v0.4提交

tasks/update_rootSourceId.py

@@ -3,11 +3,9 @@
 迁移rootSourceId
 """
 import json
-import time
 
 import pymysql
 import datetime
-import schedule
 from concurrent.futures.thread import ThreadPoolExecutor
 
 source_list = [
@@ -29,7 +27,13 @@ source_id_list = {
     "touliu_tencentGzhArticle_a18c11dd294df014334f7db72830221a": 1,
     "touliu_tencentGzhArticle_c2debdc233827497e24b633dea36c57c": 1,
     "touliu_tencentGzhArticle_d66796826916665a23c667472ef4dd56": 1,
-    "touliu_tencentGzhArticle_f8e97355f3687f57fd4efeb635a7a3a2": 1
+    "touliu_tencentGzhArticle_f8e97355f3687f57fd4efeb635a7a3a2": 1,
+    "touliu_tencentGzhArticle_gh_68e7fdc09fe4_90bb12e53f6628fd5330310c7c3cc344": 1,
+    "touliu_tencentGzhArticle_gh_68e7fdc09fe4_cd602a61ea073e41404572fce51eb297": 1,
+    "touliu_tencentGzhArticle_gh_68e7fdc09fe4_d8fca9b2712f829d625d98bec37db228": 1,
+    "touliu_tencentGzhArticle_gh_77f36c109fb1_1401a97f6537f32b14496cd5fe6caa70": 1,
+    "touliu_tencentGzhArticle_gh_77f36c109fb1_926713998cd1513370b910ba20adda44": 1,
+    "touliu_tencentGzhArticle_gh_77f36c109fb1_4ca7c1c6223501ff4f80913f8363309f": 1
 }
 
 
@@ -87,11 +91,12 @@ def process_each_data(data_tuple):
                 charset="utf8mb4"
             )
             source_id = json.loads(result)['productionPath'].split("rootSourceId%3D")[1]
+            video_id = json.loads(result)['productionPath'].split("videos%3Fid%3D")[1].split("%26su%")[0]
             sql = f"""
             INSERT INTO long_articles_root_source_id
-            (rootSourceId, accountName, ghId, articleTitle, requestTime, trace_id, push_type)
+            (rootSourceId, accountName, ghId, articleTitle, requestTime, trace_id, push_type, video_id)
             values 
-            (%s, %s, %s, %s, %s, %s, %s);
+            (%s, %s, %s, %s, %s, %s, %s, %s);
             """
             cursor = connection.cursor()
             cursor.execute(
@@ -103,7 +108,8 @@ def process_each_data(data_tuple):
                     title,
                     request_time_stamp,
                     trace_id,
-                    source_id_list.get(source_id, 2)
+                    source_id_list.get(source_id, 2),
+                    video_id
                 )
             )
             connection.commit()