语言判断

comData/noticeReport/东方财富网-公告.py

-import os
+import os
@@ -192,7 +192,9 @@ def GetContent(pdf_url,info_url, pdf_name, social_code, year, pub_time, start_ti
     else:
         return False
     content = retData['content']
-
+    lang = baseCore.detect_language(content)
+    if lang == 'cn':
+        lang = 'zh'
     time_now = time.strftime("%Y-%m-%d %H:%M:%S", time.localtime())
     dic_news = {
         'attachmentIds': att_id,
@@ -203,7 +205,7 @@ def GetContent(pdf_url,info_url, pdf_name, social_code, year, pub_time, start_ti
         'deleteFlag': '0',
         'id': '',
         'keyWords': '',
-        'lang': 'zh',
+        'lang': lang,
         'origin': '东方财富网',
         'publishDate': pub_time,
         'sid': '1684032033495392257',

comData/noticeReport/东方财富网-港股公告.py

-import os
+import os
@@ -278,7 +278,9 @@ def spider(browser, code, social_code, com_name):
             att_id = ''
 
         browser2.quit()
-
+        lang = baseCore.detect_language(content)
+        if lang == 'cn':
+            lang = 'zh'
         time_now = time.strftime("%Y-%m-%d %H:%M:%S", time.localtime())
         dic_news = {
             'attachmentIds': att_id,
@@ -289,7 +291,7 @@ def spider(browser, code, social_code, com_name):
             'deleteFlag': '0',
             'id': '',
             'keyWords': '',
-            'lang': 'zh',
+            'lang': lang,
             'origin': '东方财富网',
             'publishDate': publishDate,
             'sid': '1684032033495392257',

comData/noticeReport/证监会-公告.py

-
+
@@ -308,7 +308,9 @@ def GetContent(pdf_url, pdf_name, social_code, year, pub_time, start_time,com_na
     else:
         return False
     content = retData['content']
-
+    lang = baseCore.detect_language(content)
+    if lang == 'cn':
+        lang = 'zh'
     time_now = time.strftime("%Y-%m-%d %H:%M:%S", time.localtime())
     dic_news = {
         'attachmentIds': att_id,
@@ -319,7 +321,7 @@ def GetContent(pdf_url, pdf_name, social_code, year, pub_time, start_time,com_na
         'deleteFlag': '0',
         'id': '',
         'keyWords': '',
-        'lang': 'zh',
+        'lang': lang,
         'origin': origin,
         'publishDate': pub_time,
         'sid': '1684032033495392257',