Skip to content

P
physical
Commit 8959570e authored by 段英荣's avatar 段英荣
Browse files
Options

modify

parent 8fcf5a5c
Hide whitespace changes
Inline Side-by-side
Showing with 255 additions and 64 deletions
search/utils/pick.py
View file @ 8959570e
......@@ -4,6 +4,7 @@
import logging
import traceback
import datetime
import json
from libs.es import ESPerform
from libs.cache import redis_client
......@@ -13,7 +14,7 @@ class PickUtils(object):
def refresh_redis_data(cls,have_read_celebrity_list,redis_cli,redis_key,key_exist):
try:
if key_exist:
redis_cli.set(redis_key,have_read_celebrity_list)
redis_cli.set(redis_key,json.dumps(have_read_celebrity_list))
else:
now = datetime.datetime.now()
......@@ -22,7 +23,7 @@ class PickUtils(object):
valid_seconds = (end_time - now).seconds
redis_cli.setex(redis_key,have_read_celebrity_list,valid_seconds)
redis_cli.setex(redis_key,json.dumps(have_read_celebrity_list),valid_seconds)
return True
except:
......
search/utils/topic.py
View file @ 8959570e
......@@ -23,9 +23,9 @@ class TopicUtils(object):
"user_id": user_id
}
q["_souce"] = ["attention_user_id_list", "pick_user_id_list", "same_group_user_id_list"]
q["_souce"] = ["tag_list","attention_user_id_list", "pick_user_id_list", "same_group_user_id_list"]
result_dict = ESPerform.get_search_results(ESPerform.get_cli(), "topic", q, offset, size)
result_dict = ESPerform.get_search_results(ESPerform.get_cli(), "user", q, offset, size)
return result_dict
except:
......@@ -56,4 +56,122 @@ class TopicUtils(object):
return (chinese_user_id_list,japan_user_id_list,korea_user_id_list)
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return ([],[],[])
\ No newline at end of file
return ([],[],[])
@classmethod
def refresh_redis_hash_data(cls, redis_cli,redis_key,redis_data_dict):
try:
redis_cli.hmset(redis_key, redis_data_dict)
return True
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return False
@classmethod
def get_recommend_topic_ids(cls,user_id,offset,size,is_first_time,group_topic_ids_index,not_group_topic_ids_index):
"""
:需增加打散逻辑
:remark:获取首页推荐帖子列表
:param user_id:
:param offset:
:param size:
:param is_first_time:
:param group_topic_ids_index:
:param not_group_topic_ids_index:
:return:
"""
try:
#重写offset,size
size = 1000
if is_first_time:
offset=0
else:
if group_topic_ids_index>0:
offset = 1000 * group_topic_ids_index
else:
offset = 1000 * not_group_topic_ids_index
result_dict = TopicUtils.get_related_user_info(user_id, 0, 1)
if len(result_dict["hits"]) == 0:
logging.error("not find user_id:%d in es!" % int(user_id))
return {"recommend_topic_ids": []}
attention_user_info_list = result_dict["hits"][0]["attention_user_id_list"]
(attention_chinese_user_id_list, attention_japan_user_id_list,
attention_korea_user_id_list) = TopicUtils.analyze_related_user_id_list(
related_user_id_list=attention_user_info_list)
pick_user_info_list = result_dict["hits"][0]["pick_user_id_list"]
(pick_chinese_user_id_list, pick_japan_user_id_list,
pick_korea_user_id_list) = TopicUtils.analyze_related_user_id_list(pick_user_info_list)
same_group_user_info_list = result_dict["hits"][0]["same_group_user_id_list"]
(same_group_chinese_user_id_list, same_group_japan_user_id_list,
same_group_korea_user_id_list) = TopicUtils.analyze_related_user_id_list(same_group_user_info_list)
user_tag_list = result_dict["hits"][0]["tag_list"]
q = dict()
q["filter"] = {
"range": {"content_level": {"gte": 3, "lte": 5}}
}
if not is_first_time:
if group_topic_ids_index:
q["filter"]["range"]["group_id"] = {
"gt": 0
}
else:
q["filter"]["range"]["group_id"] = {
"lt": 0
}
functions_list = [
{
"filter": {"match": {
"user_id": attention_chinese_user_id_list + attention_japan_user_id_list + attention_korea_user_id_list}},
"weight": 5,
},
{
"filter": {"match": {
"user_id": pick_chinese_user_id_list + pick_japan_user_id_list + pick_korea_user_id_list}},
"weight": 3
},
{
"filter": {"match": {
"user_id": same_group_chinese_user_id_list + same_group_japan_user_id_list + same_group_korea_user_id_list}},
"weight": 2
}
]
query_function_score = {
"query": {
"match": {
"tag_list": user_tag_list
}
},
"score_mode": "sum",
"boost_mode": "sum",
"functions": functions_list
}
q["query"] = {
"function_score": query_function_score
}
q["_source"] = {
"include":["id","group_id"]
}
logging.info("duan add,home_recommend es query:%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)
group_topic_ids = list()
not_group_topic_ids = list()
for item in result_dict["hits"]:
if item["group_id"] > 0:
group_topic_ids.append(item["id"])
else:
not_group_topic_ids.append(item["id"])
return (group_topic_ids,not_group_topic_ids)
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return ([],[])
\ No newline at end of file
search/views/pick.py
View file @ 8959570e
......@@ -18,9 +18,10 @@ def pick_celebrity(user_id,pick_id,offset=0,size=10):
try:
redis_celebrity_info = redis_client.get(redis_key)
have_read_pick_celebrity_ids = json.loads(redis_celebrity_info)
have_read_pick_celebrity_ids = json.loads(redis_celebrity_info) if redis_celebrity_info
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
pass
q = dict()
q["query"] = {
......
search/views/topic.py
View file @ 8959570e
......@@ -4,7 +4,10 @@
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
@bind("physical/search/home_recommend")
......@@ -14,61 +17,48 @@ def home_recommend(user_id=-1,offset=0,size=10):
:return:
"""
try:
# 马甲账号集合
vest_account_list = list()
redis_key = "physical:home_recommend:" + "user_id:" + str(user_id)
redis_val_dict = redis_client.hgetall(redis_key)
if len(redis_val_dict) > 0:
group_topic_ids = json.loads(redis_val_dict["group_topic_ids"])
not_group_topic_ids = json.loads(redis_val_dict["not_group_topic_ids"])
group_topic_ids_index = redis_val_dict["group_topic_ids_index"]
not_group_topic_ids_index = redis_val_dict["not_group_topic_ids_index"]
result_dict = TopicUtils.get_related_user_info(user_id,offset,size)
if len(result_dict["hits"]) == 0:
logging.error("Not find user_id:%d in es!" % int(user_id))
return {"recommend_topic_ids":[]}
if len(group_topic_ids) < (size-1) or len(not_group_topic_ids) < 1:
if len(group_topic_ids) < (size-1):
group_topic_ids_index += 1
(new_group_topic_ids, new_not_group_topic_ids) = TopicUtils.get_recommend_topic_ids(user_id, offset, size,
False, group_topic_ids_index, not_group_topic_ids_index)
group_topic_ids += new_group_topic_ids
else:
not_group_topic_ids_index += 1
(new_group_topic_ids, new_not_group_topic_ids) = TopicUtils.get_recommend_topic_ids(user_id, offset, size,
False, group_topic_ids_index, not_group_topic_ids_index)
not_group_topic_ids += new_not_group_topic_ids
attention_user_info_list = result_dict["hits"][0]["attention_user_id_list"]
(attention_chinese_user_id_list, attention_japan_user_id_list, attention_korea_user_id_list) = TopicUtils.analyze_related_user_id_list(related_user_id_list=attention_user_info_list)
pick_user_info_list = result_dict["hits"][0]["pick_user_id_list"]
(pick_chinese_user_id_list, pick_japan_user_id_list, pick_korea_user_id_list) = TopicUtils.analyze_related_user_id_list(pick_user_info_list)
same_group_user_info_list = result_dict["hits"][0]["same_group_user_id_list"]
(same_group_chinese_user_id_list, same_group_japan_user_id_list, same_group_korea_user_id_list) = TopicUtils.analyze_related_user_id_list(same_group_user_info_list)
q = dict()
q["filter"] = {
"range":{"content_level":{"gte":3,"lte":5}}
}
query_function_score = dict()
functions = [
{
"filter":{"match":{"user_id":attention_chinese_user_id_list}},
"weight": 5,
},
{
"filter":{"match":{"user_id":pick_chinese_user_id_list}},
"weight":3
},
{
"filter": {"match": {"user_id": same_group_chinese_user_id_list}},
"weight": 2
},
{
"filter": {"match": {"user_id": vest_account_list}},
"weight": 0.5
},
{
"filter":{"match":{"content_level":5}},
"weight":5
},
{
"filter":{"match":{"content_level":4}},
"weight":3
},
{
"filter":{"match":{"content_level":3}},
"weight":2
recommend_topic_ids = group_topic_ids[:(size-1)] + not_group_topic_ids[:1]
redis_hash_dict = {
"group_topic_ids": group_topic_ids[(size-1):],
"not_group_topic_ids": not_group_topic_ids[1:],
"group_topic_ids_index":group_topic_ids_index,
"not_group_topic_ids_index":not_group_topic_ids_index
}
]
TopicUtils.refresh_redis_hash_data(redis_client,redis_key,redis_hash_dict)
return {"recommend_topic_ids":recommend_topic_ids}
else:
(group_topic_ids,not_group_topic_ids) = TopicUtils.get_recommend_topic_ids(user_id,offset,size,True,0,0)
recommend_topic_ids = group_topic_ids[:(size - 1)] + not_group_topic_ids[:1]
redis_hash_dict = {
"group_topic_ids": group_topic_ids[(size - 1):],
"not_group_topic_ids": not_group_topic_ids[1:],
"group_topic_ids_index": 0,
"not_group_topic_ids_index": 0
}
TopicUtils.refresh_redis_hash_data(redis_client, redis_key, redis_hash_dict)
return {"recommend_topic_ids": recommend_topic_ids}
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
......
trans2es/mapping/topic.json
View file @ 8959570e
......@@ -10,9 +10,11 @@
"content":{"type":"text"},
"content_level":{"type":"text"},
"user_id":{"type":"long"},
"group_id":{"type":"long"}, //所在组ID
"tag_list":{"type":"long"},//标签属性
"share_num":{"type":"long"},
"pick_id_list":{"type":"long"},
"offline_score":{"type":"double"},//离线算分
"create_time":{"type":"date", "format":"date_time_no_millis"},
"update_time":{"type":"date", "format":"date_time_no_millis"}
}
......
trans2es/models/group.py
View file @ 8959570e
......@@ -14,14 +14,15 @@ class Group(models.Model):
verbose_name = u'小组'
db_table = 'group'
id = models.IntegerField(verbose_name=u"小组ID",primary_key=True)
name = models.CharField(verbose_name=u'名称', max_length=100, default='')
description = models.CharField(verbose_name=u'描述', max_length=200, default='')
creator_id = models.IntegerField(verbose_name=u'创建者ID')
#icon = ImgUrlField('图标', max_length=255, img_type=IMG_TYPE.WATERMARK, default='')
icon = models.CharField(verbose_name=u'图标', max_length=128, default='')
topic_num = models.IntegerField(verbose_name=u'帖子数', default=0)
is_online = models.BooleanField(verbose_name=u"是否有效", default=True, db_index=True)
is_recommend = models.BooleanField(verbose_name=u"是否推荐", default=False, db_index=True)
is_online = models.BooleanField(verbose_name=u"是否有效", default=True)
is_recommend = models.BooleanField(verbose_name=u"是否推荐", default=False)
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))
......
trans2es/models/topic.py
View file @ 8959570e
......@@ -10,6 +10,21 @@ from django.db import models
import datetime
from .pick_topic import PickTopic
from .tag import TopicTag
from .user import UserExtra
from .group import Group
class ActionSumAboutTopic(models.Model):
class Meta:
verbose_name=u"帖子埋点数据汇总"
db_table="action_sum_about_topic"
partiton_date = models.CharField(verbose_name=u"日期",max_length=20)
device_id = models.CharField(verbose_name=u"用户设备号",max_length=50)
topic_id = models.CharField(verbose_name=u"帖子ID",max_length=50)
user_id = models.CharField(verbose_name=u"用户ID",max_length=50)
data_type = models.IntegerField(verbose_name=u"动作类型")
data_value = models.BigIntegerField(verbose_name=u"值")
class Topic(models.Model):
class Meta:
......@@ -18,7 +33,7 @@ class Topic(models.Model):
id = models.IntegerField(verbose_name='日记ID',primary_key=True)
name = models.CharField(verbose_name='日记名称',max_length=100)
group_id = models.IntegerField(verbose_name='用户所在组ID')
group_id = models.IntegerField(verbose_name='用户所在组ID',default=-1)
user_id = models.IntegerField(verbose_name='用户ID')
description = models.CharField(verbose_name='日记本描述',max_length=200)
content = models.CharField(verbose_name='日记本内容',max_length=1000)
......@@ -56,4 +71,50 @@ class Topic(models.Model):
return topic_tag_id_list
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return None
\ No newline at end of file
return None
def get_topic_offline_score(self):
try:
offline_score = 0.0
user_is_shadow = False
#是否官方推荐用户
user_query_results = UserExtra.objects.filter(user_id=self.user_id)
if user_query_results.count() > 0:
if user_query_results[0].is_recommend:
offline_score += 2.0
elif user_query_results[0].is_shadow:
user_is_shadow = True
#是否官方推荐小组
group_query_results = Group.objects.filter(id=self.group_id)
if group_query_results.count() > 0:
if group_query_results[0].is_recommend:
offline_score += 4.0
#帖子等级
if self.content_level == '5':
offline_score += 5.0
elif self.content_level == '4':
offline_score += 3.0
elif self.content_level == '3':
offline_score += 2.0
exposure_count = ActionSumAboutTopic.objects.filter(topic_id=self.id,data_type=1).count()
click_count = ActionSumAboutTopic.objects.filter(topic_id=self.id, data_type=2).count()
uv_num = ActionSumAboutTopic.objects.filter(topic_id=self.id,data_type=3).count()
if exposure_count>0:
offline_score += click_count/exposure_count
if uv_num>0:
offline_score += (self.vote_num/uv_num + self.reply_num/uv_num)
"""
1:马甲账号是否对总分降权？
"""
if user_is_shadow:
offline_score = offline_score*0.5
return offline_score
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return 0.0
\ No newline at end of file
trans2es/models/user.py
View file @ 8959570e
......@@ -94,4 +94,20 @@ class User(models.Model):
return user_tag_id_list
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return None
\ No newline at end of file
return None
class UserExtra(models.Model):
class Meta:
verbose_name=u"推荐用户表"
db_table="user_extra"
id = models.IntegerField(verbose_name="主键ID",primary_key=True)
user_id = models.IntegerField(verbose_name=u"用户ID")
is_shadow = models.BooleanField(verbose_name=u"是否是马甲账户")
is_online = models.BooleanField(verbose_name=u"是否上线")
is_recommend = models.BooleanField(verbose_name=u"是否推荐")
has_answered = models.BooleanField(verbose_name=u"")
is_deleted = models.BooleanField(verbose_name=u"")
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))
trans2es/utils/topic_transfer.py
View file @ 8959570e
......@@ -22,10 +22,12 @@ class TopicTransfer(object):
res["content"] = instance.content
res["content_level"] = instance.content_level
res["user_id"] = instance.user_id
res["group_id"] = instance.group_id
res["share_num"] = instance.share_num
res["pick_id_list"] = instance.get_pick_id_info()
res["tag_list"] = instance.get_topic_tag_id_list()
res["offline_score"] = instance.get_topic_offline_score()
create_time = instance.create_time
res["create_time"] = tzlc(create_time)
......
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