Skip to content

P
physical
Commit 3fec999d authored by lixiaofang's avatar lixiaofang
Browse files
Options

add

parents 595cecb8 a7cee0ca
Hide whitespace changes
Inline Side-by-side
Showing with 135 additions and 61 deletions
linucb/views/collect_data.py
View file @ 3fec999d
...@@ -189,11 +189,11 @@ class CollectData(object): ...@@ -189,11 +189,11 @@ class CollectData(object):
click_results = TopicTag.objects.using(settings.SLAVE_DB_NAME).filter( click_results = TopicTag.objects.using(settings.SLAVE_DB_NAME).filter(
topic_id=topic_id, is_online=True).values_list("tag_id", "is_collection") topic_id=topic_id, is_online=True).values_list("tag_id", "is_collection")
for tag_id, is_collection in click_results: for tag_id, is_collection in click_results:
topic_tag_list.append(tag_id) # topic_tag_list.append(tag_id)
if is_collection: if is_collection:
click_topic_tag_list.append(tag_id) topic_tag_list.append(tag_id)
tag_query_results = Tag.objects.using(settings.SLAVE_DB_NAME).filter( tag_query_results = Tag.objects.using(settings.SLAVE_DB_NAME).filter(
id__in=topic_tag_list, is_online=True, is_deleted=False).values_list("id", id__in=topic_tag_list, is_online=True, is_deleted=False, is_category=False).values_list("id",
"is_ai") "is_ai")
for id, is_ai in tag_query_results: for id, is_ai in tag_query_results:
if is_ai: if is_ai:
...@@ -207,7 +207,7 @@ class CollectData(object): ...@@ -207,7 +207,7 @@ class CollectData(object):
device_id = raw_val_dict["device"]["device_id"] device_id = raw_val_dict["device"]["device_id"]
user_id = raw_val_dict["user_id"] if "user_id" in raw_val_dict else None user_id = raw_val_dict["user_id"] if "user_id" in raw_val_dict else None
if query_type=="do": if query_type=="do":
tag_list = list(Tag.objects.using(settings.SLAVE_DB_NAME).filter(name=tag_name,is_online=True,is_deleted=False).values_list("id",flat=True)) tag_list = list(Tag.objects.using(settings.SLAVE_DB_NAME).filter(name=tag_name,is_online=True,is_deleted=False, is_category=False).values_list("id",flat=True))
click_topic_tag_list.extend(tag_list) click_topic_tag_list.extend(tag_list)
logging.info("query tag attention,positive tag_list,device_id:%s,query_name:%s,tag_list:%s" % ( logging.info("query tag attention,positive tag_list,device_id:%s,query_name:%s,tag_list:%s" % (
str(device_id), tag_name, str(click_topic_tag_list))) str(device_id), tag_name, str(click_topic_tag_list)))
...@@ -292,17 +292,21 @@ class CollectData(object): ...@@ -292,17 +292,21 @@ class CollectData(object):
# if len(exposure_sql_query_results)>0: # if len(exposure_sql_query_results)>0:
if len(tagid_list) > 0: if len(tagid_list) > 0:
tag_query_results = Tag.objects.using(settings.SLAVE_DB_NAME).filter(
id__in=tagid_list, is_online=True, is_deleted=False,
is_category=False).values_list("id",flat =True)
is_click = 1 is_click = 1
is_vote = 0 is_vote = 0
reward = 1 if is_click or is_vote else 0 reward = 1 if is_click or is_vote else 0
for tag_id in tagid_list: for tag_id in tag_query_results:
self.update_user_linucb_tag_info(reward, device_id, tag_id, user_feature) self.update_user_linucb_tag_info(reward, device_id, tag_id, user_feature)
# 更新该用户的推荐tag数据，放在 更新完成user tag行为信息之后 # 更新该用户的推荐tag数据，放在 更新完成user tag行为信息之后
self.update_recommend_tag_list(device_id, user_feature, user_id,new_user_click_tag_list=tagid_list) self.update_recommend_tag_list(device_id, user_feature, user_id,
else: new_user_click_tag_list=tag_query_results)
logging.warning("unknown type msg:%s" % raw_val_dict.get("type", "missing type")) else:
logging.warning("unknown type msg:%s" % raw_val_dict.get("type", "missing type"))
except: except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc()) logging.error("catch exception,err_msg:%s" % traceback.format_exc())
......
search/utils/user.py
View file @ 3fec999d
...@@ -275,21 +275,21 @@ class UserUtils(object): ...@@ -275,21 +275,21 @@ class UserUtils(object):
for query_item in batch_query_list: for query_item in batch_query_list:
query_body += "{}\n{}\n".format(json.dumps(search_header_dict),json.dumps(query_item)) query_body += "{}\n{}\n".format(json.dumps(search_header_dict),json.dumps(query_item))
result_dict = ESPerform.get_search_results(es_cli_obj, sub_index_name="user", query_body=query_body,
batch_search=True)
ret_dict = dict() ret_dict = dict()
user_index = 0 if len(query_body)>0:
for res_item in result_dict["responses"]: result_dict = ESPerform.get_search_results(es_cli_obj, sub_index_name="user", query_body=query_body,
recommend_user_list = list() batch_search=True)
for item in res_item["hits"]["hits"]: user_index = 0
recommend_user_list.append(item["_source"]["user_id"]) for res_item in result_dict["responses"]:
recommend_user_list = list()
for item in res_item["hits"]["hits"]:
recommend_user_list.append(item["_source"]["user_id"])
logging.info("duan add,attention_user_id_list:%s,user_index:%d" % (str(attention_user_id_list), user_index)) logging.info("duan add,attention_user_id_list:%s,user_index:%d" % (str(attention_user_id_list), user_index))
ret_dict[str(attention_user_id_list[user_index])] = recommend_user_list ret_dict[str(attention_user_id_list[user_index])] = recommend_user_list
user_index += 1 user_index += 1
logging.info("duan add,ret_dict:%s" % str(ret_dict))
return ret_dict return ret_dict
except: except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc()) logging.error("catch exception,err_msg:%s" % traceback.format_exc())
......
search/views/group.py
View file @ 3fec999d
...@@ -13,6 +13,7 @@ from libs.es import ESPerform ...@@ -13,6 +13,7 @@ from libs.es import ESPerform
from search.utils.group import GroupUtils from search.utils.group import GroupUtils
from search.utils.common import GroupSortTypes from search.utils.common import GroupSortTypes
from libs.es import ESPerform from libs.es import ESPerform
from trans2es.models.topic import CommunityCategoryTagRelation
from trans2es.models.pictorial import PictorialTopics from trans2es.models.pictorial import PictorialTopics
from trans2es.models.pictorial import CommunityPictorialActivity from trans2es.models.pictorial import CommunityPictorialActivity
from alpha_types.venus import PICTORIAL_ACTIVITY_SORT from alpha_types.venus import PICTORIAL_ACTIVITY_SORT
...@@ -96,6 +97,10 @@ def pictorial_sort(user_id=-1, sort_type=GroupSortTypes.HOT_RECOMMEND, offset=0, ...@@ -96,6 +97,10 @@ def pictorial_sort(user_id=-1, sort_type=GroupSortTypes.HOT_RECOMMEND, offset=0,
@bind("physical/search/pictorial_topic") @bind("physical/search/pictorial_topic")
def pictorial_topic(topic_id=-1, offset=0, size=10): def pictorial_topic(topic_id=-1, offset=0, size=10):
""" """
帖子详情页的榜单推荐
先获取该帖子所属榜单 过滤下线的无帖子的被删除的榜单
再根据该帖子下的标签去获取有这些标签的榜单
最后根据该标签对应的分类标签去获取榜单
:remark 入选画报 :remark 入选画报
:param user_id: :param user_id:
:param sort_type: :param sort_type:
...@@ -108,7 +113,7 @@ def pictorial_topic(topic_id=-1, offset=0, size=10): ...@@ -108,7 +113,7 @@ def pictorial_topic(topic_id=-1, offset=0, size=10):
# 获取es链接对象 # 获取es链接对象
es_cli_obj = ESPerform.get_cli() es_cli_obj = ESPerform.get_cli()
q = {} q = {}
# 获取帖子从属的画报 # 获取帖子及该帖子的标签
q["query"] = { q["query"] = {
"bool": { "bool": {
"must": [ "must": [
...@@ -143,26 +148,30 @@ def pictorial_topic(topic_id=-1, offset=0, size=10): ...@@ -143,26 +148,30 @@ def pictorial_topic(topic_id=-1, offset=0, size=10):
for item in result_dict["hits"]: for item in result_dict["hits"]:
pict_pictorial_ids_list = item["_source"]["pictorial_id"] pict_pictorial_ids_list = item["_source"]["pictorial_id"]
topic_tag_list = item["_source"]["tag_list"] topic_tag_list = item["_source"]["tag_list"]
##先拿到当前帖子对应的标签的分类ID
tag_type_ids = CommunityCategoryTagRelation.objects.filter(tag_id__in=topic_tag_list,
is_online=True, is_deleted=False).values_list(
"category_tag_id", flat=True)
logging.info("get tag_type_ids:%s" % tag_type_ids)
# 拿到所属榜单
if len(pict_pictorial_ids_list) > 0:
q["query"] = { q["query"] = {
"bool": { "bool": {
"must": [{ "must": [{
"terms": { "terms": {
"id": pict_pictorial_ids_list "id": pict_pictorial_ids_list
} }
}, },
{ {
"term": { "term": {
"is_online": True "is_online": True
} }
}, },
{ {
"term": { "term": {
"is_deleted": False "is_deleted": False
} }
}, },
] ]
} }
...@@ -183,13 +192,18 @@ def pictorial_topic(topic_id=-1, offset=0, size=10): ...@@ -183,13 +192,18 @@ def pictorial_topic(topic_id=-1, offset=0, size=10):
pictorial_id = item["_source"]["id"] pictorial_id = item["_source"]["id"]
pictorial_id_list.append(pictorial_id) pictorial_id_list.append(pictorial_id)
if len(pictorial_id_list) < 10: functions_list = []
all_tag = []
all_tag.extend(topic_tag_list)
all_tag.extend(tag_type_ids)
# 如果不够10个去拿所属标签的榜单和所属分类的标签的榜单
if len(pictorial_id_list) < 10 and len(all_tag) > 0:
num = 10 - len(pictorial_id_list) num = 10 - len(pictorial_id_list)
q["query"] = { query = {
"bool": { "bool": {
"must": [{ "must": [{
"terms": { "terms": {
"tag_id": topic_tag_list "tag_id": all_tag
}}, { }}, {
"term": { "term": {
"is_online": True "is_online": True
...@@ -208,31 +222,56 @@ def pictorial_topic(topic_id=-1, offset=0, size=10): ...@@ -208,31 +222,56 @@ def pictorial_topic(topic_id=-1, offset=0, size=10):
}] }]
} }
} }
functions_list.append({
"filter": {
"constant_score": {
"filter": {
"terms": {
"tag_id": topic_tag_list
}
}
}
},
"weight": 200
})
q = {
"query": {
"function_score": {
"query": query,
"score_mode": "sum",
"boost_mode": "replace",
"functions": functions_list
}
}
}
q["_source"] = { q["_source"] = {
"includes": ["id", "tag_id"]} "includes": ["id", "tag_id"]}
q["sort"] = { q["sort"] = [
'update_time': { {
'order': 'desc' "_score": {
"order": "desc"
}} }
q["from"] = 0 },
q["size"] = 10 {
'update_time': {
'order': 'desc'
result_dict = ESPerform.get_search_results(es_cli_obj, "pictorial", q, offset, size) }
}
]
result_dict = ESPerform.get_search_results(es_cli_obj, "pictorial", q, 0, 10)
logging.info("get result_dict:%s" % result_dict)
if len(result_dict["hits"]) > 0: if len(result_dict["hits"]) > 0:
for item in result_dict["hits"]: for item in result_dict["hits"]:
id = item["_source"]["id"] id = item["_source"]["id"]
if id not in pictorial_id_list: if id not in pictorial_id_list:
pictorial_id_list.append(id) # pictorial_id_list.append(id) #
pictorial_list = pictorial_id_list if len(pictorial_id_list) < 10 else pictorial_id_list[:10] pictorial_list = pictorial_id_list if len(pictorial_id_list) < 10 else pictorial_id_list[:10]
return {"pictorial_ids_list": pictorial_list} return {"pictorial_ids_list": pictorial_list}
else: else:
return {"pictorial_ids_list": pictorial_id_list} return {"pictorial_ids_list": pictorial_id_list}
except: except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc()) logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return {"pictorial_ids_list": []} return {"pictorial_ids_list": []}
......
search/views/topic.py
View file @ 3fec999d
...@@ -15,6 +15,7 @@ from libs.es import ESPerform ...@@ -15,6 +15,7 @@ from libs.es import ESPerform
from django.conf import settings from django.conf import settings
from libs.tools import get_have_read_lin_pictorial_id_list from libs.tools import get_have_read_lin_pictorial_id_list
import datetime import datetime
from trans2es.models.tag import Tag
def get_discover_page_topic_ids(user_id, device_id, size, query_type=TopicPageType.FIND_PAGE): def get_discover_page_topic_ids(user_id, device_id, size, query_type=TopicPageType.FIND_PAGE):
...@@ -430,35 +431,64 @@ def topic_detail_page_recommend(device_id="", user_id=-1, topic_id=-1, topic_pic ...@@ -430,35 +431,64 @@ def topic_detail_page_recommend(device_id="", user_id=-1, topic_id=-1, topic_pic
have_read_topic_list.append(topic_id) have_read_topic_list.append(topic_id)
topic_user_result = list() topic_user_result = list()
topic_tag_result = list() topic_tag_result = list()
normal_tag_result = list()
category_tag_reuslt = list()
result = list() result = list()
if len(topic_tag_list) != 0: tag_query_results = Tag.objects.using(settings.SLAVE_DB_NAME).filter(
topic_tag_result = TopicUtils.top_get_topic_detail_recommend_list(user_id, topic_id, have_read_topic_list, id__in=topic_tag_list, is_online=True, is_deleted=False).values_list("id", "is_category")
size, es_cli_obj,
index_type="topic", routing="3,4,5,6", category_tag_list = list()
topic_tag_list=topic_tag_list) normal_tag_list = list()
topic_tag_size = len(topic_tag_result)
have_read_topic_list.extend(topic_tag_result) for id, is_category in tag_query_results:
if is_category:
category_tag_list.append(id)
else:
normal_tag_list.append(id)
if len(normal_tag_list) != 0:
normal_tag_result = TopicUtils.top_get_topic_detail_recommend_list(user_id, topic_id, have_read_topic_list,
size, es_cli_obj,
index_type="topic", routing="3,4,5,6",
topic_tag_list=normal_tag_list)
normal_tag_size = len(normal_tag_result)
have_read_topic_list.extend(normal_tag_result)
else: else:
topic_tag_size = 0 normal_tag_size = 0
if topic_tag_size < size:
size = size - topic_tag_size if normal_tag_size < size:
if topic_user_id != -1: size = size - normal_tag_size
topic_user_result = TopicUtils.top_get_topic_detail_recommend_list(user_id, topic_id, if len(category_tag_list) != 0:
category_tag_reuslt = TopicUtils.top_get_topic_detail_recommend_list(user_id, topic_id,
have_read_topic_list,
size, es_cli_obj,
index_type="topic",
routing="3,4,5,6",
topic_tag_list=category_tag_list)
category_tag_size = len(category_tag_reuslt)
have_read_topic_list.extend(category_tag_reuslt)
else:
category_tag_size = 0
if category_tag_size < size:
size = size - category_tag_size
if topic_user_id != -1:
topic_user_result = TopicUtils.top_get_topic_detail_recommend_list(user_id, topic_id,
have_read_topic_list, have_read_topic_list,
size, es_cli_obj, size, es_cli_obj,
index_type="topic", index_type="topic",
routing="3,4,5,6", routing="3,4,5,6",
topic_user_id=topic_user_id topic_user_id=topic_user_id
) )
topic_user_size = len(topic_user_result) topic_user_size = len(topic_user_result)
have_read_topic_list.extend(topic_user_result) have_read_topic_list.extend(topic_user_result)
if topic_user_size < size: if topic_user_size < size:
size = size - topic_user_size size = size - topic_user_size
result = TopicUtils.top_get_topic_detail_recommend_list(user_id, topic_id, result = TopicUtils.top_get_topic_detail_recommend_list(user_id, topic_id,
have_read_topic_list, have_read_topic_list,
size, es_cli_obj, size, es_cli_obj,
index_type="topic", routing="4,5,6") index_type="topic", routing="4,5,6")
have_read_topic_list.extend(result) have_read_topic_list.extend(result)
# have_read_topic_redis_data = redis_client.get(redis_key) # have_read_topic_redis_data = redis_client.get(redis_key)
# have_read_topic_list = json.loads(have_read_topic_redis_data) if have_read_topic_redis_data else [] # have_read_topic_list = json.loads(have_read_topic_redis_data) if have_read_topic_redis_data else []
...@@ -494,7 +524,8 @@ def topic_detail_page_recommend(device_id="", user_id=-1, topic_id=-1, topic_pic ...@@ -494,7 +524,8 @@ def topic_detail_page_recommend(device_id="", user_id=-1, topic_id=-1, topic_pic
logging.warning("topic_user_result:%s" % str(topic_user_result)) logging.warning("topic_user_result:%s" % str(topic_user_result))
logging.warning("result:%s" % str(result)) logging.warning("result:%s" % str(result))
recommend_topic_ids_list = list() recommend_topic_ids_list = list()
recommend_topic_ids_list.extend(topic_tag_result) recommend_topic_ids_list.extend(normal_tag_result)
recommend_topic_ids_list.extend(category_tag_reuslt)
recommend_topic_ids_list.extend(topic_user_result) recommend_topic_ids_list.extend(topic_user_result)
recommend_topic_ids_list.extend(result) recommend_topic_ids_list.extend(result)
return {"recommend_topic_ids": recommend_topic_ids_list} return {"recommend_topic_ids": recommend_topic_ids_list}
......
trans2es/models/tag.py
View file @ 3fec999d
...@@ -53,6 +53,7 @@ class Tag(models.Model): ...@@ -53,6 +53,7 @@ class Tag(models.Model):
is_own = models.IntegerField(verbose_name=u"是否ins上自带") is_own = models.IntegerField(verbose_name=u"是否ins上自带")
create_time = models.DateTimeField(verbose_name=u'创建时间', default=datetime.datetime.fromtimestamp(0)) create_time = models.DateTimeField(verbose_name=u'创建时间', default=datetime.datetime.fromtimestamp(0))
update_time = models.DateTimeField(verbose_name=u'更新时间', default=datetime.datetime.fromtimestamp(0)) update_time = models.DateTimeField(verbose_name=u'更新时间', default=datetime.datetime.fromtimestamp(0))
is_category = models.IntegerField(verbose_name=u"是否大分类")
class CommunityTagFollow(models.Model): class CommunityTagFollow(models.Model):
......
trans2es/models/topic.py
View file @ 3fec999d
...@@ -441,7 +441,6 @@ class CommunityTopicProduct(models.Model): ...@@ -441,7 +441,6 @@ class CommunityTopicProduct(models.Model):
create_time = models.DateTimeField(verbose_name=u'创建时间') create_time = models.DateTimeField(verbose_name=u'创建时间')
is_deleted = models.BooleanField(verbose_name=u'是否删除') is_deleted = models.BooleanField(verbose_name=u'是否删除')
class CommunityCategoryTagRelation(models.Model): class CommunityCategoryTagRelation(models.Model):
class Meta: class Meta:
verbose_name = u"标签分类对应关系" verbose_name = u"标签分类对应关系"
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment