def load_tree_from_file(filename, tokenize = False): def preprocess_leaf(leaf): leaf = re.sub('_!(.+?)!_', '\\1', leaf) if tokenize: return leaf.split(' ') else: return leaf if filename.endswith('.dis'): pt = rst_lib.load_tree(filename) elif filename.endswith('.tree'): pt = ParseTree.parse(open(filename).read(), leaf_pattern = '_!.+?!_', parse_leaf = preprocess_leaf) return pt