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)
436 if (path.rfind("./") == -1) { return fileName; } //already complete name
437 else { newFileName = fileName.substr(fileName.rfind("./")+2); } //save the complete part of the name
439 char* cwdpath = new char[1024];
442 cwdpath=getcwd(cwdpath,size);
448 if (cwd.length() > 0) { simpleCWD = cwd.substr(1); }
450 //break apart the current working directory
452 while (simpleCWD.find_first_of('/') != -1) {
453 string dir = simpleCWD.substr(0,simpleCWD.find_first_of('/'));
454 simpleCWD = simpleCWD.substr(simpleCWD.find_first_of('/')+1, simpleCWD.length());
457 //get last one // ex. ../../../filename = /user/work/desktop/filename
458 dirs.push_back(simpleCWD); //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
461 int index = dirs.size()-1;
463 while((pos = path.rfind("./")) != -1) { //while you don't have a complete path
464 if (path[(pos-1)] == '.') { //you want your parent directory ../
465 path = path.substr(0, pos-1);
467 if (index == 0) { break; }
468 }else if (path[(pos-1)] == '/') { //you want the current working dir ./
469 path = path.substr(0, pos);
470 }else if (pos == 1) { break;
471 }else { cout << "cannot resolve path for " << fileName << endl; return fileName; }
474 for (int i = index; i >= 0; i--) {
475 newFileName = dirs[i] + "/" + newFileName;
478 newFileName = "/" + newFileName;
482 if (path.rfind(".\\") == -1) { return fileName; } //already complete name
483 else { newFileName = fileName.substr(fileName.rfind(".\\")+2); } //save the complete part of the name
485 char *cwdpath = NULL;
486 cwdpath = getcwd(NULL, 0); // or _getcwd
487 if ( cwdpath != NULL) { cwd = cwdpath; }
490 //break apart the current working directory
492 while (cwd.find_first_of('\\') != -1) {
493 string dir = cwd.substr(0,cwd.find_first_of('\\'));
494 cwd = cwd.substr(cwd.find_first_of('\\')+1, cwd.length());
499 dirs.push_back(cwd); //ex. dirs[0] = user, dirs[1] = work, dirs[2] = desktop
501 int index = dirs.size()-1;
503 while((pos = path.rfind(".\\")) != -1) { //while you don't have a complete path
504 if (path[(pos-1)] == '.') { //you want your parent directory ../
505 path = path.substr(0, pos-1);
507 if (index == 0) { break; }
508 }else if (path[(pos-1)] == '\\') { //you want the current working dir ./
509 path = path.substr(0, pos);
510 }else if (pos == 1) { break;
511 }else { cout << "cannot resolve path for " << fileName << endl; return fileName; }
514 for (int i = index; i >= 0; i--) {
515 newFileName = dirs[i] + "\\" + newFileName;
523 catch(exception& e) {
524 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function getFullPathName. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
528 /***********************************************************************/
530 inline int openInputFile(string fileName, ifstream& fileHandle, string m){
533 string completeFileName = getFullPathName(fileName);
535 fileHandle.open(completeFileName.c_str());
539 //check for blank file
544 /***********************************************************************/
546 inline int openInputFile(string fileName, ifstream& fileHandle){
549 string completeFileName = getFullPathName(fileName);
551 fileHandle.open(completeFileName.c_str());
553 cout << "Error: Could not open " << completeFileName << endl;
557 //check for blank file
559 if (fileHandle.eof()) { cout << completeFileName << " is blank. Please correct." << endl; return 1; }
565 /***********************************************************************/
567 inline int renameFile(string oldName, string newName){
570 int exist = openInputFile(newName, inTest, "");
572 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
573 if (exist == 0) { //you could open it so you want to delete it
575 string command = "rm " + newName;
576 system(command.c_str());
579 string command = "mv " + oldName + " " + newName;
580 system(command.c_str());
582 remove(newName.c_str());
583 int renameOk = rename(oldName.c_str(), newName.c_str());
588 /***********************************************************************/
590 inline int openOutputFile(string fileName, ofstream& fileHandle){
592 string completeFileName = getFullPathName(fileName);
594 fileHandle.open(completeFileName.c_str(), ios::trunc);
596 cout << "Error: Could not open " << completeFileName << endl;
605 /***********************************************************************/
607 inline int getNumSeqs(ifstream& file){
609 int numSeqs = count(istreambuf_iterator<char>(file),istreambuf_iterator<char>(), '>');
614 /***********************************************************************/
616 inline bool inVector(string member, vector<string> group){
618 for (int i = 0; i < group.size(); i++) {
619 if (group[i] == member) { return true; }
624 /***********************************************************************/
626 //This function parses the estimator options and puts them in a vector
627 inline void splitAtDash(string& estim, vector<string>& container) {
631 while (estim.find_first_of('-') != -1) {
632 individual = estim.substr(0,estim.find_first_of('-'));
633 if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
634 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
635 container.push_back(individual);
639 container.push_back(estim);
641 catch(exception& e) {
642 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
647 /***********************************************************************/
648 //This function parses the label options and puts them in a set
649 inline void splitAtDash(string& estim, set<string>& container) {
653 while (estim.find_first_of('-') != -1) {
654 individual = estim.substr(0,estim.find_first_of('-'));
655 if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
656 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
657 container.insert(individual);
661 container.insert(estim);
663 catch(exception& e) {
664 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
668 /***********************************************************************/
669 //This function parses the line options and puts them in a set
670 inline void splitAtDash(string& estim, set<int>& container) {
675 while (estim.find_first_of('-') != -1) {
676 individual = estim.substr(0,estim.find_first_of('-'));
677 if ((estim.find_first_of('-')+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
678 estim = estim.substr(estim.find_first_of('-')+1, estim.length());
679 convert(individual, lineNum); //convert the string to int
680 container.insert(lineNum);
684 convert(estim, lineNum); //convert the string to int
685 container.insert(lineNum);
687 catch(exception& e) {
688 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtDash. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
692 /***********************************************************************/
693 //This function parses the a string and puts peices in a vector
694 inline void splitAtComma(string& estim, vector<string>& container) {
698 while (estim.find_first_of(',') != -1) {
699 individual = estim.substr(0,estim.find_first_of(','));
700 if ((estim.find_first_of(',')+1) <= estim.length()) { //checks to make sure you don't have comma at end of string
701 estim = estim.substr(estim.find_first_of(',')+1, estim.length());
702 container.push_back(individual);
706 container.push_back(estim);
708 catch(exception& e) {
709 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtComma. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
713 /***********************************************************************/
715 //This function splits up the various option parameters
716 inline void splitAtComma(string& prefix, string& suffix){
718 prefix = suffix.substr(0,suffix.find_first_of(','));
719 if ((suffix.find_first_of(',')+2) <= suffix.length()) { //checks to make sure you don't have comma at end of string
720 suffix = suffix.substr(suffix.find_first_of(',')+1, suffix.length());
722 while(suffix.at(0) == ' ')
723 suffix = suffix.substr(1, suffix.length());
727 catch(exception& e) {
728 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtComma. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
732 /***********************************************************************/
734 //This function separates the key value from the option value i.e. dist=96_...
735 inline void splitAtEquals(string& key, string& value){
737 if(value.find_first_of('=') != -1){
738 key = value.substr(0,value.find_first_of('='));
739 if ((value.find_first_of('=')+1) <= value.length()) {
740 value = value.substr(value.find_first_of('=')+1, value.length());
747 catch(exception& e) {
748 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function splitAtEquals. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
752 /**************************************************************************************************/
754 inline bool inUsersGroups(string groupname, vector<string> Groups) {
756 for (int i = 0; i < Groups.size(); i++) {
757 if (groupname == Groups[i]) { return true; }
761 catch(exception& e) {
762 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function inUsersGroups. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
766 /**************************************************************************************************/
767 //returns true if any of the strings in first vector are in second vector
768 inline bool inUsersGroups(vector<string> groupnames, vector<string> Groups) {
771 for (int i = 0; i < groupnames.size(); i++) {
772 if (inUsersGroups(groupnames[i], Groups)) { return true; }
776 catch(exception& e) {
777 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function inUsersGroups. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
781 /***********************************************************************/
782 //this function determines if the user has given us labels that are smaller than the given label.
783 //if so then it returns true so that the calling function can run the previous valid distance.
784 //it's a "smart" distance function. It also checks for invalid labels.
785 inline bool anyLabelsToProcess(string label, set<string>& userLabels, string errorOff) {
788 set<string>::iterator it;
789 vector<float> orderFloat;
790 map<string, float> userMap; //the conversion process removes trailing 0's which we need to put back
791 map<string, float>::iterator it2;
793 bool smaller = false;
795 //unique is the smallest line
796 if (label == "unique") { return false; }
797 else { convert(label, labelFloat); }
799 //go through users set and make them floats
800 for(it = userLabels.begin(); it != userLabels.end(); ++it) {
803 if ((*it != "unique") && (convertTestFloat(*it, temp) == true)){
805 orderFloat.push_back(temp);
807 }else if (*it == "unique") {
808 orderFloat.push_back(-1.0);
809 userMap["unique"] = -1.0;
811 if (errorOff == "") { cout << *it << " is not a valid label." << endl; }
812 userLabels.erase(*it);
818 sort(orderFloat.begin(), orderFloat.end());
820 /*************************************************/
821 //is this label bigger than any of the users labels
822 /*************************************************/
824 //loop through order until you find a label greater than label
825 for (int i = 0; i < orderFloat.size(); i++) {
826 if (orderFloat[i] < labelFloat) {
828 if (orderFloat[i] == -1) {
829 if (errorOff == "") { cout << "Your file does not include the label unique." << endl; }
830 userLabels.erase("unique");
833 if (errorOff == "") { cout << "Your file does not include the label " << endl; }
835 for (it2 = userMap.begin(); it2!= userMap.end(); it2++) {
836 if (it2->second == orderFloat[i]) {
838 //remove small labels
843 if (errorOff == "") {cout << s << ". I will use the next smallest distance. " << endl; }
845 //since they are sorted once you find a bigger one stop looking
852 catch(exception& e) {
853 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function anyLabelsToProcess. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
858 /**************************************************************************************************/
859 inline void appendFiles(string temp, string filename) {
864 //open output file in append mode
865 openOutputFileAppend(filename, output);
866 openInputFile(temp, input);
868 while(char c = input.get()){
869 if(input.eof()) { break; }
870 else { output << c; }
876 catch(exception& e) {
877 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function appendFiles. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
882 /**************************************************************************************************/
883 inline string sortFile(string distFile){
885 string outfile = getRootName(distFile) + "sorted.dist";
887 //if you can, use the unix sort since its been optimized for years
888 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
889 string command = "sort -n -k +3 " + distFile + " -o " + outfile;
890 system(command.c_str());
891 #else //you are stuck with my best attempt...
892 //windows sort does not have a way to specify a column, only a character in the line
893 //since we cannot assume that the distance will always be at the the same character location on each line
894 //due to variable sequence name lengths, I chose to force the distance into first position, then sort and then put it back.
896 //read in file line by file and put distance first
897 string tempDistFile = distFile + ".temp";
900 openInputFile(distFile, input);
901 openOutputFile(tempDistFile, output);
903 string firstName, secondName;
906 input >> firstName >> secondName >> dist;
907 output << dist << '\t' << firstName << '\t' << secondName << endl;
914 //sort using windows sort
915 string tempOutfile = outfile + ".temp";
916 string command = "sort " + tempDistFile + " /O " + tempOutfile;
917 system(command.c_str());
919 //read in sorted file and put distance at end again
921 openInputFile(tempOutfile, input2);
922 openOutputFile(outfile, output);
925 input2 >> dist >> firstName >> secondName;
926 output << firstName << '\t' << secondName << '\t' << dist << endl;
933 remove(tempDistFile.c_str());
934 remove(tempOutfile.c_str());
939 catch(exception& e) {
940 cout << "Standard Error: " << e.what() << " has occurred in the mothur.h function sortfile. Please contact Pat Schloss at mothur.bugs@gmail.com." << "\n";
944 /**************************************************************************************************/