precluster command finished

[mothur.git] / mothur.h

diff --git a/mothur.h b/mothur.h
index f839328cd2c304330fb18f31fb97eea531ef625a..cf6ae69dc85fc866c34588118575222772897dcf 100644 (file)
--- a/mothur.h
+++ b/mothur.h
@@ -100,7 +100,15 @@ struct clusterNode {
        int smallChild; //used to make linkTable work with list and rabund. represents bin number of this cluster node
        clusterNode(int num, int par, int kid) : numSeq(num), parent(par), smallChild(kid) {};
 };
-
+/************************************************************/
+struct seqDist {
+       int seq1;
+       int seq2;
+       float dist;
+       seqDist() {}
+       seqDist(int s1, int s2, float d) : seq1(s1), seq2(s2), dist(d) {}
+       ~seqDist() {}
+};
 /***********************************************************************/
 
 // snagged from http://www.parashift.com/c++-faq-lite/misc-technical-issues.html#faq-39.2
@@ -775,6 +783,92 @@ inline bool anyLabelsToProcess(string label, set<string>& userLabels, string err
        }
 }
 
+/**************************************************************************************************/
+inline void appendFiles(string temp, string filename) {
+       try{
+               ofstream output;
+               ifstream input;
+       
+               //open output file in append mode
+               openOutputFileAppend(filename, output);
+               openInputFile(temp, input);
+               
+               while(char c = input.get()){
+                       if(input.eof())         {       break;                  }
+                       else                            {       output << c;    }
+               }
+               
+               input.close();
+               output.close();
+       }
+       catch(exception& e) {
+               errorOut(e, "mothur", "appendFiles");
+               exit(1);
+       }
+}
+
+/**************************************************************************************************/
+inline string sortFile(string distFile){
+       try {   
+               string outfile = getRootName(distFile) + "sorted.dist";
+               
+               //if you can, use the unix sort since its been optimized for years
+               #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+                       string command = "sort -n -k +3 " + distFile + " -o " + outfile;
+                       system(command.c_str());
+               #else //you are stuck with my best attempt...
+                       //windows sort does not have a way to specify a column, only a character in the line
+                       //since we cannot assume that the distance will always be at the the same character location on each line
+                       //due to variable sequence name lengths, I chose to force the distance into first position, then sort and then put it back.
+               
+                       //read in file line by file and put distance first
+                       string tempDistFile = distFile + ".temp";
+                       ifstream input;
+                       ofstream output;
+                       openInputFile(distFile, input);
+                       openOutputFile(tempDistFile, output);
+
+                       string firstName, secondName;
+                       float dist;
+                       while (input) {
+                               input >> firstName >> secondName >> dist;
+                               output << dist << '\t' << firstName << '\t' << secondName << endl;
+                               gobble(input);
+                       }
+                       input.close();
+                       output.close();
+               
+       
+                       //sort using windows sort
+                       string tempOutfile = outfile + ".temp";
+                       string command = "sort " + tempDistFile + " /O " + tempOutfile;
+                       system(command.c_str());
+               
+                       //read in sorted file and put distance at end again
+                       ifstream input2;
+                       openInputFile(tempOutfile, input2);
+                       openOutputFile(outfile, output);
+               
+                       while (input2) {
+                               input2 >> dist >> firstName >> secondName;
+                               output << firstName << '\t' << secondName << '\t' << dist << endl;
+                               gobble(input2);
+                       }
+                       input2.close();
+                       output.close();
+               
+                       //remove temp files
+                       remove(tempDistFile.c_str());
+                       remove(tempOutfile.c_str());
+               #endif
+               
+               return outfile;
+       }
+       catch(exception& e) {
+               errorOut(e, "mothur", "sortFile");
+               exit(1);
+       }
+}
 /**************************************************************************************************/
 #endif