def set_up(self):
basedir_contents = os.listdir(self.basedir)
files = [f for f in basedir_contents if os.path.isfile(join(self.basedir, f))]
directories = [f for f in basedir_contents if os.path.isdir(join(self.basedir, f))]
if "feature_list.txt" not in files or\
"tree_list.txt" not in files:
sys.exit("You need feature_list.txt and tree_list.txt in the base directory")
if "gold_files" not in directories:
os.mkdir(join(self.basedir, "gold_files"))
if "models" not in directories:
os.mkdir(join(self.basedir, "models"))
if "tagged_files" not in directories:
os.mkdir(join(self.basedir, "tagged_files"))
self.tree_funcs = feature_list_reader(
join(self.basedir, "tree_list.txt"), globals()
)
self.feature_funcs = feature_list_reader(
join(self.basedir, "feature_list.txt"), globals()
)
parser.add_argument("output_dir")
parser.add_argument("file_suffix")
parser.add_argument("tree_list")
parser.add_argument("feature_list")
parser.add_argument("-a",
"--answers",
help="the input file has the answers",
action="store_true")
parser.add_argument("-m",
"--mallet",
help="prepare mallet output",
action="store_true")
all_args = parser.parse_args()
tree_funcs = feature_list_reader(all_args.tree_list, locals())
feature_funcs = feature_list_reader(all_args.feature_list, locals())
if all_args.mallet:
feature_funcs.insert(0, relation_type)
data = get_original_data(all_args.input_file)
f = Featurizer(data, tree_funcs, feature_funcs, not all_args.answers)
if all_args.mallet:
f.build_mallet_features()
f.write_no_tag(all_args.output_dir, all_args.file_suffix)
else:
f.build_features()
if all_args.answers:
f.build_relation_class_vectors()
f.write_multiple_vectors(all_args.output_dir, all_args.file_suffix)
for row in self.new_features:
f_out.write("{}\n".format(" ".join(row)))
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("input_file")
parser.add_argument("output_dir")
parser.add_argument("file_suffix")
parser.add_argument("tree_list")
parser.add_argument("feature_list")
parser.add_argument("-a", "--answers", help="the input file has the answers", action="store_true")
parser.add_argument("-m", "--mallet", help="prepare mallet output", action="store_true")
all_args = parser.parse_args()
tree_funcs = feature_list_reader(all_args.tree_list, locals())
feature_funcs = feature_list_reader(all_args.feature_list, locals())
if all_args.mallet:
feature_funcs.insert(0, relation_type)
data = get_original_data(all_args.input_file)
f = Featurizer(data, tree_funcs, feature_funcs, not all_args.answers)
if all_args.mallet:
f.build_mallet_features()
f.write_no_tag(all_args.output_dir, all_args.file_suffix)
else:
f.build_features()
if all_args.answers:
f.build_relation_class_vectors()
f.write_multiple_vectors(all_args.output_dir, all_args.file_suffix)
