def download_page_info(self, json_data):
if json_data.has_key("poi"):
content = json_data["poi"]
value_list = []
for item in content:
company_name = item["name"]
addr = item["address"]
if item.has_key("tel"):
phone_num = item["tel"]
if phone_num == "":
continue
print(company_name, phone_num, addr)
try:
with open("temp_ts.dat", "a") as f:
f.write(phone_num + "*" + company_name + "\n")
except:
pass
res = self.get_phone_info(phone_num)
time.sleep(0.5)
if res:
province = res[0]
city = res[1]
yys = res[2]
else:
province = "未知"
city = "未知"
yys = "未知"
db = SqlLiteHandle()
sql = "insert into ts_1 values (?,?,?,?,?,?,?)"
db.insert(sql, [(phone_num, company_name, company_name,
addr, province, city, yys)])
db.close()
def download_page_info_nm(self, json_data):
if json_data.has_key("data"):
data = json_data["data"]
if data.has_key("poi_list"):
data_list = data["poi_list"]
for poi in data_list:
company_name = poi["disp_name"]
addr = poi["address"]
phone_num = poi["tel"]
if phone_num:
print(company_name, phone_num, addr)
try:
with open("temp_ts.dat", "a") as f:
f.write(phone_num + "*" + company_name + "\n")
except:
pass
res = self.get_phone_info(phone_num)
time.sleep(5)
if res:
province = res[0]
city = res[1]
yys = res[2]
else:
province = "未知"
city = "未知"
yys = "未知"
db = SqlLiteHandle()
sql = "insert into ts_1 values (?,?,?,?,?,?,?)"
db.insert(sql, [(phone_num, company_name, company_name,
addr, province, city, yys)])
db.close()
def create_table(self):
# time_str = time.time()
sql = "drop table tt_1"
db = SqlLiteHandle()
db.excute(sql)
db.close()
sql = """
CREATE TABLE tt_1 (
"phoneNum" TEXT(20),
"contact" TEXT(200),
"companyName" TEXT(400),
"address" TEXT(600),
"provice" TEXT(40),
"city" TEXT(80),
"yys" TEXT(40));
"""
db = SqlLiteHandle()
db.excute(sql)
db.close()
def download_company_page(self, url):
downloader = PageDownload()
page = downloader.simple_download(url)
if not page:
return False
phone_num = re.findall(self.hq_phoneNum_filter, page)
contact = re.findall(self.hq_contact_filter, page)
company_name = re.findall(self.hq_companyName_filter, page)
address = re.findall(self.hq_address_filter, page)
if company_name:
company_name = company_name[0]
else:
company_name = "未知"
if address:
address = address[0]
else:
address = "未知"
if contact:
contact = contact[0]
else:
contact = "未知"
if phone_num:
phone_num = phone_num[0]
else:
return False
try:
with open("temp_tt.dat", "a") as f:
f.write(phone_num + "*" + contact + "\n")
except:
pass
res = self.get_phone_info(phone_num)
if res:
province = res[0]
city = res[1]
yys = res[2]
else:
province = "未知"
city = "未知"
yys = "未知"
print url, phone_num, contact, company_name, address
db = SqlLiteHandle()
sql = "insert into tt_1 values (?,?,?,?,?,?,?)"
db.insert(
sql,
[(phone_num, contact, company_name, address, province, city, yys)])
db.close()
time.sleep(0.2)
return True
def download_page_info(self, json_data):
if json_data.has_key("searchResult"):
content = json_data["searchResult"]
value_list = []
for item in content:
id = item["id"]
company_name = item["title"]
addr = item["address"]
showType = item["showType"]
try:
url = "http://www.meituan.com/%s/%s/" % (
meituan_cls[showType], id)
except:
continue
phone_num = self.get_phone_num(url)
time.sleep(5)
if phone_num:
print(company_name, phone_num, addr)
try:
with open("temp_ts.dat", "a") as f:
f.write(phone_num + "*" + company_name + "\n")
except:
pass
res = self.get_phone_info(phone_num)
time.sleep(0.5)
if res:
province = res[0]
city = res[1]
yys = res[2]
else:
province = "未知"
city = "未知"
yys = "未知"
db = SqlLiteHandle()
sql = "insert into ts_1 values (?,?,?,?,?,?,?)"
db.insert(sql, [(phone_num, company_name, company_name,
addr, province, city, yys)])
db.close()