X-Git-Url: https://git.donarmstrong.com/?p=mothur.git;a=blobdiff_plain;f=sequenceparser.cpp;h=37891eb44a0b42c2b9f0879dfd084da1893c289b;hp=3eb508dd737e15d4f06cb04c202608b5d5e1fe7c;hb=b206f634aae1b4ce13978d203247fb64757d5482;hpb=2c97dd48b8e27ee0a6a86c7a082f4c504c3357c6

diff --git a/sequenceparser.cpp b/sequenceparser.cpp
index 3eb508d..37891eb 100644
--- a/sequenceparser.cpp
+++ b/sequenceparser.cpp
@@ -59,81 +59,167 @@ SequenceParser::SequenceParser(string groupFile, string fastaFile, string nameFi
 		in.close();
 				 
 		if (error == 1) { m->control_pressed = true; }
-				 
+		
 		//read name file
 		ifstream inName;
 		m->openInputFile(nameFile, inName);
 		
-		string first, second;
+		//string first, second;
 		int countName = 0;
 		set<string> thisnames1;
 		
-		while(!inName.eof()) {
-			
+        string rest = "";
+        char buffer[4096];
+        bool pairDone = false;
+        bool columnOne = true;
+        string firstCol, secondCol;
+        
+		while (!inName.eof()) {
 			if (m->control_pressed) { break; }
 			
-			inName >> first; m->gobble(inName);
-			inName >> second; m->gobble(inName);
-			
-			vector<string> names;
-			m->splitAtChar(second, names, ',');
-			
-			//get aligned string for these seqs from the fasta file
-			string alignedString = "";
-			map<string, string>::iterator itAligned = seqName.find(names[0]);
-			if (itAligned == seqName.end()) {
-				error = 1; m->mothurOut("[ERROR]: " + names[0] + " is in your name file and not in your fasta file, please correct."); m->mothurOutEndLine();
-			}else {
-				alignedString = itAligned->second;
-			}
-			
-			//separate by group - parse one line in name file
-			map<string, string> splitMap; //group -> name1,name2,...
-			map<string, string>::iterator it;
-			for (int i = 0; i < names.size(); i++) {
-				
-				string group = groupMap->getGroup(names[i]);
-				if (group == "not found") {  error = 1; m->mothurOut("[ERROR]: " + names[i] + " is in your name file and not in your groupfile, please correct."); m->mothurOutEndLine();  }
-				else {	
-					
-					it = splitMap.find(group);
-					if (it != splitMap.end()) { //adding seqs to this group
-						(it->second) += "," + names[i];
-						thisnames1.insert(names[i]);
-						countName++;
-					}else { //first sighting of this group
-						splitMap[group] = names[i];
-						countName++;
-						thisnames1.insert(names[i]);
-						
-						//is this seq in the fasta file?
-						if (i != 0) { //if not then we need to add a duplicate sequence to the seqs for this group so the new "fasta" and "name" files will match
-							Sequence tempSeq(names[i], alignedString); //get the first guys sequence string since he's in the fasta file.
-							seqs[group].push_back(tempSeq);
-						}
-					}
-				}
-				
-				allSeqsMap[names[i]] = names[0];
-			}
-			
-			
-			//fill nameMapPerGroup - holds all lines in namefile separated by group
-			for (it = splitMap.begin(); it != splitMap.end(); it++) {
-				//grab first name
-				string firstName = "";
-				for(int i = 0; i < (it->second).length(); i++) {
-					if (((it->second)[i]) != ',') {
-						firstName += ((it->second)[i]);
-					}else { break; }
-				}
-				
-				//group1 -> seq1 -> seq1,seq2,seq3
-				nameMapPerGroup[it->first][firstName] = it->second;
-			}
+            inName.read(buffer, 4096);
+            vector<string> pieces = m->splitWhiteSpace(rest, buffer, inName.gcount());
+            
+            for (int i = 0; i < pieces.size(); i++) {
+                if (columnOne) {  firstCol = pieces[i]; columnOne=false; }
+                else  { secondCol = pieces[i]; pairDone = true; columnOne=true; }
+                
+                if (pairDone) { //save one line
+                    if (m->debug) { m->mothurOut("[DEBUG]: reading names: " + firstCol + '\t' + secondCol + ".\n"); }
+                    vector<string> names;
+                    m->splitAtChar(secondCol, names, ',');
+                    
+                    //get aligned string for these seqs from the fasta file
+                    string alignedString = "";
+                    map<string, string>::iterator itAligned = seqName.find(names[0]);
+                    if (itAligned == seqName.end()) {
+                        error = 1; m->mothurOut("[ERROR]: " + names[0] + " is in your name file and not in your fasta file, please correct."); m->mothurOutEndLine();
+                    }else {
+                        alignedString = itAligned->second;
+                    }
+                    
+                    //separate by group - parse one line in name file
+                    map<string, string> splitMap; //group -> name1,name2,...
+                    map<string, string>::iterator it;
+                    for (int i = 0; i < names.size(); i++) {
+                        
+                        string group = groupMap->getGroup(names[i]);
+                        if (group == "not found") {  error = 1; m->mothurOut("[ERROR]: " + names[i] + " is in your name file and not in your groupfile, please correct."); m->mothurOutEndLine();  }
+                        else {	
+                            
+                            it = splitMap.find(group);
+                            if (it != splitMap.end()) { //adding seqs to this group
+                                (it->second) += "," + names[i];
+                                thisnames1.insert(names[i]);
+                                countName++;
+                            }else { //first sighting of this group
+                                splitMap[group] = names[i];
+                                countName++;
+                                thisnames1.insert(names[i]);
+                                
+                                //is this seq in the fasta file?
+                                if (i != 0) { //if not then we need to add a duplicate sequence to the seqs for this group so the new "fasta" and "name" files will match
+                                    Sequence tempSeq(names[i], alignedString); //get the first guys sequence string since he's in the fasta file.
+                                    seqs[group].push_back(tempSeq);
+                                }
+                            }
+                        }
+                        
+                        allSeqsMap[names[i]] = names[0];
+                    }
+                    
+                    
+                    //fill nameMapPerGroup - holds all lines in namefile separated by group
+                    for (it = splitMap.begin(); it != splitMap.end(); it++) {
+                        //grab first name
+                        string firstName = "";
+                        for(int i = 0; i < (it->second).length(); i++) {
+                            if (((it->second)[i]) != ',') {
+                                firstName += ((it->second)[i]);
+                            }else { break; }
+                        }
+                        
+                        //group1 -> seq1 -> seq1,seq2,seq3
+                        nameMapPerGroup[it->first][firstName] = it->second;
+                    }
+
+                    pairDone = false; 
+                }
+            }
 		}
-		
 		inName.close();
+        
+        //in case file does not end in white space
+        if (rest != "") {
+            vector<string> pieces = m->splitWhiteSpace(rest);
+            
+            for (int i = 0; i < pieces.size(); i++) {
+                if (columnOne) {  firstCol = pieces[i]; columnOne=false; }
+                else  { secondCol = pieces[i]; pairDone = true; columnOne=true; }
+                
+                if (pairDone) { //save one line
+                    if (m->debug) { m->mothurOut("[DEBUG]: reading names: " + firstCol + '\t' + secondCol + ".\n"); }
+                    vector<string> names;
+                    m->splitAtChar(secondCol, names, ',');
+                    
+                    //get aligned string for these seqs from the fasta file
+                    string alignedString = "";
+                    map<string, string>::iterator itAligned = seqName.find(names[0]);
+                    if (itAligned == seqName.end()) {
+                        error = 1; m->mothurOut("[ERROR]: " + names[0] + " is in your name file and not in your fasta file, please correct."); m->mothurOutEndLine();
+                    }else {
+                        alignedString = itAligned->second;
+                    }
+                    
+                    //separate by group - parse one line in name file
+                    map<string, string> splitMap; //group -> name1,name2,...
+                    map<string, string>::iterator it;
+                    for (int i = 0; i < names.size(); i++) {
+                        
+                        string group = groupMap->getGroup(names[i]);
+                        if (group == "not found") {  error = 1; m->mothurOut("[ERROR]: " + names[i] + " is in your name file and not in your groupfile, please correct."); m->mothurOutEndLine();  }
+                        else {	
+                            
+                            it = splitMap.find(group);
+                            if (it != splitMap.end()) { //adding seqs to this group
+                                (it->second) += "," + names[i];
+                                thisnames1.insert(names[i]);
+                                countName++;
+                            }else { //first sighting of this group
+                                splitMap[group] = names[i];
+                                countName++;
+                                thisnames1.insert(names[i]);
+                                
+                                //is this seq in the fasta file?
+                                if (i != 0) { //if not then we need to add a duplicate sequence to the seqs for this group so the new "fasta" and "name" files will match
+                                    Sequence tempSeq(names[i], alignedString); //get the first guys sequence string since he's in the fasta file.
+                                    seqs[group].push_back(tempSeq);
+                                }
+                            }
+                        }
+                        
+                        allSeqsMap[names[i]] = names[0];
+                    }
+                    
+                    
+                    //fill nameMapPerGroup - holds all lines in namefile separated by group
+                    for (it = splitMap.begin(); it != splitMap.end(); it++) {
+                        //grab first name
+                        string firstName = "";
+                        for(int i = 0; i < (it->second).length(); i++) {
+                            if (((it->second)[i]) != ',') {
+                                firstName += ((it->second)[i]);
+                            }else { break; }
+                        }
+                        
+                        //group1 -> seq1 -> seq1,seq2,seq3
+                        nameMapPerGroup[it->first][firstName] = it->second;
+                    }
+                    
+                    pairDone = false; 
+                }
+            }
+        }
 		
 		if (error == 1) { m->control_pressed = true; }
 		        
@@ -224,8 +310,6 @@ vector<string> SequenceParser::getNamesOfGroups(){ return groupMap->getNamesOfGr
 /************************************************************/
 bool SequenceParser::isValidGroup(string g){ return groupMap->isValidGroup(g); }
 /************************************************************/
-string SequenceParser::getGroup(string g){ return groupMap->getGroup(g); }
-/************************************************************/
 int SequenceParser::getNumSeqs(string g){ 
 	try {
 		map<string, vector<Sequence> >::iterator it;
@@ -316,7 +400,7 @@ int SequenceParser::getSeqs(string g, string filename, bool uchimeFormat=false){
 					
 					if(m->control_pressed) { out.close(); m->mothurRemove(filename); return 1; }
 					
-					out << ">" << nameVector[i].name  << "/ab=" << nameVector[i].numIdentical << "/" << endl << nameVector[i].seq << endl;
+					out << ">" <<  nameVector[i].name << "/ab=" << nameVector[i].numIdentical << "/" << endl << nameVector[i].seq << endl; //
 				}
 				
 			}else {