|
|
|
from pathlib import Path
|
|
|
|
from corpusparser import Parser
|
|
|
|
import argparse
|
|
|
|
import logging
|
|
|
|
import json
|
|
|
|
|
|
|
|
logging.basicConfig(level=logging.INFO)
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
|
|
|
|
## Main handles command line arguments and writing to files / DB.
|
|
|
|
|
|
|
|
def ssj_to_json_file(sentence_generator, outfile):
|
|
|
|
# this funciton is based on the fact that files are parsed sequentially
|
|
|
|
data_buffer = []
|
|
|
|
for s in sentence_generator:
|
|
|
|
sdata = s[1]
|
|
|
|
data_buffer += [sdata]
|
|
|
|
|
|
|
|
# outfile = Path(outfile)
|
|
|
|
with open(outfile, "w") as fp:
|
|
|
|
logger.info("Writing to {}".format(outfile))
|
|
|
|
json.dump(data_buffer, fp)
|
|
|
|
|
|
|
|
def kres_to_json_files()
|
|
|
|
return "TODO"
|
|
|
|
|
|
|
|
def to_db():
|
|
|
|
return "TODO"
|
|
|
|
|
|
|
|
if __name__ == "__main__":
|
|
|
|
parser = argparse.ArgumentParser(description="Parsing corpora kres and ssj500k.")
|
|
|
|
parser.add_argument('--kres-folder', required=True)
|
|
|
|
parser.add_argument('--kres-srl-folder', required=True)
|
|
|
|
parser.add_argument('--ssj-file', required=True)
|
|
|
|
parser.add_argument('--output', required=False, default=None)
|
|
|
|
parser.add_argument('--outdir', required=False, default=None)
|
|
|
|
parser.add_argument('--dbaddr', required=False, default=None)
|
|
|
|
args = parser.parse_args()
|
|
|
|
|
|
|
|
# parse ssj
|
|
|
|
logger.info("Parsing ssj500k: {}".format(args.ssj_file))
|
|
|
|
ssj_parser = Parser(
|
|
|
|
corpus="ssj",
|
|
|
|
infiles=[args.ssj_file],
|
|
|
|
)
|
|
|
|
# res = [x[1]["sid"] for x in ssj_parser.sentence_generator()]
|
|
|
|
# logger.info("Parsed {} sentences (ssj500k)".format(len(res)))
|
|
|
|
|
|
|
|
# ssj to json
|
|
|
|
ssj_to_json_file(ssj_parser.sentence_generator(), "/home/voje/workdir/ssj.json")
|
|
|
|
|
|
|
|
# parse kres
|
|
|
|
logger.info("Parsing Kres: {}".format(args.ssj_file))
|
|
|
|
kres_parser = Parser(
|
|
|
|
corpus="kres",
|
|
|
|
infiles=[args.kres_folder, args.kres_srl_folder],
|
|
|
|
)
|
|
|
|
res = [x[1]["sid"] for x in kres_parser.sentence_generator()]
|
|
|
|
logger.info("Parsed {} sentences (kres)".format(len(res)))
|
|
|
|
|
|
|
|
|
|
|
|
## Handling output is situational --- implement it outside of Parser.
|
|
|
|
## Parser returns tuples (orig_file, element)
|
|
|
|
# 1. parse per-file and output to file (JSON)
|
|
|
|
# 2. parse and save to DB
|
|
|
|
|
|
|
|
# TODO
|