def law_item_parse(item):
templates = dict()
lines = item.strip().replace('<p>', '').replace('\u3000', '').split('</p>')
if lines:
first_item = lines[0]
first_item = item_title_filter(first_item)
if len(lines) > 1:
items = lines[1:]
else:
items = []
items = [number_zh_filter(it) for it in items]
ltp_result_dict = ltp_tool(first_item, 'srl')
first_segs = first_item_filter(first_item)
if not first_segs:
return templates
seg = ltp_result_dict['seg']
key_id = 0
if seg:
for n in seg:
if n['word'] in key1: # 应该为关键字列表
key_id = n['id']
break
roles = ltp_result_dict['role']
if roles:
for role in roles[::-1]:
role_type = role['type']
beg = role['beg']
end = role['end']
result = remove_special_character(first_segs[1])
if role_type == 'A0' and end < key_id:
sub = ''.join([n['word'] for n in seg[beg:end + 1]])
condition = remove_special_character(first_segs[0].replace(
sub, ''))
template = st.SentenceTemplate(subject=[sub],
condition=condition,
result=result,
flag=0)
break
elif role_type == 'A1' and end < key_id:
segs_ = remove_special_character(first_segs[0])
template = st.SentenceTemplate(subject=[],
condition=segs_,
result=result,
flag=1)
continue
else:
template = st.SentenceTemplate(subject=[],
condition='',
result=first_segs[1],
flag=1)
if template:
condition, subject, behavior, result = template.parse_items(
items)
templates['condition'], templates['subject'], templates['behavior'], templates['result'] = \
condition, subject, behavior, result
return templates
def law_item_parse_j(lines):
global template
templates = dict()
# 按照</p>拆
lines = lines.strip().replace('<p>', '').replace('\u3000',
'').split('</p>')
# 非空行
if lines:
# first_item是（一）到（十）前面的说明性文字
first_item = lines[0]
# 去掉所有的第……条
first_item = item_title_filter(first_item)
items = []
items_ap = []
if len(lines) > 1:
# 第0行一般是“有下列情形之一的……”
for i, word in enumerate(lines[1:]):
if has_key_one(word):
items.append(word)
else:
items_ap = lines[(i + 1):]
break
# 去掉（一）等
items = [
number_zh_filter(remove_special_character(it)) for it in items
]
# 对第一条进行语意角色标注
# 对应的字典有两个key分别为role 和 seg，role部分是每个词（可能不止一个词，具体几个词由beg和end决定）
# type代表角色对应类型，id代表这个类型的角色在role里服务的个体
ltp_result_dict = ltp_tool(first_item, 'srl')
# ltp_jufa_dict = ltp_parse(first_item, 'parse')
# 这个用来将“有下列情形之一的”的主客体分开
first_segs = first_item_filter(first_item)
if not first_segs:
return templates
seg = ltp_result_dict['seg']
key_id = 0
if seg:
for n in seg:
if n['word'] in key1:
key_id = n['id']
break
roles = ltp_result_dict['role']
# key_id之前的部分是subject，之后的都是result
if roles:
# 逆序roles
for role in roles[::-1]:
role_type = role['type']
beg = role['beg']
end = role['end']
result = remove_special_character(first_segs[1])
# 实际上是直到找到A0为止，否则会一直循环下去
if role_type == 'A0' and end < key_id:
sub = ''.join([n['word'] for n in seg[beg:end + 1]])
condition = remove_special_character(first_segs[0].replace(
sub, ''))
template = st.SentenceTemplate(subject=sub,
condition=condition,
result=result,
flag=0)
break
elif role_type == 'A1' and end < key_id:
segs_ = remove_special_character(first_segs[0])
template = st.SentenceTemplate(subject='',
condition=segs_,
result=result,
flag=1)
continue
else:
template = st.SentenceTemplate(subject='',
condition='',
result=result,
flag=1)
if template:
# condition, subject, behavior, result = template.parse_items(items)
beh = []
for tiao in items:
beh.append(tiao)
templates['condition'], templates['subject'], templates['behavior'], templates['result'] = \
template.condition, template.subject, beh, template.result
return templates
def law_item_parse_j(lines):
global template
templates = dict()
# 按照</p>拆
result_list = []
key_list = []
last_behavior = '' # 如果behavior，说的是同一个behavior
key_item = ''
last_beh = '' # 必须对其所有的或者所经营的船舶、排筏、设施的安全负责，并且应当 '做到' 下列各项 拿出做到 加到behavior上
lines = lines.strip().replace('<p>', '').replace('\u3000',
'').split('</p>')
# 非空行
if lines:
# first_item是（一）到（十）前面的说明性文字
first_item = lines[0]
# 去掉所有的第……条
first_item = item_title_filter(first_item)
items = []
items_ap = []
if len(lines) > 1:
# 第0行一般是“有下列情形之一的……”
for i, word in enumerate(lines[1:]):
if has_key_one(word) or has_key_one_v2(word) or has_key_one_v4(
word): # 有（一）或 一或 １、
items.append(word)
else:
items_ap = lines[(i + 1):]
break
# 去掉（一）等
items = [
number_zh_filter_plus(
number_zh_filter(remove_special_character(it))) for it in items
]
# 对第一条进行语意角色标注
# 对应的字典有两个key分别为role 和 seg，role部分是每个词（可能不止一个词，具体几个词由beg和end决定）
# type代表角色对应类型，id代表这个类型的角色在role里服务的个体
ltp_result_dict = ltp_tool(first_item, 'srl')
# ltp_jufa_dict = ltp_parse(first_item, 'parse')
# 这个用来将“有下列情形之一的”的主客体分开
first_segs = first_item_filter(first_item)
if not first_segs:
return templates
if ltp_result_dict:
seg = ltp_result_dict['seg']
key_id = 0
if seg:
for n in seg:
if n['word'] in key2 or n['word'] in key:
key_id = n['id']
key_item = n['word']
if key_item in key: # 找到key2中的关键词
key_item = ''
continue
if key_item in key2:
break
roles = ltp_result_dict['role']
# key_id之前的部分是subject，之后的都是result
if roles:
# 逆序roles
for role in roles[::-1]:
role_type = role['type']
beg = role['beg']
end = role['end']
result = remove_special_character(first_segs[1]).replace(
key_item, '') # 除去result中的key
# 实际上是直到找到A0为止，否则会一直循环下去
# 优先判断特殊情况
if check_sub_v2(first_segs[0]):
try:
sub = ''
condition = remove_special_character(
(first_segs[0].replace(key_item, '')))
if second_item_filter(
first_segs[0]
): # 未经自治县交通主管部门批准，在乡道和乡道用地范围内，不得从事下列活动：
second_result = second_item_filter(
first_segs[0])
condition = remove_special_character(
second_result[0])
last_beh = second_result[1]
template = st.SentenceTemplate(subject=sub,
condition=condition,
result=result,
flag=0)
except Exception:
write_to_file_append(lines, 'model_1_error.txt')
elif role_type == 'A0' and end < key_id:
sub = ''.join([n['word'] for n in seg[beg:end + 1]])
if sub in sub_list_2:
sub = ''
condition = remove_special_character(
first_segs[0].replace(sub, ''))
if second_item_filter(condition):
second_result = second_item_filter(condition)
condition = remove_special_character(
second_result[0])
last_beh = second_result[1]
template = st.SentenceTemplate(subject=sub,
condition=condition,
result=result,
flag=0)
break
elif role_type == 'A1' and end < key_id:
segs_ = remove_special_character(first_segs[0])
template = st.SentenceTemplate(subject='',
condition=segs_,
result=result,
flag=1)
continue
else:
template = st.SentenceTemplate(subject='',
condition='',
result=result,
flag=1)
if template:
# condition, subject, behavior, result = template.parse_items(items)
beh = []
for i, tiao in enumerate(items):
try:
pre_behavior = ''
if get_sentence_key(tiao):
get_result = get_sentence_key(tiao)
key_list.append(''.join(
s for s in get_result[0]))
behavior = last_beh + remove_last_de(
tiao.replace(key_list[i], ''))
if filter_key_one_behv(
behavior
) and template.subject not in sub_list: # 对于behavior中是result的情况进行了过滤
filter_result = get_result_from_beh(
behavior)
if filter_result:
if filter_result[0]:
pre_behavior = remove_last_de(
remove_special_character(
filter_result[0]))
if not pre_behavior:
pre_behavior = last_behavior
beh.append(pre_behavior)
if filter_result[2]:
result_list.append(
filter_result[1] +
filter_result[2])
else:
result_list.append(
filter_result[1])
last_behavior = pre_behavior
else:
beh.append(behavior)
else:
key_item = '' if key_item == '由' else key_item
key_list.append(key_item)
behavior = last_beh + remove_last_de(tiao)
if filter_key_one_behv(behavior) and filter_key_one_behv_plus(behavior) and template.subject not in sub_list \
and template.condition not in condition_list_2: # 对于behavior中是result的情况进行了过滤
filter_result = get_result_from_beh(
behavior)
if filter_result:
if filter_result[0]:
pre_behavior = remove_last_de(
remove_special_character(
filter_result[0]))
if not pre_behavior:
pre_behavior = last_behavior
beh.append(pre_behavior)
if filter_result[2]:
result_list.append(
filter_result[1] +
filter_result[2])
else:
result_list.append(
filter_result[1])
last_behavior = pre_behavior
else:
beh.append(behavior)
except Exception:
write_to_file_append(lines, 'model_1_error.txt')
if len(template.condition
) <= 1 or template.condition in condition_list:
template.condition = ''
if result_list:
templates['condition'], templates['subject'], templates['key'], templates['behavior'], \
templates['result'] = \
template.condition+template.result, template.subject, key_list, beh, result_list
else:
templates['condition'], templates['subject'], templates['key'], templates['behavior'], templates['result'] = \
template.condition, template.subject, key_list, beh, template.result.replace(template.subject, '')
return templates