def main(args):
if not args.as_array and not args.as_extra:
args.as_extra = True
for spec in read_specs(args, converters=FROM_FORMAT_NO_PLACEHOLDERS):
kwargs = dict(as_array=args.as_array,
as_extra=args.as_extra,
verbose=args.verbose,
lang=spec.lang)
passages = spec.passages
if spec.conllu:
passages = copy_annotation(passages,
spec.conllu,
by_id=args.by_id,
**kwargs)
elif spec.udpipe:
passages = annotate_udpipe(passages, spec.udpipe, **kwargs)
elif spec.stanfordnlp:
passages = annotate_stanfordnlp(passages, spec.stanfordnlp,
**kwargs)
for passage in annotate_all(passages if args.verbose else tqdm(
passages, unit=" passages", desc="Annotating " + spec.out_dir),
replace=spec.conllu
or not (spec.udpipe or spec.stanfordnlp),
**kwargs):
if passage.extra.get("format") == "amr" and args.as_array:
from semstr.conversion.amr import AmrConverter
AmrConverter.introduce_placeholders(passage)
write_passage(passage,
outdir=spec.out_dir,
verbose=args.verbose,
binary=args.binary)
def to_amr(passage,
metadata=True,
wikification=True,
use_original=True,
verbose=False,
default_label=None,
**kwargs):
""" Convert from a Passage object to a string in AMR PENMAN format (export)
:param passage: the Passage object to convert
:param metadata: whether to print ::id and ::tok lines
:param wikification: whether to wikify named concepts, adding a :wiki triple
:param use_original: whether to use original AMR text from passage.extra
:param verbose: whether to print extra information
:param default_label: label to use in case node has no label attribute
:return list of lines representing an AMR in PENMAN format, constructed from the passage
"""
from semstr.conversion.amr import AmrConverter
return AmrConverter().to_format(passage,
metadata,
wikification,
verbose,
use_original=use_original,
default_label=default_label,
format=kwargs.get("format"))
def from_amr(lines, passage_id=None, return_original=False, save_original=True, wikification=False, placeholders=True,
**kwargs):
"""Converts from parsed text in AMR PENMAN format to a Passage object.
:param lines: iterable of lines in AMR PENMAN format, describing a single passage.
:param passage_id: ID to set for passage, overriding the ID from the file
:param save_original: whether to save original AMR text in passage.extra
:param return_original: return triple of (UCCA passage, AMR string, AMR ID)
:param wikification: whether to use wikification for replacing node labels with placeholders based on tokens
:param placeholders: introduce placeholders into node labels when they include the terminal's text?
:return generator of Passage objects
"""
from semstr.conversion.amr import AmrConverter
return AmrConverter().from_format(lines, passage_id=passage_id, return_original=return_original,
save_original=save_original, wikification=wikification, placeholders=placeholders,
format=kwargs.get("format"))
def from_amr(lines,
passage_id=None,
return_original=False,
save_original=True,
*args,
**kwargs):
"""Converts from parsed text in AMR PENMAN format to a Passage object.
:param lines: iterable of lines in AMR PENMAN format, describing a single passage.
:param passage_id: ID to set for passage, overriding the ID from the file
:param save_original: whether to save original AMR text in passage.extra
:param return_original: return triple of (UCCA passage, AMR string, AMR ID)
:return generator of Passage objects
"""
del args, kwargs
from semstr.conversion.amr import AmrConverter
return AmrConverter().from_format(lines,
passage_id,
return_original=return_original,
save_original=save_original)
def to_amr(passage,
metadata=True,
wikification=True,
use_original=True,
verbose=False,
*args,
**kwargs):
""" Convert from a Passage object to a string in AMR PENMAN format (export)
:param passage: the Passage object to convert
:param metadata: whether to print ::id and ::tok lines
:param wikification: whether to wikify named concepts, adding a :wiki triple
:param use_original: whether to use original AMR text from passage.extra
:param verbose: whether to print extra information
:return list of lines representing an AMR in PENMAN format, constructed from the passage
"""
del args, kwargs
from semstr.conversion.amr import AmrConverter
return AmrConverter().to_format(passage,
metadata,
wikification,
verbose,
use_original=use_original)