def determine_feature_value_for_instance(self, igt):
function_list = []
try:
words = igt["w"]
dparse = igt["w-ds"]
except KeyError:
return
found_root = False
found_subj = False
found_obj = False
root_word = ""
if len(dparse) > 0 and dparse[0].text != "root":
return # early exit if it doesn't have a root. unusable
else:
found_root = True
pos = dparse[0]
root_word = pos.attributes["dep"]
word = words[pos.attributes["dep"]]
segmentation = word.segmentation
word_token = ref.resolve(igt, segmentation)
function_list.append(WordPos(word_token, "V", segmentation))
# loop looking for subject and object of that root
for pos in dparse:
if pos.text != "root":
word = words[pos.attributes["dep"]]
segmentation = word.segmentation
word_token = ref.resolve(igt, segmentation)
head_word = pos.attributes["head"]
if pos.text == "nsubj" and not found_subj and head_word == root_word:
function_list.append(WordPos(word_token, "S", segmentation))
found_subj = True
if found_obj:
break # break out of the loop. we have everything we need
elif pos.text == "dobj" and not found_obj and head_word == root_word:
function_list.append(WordPos(word_token, "O", segmentation))
found_obj = True
if found_subj:
break # break out of the loop. we have everything we need
# if we've found something, let's add some probabilities
if found_root and len(function_list) > 1:
sorted_list = sorted(function_list)
word_order = FeatureProbe.list_to_word_order(sorted_list)
if found_subj and found_obj:
self.instance_count += 1
self.order_counts[word_order] += 1.0
if self.debug:
self.debug_log("WORD-ORDER: " + word_order)
def determine_feature_value_for_instance(self, igt):
try:
words = igt["w"]
dparse = igt["w-ds"]
except KeyError:
return
if len(dparse) < 0:
return # early exit. wtf.
# loop looking for all adjectives
for pos in dparse:
if pos.text == "amod":
word = words[pos.attributes["dep"]]
word_segmentation = word.segmentation
word_token = ref.resolve(igt, word_segmentation)
try:
if not self.is_head_noun(pos, igt, dparse):
continue
head_word = words[pos.attributes["head"]]
head_word_segmentation = head_word.segmentation
adj = WordPos(word_token, "adj", word_segmentation)
noun = WordPos(head_word, "noun", head_word_segmentation)
if adj > noun:
self.order_counts["Noun-Adjective"] += 1.0
else: # assume noun>adj
self.order_counts["Adjective-Noun"] += 1.0
self.instance_count += 1
except KeyError:
continue
def test_resolve(self):
self.assertRaises(XigtError, ref.resolve, self.xc1, 'p1')
self.assertRaises(XigtError, ref.resolve, self.xc1[0]['t'], 'p1')
self.assertEqual(ref.resolve(self.xc1[0], 'p1'),
'inu=ga san-biki hoe-ru')
self.assertEqual(ref.resolve(self.xc1[0]['p'], 'p1'),
'inu=ga san-biki hoe-ru')
self.assertEqual(ref.resolve(self.xc1[0], 't1'),
'Three dogs bark.')
self.assertEqual(ref.resolve(self.xc1[0], 'p1[0:6]'),
'inu=ga')
self.assertEqual(ref.resolve(self.xc1[0], 'p1[0:6,7:15]'),
'inu=ga san-biki')
self.assertEqual(ref.resolve(self.xc3[0], 'w1'), 'inu=ga')
self.assertEqual(ref.resolve(self.xc3[0], 'm1'), 'inu')
self.assertEqual(ref.resolve(self.xc3[0], 'g1'), 'dog')
def resolve_ref(self, refattr):
algnexpr = self.attributes[refattr]
if self.tier is None:
raise XigtStructureError(
'Cannot resolve item reference; item (id: {}) is not '
'contained by a Tier.'.format(self.id))
reftier_id = self.tier.attributes[refattr]
if self.igt is None:
raise XigtStructureError(
'Cannot resolve item reference; item\'s tier (id: {}) '
'is not contained by an Igt.'.format(self.tier.id))
reftier = self.igt.get(reftier_id)
if reftier is None:
raise XigtStructureError(
'Referred tier (id: {}) does not exist in the Igt.'.format(
reftier_id))
value = ref.resolve(reftier, algnexpr)
return value
def resolve_ref(self, refattr):
algnexpr = self.attributes[refattr]
if self.tier is None:
raise XigtStructureError(
'Cannot resolve item reference; item (id: {}) is not '
'contained by a Tier.'
.format(self.id)
)
reftier_id = self.tier.attributes[refattr]
if self.igt is None:
raise XigtStructureError(
'Cannot resolve item reference; item\'s tier (id: {}) '
'is not contained by an Igt.'
.format(self.tier.id)
)
reftier = self.igt.get(reftier_id)
if reftier is None:
raise XigtStructureError(
'Referred tier (id: {}) does not exist in the Igt.'
.format(reftier_id)
)
value = ref.resolve(reftier, algnexpr)
return value
def test_resolve(self):
with pytest.raises(XigtError): ref.resolve(self.xc1, 'p1')
with pytest.raises(XigtError): ref.resolve(self.xc1[0]['t'], 'p1')
assert ref.resolve(self.xc1[0], 'p1') == 'inu=ga san-biki hoe-ru'
assert ref.resolve(self.xc1[0]['p'], 'p1') == 'inu=ga san-biki hoe-ru'
assert ref.resolve(self.xc1[0], 't1') == 'Three dogs bark.'
assert ref.resolve(self.xc1[0], 'p1[0:6]') == 'inu=ga'
assert ref.resolve(self.xc1[0], 'p1[0:6,7:15]') == 'inu=ga san-biki'
assert ref.resolve(self.xc3[0], 'w1') == 'inu=ga'
assert ref.resolve(self.xc3[0], 'm1') == 'inu'
assert ref.resolve(self.xc3[0], 'g1') == 'dog'
