def check(self, derivs_file, gs_file):
with open(derivs_file) as f:
with open(gs_file) as gs:
file = f.readlines()
gsdeps_file = gs.readlines()
while file and gsdeps_file:
_, deriv = file.pop(0), file.pop(0)
gsdeps_line = gsdeps_file.pop(0)
if deriv.startswith('#'): continue
t = naive_label_derivation(parse_tree(deriv))
# only take the first two elements (filler lex, arg lex)
deps = set(imap(lambda v: tuple(e.split(IndexSeparator)[0] for e in v[0:2]), mkdeps(t)))
gsdeps = parse_gsdeps(gsdeps_line)
try:
self.assertEqual(deps, gsdeps)
except AssertionError:
print "EXPECTED\n-------"
for depl, depr in sorted(gsdeps):
print depl, depr
print "GOT\n---"
for depl, depr in sorted(deps):
print depl, depr
print "DIFF\n----"
print "false negatives: %s" % ' '.join('|'.join((u, v)) for u, v in list(set(gsdeps) - set(deps)))
print "false positives: %s" % ' '.join('|'.join((u, v)) for u, v in list(set(deps) - set(gsdeps)))
raise
def from_header_and_derivation(header, deriv_string):
"""Creates a Derivation object based on a header line and a derivation representation.
This retrieves the section, document and derivation number from the header line,
expecting it to be of the form
ID=wsj_SSDD.dd PARSER=GOLD NUMPARSE=1"""
matches = re.match(r"ID=wsj_(\d\d)(\d\d).(\d+)", header)
if matches and len(matches.groups()) == 3:
sec_no, doc_no, der_no = [int(i) for i in matches.groups()]
derivation = parse_tree(deriv_string)
return Derivation(sec_no, doc_no, der_no, derivation)
raise CCGbankParseException, "Malformed CCGbank header: %s" % header
def from_header_and_derivation(header, deriv_string):
'''Creates a Derivation object based on a header line and a derivation representation.
This retrieves the section, document and derivation number from the header line,
expecting it to be of the form
ID=wsj_SSDD.dd PARSER=GOLD NUMPARSE=1'''
matches = re.match(r'ID=wsj_(\d\d)(\d\d).(\d+)', header)
if matches and len(matches.groups()) == 3:
sec_no, doc_no, der_no = [int(i) for i in matches.groups()]
derivation = parse_tree(deriv_string)
return Derivation(sec_no, doc_no, der_no, derivation)
raise CCGbankParseException, "Malformed CCGbank header: %s" % header
def check(self, derivs_file, gs_file):
with open(derivs_file) as f:
with open(gs_file) as gs:
file = f.readlines()
gsdeps_file = gs.readlines()
while file and gsdeps_file:
_, deriv = file.pop(0), file.pop(0)
gsdeps_line = gsdeps_file.pop(0)
if deriv.startswith('#'): continue
t = naive_label_derivation(parse_tree(deriv))
# only take the first two elements (filler lex, arg lex)
deps = set(
imap(
lambda v: tuple(
e.split(IndexSeparator)[0] for e in v[0:2]),
mkdeps(t)))
gsdeps = parse_gsdeps(gsdeps_line)
try:
self.assertEqual(deps, gsdeps)
except AssertionError:
print "EXPECTED\n-------"
for depl, depr in sorted(gsdeps):
print depl, depr
print "GOT\n---"
for depl, depr in sorted(deps):
print depl, depr
print "DIFF\n----"
print "false negatives: %s" % ' '.join(
'|'.join((u, v))
for u, v in list(set(gsdeps) - set(deps)))
print "false positives: %s" % ' '.join(
'|'.join((u, v))
for u, v in list(set(deps) - set(gsdeps)))
raise
def testWriteDerivation(self):
tree = parse_tree(self.from_ccgbank)
write_graph(tree, 'ccg_deriv.dot')
self.assert_(os.path.exists('ccg_deriv.dot'))
def testParseInverseParse(self):
deriv_string = '(<T dummycat head_index child_count> (<L Ldummycat1 Lpos1 Lpos2 Llex Lcatfix>) (<L Rdummycat1 Rpos1 Rpos2 Rlex Rcatfix>) )'
self.assertEqual(repr(self.n), deriv_string)
self.assertEqual(self.n, parse_tree(deriv_string))
def testWriteDerivation(self):
tree = parse_tree(self.from_ccgbank)
write_graph(tree, 'ccg_deriv.dot')
self.assert_(os.path.exists('ccg_deriv.dot'))
def testParseInverseParse(self):
deriv_string = '(<T dummycat head_index child_count> (<L Ldummycat1 Lpos1 Lpos2 Llex Lcatfix>) (<L Rdummycat1 Rpos1 Rpos2 Rlex Rcatfix>) )'
self.assertEqual(repr(self.n), deriv_string)
self.assertEqual(self.n, parse_tree(deriv_string))