10 * Created by Sarah Westcott on 2/19/09.
11 * Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
15 /* This file contains all the standard incudes we use in the project as well as some common utilities. */
53 /***********************************************************************/
55 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
60 #include <readline/readline.h>
61 #include <readline/history.h>
65 #include <conio.h> //allows unbuffered screen capture from stdin
66 #include <direct.h> //get cwd
71 #define exp(x) (exp((double) x))
72 #define sqrt(x) (sqrt((double) x))
73 #define log10(x) (log10((double) x))
74 #define log2(x) (log10(x)/log10(2))
75 #define isnan(x) ((x) != (x))
76 #define isinf(x) (fabs(x) == std::numeric_limits<double>::infinity())
78 typedef unsigned long ull;
88 IntNode(int lv, int rv, IntNode* l, IntNode* r) : lvalue(lv), rvalue(rv), left(l), right(r) {};
98 /************************************************************/
102 int smallChild; //used to make linkTable work with list and rabund. represents bin number of this cluster node
103 clusterNode(int num, int par, int kid) : numSeq(num), parent(par), smallChild(kid) {};
105 /************************************************************/
111 seqDist(int s1, int s2, float d) : seq1(s1), seq2(s2), dist(d) {}
114 //********************************************************************************************************************
115 //sorts lowest to highest
116 inline bool compareSequenceDistance(seqDist left, seqDist right){
117 return (left.dist < right.dist);
119 /***********************************************************************/
121 // snagged from http://www.parashift.com/c++-faq-lite/misc-technical-issues.html#faq-39.2
122 // works for now, but there should be a way to do it without killing the whole program
124 class BadConversion : public runtime_error {
126 BadConversion(const string& s) : runtime_error(s){ }
129 //**********************************************************************************************************************
132 inline void convert(const string& s, T& x, bool failIfLeftoverChars = true){
135 if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
136 throw BadConversion(s);
139 //**********************************************************************************************************************
142 inline bool convertTestFloat(const string& s, T& x, bool failIfLeftoverChars = true){
145 if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
152 //**********************************************************************************************************************
155 inline bool convertTest(const string& s, T& x, bool failIfLeftoverChars = true){
158 if (!(i >> x) || (failIfLeftoverChars && i.get(c)))
160 cout << "unable to be converted into an integer.\n" << endl;
166 //**********************************************************************************************************************
169 string toString(const T&x){
175 //**********************************************************************************************************************
178 string toHex(const T&x){
185 //**********************************************************************************************************************
188 string toString(const T&x, int i){
192 output << fixed << x;
196 /***********************************************************************/
197 inline int openOutputFileAppend(string fileName, ofstream& fileHandle){
199 fileHandle.open(fileName.c_str(), ios::app);
201 cout << "Error: Could not open " << fileName << endl;
209 /***********************************************************************/
211 inline void gobble(istream& f){
214 while(isspace(d=f.get())) {;}
218 /***********************************************************************/
220 inline string getline(ifstream& fileHandle) {
225 while (!fileHandle.eof()) {
227 char c = fileHandle.get();
229 //are you at the end of the line
230 if ((c == '\n') || (c == '\r') || (c == '\f')){ break; }
237 catch(exception& e) {
238 cout << "Error in mothur function getline" << endl;
243 /***********************************************************************/
245 inline bool isTrue(string f){
247 if ((f == "TRUE") || (f == "T") || (f == "true") || (f == "t")) { return true; }
248 else { return false; }
251 /***********************************************************************/
253 inline float roundDist(float dist, int precision){
255 return int(dist * precision + 0.5)/float(precision);
259 /***********************************************************************/
261 inline int getNumNames(string names){
267 for(int i=0;i<names.size();i++){
278 /**************************************************************************************************/
280 inline vector<vector<double> > binomial(int maxOrder){
282 vector<vector<double> > binomial(maxOrder+1);
284 for(int i=0;i<=maxOrder;i++){
285 binomial[i].resize(maxOrder+1);
294 for(int i=2;i<=maxOrder;i++){
298 for(int i=2;i<=maxOrder;i++){
299 for(int j=1;j<=maxOrder;j++){
300 if(i==j){ binomial[i][j]=1; }
301 if(j>i) { binomial[i][j]=0; }
302 else { binomial[i][j]=binomial[i-1][j-1]+binomial[i-1][j]; }
309 /***********************************************************************/
311 inline string getRootName(string longName){
313 string rootName = longName;
315 if(longName.find_last_of(".") != longName.npos){
316 int pos = longName.find_last_of('.')+1;
317 rootName = longName.substr(0, pos);
322 /***********************************************************************/
324 inline string getSimpleName(string longName){
326 string simpleName = longName;
329 found=longName.find_last_of("/\\");
331 if(found != longName.npos){
332 simpleName = longName.substr(found+1);
335 //if(longName.find_last_of("/") != longName.npos){
336 // int pos = longName.find_last_of('/')+1;
337 // simpleName = longName.substr(pos, longName.length());
343 /***********************************************************************/
345 inline int factorial(int num){
348 for (int i = 1; i <= num; i++) {
354 /**************************************************************************************************
356 double min(double x, double y)
362 /***********************************************************************/
364 inline string getPathName(string longName){
366 string rootPathName = longName;
368 if(longName.find_last_of("/\\") != longName.npos){
369 int pos = longName.find_last_of("/\\")+1;
370 rootPathName = longName.substr(0, pos);
375 /***********************************************************************/
377 inline string hasPath(string longName){
382 found=longName.find_last_of("~/\\");
384 if(found != longName.npos){
385 path = longName.substr(0, found+1);
391 /***********************************************************************/
393 inline string getExtension(string longName){
395 string extension = longName;
397 if(longName.find_last_of('.') != longName.npos){
398 int pos = longName.find_last_of('.');
399 extension = longName.substr(pos, longName.length());
404 /***********************************************************************/
405 inline bool isBlank(string fileName){
408 fileHandle.open(fileName.c_str());
410 cout << "Error: Could not open " << fileName << endl;
413 //check for blank file
415 if (fileHandle.eof()) { fileHandle.close(); return true; }
419 /***********************************************************************/
421 inline string getFullPathName(string fileName){
424 string path = hasPath(fileName);
428 if (path == "") { return fileName; } //its a simple name
429 else { //we need to complete the pathname
430 // ex. ../../../filename
431 // cwd = /user/work/desktop
434 //get current working directory
435 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
437 if (path.find("~") != -1) { //go to home directory
438 string homeDir = getenv ("HOME");
439 newFileName = homeDir + fileName.substr(fileName.find("~")+1);
442 if (path.rfind("./") == -1) { return fileName; } //already complete name
443 else { newFileName = fileName.substr(fileName.rfind("./")+2); } //save the complete part of the name
445 char* cwdpath = new char[1024];
448 cwdpath=getcwd(cwdpath,size);
454 if (cwd.length() > 0) { simpleCWD = cwd.substr(1); }
456 //break apart the current working directory
458 while (simpleCWD.find_first_of('/') != -1) {
459 string dir = simpleCWD.substr(0,simpleCWD.find_first_of('/'));
460 simpleCWD = simpleCWD.substr(simpleCWD.find_first_of('/')+1, simpleCWD.length());
463 //get last one // ex. ../../../filename = /user/work/desktop/filename
464 dirs.push_back(simpleCWD); //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
467 int index = dirs.size()-1;
469 while((pos = path.rfind("./")) != -1) { //while you don't have a complete path
470 if (pos == 0) { break; //you are at the end
471 }else if (path[(pos-1)] == '.') { //you want your parent directory ../
472 path = path.substr(0, pos-1);
474 if (index == 0) { break; }
475 }else if (path[(pos-1)] == '/') { //you want the current working dir ./
476 path = path.substr(0, pos);
477 }else if (pos == 1) { break; //you are at the end
478 }else { cout << "cannot resolve path for " << fileName << endl; return fileName; }
481 for (int i = index; i >= 0; i--) {
482 newFileName = dirs[i] + "/" + newFileName;
485 newFileName = "/" + newFileName;
489 if (path.find("~") != -1) { //go to home directory
490 string homeDir = getenv ("HOMEPATH");
491 newFileName = homeDir + fileName.substr(fileName.find("~")+1);
494 if (path.rfind(".\\") == -1) { return fileName; } //already complete name
495 else { newFileName = fileName.substr(fileName.rfind(".\\")+2); } //save the complete part of the name
497 char *cwdpath = NULL;
498 cwdpath = getcwd(NULL, 0); // or _getcwd
499 if ( cwdpath != NULL) { cwd = cwdpath; }
502 //break apart the current working directory
504 while (cwd.find_first_of('\\') != -1) {
505 string dir = cwd.substr(0,cwd.find_first_of('\\'));
506 cwd = cwd.substr(cwd.find_first_of('\\')+1, cwd.length());
511 dirs.push_back(cwd); //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
513 int index = dirs.size()-1;
515 while((pos = path.rfind(".\\")) != -1) { //while you don't have a complete path
516 if (pos == 0) { break; //you are at the end
517 }else if (path[(pos-1)] == '.') { //you want your parent directory ../
518 path = path.substr(0, pos-1);
520 if (index == 0) { break; }
521 }else if (path[(pos-1)] == '\\') { //you want the current working dir ./
522 path = path.substr(0, pos);
523 }else if (pos == 1) { break; //you are at the end
524 }else { cout << "cannot resolve path for " << fileName << endl; return fileName; }
527 for (int i = index; i >= 0; i--) {
528 newFileName = dirs[i] + "\\" + newFileName;
537 catch(exception& e) {
538 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function getFullPathName. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
542 /***********************************************************************/
544 inline int openInputFile(string fileName, ifstream& fileHandle, string m){
547 string completeFileName = getFullPathName(fileName);
549 fileHandle.open(completeFileName.c_str());
553 //check for blank file
558 /***********************************************************************/
560 inline int openInputFile(string fileName, ifstream& fileHandle){
563 string completeFileName = getFullPathName(fileName);
565 fileHandle.open(completeFileName.c_str());
567 cout << "Error: Could not open " << completeFileName << endl;
571 //check for blank file
573 if (fileHandle.eof()) { cout << completeFileName << " is blank. Please correct." << endl; return 1; }
579 /***********************************************************************/
581 inline int renameFile(string oldName, string newName){
584 int exist = openInputFile(newName, inTest, "");
586 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
587 if (exist == 0) { //you could open it so you want to delete it
589 string command = "rm " + newName;
590 system(command.c_str());
593 string command = "mv " + oldName + " " + newName;
594 system(command.c_str());
596 remove(newName.c_str());
597 int renameOk = rename(oldName.c_str(), newName.c_str());
602 /***********************************************************************/
604 inline int openOutputFile(string fileName, ofstream& fileHandle){
606 string completeFileName = getFullPathName(fileName);
608 fileHandle.open(completeFileName.c_str(), ios::trunc);
610 cout << "Error: Could not open " << completeFileName << endl;
619 /***********************************************************************/
621 inline int getNumSeqs(ifstream& file){
623 int numSeqs = count(istreambuf_iterator<char>(file),istreambuf_iterator<char>(), '>');
628 /***********************************************************************/
630 inline bool inVector(string member, vector<string> group){
632 for (int i = 0; i < group.size(); i++) {
633 if (group[i] == member) { return true; }
638 /***********************************************************************/
640 //This function parses the estimator options and puts them in a vector
641 inline void splitAtDash(string& estim, vector<string>& container) {
645 while (estim.find_first_of('-') != -1) {
646 individual = estim.substr(0,estim.find_first_of('-'));
647 if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
648 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
649 container.push_back(individual);
653 container.push_back(estim);
655 catch(exception& e) {
656 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
661 /***********************************************************************/
662 //This function parses the label options and puts them in a set
663 inline void splitAtDash(string& estim, set<string>& container) {
667 while (estim.find_first_of('-') != -1) {
668 individual = estim.substr(0,estim.find_first_of('-'));
669 if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
670 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
671 container.insert(individual);
675 container.insert(estim);
677 catch(exception& e) {
678 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
682 /***********************************************************************/
683 //This function parses the line options and puts them in a set
684 inline void splitAtDash(string& estim, set<int>& container) {
689 while (estim.find_first_of('-') != -1) {
690 individual = estim.substr(0,estim.find_first_of('-'));
691 if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
692 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
693 convert(individual, lineNum); //convert the string to int
694 container.insert(lineNum);
698 convert(estim, lineNum); //convert the string to int
699 container.insert(lineNum);
701 catch(exception& e) {
702 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
706 /***********************************************************************/
707 //This function parses the a string and puts peices in a vector
708 inline void splitAtComma(string& estim, vector<string>& container) {
712 while (estim.find_first_of(',') != -1) {
713 individual = estim.substr(0,estim.find_first_of(','));
714 if ((estim.find_first_of(',')+1) <= estim.length()) { //checks to make sure you don't have comma at end of string
715 estim = estim.substr(estim.find_first_of(',')+1, estim.length());
716 container.push_back(individual);
720 container.push_back(estim);
722 catch(exception& e) {
723 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtComma. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
727 /***********************************************************************/
729 //This function splits up the various option parameters
730 inline void splitAtComma(string& prefix, string& suffix){
732 prefix = suffix.substr(0,suffix.find_first_of(','));
733 if ((suffix.find_first_of(',')+2) <= suffix.length()) { //checks to make sure you don't have comma at end of string
734 suffix = suffix.substr(suffix.find_first_of(',')+1, suffix.length());
736 while(suffix.at(0) == ' ')
737 suffix = suffix.substr(1, suffix.length());
741 catch(exception& e) {
742 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtComma. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
746 /***********************************************************************/
748 //This function separates the key value from the option value i.e. dist=96_...
749 inline void splitAtEquals(string& key, string& value){
751 if(value.find_first_of('=') != -1){
752 key = value.substr(0,value.find_first_of('='));
753 if ((value.find_first_of('=')+1) <= value.length()) {
754 value = value.substr(value.find_first_of('=')+1, value.length());
761 catch(exception& e) {
762 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtEquals. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
766 /**************************************************************************************************/
768 inline bool inUsersGroups(string groupname, vector<string> Groups) {
770 for (int i = 0; i < Groups.size(); i++) {
771 if (groupname == Groups[i]) { return true; }
775 catch(exception& e) {
776 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function inUsersGroups. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
780 /**************************************************************************************************/
781 //returns true if any of the strings in first vector are in second vector
782 inline bool inUsersGroups(vector<string> groupnames, vector<string> Groups) {
785 for (int i = 0; i < groupnames.size(); i++) {
786 if (inUsersGroups(groupnames[i], Groups)) { return true; }
790 catch(exception& e) {
791 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function inUsersGroups. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
795 /***********************************************************************/
796 //this function determines if the user has given us labels that are smaller than the given label.
797 //if so then it returns true so that the calling function can run the previous valid distance.
798 //it's a "smart" distance function. It also checks for invalid labels.
799 inline bool anyLabelsToProcess(string label, set<string>& userLabels, string errorOff) {
802 set<string>::iterator it;
803 vector<float> orderFloat;
804 map<string, float> userMap; //the conversion process removes trailing 0's which we need to put back
805 map<string, float>::iterator it2;
807 bool smaller = false;
809 //unique is the smallest line
810 if (label == "unique") { return false; }
811 else { convert(label, labelFloat); }
813 //go through users set and make them floats
814 for(it = userLabels.begin(); it != userLabels.end(); ++it) {
817 if ((*it != "unique") && (convertTestFloat(*it, temp) == true)){
819 orderFloat.push_back(temp);
821 }else if (*it == "unique") {
822 orderFloat.push_back(-1.0);
823 userMap["unique"] = -1.0;
825 if (errorOff == "") { cout << *it << " is not a valid label." << endl; }
826 userLabels.erase(*it);
832 sort(orderFloat.begin(), orderFloat.end());
834 /*************************************************/
835 //is this label bigger than any of the users labels
836 /*************************************************/
838 //loop through order until you find a label greater than label
839 for (int i = 0; i < orderFloat.size(); i++) {
840 if (orderFloat[i] < labelFloat) {
842 if (orderFloat[i] == -1) {
843 if (errorOff == "") { cout << "Your file does not include the label unique." << endl; }
844 userLabels.erase("unique");
847 if (errorOff == "") { cout << "Your file does not include the label " << endl; }
849 for (it2 = userMap.begin(); it2!= userMap.end(); it2++) {
850 if (it2->second == orderFloat[i]) {
852 //remove small labels
857 if (errorOff == "") {cout << s << ". I will use the next smallest distance. " << endl; }
859 //since they are sorted once you find a bigger one stop looking
866 catch(exception& e) {
867 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function anyLabelsToProcess. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
872 /**************************************************************************************************/
873 inline void appendFiles(string temp, string filename) {
878 //open output file in append mode
879 openOutputFileAppend(filename, output);
880 int ableToOpen = openInputFile(temp, input, "no error");
882 if (ableToOpen == 0) { //you opened it
883 while(char c = input.get()){
884 if(input.eof()) { break; }
885 else { output << c; }
892 catch(exception& e) {
893 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function appendFiles. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
898 /**************************************************************************************************/
899 inline string sortFile(string distFile, string outputDir){
902 //if (outputDir == "") { outputDir += hasPath(distFile); }
903 string outfile = getRootName(distFile) + "sorted.dist";
906 //if you can, use the unix sort since its been optimized for years
907 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
908 string command = "sort -n -k +3 " + distFile + " -o " + outfile;
909 system(command.c_str());
910 #else //you are stuck with my best attempt...
911 //windows sort does not have a way to specify a column, only a character in the line
912 //since we cannot assume that the distance will always be at the the same character location on each line
913 //due to variable sequence name lengths, I chose to force the distance into first position, then sort and then put it back.
915 //read in file line by file and put distance first
916 string tempDistFile = distFile + ".temp";
919 openInputFile(distFile, input);
920 openOutputFile(tempDistFile, output);
922 string firstName, secondName;
925 input >> firstName >> secondName >> dist;
926 output << dist << '\t' << firstName << '\t' << secondName << endl;
933 //sort using windows sort
934 string tempOutfile = outfile + ".temp";
935 string command = "sort " + tempDistFile + " /O " + tempOutfile;
936 system(command.c_str());
938 //read in sorted file and put distance at end again
940 openInputFile(tempOutfile, input2);
941 openOutputFile(outfile, output);
944 input2 >> dist >> firstName >> secondName;
945 output << firstName << '\t' << secondName << '\t' << dist << endl;
952 remove(tempDistFile.c_str());
953 remove(tempOutfile.c_str());
958 catch(exception& e) {
959 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function sortfile. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
963 /**************************************************************************************************/
964 inline vector<long> setFilePosFasta(string filename, int& num) {
966 vector<long> positions;
968 openInputFile(filename, inFASTA);
971 while(!inFASTA.eof()){
972 input = getline(inFASTA); gobble(inFASTA);
973 if (input.length() != 0) {
974 if(input[0] == '>'){ long pos = inFASTA.tellg(); positions.push_back(pos - input.length() - 1); }
979 num = positions.size();
984 //get num bytes in file
985 pFile = fopen (filename.c_str(),"rb");
986 if (pFile==NULL) perror ("Error opening file");
988 fseek (pFile, 0, SEEK_END);
993 long size = positions[(positions.size()-1)];
995 openInputFile(filename, in);
999 while(char c = in.get()){
1000 if(in.eof()) { break; }
1005 positions.push_back(size);
1009 /**************************************************************************************************/
1010 inline vector<long> setFilePosEachLine(string filename, int& num) {
1012 vector<long> positions;
1014 openInputFile(filename, in);
1018 long lastpos = in.tellg();
1019 input = getline(in); gobble(in);
1020 if (input.length() != 0) {
1021 long pos = in.tellg();
1022 if (pos != -1) { positions.push_back(pos - input.length() - 1); }
1023 else { positions.push_back(lastpos); }
1028 num = positions.size();
1033 //get num bytes in file
1034 pFile = fopen (filename.c_str(),"rb");
1035 if (pFile==NULL) perror ("Error opening file");
1037 fseek (pFile, 0, SEEK_END);
1042 positions.push_back(size);
1047 /**************************************************************************************************/