def parse_spacy(passages, lang, verbose=False):
for passage, in annotate_all(zip(passages),
as_array=True,
as_tuples=True,
lang=lang,
verbose=verbose):
terminals = sorted(passage.layer(layer0.LAYER_ID).all,
key=operator.attrgetter("position"))
dep_nodes = [ConlluConverter.Node()] + [
ConlluConverter.Node(t.position,
terminal=t,
token=ConlluConverter.Token(t.text, t.tag))
for t in terminals
]
for dep_node in dep_nodes[1:]:
dep_node.token.paragraph = dep_node.terminal.paragraph
head = Attr.HEAD(dep_node.terminal.tok[Attr.HEAD.value])
if head:
head += dep_node.position
rel = Attr.DEP(dep_node.terminal.tok[Attr.DEP.value],
lang=passage.attrib.get("lang", lang))
assert head is not None and rel is not None, \
"head=%r, rel=%r for token %d in:\n%s" % (head, rel, dep_node.position, " ".join(map(str, terminals)))
edge = ConlluConverter.Edge(head, rel, remote=False)
dep_node.terminal = None
edge.link_head(dep_nodes)
dep_node.add_edges([edge])
parsed = ConlluConverter().build_passage(dep_nodes, passage.ID)
yield passage, parsed
def from_conllu(lines,
passage_id=None,
return_original=False,
annotate=False,
terminals_only=False,
dep=False,
**kwargs):
"""Converts from parsed text in Universal Dependencies format to a Passage object.
:param lines: iterable of lines in Universal Dependencies format, describing a single passage.
:param passage_id: ID to set for passage
:param return_original: return triple of (UCCA passage, Universal Dependencies string, sentence ID)
:param annotate: whether to save dependency annotations in "extra" dict of layer 0
:param terminals_only: create only terminals (with any annotation if specified), no non-terminals
:param dep: return dependency graph rather than converted UCCA passage
:return generator of Passage objects
"""
from semstr.conversion.conllu import ConlluConverter
return ConlluConverter().from_format(lines,
passage_id=passage_id,
return_original=return_original,
annotate=annotate,
terminals_only=terminals_only,
dep=dep,
format=kwargs.get("format"))
def to_conllu(passage, test=False, *args, **kwargs):
""" Convert from a Passage object to a string in Universal Dependencies format (conllu)
:param passage: the Passage object to convert
:param test: whether to omit the head and deprel columns. Defaults to False
:return list of lines representing the semantic dependencies in the passage
"""
del args, kwargs
from semstr.conversion.conllu import ConlluConverter
return ConlluConverter().to_format(passage, test, tree=True)
def to_conllu(passage, test=False, enhanced=True, preprocess=True, **kwargs):
""" Convert from a Passage object to a string in Universal Dependencies format (conllu)
:param passage: the Passage object to convert
:param test: whether to omit the head and deprel columns. Defaults to False
:param enhanced: whether to include enhanced edges
:param preprocess: preprocess the converted dependency graph before returning it?
:return list of lines representing the semantic dependencies in the passage
"""
from semstr.conversion.conllu import ConlluConverter
return ConlluConverter(enhanced=enhanced).to_format(passage, test=test, preprocess=preprocess,
format=kwargs.get("format"))
def from_conllu(lines,
passage_id=None,
split=True,
return_original=False,
annotate=False,
*args,
**kwargs):
"""Converts from parsed text in Universal Dependencies format to a Passage object.
:param lines: iterable of lines in Universal Dependencies format, describing a single passage.
:param passage_id: ID to set for passage
:param split: split each sentence to its own passage?
:param return_original: return triple of (UCCA passage, Universal Dependencies string, sentence ID)
:param annotate: whether to save dependency annotations in "extra" dict of layer 0
:return generator of Passage objects
"""
del args, kwargs
from semstr.conversion.conllu import ConlluConverter
return ConlluConverter().from_format(lines,
passage_id,
split,
return_original=return_original,
annotate=annotate)