天眼查企业动态

comData/tcyQydt/tyc_qydt_add.py

@@ -14,9 +14,9 @@ from base.smart import smart_extractor
 import urllib3
 
 urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)
-jieba.cut("必须加载jieba")
 # 初始化，设置中文分词
-
+jieba.cut("必须加载jieba")
+smart =smart_extractor.SmartExtractor('cn')
 baseCore = BaseCore()
 log = baseCore.getLogger()
 cnx = pymysql.connect(host='114.116.44.11', user='root', password='f7s0&7qqtK', db='dbScore', charset='utf8mb4')
@@ -154,8 +154,8 @@ def beinWork(tyc_code, social_code,start_time):
                 time_format = baseCore.getNowTime(1)
             try:
                 # 开始进行智能解析
-                lang = baseCore.detect_language(title)
-                smart = smart_extractor.SmartExtractor(lang)
+                # lang = baseCore.detect_language(title)
+                # smart = smart_extractor.SmartExtractor(lang)
                 contentText = smart.extract_by_url(link).text
                 # time.sleep(3)
             except Exception as e: