def get_args(rel, comparison_arg):
try:
sbj = _get_sbj(rel)[0].most_specific_head()
if len(comparison_arg) > 1:
veh = comparison_arg[1].most_specific_head()
else:
veh = comparison_arg[0][1].most_specific_head()
if comparison_arg.node == "ADJP":
gnd = comparison_arg.head()[0].head()
veh = comparison_arg[1, 0, 1, 0, 1, 0].head()
elif comparison_arg[0].node == "ADJP":
gnd = comparison_arg.most_specific_head()
veh = comparison_arg[0, 1, 0, 1, 0, 1, 0].head()
else:
gnd = None # we accept partial matches that don't find the gnd
except:
sbj = GlarfTree("", [])
veh = GlarfTree("", [])
gnd = None
return {
"T": sbj[0].print_flat(False, False, False, False) if sbj else "",
"E": rel.head,
"C": comparison_arg.leaves()[0],
"V": veh.print_flat(False, False, False, False) if veh else "",
"P": gnd.print_flat(False, False, False, False) if gnd else "",
}
def get_args(rel, comparison_arg):
try:
sbj = _get_sbj(rel)[0].most_specific_head()
if len(comparison_arg) > 1:
veh = comparison_arg[1].most_specific_head()
else:
veh = comparison_arg[0][1].most_specific_head()
if comparison_arg.node == 'ADJP':
gnd = comparison_arg.head()[0].head()
veh = comparison_arg[1, 0, 1, 0, 1, 0].head()
elif comparison_arg[0].node == 'ADJP':
gnd = comparison_arg.most_specific_head()
veh = comparison_arg[0, 1, 0, 1, 0, 1, 0].head()
else:
gnd = None # we accept partial matches that don't find the gnd
except:
sbj = GlarfTree('', [])
veh = GlarfTree('', [])
gnd = None
return {
'T': sbj[0].print_flat(False, False, False, False) if sbj else "",
'E': rel.head,
'C': comparison_arg.leaves()[0],
'V': veh.print_flat(False, False, False, False) if veh else "",
'P': gnd.print_flat(False, False, False, False) if gnd else "",
}
def __init__(self, glarf_parses, glarf_tuples=None):
if not glarf_tuples:
glarf_tuples = [[] for _ in glarf_parses]
for s, t in zip(glarf_parses, glarf_tuples):
tree = GlarfTree.glarf_parse(s, t)
tree._forest = self
self.append(tree)
# My wife, Mary, is beautiful.
from pyglarf import GlarfTree
apposite = GlarfTree(
'S', [GlarfTree('S-SBJ', [GlarfTree('NP', [GlarfTree('HEAD', [GlarfTree(
'NP', [GlarfTree('T-POS', [GlarfTree('PRP$', ['My', 'i', '0'])]), GlarfTree(
'HEAD', [GlarfTree('NN', ['wife', 'wife', '1'])]), GlarfTree('PTB2-POINTER',
['|0+1|'])])]), GlarfTree('PUNCTUATION', [GlarfTree('|,|', ['|,|', '', '2'])
]), GlarfTree('APPOSITE', [GlarfTree('NP', [GlarfTree('NAME', [GlarfTree(
'NNP', ['Mary', 'mary', '3'])]), GlarfTree('PTB2-POINTER', ['|3+1|']),
GlarfTree('SEM-FEATURE', ['NHUMAN']), GlarfTree('NE-TYPE', ['PERSON']),
GlarfTree('PATTERN', ['NAME']), GlarfTree('INDEX', ['5'])])]), GlarfTree(
'PTB2-POINTER', ['|0+2|']), GlarfTree('INDEX', ['4'])])]), GlarfTree(
'PUNCTUATION1', [GlarfTree('|,|', ['|,|', '', '4'])]), GlarfTree('PRD', [
GlarfTree('VP', [GlarfTree('L-SBJ', [GlarfTree('NP', [GlarfTree('EC-TYPE',
['INF']), GlarfTree('INDEX', ['4'])])]), GlarfTree('HEAD', [GlarfTree('VG',
[GlarfTree('HEAD', [GlarfTree('VBZ', ['is', 'be', '5'])]), GlarfTree(
'P-ARG2', [GlarfTree('ADJP', [GlarfTree('EC-TYPE', ['PB']), GlarfTree(
'INDEX', ['7'])])]), GlarfTree('P-ARG1', [GlarfTree('NP', [GlarfTree(
'EC-TYPE', ['PB']), GlarfTree('INDEX', ['4'])])]), GlarfTree('INDEX',
['6']), GlarfTree('BASE', ['BE']), GlarfTree('VERB-SENSE', ['1']),
GlarfTree('SENSE-NAME', ['"COPULA"'])])]), GlarfTree('PRD', [GlarfTree(
'ADJP', [GlarfTree('HEAD', [GlarfTree('JJ', ['beautiful', 'beautiful', '6'])
]), GlarfTree('PTB2-POINTER', ['|6+1|']), GlarfTree('INDEX', ['7'])])]),
GlarfTree('PTB2-POINTER', ['|5+1|']), GlarfTree('TRANSPARENT', ['T'])])]),
GlarfTree('PUNCTUATION2', [GlarfTree('|.|', ['|.|', '', '7'])]), GlarfTree(
'PTB2-POINTER', ['|0+3|']), GlarfTree('TREE-NUM', ['0']), GlarfTree(
'FILE-NAME', ['"tmp"']), GlarfTree('INDEX', ['0']), GlarfTree(
'SENTENCE-OFFSET', ['0'])])
nps = [np for np, _ in apposite.nps()]
"w",
encoding="utf-8")
matches = 0
dep_matches = 0
for ii, (sent, ctx, gf, gt,
dep) in enumerate(zip(sents, ctxs, gfs, gts, dep_parse)):
dep = dep[1]
if ii % 20 == 0:
print('.')
if f:
f.close()
f = open('bnc_similes/{}/{:03d}.txt'.format(sys.argv[1], ii / 20),
'w',
encoding='utf-8')
try:
tree = GlarfTree.glarf_parse(gf, gt)
args = [get_args(*node) for node in find_comparison_nodes(tree)]
args = [
arg_dict for arg_dict in args
if arg_dict['C'].lower() == sys.argv[1] and 'V' in arg_dict
and arg_dict['V'].strip() != ""
]
except ValueError:
args = []
continue
dep_args = [
m for pat in patterns[:2] for m in match(dep, pat)
if m['C'].form.lower() == sys.argv[1]
]
encoding="utf-8")
only_dep = open("bnc_similes/{}/only_dep.txt".format(sys.argv[1]), "w",
encoding="utf-8")
matches = 0
dep_matches = 0
for ii, (sent, ctx, gf, gt, dep) in enumerate(zip(sents, ctxs, gfs, gts,
dep_parse)):
dep = dep[1]
if ii % 20 == 0:
print '.'
if f:
f.close()
f = open('bnc_similes/{}/{:03d}.txt'.format(sys.argv[1], ii / 20), 'w',
encoding='utf-8')
try:
tree = GlarfTree.glarf_parse(gf, gt)
args = [get_args(*node) for node in find_comparison_nodes(tree)]
args = [arg_dict for arg_dict in args
if arg_dict['C'].lower() == sys.argv[1]
and 'V' in arg_dict and arg_dict['V'].strip() != ""]
except ValueError:
args = []
continue
dep_args = [m for pat in patterns[:2] for m in match(dep, pat)
if m['C'].form.lower() == sys.argv[1]]
if args:
matches += 1
if dep_args:
dep_matches += 1
def test_error_parse():
"""Test that GLARF error parses raise exceptions"""
GlarfTree.glarf_parse("((***ERROR***))")
def test_leaves():
"""Test that all PTB leaves are captured"""
assert_equal(GlarfTree.glarf_parse(test_sentence).print_flat(
pos=False, lemma=False, structure=False),
'In/0 |2003|/1 |,|/2 Yahoo/3 !/4 acquired/5 Overture/6 for/7 '
'$/8 |1.63|/9 billion/10 |.|/11')
def test_unsuccessful_parse():
"""...where the default parse fails"""
GlarfTree.parse(test_sentence)
def test_successful_parse():
"""Test that glarf_parse succeeds..."""
GlarfTree.glarf_parse(test_sentence)