求是网领导讲话

qiushi_leaderspeech.py

@@ -119,17 +119,17 @@ if __name__=='__main__':
                     #     or '中共' in author or '记者' in author or '新闻社' in author\
                     #     or '党委' in author or '调研组' in author or '研究中心' in author\
                     #     or '委员会' in author or '博物' in author or '大学' in author or '联合会' in author :
-                    # if '（' in author or '本刊' in author  \
-                    #         or '记者' in author or '新闻社' in author \
-                    #         or '”' in author\
-                    #         or '大学' in author or '洛桑江村' in author:
-                    #     continue
-                    if '国资委党委' in author:
-                        pass
-                    else:
+                    if '（' in author or '本刊' in author  \
+                            or '记者' in author or '新闻社' in author \
+                            or '”' in author\
+                            or '大学' in author or '洛桑江村' in author:
                         continue
+                    # if '国资委党委' in author:
+                    #     pass
+                    # else:
+                    #     continue
                     new_href = new.find('a')['href']
-                    is_member = r.sismember('qiushileaderspeech_two::' + period_title, new_href)
+                    is_member = r.sismember('qiushileaderspeech::' + period_title, new_href)
                     if is_member:
                         continue
                     new_title = new.find('a').text.replace('\u3000',' ').lstrip(' ').replace('——', '').replace('\xa0', '')
@@ -165,7 +165,7 @@ if __name__=='__main__':
                 }
                 log.info(dic_news)
                 if sendKafka(dic_news):
-                    r.sadd('qiushileaderspeech_two::' + period_title, new_href)
+                    r.sadd('qiushileaderspeech::' + period_title, new_href)
                     log.info(f'采集成功----{dic_news["sourceAddress"]}')