fixed problem with class.fy.seqs where group totals did not add up. added constructo...

[mothur.git] / sequence.cpp

diff --git a/sequence.cpp b/sequence.cpp
index 752e081e367fc6d572b5622e0b4eb6003a28f9bb..71dd3a925fab8eb7771803c6db557dfb9d10200b 100644 (file)
--- a/sequence.cpp
+++ b/sequence.cpp
@@ -10,55 +10,194 @@
 #include "sequence.hpp"
 
 /***********************************************************************/
-
 Sequence::Sequence(){
+       m = MothurOut::getInstance();
        initialize();
 }
-
 /***********************************************************************/
-
 Sequence::Sequence(string newName, string sequence) {
+       try {
+               m = MothurOut::getInstance();
+               initialize();   
+               name = newName;
+               
+               //setUnaligned removes any gap characters for us
+               setUnaligned(sequence);
+               setAligned(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);
+       }                       
+}
 
-       initialize();   
-       name = newName;
+//********************************************************************************************************************
+//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();
        
-       //setUnaligned removes any gap characters for us
-       setUnaligned(sequence);
-       setAligned(sequence);
+               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(ifstream& fastaFile){
-
-       initialize();
-       fastaFile >> name;
-       name = name.substr(1);
-       string sequence;
+Sequence::Sequence(istringstream& fastaString, string JustUnaligned){
+       try {
+               m = MothurOut::getInstance();
+       
+               initialize();
+               fastaString >> 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);
+               //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
                
-               if (fastaFile) {  
-                       fastaFile >> name;  
-                       name = name.substr(1);  
-               }else { 
-                       name = "";
-                       break;
+               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); 
        }
-       
-       //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
+       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);                
-       
-       setAligned(sequence);   
-       //setUnaligned removes any gap characters for us                                                
-       setUnaligned(sequence);                                                         
+               sequence = getSequenceString(fastaFile);                
+               
+               //setUnaligned removes any gap characters for us                                                
+               setUnaligned(sequence); 
+       }
+       catch(exception& e) {
+               m->errorOut(e, "Sequence", "Sequence");
+               exit(1);
+       }                                                       
 }
+
 //********************************************************************************************************************
 string Sequence::getSequenceString(ifstream& fastaFile) {
        try {
@@ -81,7 +220,7 @@ string Sequence::getSequenceString(ifstream& fastaFile) {
                return sequence;
        }
        catch(exception& e) {
-               errorOut(e, "Sequence", "getSequenceString");
+               m->errorOut(e, "Sequence", "getSequenceString");
                exit(1);
        }
 }
@@ -103,11 +242,59 @@ string Sequence::getCommentString(ifstream& fastaFile) {
                return sequence;
        }
        catch(exception& e) {
-               errorOut(e, "Sequence", "getCommentString");
+               m->errorOut(e, "Sequence", "getCommentString");
+               exit(1);
+       }
+}
+//********************************************************************************************************************
+string Sequence::getSequenceString(istringstream& fastaFile) {
+       try {
+               char letter;
+               string sequence = "";   
+               
+               while(!fastaFile.eof()){
+                       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(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;
+       }
+       catch(exception& e) {
+               m->errorOut(e, "Sequence", "getCommentString");
                exit(1);
        }
 }
-
 //********************************************************************************************************************
 
 void Sequence::initialize(){
@@ -216,7 +403,8 @@ string Sequence::getName(){
 //********************************************************************************************************************
 
 string Sequence::getAligned(){
-       return aligned;
+       if(isAligned == 0)      { return unaligned; }
+       else                            {  return aligned;  }
 }
 
 //********************************************************************************************************************
@@ -345,5 +533,4 @@ void Sequence::reverseComplement(){
        aligned = temp;
        
 }
-
-//********************************************************************************************************************
+/**************************************************************************************************/