Merge remote-tracking branch 'origin/master'

comData/sinafinance_news/nyse_news_gn.py

@@ -27,7 +27,7 @@ headers = {
     'Cache-Control': 'no-cache',
     'Pragma': 'no-cache'
 }
-taskType = '企业动态/新浪财经'
+taskType = '企业动态/新浪财经/国内'
 pattern = r"\d{4}-\d{2}-\d{2}\s+\d{2}:\d{2}"
 
 # 获取响应页面

comData/sinafinance_news/nyse_news_xg.py

@@ -28,7 +28,7 @@ headers = {
     'Cache-Control': 'no-cache',
     'Pragma': 'no-cache'
 }
-taskType = '企业动态/新浪财经'
+taskType = '企业动态/新浪财经/香港'
 
 
 # 判断时间是否是正确格式
@@ -51,7 +51,7 @@ def format_time(time_str):
 def getrequests(url):
     ip = baseCore.get_proxy()
     req = requests.get(url, headers=headers,proxies=ip)
-    req.encoding = req.apparent_encoding
+    req.encoding = 'gbk'
     soup = BeautifulSoup(req.text, 'html.parser')
     return soup
 
@@ -117,7 +117,7 @@ def getDic(social_code, title, href, pub_time):
     #     state = 0
     #     takeTime = baseCore.getTimeCost(start_time, time.time())
     #     baseCore.recordLog(social_code, taskType, state, takeTime, href, f'{href}===发送Kafka失败')
-    # return 1
+    return 1
 
 
 # 数据发送至Kafka
@@ -165,10 +165,10 @@ def selectUrl(url, social_code):
 
 
 def doJob():
-    # while True:
+    while True:
         start_time = time.time()
-    # social_code = baseCore.redicPullData('NewsEnterprise:xgqy_nyse_socialCode')
-    social_code = '91330000747735638J'
+        social_code = baseCore.redicPullData('NewsEnterprise:xgqy_nyse_socialCode')
+        # social_code = '91330000747735638J'
         if not social_code or social_code == 'None':
             time.sleep(20)
         data = baseCore.getInfomation(social_code)