Skip to content

P
physical
Commit 5995ef9a authored by lixiaofang's avatar lixiaofang
Browse files
Options

新增搜索接口

parent 53a3904c
Hide whitespace changes
Inline Side-by-side
Showing with 275 additions and 102 deletions
search/views/search_hotword.py 0 → 100644
View file @ 5995ef9a
#!/usr/bin/env python
# -*- coding: utf-8 -*-
from gm_rpcd.all import bind
import logging
import traceback
import json
from libs.cache import redis_client
from libs.es import ESPerform
from search.utils.group import GroupUtils
from search.utils.common import GroupSortTypes
from libs.es import ESPerform
from trans2es.models.pictorial import PictorialTopics
from trans2es.models.tag import RegisterShowTag, Tag
import random
@bind("physical/search/search_hotword")
def search_hotword(device_id=None, offset=0, size=10):
"""
:remark:搜索页的热门灵感
内容组成：搜索推荐热词register_show_tag
个性化标签physical:linucb:tag_recommend:device_id:
核心词，需要去重community_tag collection 1
①先从后台随机取最多6个搜索推荐热词，不够6个则有多少取多少
②然后随机取和推荐热词一样数量的linUCB标签，取不够数则取核心标签，linUCB+核心词去重后的数量要和搜索推荐热词的数量一样
③最后打乱顺序填入到热门灵感中，超过3行的标签弃掉，不够3行不用管
:param query:
:param offset:
:param size:
:return:
"""
try:
all_tag_name_list = set()
# 获取搜索推荐热词
results_registr_tag = RegisterShowTag.objects.filter(is_deleted=False, is_online=1).count()
results_tag = Tag.objects.filter(is_online=True, collection=1).count()
for i in range(0, 6):
tag_id = random.randint(0, results_registr_tag)
results_tag_chose = Tag.objects.filter(id=tag_id).values_list("name", flat=True)
all_tag_name_list.add(results_tag_chose)
logging.info("get all_tag_name_list01:%s" % all_tag_name_list)
# 获取个性化标签
linucb_recommend_redis_prefix = "physical:linucb:tag_recommend:device_id:"
tag_recommend_redis_key = linucb_recommend_redis_prefix + str(device_id)
linucb_recommend_tag_data = redis_client.get(tag_recommend_redis_key)
linucb_recommend_tag_list = json.loads(linucb_recommend_tag_data) if linucb_recommend_tag_data else []
for item in linucb_recommend_tag_list:
all_tag_name_list.add(item)
logging.info("get all_tag_name_list02:%s" % all_tag_name_list)
# 取不够数则取核心标签
if len(all_tag_name_list) < 12:
for i in range(0, 12):
tag_id = random.randint(0, results_tag)
results_tag_hexin = Tag.objects.filter(id=tag_id).values_list("name", flat=True)
if results_tag_hexin not in all_tag_name_list:
all_tag_name_list.add(results_tag_hexin)
if len(all_tag_name_list) >= 12:
logging.info("get all_tag_name_list03:%s" % all_tag_name_list)
return {"recommend_name": all_tag_name_list}
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return {"recommend_name": []}
search/views/user.py
View file @ 5995ef9a
......@@ -14,7 +14,7 @@ from libs.es import ESPerform
@bind("physical/search/recommend_user")
def recommend_user(self_user_id,interesting_user_id,offset=0,size=10):
def recommend_user(self_user_id, interesting_user_id, offset=0, size=10):
"""
:remark 点关注推荐用户
:param self_user_id:
......@@ -29,25 +29,29 @@ def recommend_user(self_user_id,interesting_user_id,offset=0,size=10):
if not isinstance(interesting_user_id, int):
interesting_user_id = -1
#获取es链接对象
# 获取es链接对象
es_cli_obj = ESPerform.get_cli()
#获取关注用户列表
(self_attention_user_id_list,recursion_attention_user_id_list) = UserUtils.get_attention_user_list([self_user_id,interesting_user_id],self_user_id,es_cli_obj)
# 获取关注用户列表
(self_attention_user_id_list, recursion_attention_user_id_list) = UserUtils.get_attention_user_list(
[self_user_id, interesting_user_id], self_user_id, es_cli_obj)
#去除自身及感兴趣的用户ID
# 去除自身及感兴趣的用户ID
self_attention_user_id_list.append(self_user_id)
self_attention_user_id_list.append(interesting_user_id)
recommend_user_list = UserUtils.get_recommend_user_list(self_attention_user_id_list,recursion_attention_user_id_list,offset,size,es_cli_obj)
recommend_user_list = UserUtils.get_recommend_user_list(self_attention_user_id_list,
recursion_attention_user_id_list, offset, size,
es_cli_obj)
return recommend_user_list
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return []
@bind("physical/search/batch_recommend_user")
def batch_recommend_user(self_user_id,interesting_user_id_list,offset=0,size=10):
def batch_recommend_user(self_user_id, interesting_user_id_list, offset=0, size=10):
"""
:remark 点关注推荐用户 - 批量接口
:param self_user_id:
......@@ -57,23 +61,92 @@ def batch_recommend_user(self_user_id,interesting_user_id_list,offset=0,size=10)
:return:
"""
try:
if not isinstance(self_user_id,int):
if not isinstance(self_user_id, int):
self_user_id = -1
#获取es链接对象
# 获取es链接对象
es_cli_obj = ESPerform.get_cli()
#获取关注用户列表
(need_filter_attention_user_id_list, attention_user_dict_list,attention_user_id_list) = UserUtils.get_batch_attention_user_list(interesting_user_id_list,self_user_id,es_cli_obj)
# 获取关注用户列表
(need_filter_attention_user_id_list, attention_user_dict_list,
attention_user_id_list) = UserUtils.get_batch_attention_user_list(interesting_user_id_list, self_user_id,
es_cli_obj)
#去除自身及感兴趣的用户ID
# 去除自身及感兴趣的用户ID
need_filter_attention_user_id_list.append(self_user_id)
recommend_user_dict = UserUtils.get_batch_recommend_user_dict(need_filter_attention_user_id_list=need_filter_attention_user_id_list,attention_user_id_list=attention_user_id_list,attention_user_dict_list=attention_user_dict_list,self_user_id=self_user_id,offset=offset,size=size,es_cli_obj=es_cli_obj)
recommend_user_dict = UserUtils.get_batch_recommend_user_dict(
need_filter_attention_user_id_list=need_filter_attention_user_id_list,
attention_user_id_list=attention_user_id_list, attention_user_dict_list=attention_user_dict_list,
self_user_id=self_user_id, offset=offset, size=size, es_cli_obj=es_cli_obj)
logging.info("duan add,recommend_user_dict:%s" % str(recommend_user_dict))
return recommend_user_dict
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return {}
\ No newline at end of file
return {}
@bind("physical/search/search_user")
def search_user(query="", offset=0, size=10):
"""
:remark 搜索用户
:param query:
:param offset:
:param size:
:return:
搜索域：用户名称
1.识别query：允许模糊匹配
2.召回部分：全部 线上状态用户
3.rank部分：一级-是否推荐，二级-发帖数
"""
try:
multi_match = {
"fields": ["nick_name"],
"type": "cross_fields",
"operator": "and",
"query": query
}
q = {}
q["query"] = {
"bool": {
"must": [{
"multi_match": multi_match
}, {
"term": {
"is_online": True
}
}
]
}
}
q["sort"] = {
"is_recommend": {
"order": "asc"
},
"count_topic": {
"order": "desc"
},
}
result_dict = ESPerform.get_analyze_results(
ESPerform.get_cli(), sub_index_name="user", query_body=q, offset=offset, size=size
)
res = result_dict["hits"]["hits"]
user_id = [item["_source"]["user_id"] for item in res]
return {"search_user_id": user_id}
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return {"search_user_id": []}
trans2es/mapping/user.json
View file @ 5995ef9a
......@@ -60,6 +60,7 @@
"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"}
"update_time_val":{"type":"long"},
"count_topic":{"type":"long"}
}
}
\ No newline at end of file
trans2es/models/tag.py
View file @ 5995ef9a
......@@ -12,10 +12,10 @@ import datetime
class TopicTag(models.Model):
class Meta:
verbose_name=u"帖子标签"
db_table="community_topictag"
verbose_name = u"帖子标签"
db_table = "community_topictag"
id = models.IntegerField(primary_key=True,verbose_name=u"主键ID")
id = models.IntegerField(primary_key=True, verbose_name=u"主键ID")
tag_id = models.IntegerField(verbose_name=u"标签ID")
topic_id = models.IntegerField(verbose_name=u"帖子ID")
is_online = models.BooleanField(verbose_name=u"是否在线")
......@@ -25,73 +25,84 @@ class TopicTag(models.Model):
class AccountUserTag(models.Model):
class Meta:
verbose_name=u"用户标签"
db_table="account_user_tag"
verbose_name = u"用户标签"
db_table = "account_user_tag"
id = models.IntegerField(primary_key=True,verbose_name=u"主键ID")
id = models.IntegerField(primary_key=True, verbose_name=u"主键ID")
tag_id = models.IntegerField(verbose_name=u"标签ID")
user = models.IntegerField(verbose_name=u"用户ID")
is_deleted = models.BooleanField(verbose_name=u"是否删除")
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))
is_body_esthetics = models.IntegerField(verbose_name=u"是否有用标签")
class Tag(models.Model):
class Meta:
verbose_name=u"标签"
db_table="community_tag"
verbose_name = u"标签"
db_table = "community_tag"
id = models.IntegerField(primary_key=True,verbose_name=u"主键ID")
id = models.IntegerField(primary_key=True, verbose_name=u"主键ID")
is_deleted = models.BooleanField(verbose_name=u"是否删除")
is_online = models.BooleanField(verbose_name=u"是否上线")
name = models.CharField(verbose_name=u"标签名称",max_length=128)
name = models.CharField(verbose_name=u"标签名称", max_length=128)
description = models.TextField(verbose_name=u"标签描述")
icon_url=models.CharField(verbose_name=u"icon_url",max_length=120)
icon_url = models.CharField(verbose_name=u"icon_url", max_length=120)
collection = models.IntegerField(verbose_name=u"是否编辑")
is_ai = models.IntegerField(verbose_name=u"是否ai")
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))
class CommunityTagFollow(models.Model):
class Meta:
verbose_name=u"用户关注标签"
db_table="community_tag_follow"
id = models.IntegerField(primary_key=True,verbose_name=u"主键ID")
verbose_name = u"用户关注标签"
db_table = "community_tag_follow"
id = models.IntegerField(primary_key=True, verbose_name=u"主键ID")
is_deleted = models.BooleanField(verbose_name=u"是否删除")
is_online = models.BooleanField(verbose_name=u"是否上线")
user_id = models.IntegerField(verbose_name=u"用户ID")
tag_id = models.IntegerField(verbose_name=u"帖子ID")
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))
class CommunityTagTypeRelation(models.Model):
class Meta:
verbose_name=u"标签类型对应关系"
db_table="community_tag_type_relation"
verbose_name = u"标签类型对应关系"
db_table = "community_tag_type_relation"
id = models.IntegerField(primary_key=True,verbose_name=u"主键ID")
id = models.IntegerField(primary_key=True, verbose_name=u"主键ID")
is_deleted = models.BooleanField(verbose_name=u"是否删除")
is_online = models.BooleanField(verbose_name=u"是否上线")
tag_id = models.IntegerField(verbose_name=u"标签ID")
tag_type_id = models.IntegerField(verbose_name=u"标签类型ID")
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))
class CommunityTagType(models.Model):
class Meta:
verbose_name=u"标签类型"
db_table="community_tag_type"
verbose_name = u"标签类型"
db_table = "community_tag_type"
id = models.IntegerField(primary_key=True,verbose_name=u"主键ID")
id = models.IntegerField(primary_key=True, verbose_name=u"主键ID")
is_deleted = models.BooleanField(verbose_name=u"是否删除")
is_online = models.BooleanField(verbose_name=u"是否上线")
name = models.CharField(verbose_name=u"名称",max_length=64)
create_time = models.DateTimeField(verbose_name=u'创建时间',default=datetime.datetime.fromtimestamp(0))
name = models.CharField(verbose_name=u"名称", max_length=64)
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))
class RegisterShowTag(models.Model):
class Meta:
verbose_name = "搜索热词"
db_table = "register_show_tag"
id = models.IntegerField(primary_key=True, verbose_name=u"主键ID")
is_deleted = models.IntegerField(verbose_name=u"是否删除")
is_online = models.IntegerField(verbose_name=u"是否在线")
tag_id = models.IntegerField(verbose_name="标签ID")
trans2es/models/user.py
View file @ 5995ef9a
......@@ -16,6 +16,8 @@ from .tag import AccountUserTag
from .topic import Topic
from .user_extra import UserExtra
from .pictorial import PictorialFollow
from .topic import Topic
class User(models.Model):
class Meta:
......@@ -35,9 +37,10 @@ class User(models.Model):
update_time = models.DateTimeField(verbose_name=u'更新时间', default=datetime.datetime.fromtimestamp(0))
@classmethod
def get_user_nick_name(cls,user_id):
def get_user_nick_name(cls, user_id):
try:
nick_name = User.objects.using(settings.SLAVE_DB_NAME).filter(user_id=user_id).values_list("nick_name").first()
nick_name = User.objects.using(settings.SLAVE_DB_NAME).filter(user_id=user_id).values_list(
"nick_name").first()
return nick_name[0]
except:
......@@ -47,7 +50,8 @@ class User(models.Model):
def get_is_recommend_flag(self):
is_shadow = False
is_recommend = False
query_sql = UserExtra.objects.using(settings.SLAVE_DB_NAME).filter(user_id=self.user_id, is_deleted=False, is_online=True)
query_sql = UserExtra.objects.using(settings.SLAVE_DB_NAME).filter(user_id=self.user_id, is_deleted=False,
is_online=True)
for record in query_sql:
is_recommend = record.is_recommend
is_shadow = record.is_shadow
......@@ -58,8 +62,9 @@ class User(models.Model):
latest_topic_time_val = -1
# 获取该用户最新发帖时间
topic_records = Topic.objects.using(settings.SLAVE_DB_NAME).filter(user_id=self.user_id).order_by("-update_time").values_list("update_time",
flat=True).first()
topic_records = Topic.objects.using(settings.SLAVE_DB_NAME).filter(user_id=self.user_id).order_by(
"-update_time").values_list("update_time",
flat=True).first()
if topic_records:
tzlc_topic_update_time = tzlc(topic_records)
latest_topic_time_val = int(time.mktime(tzlc_topic_update_time.timetuple()))
......@@ -75,7 +80,8 @@ class User(models.Model):
logging.info("get follow_user_id_list :%s" % follow_user_id_list)
for i in range(0, len(follow_user_id_list), 1000):
logging.info("get follow_user_id_list :%s" % follow_user_id_list[i:i + 1000])
sql_data_list = User.objects.using(settings.SLAVE_DB_NAME).filter(user_id__in=follow_user_id_list[i:i + 1000])
sql_data_list = User.objects.using(settings.SLAVE_DB_NAME).filter(
user_id__in=follow_user_id_list[i:i + 1000])
for detail_data in sql_data_list:
item = {
"user_id": detail_data.user_id,
......@@ -88,7 +94,8 @@ class User(models.Model):
def get_attention_group_id_list(self):
try:
attention_group_id_list = list()
query_results = GroupUserRole.objects.using(settings.SLAVE_DB_NAME).filter(is_online=True, user_id=self.user_id)
query_results = GroupUserRole.objects.using(settings.SLAVE_DB_NAME).filter(is_online=True,
user_id=self.user_id)
for item in query_results:
item_dict = {
"group_id": item.group_id,
......@@ -104,15 +111,16 @@ class User(models.Model):
def get_attention_pictorial_id_list(self):
try:
attention_pictorial_id_list = list()
query_results = PictorialFollow.objects.using(settings.SLAVE_DB_NAME).filter(is_online=True, user_id=self.user_id)
logging.info("get PictorialFollow:%s"%query_results)
query_results = PictorialFollow.objects.using(settings.SLAVE_DB_NAME).filter(is_online=True,
user_id=self.user_id)
logging.info("get PictorialFollow:%s" % query_results)
for item in query_results:
item_dict = {
"pictorial_id": item.pictorial_id,
"update_time_val": time.mktime(tzlc(item.update_time).timetuple())
}
attention_pictorial_id_list.append(item_dict)
logging.info("get user_id:%s" %self.user_id)
logging.info("get user_id:%s" % self.user_id)
logging.info("get same_pictorial_user_id_list:%s" % attention_pictorial_id_list)
......@@ -149,25 +157,29 @@ class User(models.Model):
def get_same_pictorial_user_id_list(self):
#todo 有的同组数据过大，导致celery cpu过高，暂时限制同组的数据大小，后续可能会去掉同组的数据
# todo 有的同组数据过大，导致celery cpu过高，暂时限制同组的数据大小，后续可能会去掉同组的数据
same_pictorial_user_id_list = list()
pictorial_items_list = list(PictorialFollow.objects.using(settings.SLAVE_DB_NAME).filter(user_id=self.user_id).values_list("pictorial_id",flat=True))
pictorial_items_list = list(
PictorialFollow.objects.using(settings.SLAVE_DB_NAME).filter(user_id=self.user_id).values_list(
"pictorial_id", flat=True))
for pictorial_id in pictorial_items_list:
user_items_list = list(PictorialFollow.objects.using(settings.SLAVE_DB_NAME).filter(pictorial_id=pictorial_id).values_list("user_id",flat=True))
user_items_list = list(
PictorialFollow.objects.using(settings.SLAVE_DB_NAME).filter(pictorial_id=pictorial_id).values_list(
"user_id", flat=True))
for user_id in user_items_list:
same_pictorial_user_id_list.append(user_id)
if len(same_pictorial_user_id_list)>=100:
if len(same_pictorial_user_id_list) >= 100:
break
if len(same_pictorial_user_id_list)>=100:
if len(same_pictorial_user_id_list) >= 100:
break
logging.info("get same user_id:%s"%self.user_id)
logging.info("get same_pictorial_user_id_list:%s"%same_pictorial_user_id_list)
logging.info("get same user_id:%s" % self.user_id)
logging.info("get same_pictorial_user_id_list:%s" % same_pictorial_user_id_list)
same_pictorial_detail_list = list()
for i in range(0, len(same_pictorial_user_id_list), 200):
sql_data_list = User.objects.using(settings.SLAVE_DB_NAME).filter(user_id__in=same_pictorial_user_id_list[i:i + 1000])
sql_data_list = User.objects.using(settings.SLAVE_DB_NAME).filter(
user_id__in=same_pictorial_user_id_list[i:i + 1000])
for detail_data in sql_data_list:
item = {
"user_id": detail_data.user_id,
......@@ -181,7 +193,8 @@ class User(models.Model):
try:
user_tag_id_list = list()
query_results = AccountUserTag.objects.using(settings.SLAVE_DB_NAME).filter(user=self.user_id, is_deleted=False)
query_results = AccountUserTag.objects.using(settings.SLAVE_DB_NAME).filter(user=self.user_id,
is_deleted=False)
for item in query_results:
user_tag_id_list.append(item.tag_id)
......@@ -194,11 +207,21 @@ class User(models.Model):
try:
user_useful_tag_id_list = list()
query_results = AccountUserTag.objects.using(settings.SLAVE_DB_NAME).filter(user=self.user_id,is_body_esthetics =1, is_deleted=False)
query_results = AccountUserTag.objects.using(settings.SLAVE_DB_NAME).filter(user=self.user_id,
is_body_esthetics=1,
is_deleted=False)
for item in query_results:
user_useful_tag_id_list.append(item.tag_id)
return user_useful_tag_id_list
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return []
\ No newline at end of file
return []
def get_topic_number(self):
try:
number = Topic.objects.filter(user_id=self.user_id).count()
return number
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return 0
trans2es/type_info.py
View file @ 5995ef9a
......@@ -14,7 +14,7 @@ import sys
from libs.cache import redis_client
import copy
from trans2es.models import topic, user, pick_celebrity, group, celebrity, tag, contrast_similar,pictorial
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
......@@ -196,7 +196,7 @@ class TypeInfo(object):
old_data["is_history"] = True
data_list.append(old_data)
if int_ori_topic_star>=4:
if int_ori_topic_star >= 4:
topic_data_high_star_list.append(old_data)
redis_client.hset(self.physical_topic_star, data["id"], data["content_level"])
# data_list = [
......@@ -207,12 +207,11 @@ class TypeInfo(object):
# ]
# ESPerform.es_helpers_bulk(ESPerform.get_cli(), data_list, "topic-star")
if data["content_level"] and int(data["content_level"])>=4:
if data["content_level"] and int(data["content_level"]) >= 4:
topic_data_high_star_list.append(data)
data_list.append(data)
return (data_list,topic_data_high_star_list)
return (data_list, topic_data_high_star_list)
def elasticsearch_bulk_insert_data(self, sub_index_name, data_list, es=None):
......@@ -255,17 +254,17 @@ class TypeInfo(object):
else:
qs = self.model.objects.all()
end = time.time()
time0=end-begin
time0 = end - begin
begin = time.time()
instance_list = qs.filter(pk__in=pk_list)
end = time.time()
time1=end-begin
time1 = end - begin
begin = time.time()
data_list, topic_data_high_star_list = self.bulk_get_data(instance_list)
end = time.time()
time2=end-begin
time2 = end - begin
begin = time.time()
# logging.info("get sub_index_name:%s"%sub_index_name)
......@@ -277,7 +276,7 @@ class TypeInfo(object):
es=es,
)
if sub_index_name=="topic":
if sub_index_name == "topic":
self.elasticsearch_bulk_insert_data(
sub_index_name="topic-star-routing",
data_list=data_list,
......@@ -285,7 +284,7 @@ class TypeInfo(object):
)
# 同时写4星及以上的帖子
if len(topic_data_high_star_list)>0:
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,
......@@ -293,9 +292,8 @@ class TypeInfo(object):
)
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))
time3 = end - begin
logging.info("duan add,insert_table_by_pk_list time cost:%ds,%ds,%ds,%ds" % (time0, time1, time2, time3))
def insert_table_chunk(self, sub_index_name, table_chunk, es=None):
try:
......@@ -317,7 +315,7 @@ class TypeInfo(object):
auto_create_index=True
)
logging.info("es_helpers_bulk,sub_index_name:%s,data_list len:%d" % (sub_index_name,len(data_list)))
logging.info("es_helpers_bulk,sub_index_name:%s,data_list len:%d" % (sub_index_name, len(data_list)))
stage_3_time = time.time()
end_clock = time.clock()
......@@ -355,8 +353,8 @@ def get_type_info_map():
name='topic-star',
type='topic-star',
model=topic.Topic,
query_deferred=lambda: topic.Topic.objects.all().query,#假的
get_data_func=TopicTransfer.get_topic_data,#假的
query_deferred=lambda: topic.Topic.objects.all().query, # 假的
get_data_func=TopicTransfer.get_topic_data, # 假的
bulk_insert_chunk_size=100,
round_insert_chunk_size=5,
round_insert_period=2,
......@@ -525,4 +523,3 @@ def get_type_info_map():
_get_type_info_map_result = type_info_map
return type_info_map
trans2es/utils/tag_transfer.py
View file @ 5995ef9a
......@@ -8,16 +8,15 @@ import logging
import traceback
from libs.tools import tzlc
from trans2es.models.topic import Topic
from trans2es.models.tag import TopicTag,CommunityTagType,CommunityTagTypeRelation
from trans2es.models.tag import TopicTag, CommunityTagType, CommunityTagTypeRelation
import datetime
from django.conf import settings
class TagTransfer(object):
@classmethod
def get_tag_name_data(cls,instance):
def get_tag_name_data(cls, instance):
try:
res = dict()
res["name"] = instance.name
......@@ -28,7 +27,7 @@ class TagTransfer(object):
return dict()
@classmethod
def get_tag_data(cls,instance):
def get_tag_data(cls, instance):
try:
res = dict()
......@@ -36,14 +35,14 @@ class TagTransfer(object):
tag_name_terms_list = list()
for i in range(len(instance.name)):
for j in range(i,len(instance.name)+1):
for j in range(i, len(instance.name) + 1):
name_term = instance.name[i:j].strip()
if name_term:
tag_name_terms_list.append(name_term.lower())
res["suggest"] = {
"input":tag_name_terms_list,
"contexts":{
"input": tag_name_terms_list,
"contexts": {
"is_online": [instance.is_online],
"is_deleted": [instance.is_deleted]
}
......@@ -54,24 +53,27 @@ class TagTransfer(object):
topic_num = 0
res["near_new_topic_num"] = topic_num
if instance.is_online==True and instance.is_deleted==False:
if instance.is_online == True and instance.is_deleted == False:
topic_id_list = list()
sql_result_results = list(TopicTag.objects.using(settings.SLAVE_DB_NAME).filter(
tag_id=instance.id).values_list("topic_id", "is_online"))
for topic_id,is_online in sql_result_results:
for topic_id, is_online in sql_result_results:
if is_online:
topic_id_list.append(topic_id)
time_base_val = datetime.datetime.strftime(datetime.datetime.now()+datetime.timedelta(-7), "%Y-%m-%d")
time_base_val = datetime.datetime.strftime(datetime.datetime.now() + datetime.timedelta(-7), "%Y-%m-%d")
for topic_begin_index in range(0,len(topic_id_list),100):
cur_topic_num = Topic.objects.using(settings.SLAVE_DB_NAME).filter(id__in=topic_id_list[topic_begin_index:topic_begin_index+100],create_time__gte=time_base_val).count()
for topic_begin_index in range(0, len(topic_id_list), 100):
cur_topic_num = Topic.objects.using(settings.SLAVE_DB_NAME).filter(
id__in=topic_id_list[topic_begin_index:topic_begin_index + 100],
create_time__gte=time_base_val).count()
topic_num += cur_topic_num
res["near_new_topic_num"] = topic_num
tag_type_sql_list = CommunityTagTypeRelation.objects.using(settings.SLAVE_DB_NAME).filter(tag_id=instance.id).values_list("tag_type_id",flat=True)
tag_type_sql_list = CommunityTagTypeRelation.objects.using(settings.SLAVE_DB_NAME).filter(
tag_id=instance.id).values_list("tag_type_id", flat=True)
tag_type_list = list()
for tag_type_id in tag_type_sql_list:
tag_type_list.append(tag_type_id)
......@@ -83,4 +85,4 @@ class TagTransfer(object):
return res
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
return dict()
\ No newline at end of file
return dict()
trans2es/utils/user_transfer.py
View file @ 5995ef9a
......@@ -21,8 +21,9 @@ class UserTransfer(object):
follow_user_detail_list = list()
for i in range(0, len(follow_user_id_list), 1000):
sql_data_list = User.objects.using(settings.SLAVE_DB_NAME).filter(user_id__in=follow_user_id_list[i:i + 1000], is_online=True,
is_deleted=False)
sql_data_list = User.objects.using(settings.SLAVE_DB_NAME).filter(
user_id__in=follow_user_id_list[i:i + 1000], is_online=True,
is_deleted=False)
for detail_data in sql_data_list:
item = {
"user_id": detail_data.user_id,
......@@ -46,7 +47,7 @@ class UserTransfer(object):
res["country_id"] = instance.country_id
res["is_online"] = instance.is_online
res["is_deleted"] = instance.is_deleted
res["count_topic"] = instance.get_topic_number()
try:
(is_recommend, is_shadow) = instance.get_is_recommend_flag()
res["is_recommend"] = is_recommend
......@@ -95,7 +96,6 @@ class UserTransfer(object):
res["attention_pictorial_id_list"] = []
res["same_pictorial_user_id_list"] = []
return res
except:
logging.error("catch exception,err_msg:%s" % traceback.format_exc())
......
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