提交 5899c4fb 作者: 薛凌堃

天眼查企业动态维护

上级 d43a13da
import json import json
import random
import requests, time, pymysql import requests, time, pymysql
import jieba import jieba
import sys import sys
...@@ -45,24 +47,21 @@ def beinWork(tyc_code, social_code,start_time): ...@@ -45,24 +47,21 @@ def beinWork(tyc_code, social_code,start_time):
retData = {'total': 0, 'up_okCount': 0, 'up_errorCount': 0, 'up_repetCount': 0} retData = {'total': 0, 'up_okCount': 0, 'up_errorCount': 0, 'up_repetCount': 0}
t = time.time() t = time.time()
url = f'https://capi.tianyancha.com/cloud-yq-news/company/detail/publicmsg/news/webSimple?_={t}&id={tyc_code}&ps={pageSize}&pn=1&emotion=-100&event=-100' url = f'https://capi.tianyancha.com/cloud-yq-news/company/detail/publicmsg/news/webSimple?_={t}&id={tyc_code}&ps={pageSize}&pn=1&emotion=-100&event=-100'
for m in range(0, 3): try:
try: for m in range(0, 3):
ip = baseCore.get_proxy() ip = baseCore.get_proxy()
headers['User-Agent'] = baseCore.getRandomUserAgent() headers['User-Agent'] = baseCore.getRandomUserAgent()
response = requests.get(url=url, headers=headers, proxies=ip, verify=False) response = requests.get(url=url, headers=headers, proxies=ip, verify=False)
# time.sleep(random.randint(3, 5)) time.sleep(random.randint(3, 5))
break break
except Exception as e: if (response.status_code == 200):
pass pass
except Exception as e:
if (response.status_code == 200):
pass
else:
log.error(f"{tyc_code}-----获取总数接口失败") log.error(f"{tyc_code}-----获取总数接口失败")
e = '获取总数接口失败' error = '获取总数接口失败'
state = 0 state = 0
takeTime = baseCore.getTimeCost(start_time, time.time()) takeTime = baseCore.getTimeCost(start_time, time.time())
baseCore.recordLog(social_code, taskType, state, takeTime, url, e) baseCore.recordLog(social_code, taskType, state, takeTime, url, f'{error}----{e}')
return retData return retData
try: try:
json_1 = json.loads(response.content.decode('utf-8')) json_1 = json.loads(response.content.decode('utf-8'))
...@@ -177,7 +176,7 @@ def beinWork(tyc_code, social_code,start_time): ...@@ -177,7 +176,7 @@ def beinWork(tyc_code, social_code,start_time):
pass pass
continue continue
try: try:
insert_sql = '''insert into brpa_source_article(social_credit_code,source_address,origin,type,create_time) values(%s,%s,%s,%s,now())''' insert_sql = '''insert into brpa_source_article(social_credit_code,source_address,origin,type,publish_time,create_time) values(%s,%s,%s,%s,%s,now())'''
# 动态信息列表 # 动态信息列表
up_okCount = up_okCount + 1 up_okCount = up_okCount + 1
list_info = [ list_info = [
...@@ -185,6 +184,7 @@ def beinWork(tyc_code, social_code,start_time): ...@@ -185,6 +184,7 @@ def beinWork(tyc_code, social_code,start_time):
link, link,
'天眼查', '天眼查',
'2', '2',
time_format
] ]
cursor_.execute(insert_sql, tuple(list_info)) cursor_.execute(insert_sql, tuple(list_info))
cnx_.commit() cnx_.commit()
...@@ -214,10 +214,10 @@ def beinWork(tyc_code, social_code,start_time): ...@@ -214,10 +214,10 @@ def beinWork(tyc_code, social_code,start_time):
} }
except Exception as e: except Exception as e:
log.info(f'传输失败:{social_code}----{link}') log.info(f'传输失败:{social_code}----{link}')
e = '数据库传输失败' error = '数据库传输失败'
state = 0 state = 0
takeTime = baseCore.getTimeCost(start_time, time.time()) takeTime = baseCore.getTimeCost(start_time, time.time())
baseCore.recordLog(social_code, taskType, state, takeTime, link, e) baseCore.recordLog(social_code, taskType, state, takeTime, link, f'{error}----{e}')
continue continue
# print(dic_news) # print(dic_news)
# 将相应字段通过kafka传输保存 # 将相应字段通过kafka传输保存
......
Markdown 格式
0%
您添加了 0 到此讨论。请谨慎行事。
请先完成此评论的编辑!
注册 或者 后发表评论