]> git.donarmstrong.com Git - mothur.git/blobdiff - mothurout.cpp
fixed bug with trim.seqs- when a file is blank for a grouping mothur removed it,...
[mothur.git] / mothurout.cpp
index f6b2a1d6312421da8405626b7aea19e71cd2b985..54cf25d891e4528b1a6c1f08bc350a58b8aeb9e6 100644 (file)
@@ -9,6 +9,7 @@
 
 #include "mothurout.h"
 
+
 /******************************************************/
 MothurOut* MothurOut::getInstance() {
        if( _uniqueInstance == 0) {
@@ -60,6 +61,16 @@ void MothurOut::setDefaultPath(string pathname)  {
        }
 }
 /*********************************************************************************************/
+void MothurOut::setOutputDir(string pathname)  {
+       try {
+               outputDir = pathname;
+       }
+       catch(exception& e) {
+               errorOut(e, "MothurOut", "setOutputDir");
+               exit(1);
+       }
+}
+/*********************************************************************************************/
 void MothurOut::closeLog()  {
        try {
                
@@ -276,8 +287,9 @@ int MothurOut::openOutputFileAppend(string fileName, ofstream& fileHandle){
 /***********************************************************************/
 void MothurOut::gobble(istream& f){
        try {
+               
                char d;
-               while(isspace(d=f.get()))               {;}
+               while(isspace(d=f.get()))               { ;}
                f.putback(d);
        }
        catch(exception& e) {
@@ -348,14 +360,32 @@ string MothurOut::getline(ifstream& fileHandle) {
 }
 /***********************************************************************/
 
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#ifdef USE_COMPRESSION
+inline bool endsWith(string s, const char * suffix){
+  size_t suffixLength = strlen(suffix);
+  return s.size() >= suffixLength && s.substr(s.size() - suffixLength, suffixLength).compare(suffix) == 0;
+}
+#endif
+#endif
+
 string MothurOut::getRootName(string longName){
        try {
        
                string rootName = longName;
-               
-               if(longName.find_last_of(".") != longName.npos){
-                       int pos = longName.find_last_of('.')+1;
-                       rootName = longName.substr(0, pos);
+
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#ifdef USE_COMPRESSION
+    if (endsWith(rootName, ".gz") || endsWith(rootName, ".bz2")) {
+      int pos = rootName.find_last_of('.');
+      rootName = rootName.substr(0, pos);
+      cerr << "shortening " << longName << " to " << rootName << "\n";
+    }
+#endif
+#endif
+               if(rootName.find_last_of(".") != rootName.npos){
+                       int pos = rootName.find_last_of('.')+1;
+                       rootName = rootName.substr(0, pos);
                }
 
                return rootName;
@@ -459,6 +489,7 @@ bool MothurOut::isBlank(string fileName){
                        //check for blank file
                        gobble(fileHandle);
                        if (fileHandle.eof()) { fileHandle.close(); return true;  }
+                       fileHandle.close();
                }
                return false;
        }
@@ -486,15 +517,20 @@ string MothurOut::getFullPathName(string fileName){
                #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)   
                        
                        if (path.find("~") != -1) { //go to home directory
-                               string homeDir = getenv ("HOME");
+                               string homeDir;
+                       
+                               char *homepath = NULL;
+                               homepath = getenv ("HOME");
+                               if ( homepath != NULL) { homeDir = homepath; }
+                               else { homeDir = "";  }
+
                                newFileName = homeDir + fileName.substr(fileName.find("~")+1);
                                return newFileName;
                        }else { //find path
-                               if (path.rfind("./") == -1) { return fileName; } //already complete name
+                               if (path.rfind("./") == string::npos) { return fileName; } //already complete name
                                else { newFileName = fileName.substr(fileName.rfind("./")+2); } //save the complete part of the name
                                
                                //char* cwdpath = new char[1024];
-
                                //size_t size;
                                //cwdpath=getcwd(cwdpath,size);
                                //cwd = cwdpath;
@@ -503,6 +539,7 @@ string MothurOut::getFullPathName(string fileName){
                                cwdpath = getcwd(NULL, 0); // or _getcwd
                                if ( cwdpath != NULL) { cwd = cwdpath; }
                                else { cwd = "";  }
+
                                
                                //rip off first '/'
                                string simpleCWD;
@@ -510,7 +547,7 @@ string MothurOut::getFullPathName(string fileName){
                                
                                //break apart the current working directory
                                vector<string> dirs;
-                               while (simpleCWD.find_first_of('/') != -1) {
+                               while (simpleCWD.find_first_of('/') != string::npos) {
                                        string dir = simpleCWD.substr(0,simpleCWD.find_first_of('/'));
                                        simpleCWD = simpleCWD.substr(simpleCWD.find_first_of('/')+1, simpleCWD.length());
                                        dirs.push_back(dir);
@@ -521,7 +558,7 @@ string MothurOut::getFullPathName(string fileName){
                        
                                int index = dirs.size()-1;
                
-                               while((pos = path.rfind("./")) != -1) { //while you don't have a complete path
+                               while((pos = path.rfind("./")) != string::npos) { //while you don't have a complete path
                                        if (pos == 0) { break;  //you are at the end
                                        }else if (path[(pos-1)] == '.') { //you want your parent directory ../
                                                path = path.substr(0, pos-1);
@@ -539,15 +576,14 @@ string MothurOut::getFullPathName(string fileName){
                                
                                newFileName =  "/" +  newFileName;
                                return newFileName;
-               
                        }       
                #else
-                       if (path.find("~") != -1) { //go to home directory
+                       if (path.find("~") != string::npos) { //go to home directory
                                string homeDir = getenv ("HOMEPATH");
                                newFileName = homeDir + fileName.substr(fileName.find("~")+1);
                                return newFileName;
                        }else { //find path
-                               if (path.rfind(".\\") == -1) { return fileName; } //already complete name
+                               if (path.rfind(".\\") == string::npos) { return fileName; } //already complete name
                                else { newFileName = fileName.substr(fileName.rfind(".\\")+2); } //save the complete part of the name
                                                        
                                char *cwdpath = NULL;
@@ -568,7 +604,7 @@ string MothurOut::getFullPathName(string fileName){
                                        
                                int index = dirs.size()-1;
                                        
-                               while((pos = path.rfind(".\\")) != -1) { //while you don't have a complete path
+                               while((pos = path.rfind(".\\")) != string::npos) { //while you don't have a complete path
                                        if (pos == 0) { break;  //you are at the end
                                        }else if (path[(pos-1)] == '.') { //you want your parent directory ../
                                                path = path.substr(0, pos-1);
@@ -596,14 +632,38 @@ string MothurOut::getFullPathName(string fileName){
        }       
 }
 /***********************************************************************/
-//no error open
+
 int MothurOut::openInputFile(string fileName, ifstream& fileHandle, string m){
        try {
                        //get full path name
                        string completeFileName = getFullPathName(fileName);
-
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#ifdef USE_COMPRESSION
+      // check for gzipped or bzipped file
+      if (endsWith(completeFileName, ".gz") || endsWith(completeFileName, ".bz2")) {
+        string tempName = string(tmpnam(0));
+        mkfifo(tempName.c_str(), 0666);
+        int fork_result = fork();
+        if (fork_result < 0) {
+          cerr << "Error forking.\n";
+          exit(1);
+        } else if (fork_result == 0) {
+          string command = (endsWith(completeFileName, ".gz") ? "zcat " : "bzcat ") + completeFileName + string(" > ") + tempName;
+          cerr << "Decompressing " << completeFileName << " via temporary named pipe " << tempName << "\n";
+          system(command.c_str());
+          cerr << "Done decompressing " << completeFileName << "\n";
+          remove(tempName.c_str());
+          exit(EXIT_SUCCESS);
+        } else {
+          cerr << "waiting on child process " << fork_result << "\n";
+          completeFileName = tempName;
+        }
+      }
+#endif
+#endif
                        fileHandle.open(completeFileName.c_str());
                        if(!fileHandle) {
+                               //mothurOut("[ERROR]: Could not open " + completeFileName); mothurOutEndLine();
                                return 1;
                        }else {
                                //check for blank file
@@ -620,8 +680,33 @@ int MothurOut::openInputFile(string fileName, ifstream& fileHandle, string m){
 
 int MothurOut::openInputFile(string fileName, ifstream& fileHandle){
        try {
+
                //get full path name
                string completeFileName = getFullPathName(fileName);
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#ifdef USE_COMPRESSION
+  // check for gzipped or bzipped file
+  if (endsWith(completeFileName, ".gz") || endsWith(completeFileName, ".bz2")) {
+    string tempName = string(tmpnam(0));
+    mkfifo(tempName.c_str(), 0666);
+    int fork_result = fork();
+    if (fork_result < 0) {
+      cerr << "Error forking.\n";
+      exit(1);
+    } else if (fork_result == 0) {
+      string command = (endsWith(completeFileName, ".gz") ? "zcat " : "bzcat ") + completeFileName + string(" > ") + tempName;
+      cerr << "Decompressing " << completeFileName << " via temporary named pipe " << tempName << "\n";
+      system(command.c_str());
+      cerr << "Done decompressing " << completeFileName << "\n";
+      remove(tempName.c_str());
+      exit(EXIT_SUCCESS);
+    } else {
+      cerr << "waiting on child process " << fork_result << "\n";
+      completeFileName = tempName;
+    }
+  }
+#endif
+#endif
 
                fileHandle.open(completeFileName.c_str());
                if(!fileHandle) {
@@ -676,7 +761,27 @@ int MothurOut::openOutputFile(string fileName, ofstream& fileHandle){
        try { 
        
                string completeFileName = getFullPathName(fileName);
-               
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#ifdef USE_COMPRESSION
+    // check for gzipped file
+    if (endsWith(completeFileName, ".gz") || endsWith(completeFileName, ".bz2")) {
+      string tempName = string(tmpnam(0));
+      mkfifo(tempName.c_str(), 0666);
+      cerr << "Compressing " << completeFileName << " via temporary named pipe " << tempName << "\n";
+      int fork_result = fork();
+      if (fork_result < 0) {
+        cerr << "Error forking.\n";
+        exit(1);
+      } else if (fork_result == 0) {
+        string command = string(endsWith(completeFileName, ".gz") ?  "gzip" : "bzip2") + " -v > " + completeFileName + string(" < ") + tempName;
+        system(command.c_str());
+        exit(0);
+      } else {
+        completeFileName = tempName;
+      }
+    }
+#endif
+#endif
                fileHandle.open(completeFileName.c_str(), ios::trunc);
                if(!fileHandle) {
                        mothurOut("[ERROR]: Could not open " + completeFileName); mothurOutEndLine();
@@ -702,6 +807,7 @@ void MothurOut::appendFiles(string temp, string filename) {
                //open output file in append mode
                openOutputFileAppend(filename, output);
                int ableToOpen = openInputFile(temp, input, "no error");
+               //int ableToOpen = openInputFile(temp, input);
                
                if (ableToOpen == 0) { //you opened it
                        while(char c = input.get()){
@@ -906,7 +1012,7 @@ vector<unsigned long int> MothurOut::divideFile(string filename, int& proc) {
                //estimate file breaks
                unsigned long int chunkSize = 0;
                chunkSize = size / proc;
-       
+
                //file to small to divide by processors
                if (chunkSize == 0)  {  proc = 1;       filePos.push_back(size); return filePos;        }
        
@@ -924,7 +1030,7 @@ vector<unsigned long int> MothurOut::divideFile(string filename, int& proc) {
                           char c = in.get();
                           if (c == '>') {   in.putback(c); newSpot = in.tellg(); break;  }
                        }
-                       
+               
                        //there was not another sequence before the end of the file
                        unsigned long int sanityPos = in.tellg();
 
@@ -1050,7 +1156,90 @@ vector<vector<double> > MothurOut::binomial(int maxOrder){
                exit(1);
        }
 }
-
+/**************************************************************************************************/
+unsigned int MothurOut::fromBase36(string base36){
+       try {
+               unsigned int num = 0;
+               
+               map<char, int> converts;
+               converts['A'] = 0;
+               converts['a'] = 0;
+               converts['B'] = 1;
+               converts['b'] = 1;
+               converts['C'] = 2;
+               converts['c'] = 2;
+               converts['D'] = 3;
+               converts['d'] = 3;
+               converts['E'] = 4;
+               converts['e'] = 4;
+               converts['F'] = 5;
+               converts['f'] = 5;
+               converts['G'] = 6;
+               converts['g'] = 6;
+               converts['H'] = 7;
+               converts['h'] = 7;
+               converts['I'] = 8;
+               converts['i'] = 8;
+               converts['J'] = 9;
+               converts['j'] = 9;
+               converts['K'] = 10;
+               converts['k'] = 10;
+               converts['L'] = 11;
+               converts['l'] = 11;
+               converts['M'] = 12;
+               converts['m'] = 12;
+               converts['N'] = 13;
+               converts['n'] = 13;
+               converts['O'] = 14;
+               converts['o'] = 14;
+               converts['P'] = 15;
+               converts['p'] = 15;
+               converts['Q'] = 16;
+               converts['q'] = 16;
+               converts['R'] = 17;
+               converts['r'] = 17;
+               converts['S'] = 18;
+               converts['s'] = 18;
+               converts['T'] = 19;
+               converts['t'] = 19;
+               converts['U'] = 20;
+               converts['u'] = 20;
+               converts['V'] = 21;
+               converts['v'] = 21;
+               converts['W'] = 22;
+               converts['w'] = 22;
+               converts['X'] = 23;
+               converts['x'] = 23;
+               converts['Y'] = 24;
+               converts['y'] = 24;
+               converts['Z'] = 25;
+               converts['z'] = 25;
+               converts['0'] = 26;
+               converts['1'] = 27;
+               converts['2'] = 28;
+               converts['3'] = 29;
+               converts['4'] = 30;
+               converts['5'] = 31;
+               converts['6'] = 32;
+               converts['7'] = 33;
+               converts['8'] = 34;
+               converts['9'] = 35;             
+               
+               int i = 0;
+               while (i < base36.length()) {
+                       char c = base36[i];
+                       num = 36 * num + converts[c];
+                       i++;
+               }
+               
+               return num;
+               
+       }
+       catch(exception& e) {
+               errorOut(e, "MothurOut", "fromBase36");
+               exit(1);
+       }
+}
 /***********************************************************************/
 
 int MothurOut::factorial(int num){
@@ -1103,17 +1292,19 @@ void MothurOut::getNumSeqs(ifstream& file, int& numSeqs){
 //This function parses the estimator options and puts them in a vector
 void MothurOut::splitAtChar(string& estim, vector<string>& container, char symbol) {
        try {
-               string individual;
-               
-               while (estim.find_first_of(symbol) != -1) {
-                       individual = estim.substr(0,estim.find_first_of(symbol));
-                       if ((estim.find_first_of(symbol)+1) <= estim.length()) { //checks to make sure you don't have dash at end of string
-                               estim = estim.substr(estim.find_first_of(symbol)+1, estim.length());
+               string individual = "";
+               int estimLength = estim.size();
+               for(int i=0;i<estimLength;i++){
+                       if(estim[i] == symbol){
                                container.push_back(individual);
+                               individual = "";                                
+                       }
+                       else{
+                               individual += estim[i];
                        }
                }
-               //get last one
-               container.push_back(estim);
+               container.push_back(individual);
+
        }
        catch(exception& e) {
                errorOut(e, "MothurOut", "splitAtChar");
@@ -1348,20 +1539,21 @@ bool MothurOut::anyLabelsToProcess(string label, set<string>& userLabels, string
                }
                
                //go through users set and make them floats
-               for(it = userLabels.begin(); it != userLabels.end(); ++it) {
+               for(it = userLabels.begin(); it != userLabels.end();) {
                        
                        float temp;
                        if ((*it != "unique") && (convertTestFloat(*it, temp) == true)){
                                convert(*it, temp);
                                orderFloat.push_back(temp);
                                userMap[*it] = temp;
+                               it++;
                        }else if (*it == "unique") { 
                                orderFloat.push_back(-1.0);
                                userMap["unique"] = -1.0;
+                               it++;
                        }else {
-                               if (errorOff == "") {  cout << *it << " is not a valid label." << endl;  }
-                               userLabels.erase(*it); 
-                               it--;
+                               if (errorOff == "") {  mothurOut(*it + " is not a valid label."); mothurOutEndLine();  }
+                               userLabels.erase(it++); 
                        }
                }
                
@@ -1377,11 +1569,11 @@ bool MothurOut::anyLabelsToProcess(string label, set<string>& userLabels, string
                        if (orderFloat[i] < labelFloat) {
                                smaller = true;
                                if (orderFloat[i] == -1) { 
-                                       if (errorOff == "") { cout << "Your file does not include the label unique." << endl; }
+                                       if (errorOff == "") { mothurOut("Your file does not include the label unique."); mothurOutEndLine(); }
                                        userLabels.erase("unique");
                                }
                                else {  
-                                       if (errorOff == "") { cout << "Your file does not include the label " << endl; }
+                                       if (errorOff == "") { mothurOut("Your file does not include the label "); mothurOutEndLine(); }
                                        string s = "";
                                        for (it2 = userMap.begin(); it2!= userMap.end(); it2++) {  
                                                if (it2->second == orderFloat[i]) {  
@@ -1391,7 +1583,7 @@ bool MothurOut::anyLabelsToProcess(string label, set<string>& userLabels, string
                                                        break;
                                                }
                                        }
-                                       if (errorOff == "") {cout << s <<  ". I will use the next smallest distance. " << endl; }
+                                       if (errorOff == "") {mothurOut( s +  ". I will use the next smallest distance. "); mothurOutEndLine(); }
                                }
                        //since they are sorted once you find a bigger one stop looking
                        }else { break; }
@@ -1451,6 +1643,7 @@ bool MothurOut::checkReleaseVersion(ifstream& file, string version) {
                exit(1);
        }
 }
+
 /**************************************************************************************************/