X-Git-Url: https://git.donarmstrong.com/?p=mothur.git;a=blobdiff_plain;f=fastamap.cpp;h=bf55493faece21cbd6c0043ac89149826c83c166;hp=4f4bf33206d77e437dd32bd2ff14b7f5f1290042;hb=d1c97b8c04bb75faca1e76ffad60b37a4d789d3d;hpb=20a2d0350a737a434c89f303662d64a8eeea7b05

diff --git a/fastamap.cpp b/fastamap.cpp
index 4f4bf33..bf55493 100644
--- a/fastamap.cpp
+++ b/fastamap.cpp
@@ -8,76 +8,190 @@
  */
 
 #include "fastamap.h"
+#include "sequence.hpp"
 
 /*******************************************************************************/
- FastaMap::FastaMap(ifstream& in) {
-	//int numberOfSequences = 0;
+
+void FastaMap::readFastaFile(string inFileName) {
+	try {
+		ifstream in;
+		m->openInputFile(inFileName, in);
+		string name, sequence, line;
+		sequence = "";
+		string temp;
+		map<string, string>::iterator itName;
+		
+		
+		while(!in.eof()){
+			if (m->control_pressed) { break; }
+			
+			Sequence currSeq(in);
+			name = currSeq.getName();
+			
+			if (name != "") {
+				if(currSeq.getIsAligned())	{	sequence = currSeq.getAligned();	}
+				else						{	sequence = currSeq.getUnaligned();	}
+				
+				itName = seqmap.find(name);
+				if (itName == seqmap.end()) { seqmap[name] = sequence;  }
+				else { m->mothurOut("You already have a sequence named " + name + ", sequence names must be unique, please correct."); m->mothurOutEndLine(); }
+				
+				map<string,group>::iterator it = data.find(sequence);
+				if (it == data.end()) { 	//it's unique.
+					data[sequence].groupname = name;  //group name will be the name of the first duplicate sequence found.
+					//				data[sequence].groupnumber = 1;
+					data[sequence].names = name;
+				}else { // its a duplicate.
+					data[sequence].names += "," + name;
+					//				data[sequence].groupnumber++;
+				}	
+			}
+			m->gobble(in);
+		}
+		in.close();		
+	}
+	catch(exception& e) {
+		m->errorOut(e, "FastaMap", "readFastaFile");
+		exit(1);
+	}
+}
+
+/*******************************************************************************/
+
+void FastaMap::readFastaFile(string inFastaFile, string oldNameFileName){ //prints data
 	
-	string name, sequence, line;
-	sequence = "";
+	ifstream oldNameFile;
+	m->openInputFile(oldNameFileName, oldNameFile);
 	
-	getline(in, line);
-	name = line.substr(1, line.length());  //rips off '>'
+	map<string,string> oldNameMap;
+	map<string, string>::iterator itName;
+	string name, list;
+	while(!oldNameFile.eof()){
+		if (m->control_pressed) { break; }
+		
+		oldNameFile >> name; m->gobble(oldNameFile);
+		oldNameFile >> list;
+		oldNameMap[name] = list;
+		m->gobble(oldNameFile);
+	}
+	oldNameFile.close();
 	
-	//read through file
-	while (getline(in, line)) {
-		if (isalnum(line.at(0))){  //if it's a sequence line
-			sequence += line;
-		}
-		else{
-			//input sequence info into map
-			it = data.find(sequence);
+	ifstream inFASTA;
+	m->openInputFile(inFastaFile, inFASTA);
+	string sequence;
+	while(!inFASTA.eof()){
+		if (m->control_pressed) { break; }
+		
+		Sequence currSeq(inFASTA);
+		name = currSeq.getName();
+		
+		if (name != "") {
+			if(currSeq.getIsAligned())	{	sequence = currSeq.getAligned();	}
+			else						{	sequence = currSeq.getUnaligned();	}
+			
+			itName = seqmap.find(name);
+			if (itName == seqmap.end()) { seqmap[name] = sequence;  }
+			else { m->mothurOut("You already have a sequence named " + name + ", sequence names must be unique, please correct."); m->mothurOutEndLine(); }
+			
+			seqmap[name] = sequence;  
+			map<string,group>::iterator it = data.find(sequence);
 			if (it == data.end()) { 	//it's unique.
 				data[sequence].groupname = name;  //group name will be the name of the first duplicate sequence found.
-				data[sequence].groupnumber = 1;
-				data[sequence].names = name;
+				//			data[sequence].groupnumber = 1;
+				data[sequence].names = oldNameMap[name];
 			}else { // its a duplicate.
-				data[sequence].names += "," + name;
-				data[sequence].groupnumber++;	
-			}
-			name = (line.substr(1, (line.npos))); //The line you just read is a new name so rip off '>'
+				data[sequence].names += "," + oldNameMap[name];
+				//			data[sequence].groupnumber++;
+			}	
 		}
+		m->gobble(inFASTA);
 	}
 	
-	//store last sequence and name info.
-	it = data.find(sequence);
-	if (it == data.end()) { 	//it's unique.
-		data[sequence].groupname = name;  //group name will be the name of the first duplicate sequence found.
-		data[sequence].groupnumber = 1;
-		data[sequence].names = name;
-	}else { // its a duplicate.
-		data[sequence].names += "," + name;
-		data[sequence].groupnumber++;	
-	}	
+	
+	inFASTA.close();
 }
+
 /*******************************************************************************/
+
 string FastaMap::getGroupName(string seq) {  //pass a sequence name get its group
 	return data[seq].groupname;
 }
+
 /*******************************************************************************/
-int FastaMap::getGroupNumber(string seq) {  //pass a sequence name get number of sequence in its group
-	return data[seq].groupnumber;
-}
-/*******************************************************************************/
+
 string FastaMap::getNames(string seq) {	//pass a sequence get the string of names in the group separated by ','s.
 	return data[seq].names;
 }
+
 /*******************************************************************************/
-void FastaMap::push_back(string seq, string Name) {//sequencename, name
-	data[seq].groupname = Name;
-	data[seq].groupnumber = 1;
-	data[seq].names = Name;
-}
+
+string FastaMap::getSequence(string name) {
+	
+	map<string,string>::iterator it = seqmap.find(name);
+	if (it == seqmap.end()) { 	return "not found";		}
+	else					{	return it->second;		}
+	
+}	
+
 /*******************************************************************************/
-void FastaMap::clear() { //clears out data
-	data.clear();
+
+void FastaMap::push_back(string name, string seq) {
+	
+	map<string,group>::iterator it = data.find(seq);
+	if (it == data.end()) { 	//it's unique.
+		data[seq].groupname = name;  //group name will be the name of the first duplicate sequence found.
+		data[seq].names = name;
+	}else { // its a duplicate.
+		data[seq].names += "," + name;
+	}
+	seqmap[name] = seq;
 }
+
 /*******************************************************************************/
-int FastaMap::size(){ //returns datas size which is the number of unique sequences
+
+int FastaMap::sizeUnique(){ //returns datas size which is the number of unique sequences
 	return data.size();
 }
+
 /*******************************************************************************/
-void FastaMap::print(ostream&){ //prints data
 
+void FastaMap::printNamesFile(string outFileName){ //prints data
+	try {
+		ofstream outFile;
+		m->openOutputFile(outFileName, outFile);
+		
+		// two column file created with groupname and them list of identical sequence names
+		for (map<string,group>::iterator it = data.begin(); it != data.end(); it++) {
+			if (m->control_pressed) { break; }
+			outFile << it->second.groupname << '\t' << it->second.names << endl;
+		}
+		outFile.close();
+	}
+	catch(exception& e) {
+		m->errorOut(e, "FastaMap", "printNamesFile");
+		exit(1);
+	}
 }
+
 /*******************************************************************************/
+
+void FastaMap::printCondensedFasta(string outFileName){ //prints data
+	try {
+		ofstream out;
+		m->openOutputFile(outFileName, out);
+		//creates a fasta file
+		for (map<string,group>::iterator it = data.begin(); it != data.end(); it++) {
+			if (m->control_pressed) { break; }
+			out << ">" << it->second.groupname << endl;
+			out << it->first << endl;
+		}
+		out.close();
+	}
+	catch(exception& e) {
+		m->errorOut(e, "FastaMap", "printCondensedFasta");
+		exit(1);
+	}
+}
+
+/*******************************************************************************/
+