blob: 9d7754313dfa2dc7d6968bdfef9349e2ef88c51e [file] [log] [blame]
import argparse
from lib.CoNLL_Annotation import read_conll_generator, CoNLL09_Token
# TODO: Parallelize this for HUGE Files: All sentences can be processed independently
if __name__ == "__main__":
"""
EXAMPLE:
For TreeTagger:
python my_utils/conll_to_tok.py -s /vol/netapp/daza/datasets/TIGER_conll/tiger_release_aug07.corrected.16012013.conll09 -ss "</S>"
For RNNTagger
python my_utils/conll_to_tok.py -s /vol/netapp/daza/datasets/TIGER_conll/tiger_release_aug07.corrected.16012013.conll09
"""
parser = argparse.ArgumentParser()
parser.add_argument("-s", "--src_file", help="CoNLLU File to Convert into the .tok input for RNNTagger/TreeTagger", required=True)
parser.add_argument("-ss", "--sent_sep", help="Special separator to distinguish sentence boundaries", default="")
args = parser.parse_args()
output_file = open(f"{args.src_file}.tok","w")
for conll_obj in read_conll_generator(args.src_file, token_class=CoNLL09_Token):
for tok in conll_obj.tokens:
output_file.write(tok.word+"\n")
output_file.write(args.sent_sep+"\n")