merge

trans2es/mapping/topic-v1.json

+{
+  "dynamic":"strict",
+  "properties": {
+    "id":{"type":"long"},
+    "is_online":{"type":"boolean"},//上线
+    "is_deleted":{"type":"boolean"},
+    "vote_num":{"type":"long"},
+    "reply_num":{"type":"long"},
+    "name":{"type":"text","analyzer":"gm_default_index","search_analyzer":"gm_default_index"},
+    "description":{"type":"text","analyzer":"gm_default_index","search_analyzer":"gm_default_index"},
+    "content":{"type":"text","analyzer":"gm_default_index","search_analyzer":"gm_default_index"},
+    "content_level":{"type":"text"},
+    "user_id":{"type":"long"},
+    "group_id":{"type":"long"}, //所在组ID
+    "tag_list":{"type":"long"},//标签属性
+    "edit_tag_list":{"type":"long"},//编辑标签
+    "tag_name_list":{"type":"text","analyzer":"gm_default_index","search_analyzer":"gm_default_index"},
+    "share_num":{"type":"long"},
+    "pick_id_list":{"type":"long"},
+    "offline_score":{"type":"double"},//离线算分
+    "manual_score":{"type":"double"},//人工赋分
+    "has_image":{"type":"boolean"},//是否有图
+    "has_video":{"type":"boolean"},//是否是视频
+    "create_time":{"type":"date", "format":"date_time_no_millis"},
+    "update_time":{"type":"date", "format":"date_time_no_millis"},
+    "create_time_val":{"type":"long"},
+    "update_time_val":{"type":"long"},
+    "language_type":{"type":"long"},
+
+    "is_shadow": {"type": "boolean"},
+    "is_recommend": {"type": "boolean"},
+    "is_complaint": {"type": "boolean"},  // 是否被举报
+
+    "virtual_content_level":{"type": "text"},
+    "like_num_crawl": {"type": "long"}, // 爬取点赞数
+    "comment_num_crawl": {"type": "long"}, // 爬取评论数
+
+    "is_crawl": {"type": "boolean"},
+    "platform": {"type": "long"},
+    "platform_id": {"type": "long"},
+
+    "drop_score":{"type": "double"},  // 人工降分
+    "sort_score":{"type": "double"},  // 排序分
+
+    "pictorial_id":{"type": "long"}, //所在组ID
+    "pictorial_name":{  // 所在组名称
+        "type": "text",
+        "analyzer": "gm_default_index",
+        "search_analyzer": "gm_default_index"
+    }
+
+  }
+}

trans2es/mapping/topic.json

@@ -50,4 +50,4 @@
     }
 
   }
-}
\ No newline at end of file
+}

trans2es/mapping/user.json

@@ -28,17 +28,31 @@
         "country_id":{"type":"text"}
       }
     },
-    "same_group_user_id_list":{//同组用户列表
+//    "same_group_user_id_list":{//同组用户列表
+//      "type":"nested",
+//      "properties":{
+//        "user_id":{"type":"long"},
+//        "country_id":{"type":"text"}
+//      }
+//    },
+//    "attention_group_id_list":{//关注小组列表
+//      "type":"nested",
+//      "properties":{
+//        "group_id":{"type":"long"},
+//        "update_time_val":{"type":"long"}
+//      }
+//    },
+    "same_pictorial_user_id_list":{//同画报用户列表
       "type":"nested",
       "properties":{
         "user_id":{"type":"long"},
         "country_id":{"type":"text"}
       }
     },
-    "attention_group_id_list":{//关注小组列表
+    "attention_pictorial_id_list":{//关注画报列表
       "type":"nested",
       "properties":{
-        "group_id":{"type":"long"},
+        "pictorial_id":{"type":"long"},
         "update_time_val":{"type":"long"}
       }
     },

trans2es/type_info.py

@@ -12,11 +12,12 @@ import elasticsearch
 import elasticsearch.helpers
 import sys
 
-from trans2es.models import topic, user, pick_celebrity, group, celebrity, tag, contrast_similar
+from trans2es.models import topic, user, pick_celebrity, group, celebrity, tag, contrast_similar,pictorial
 from trans2es.utils.user_transfer import UserTransfer
 from trans2es.utils.pick_celebrity_transfer import PickCelebrityTransfer
 from trans2es.utils.group_transfer import GroupTransfer
 from trans2es.utils.topic_transfer import TopicTransfer
+from trans2es.utils.pictorial_transfer import PictorialTransfer
 from trans2es.utils.celebrity_transfer import CelebrityTransfer
 from trans2es.utils.tag_transfer import TagTransfer
 from trans2es.utils.contrast_similar_transfer import Contrast_Similar_Transfer
@@ -80,8 +81,6 @@ class TypeInfo(object):
 
     def bulk_get_data(self, instance_iterable):
         data_list = []
-        # 4，5星帖子单独索引
-        topic_data_high_star_list = list()
         if self.batch_get_data_func:
             _pk_list = [getattr(instance, 'pk', None) for instance in instance_iterable]
             not_found_pk_list = []
@@ -138,9 +137,7 @@ class TypeInfo(object):
                     ))
                 else:
                     data_list.append(data)
-                    if self.type=="topic" and instance.content_level and int(instance.content_level)>=4:
-                        topic_data_high_star_list.append(data)
-        return (data_list,topic_data_high_star_list)
+        return data_list
 
     def elasticsearch_bulk_insert_data(self, sub_index_name, data_list, es=None):
 
@@ -165,10 +162,10 @@ class TypeInfo(object):
         #             traceback.print_exc()
         #             es_result = 'error'
 
-        return ESPerform.es_helpers_bulk(es, data_list, sub_index_name)
+        return ESPerform.es_helpers_bulk(es, data_list, sub_index_name, True)
 
     def elasticsearch_bulk_insert(self, sub_index_name, instance_iterable, es=None):
-        data_list, topic_data_high_star_list = self.bulk_get_data(instance_iterable)
+        data_list = self.bulk_get_data(instance_iterable)
         return self.elasticsearch_bulk_insert_data(
             sub_index_name=sub_index_name,
             data_list=data_list,
@@ -191,24 +188,19 @@ class TypeInfo(object):
         time1=end-begin
 
         begin = time.time()
-        data_list, topic_data_high_star_list = self.bulk_get_data(instance_list)
+        data_list = self.bulk_get_data(instance_list)
         end = time.time()
         time2=end-begin
 
         begin = time.time()
+        logging.info("get sub_index_name:%s"%sub_index_name)
+        logging.info("get data_list:%s"%data_list)
+
         self.elasticsearch_bulk_insert_data(
             sub_index_name=sub_index_name,
             data_list=data_list,
             es=es,
         )
-
-        # 同时写4星及以上的帖子
-        if len(topic_data_high_star_list)>0:
-            self.elasticsearch_bulk_insert_data(
-                sub_index_name="topic-high-star",
-                data_list=topic_data_high_star_list,
-                es=es,
-            )
         end = time.time()
         time3=end-begin
         logging.info("duan add,insert_table_by_pk_list time cost:%ds,%ds,%ds,%ds" % (time0,time1,time2,time3))
@@ -223,7 +215,7 @@ class TypeInfo(object):
 
             stage_1_time = time.time()
 
-            data_list, topic_data_high_star_list = self.bulk_get_data(instance_list)
+            data_list = self.bulk_get_data(instance_list)
 
             stage_2_time = time.time()
 
@@ -268,9 +260,10 @@ def get_type_info_map():
         return _get_type_info_map_result
 
     type_info_list = [
+
         TypeInfo(
-            name='topic-high-star',  # >=4星日记
-            type='topic-high-star',
+            name='topic',  # 日记
+            type='topic',
             model=topic.Topic,
             query_deferred=lambda: topic.Topic.objects.all().query,
             get_data_func=TopicTransfer.get_topic_data,
@@ -279,8 +272,8 @@ def get_type_info_map():
             round_insert_period=2,
         ),
         TypeInfo(
-            name='topic',  # 日记
-            type='topic',
+            name='topic-v1',  # 日记
+            type='topic-v1',
             model=topic.Topic,
             query_deferred=lambda: topic.Topic.objects.all().query,
             get_data_func=TopicTransfer.get_topic_data,
@@ -352,7 +345,18 @@ def get_type_info_map():
             bulk_insert_chunk_size=100,
             round_insert_chunk_size=5,
             round_insert_period=2
+        ),
+        TypeInfo(
+            name="pictorial",  # 画报
+            type="pictorial",
+            model=pictorial.Pictorial,
+            query_deferred=lambda: pictorial.Pictorial.objects.all().query,
+            get_data_func=PictorialTransfer.get_poctorial_data,
+            bulk_insert_chunk_size=100,
+            round_insert_chunk_size=5,
+            round_insert_period=2,
         )
+
     ]
 
     type_info_map = {
@@ -362,3 +366,4 @@ def get_type_info_map():
 
     _get_type_info_map_result = type_info_map
     return type_info_map
+