def parse(cls, raw, item, process=None):
if process is not None:
law_state_list = process.get('lawStateList')
if law_state_list is not None:
tmp_list = []
for law in law_state_list:
mean = law.get('lawStateCNMeaning')
law_date = law.get('prsDate')
part = (ResultItem(table=cls.table_name, title=cls.title[0], value=mean),
ResultItem(table=cls.table_name, title=cls.title[1], value=law_date))
tmp_list.append(part)
item.law_state_list = tmp_list
return item
def parse(cls, raw, item, process=None):
if process is not None:
abstract = BeautifulSoup(
process.get('abstractInfoDTO').get('abIndexList')[0].get(
'value'), 'lxml').text.replace('\n', '').strip()
item.abstract = ResultItem(title=cls.title, value=abstract)
return item
def parse(cls, raw, item, process=None):
if process is not None:
patent_name = process.get('abstractInfoDTO').get('tioIndex').get(
'value')
item.patent_name = ResultItem(title=cls.title,
value=str(patent_name))
return item
def parse(cls, raw, item, process=None):
if process is not None:
item.full_text = ResultItem(
table=cls.table_name,
title=cls.title,
value=BeautifulSoup(
str(process.get('fullTextDTO').get('literaInfohtml')),
'lxml').get_text().replace("'", '"').replace(';', ','))
return item
def push_item(json_list, item: DataItem, title, name):
"""
detail页面的解析函数
:param json_list:
:param item:
:param title:
:param name:
:return:
"""
if json_list is not None:
aitem_list = json_list.get('abstractInfoDTO').get('abstractItemList')
for a_item in aitem_list:
if a_item.get('indexCnName').find(name) != -1:
item.__setattr__(
title, ResultItem(title=name, value=a_item.get('value')))
break
if not hasattr(item, title):
item.__setattr__(title, ResultItem(title=name, value=""))
return item
def parse(cls, raw, item, process=None):
if process is not None:
cognation_list = process.get('cognationList')
# print('cognation', cognation_list)
if cognation_list is not None:
pn_list = []
for cog in cognation_list:
pn_list.append(cog.get('pn'))
item.cognation_list = ResultItem(table=cls.table_name, title=cls.title, value=pn_list)
return item
def parse(cls, raw, item, process=None):
if process is not None:
patent_id = process.find(attrs={'name': 'idHidden'}).get('value')
item.patent_id = ResultItem(title=cls.title, value=str(patent_id))
return item
def push_item(details_str, item: DataItem, title, name):
item.__setattr__(title, ResultItem(title=name, value=details_str))
return item
def parse(cls, raw, item, process:DetaildsRecord=None):
if process is not None:
abstract = process.get_abstract()
item.abstract = ResultItem(title=cls.title, value=abstract)
return item
def parse(cls, raw, item, process: DetaildsRecord=None):
if process is not None:
patent_name = process.get_patent_name()
item.patent_name = ResultItem(title=cls.title, value=str(patent_name))
return item
def parse(cls, raw, item, process=None):
if process is not None:
patent_id = process
item.patent_id = ResultItem(title=cls.title, value=str(patent_id))
return item
