changed reading of name file to use buffered reads. note the splitAtWhiteSpace functi...

[mothur.git] / screenseqscommand.cpp

diff --git a/screenseqscommand.cpp b/screenseqscommand.cpp
index 9494865c4f66e302182b8fd221e0b44951cb2518..7ac910d4a74b9fc1bffeacb32f5893941f3eb60f 100644 (file)
--- a/screenseqscommand.cpp
+++ b/screenseqscommand.cpp
@@ -288,7 +288,7 @@ int ScreenSeqsCommand::execute(){
                        getSummary(positions); 
                } 
                else { 
-                       #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+                       #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                 positions = m->divideFile(fastafile, processors);
                 for (int i = 0; i < (positions.size()-1); i++) { lines.push_back(linePair(positions[i], positions[(i+1)])); }
                        #else 
@@ -296,6 +296,7 @@ int ScreenSeqsCommand::execute(){
                 else {
                     int numFastaSeqs = 0;
                     positions = m->setFilePosFasta(fastafile, numFastaSeqs); 
+                    if (positions.size() < processors) { processors = positions.size(); }
                 
                     //figure out how many sequences you have to process
                     int numSeqsPerProcessor = numFastaSeqs / processors;
@@ -398,17 +399,10 @@ int ScreenSeqsCommand::execute(){
                        MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
                                        
 #else
-                                               
-       //#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
-                       if(processors == 1){
-                               numFastaSeqs = driver(lines[0], goodSeqFile, badAccnosFile, fastafile, badSeqNames);            
-                       }else{
-                               numFastaSeqs = createProcesses(goodSeqFile, badAccnosFile, fastafile, badSeqNames); 
-            }
-       //#else
-       //              numFastaSeqs = driver(lines[0], goodSeqFile, badAccnosFile, fastafile, badSeqNames);            
-       //#endif
-            if (m->control_pressed) { m->mothurRemove(goodSeqFile); return 0; }
+        if(processors == 1){ numFastaSeqs = driver(lines[0], goodSeqFile, badAccnosFile, fastafile, badSeqNames);      }       
+        else{ numFastaSeqs = createProcesses(goodSeqFile, badAccnosFile, fastafile, badSeqNames); }
+        
+        if (m->control_pressed) { m->mothurRemove(goodSeqFile); return 0; }
 #endif         
 
                #ifdef USE_MPI
@@ -618,7 +612,7 @@ int ScreenSeqsCommand::getSummary(vector<unsigned long long>& positions){
                vector<int> longHomoPolymer;
                
         vector<unsigned long long> positions;
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                positions = m->divideFile(fastafile, processors);
                for (int i = 0; i < (positions.size()-1); i++) { lines.push_back(linePair(positions[i], positions[(i+1)])); }   
 #else
@@ -626,6 +620,7 @@ int ScreenSeqsCommand::getSummary(vector<unsigned long long>& positions){
         else {
             int numFastaSeqs = 0;
             positions = m->setFilePosFasta(fastafile, numFastaSeqs); 
+            if (positions.size() < processors) { processors = positions.size(); }
             
             //figure out how many sequences you have to process
             int numSeqsPerProcessor = numFastaSeqs / processors;
@@ -645,7 +640,7 @@ int ScreenSeqsCommand::getSummary(vector<unsigned long long>& positions){
                        driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[0]);
 #else
                int numSeqs = 0;
-               //#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+               //#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                        if(processors == 1){
                                numSeqs = driverCreateSummary(startPosition, endPosition, seqLength, ambigBases, longHomoPolymer, fastafile, lines[0]);
                        }else{
@@ -750,7 +745,7 @@ int ScreenSeqsCommand::driverCreateSummary(vector<int>& startPosition, vector<in
                                count++;
                        }
                        //if((count) % 100 == 0){       m->mothurOut("Optimizing sequence: " + toString(count)); m->mothurOutEndLine();         }
-                       #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+                       #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                                unsigned long long pos = in.tellg();
                                if ((pos == -1) || (pos >= filePos.end)) { break; }
                        #else
@@ -776,7 +771,7 @@ int ScreenSeqsCommand::createProcessesCreateSummary(vector<int>& startPosition,
                int num = 0;
                vector<int> processIDS;
 
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                                
                //loop through and create all the processes you want
                while (process != processors) {
@@ -1170,7 +1165,7 @@ int ScreenSeqsCommand::driver(linePair filePos, string goodFName, string badAccn
                        count++;
                        }
                        
-                       #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+                       #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                                unsigned long long pos = inFASTA.tellg();
                                if ((pos == -1) || (pos >= filePos.end)) { break; }
                        #else
@@ -1282,7 +1277,7 @@ int ScreenSeqsCommand::createProcesses(string goodFileName, string badAccnos, st
         int process = 1;
                int num = 0;
 
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
+#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux) || (__linux__) || (__unix__) || (__unix)
                                
                //loop through and create all the processes you want
                while (process != processors) {
@@ -1363,7 +1358,7 @@ int ScreenSeqsCommand::createProcesses(string goodFileName, string badAccnos, st
             if (i!=0) {extension += toString(i) + ".temp"; processIDS.push_back(i); }
             
                        // Allocate memory for thread data.
-                       sumScreenData* tempSum = new sumScreenData(startPos, endPos, maxAmbig, maxHomoP, minLength, maxLength, filename, m, lines[i].start, lines[i].end,goodFileName+extension, badAccnos+extension, &badSeqNames);
+                       sumScreenData* tempSum = new sumScreenData(startPos, endPos, maxAmbig, maxHomoP, minLength, maxLength, filename, m, lines[i].start, lines[i].end,goodFileName+extension, badAccnos+extension);
                        pDataArray.push_back(tempSum);
                        
                        //default security attributes, thread function name, argument to thread function, use default creation flags, returns the thread identifier
@@ -1380,6 +1375,7 @@ int ScreenSeqsCommand::createProcesses(string goodFileName, string badAccnos, st
                //Close all thread handles and free memory allocations.
                for(int i=0; i < pDataArray.size(); i++){
                        num += pDataArray[i]->count;
+            for (set<string>::iterator it = pDataArray[i]->badSeqNames.begin(); it != pDataArray[i]->badSeqNames.end(); it++) {        badSeqNames.insert(*it);       }
                        CloseHandle(hThreadArray[i]);
                        delete pDataArray[i];
                }