def get_company_info(self, response):
compass_name = response.xpath(
'//input[@id="corpName"]/@value').extract_first()
base_nodes = response.xpath('//div[@class="t_end"]/ul//tr/td/text()')
info_list = []
for i, node in enumerate(base_nodes):
if i % 2 == 0:
continue
info_list.append(node.extract())
[honor_code, representative, compass_type, provice,
operating_addr] = info_list
company_item = CompassItem({ # 自动检查key是否合法
'compass_name': compass_name,
'compass_link': response.url,
'honor_code': honor_code,
'representative': representative,
'compass_type': compass_type,
'provice': provice,
'operating_addr': operating_addr,
'establish_time': None,
'register_capital': None,
'net_asset': None,
# 'crawl_time': self.fmt_time()
})
return [company_item]
def get_company_info(self, response):
nodes = response.xpath('//div[@class="basic_infor"]//tbody/tr')
compass_name = nodes[0].xpath(
'./td[@class="name_level3"]/text()').extract_first().strip()
honor_code = nodes[1].xpath(
'./td[@id="LicenseNum"]/text()').extract_first().strip()
representive = nodes[2].xpath(
'./td[@id="LegalMan"]/text()').extract_first().strip()
compass_type = nodes[2].xpath(
'./td[@id="EconType"]/text()').extract_first().strip()
provice = nodes[3].xpath(
'./td[@id="Td1"]/text()').extract_first().strip()
operating_addr = nodes[3].xpath(
'./td[@id="Description"]').extract_first().strip()
company_item = CompassItem({ # 自动检查key是否合法
'compass_name': compass_name,
'compass_link': response.url,
'honor_code': honor_code, # 信用代码
'representative': representive, # 法人
'compass_type': compass_type, # 公司类型
'provice': provice,
'operating_addr': operating_addr, # 运营地址
'establish_time': None,
'register_capital': None,
'net_asset': None,
})
return [company_item]
def parse_compass_info(self, unit, url):
company_item = CompassItem({ # 自动检查key是否合法
'compass_name': unit['CorpName'],
'compass_link': url,
'honor_code': unit['CorpCode'], # 信用代码
'representative': unit['LegalMan'], # 法人
'compass_type': unit['EconomicNum'], # 公司类型
'provice': ''.join(unit['AreaName'].split('·')[:1]),
'operating_addr': unit['Address'], # 运营地址
'establish_time': 'None',
'register_capital': unit['RegPrin'],
'net_asset': None,
})
return [company_item]
def get_company_info(self, response):
nodes = response.xpath('//div[@class="basic_infor"]//tbody/tr')
company_item = CompassItem({ # 自动检查key是否合法
'compass_name': nodes.xpath('./td[@class="name_level3"]').extract()[0],
'compass_link': response.url,
'honor_code': nodes.xpath('./td[@id="LicenseNum"]').extract()[0], # 信用代码
'representative': nodes.xpath('./td[@id="LegalMan"]').extract()[0], # 法人
'compass_type': nodes.xpath('./td[@id="EconType"]').extract()[0], # 公司类型
'provice': ''.join(nodes.xpath('./td[@id="Td1"]').extract()),
'operating_addr': ''.join(nodes.xpath('./td[@id="Description"]')), # 运营地址
'establish_time': None,
'register_capital': None,
'net_asset': None,
})
return [company_item]
def extract_compass_info(self, resp_detail, com_rules):
response = resp_detail
node = response.xpath(com_rules.get('cnodes')[0])[0]
company_item = CompassItem()
company_item['compass_link'] = response.url
for k, v in com_rules.items():
if 'node' in k:
continue
rule, map_key = v[0], v[1]
if v[0] is None:
company_item[map_key] = ''
else:
company_item[map_key] = node.xpath(
rule).extract_first().replace('\n', '').replace(
'\t', '').replace('\r', '').replace(' ', '')
return [company_item]
def get_company_info(self, response):
compass_name = ''.join(
response.xpath(
'//div[@class="tLayer-1"]/h3/text()').extract()).strip()
honor_code, register_capital = response.xpath(
'//div[@class="tLayer-1"]/table/tr[1]/td[not(@class)]/text()'
).extract()
honor_code = 'None' if len(honor_code) < 7 else honor_code
representive = ''.join(
response.xpath(
'//div[@class="tLayer-1"]/table/tr[2]/td[not(@class)][1]/text()'
).extract())
compass_type = response.xpath(
'//div[@class="tLayer-1"]/table/tr[3]/td[not(@class)]/text()'
).extract()[0]
establish_time = ''.join(
response.xpath(
'//div[@class="tLayer-1"]/table/tr[4]/td[not(@class)][2]/text()'
).extract()).strip()
provice = ''.join(
response.xpath(
'//div[@class="tLayer-1"]/table/tr[5]/td[not(@class)][2]/text()'
).extract())
operating_addr = ''.join(
response.xpath(
'//div[@class="tLayer-1"]/table/tr[6]/td[not(@class)][1]/text()'
).extract())
company_item = CompassItem({ # 自动检查key是否合法
'compass_name': compass_name,
'compass_link': response.url,
'honor_code': honor_code, # 信用代码
'representative': representive, # 法人
'compass_type': compass_type, # 公司类型
'provice': provice,
'operating_addr': operating_addr, # 运营地址
'establish_time': establish_time,
'register_capital': register_capital,
'net_asset': None,
})
# print company_item
return [company_item]