X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=classify.cpp;h=875628d85d8dbcb4caf6d712b41be517eefef733;hb=03acfa01ede71b107817966fcf27cd728038b3c9;hp=e07344d4678a0ca263008e4b10d4d125ad8ef122;hpb=725a3d4ff2442c79bfde0a75ed3e0904edcf03b7;p=mothur.git

diff --git a/classify.cpp b/classify.cpp
index e07344d..875628d 100644
--- a/classify.cpp
+++ b/classify.cpp
@@ -28,7 +28,7 @@ void Classify::generateDatabaseAndNames(string tfile, string tempFile, string me
 		m->mothurOut("Generating search database...    "); cout.flush();
 #ifdef USE_MPI	
 			int pid, processors;
-			vector<long> positions;
+			vector<unsigned long int> positions;
 			int tag = 2001;
 		
 			MPI_Status status; 
@@ -46,7 +46,7 @@ void Classify::generateDatabaseAndNames(string tfile, string tempFile, string me
 			//delete inFileName;
 
 			if (pid == 0) { //only one process needs to scan file
-				positions = setFilePosFasta(tempFile, numSeqs); //fills MPIPos, returns numSeqs
+				positions = m->setFilePosFasta(tempFile, numSeqs); //fills MPIPos, returns numSeqs
 
 				//send file positions to all processes
 				for(int i = 1; i < processors; i++) { 
@@ -96,8 +96,8 @@ void Classify::generateDatabaseAndNames(string tfile, string tempFile, string me
 		//need to know number of template seqs for suffixdb
 		if (method == "suffix") {
 			ifstream inFASTA;
-			openInputFile(tempFile, inFASTA);
-			numSeqs = count(istreambuf_iterator<char>(inFASTA),istreambuf_iterator<char>(), '>');
+			m->openInputFile(tempFile, inFASTA);
+			m->getNumSeqs(inFASTA, numSeqs);
 			inFASTA.close();
 		}
 
@@ -108,7 +108,10 @@ void Classify::generateDatabaseAndNames(string tfile, string tempFile, string me
 			
 			kmerDBName = tempFile.substr(0,tempFile.find_last_of(".")+1) + char('0'+ kmerSize) + "mer";
 			ifstream kmerFileTest(kmerDBName.c_str());
-			if(kmerFileTest){	needToGenerate = false;		}
+			if(kmerFileTest){	
+				bool GoodFile = m->checkReleaseVersion(kmerFileTest, m->getVersion());
+				if (GoodFile) {  needToGenerate = false;	}
+			}
 		}
 		else if(method == "suffix")		{	database = new SuffixDB(numSeqs);								}
 		else if(method == "blast")		{	database = new BlastDB(gapOpen, gapExtend, match, misMatch);	}
@@ -121,14 +124,14 @@ void Classify::generateDatabaseAndNames(string tfile, string tempFile, string me
 		
 		if (needToGenerate) {
 			ifstream fastaFile;
-			openInputFile(tempFile, fastaFile);
+			m->openInputFile(tempFile, fastaFile);
 			
 			while (!fastaFile.eof()) {
 				Sequence temp(fastaFile);
-				gobble(fastaFile);
+				m->gobble(fastaFile);
 			
 				names.push_back(temp.getName());
-								
+							
 				database->addSequence(temp);	
 			}
 			fastaFile.close();
@@ -138,21 +141,27 @@ void Classify::generateDatabaseAndNames(string tfile, string tempFile, string me
 		}else if ((method == "kmer") && (!needToGenerate)) {	
 			ifstream kmerFileTest(kmerDBName.c_str());
 			database->readKmerDB(kmerFileTest);	
-			
+		
 			ifstream fastaFile;
-			openInputFile(tempFile, fastaFile);
+			m->openInputFile(tempFile, fastaFile);
 			
 			while (!fastaFile.eof()) {
 				Sequence temp(fastaFile);
-				gobble(fastaFile);
+				m->gobble(fastaFile);
 
 				names.push_back(temp.getName());
 			}
 			fastaFile.close();
 		}
-#endif		
+#endif	
+	
 		database->setNumSeqs(names.size());
 		
+		//sanity check
+		bool okay = phyloTree->ErrorCheck(names);
+		
+		if (!okay) { m->control_pressed = true; }
+		
 		m->mothurOut("DONE."); m->mothurOutEndLine();
 		m->mothurOut("It took " + toString(time(NULL) - start) + " seconds generate search database. "); m->mothurOutEndLine();
 
@@ -177,7 +186,7 @@ int Classify::readTaxonomy(string file) {
 
 #ifdef USE_MPI	
 		int pid, num, processors;
-		vector<long> positions;
+		vector<unsigned long int> positions;
 		int tag = 2001;
 		
 		MPI_Status status; 
@@ -195,7 +204,7 @@ int Classify::readTaxonomy(string file) {
 		//delete inFileName;
 
 		if (pid == 0) {
-			positions = setFilePosEachLine(file, num);
+			positions = m->setFilePosEachLine(file, num);
 			
 			//send file positions to all processes
 			for(int i = 1; i < processors; i++) { 
@@ -230,7 +239,7 @@ int Classify::readTaxonomy(string file) {
 		MPI_Barrier(MPI_COMM_WORLD); //make everyone wait - just in case
 #else				
 		ifstream inTax;
-		openInputFile(file, inTax);
+		m->openInputFile(file, inTax);
 	
 		//read template seqs and save
 		while (!inTax.eof()) {
@@ -240,7 +249,7 @@ int Classify::readTaxonomy(string file) {
 			
 			phyloTree->addSeqToTree(name, taxInfo);
 		
-			gobble(inTax);
+			m->gobble(inTax);
 		}
 		inTax.close();
 #endif	
@@ -248,7 +257,7 @@ int Classify::readTaxonomy(string file) {
 		phyloTree->assignHeirarchyIDs(0);
 		
 		phyloTree->setUp(file);
-		
+	
 		m->mothurOut("DONE.");
 		m->mothurOutEndLine();	cout.flush();