enrichment_done = False
for lang in ef.langs:
filtered_f = ef.get_filtered_file(lang)
enriched_f = ef.get_enriched_file(lang)
if not os.path.exists(enriched_f):
enrichment_done = True
if USE_CONDOR:
model_prefix, name = ef.get_condor_enrich(lang)
run_cmd([p3path, intent_script, 'enrich',
'--align', 'heur,heurpos,giza,gizaheur',
'--pos class', '--parse trans',
filtered_f, enriched_f],
model_prefix, name, False)
else:
enrich(**{ARG_INFILE:filtered_f, ARG_OUTFILE:enriched_f, ALN_VAR:ARG_ALN_METHODS, POS_VAR:ARG_POS_CLASS, PARSE_VAR:ARG_PARSE_TRANS})
if USE_CONDOR and enrichment_done:
condor_wait_notify("Data has been enriched.", email_address, "CONDOR: Enrichment Complete.")
# -------------------------------------------
# 3) Re-project the data...
# -------------------------------------------
projection_done = False
for lang in ef.langs:
for aln_method in aln_methods:
enriched_f = ef.get_enriched_file(lang)
projected_f = ef.get_projected_file(aln_method, lang)
if not os.path.exists(projected_f):
projection_done = True
def test_inst_1(self):
kwargs = {ARG_INFILE:dep_path,
ARG_OUTFILE:'/dev/null',
ALN_VAR:[ARG_ALN_HEUR],
PARSE_VAR:[ARG_PARSE_PROJ, ARG_PARSE_TRANS]}
self.assertIsNone(enrich(**kwargs))
def test_ger(self):
enrich(IN_FILE=ger_file, **no_enrich_args)
except PathArgException as pae: # If we get some kind of invalid file in the arguments, print it and exit.
MAIN_LOG.critical(str(pae))
# sys.stderr.write(str(pae)+'\n')
sys.exit(2)
# Decide on action based on subcommand and args. -------------------------------
# ===============================================================================
# Set verbosity level
# ===============================================================================
logging.getLogger().setLevel(logging.WARNING - 10 * (min(args.verbose, 2)))
# ENRICH
if args.subcommand == CMD_ENRICH:
enrich(**vars(args))
# 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
def test_ctn(self):
enrich(IN_FILE=ctn_file, **no_enrich_args)
def test_kor(self):
enrich(IN_FILE=kor_file, **no_enrich_args)
def test_814(self):
d = all_enrich_args.copy()
d['IN_FILE'] = os.path.join(testfile_dir, 'xigt/814.xml')
self.assertIsNone(enrich(**d))
def test_911(self):
d = all_enrich_args.copy()
d['IN_FILE'] = os.path.join(testfile_dir, 'xigt/multiple_line_tests.xml')
self.assertIsNone(enrich(**d))
def harness(self, name):
d = all_enrich_args.copy()
d['IN_FILE'] = os.path.join(testfile_dir, name)
self.assertIsNone(enrich(**d))
def test_hanging(self):
xp = xigt_testfile('hang_test.xml')
def_enrich_args[ARG_INFILE] = xp
enrich(**def_enrich_args)
def test_encoding(self):
xp = xigt_testfile('encoding-error-test.xml')
def_enrich_args[ARG_INFILE] = xp
enrich(**def_enrich_args)