【python爬虫】爬取汽车之家各品牌汽车价格、销量和评论数据

([sS]*?)

def get_ea_sale(self):

try:

ea_sale_data_list = []

for year in range(2018,2021):

for month in range(1, 13):

if month>9:

date_ = str(year) + str(month)

else:

date_ = str(year) + \"0\" + str(month)

for i in range(1,3):

ea_sale_api = f\"/ev-{date_}-{date_}-{i}.html\"

print(ea_sale_api)

response = _html(ea_sale_api)

if not response:

print(\'城市页请求失败\')

return 0

re_no = l(r\'

暂时没有 电动车 d+.d+ 的销量数据

if re_no and len(re_no):

print(\"没第?页\")

break

# 销量数据块

return 0

ea_com_json=()

# print(ea_com_json)

result_list = ea_com_json[\'result\']

for result in result_list:

ea_com_data_list = []

car_id = int(result[\'SeriesId\'])

car_name = result[\'SeriesName\']

print(car_name)

com_api = f\"/{car_id}/index_\"

com_resp = _html(com_api)

if not com_resp:

print(\'?碑列表请求失败\')

continue

#查看?碑的条数

com_num_list = l(r\'共有(d+)条?碑\',com_)

if not com_num_list or not len(com_num_list):

print(\"没有?碑\")

api_ip = \'/switch-ip\'

api_ip_resp = _html(api_ip)

(1)

com_api = f\"/{car_id}/index_\"

com_resp = _html(com_api)

if not com_resp:

print(\'?碑列表请求失败=========\')

continue

# 查看?碑的条数

com_num_list = l(r\'共有(d+)条?碑\', com_)

if not com_num_list or not len(com_num_list):

print(\"没有?碑=========\")

continue

com_num = int(com_num_list[0])

if com_num>15:

#翻页

continue

# 评论id和评论链接

com_id_url_list = l(r\'发表了?碑s+

if not com_id_url_list or not len(com_id_url_list):

print(\"没有?碑id======\")

continue

for com_id_url in com_id_url_list:

com_url = com_id_url

#以时间戳作为评论id

com_id = str(4())

ea_com_data_((car_id,car_name,com_id,com_url))

# 没有列表数据

if not ea_com_data_list or not len(ea_com_data_list):

return 0

print(f\"汽车之家{car_name}评论id已经爬取完成\")

return 1

except:

t_error()