fixed read.tree so that it can read trees generated by fasttree

[mothur.git] / readtree.cpp

diff --git a/readtree.cpp b/readtree.cpp
index 0d25f7e2f4c4e0232cd137b184da0077585edefb..1dd77f9152b241321f6658d0e0efc025c01a1a71 100644 (file)
--- a/readtree.cpp
+++ b/readtree.cpp
@@ -165,6 +165,7 @@ int ReadNewickTree::read() {
                if (error != 0) { readOk = error; } 
                
                filehandle.close();
+
                return readOk;
        }
        catch(exception& e) {
@@ -237,14 +238,15 @@ int ReadNewickTree::readTreeString() {
                        // ';' means end of tree.                                                                                               
                        else if((ch=filehandle.peek())==';' || ch=='['){                
                                rooted = 1;                                                                     
-                       }                                                                                               
+                       }       
+               
                        if(rooted != 1){                                                                
                                rc = readNewickInt(filehandle, n, T);
                                if (rc == -1) { m->mothurOut("error with rc"); m->mothurOutEndLine(); return -1; } //reports an error in reading
                                if(filehandle.peek() == ')'){                                   
                                        readSpecialChar(filehandle,')',"right parenthesis");
                                }                                                                                       
-                       }                                                                                               
+                       }       
                }
                //note: treeclimber had the code below added - not sure why?
                else{
@@ -263,7 +265,8 @@ int ReadNewickTree::readTreeString() {
                        lc = rc = -1;
                } 
                
-               while(((ch=filehandle.get())!=';') && (filehandle.eof() != true)){;}                                            
+               while(((ch=filehandle.get())!=';') && (filehandle.eof() != true)){;}    
+                                                       
                if(rooted != 1){                                                                        
                        T->tree[n].setChildren(lc,rc);
                        T->tree[n].setBranchLength(0);
@@ -271,6 +274,8 @@ int ReadNewickTree::readTreeString() {
                        if(lc!=-1){             T->tree[lc].setParent(n);               }
                        if(rc!=-1){             T->tree[rc].setParent(n);               }
                }
+               
+               //T->printTree(); cout << endl;
                return 0;
        
        }
@@ -287,22 +292,50 @@ int ReadNewickTree::readNewickInt(istream& f, int& n, Tree* T) {
                if (m->control_pressed) { return -1; } 
                
                int c = readNodeChar(f);
-    
+
                if(c == '('){
-                       int lc = readNewickInt(f, n, T);
-                       if (lc == -1) { return -1; } //reports an error in reading
+               
+                       //to account for multifurcating trees generated by fasttree, we are forcing them to be bifurcating
+                       //read all children
+                       vector<int> childrenNodes;
+                       while(f.peek() != ')'){
+                               int child = readNewickInt(f, n, T);
+                               if (child == -1) { return -1; } //reports an error in reading
+               //cout << "child = " << child << endl;          
+                               childrenNodes.push_back(child);
+                               
+                               //after a child you either have , or ), check for both
+                               if(f.peek()==')'){  break;  }
+                               else if (f.peek()==',') {   readSpecialChar(f,',',"comma");  }
+                               else {;}
+                       }
+       //cout << childrenNodes.size() << endl;         
+                       if (childrenNodes.size() < 2) {  m->mothurOut("Error in tree, please correct."); m->mothurOutEndLine(); return -1; }
+                       
+                       //then force into 2 node structure
+                       for (int i = 1; i < childrenNodes.size(); i++) {
+                       
+                               int lc, rc;
+                               if (i == 1) { lc = childrenNodes[i-1]; rc = childrenNodes[i]; }
+                               else { lc = n-1; rc = childrenNodes[i]; }
+                       //cout << i << '\t' << lc << '\t' << rc << endl;        
+                               T->tree[n].setChildren(lc,rc);
+                               T->tree[lc].setParent(n);
+                               T->tree[rc].setParent(n);
+                               
+                               //T->printTree(); cout << endl;
+                               n++;
+                       }
+                       
+                       //to account for extra ++ in looping
+                       n--;
                        
-                       readSpecialChar(f,',',"comma");
-
-                       int rc = readNewickInt(f, n, T);
-                       if (rc == -1) { return -1; }  //reports an error in reading     
                        if(f.peek()==')'){      
                                readSpecialChar(f,')',"right parenthesis");     
                                //to pass over labels in trees
                                c=filehandle.get();
-                               while((c!=',') && (c != -1) && (c!= ':') && (c!=';')){ c=filehandle.get(); }
+                               while((c!=',') && (c != -1) && (c!= ':') && (c!=';')&& (c!=')')){ c=filehandle.get(); }
                                filehandle.putback(c);
-
                        }                       
                
                        if(f.peek() == ':'){                                                                          
@@ -314,12 +347,14 @@ int ReadNewickTree::readNewickInt(istream& f, int& n, Tree* T) {
                        }else{
                                T->tree[n].setBranchLength(0.0); 
                        }                                               
-               
-                       T->tree[n].setChildren(lc,rc);
-                       T->tree[lc].setParent(n);
-                       T->tree[rc].setParent(n);
-               
+                       
+                       //T->tree[n].setChildren(lc,rc);
+                       //T->tree[lc].setParent(n);
+                       //T->tree[rc].setParent(n);
+                       //T->printTree();  cout << endl;
+                       
                        return n++;
+               
                }else{
                        f.putback(c);
                        string name = "";
@@ -328,7 +363,7 @@ int ReadNewickTree::readNewickInt(istream& f, int& n, Tree* T) {
                                name += d;
                                d=f.get();
                        }
-               
+//cout << name << endl;
                        int blen = 0;
                        if(d == ':')    {               blen = 1;       }               
                
@@ -372,6 +407,7 @@ int ReadNewickTree::readNewickInt(istream& f, int& n, Tree* T) {
                        }
                
                        while((c=f.get())!=0 && (c != ':' && c != ',' && c!=')') )              {;}             
+       
                        f.putback(c);
                
                        return n1;