#!/usr/bin/env python
from dendropy import DnaCharacterMatrix, Tree
import sys

mat=sys.argv[1]
mattype=sys.argv[2]
tre=sys.argv[3]
tretype=sys.argv[4]


d = DnaCharacterMatrix.get(path=mat,
                           schema=mattype)
# make the taxon_namespace mutable,
# so that tree can be read even if different
d.taxon_namespace.is_mutable = True

tree = Tree.get(path=tre,
                schema=tretype,
                preserve_underscores=True,
                taxon_namespace=d.taxon_namespace)
# get all of the taxa associated with tips of the tree, and make sure that
#   they include all of the members of the data's taxon_namespace...
treed_taxa = [i.taxon for i in tree.leaf_nodes()]

if set(treed_taxa) != d.poll_taxa():
    tree_missing = [i.label for i in d.taxon_namespace if i not in treed_taxa]
    emf = 'Some of the taxa in the matrix are not in the tree. Missing "{}"\n'
    em = emf.format('", "'.join(tree_missing))
    raise ValueError(em)

if set(treed_taxa) != d.poll_taxa():
    mat_missing = [i.label for i in d.taxon_namespace if i not in d.poll_taxa()]
    emf = 'Some of the taxa in the tree are not in the data matrix. Missing "{}"\n'
    em = emf.format('", "'.join(mat_missing))
    raise ValueError(em)