done creating kres and ssj json files

dev
voje 5 years ago
parent 83c24609bf
commit eb83519f51

@ -12,7 +12,7 @@ SSJ_FILE = "$(MAKE_ROOT)/data/samples/ssj_example/ssj500k-sl.body.sample.xml"
KRES_FOLDER = "$(MAKE_ROOT)/data/samples/kres_example" KRES_FOLDER = "$(MAKE_ROOT)/data/samples/kres_example"
KRES_SRL_FOLDER = "$(MAKE_ROOT)/data/kres_srl" KRES_SRL_FOLDER = "$(MAKE_ROOT)/data/kres_srl"
OUTPUT = "file" OUTPUT = "file"
OUTDIR = "$(HOME)/workdir/outputfolder" OUTDIR = "/home/voje/workdir/test_out"
DBADDR = "" DBADDR = ""
export export

@ -9,20 +9,52 @@ logger = logging.getLogger(__name__)
## Main handles command line arguments and writing to files / DB. ## Main handles command line arguments and writing to files / DB.
def ssj_to_json_file(sentence_generator, outfile): def ssj_to_json_file(sentence_generator, outfolder):
# this funciton is based on the fact that files are parsed sequentially # this funciton is based on the fact that files are parsed sequentially
outfolder = Path(outfolder)
outfolder.mkdir(parents=True, exist_ok=True)
outfile = outfolder / "ssj500k.json"
data_buffer = [] data_buffer = []
for s in sentence_generator: for s in sentence_generator:
sdata = s[1] sdata = s[1]
data_buffer += [sdata] data_buffer += [sdata]
# outfile = Path(outfile) # outfile = Path(outfile)
with open(outfile, "w") as fp: with outfile.open("w") as fp:
logger.info("Writing to {}".format(outfile)) logger.info("Writing to {}".format(outfile))
json.dump(data_buffer, fp) json.dump(data_buffer, fp)
def kres_to_json_files() def kres_to_json_files(sentence_generator, outfolder):
return "TODO" outfolder = Path(outfolder) / "kres_json"
outfolder.mkdir(parents=True, exist_ok=True)
def write_buffer_to_file(outfile, outfile_buffer):
logger.info("Writing file: {}".format(outfile))
with outfile.open("w") as fp:
json.dump(outfile_buffer, fp)
outfile_buffer = None
current_outfile = None
for s in sentence_generator:
infile = s[0]
outfile = outfolder / Path(infile.name.split(".")[0]).with_suffix(".json")
# parser sequentially parses files; when we're done with a file, write it out
if current_outfile is None:
current_outfile = outfile
outfile_buffer = []
elif outfile != current_outfile:
write_buffer_to_file(current_outfile, outfile_buffer)
current_outfile = outfile
outfile_buffer = []
# update buffer
sdata = s[1]
outfile_buffer += [sdata]
write_buffer_to_file(current_outfile, outfile_buffer)
def to_db(): def to_db():
return "TODO" return "TODO"
@ -47,7 +79,7 @@ if __name__ == "__main__":
# logger.info("Parsed {} sentences (ssj500k)".format(len(res))) # logger.info("Parsed {} sentences (ssj500k)".format(len(res)))
# ssj to json # ssj to json
ssj_to_json_file(ssj_parser.sentence_generator(), "/home/voje/workdir/ssj.json") ssj_to_json_file(ssj_parser.sentence_generator(), args.outdir)
# parse kres # parse kres
logger.info("Parsing Kres: {}".format(args.ssj_file)) logger.info("Parsing Kres: {}".format(args.ssj_file))
@ -55,8 +87,11 @@ if __name__ == "__main__":
corpus="kres", corpus="kres",
infiles=[args.kres_folder, args.kres_srl_folder], infiles=[args.kres_folder, args.kres_srl_folder],
) )
res = [x[1]["sid"] for x in kres_parser.sentence_generator()] # res = [x[1]["sid"] for x in kres_parser.sentence_generator()]
logger.info("Parsed {} sentences (kres)".format(len(res))) # logger.info("Parsed {} sentences (kres)".format(len(res)))
# kres to json
kres_to_json_files(kres_parser.sentence_generator(), args.outdir)
## Handling output is situational --- implement it outside of Parser. ## Handling output is situational --- implement it outside of Parser.

Loading…
Cancel
Save