added modify names parameter to set.dir

[mothur.git] / sequence.cpp

diff --git a/sequence.cpp b/sequence.cpp
index a1e787bab8cd24ebb352593055e7d2fc10b406de..d6073d75da3ed630b27b626bbba8615d79b58858 100644 (file)
--- a/sequence.cpp
+++ b/sequence.cpp
@@ -20,6 +20,8 @@ Sequence::Sequence(string newName, string sequence) {
                m = MothurOut::getInstance();
                initialize();   
                name = newName;
+        
+        m->checkName(name);
                
                //setUnaligned removes any gap characters for us
                setUnaligned(sequence);
@@ -36,6 +38,8 @@ Sequence::Sequence(string newName, string sequence, string justUnAligned) {
                m = MothurOut::getInstance();
                initialize();   
                name = newName;
+        
+        m->checkName(name);
                
                //setUnaligned removes any gap characters for us
                setUnaligned(sequence);
@@ -53,11 +57,9 @@ Sequence::Sequence(istringstream& fastaString){
                m = MothurOut::getInstance();
        
                initialize();
-               fastaString >> name;
-               
-               if (name.length() != 0) { 
+        name = getSequenceName(fastaString);
                
-                       name = name.substr(1);
+               if (!m->control_pressed) { 
                        string sequence;
                
                        //read comments
@@ -84,8 +86,7 @@ Sequence::Sequence(istringstream& fastaString){
                        setUnaligned(sequence); 
                        
                        if ((numAmbig / (float) numBases) > 0.25) { m->mothurOut("[WARNING]: We found more than 25% of the bases in sequence " + name + " to be ambiguous. Mothur is not setup to process protein sequences."); m->mothurOutEndLine(); }
-               
-               }else{ m->mothurOut("Error in reading your fastafile, at position " + toString(fastaString.tellg()) + ". Blank name."); m->mothurOutEndLine(); }
+               }
                
        }
        catch(exception& e) {
@@ -100,11 +101,9 @@ Sequence::Sequence(istringstream& fastaString, string JustUnaligned){
                m = MothurOut::getInstance();
        
                initialize();
-               fastaString >> name;
-               
-               if (name.length() != 0) { 
+               name = getSequenceName(fastaString);
                
-                       name = name.substr(1);
+               if (!m->control_pressed) { 
                        string sequence;
                
                        //read comments
@@ -131,7 +130,7 @@ Sequence::Sequence(istringstream& fastaString, string JustUnaligned){
                        
                        if ((numAmbig / (float) numBases) > 0.25) { m->mothurOut("[WARNING]: We found more than 25% of the bases in sequence " + name + " to be ambiguous. Mothur is not setup to process protein sequences."); m->mothurOutEndLine(); }
                        
-               }else{ m->mothurOut("Error in reading your fastafile, at position " + toString(fastaString.tellg()) + ". Blank name."); m->mothurOutEndLine(); }
+               }
                
        }
        catch(exception& e) {
@@ -147,11 +146,9 @@ Sequence::Sequence(ifstream& fastaFile){
        try {
                m = MothurOut::getInstance();
                initialize();
-               fastaFile >> name;
-               
-               if (name.length() != 0) { 
+               name = getSequenceName(fastaFile);
                
-                       name = name.substr(1); 
+               if (!m->control_pressed) { 
                        
                        string sequence;
                
@@ -181,7 +178,7 @@ Sequence::Sequence(ifstream& fastaFile){
                        
                        if ((numAmbig / (float) numBases) > 0.25) { m->mothurOut("[WARNING]: We found more than 25% of the bases in sequence " + name + " to be ambiguous. Mothur is not setup to process protein sequences."); m->mothurOutEndLine(); }
                        
-               }else{ m->mothurOut("Error in reading your fastafile, at position " + toString(fastaFile.tellg()) + ". Blank name."); m->mothurOutEndLine(); }
+               }
 
        }
        catch(exception& e) {
@@ -195,13 +192,11 @@ Sequence::Sequence(ifstream& fastaFile, string& extraInfo, bool getInfo){
        try {
                m = MothurOut::getInstance();
                initialize();
-               fastaFile >> name;
         extraInfo = "";
                
-               if (name.length() != 0) { 
-            
-                       name = name.substr(1); 
-                       
+               name = getSequenceName(fastaFile);
+               
+               if (!m->control_pressed) {                      
                        string sequence;
             
                        //read comments
@@ -221,7 +216,7 @@ Sequence::Sequence(ifstream& fastaFile, string& extraInfo, bool getInfo){
                        //read info after sequence name
                        while (!fastaFile.eof())        {       
                 char c = fastaFile.get(); 
-                if (c == 10 || c == 13){  break;       }       
+                if (c == 10 || c == 13 || c == -1){  break;    }
                 extraInfo += c;
             } 
                        
@@ -233,8 +228,7 @@ Sequence::Sequence(ifstream& fastaFile, string& extraInfo, bool getInfo){
                        setUnaligned(sequence); 
                        
                        if ((numAmbig / (float) numBases) > 0.25) { m->mothurOut("[WARNING]: We found more than 25% of the bases in sequence " + name + " to be ambiguous. Mothur is not setup to process protein sequences."); m->mothurOutEndLine(); }
-                       
-               }else{ m->mothurOut("Error in reading your fastafile, at position " + toString(fastaFile.tellg()) + ". Blank name."); m->mothurOutEndLine(); }
+               }
         
        }
        catch(exception& e) {
@@ -248,10 +242,9 @@ Sequence::Sequence(ifstream& fastaFile, string JustUnaligned){
        try {
                m = MothurOut::getInstance();
                initialize();
-               fastaFile >> name;
+               name = getSequenceName(fastaFile);
                
-               if (name.length() != 0) { 
-                       name = name.substr(1);
+               if (!m->control_pressed) { 
                        string sequence;
                        
                        //read comments
@@ -279,7 +272,7 @@ Sequence::Sequence(ifstream& fastaFile, string JustUnaligned){
                        
                        if ((numAmbig / (float) numBases) > 0.25) { m->mothurOut("[WARNING]: We found more than 25% of the bases in sequence " + name + " to be ambiguous. Mothur is not setup to process protein sequences."); m->mothurOutEndLine(); }
                        
-               }else{ m->mothurOut("Error in reading your fastafile, at position " + toString(fastaFile.tellg()) + ". Blank name."); m->mothurOutEndLine(); }
+               }
                
        }
        catch(exception& e) {
@@ -287,7 +280,50 @@ Sequence::Sequence(ifstream& fastaFile, string JustUnaligned){
                exit(1);
        }                                                       
 }
-
+//********************************************************************************************************************
+string Sequence::getSequenceName(ifstream& fastaFile) {
+       try {
+               string name = "";
+               
+        fastaFile >> name;
+               
+               if (name.length() != 0) { 
+            
+                       name = name.substr(1); 
+            
+            m->checkName(name);
+            
+        }else{ m->mothurOut("Error in reading your fastafile, at position " + toString(fastaFile.tellg()) + ". Blank name."); m->mothurOutEndLine(); m->control_pressed = true;  }
+        
+               return name;
+       }
+       catch(exception& e) {
+               m->errorOut(e, "Sequence", "getSequenceName");
+               exit(1);
+       }
+}
+//********************************************************************************************************************
+string Sequence::getSequenceName(istringstream& fastaFile) {
+       try {
+               string name = "";
+               
+        fastaFile >> name;
+               
+               if (name.length() != 0) { 
+            
+                       name = name.substr(1); 
+            
+            m->checkName(name);
+            
+        }else{ m->mothurOut("Error in reading your fastafile, at position " + toString(fastaFile.tellg()) + ". Blank name."); m->mothurOutEndLine(); m->control_pressed = true;  }
+        
+               return name;
+       }
+       catch(exception& e) {
+               m->errorOut(e, "Sequence", "getSequenceName");
+               exit(1);
+       }
+}
 //********************************************************************************************************************
 string Sequence::getSequenceString(ifstream& fastaFile, int& numAmbig) {
        try {
@@ -300,7 +336,7 @@ string Sequence::getSequenceString(ifstream& fastaFile, int& numAmbig) {
                        if(letter == '>'){
                                fastaFile.putback(letter);
                                break;
-                       }
+                       }else if (letter == ' ') {;}
                        else if(isprint(letter)){
                                letter = toupper(letter);
                                if(letter == 'U'){letter = 'T';}
@@ -354,7 +390,7 @@ string Sequence::getSequenceString(istringstream& fastaFile, int& numAmbig) {
                        if(letter == '>'){
                                fastaFile.putback(letter);
                                break;
-                       }
+                       }else if (letter == ' ') {;}
                        else if(isprint(letter)){
                                letter = toupper(letter);
                                if(letter == 'U'){letter = 'T';}
@@ -531,6 +567,15 @@ string Sequence::getUnaligned(){
 int Sequence::getNumBases(){
        return numBases;
 }
+//********************************************************************************************************************
+
+int Sequence::getNumNs(){
+    int numNs = 0;
+       for (int i = 0; i < unaligned.length(); i++) {
+        if(toupper(unaligned[i]) == 'N') { numNs++; }
+    }
+    return numNs;
+}
 
 //********************************************************************************************************************
 
@@ -630,12 +675,12 @@ int Sequence::filterToPos(int start){
     
     if (start > aligned.length()) { start = aligned.length(); m->mothurOut("[ERROR]: start to large.\n"); }
     
-       for(int j = 0; j < start-1; j++) {
+       for(int j = 0; j < start; j++) {
                aligned[j] = '.';
        }
        
     //things like ......----------AT become ................AT
-    for(int j = start-1; j < aligned.length(); j++) {
+    for(int j = start; j < aligned.length(); j++) {
         if (isalpha(aligned[j])) { break; }
         else { aligned[j] = '.'; }
     }
@@ -682,7 +727,7 @@ int Sequence::getEndPos(){
 //********************************************************************************************************************
 
 void Sequence::padFromPos(int end){
-       cout << end << '\t' << endPos << endl;
+       //cout << end << '\t' << endPos << endl;
        for(int j = end; j < endPos; j++) {
                aligned[j] = '.';
        }
@@ -719,6 +764,8 @@ void Sequence::trim(int length){
        if(numBases > length){
                unaligned = unaligned.substr(0,length);
                numBases = length;
+        aligned = "";
+        isAligned = 0;
        }
        
 }