Merge branch 'dlis' into 'master'

add

See merge request alpha/physical!117

app_conf.xml

@@ -14,5 +14,6 @@
         <element value="search.views.user"/>
         <element value="search.views.tag"/>
         <element value="injection.data_sync.tasks"/>
+        <element value="search.views.contrast_similar"/>
     </config>
 </gm_rpcd_config>

search/views/contrast_similar.py

+# -*- coding: UTF-8 -*-
+# !/usr/bin/env python
+
+from __future__ import unicode_literals, absolute_import, print_function
+from gm_rpcd.all import bind
+import logging
+import traceback
+import json
+from search.utils.topic import TopicUtils
+from libs.es import ESPerform
+from libs.cache import redis_client
+from search.utils.common import *
+
+from gm_rpcd.all import bind
+import functools
+import logging
+import traceback
+from libs.es import ESPerform
+
+logger = logging.getLogger(__name__)
+
+MAX_LOAD = 100
+GROUP_SIZE = 10
+
+
+@bind('physical/search/query_contrast_similar')
+def contrast_similar(user_id, offset=0, size=10):
+    try:
+        if isinstance(user_id, int):
+            contrast_topic_id_list = get_all_topic_id_list(user_id, offset, size)
+            return {"contrast_similar_topic_id": contrast_topic_id_list}
+
+    except:
+
+        logging.error("catch exception,err_msg:%s" % traceback.format_exc())
+        return {"contrast_similar_topic_id": []}
+
+
+def get_all_topic_id_list(user_id, offset, size):
+    try:
+        if user_id:
+            redis_key = "physical:user_similar:participant_user_id:" + str(user_id)
+            redis_field_val_list = redis_client.get(redis_key)
+            have_read_topic_id_list = json.loads(redis_field_val_list) if redis_field_val_list else []
+            read_topic_id_list = list()
+            if len(have_read_topic_id_list) > 0:
+                for i in have_read_topic_id_list:
+                    read_topic_id_list.append(i.get("contrast_user_id"))
+                q = {
+                    "query": {
+                        "function_score": {
+                            "query": {
+                                "bool": {
+                                    "must": [{
+                                        "term": {
+                                            "is_online": True
+                                        }
+                                    }, {
+                                        "term": {
+                                            "is_deleted": False
+                                        }
+                                    },
+                                        {
+                                            "terms": {
+                                                "content_level": [3, 4, 5]
+                                            }
+                                        }],
+                                    "minimum_should_match": 1,
+                                    "should": {
+                                        "terms": {
+                                            "user_id": read_topic_id_list
+                                        }
+                                    }
+                                }
+                            }
+
+                        }
+                    }
+
+                }
+
+                logging.info('contrast_similar,as q:%s' % str(q).encode('utf-8'))
+
+                result_dict = ESPerform.get_search_results(ESPerform.get_cli(), sub_index_name="topic",
+                                                           query_body=q,
+                                                           offset=offset, size=size)
+                contrast_topic_ids = []
+
+                if len(result_dict['hits']) > 0:
+
+                    res_hit = result_dict['hits']
+
+                    for item in res_hit:
+                        topic_id = item['_source']['id']
+                        contrast_topic_ids.append(topic_id)
+
+                    logging.info('contrast_similar_topic_id:%s' % contrast_topic_ids)
+
+                return contrast_topic_ids
+
+
+    except:
+
+        logging.error("catch exception,err_msg:%s" % traceback.format_exc())
+        return {"contrast_similar_topic_id": []}

trans2es/management/commands/trans2es_data2es_parallel.py

@@ -73,8 +73,8 @@ class SyncDataToRedis(object):
                 redis_key = redis_key_prefix + str(participant_user_id)
                 similar_result_items = FaceUserContrastSimilar.objects.filter(is_online=True, is_deleted=False,
                                                                               participant_user_id=participant_user_id,
-                                                                              similarity__gt=0.4).order_by(
-                    "-similarity").limit(100)
+                                                                              similarity__gt=0.3).order_by(
+                    "-similarity")
 
                 item_list = list()
                 for item in similar_result_items:

trans2es/models/topic.py

@@ -105,18 +105,36 @@ class Topic(models.Model):
             logging.error("catch exception,err_msg:%s" % traceback.format_exc())
             return []
 
-    def get_tag_name_list(self,tag_id_list):
+    def get_tag_name_list(self, tag_id_list):
         try:
             tag_name_list = list()
-            query_results = Tag.objects.filter(id__in=tag_id_list).values_list("name",flat=True)
-            for item in query_results:
-                tag_name_list.append(item)
+            logging.info("get tag_id_list :%s"%tag_id_list)
+            if len(tag_id_list) > 100:
+                lens = int(len(tag_id_list) / 100)
+                for i in range(lens+1):
+                    if i != 0:
+                        tag_id_100 = tag_id_list[i * 100:i * 100 + 100]
+                    if i == 0:
+                        tag_id_100 = tag_id_list[i:i + 100]
+                    if i * 100 == lens * 100 and len(tag_id_list[lens * 100:]) > 0:
+                        tag_id_100 = tag_id_list[lens * 100:]
+                    if tag_id_100:
+                        logging.info("get tag_id_100 :%s" % tag_id_100)
+
+                        query_results = Tag.objects.filter(id__in=tag_id_100).values_list("name", flat=True)
+                        for item in query_results:
+                            tag_name_list.append(item)
+            else:
+                query_results = Tag.objects.filter(id__in=tag_id_list).values_list("name", flat=True)
+                for item in query_results:
+                    tag_name_list.append(item)
 
             return tag_name_list
         except:
             logging.error("catch exception,err_msg:%s" % traceback.format_exc())
             return []
 
+
     def get_topic_offline_score(self):
         try:
             offline_score = 0.0