cjvt-srl-tagging/tools/parse_all.py

from parser.parser import Parser
import os
from os.path import join, dirname
from pathlib import Path
import re
import sys
import cProfile
import configparser
import logging
from multiprocessing import Pool

SSJ500K_2_1 = 27829  # number of sentences
par = Parser()

# path to data
config = configparser.ConfigParser()
config.read("tools.cfg")
INDIR = Path(config["tools"]["kres_orig"])
OUTDIR = Path(config["tools"]["kres_tsv"])
CPU_CORES = int(config["tools"]["cpu_cores"])

LOGFILE = Path(config["tools"]["logfile"]).absolute()
LOGFILE.touch(exist_ok=True)
LOGFILE.resolve()

logging.basicConfig(filename=str(LOGFILE), level=logging.INFO)

"""
print("parsing ssj")
ssj_file = "../data/ssj500k-sl.sample.xml"
ssj_dict = par.parse_tei(ssj_file)
# assert (len(ssj_dict) == 27829), "Parsed wrong number of sentences."
print("end parsing ssj")
"""

# kres_file = "../data/kres_example/F0019343.xml.parsed.xml"
OUTDIR.mkdir(exist_ok=True)

infiles = list(enumerate([x for x in INDIR.iterdir() if x.is_file()]))
logging.info("Parsing kres: {} files.".format(len(infiles)))

def handle_file(infile):
    i = infile[0]
    kres_file = infile[1]
    outfile = (OUTDIR / kres_file.name).with_suffix(".tsv")

    if outfile.is_file():
        logging.info("Skipping existing file: {}.".format(str(kres_file)))
        return True

    try:
        res_dict = par.parse_tei(kres_file)
        kres_out_str = ""
        for _, sentence in res_dict.items():
            kres_out_str += par.to_conll_2009_SRL(sentence)
    except Exception as exc:
        logging.info("Failed processing file: {}".format(str(kres_file)))
        logging.error(exc)
        return False


    with outfile.open("wb+") as fp:
        fp.write(kres_out_str.encode("utf-8"))
        logging.info("Processed file ({}/{}): {}".format(i+1, len(infiles), str(kres_file)))
        return True
    return False

with Pool(CPU_CORES) as p:
    p.map(handle_file, infiles)


logging.info("end parsing kres")