def detect(and_or_word):
ao_indices = [word_indices_in_group(and_or_word, group) for group in groups]
for i, aos in enumerate(ao_indices):
num_of_ao = len(aos)
if num_of_ao >= 2:
upper = and_or_word.upper()
# print " %s-%s- found in #%d" % (upper, upper, i), aos
cl = AndOr(and_or_word)
# 区間が確定しているもの（＝最後の１つ以外）をまず追加
for j in range(num_of_ao-1):
this_et_idx = aos[j]
next_et_idx = aos[j+1]
cl.add(words[this_et_idx+1:next_et_idx])
# 最後の１つは、どこで終わるか確かめながら追加
last_et_idx = aos[num_of_ao-1]
end_idx = groups[i][-1] + 1
ws = []
for idx in range(last_et_idx+1, end_idx):
word = words[idx]
first_item = word.items[0]
if cl.pos == 'noun':
# 格変化のある語に限る
if first_item._ is None: break
if non_genitive(first_item._):
cases = [x[0] for x in first_item._]
# これまでの物と格が一致する可能性がなければ排除
cases_x = filter(lambda case:case in cases, cl.cases)
if not cases_x: break
ws.append(word)
idx += 1
if ws:
cl.add(ws)
else:
pass # ERROR: type mismatch
# print "CL: [%d..%d)" % (aos[0], idx)
# cl.dump()
for j in range(aos[0], idx):
# words[j] = None
# visited[j] = True
visited.add(j)
cl.restrict()
words[aos[0]] = cl
ao_loc.add(aos[0])
def detect_genitive_correspondances(words):
M = len(words)
targets = {}
gen = []
blocks = {}
for i in range(M):
word = words[i]
if isinstance(word, Predicate):
pass #blocks[i] = word
elif isinstance(word, AndOr):
# targets[i] = word
if not non_genitive(word._) and word._[0][0] == 'Gen':
gen.append(i)
elif isinstance(word, Word):
if not word.items:
blocks[i] = word
continue
first_item = word.items[0]
if word.surface in (u'et', u'neque') or first_item.pos == 'preposition':
blocks[i] = word
continue
if first_item.pos == 'adj' and first_item.attrib('base') == u'plēnus':
targets[i] = word
continue
if first_item.pos != 'noun':
continue
targets[i] = word
if not non_genitive(first_item._) and first_item._[0][0] == 'Gen':
gen.append(i)
else:
# blocks[i] = word
pass
if not gen: return (words, [])
# print "non-Gen:", [(ix, word.surface_utf8()) for ix,word in targets.items()]
# print " Gen:", gen
def find_target(gen_ix):
def sub(fr, to, step):
for i in range(fr, to, step):
if blocks.has_key(i):
return -1
if not targets.has_key(i):
continue
return i
return -1
pre = sub(gen_ix-1, -1, -1)
if pre >= 0: return pre
post = sub(gen_ix+1, M, 1)
if post >= 0: return post
return -1
gen.sort(reverse=True)
non_gen = set()
for gen_ix in gen:
print "// GEN#%d (%s)" % (gen_ix, words[gen_ix].surface_utf8()),
target_ix = find_target(gen_ix)
if target_ix >= 0:
print "-> TARGET#%d (%s)" % (target_ix, words[target_ix].surface_utf8())
words[target_ix].add_genitive(words[gen_ix])
# words[gen_ix] = None
else:
print "-> no target noun detected"
non_gen.add(gen_ix)
words[gen_ix].restrict_cases(('Nom','Voc','Acc','Dat','Abl','Loc'))
return (words, filter(lambda ix:ix not in non_gen, gen))