def failed_insertion_test(self):
t = IdTree.fromstring('''(ROOT
(SBARQ
(WHNP (WDT What) (NP (NN kind) (PP (IN of) (NP (NNP work,)))))
(SQ (VP (VBZ then?)))))''')
tgt_w = create_words_tier_from_string('kam-a na them lis-no-kha hou')
aln = Alignment([(1, 3), (2, 5), (4, 1), (5, 5)])
project_ps(t, tgt_w, aln)
def ordering_test(self):
"""
This particular tree structure results in changing a child of a tree while iterating through
the children and required a fix such that if such a change is detected, we start iterating
over the children again, so we're not holding onto a stale pointer.
"""
src_t = IdTree.fromstring('''(ROOT (FRAG
(ADVP (RB Probably))
(SBAR (S
(NP (PRP you))
(VP (VBP find)
(ADJP (JJ something))
)
)
)
))''')
tgt_t = IdTree.fromstring('''(ROOT (FRAG
(VBP chitt-u-m)
(ADVP (RB hola))
(UNK ni)
(UNK hou)
(VBP chitt-u-m)
))''')
tgt_w = create_words_tier_from_string('''chitt-u-m hola ni hou chitt-u-m''')
aln = Alignment([(1,2),(3,1),(3,5)])
proj = project_ps(src_t, tgt_w, aln)
self.assertTrue(tgt_t.similar(proj))
def test_projection(self):
proj = project_ps(self.t, create_words_tier_from_string("rhoddodd yr athro lyfr i'r bachgen ddoe"), self.aln)
# Reassign the ids after everything has moved around.
proj.assign_ids()
self.assertEqual(self.proj, proj)
def ctn_merge_2_test(self):
src_t = IdTree.fromstring('''(ROOT
(UCP
(S
(NP (PRP They))
(VP
(VBP are)
(RB also)
(ADJP
(RB too)
(JJ lazy)
(S
(VP
(TO to)
(VP (VB take) (NP (PRP it)) (ADVP (RB out,))))))))
(CC and)
(SBAR
(IN so)
(S
(NP (PRP they))
(VP (VBP do) (RB not) (VP (VB drink) (NP (NN it.))))))))''')
tgt_w = create_words_tier_from_string('loĩs-ma yaŋ hunci-suma kat-a-ŋs-e kina u-tus-u-kV-nɨŋ')
aln = Alignment([(16, 6), (3, 2), (7, 1), (15, 3), (9, 3), (11, 3), (12, 6), (14, 6), (13, 6), (4, 3), (5, 3)])
proj = project_ps(src_t, tgt_w, aln)
self.assertEqual(len(proj.leaves()), 6)
def test_duplicates(self):
"""
Test the case where an English word aligns to multiple language words.
"""
src_t = IdTree.fromstring('(ROOT (SBARQ (WHNP (WP Who)) (SQ (VP (VBZ else?)))))')
tgt_w = create_words_tier_from_string('sa-lo sa-lo')
tgt_t = IdTree.fromstring('(ROOT (SBARQ (WHNP (WP sa-lo) (WP sa-lo))))')
aln = Alignment([(1,1),(1,2)])
result = project_ps(src_t, tgt_w, aln)
self.assertTrue(tgt_t.similar(result))
