X-Git-Url: https://git.donarmstrong.com/?p=mothur.git;a=blobdiff_plain;f=readcluster.cpp;h=a6adabb5b8aa155d3b633f885fd37f7ba051465f;hp=11b8d012a22e592d99bb71e3063c0c89a710252a;hb=a8e2df1b96a57f5f29576b08361b86a96a8eff4f;hpb=b5a791c81d432082bf38755a08b33863f255341d

diff --git a/readcluster.cpp b/readcluster.cpp
index 11b8d01..a6adabb 100644
--- a/readcluster.cpp
+++ b/readcluster.cpp
@@ -11,31 +11,59 @@
 
 /***********************************************************************/
 
-ReadCluster::ReadCluster(string distfile, float c){
-		globaldata = GlobalData::getInstance();
+ReadCluster::ReadCluster(string distfile, float c, string o, bool s){
+		m = MothurOut::getInstance();
         distFile = distfile;
 		cutoff = c;
+		outputDir = o;
+		sortWanted = s;
+		list = NULL;
 }
 
 /***********************************************************************/
 
-void ReadCluster::read(NameAssignment* nameMap){
+int ReadCluster::read(NameAssignment*& nameMap){
 	try {
         
 		if (format == "phylip") { convertPhylip2Column(nameMap); }
 		else { list = new ListVector(nameMap->getListVector());  }
 		
-		OutPutFile = sortFile(distFile);
+		if (m->control_pressed) { return 0; }
+		
+		if (sortWanted) {  OutPutFile = m->sortFile(distFile, outputDir);  }
+		else {  OutPutFile = distFile;   } //for use by clusters splitMatrix to convert a phylip matrix to column
+		
+		return 0;
 			
 	}
 	catch(exception& e) {
-		errorOut(e, "ReadCluster", "read");
+		m->errorOut(e, "ReadCluster", "read");
+		exit(1);
+	}
+}
+/***********************************************************************/
+int ReadCluster::read(CountTable*& ct){
+	try {
+        
+		if (format == "phylip") { convertPhylip2Column(ct); }
+		else { list = new ListVector(ct->getListVector());  }
+		
+		if (m->control_pressed) { return 0; }
+		
+		if (sortWanted) {  OutPutFile = m->sortFile(distFile, outputDir);  }
+		else {  OutPutFile = distFile;   } //for use by clusters splitMatrix to convert a phylip matrix to column
+		
+		return 0;
+        
+	}
+	catch(exception& e) {
+		m->errorOut(e, "ReadCluster", "read");
 		exit(1);
 	}
 }
 /***********************************************************************/
 
-void ReadCluster::convertPhylip2Column(NameAssignment* nameMap){
+int ReadCluster::convertPhylip2Column(NameAssignment*& nameMap){
 	try {	
 		//convert phylip file to column file
 		map<int, string> rowToName;
@@ -45,15 +73,20 @@ void ReadCluster::convertPhylip2Column(NameAssignment* nameMap){
 		ofstream out;
 		string tempFile = distFile + ".column.temp";
 		
-		openInputFile(distFile, in);
-		openOutputFile(tempFile, out);
+		m->openInputFile(distFile, in);  m->gobble(in);
+		m->openOutputFile(tempFile, out);
 		
 		float distance;
 		int square, nseqs;
 		string name;
 		vector<string> matrixNames;
-        
-		in >> nseqs >> name;
+		
+		string numTest;
+		in >> numTest >> name;
+		
+		if (!m->isContainingOnlyDigits(numTest)) { m->mothurOut("[ERROR]: expected a number and got " + numTest + ", quitting."); m->mothurOutEndLine(); exit(1); }
+		else { convert(numTest, nseqs); }
+		
 		rowToName[0] = name;
 		matrixNames.push_back(name);
 		
@@ -63,7 +96,7 @@ void ReadCluster::convertPhylip2Column(NameAssignment* nameMap){
 		}
 		else{
 			list = new ListVector(nameMap->getListVector());
-			if(nameMap->count(name)==0){        mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); mothurOutEndLine(); }
+			if(nameMap->count(name)==0){        m->mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); m->mothurOutEndLine(); }
 		}
         
 		char d;
@@ -95,6 +128,9 @@ void ReadCluster::convertPhylip2Column(NameAssignment* nameMap){
 					list->set(i, name);
 					
 					for(int j=0;j<i;j++){
+					
+						if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(tempFile); return 0; }
+						
 						in >> distance;
 						
 						if (distance == -1) { distance = 1000000; }
@@ -106,9 +142,12 @@ void ReadCluster::convertPhylip2Column(NameAssignment* nameMap){
 					
 				}
 				else{
-					if(nameMap->count(name)==0){        mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); mothurOutEndLine(); }
+					if(nameMap->count(name)==0){        m->mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); m->mothurOutEndLine(); }
 					
 					for(int j=0;j<i;j++){
+						
+						if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(tempFile); return 0; }
+						
 						in >> distance;
 						
 						if (distance == -1) { distance = 1000000; }
@@ -129,6 +168,8 @@ void ReadCluster::convertPhylip2Column(NameAssignment* nameMap){
 				if(nameMap == NULL){
 					list->set(i, name);
 					for(int j=0;j<nseqs;j++){
+						if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(tempFile); return 0; }
+						
 						in >> distance;
 					
 						if (distance == -1) { distance = 1000000; }
@@ -139,9 +180,11 @@ void ReadCluster::convertPhylip2Column(NameAssignment* nameMap){
 					}
 				}
 				else{
-					if(nameMap->count(name)==0){        mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); mothurOutEndLine(); }
+					if(nameMap->count(name)==0){        m->mothurOut("Error: Sequence '" + name + "' was not found in the names file, please correct"); m->mothurOutEndLine(); }
 					
 					for(int j=0;j<nseqs;j++){
+						if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(tempFile); return 0; }
+						
 						in >> distance;
                         
 						if (distance == -1) { distance = 1000000; }
@@ -158,39 +201,219 @@ void ReadCluster::convertPhylip2Column(NameAssignment* nameMap){
 		list->setLabel("0");
 		in.close();
 		out.close();
-		
+	
 		if(nameMap == NULL){
 			nameMap = new NameAssignment();
 			for(int i=0;i<matrixNames.size();i++){
 				nameMap->push_back(matrixNames[i]);
 			}
-			globaldata->nameMap = nameMap;
 		}
 		
 	
 		ifstream in2;
 		ofstream out2;
 		
-		string outputFile = getRootName(distFile) + "column.dist";
-		openInputFile(tempFile, in2);
-		openOutputFile(outputFile, out2);
+		string outputFile = m->getRootName(distFile) + "column.dist";
+		m->openInputFile(tempFile, in2);
+		m->openOutputFile(outputFile, out2);
 		
 		int first, second;
 		float dist;
 		
 		while (in2) {
+			if (m->control_pressed) { in2.close(); out2.close(); m->mothurRemove(tempFile); m->mothurRemove(outputFile); return 0; }
+			
 			in2 >> first >> second >> dist;
 			out2 << rowToName[first] << '\t' << rowToName[second] << '\t' << dist << endl;
-			gobble(in2);
+			m->gobble(in2);
 		}
 		in2.close();
 		out2.close();
 		
-		remove(tempFile.c_str());
+		m->mothurRemove(tempFile);
 		distFile = outputFile;
+	
+		if (m->control_pressed) {  m->mothurRemove(outputFile);  }
+
+		return 0;
+	}
+	catch(exception& e) {
+		m->errorOut(e, "ReadCluster", "convertPhylip2Column");
+		exit(1);
+	}
+}
+/***********************************************************************/
+
+int ReadCluster::convertPhylip2Column(CountTable*& ct){
+	try {	
+		//convert phylip file to column file
+		map<int, string> rowToName;
+		map<int, string>::iterator it;
+		
+		ifstream in;
+		ofstream out;
+		string tempFile = distFile + ".column.temp";
+		
+		m->openInputFile(distFile, in);  m->gobble(in);
+		m->openOutputFile(tempFile, out);
+		
+		float distance;
+		int square, nseqs;
+		string name;
+		vector<string> matrixNames;
+		
+		string numTest;
+		in >> numTest >> name;
+		
+		if (!m->isContainingOnlyDigits(numTest)) { m->mothurOut("[ERROR]: expected a number and got " + numTest + ", quitting."); m->mothurOutEndLine(); exit(1); }
+		else { convert(numTest, nseqs); }
+		
+		rowToName[0] = name;
+		matrixNames.push_back(name);
+		
+		if(ct == NULL){
+			list = new ListVector(nseqs);
+			list->set(0, name);
+		}
+		else{  list = new ListVector(ct->getListVector()); }
+        
+		char d;
+		while((d=in.get()) != EOF){
+			
+			if(isalnum(d)){
+				square = 1;
+				in.putback(d);
+				for(int i=0;i<nseqs;i++){
+					in >> distance;
+				}
+				break;
+			}
+			if(d == '\n'){
+				square = 0;
+				break;
+			}
+		}
+        
+		if(square == 0){
+            
+			for(int i=1;i<nseqs;i++){
+				in >> name;
+				rowToName[i] = name;
+				matrixNames.push_back(name);
+				
+				//there's A LOT of repeated code throughout this method...
+				if(ct == NULL){
+					list->set(i, name);
+					
+					for(int j=0;j<i;j++){
+                        
+						if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(tempFile); return 0; }
+						
+						in >> distance;
+						
+						if (distance == -1) { distance = 1000000; }
+						
+						if(distance < cutoff){
+							out << i << '\t' << j << '\t' << distance << endl;
+						}
+					}
+					
+				}
+				else{
+					
+					for(int j=0;j<i;j++){
+						
+						if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(tempFile); return 0; }
+						
+						in >> distance;
+						
+						if (distance == -1) { distance = 1000000; }
+						
+						if(distance < cutoff){
+							out << i << '\t' << j << '\t' << distance << endl;
+						}
+					}
+				}
+			}
+		}
+		else{
+			for(int i=1;i<nseqs;i++){
+				in >> name;                
+				rowToName[i] = name;
+				matrixNames.push_back(name);
+                
+				if(ct == NULL){
+					list->set(i, name);
+					for(int j=0;j<nseqs;j++){
+						if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(tempFile); return 0; }
+						
+						in >> distance;
+                        
+						if (distance == -1) { distance = 1000000; }
+						
+						if(distance < cutoff && j < i){
+							out << i << '\t' << j << '\t' << distance << endl;
+						}
+					}
+				}
+				else{
+					for(int j=0;j<nseqs;j++){
+						if (m->control_pressed) { in.close(); out.close(); m->mothurRemove(tempFile); return 0; }
+						
+						in >> distance;
+                        
+						if (distance == -1) { distance = 1000000; }
+						
+						if(distance < cutoff && j < i){
+							out << i << '\t' << j << '\t' << distance << endl;
+						}
+						
+					}
+				}
+			}
+		}
+		
+		list->setLabel("0");
+		in.close();
+		out.close();
+        
+		if(ct == NULL){
+			ct = new CountTable();
+			for(int i=0;i<matrixNames.size();i++){
+				ct->push_back(matrixNames[i]);
+			}
+		}
+		
+        
+		ifstream in2;
+		ofstream out2;
+		
+		string outputFile = m->getRootName(distFile) + "column.dist";
+		m->openInputFile(tempFile, in2);
+		m->openOutputFile(outputFile, out2);
+		
+		int first, second;
+		float dist;
+		
+		while (in2) {
+			if (m->control_pressed) { in2.close(); out2.close(); m->mothurRemove(tempFile); m->mothurRemove(outputFile); return 0; }
+			
+			in2 >> first >> second >> dist;
+			out2 << rowToName[first] << '\t' << rowToName[second] << '\t' << dist << endl;
+			m->gobble(in2);
+		}
+		in2.close();
+		out2.close();
+		
+		m->mothurRemove(tempFile);
+		distFile = outputFile;
+        
+		if (m->control_pressed) {  m->mothurRemove(outputFile);  }
+        
+		return 0;
 	}
 	catch(exception& e) {
-		errorOut(e, "ReadCluster", "convertPhylip2Column");
+		m->errorOut(e, "ReadCluster", "convertPhylip2Column");
 		exit(1);
 	}
 }