def copy_annotation(passages,
conllu,
by_id=False,
as_array=True,
as_extra=True,
verbose=False,
lang=None):
conllu_sentences = {annotated.ID: annotated for annotated in
get_passages_with_progress_bar(conllu, converters=CONVERTERS, desc="Reading '%s'" % conllu)} \
if by_id else get_passages(conllu, converters=CONVERTERS)
for passage in passages:
try:
annotated = conllu_sentences[passage.ID] if by_id else next(
conllu_sentences)
except (KeyError, StopIteration) as e:
raise ValueError(
"Missing annotation for passage ID '%s', by_id=%s" %
(passage.ID, by_id)) from e
if verbose:
with external_write_mode():
print("Reading annotation from '%s'" % annotated.ID)
if as_array:
passage.layer(layer0.LAYER_ID).docs()[:] = annotated.layer(
layer0.LAYER_ID).docs()
if as_extra:
for terminal, annotated_terminal in zip(
passage.layer(layer0.LAYER_ID).all,
annotated.layer(layer0.LAYER_ID).all):
copy_tok_to_extra(annotated_terminal, terminal, lang=lang)
yield passage
def main(args):
if args.out_dir:
os.makedirs(args.out_dir, exist_ok=True)
if not args.tikz:
import matplotlib
matplotlib.use('Agg')
to_stdout = (args.tikz or args.standoff) and not args.out_dir
t = args.passages
t = get_passages(t) if to_stdout else get_passages_with_progress_bar(
t, desc="Visualizing")
if args.sentences:
t = (sentence for passage in t
for sentence in split2sentences(passage))
for passage in t:
if args.tikz:
print_text(args, visualization.tikz(passage),
passage.ID + ".tikz.txt")
elif args.standoff:
print_text(args, visualization.standoff(passage),
passage.ID + ".ann")
else:
import matplotlib.pyplot as plt
width = len(passage.layer(layer0.LAYER_ID).all) * 19 / 27
plt.figure(passage.ID, figsize=(width, width * 10 / 19))
visualization.draw(passage, node_ids=args.node_ids)
if args.out_dir:
plt.savefig(
os.path.join(args.out_dir, passage.ID + "." + args.format))
plt.close()
else:
plt.show()
def read(fp, text=None, prefix=None):
parent = Path(fp.name).parent
paths = [parent / file.strip() for file in fp]
for passage in get_passages(map(str, paths)):
try:
graph = passage2graph(passage, text, prefix)
except Exception as exception:
print(exception)
continue
yield graph, None
def run(self, guessed: List[str], ref: List[str], **kwargs):
del kwargs
guessed, ref = [{
p.ID: p
for p in get_passages(f, converters=self.converters())
} for f in (guessed, ref)]
stats = SummaryStatistics.aggregate([
self.evaluate(g, ref[i]) for i, g in sorted(guessed.items())
if i in ref
])
stats.print()
def main(args):
passages = list(get_passages(args.filenames))
if args.join_by_prefix:
subsets = defaultdict(list)
for passage in passages:
subsets[passage.ID[:-3]].append(passage)
else:
subsets = {passages[0].ID: passages}
for passage_id, subset in sorted(subsets.items()):
print("Joining passages " + ", ".join(passage.ID for passage in subset), file=sys.stderr)
joined = ucca.convert.join_passages(passages, passage_id=passage_id, remarks=args.remarks)
outfile = "%s/%s.%s" % (args.outdir, args.prefix + joined.ID, "pickle" if args.binary else "xml")
print("Writing joined passage file '%s'..." % outfile, file=sys.stderr)
passage2file(joined, outfile, args.binary)
def read_specs(args, converters=None):
specs = [(pattern, args.out_dir, args.lang, args.udpipe, args.stanfordnlp,
args.conllu, args.join) for pattern in args.filenames]
if args.list_file:
with open(args.list_file, encoding="utf-8") as f:
specs += [l.strip().split() for l in f if not l.startswith("#")]
for spec in specs:
pattern = spec[0]
filenames = sorted(glob(pattern))
if not filenames:
raise IOError("Not found: " + pattern)
yield AnnotationSpecification(
passages=get_passages(filenames, converters=converters),
out_dir=spec[1] if len(spec) > 1 else args.out_dir,
lang=spec[2] if len(spec) > 2 else args.lang,
udpipe=spec[3] if len(spec) > 3 else args.udpipe,
stanfordnlp=spec[4] if len(spec) > 4 else args.stanfordnlp,
conllu=spec[5] if len(spec) > 5 else args.conllu,
join=spec[6] if len(spec) > 6 else args.join)