pre-david commit

src/PhyloNetwork/classes.py

@@ -62,6 +62,13 @@ def node_by_taxa(self,taxa):
                 return node
         return None
 
+    @memoize_method
+    def all_nodes_by_taxa(self,taxa):
+        """
+        Returns all nodes labelled by taxa.
+        """
+        return [node for node in self.labelled_nodes() if self.label(node)==taxa]
+
     def is_tree_node(self,u):
         """
         Returns True if u is a tree node, False otherwise.
@@ -194,10 +201,11 @@ def walk(self,parsed,ignore_prefix=None):
             return internal_label
 
     def from_eNewick(self,string,ignore_prefix=None):
+        #parsed=eNewickParser(string)[0]
         try:
             parsed=eNewickParser(string)[0]
         except pyparsing.ParseException:
-            print 'Malformed eNewick'
+            raise 'Malformed eNewick'
             #raise JoQueSe
             return False
         self.walk(parsed,ignore_prefix=ignore_prefix)
@@ -328,8 +336,8 @@ def topological_restriction(self,subtaxa):
         restricted=copy.deepcopy(self)
         for taxon in restricted.taxa():
             if not taxon in subtaxa:
-                u=restricted.node_by_taxa(taxon)
-                del restricted._labels[u]
+                for u in restricted.all_nodes_by_taxa(taxon):
+                    del restricted._labels[u]
         restricted.cache = {}
 
         while True:

src/PhyloNetwork/distances.py

@@ -55,13 +55,15 @@ def nodal_distance_unsplitted(net1,net2,p=1,take_root=False):
         mat=net1.nodal_matrix()-net2.nodal_matrix()
     except:
         raise Exception("Networks over different set of taxa")
+    n = len(net1.taxa())
+    result = sum([ (mat1[i,j]+mat1[j,i]-mat2[i,j]-mat2[j,i])**p for i in range(n) for j in range(i+1,n)])
     if p==1:
-        return sum(abs(mat.flatten()))/2
+        return result
     else:
         if take_root:
-            return (sum(abs(mat.flatten())**p)/2)**(1.0/p)
+            return (result)**(1.0/p)
         else:
-            return (sum(abs(mat.flatten())**p)/2)
+            return result
 
 def cophenetic_distance(net1,net2,p=1,take_root=False):
     mat1=net1.cophenetic_matrix()
@@ -71,7 +73,7 @@ def cophenetic_distance(net1,net2,p=1,take_root=False):
     except:
         raise Exception("Networks over different set of taxa")
     if p==1:
-        return sum(abs(mat.flatten()))/2
+        return sum(abs(mat.flatten()))
     else:
         if take_root:
             return (sum(abs(mat.flatten())**p))**(1.0/p)

src/PhyloNetwork/eNewick.py

@@ -18,7 +18,7 @@ def makeeNewickParser():
     sharp     = Literal("#").suppress()
 
     # terminal
-    name    = Word(alphanums + alphas8bit + "_" + "-" + "." + "+" + "~" + "{" + "}" + "*" )
+    name    = Word(alphanums + alphas8bit + "_" + "-" + "." + "+" + "&" + "/" + "~" + "{" + "}" + "*" + "'" + '"' + '\\' + '?')
     string  = Word(alphas)
     fnumber = Combine(
         Word("+-"+nums, nums) + 

src/PhyloNetwork/generators.py

@@ -288,6 +288,22 @@ def Tree_generator_random(taxa,binary=False,nested_taxa=True,id_offset=0):
     while True:
         yield f(taxa,id_offset)
 
+def tree_generator_random_yule(taxa):
+    n = len(taxa)
+    if n == 1:
+        return PhyloNetwork(eNewick=('%s;' % taxa[0]))
+    tree = PhyloNetwork(eNewick=('(%s,%s);' % (taxa[0],taxa[1])))
+    for i in range(2,n):
+        leaf = random.choice(tree.leaves())
+        tree = push_and_hang(tree,leaf,taxa[i])
+    labels = tree._labels
+    keys = labels.keys()
+    values = labels.values()
+    random.shuffle(values)
+    tree._labels = dict(zip(keys,values))
+    tree.cache = {}
+    return tree
+
 if __name__ == "__main__":
     tg = Tree_generator(['1','2','3'])
     while tg: