def __init__(self, userId, password):
self._qianzhan_client = QianzhanClient(userId, password)
self._company_detail_url_list = []
self._txt = get_1000_txt()
# self._txt_len = len(self._txt)
# self._i = 0
# self._j = 0
logging.info("txt len:->%d" % len(self._txt))
pass
def __init__(self, userId, password):
self._qianzhan_client = QianzhanClient(userId, password)
pass
class Spider(object):
def __init__(self, userId, password):
self._qianzhan_client = QianzhanClient(userId, password)
pass
def _get_company(self, url):
response = self._qianzhan_client.get_company(url)
# print(response.text)
soup = BeautifulSoup(response.text, 'lxml')
company = {}
company.update({
'company_name':
soup.select_one('h1[class="ct_name"]').contents[0]
})
try:
company.update({'url': soup.select_one('a[class="url"]').text})
except Exception, e:
pass
company.update({
'item_update_time':
time.strftime('%Y-%m-%d', time.localtime(time.time()))
})
span_list = soup.select('ul[class="art-basic"] li span[class="info"]')
company.update({
'organization_registration_code': span_list[0].text,
'registration_number': span_list[1].text,
'legal_representative': span_list[2].text,
'business_status': span_list[3].text,
'registered_capital': span_list[4].text,
'business_type': span_list[5].text,
'register_date': span_list[6].text,
'operating_period': span_list[7].text,
'business_address': span_list[8].text,
'business_scope': span_list[9].text
})
span_list_2 = soup.select('ul[class="art-org"] li span[class=""info]')
company.update({
'province': span_list_2[1].text,
'registration_authority': span_list_2[4].text
})
# use
company.update({
'hdencryptCode':
soup.select_one('input[id="hdencryptCode"]')['value'],
'hdoc_area':
soup.select_one('input[id="hdoc_area"]')['value']
})
# logging.debug("company:->%s" % company)
# company.update({'getcommentlist': self._qianzhan_client.post_getcommentlist(company['hdencryptCode'])})
# company.update({'SearchItemCCXX': self._qianzhan_client.post_SearchItemCCXX(company['hdencryptCode'],
# company['hdoc_area'])})
# company.update({'searchitemdftz': self._qianzhan_client.post_searchitemdftz(company['company_name'])})
company.update({
'searchitemnbinfo':
self._qianzhan_client.post_searchitemnbinfo(
company['hdencryptCode'], company['hdoc_area'])
})
if company['searchitemnbinfo'] and len(
company['searchitemnbinfo']) > 0:
company.update({
'searchitemnb':
self._qianzhan_client.post_searchitemnb(
company['hdencryptCode'], company['hdoc_area'],
company['searchitemnbinfo'][0].get('year'))
})
# company.update({'searchitemsite': self._qianzhan_client.post_searchitemsite(company['hdencryptCode'])})
# print "company:->", company
return company