from makePickle import pickleLoad
for bun in pickleLoad('outchunk.pickle'):
if not bun:
continue
for chunk in bun:
if chunk.dst == -1 \
or '名詞' not in {morph.pos for morph in chunk.morphs} \
or '動詞' not in {morph.pos for morph in bun[chunk.dst].morphs}:
# 係り先が無い時 or 係り元に「名詞」を含まない時 or 係り先に「動詞」を含まない時
continue
print(''.join(
[morph.surface
for morph in chunk.morphs if not morph.pos == '記号']) + '\t' +
''.join([
morph.surface
for morph in bun[chunk.dst].morphs if not morph.pos == '記号'
]))
from makePickle import pickleLoad
input_sentence = pickleLoad('outchunk.pickle')
output_sentence = []
for bun in input_sentence:
if not bun: # bunが空（[]）の時
continue
strings_list = []
for chunks in bun:
if chunks.dst == -1: # 係り先が無い時
continue
origin = ''.join([
morph.surface for morph in chunks.morphs if not morph.pos == '記号'
]) # 記号を除去
end = ''.join([
morph.surface for morph in bun[chunks.dst].morphs
if not morph.pos == '記号'
]) # 上に同じ
strings_list.append(origin + end)
string = '\t'.join(strings_list) # タブ区切りで文字列にする
output_sentence.append(string)
for sentence in output_sentence:
print(sentence)
from makePickle import pickleLoad
lines = pickleLoad('out.pickle')
for l in lines[2]:
print(l.surface, l.base, l.pos, l.pos1)
chunk = Chunk()
chunk.dst = int(status[2][:-1]) # D抜いてintにキャスト
temp_dic[chunk.dst].append(index) # 係り元を集計
else:
if 'EOS' in line: # EOSの時
chunk.srcs = temp_dic[index]
inner_list.append(chunk)
outer_list.append(inner_list[1:])
chunk = Chunk()
inner_list = []
temp_dic = defaultdict(list)
index = -1
else:
elements = re.split('[,\t]', line.rstrip())
morph = Morph()
morph.surface = elements[0]
morph.base = elements[7]
morph.pos = elements[1]
morph.pos1 = elements[2]
chunk.morphs.append(morph)
return outer_list
if __name__ == '__main__':
chunk8 = pickleLoad('outchunk.pickle')[7]
df = pd.DataFrame(
[[''.join([morph.surface for morph in chunk.morphs]),
str(chunk.dst)] for chunk in chunk8],
columns=['文節', '係り先'])
print(df)
import makePickle as mp
from k40 import create_morph
mp.pickleDump(create_morph(), 'out')
dic = mp.pickleLoad('out.pickle')
print(list(map(lambda x: x.surface, dic[2])))