X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=sequence.cpp;h=71dd3a925fab8eb7771803c6db557dfb9d10200b;hb=b447f829850ae054e42560c7c3ed71b14f3f40bb;hp=6a15eb5f7d23f93272811a975beedc649a0d9ce9;hpb=64eee3a595ae53817f52807d7393b22e74e31f56;p=mothur.git

diff --git a/sequence.cpp b/sequence.cpp
index 6a15eb5..71dd3a9 100644
--- a/sequence.cpp
+++ b/sequence.cpp
@@ -10,56 +10,291 @@
 #include "sequence.hpp"
 
 /***********************************************************************/
-
 Sequence::Sequence(){
+	m = MothurOut::getInstance();
 	initialize();
 }
-
 /***********************************************************************/
-
 Sequence::Sequence(string newName, string sequence) {
-
-	initialize();	
-	name = newName;
-	if(sequence.find_first_of('-') != string::npos) {
+	try {
+		m = MothurOut::getInstance();
+		initialize();	
+		name = newName;
+		
+		//setUnaligned removes any gap characters for us
+		setUnaligned(sequence);
 		setAligned(sequence);
 	}
-	setUnaligned(sequence);
+	catch(exception& e) {
+		m->errorOut(e, "Sequence", "Sequence");
+		exit(1);
+	}			
+}
+/***********************************************************************/
+Sequence::Sequence(string newName, string sequence, string justUnAligned) {
+	try {
+		m = MothurOut::getInstance();
+		initialize();	
+		name = newName;
+		
+		//setUnaligned removes any gap characters for us
+		setUnaligned(sequence);
+	}
+	catch(exception& e) {
+		m->errorOut(e, "Sequence", "Sequence");
+		exit(1);
+	}			
+}
+
+//********************************************************************************************************************
+//this function will jump over commented out sequences, but if the last sequence in a file is commented out it makes a blank seq
+Sequence::Sequence(istringstream& fastaString){
+	try {
+		m = MothurOut::getInstance();
+	
+		initialize();
+		fastaString >> name;
+		name = name.substr(1);
+		string sequence;
 	
+		//read comments
+		while ((name[0] == '#') && fastaString) { 
+			while (!fastaString.eof())	{	char c = fastaString.get(); if (c == 10 || c == 13){	break;	}	} // get rest of line if there's any crap there
+			sequence = getCommentString(fastaString);
+			
+			if (fastaString) {  
+				fastaString >> name;  
+				name = name.substr(1);	
+			}else { 
+				name = "";
+				break;
+			}
+		}
+		
+		while (!fastaString.eof())	{	char c = fastaString.get();  if (c == 10 || c == 13){	break;	}	} // get rest of line if there's any crap there
+		
+		sequence = getSequenceString(fastaString);		
+		setAligned(sequence);	
+		//setUnaligned removes any gap characters for us						
+		setUnaligned(sequence);		
+	}
+	catch(exception& e) {
+		m->errorOut(e, "Sequence", "Sequence");
+		exit(1);
+	}								
 }
 //********************************************************************************************************************
+//this function will jump over commented out sequences, but if the last sequence in a file is commented out it makes a blank seq
+Sequence::Sequence(istringstream& fastaString, string JustUnaligned){
+	try {
+		m = MothurOut::getInstance();
+	
+		initialize();
+		fastaString >> name;
+		name = name.substr(1);
+		string sequence;
+	
+		//read comments
+		while ((name[0] == '#') && fastaString) { 
+			while (!fastaString.eof())	{	char c = fastaString.get(); if (c == 10 || c == 13){	break;	}	} // get rest of line if there's any crap there
+			sequence = getCommentString(fastaString);
+			
+			if (fastaString) {  
+				fastaString >> name;  
+				name = name.substr(1);	
+			}else { 
+				name = "";
+				break;
+			}
+		}
+		
+		while (!fastaString.eof())	{	char c = fastaString.get();  if (c == 10 || c == 13){	break;	}	} // get rest of line if there's any crap there
+		
+		sequence = getSequenceString(fastaString);		
+		
+		//setUnaligned removes any gap characters for us						
+		setUnaligned(sequence);		
+	}
+	catch(exception& e) {
+		m->errorOut(e, "Sequence", "Sequence");
+		exit(1);
+	}								
+}
 
+
+//********************************************************************************************************************
+//this function will jump over commented out sequences, but if the last sequence in a file is commented out it makes a blank seq
 Sequence::Sequence(ifstream& fastaFile){
+	try {
+		m = MothurOut::getInstance();
+		initialize();
+		fastaFile >> name;
+		name = name.substr(1);
+		string sequence;
+		
+		//read comments
+		while ((name[0] == '#') && fastaFile) { 
+			while (!fastaFile.eof())	{	char c = fastaFile.get(); if (c == 10 || c == 13){	break;	}	} // get rest of line if there's any crap there
+			sequence = getCommentString(fastaFile);
+			
+			if (fastaFile) {  
+				fastaFile >> name;  
+				name = name.substr(1);	
+			}else { 
+				name = "";
+				break;
+			}
+		}
+		
+		//read real sequence
+		while (!fastaFile.eof())	{	char c = fastaFile.get(); if (c == 10 || c == 13){	break;	}	} // get rest of line if there's any crap there
+		
+		sequence = getSequenceString(fastaFile);		
+		
+		setAligned(sequence);	
+		//setUnaligned removes any gap characters for us						
+		setUnaligned(sequence);	
+	}
+	catch(exception& e) {
+		m->errorOut(e, "Sequence", "Sequence");
+		exit(1);
+	}							
+}
+//********************************************************************************************************************
+//this function will jump over commented out sequences, but if the last sequence in a file is commented out it makes a blank seq
+Sequence::Sequence(ifstream& fastaFile, string JustUnaligned){
+	try {
+		m = MothurOut::getInstance();
+		initialize();
+		fastaFile >> name;
+		name = name.substr(1);
+		string sequence;
+		
+		//read comments
+		while ((name[0] == '#') && fastaFile) { 
+			while (!fastaFile.eof())	{	char c = fastaFile.get(); if (c == 10 || c == 13){	break;	}	} // get rest of line if there's any crap there
+			sequence = getCommentString(fastaFile);
+			
+			if (fastaFile) {  
+				fastaFile >> name;  
+				name = name.substr(1);	
+			}else { 
+				name = "";
+				break;
+			}
+		}
+		
+		//read real sequence
+		while (!fastaFile.eof())	{	char c = fastaFile.get(); if (c == 10 || c == 13){	break;	}	} // get rest of line if there's any crap there
+		
+		sequence = getSequenceString(fastaFile);		
+		
+		//setUnaligned removes any gap characters for us						
+		setUnaligned(sequence);	
+	}
+	catch(exception& e) {
+		m->errorOut(e, "Sequence", "Sequence");
+		exit(1);
+	}							
+}
 
-	initialize();
-	fastaFile >> name;
-	name = name.substr(1);
-	char c;
-	
-	while ((c = fastaFile.get()) != EOF)	{	if (c == 10){	break;	}	} // get rest of line if there's any crap there
-	
-	char letter;
-	string sequence;
+//********************************************************************************************************************
+string Sequence::getSequenceString(ifstream& fastaFile) {
+	try {
+		char letter;
+		string sequence = "";	
+		
+		while(fastaFile){
+			letter= fastaFile.get();
+			if(letter == '>'){
+				fastaFile.putback(letter);
+				break;
+			}
+			else if(isprint(letter)){
+				letter = toupper(letter);
+				if(letter == 'U'){letter = 'T';}
+				sequence += letter;
+			}
+		}
+		
+		return sequence;
+	}
+	catch(exception& e) {
+		m->errorOut(e, "Sequence", "getSequenceString");
+		exit(1);
+	}
+}
+//********************************************************************************************************************
+//comment can contain '>' so we need to account for that
+string Sequence::getCommentString(ifstream& fastaFile) {
+	try {
+		char letter;
+		string sequence = "";
+		
+		while(fastaFile){
+			letter=fastaFile.get();
+			if((letter == '\r') || (letter == '\n')){  
+				gobble(fastaFile);  //in case its a \r\n situation
+				break;
+			}
+		}
+		
+		return sequence;
+	}
+	catch(exception& e) {
+		m->errorOut(e, "Sequence", "getCommentString");
+		exit(1);
+	}
+}
+//********************************************************************************************************************
+string Sequence::getSequenceString(istringstream& fastaFile) {
+	try {
+		char letter;
+		string sequence = "";	
+		
+		while(!fastaFile.eof()){
+			letter= fastaFile.get();
 	
-	while(fastaFile){
-		letter= fastaFile.get();
-		if(letter == '>'){
-			fastaFile.putback(letter);
-			break;
+			if(letter == '>'){
+				fastaFile.putback(letter);
+				break;
+			}
+			else if(isprint(letter)){
+				letter = toupper(letter);
+				if(letter == 'U'){letter = 'T';}
+				sequence += letter;
+			}
 		}
-		else if(isprint(letter)){
-			letter = toupper(letter);
-			if(letter == 'U'){letter = 'T';}
-			sequence += letter;
+		
+		return sequence;
+	}
+	catch(exception& e) {
+		m->errorOut(e, "Sequence", "getSequenceString");
+		exit(1);
+	}
+}
+//********************************************************************************************************************
+//comment can contain '>' so we need to account for that
+string Sequence::getCommentString(istringstream& fastaFile) {
+	try {
+		char letter;
+		string sequence = "";
+		
+		while(fastaFile){
+			letter=fastaFile.get();
+			if((letter == '\r') || (letter == '\n')){  
+				gobble(fastaFile);  //in case its a \r\n situation
+				break;
+			}
 		}
+		
+		return sequence;
 	}
-
-	if(sequence.find_first_of('-') != string::npos){	//	if there are any gaps in the sequence, assume that it is
-		setAligned(sequence);							//	an alignment file
+	catch(exception& e) {
+		m->errorOut(e, "Sequence", "getCommentString");
+		exit(1);
 	}
-	setUnaligned(sequence);								//	also set the unaligned sequence file
 }
-
 //********************************************************************************************************************
 
 void Sequence::initialize(){
@@ -90,7 +325,7 @@ void Sequence::setName(string seqName) {
 
 void Sequence::setUnaligned(string sequence){
 	
-	if(sequence.find_first_of('-') != string::npos) {
+	if(sequence.find_first_of('.') != string::npos || sequence.find_first_of('-') != string::npos) {
 		string temp = "";
 		for(int j=0;j<sequence.length();j++) {
 			if(isalpha(sequence[j]))	{	temp += sequence[j];	}
@@ -111,6 +346,7 @@ void Sequence::setAligned(string sequence){
 	//if the alignment starts or ends with a gap, replace it with a period to indicate missing data
 	aligned = sequence;
 	alignmentLength = aligned.length();
+	setUnaligned(sequence);	
 
 	if(aligned[0] == '-'){
 		for(int i=0;i<alignmentLength;i++){
@@ -167,7 +403,8 @@ string Sequence::getName(){
 //********************************************************************************************************************
 
 string Sequence::getAligned(){
-	return aligned;
+	if(isAligned == 0)	{ return unaligned; }
+	else				{  return aligned;  }
 }
 
 //********************************************************************************************************************
@@ -293,7 +530,7 @@ void Sequence::reverseComplement(){
 		else						{	temp += 'N';	}
 	}
 	unaligned = temp;
+	aligned = temp;
 	
 }
-
-//********************************************************************************************************************
+/**************************************************************************************************/