {
1302                         if (groupname == Groups[i]) { return true; }
1303                 }
1304                 return false;
1305         }
1306         catch(exception& e) {
1307                 errorOut(e, "MothurOut", "inUsersGroups");
1308                 exit(1);
1309         }
1310 }
1311 /**************************************************************************************************/
1312 //returns true if any of the strings in first vector are in second vector
1313 bool MothurOut::inUsersGroups(vector<string> groupnames, vector<string> Groups) {
1314         try {
1315
1316                 for (int i = 0; i < groupnames.size(); i++) {
1317                         if (inUsersGroups(groupnames[i], Groups)) { return true; }
1318                 }
1319                 return false;
1320         }
1321         catch(exception& e) {
1322                 errorOut(e, "MothurOut", "inUsersGroups");
1323                 exit(1);
1324         }
1325 }
1326 /***********************************************************************/
1327 //this function determines if the user has given us labels that are smaller than the given label.
1328 //if so then it returns true so that the calling function can run the previous valid distance.
1329 //it's a "smart" distance function.  It also checks for invalid labels.
1330 bool MothurOut::anyLabelsToProcess(string label, set<string>& userLabels, string errorOff) {
1331         try {
1332
1333                 set<string>::iterator it;
1334                 vector<float> orderFloat;
1335                 map<string, float> userMap;  //the conversion process removes trailing 0's which we need to put back
1336                 map<string, float>::iterator it2;
1337                 float labelFloat;
1338                 bool smaller = false;
1339
1340                 //unique is the smallest line
1341                 if (label == "unique") {  return false;  }
1342                 else {
1343                         if (convertTestFloat(label, labelFloat)) {
1344                                 convert(label, labelFloat);
1345                         }else { //cant convert
1346                                 return false;
1347                         }
1348                 }
1349
1350                 //go through users set and make them floats
1351                 for(it = userLabels.begin(); it != userLabels.end(); ++it) {
1352
1353                         float temp;
1354                         if ((*it != "unique") && (convertTestFloat(*it, temp) == true)){
1355                                 convert(*it, temp);
1356                                 orderFloat.push_back(temp);
1357                                 userMap[*it] = temp;
1358                         }else if (*it == "unique") {
1359                                 orderFloat.push_back(-1.0);
1360                                 userMap["unique"] = -1.0;
1361                         }else {
1362                                 if (errorOff == "") {  cout << *it << " is not a valid label." << endl;  }
1363                                 userLabels.erase(*it);
1364                                 it--;
1365                         }
1366                 }
1367
1368                 //sort order
1369                 sort(orderFloat.begin(), orderFloat.end());
1370
1371                 /*************************************************/
1372                 //is this label bigger than any of the users labels
1373                 /*************************************************/
1374
1375                 //loop through order until you find a label greater than label
1376                 for (int i = 0; i < orderFloat.size(); i++) {
1377                         if (orderFloat[i] < labelFloat) {
1378                                 smaller = true;
1379                                 if (orderFloat[i] == -1) {
1380                                         if (errorOff == "") { cout << "Your file does not include the label unique." << endl; }
1381                                         userLabels.erase("unique");
1382                                 }
1383                                 else {
1384                                         if (errorOff == "") { cout << "Your file does not include the label " << endl; }
1385                                         string s = "";
1386                                         for (it2 = userMap.begin(); it2!= userMap.end(); it2++) {
1387                                                 if (it2->second == orderFloat[i]) {
1388                                                         s = it2->first;
1389                                                         //remove small labels
1390                                                         userLabels.erase(s);
1391                                                         break;
1392                                                 }
1393                                         }
1394                                         if (errorOff == "") {cout << s <<  ". I will use the next smallest distance. " << endl; }
1395                                 }
1396                         //since they are sorted once you find a bigger one stop looking
1397                         }else { break; }
1398                 }
1399
1400                 return smaller;
1401
1402         }
1403         catch(exception& e) {
1404                 errorOut(e, "MothurOut", "anyLabelsToProcess");
1405                 exit(1);
1406         }
1407 }
1408
1409 /**************************************************************************************************/
1410 bool MothurOut::checkReleaseVersion(ifstream& file, string version) {
1411         try {
1412
1413                 bool good = true;
1414
1415                 string line = getline(file);
1416
1417                 //before we added this check
1418                 if (line[0] != '#') {  good = false;  }
1419                 else {
1420                         //rip off #
1421                         line = line.substr(1);
1422
1423                         vector<string> versionVector;
1424                         splitAtChar(version, versionVector, '.');
1425
1426                         //check file version
1427                         vector<string> linesVector;
1428                         splitAtChar(line, linesVector, '.');
1429
1430                         if (versionVector.size() != linesVector.size()) { good = false; }
1431                         else {
1432                                 for (int j = 0; j < versionVector.size(); j++) {
1433                                         int num1, num2;
1434                                         convert(versionVector[j], num1);
1435                                         convert(linesVector[j], num2);
1436
1437                                         //if mothurs version is newer than this files version, then we want to remake it
1438                                         if (num1 > num2) {  good = false; break;  }
1439                                 }
1440                         }
1441
1442                 }
1443
1444                 if (!good) {  file.close();  }
1445                 else { file.seekg(0);  }
1446
1447                 return good;
1448         }
1449         catch(exception& e) {
1450                 errorOut(e, "MothurOut", "checkReleaseVersion");
1451                 exit(1);
1452         }
1453 }
1454 /**************************************************************************************************/
1455
1456
1457
1458
1459