def main(): print("Reads a Wikidictionary XML dump and extracts verbs defintions") init_logging() start_time = datetime.datetime.now() definitions = Definitions() definitions.generate('definitions/cawiktionary-latest-pages-meta-current.xml') msg = 'Time {0}'.format(datetime.datetime.now() - start_time) logging.info(msg)
def test_generate(self): with tempfile.TemporaryDirectory() as dirpath: definitions = Definitions() definitions.generate("test/data/definitions.xml", infinitives = 'test/data/infinitives.txt', save_dir = dirpath) def_gen = self._hash_file(os.path.join(dirpath, "definitions.txt")) def_ref = self._hash_file("test/data/definitions.txt") defj_gen = self._hash_file(os.path.join(dirpath, "definitions.json")) defj_ref = self._hash_file("test/data/definitions.json") self.assertEquals(def_ref, def_gen) self.assertEquals(defj_ref, defj_gen)