update

utils/func_from_es_get_article.py

@@ -186,7 +186,7 @@ def get_device_num_from_es(word):
     return tractate_content_num
 
 
-def get_tractate_tags_from_es(doc_id):
+def get_tractate_tags_from_es(doc_id,id_type='tractate'):
     # print(doc_id)
     body = {
         "query": {
@@ -197,14 +197,22 @@ def get_tractate_tags_from_es(doc_id):
             }
         }
     }
+    if id_type == 'tractate':
     # tractate
-    results = es.search(
-        index='gm-dbmw-tractate-read',
-        doc_type='tractate',
-        timeout='10s',
-        body=body
-    )
+        results = es.search(
+            index='gm-dbmw-tractate-read',
+            doc_type='tractate',
+            timeout='10s',
+            body=body
+        )
     # print(results)
+    elif id_type == 'diary':
+        results = es.search(
+            index='gm-dbmw-diary-read',
+            doc_type='diary',
+            timeout='10s',
+            body=body
+        )
     return results["hits"]['hits'][0]
 
 
@@ -365,3 +373,34 @@ def get_user_post_from_mysql(sql):
                 tags_v3_card_id_list.append(int(card_id[0]))
     # print(card_id_list)
     return second_demands_count_dict, tags_v3_count_dict, second_demands_card_id_list,tags_v3_card_id_list,second_demands_tractate_dict,tags_v3_tractate_dict
+
+
+
+def get_diary_from_mysql(sql,id_type='tractate'):
+    second_demands_card_id_list = []
+    tags_v3_card_id_list = []
+    second_demands_count_dict = {}
+    tags_v3_count_dict = {}
+    second_demands_tractate_dict = {}
+    tags_v3_tractate_dict = {}
+    sql_res = con_sql_doris_prod(sql)
+    for card_id in sql_res:
+        # print(type(card_id[0]),card_id)
+        es_res = get_tractate_tags_from_es(card_id[0],id_type='diary')
+        second_demands_tractate_dict[card_id[0]] = es_res["_source"]["second_demands"]
+        tags_v3_tractate_dict[card_id[0]] = es_res["_source"]["tags_v3"]
+
+        for position in es_res["_source"]["second_demands"]:
+            if position in second_demands_count_dict:
+                second_demands_count_dict[position] += 1
+            else:
+                second_demands_count_dict[position] = 1
+                second_demands_card_id_list.append(int(card_id[0]))
+        for tag in es_res["_source"]["tags_v3"]:
+            if tag in tags_v3_count_dict:
+                tags_v3_count_dict[tag] += 1
+            else:
+                tags_v3_count_dict[tag] = 1
+                tags_v3_card_id_list.append(int(card_id[0]))
+    # print(card_id_list)
+    return second_demands_count_dict, tags_v3_count_dict, second_demands_card_id_list,tags_v3_card_id_list,second_demands_tractate_dict,tags_v3_tractate_dict