5 * Created by westcott on 2/25/10.
6 * Copyright 2010 Schloss Lab. All rights reserved.
10 #include "mothurout.h"
13 /******************************************************/
14 MothurOut* MothurOut::getInstance() {
15 if( _uniqueInstance == 0) {
16 _uniqueInstance = new MothurOut();
18 return _uniqueInstance;
20 /*********************************************************************************************/
21 void MothurOut::printCurrentFiles() {
23 if (accnosfile != "") { mothurOut("accnos=" + accnosfile); mothurOutEndLine(); }
24 if (columnfile != "") { mothurOut("column=" + columnfile); mothurOutEndLine(); }
25 if (designfile != "") { mothurOut("design=" + designfile); mothurOutEndLine(); }
26 if (fastafile != "") { mothurOut("fasta=" + fastafile); mothurOutEndLine(); }
27 if (groupfile != "") { mothurOut("group=" + groupfile); mothurOutEndLine(); }
28 if (listfile != "") { mothurOut("list=" + listfile); mothurOutEndLine(); }
29 if (namefile != "") { mothurOut("name=" + namefile); mothurOutEndLine(); }
30 if (oligosfile != "") { mothurOut("oligos=" + oligosfile); mothurOutEndLine(); }
31 if (orderfile != "") { mothurOut("order=" + orderfile); mothurOutEndLine(); }
32 if (ordergroupfile != "") { mothurOut("ordergroup=" + ordergroupfile); mothurOutEndLine(); }
33 if (phylipfile != "") { mothurOut("phylip=" + phylipfile); mothurOutEndLine(); }
34 if (qualfile != "") { mothurOut("qfile=" + qualfile); mothurOutEndLine(); }
35 if (rabundfile != "") { mothurOut("rabund=" + rabundfile); mothurOutEndLine(); }
36 if (relabundfile != "") { mothurOut("relabund=" + relabundfile); mothurOutEndLine(); }
37 if (sabundfile != "") { mothurOut("sabund=" + sabundfile); mothurOutEndLine(); }
38 if (sfffile != "") { mothurOut("sff=" + sfffile); mothurOutEndLine(); }
39 if (sharedfile != "") { mothurOut("shared=" + sharedfile); mothurOutEndLine(); }
40 if (taxonomyfile != "") { mothurOut("taxonomy=" + taxonomyfile); mothurOutEndLine(); }
41 if (treefile != "") { mothurOut("tree=" + treefile); mothurOutEndLine(); }
42 if (flowfile != "") { mothurOut("flow=" + flowfile); mothurOutEndLine(); }
43 if (processors != "1") { mothurOut("processors=" + processors); mothurOutEndLine(); }
47 errorOut(e, "MothurOut", "printCurrentFiles");
51 /*********************************************************************************************/
52 bool MothurOut::hasCurrentFiles() {
54 bool hasCurrent = false;
56 if (accnosfile != "") { return true; }
57 if (columnfile != "") { return true; }
58 if (designfile != "") { return true; }
59 if (fastafile != "") { return true; }
60 if (groupfile != "") { return true; }
61 if (listfile != "") { return true; }
62 if (namefile != "") { return true; }
63 if (oligosfile != "") { return true; }
64 if (orderfile != "") { return true; }
65 if (ordergroupfile != "") { return true; }
66 if (phylipfile != "") { return true; }
67 if (qualfile != "") { return true; }
68 if (rabundfile != "") { return true; }
69 if (relabundfile != "") { return true; }
70 if (sabundfile != "") { return true; }
71 if (sfffile != "") { return true; }
72 if (sharedfile != "") { return true; }
73 if (taxonomyfile != "") { return true; }
74 if (treefile != "") { return true; }
75 if (flowfile != "") { return true; }
76 if (processors != "1") { return true; }
82 errorOut(e, "MothurOut", "hasCurrentFiles");
87 /*********************************************************************************************/
88 void MothurOut::clearCurrentFiles() {
112 catch(exception& e) {
113 errorOut(e, "MothurOut", "clearCurrentFiles");
117 /*********************************************************************************************/
118 void MothurOut::setFileName(string filename) {
120 logFileName = filename;
124 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
126 if (pid == 0) { //only one process should output to screen
129 openOutputFile(filename, out);
135 catch(exception& e) {
136 errorOut(e, "MothurOut", "setFileName");
140 /*********************************************************************************************/
141 void MothurOut::setDefaultPath(string pathname) {
144 //add / to name if needed
145 string lastChar = pathname.substr(pathname.length()-1);
146 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
147 if (lastChar != "/") { pathname += "/"; }
149 if (lastChar != "\\") { pathname += "\\"; }
152 defaultPath = pathname;
155 catch(exception& e) {
156 errorOut(e, "MothurOut", "setDefaultPath");
160 /*********************************************************************************************/
161 void MothurOut::setOutputDir(string pathname) {
163 outputDir = pathname;
165 catch(exception& e) {
166 errorOut(e, "MothurOut", "setOutputDir");
170 /*********************************************************************************************/
171 void MothurOut::closeLog() {
176 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
178 if (pid == 0) { //only one process should output to screen
187 catch(exception& e) {
188 errorOut(e, "MothurOut", "closeLog");
193 /*********************************************************************************************/
194 MothurOut::~MothurOut() {
199 catch(exception& e) {
200 errorOut(e, "MothurOut", "MothurOut");
204 /*********************************************************************************************/
205 void MothurOut::mothurOut(string output) {
210 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
212 if (pid == 0) { //only one process should output to screen
222 catch(exception& e) {
223 errorOut(e, "MothurOut", "MothurOut");
227 /*********************************************************************************************/
228 void MothurOut::mothurOutEndLine() {
232 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
234 if (pid == 0) { //only one process should output to screen
244 catch(exception& e) {
245 errorOut(e, "MothurOut", "MothurOutEndLine");
249 /*********************************************************************************************/
250 void MothurOut::mothurOut(string output, ofstream& outputFile) {
255 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
257 if (pid == 0) { //only one process should output to screen
262 outputFile << output;
268 catch(exception& e) {
269 errorOut(e, "MothurOut", "MothurOut");
273 /*********************************************************************************************/
274 void MothurOut::mothurOutEndLine(ofstream& outputFile) {
278 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
280 if (pid == 0) { //only one process should output to screen
291 catch(exception& e) {
292 errorOut(e, "MothurOut", "MothurOutEndLine");
296 /*********************************************************************************************/
297 void MothurOut::mothurOutJustToLog(string output) {
301 MPI_Comm_rank(MPI_COMM_WORLD, &pid);
303 if (pid == 0) { //only one process should output to screen
312 catch(exception& e) {
313 errorOut(e, "MothurOut", "MothurOutJustToLog");
317 /*********************************************************************************************/
318 void MothurOut::errorOut(exception& e, string object, string function) {
320 //mem_usage(vm, rss);
322 mothurOut("[ERROR]: ");
323 mothurOut(toString(e.what()));
324 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.");
327 /*********************************************************************************************/
328 //The following was originally from http://stackoverflow.com/questions/669438/how-to-get-memory-usage-at-run-time-in-c
329 // process_mem_usage(double &, double &) - takes two doubles by reference,
330 // attempts to read the system-dependent data for a process' virtual memory
331 // size and resident set size, and return the results in KB.
333 // On failure, returns 0.0, 0.0
334 int MothurOut::mem_usage(double& vm_usage, double& resident_set) {
335 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
340 // 'file' stat seems to give the most reliable results
342 ifstream stat_stream("/proc/self/stat",ios_base::in);
344 // dummy vars for leading entries in stat that we don't care about
346 string pid, comm, state, ppid, pgrp, session, tty_nr;
347 string tpgid, flags, minflt, cminflt, majflt, cmajflt;
348 string utime, stime, cutime, cstime, priority, nice;
349 string O, itrealvalue, starttime;
351 // the two fields we want
356 stat_stream >> pid >> comm >> state >> ppid >> pgrp >> session >> tty_nr
357 >> tpgid >> flags >> minflt >> cminflt >> majflt >> cmajflt
358 >> utime >> stime >> cutime >> cstime >> priority >> nice
359 >> O >> itrealvalue >> starttime >> vsize >> rss; // don't care about the rest
361 long page_size_kb = sysconf(_SC_PAGE_SIZE) / 1024; // in case x86-64 is configured to use 2MB pages
362 vm_usage = vsize / 1024.0;
363 resident_set = rss * page_size_kb;
365 mothurOut("Memory Usage: vm = " + toString(vm_usage) + " rss = " + toString(resident_set) + "\n");
369 /* //windows memory usage
370 // Get the list of process identifiers.
371 DWORD aProcesses[1024], cbNeeded, cProcesses;
373 if ( !EnumProcesses( aProcesses, sizeof(aProcesses), &cbNeeded ) ){ return 1; }
375 // Calculate how many process identifiers were returned.
376 cProcesses = cbNeeded / sizeof(DWORD);
378 // Print the memory usage for each process
379 for (int i = 0; i < cProcesses; i++ ) {
380 DWORD processID = aProcesses[i];
382 PROCESS_MEMORY_COUNTERS pmc;
384 HANDLE hProcess = OpenProcess((PROCESS_QUERY_INFORMATION | PROCESS_VM_READ), FALSE, processID);
386 // Print the process identifier.
387 printf( "\nProcess ID: %u\n", processID);
389 if (NULL != hProcess) {
391 if ( GetProcessMemoryInfo( hProcess, &pmc, sizeof(pmc)) ) {
392 printf( "\tPageFaultCount: 0x%08X\n", pmc.PageFaultCount );
393 printf( "\tPeakWorkingSetSize: 0x%08X\n", pmc.PeakWorkingSetSize );
394 printf( "\tWorkingSetSize: 0x%08X\n", pmc.WorkingSetSize );
395 printf( "\tQuotaPeakPagedPoolUsage: 0x%08X\n", pmc.QuotaPeakPagedPoolUsage );
396 printf( "\tQuotaPagedPoolUsage: 0x%08X\n", pmc.QuotaPagedPoolUsage );
397 printf( "\tQuotaPeakNonPagedPoolUsage: 0x%08X\n", pmc.QuotaPeakNonPagedPoolUsage );
398 printf( "\tQuotaNonPagedPoolUsage: 0x%08X\n", pmc.QuotaNonPagedPoolUsage );
399 printf( "\tPagefileUsage: 0x%08X\n", pmc.PagefileUsage );
400 printf( "\tPeakPagefileUsage: 0x%08X\n", pmc.PeakPagefileUsage );
402 CloseHandle(hProcess);
412 /***********************************************************************/
413 int MothurOut::openOutputFileAppend(string fileName, ofstream& fileHandle){
415 fileName = getFullPathName(fileName);
417 fileHandle.open(fileName.c_str(), ios::app);
419 mothurOut("[ERROR]: Could not open " + fileName); mothurOutEndLine();
426 catch(exception& e) {
427 errorOut(e, "MothurOut", "openOutputFileAppend");
431 /***********************************************************************/
432 void MothurOut::gobble(istream& f){
436 while(isspace(d=f.get())) { ;}
439 catch(exception& e) {
440 errorOut(e, "MothurOut", "gobble");
444 /***********************************************************************/
445 void MothurOut::gobble(istringstream& f){
448 while(isspace(d=f.get())) {;}
451 catch(exception& e) {
452 errorOut(e, "MothurOut", "gobble");
457 /***********************************************************************/
459 string MothurOut::getline(istringstream& fileHandle) {
464 while (!fileHandle.eof()) {
466 char c = fileHandle.get();
468 //are you at the end of the line
469 if ((c == '\n') || (c == '\r') || (c == '\f')){ break; }
476 catch(exception& e) {
477 errorOut(e, "MothurOut", "getline");
481 /***********************************************************************/
483 string MothurOut::getline(ifstream& fileHandle) {
490 char c = fileHandle.get();
492 //are you at the end of the line
493 if ((c == '\n') || (c == '\r') || (c == '\f') || (c == EOF)){ break; }
500 catch(exception& e) {
501 errorOut(e, "MothurOut", "getline");
505 /***********************************************************************/
507 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
508 #ifdef USE_COMPRESSION
509 inline bool endsWith(string s, const char * suffix){
510 size_t suffixLength = strlen(suffix);
511 return s.size() >= suffixLength && s.substr(s.size() - suffixLength, suffixLength).compare(suffix) == 0;
516 string MothurOut::getRootName(string longName){
519 string rootName = longName;
521 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
522 #ifdef USE_COMPRESSION
523 if (endsWith(rootName, ".gz") || endsWith(rootName, ".bz2")) {
524 int pos = rootName.find_last_of('.');
525 rootName = rootName.substr(0, pos);
526 cerr << "shortening " << longName << " to " << rootName << "\n";
530 if(rootName.find_last_of(".") != rootName.npos){
531 int pos = rootName.find_last_of('.')+1;
532 rootName = rootName.substr(0, pos);
537 catch(exception& e) {
538 errorOut(e, "MothurOut", "getRootName");
542 /***********************************************************************/
544 string MothurOut::getSimpleName(string longName){
546 string simpleName = longName;
549 found=longName.find_last_of("/\\");
551 if(found != longName.npos){
552 simpleName = longName.substr(found+1);
557 catch(exception& e) {
558 errorOut(e, "MothurOut", "getSimpleName");
563 /***********************************************************************/
565 string MothurOut::getPathName(string longName){
567 string rootPathName = longName;
569 if(longName.find_last_of("/\\") != longName.npos){
570 int pos = longName.find_last_of("/\\")+1;
571 rootPathName = longName.substr(0, pos);
576 catch(exception& e) {
577 errorOut(e, "MothurOut", "getPathName");
582 /***********************************************************************/
584 string MothurOut::hasPath(string longName){
589 found=longName.find_last_of("~/\\");
591 if(found != longName.npos){
592 path = longName.substr(0, found+1);
597 catch(exception& e) {
598 errorOut(e, "MothurOut", "hasPath");
603 /***********************************************************************/
605 string MothurOut::getExtension(string longName){
607 string extension = longName;
609 if(longName.find_last_of('.') != longName.npos){
610 int pos = longName.find_last_of('.');
611 extension = longName.substr(pos, longName.length());
616 catch(exception& e) {
617 errorOut(e, "MothurOut", "getExtension");
621 /***********************************************************************/
622 bool MothurOut::isBlank(string fileName){
625 fileName = getFullPathName(fileName);
628 fileHandle.open(fileName.c_str());
630 mothurOut("[ERROR]: Could not open " + fileName); mothurOutEndLine();
633 //check for blank file
635 if (fileHandle.eof()) { fileHandle.close(); return true; }
640 catch(exception& e) {
641 errorOut(e, "MothurOut", "isBlank");
645 /***********************************************************************/
647 string MothurOut::getFullPathName(string fileName){
650 string path = hasPath(fileName);
654 if (path == "") { return fileName; } //its a simple name
655 else { //we need to complete the pathname
656 // ex. ../../../filename
657 // cwd = /user/work/desktop
660 //get current working directory
661 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
663 if (path.find("~") != -1) { //go to home directory
666 char *homepath = NULL;
667 homepath = getenv ("HOME");
668 if ( homepath != NULL) { homeDir = homepath; }
669 else { homeDir = ""; }
671 newFileName = homeDir + fileName.substr(fileName.find("~")+1);
674 if (path.rfind("./") == string::npos) { return fileName; } //already complete name
675 else { newFileName = fileName.substr(fileName.rfind("./")+2); } //save the complete part of the name
677 //char* cwdpath = new char[1024];
679 //cwdpath=getcwd(cwdpath,size);
682 char *cwdpath = NULL;
683 cwdpath = getcwd(NULL, 0); // or _getcwd
684 if ( cwdpath != NULL) { cwd = cwdpath; }
690 if (cwd.length() > 0) { simpleCWD = cwd.substr(1); }
692 //break apart the current working directory
694 while (simpleCWD.find_first_of('/') != string::npos) {
695 string dir = simpleCWD.substr(0,simpleCWD.find_first_of('/'));
696 simpleCWD = simpleCWD.substr(simpleCWD.find_first_of('/')+1, simpleCWD.length());
699 //get last one // ex. ../../../filename = /user/work/desktop/filename
700 dirs.push_back(simpleCWD); //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
703 int index = dirs.size()-1;
705 while((pos = path.rfind("./")) != string::npos) { //while you don't have a complete path
706 if (pos == 0) { break; //you are at the end
707 }else if (path[(pos-1)] == '.') { //you want your parent directory ../
708 path = path.substr(0, pos-1);
710 if (index == 0) { break; }
711 }else if (path[(pos-1)] == '/') { //you want the current working dir ./
712 path = path.substr(0, pos);
713 }else if (pos == 1) { break; //you are at the end
714 }else { cout << "cannot resolve path for " << fileName << endl; return fileName; }
717 for (int i = index; i >= 0; i--) {
718 newFileName = dirs[i] + "/" + newFileName;
721 newFileName = "/" + newFileName;
725 if (path.find("~") != string::npos) { //go to home directory
726 string homeDir = getenv ("HOMEPATH");
727 newFileName = homeDir + fileName.substr(fileName.find("~")+1);
730 if (path.rfind(".\\") == string::npos) { return fileName; } //already complete name
731 else { newFileName = fileName.substr(fileName.rfind(".\\")+2); } //save the complete part of the name
733 char *cwdpath = NULL;
734 cwdpath = getcwd(NULL, 0); // or _getcwd
735 if ( cwdpath != NULL) { cwd = cwdpath; }
738 //break apart the current working directory
740 while (cwd.find_first_of('\\') != -1) {
741 string dir = cwd.substr(0,cwd.find_first_of('\\'));
742 cwd = cwd.substr(cwd.find_first_of('\\')+1, cwd.length());
747 dirs.push_back(cwd); //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
749 int index = dirs.size()-1;
751 while((pos = path.rfind(".\\")) != string::npos) { //while you don't have a complete path
752 if (pos == 0) { break; //you are at the end
753 }else if (path[(pos-1)] == '.') { //you want your parent directory ../
754 path = path.substr(0, pos-1);
756 if (index == 0) { break; }
757 }else if (path[(pos-1)] == '\\') { //you want the current working dir ./
758 path = path.substr(0, pos);
759 }else if (pos == 1) { break; //you are at the end
760 }else { cout << "cannot resolve path for " << fileName << endl; return fileName; }
763 for (int i = index; i >= 0; i--) {
764 newFileName = dirs[i] + "\\" + newFileName;
773 catch(exception& e) {
774 errorOut(e, "MothurOut", "getFullPathName");
778 /***********************************************************************/
780 int MothurOut::openInputFile(string fileName, ifstream& fileHandle, string m){
783 string completeFileName = getFullPathName(fileName);
784 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
785 #ifdef USE_COMPRESSION
786 // check for gzipped or bzipped file
787 if (endsWith(completeFileName, ".gz") || endsWith(completeFileName, ".bz2")) {
788 string tempName = string(tmpnam(0));
789 mkfifo(tempName.c_str(), 0666);
790 int fork_result = fork();
791 if (fork_result < 0) {
792 cerr << "Error forking.\n";
794 } else if (fork_result == 0) {
795 string command = (endsWith(completeFileName, ".gz") ? "zcat " : "bzcat ") + completeFileName + string(" > ") + tempName;
796 cerr << "Decompressing " << completeFileName << " via temporary named pipe " << tempName << "\n";
797 system(command.c_str());
798 cerr << "Done decompressing " << completeFileName << "\n";
799 remove(tempName.c_str());
802 cerr << "waiting on child process " << fork_result << "\n";
803 completeFileName = tempName;
808 fileHandle.open(completeFileName.c_str());
810 //mothurOut("[ERROR]: Could not open " + completeFileName); mothurOutEndLine();
813 //check for blank file
818 catch(exception& e) {
819 errorOut(e, "MothurOut", "openInputFile - no Error");
823 /***********************************************************************/
825 int MothurOut::openInputFile(string fileName, ifstream& fileHandle){
829 string completeFileName = getFullPathName(fileName);
830 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
831 #ifdef USE_COMPRESSION
832 // check for gzipped or bzipped file
833 if (endsWith(completeFileName, ".gz") || endsWith(completeFileName, ".bz2")) {
834 string tempName = string(tmpnam(0));
835 mkfifo(tempName.c_str(), 0666);
836 int fork_result = fork();
837 if (fork_result < 0) {
838 cerr << "Error forking.\n";
840 } else if (fork_result == 0) {
841 string command = (endsWith(completeFileName, ".gz") ? "zcat " : "bzcat ") + completeFileName + string(" > ") + tempName;
842 cerr << "Decompressing " << completeFileName << " via temporary named pipe " << tempName << "\n";
843 system(command.c_str());
844 cerr << "Done decompressing " << completeFileName << "\n";
845 remove(tempName.c_str());
848 cerr << "waiting on child process " << fork_result << "\n";
849 completeFileName = tempName;
855 fileHandle.open(completeFileName.c_str());
857 mothurOut("[ERROR]: Could not open " + completeFileName); mothurOutEndLine();
861 //check for blank file
863 if (fileHandle.eof()) { mothurOut("[ERROR]: " + completeFileName + " is blank. Please correct."); mothurOutEndLine(); }
868 catch(exception& e) {
869 errorOut(e, "MothurOut", "openInputFile");
873 /***********************************************************************/
875 int MothurOut::renameFile(string oldName, string newName){
878 int exist = openInputFile(newName, inTest, "");
880 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
881 if (exist == 0) { //you could open it so you want to delete it
883 string command = "rm " + newName;
884 system(command.c_str());
887 string command = "mv " + oldName + " " + newName;
888 system(command.c_str());
890 remove(newName.c_str());
891 int renameOk = rename(oldName.c_str(), newName.c_str());
896 catch(exception& e) {
897 errorOut(e, "MothurOut", "renameFile");
902 /***********************************************************************/
904 int MothurOut::openOutputFile(string fileName, ofstream& fileHandle){
907 string completeFileName = getFullPathName(fileName);
908 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
909 #ifdef USE_COMPRESSION
910 // check for gzipped file
911 if (endsWith(completeFileName, ".gz") || endsWith(completeFileName, ".bz2")) {
912 string tempName = string(tmpnam(0));
913 mkfifo(tempName.c_str(), 0666);
914 cerr << "Compressing " << completeFileName << " via temporary named pipe " << tempName << "\n";
915 int fork_result = fork();
916 if (fork_result < 0) {
917 cerr << "Error forking.\n";
919 } else if (fork_result == 0) {
920 string command = string(endsWith(completeFileName, ".gz") ? "gzip" : "bzip2") + " -v > " + completeFileName + string(" < ") + tempName;
921 system(command.c_str());
924 completeFileName = tempName;
929 fileHandle.open(completeFileName.c_str(), ios::trunc);
931 mothurOut("[ERROR]: Could not open " + completeFileName); mothurOutEndLine();
938 catch(exception& e) {
939 errorOut(e, "MothurOut", "openOutputFile");
945 /**************************************************************************************************/
946 void MothurOut::appendFiles(string temp, string filename) {
951 //open output file in append mode
952 openOutputFileAppend(filename, output);
953 int ableToOpen = openInputFile(temp, input, "no error");
954 //int ableToOpen = openInputFile(temp, input);
956 if (ableToOpen == 0) { //you opened it
957 while(char c = input.get()){
958 if(input.eof()) { break; }
959 else { output << c; }
966 catch(exception& e) {
967 errorOut(e, "MothurOut", "appendFiles");
972 /**************************************************************************************************/
973 string MothurOut::sortFile(string distFile, string outputDir){
976 //if (outputDir == "") { outputDir += hasPath(distFile); }
977 string outfile = getRootName(distFile) + "sorted.dist";
980 //if you can, use the unix sort since its been optimized for years
981 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
982 string command = "sort -n -k +3 " + distFile + " -o " + outfile;
983 system(command.c_str());
984 #else //you are stuck with my best attempt...
985 //windows sort does not have a way to specify a column, only a character in the line
986 //since we cannot assume that the distance will always be at the the same character location on each line
987 //due to variable sequence name lengths, I chose to force the distance into first position, then sort and then put it back.
989 //read in file line by file and put distance first
990 string tempDistFile = distFile + ".temp";
993 openInputFile(distFile, input);
994 openOutputFile(tempDistFile, output);
996 string firstName, secondName;
999 input >> firstName >> secondName >> dist;
1000 output << dist << '\t' << firstName << '\t' << secondName << endl;
1007 //sort using windows sort
1008 string tempOutfile = outfile + ".temp";
1009 string command = "sort " + tempDistFile + " /O " + tempOutfile;
1010 system(command.c_str());
1012 //read in sorted file and put distance at end again
1014 openInputFile(tempOutfile, input2);
1015 openOutputFile(outfile, output);
1018 input2 >> dist >> firstName >> secondName;
1019 output << firstName << '\t' << secondName << '\t' << dist << endl;
1026 remove(tempDistFile.c_str());
1027 remove(tempOutfile.c_str());
1032 catch(exception& e) {
1033 errorOut(e, "MothurOut", "sortFile");
1037 /**************************************************************************************************/
1038 vector<unsigned long int> MothurOut::setFilePosFasta(string filename, int& num) {
1040 vector<unsigned long int> positions;
1042 openInputFile(filename, inFASTA);
1045 while(!inFASTA.eof()){
1046 input = getline(inFASTA);
1047 if (input.length() != 0) {
1048 if(input[0] == '>'){ unsigned long int pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); }
1050 gobble(inFASTA); //has to be here since windows line endings are 2 characters and mess up the positions
1054 num = positions.size();
1059 //get num bytes in file
1060 pFile = fopen (filename.c_str(),"rb");
1061 if (pFile==NULL) perror ("Error opening file");
1063 fseek (pFile, 0, SEEK_END);
1068 unsigned long int size = positions[(positions.size()-1)];
1070 openInputFile(filename, in);
1075 if(in.eof()) { break; }
1080 positions.push_back(size);
1084 catch(exception& e) {
1085 errorOut(e, "MothurOut", "setFilePosFasta");
1089 /**************************************************************************************************/
1090 vector<unsigned long int> MothurOut::setFilePosEachLine(string filename, int& num) {
1092 filename = getFullPathName(filename);
1094 vector<unsigned long int> positions;
1096 openInputFile(filename, in);
1100 unsigned long int lastpos = in.tellg();
1101 input = getline(in);
1102 if (input.length() != 0) {
1103 unsigned long int pos = in.tellg();
1104 if (pos != -1) { positions.push_back(pos - input.length() - 1); }
1105 else { positions.push_back(lastpos); }
1107 gobble(in); //has to be here since windows line endings are 2 characters and mess up the positions
1111 num = positions.size();
1114 unsigned long int size;
1116 //get num bytes in file
1117 pFile = fopen (filename.c_str(),"rb");
1118 if (pFile==NULL) perror ("Error opening file");
1120 fseek (pFile, 0, SEEK_END);
1125 positions.push_back(size);
1129 catch(exception& e) {
1130 errorOut(e, "MothurOut", "setFilePosEachLine");
1134 /**************************************************************************************************/
1136 vector<unsigned long int> MothurOut::divideFile(string filename, int& proc) {
1139 vector<unsigned long int> filePos;
1140 filePos.push_back(0);
1143 unsigned long int size;
1145 filename = getFullPathName(filename);
1147 //get num bytes in file
1148 pFile = fopen (filename.c_str(),"rb");
1149 if (pFile==NULL) perror ("Error opening file");
1151 fseek (pFile, 0, SEEK_END);
1156 //estimate file breaks
1157 unsigned long int chunkSize = 0;
1158 chunkSize = size / proc;
1160 //file to small to divide by processors
1161 if (chunkSize == 0) { proc = 1; filePos.push_back(size); return filePos; }
1163 //for each process seekg to closest file break and search for next '>' char. make that the filebreak
1164 for (int i = 0; i < proc; i++) {
1165 unsigned long int spot = (i+1) * chunkSize;
1168 openInputFile(filename, in);
1172 unsigned long int newSpot = spot;
1175 if (c == '>') { in.putback(c); newSpot = in.tellg(); break; }
1178 //there was not another sequence before the end of the file
1179 unsigned long int sanityPos = in.tellg();
1181 if (sanityPos == -1) { break; }
1182 else { filePos.push_back(newSpot); }
1188 filePos.push_back(size);
1190 //sanity check filePos
1191 for (int i = 0; i < (filePos.size()-1); i++) {
1192 if (filePos[(i+1)] <= filePos[i]) { filePos.erase(filePos.begin()+(i+1)); i--; }
1195 proc = (filePos.size() - 1);
1199 catch(exception& e) {
1200 errorOut(e, "MothurOut", "divideFile");
1204 /**************************************************************************************************/
1205 int MothurOut::divideFile(string filename, int& proc, vector<string>& files) {
1208 vector<unsigned long int> filePos = divideFile(filename, proc);
1210 for (int i = 0; i < (filePos.size()-1); i++) {
1214 openInputFile(filename, in);
1215 in.seekg(filePos[i]);
1216 unsigned long int size = filePos[(i+1)] - filePos[i];
1217 char* chunk = new char[size];
1218 in.read(chunk, size);
1222 string fileChunkName = filename + "." + toString(i) + ".tmp";
1224 openOutputFile(fileChunkName, out);
1226 out << chunk << endl;
1231 files.push_back(fileChunkName);
1236 catch(exception& e) {
1237 errorOut(e, "MothurOut", "divideFile");
1241 /***********************************************************************/
1243 bool MothurOut::isTrue(string f){
1246 for (int i = 0; i < f.length(); i++) { f[i] = toupper(f[i]); }
1248 if ((f == "TRUE") || (f == "T")) { return true; }
1249 else { return false; }
1251 catch(exception& e) {
1252 errorOut(e, "MothurOut", "isTrue");
1257 /***********************************************************************/
1259 float MothurOut::roundDist(float dist, int precision){
1261 return int(dist * precision + 0.5)/float(precision);
1263 catch(exception& e) {
1264 errorOut(e, "MothurOut", "roundDist");
1268 /***********************************************************************/
1270 float MothurOut::ceilDist(float dist, int precision){
1272 return int(ceil(dist * precision))/float(precision);
1274 catch(exception& e) {
1275 errorOut(e, "MothurOut", "ceilDist");
1279 /**********************************************************************************************************************/
1280 int MothurOut::readNames(string namefile, map<string, string>& nameMap) {
1285 openInputFile(namefile, in);
1288 if (control_pressed) { break; }
1290 string firstCol, secondCol;
1291 in >> firstCol >> secondCol; gobble(in);
1293 nameMap[firstCol] = secondCol;
1300 catch(exception& e) {
1301 errorOut(e, "MothurOut", "readNames");
1306 /**********************************************************************************************************************/
1307 map<string, int> MothurOut::readNames(string namefile) {
1310 map<string, int> nameMap;
1314 openInputFile(namefile, in);
1317 if (control_pressed) { break; }
1319 string firstCol, secondCol;
1320 in >> firstCol >> secondCol; gobble(in);
1322 int num = getNumNames(secondCol);
1324 nameMap[firstCol] = num;
1331 catch(exception& e) {
1332 errorOut(e, "MothurOut", "readNames");
1336 /**********************************************************************************************************************/
1337 int MothurOut::readNames(string namefile, vector<seqPriorityNode>& nameVector, map<string, string>& fastamap) {
1343 openInputFile(namefile, in);
1346 if (control_pressed) { break; }
1348 string firstCol, secondCol;
1349 in >> firstCol >> secondCol; gobble(in);
1351 int num = getNumNames(secondCol);
1353 map<string, string>::iterator it = fastamap.find(firstCol);
1354 if (it == fastamap.end()) {
1356 mothurOut("[ERROR]: " + firstCol + " is not in your fastafile, but is in your namesfile, please correct."); mothurOutEndLine();
1358 seqPriorityNode temp(num, it->second, firstCol);
1359 nameVector.push_back(temp);
1367 catch(exception& e) {
1368 errorOut(e, "MothurOut", "readNames");
1373 /***********************************************************************/
1375 int MothurOut::getNumNames(string names){
1381 for(int i=0;i<names.size();i++){
1382 if(names[i] == ','){
1390 catch(exception& e) {
1391 errorOut(e, "MothurOut", "getNumNames");
1396 /**************************************************************************************************/
1398 vector<vector<double> > MothurOut::binomial(int maxOrder){
1400 vector<vector<double> > binomial(maxOrder+1);
1402 for(int i=0;i<=maxOrder;i++){
1403 binomial[i].resize(maxOrder+1);
1412 for(int i=2;i<=maxOrder;i++){
1416 for(int i=2;i<=maxOrder;i++){
1417 for(int j=1;j<=maxOrder;j++){
1418 if(i==j){ binomial[i][j]=1; }
1419 if(j>i) { binomial[i][j]=0; }
1420 else { binomial[i][j]=binomial[i-1][j-1]+binomial[i-1][j]; }
1427 catch(exception& e) {
1428 errorOut(e, "MothurOut", "binomial");
1432 /**************************************************************************************************/
1433 unsigned int MothurOut::fromBase36(string base36){
1435 unsigned int num = 0;
1437 map<char, int> converts;
1502 while (i < base36.length()) {
1504 num = 36 * num + converts[c];
1511 catch(exception& e) {
1512 errorOut(e, "MothurOut", "fromBase36");
1516 /***********************************************************************/
1518 int MothurOut::factorial(int num){
1522 for (int i = 1; i <= num; i++) {
1528 catch(exception& e) {
1529 errorOut(e, "MothurOut", "factorial");
1533 /***********************************************************************/
1535 int MothurOut::getNumSeqs(ifstream& file){
1537 int numSeqs = count(istreambuf_iterator<char>(file),istreambuf_iterator<char>(), '>');
1541 catch(exception& e) {
1542 errorOut(e, "MothurOut", "getNumSeqs");
1546 /***********************************************************************/
1547 void MothurOut::getNumSeqs(ifstream& file, int& numSeqs){
1552 input = getline(file);
1553 if (input.length() != 0) {
1554 if(input[0] == '>'){ numSeqs++; }
1558 catch(exception& e) {
1559 errorOut(e, "MothurOut", "getNumSeqs");
1563 /***********************************************************************/
1565 //This function parses the estimator options and puts them in a vector
1566 void MothurOut::splitAtChar(string& estim, vector<string>& container, char symbol) {
1568 string individual = "";
1569 int estimLength = estim.size();
1570 for(int i=0;i<estimLength;i++){
1571 if(estim[i] == symbol){
1572 container.push_back(individual);
1576 individual += estim[i];
1579 container.push_back(individual);
1582 catch(exception& e) {
1583 errorOut(e, "MothurOut", "splitAtChar");
1588 /***********************************************************************/
1590 //This function parses the estimator options and puts them in a vector
1591 void MothurOut::splitAtDash(string& estim, vector<string>& container) {
1593 string individual = "";
1594 int estimLength = estim.size();
1595 for(int i=0;i<estimLength;i++){
1596 if(estim[i] == '-'){
1597 container.push_back(individual);
1601 individual += estim[i];
1604 container.push_back(individual);
1607 /* string individual;
1609 while (estim.find_first_of('-') != -1) {
1610 individual = estim.substr(0,estim.find_first_of('-'));
1611 if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
1612 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
1613 container.push_back(individual);
1617 container.push_back(estim); */
1619 catch(exception& e) {
1620 errorOut(e, "MothurOut", "splitAtDash");
1625 /***********************************************************************/
1626 //This function parses the label options and puts them in a set
1627 void MothurOut::splitAtDash(string& estim, set<string>& container) {
1629 string individual = "";
1630 int estimLength = estim.size();
1631 for(int i=0;i<estimLength;i++){
1632 if(estim[i] == '-'){
1633 container.insert(individual);
1637 individual += estim[i];
1640 container.insert(individual);
1642 // string individual;
1644 // while (estim.find_first_of('-') != -1) {
1645 // individual = estim.substr(0,estim.find_first_of('-'));
1646 // if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
1647 // estim = estim.substr(estim.find_first_of('-')+1, estim.length());
1648 // container.insert(individual);
1652 // container.insert(estim);
1655 catch(exception& e) {
1656 errorOut(e, "MothurOut", "splitAtDash");
1660 /***********************************************************************/
1661 //This function parses the line options and puts them in a set
1662 void MothurOut::splitAtDash(string& estim, set<int>& container) {
1667 while (estim.find_first_of('-') != -1) {
1668 individual = estim.substr(0,estim.find_first_of('-'));
1669 if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
1670 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
1671 convert(individual, lineNum); //convert the string to int
1672 container.insert(lineNum);
1676 convert(estim, lineNum); //convert the string to int
1677 container.insert(lineNum);
1679 catch(exception& e) {
1680 errorOut(e, "MothurOut", "splitAtDash");
1684 /***********************************************************************/
1685 //This function parses the a string and puts peices in a vector
1686 void MothurOut::splitAtComma(string& estim, vector<string>& container) {
1688 string individual = "";
1689 int estimLength = estim.size();
1690 for(int i=0;i<estimLength;i++){
1691 if(estim[i] == ','){
1692 container.push_back(individual);
1696 individual += estim[i];
1699 container.push_back(individual);
1704 // string individual;
1706 // while (estim.find_first_of(',') != -1) {
1707 // individual = estim.substr(0,estim.find_first_of(','));
1708 // if ((estim.find_first_of(',')+1) <= estim.length()) { //checks to make sure you don't have comma at end of string
1709 // estim = estim.substr(estim.find_first_of(',')+1, estim.length());
1710 // container.push_back(individual);
1714 // container.push_back(estim);
1716 catch(exception& e) {
1717 errorOut(e, "MothurOut", "splitAtComma");
1721 /***********************************************************************/
1723 //This function splits up the various option parameters
1724 void MothurOut::splitAtComma(string& prefix, string& suffix){
1726 prefix = suffix.substr(0,suffix.find_first_of(','));
1727 if ((suffix.find_first_of(',')+2) <= suffix.length()) { //checks to make sure you don't have comma at end of string
1728 suffix = suffix.substr(suffix.find_first_of(',')+1, suffix.length());
1730 while(suffix.at(0) == ' ')
1731 suffix = suffix.substr(1, suffix.length());
1735 catch(exception& e) {
1736 errorOut(e, "MothurOut", "splitAtComma");
1740 /***********************************************************************/
1742 //This function separates the key value from the option value i.e. dist=96_...
1743 void MothurOut::splitAtEquals(string& key, string& value){
1745 if(value.find_first_of('=') != -1){
1746 key = value.substr(0,value.find_first_of('='));
1747 if ((value.find_first_of('=')+1) <= value.length()) {
1748 value = value.substr(value.find_first_of('=')+1, value.length());
1755 catch(exception& e) {
1756 errorOut(e, "MothurOut", "splitAtEquals");
1761 /**************************************************************************************************/
1763 bool MothurOut::inUsersGroups(string groupname, vector<string> Groups) {
1765 for (int i = 0; i < Groups.size(); i++) {
1766 if (groupname == Groups[i]) { return true; }
1770 catch(exception& e) {
1771 errorOut(e, "MothurOut", "inUsersGroups");
1775 /**************************************************************************************************/
1776 //returns true if any of the strings in first vector are in second vector
1777 bool MothurOut::inUsersGroups(vector<string> groupnames, vector<string> Groups) {
1780 for (int i = 0; i < groupnames.size(); i++) {
1781 if (inUsersGroups(groupnames[i], Groups)) { return true; }
1785 catch(exception& e) {
1786 errorOut(e, "MothurOut", "inUsersGroups");
1790 /***********************************************************************/
1791 //this function determines if the user has given us labels that are smaller than the given label.
1792 //if so then it returns true so that the calling function can run the previous valid distance.
1793 //it's a "smart" distance function. It also checks for invalid labels.
1794 bool MothurOut::anyLabelsToProcess(string label, set<string>& userLabels, string errorOff) {
1797 set<string>::iterator it;
1798 vector<float> orderFloat;
1799 map<string, float> userMap; //the conversion process removes trailing 0's which we need to put back
1800 map<string, float>::iterator it2;
1802 bool smaller = false;
1804 //unique is the smallest line
1805 if (label == "unique") { return false; }
1807 if (convertTestFloat(label, labelFloat)) {
1808 convert(label, labelFloat);
1809 }else { //cant convert
1814 //go through users set and make them floats
1815 for(it = userLabels.begin(); it != userLabels.end();) {
1818 if ((*it != "unique") && (convertTestFloat(*it, temp) == true)){
1820 orderFloat.push_back(temp);
1821 userMap[*it] = temp;
1823 }else if (*it == "unique") {
1824 orderFloat.push_back(-1.0);
1825 userMap["unique"] = -1.0;
1828 if (errorOff == "") { mothurOut(*it + " is not a valid label."); mothurOutEndLine(); }
1829 userLabels.erase(it++);
1834 sort(orderFloat.begin(), orderFloat.end());
1836 /*************************************************/
1837 //is this label bigger than any of the users labels
1838 /*************************************************/
1840 //loop through order until you find a label greater than label
1841 for (int i = 0; i < orderFloat.size(); i++) {
1842 if (orderFloat[i] < labelFloat) {
1844 if (orderFloat[i] == -1) {
1845 if (errorOff == "") { mothurOut("Your file does not include the label unique."); mothurOutEndLine(); }
1846 userLabels.erase("unique");
1849 if (errorOff == "") { mothurOut("Your file does not include the label "); mothurOutEndLine(); }
1851 for (it2 = userMap.begin(); it2!= userMap.end(); it2++) {
1852 if (it2->second == orderFloat[i]) {
1854 //remove small labels
1855 userLabels.erase(s);
1859 if (errorOff == "") {mothurOut( s + ". I will use the next smallest distance. "); mothurOutEndLine(); }
1861 //since they are sorted once you find a bigger one stop looking
1868 catch(exception& e) {
1869 errorOut(e, "MothurOut", "anyLabelsToProcess");
1874 /**************************************************************************************************/
1875 bool MothurOut::checkReleaseVersion(ifstream& file, string version) {
1880 string line = getline(file);
1882 //before we added this check
1883 if (line[0] != '#') { good = false; }
1886 line = line.substr(1);
1888 vector<string> versionVector;
1889 splitAtChar(version, versionVector, '.');
1891 //check file version
1892 vector<string> linesVector;
1893 splitAtChar(line, linesVector, '.');
1895 if (versionVector.size() != linesVector.size()) { good = false; }
1897 for (int j = 0; j < versionVector.size(); j++) {
1899 convert(versionVector[j], num1);
1900 convert(linesVector[j], num2);
1902 //if mothurs version is newer than this files version, then we want to remake it
1903 if (num1 > num2) { good = false; break; }
1909 if (!good) { file.close(); }
1910 else { file.seekg(0); }
1914 catch(exception& e) {
1915 errorOut(e, "MothurOut", "checkReleaseVersion");
1919 /**************************************************************************************************/
1920 bool MothurOut::isContainingOnlyDigits(string input) {
1923 //are you a digit in ascii code
1924 for (int i = 0;i < input.length(); i++){
1925 if( input[i]>47 && input[i]<58){}
1926 else { return false; }
1931 catch(exception& e) {
1932 errorOut(e, "MothurOut", "isContainingOnlyDigits");
1936 /**************************************************************************************************/