def process(self, text):
preproc = TextTo4lang.preprocess_text(text)
deps, corefs, parse_trees = self.parser_wrapper.parse_text(preproc)
machines = self.dep_to_4lang.get_machines_from_deps_and_corefs(
deps, corefs)
# print machines
self.dep_to_4lang.lexicon.expand(machines)
graph = MachineGraph.create_from_machines(machines.values())
print graph.to_dot()
def text_to_4lang_demo(self, text, expand, fn='pic', dep_fn='deps'):
preproc_sen = TextTo4lang.preprocess_text(text.strip().decode('utf-8'))
deps, corefs = self.parser_wrapper.parse_text(preproc_sen)
words2machines = self.dep_to_4lang.get_machines_from_deps_and_corefs(
deps, corefs)
# TODO
orig_machines = set()
for machine in words2machines.itervalues():
orig_machines |= set(MachineTraverser.get_nodes(
machine, names_only=False, keep_upper=True))
# orig_machines = set([m.printname() for m in words2machines.values()])
# logging.info(u'orig_machines: {0}'.format(
# [m.printname() for m in orig_machines]))
if expand:
self.dep_to_4lang.lexicon.expand(words2machines)
pic_path = draw_text_graph(
words2machines, self.tmp_dir, fn=fn,
orig_machines=orig_machines)
dep_path = draw_dep_graph(deps[0], self.tmp_dir, dep_fn)
# deps_table = self.get_dep_table(deps[0])
return os.path.basename(dep_path), os.path.basename(pic_path)