def find_verb_subject(tnode: SimpleTree) -> Optional[SimpleTree]:
""" Starting with a verb terminal node, attempt to find
the verb's subject noun phrase """
subj = None
# First, check within the enclosing verb phrase
# (the subject may be embedded within it, as in
# ?'Í dag langaði Páli bróður að fara í sund')
p = tnode.enclosing_tag("VP").enclosing_tag("VP")
if p is not None:
try:
subj = p.NP_SUBJ
except AttributeError:
pass
if subj is None:
# If not found there, look within the
# enclosing IP (inflected phrase) node, if any
p = tnode.enclosing_tag("IP")
if p is not None:
# Found the inflected phrase:
# find the NP-SUBJ node, if any
try:
subj = p.NP_SUBJ
except AttributeError:
pass
return subj
def wrong_verb_use(
self,
match: SimpleTree,
correct_verb: str,
context: ContextType,
) -> None:
""" Annotate wrong verbs being used with nouns,
for instance 'byði hnekki' where the verb should
be 'bíða' -> 'biði hnekki' instead of 'bjóða' """
vp = match.first_match("VP > { %verb }", context)
verb = next(ch for ch in vp.children if ch.tcat == "so").own_lemma_mm
np = match.first_match("NP >> { %noun }", context)
start, end = min(vp.span[0], np.span[0]), max(vp.span[1], np.span[1])
# noun = next(ch for ch in np.leaves if ch.tcat == "no").own_lemma
text = "Hér á líklega að vera sögnin '{0}' í stað '{1}'.".format(
correct_verb, verb)
detail = "Í samhenginu '{0}' er rétt að nota sögnina '{1}' í stað '{2}'.".format(
match.tidy_text, correct_verb, verb)
suggest = ""
self._ann.append(
Annotation(
start=start,
end=end,
code="P002",
text=text,
detail=detail,
suggest=suggest,
))
def wrong_preposition_að(self, match: SimpleTree) -> None:
""" Handle a match of a suspect preposition pattern """
# Find the offending verb phrase
vp = match.first_match("VP > { %verb }", self.ctx_að)
if vp is None:
vp = match.first_match("VP >> { %verb }", self.ctx_að)
# Find the attached prepositional phrase
pp = match.first_match('P > { "að" }')
# Calculate the start and end token indices, spanning both phrases
start, end = min(vp.span[0], pp.span[0]), max(vp.span[1], pp.span[1])
text = "'{0} að' á sennilega að vera '{0} af'".format(vp.tidy_text)
detail = ("Sögnin '{0}' tekur yfirleitt með sér "
"forsetninguna 'af', ekki 'að'.".format(vp.tidy_text))
if match.tidy_text.count(" að ") == 1:
# Only one way to substitute að -> af: do it
suggest = match.tidy_text.replace(" að ", " af ")
else:
# !!! TODO: More intelligent substitution to create a suggestion
suggest = ""
self._ann.append(
Annotation(
start=start,
end=end,
code="P001",
text=text,
detail=detail,
suggest=suggest,
))
def _simple_tree(self, node: Node) -> SimpleTree:
""" Return a SimpleTree instance spanning the deep tree
of which node is the root """
first, last = self._node_span(node)
toklist = self._tokens[first:last + 1]
return SimpleTree.from_deep_tree(node,
toklist,
first_token_index=first)
def simple_parse(text):
""" No-frills parse of text, returning a SimpleTree object """
if not Settings.loaded:
Settings.read("config/Greynir.conf")
with SessionContext(read_only=True) as session:
return SimpleTree(*TreeUtility.parse_text(session, text))