def extract_pos(ef: ExperimentFiles, overwrite=False):
extraction_performed = False
for tc in ef.tagger_configs():
if not os.path.exists(tc.tagger_path()) or overwrite:
if USE_CONDOR:
args = tc.extract_args(True)
condorify(*args)
else:
args, kwargs = tc.extract_args(False)
extract_from_xigt(args, **kwargs)
extraction_performed = True
if extraction_performed and USE_CONDOR:
if condor_email:
condor.condor_wait_notify("Taggers have been extracted.", condor_email, "CONDOR: POS Tagger Extraction Complete")
else:
condor.condor_wait()
# STATS
elif args.subcommand == CMD_STATS:
igt_stats(flatten_list(args.FILE), type='xigt', show_filename=True)
# SPLIT
elif args.subcommand == CMD_SPLIT:
split_corpus(flatten_list(args.FILE), args.train, args.dev, args.test, prefix=args.prefix, overwrite=args.overwrite,
nfold=args.nfold)
# FILTER
elif args.subcommand == CMD_FILTER:
filter_corpus(flatten_list(getattr(args, ARG_INFILE)), getattr(args, ARG_OUTFILE), **vars(args))
# EXTRACT
elif args.subcommand == CMD_EXTRACT:
extract_from_xigt(input_filelist=flatten_list(args.FILE), **vars(args))
# EVAL
elif args.subcommand == CMD_EVAL:
evaluate_intent(flatten_list(args.FILE),
eval_alignment=args.alignment,
eval_ds=args.ds_projection,
eval_posproj=args.pos_projection,
classifier_path=args.classifier,
classifier_feats=args.classifier_feats,
eval_tagger=args.pos_tagger,
gold_tagmap=args.tagmap_gold, trans_tagmap=args.tagmap_trans, outpath=args.output)
# TEXT CONVERT
elif args.subcommand == CMD_TEXT:
xc = text_to_xigtxml(args.FILE)
