Django daily - model layer (2)

Several common query

Aggregate query

Aggregation, we should have this concept because the concept of aggregate functions in SQL language of the database which, polymerization, literally speaking, is to put together a class of things, but they are not the same with the grouping.

We know that when we write SQL statements, groupby grouping and aggregate functions are inseparable, because once the select statement inside the packet, then we can only inquire into the group by fields and a field consisting of a number of aggregate functions, do not written on the back of the field will select the error.

So, now that ORM is used to operate the database, then we use aggregate query time, we must follow the laws of the specific use of aggregate queries as follows:

# 首先导入聚合函数,聚合函数的关键字为aggregate
from django.db.models import Max,Min,Sum,Count,Avg

# 大概格式为:models.表名.objects.查询方式(匹配条件).aggregate(聚合函数(表中的字段))
res = models.Book.objects.all().aggregate(Avg('price'))
print(res)

Grouping queries

Keyword grouping queries to annotate, use very simple, in the actual operation of the database it is actually groupby, grouping query will follow the models behind the table name to the group, and for data packets in parentheses after the operation

# 分组查询的格式为:models.表名.objects.annotate(对组内数据做操作)
res = models.Book.objects.annotate(author_number=Count('authors').values('author_num'))
print(res)

F and Q query

The above aggregation and grouping queries are removed from the table of our field, and then we set with a constant good comparison, envisage a scenario, if we need to take the values from two different fields of the database for comparison, how to achieve it? Django which gave us the F and Q queries to solve this problem.

from django.db.models import F,Q
res = models.Book.objects.filter(sell = F('kucun'))
print(res)
# 以上语句可以实现在Book表中,我们以kucun列为依据,来查询得到自己想要的数据,甚至可以在查询的时候直接对数据做操作,比如
res = models.Book.objects.filter(sell = F('kucun')+100)
print(res)

# 而Q查询比F查询更加强大,比如以下情景
# 我们需要根据不止一个条件来筛选记录,而且这些条件并不是and关系,而是or,或者not,此时就可以使用Q查询来解决问题,比如:
res = models.Book.objects.filter(Q(title='python')|Q(kucun=666))
res1 = models.Book.objects.filter(~Q(title='python')|Q(kucun=666))
print(res,res1)
# 以上两个例子,在两个Q之间用|是or的意思,即两者只要有一个成立就算符合条件,~则是取反的意思

Q query there is a better use, we know that Q query is actually a class, so we generated by instantiating an object, and then to realize Q queries and data screening purposes by manipulating the object, such as

from django.db.models import Q
q = Q()
q.connector = 'or'  # 这里可以直接配置其查询的方式,可以是or,可以是not
q.children.append(('title','python'))
q.children.append(('kucun',666))
res = models.Book.objects.filter(q)
print(res)

Related query optimization

ORM query optimization in general there will be four more commonly used methods, they are in pairs, that only and defer, select_relater and prefetch_related, let's compare the use of introduced one by one.

only和defer('字段')
    首先他们的用法都是在关键字后面括号中加入表中有的字段.
    only会将括号内字段对应的所有数据直接封装起来,然后返回给我们的对象中,因为only()最后返回的是一个对象,然后我们对这个对象来进行操作的时候,如果点出来的是之前括号内的字段,那么此次查询就不会再经过数据库,但是如果点出来是括号内不包含的字段,那么每次都会重新查询数据库.
    做一个类比来说,only后面括号内字段所对应的数据相当于是拿出来放在了计算机的缓存中,我们在查询这个字段的时候就直接从缓存里取,而不需要再去查询数据库,可以节约大量查询的时间,当然这只是一个类比,实际情况并不是缓存的概念.
    defer的用法就和only完全相反,defer后面括号内字段对应的数据,在每次查询的时候都需要查询数据库,而括号内不包含的字段对应的数据查询则不需要查询数据库.
    个人理解only和defer的用法,only应该是用在不常修改的字段,也就是说only后面括号内的字段应该是修改次数尽量少的字段,这样既能提升效率,查询出来又不会出错.同理,defer后面括号内则是应该写频繁修改的字段,需要实时监控其变化的那种字段,比较合适.

select_relater和prefetch_related('外键关联表的表名')
    首先,select_relater和prefetch_related的共同点在于其传参的规则是相同的,其后面的括号里面都只能放外键字段,且只能放一对一和一对多的外键字段,不能放多对多的字段,而且可以叠加,即可以select_related(外键字段__外键字段__外键字段)这样放.
    select_related会自动帮我们做连表操作,这些表必须是有外键相关联的,select_related将连表之后的数据全部查询出来,然后封装起来,传给对象.
    而prefetch_related看似也是一个连表操作,其实不是,prefetch_related实际上是一种类似子查询的存在,其会多次查询不同的表,然后把最后的结果返回出来.
    这两种方法的区别在于
    第一种,内部自动连表,消耗的资源和时间都在连表上,对数据库的操作比较少,数据库压力小
    第二种,内部不做连表,消耗的资源和时间都在对数据库的查询上,数据库压力较大,但是整体效率较高

orm common field

ORM is a field in which we models.py-defined table type, the format for custom fields used, commonly used are the following:

# 自增的数字,常用来设置主键,primary_key = True就为主键
AutoField(primary_key = True)

# 最常用的字符类型,相当于python中的varchar,max_length为其最大长度,超过这个长度的会被舍弃
CharField(max_length=32)

# 整型,默认括号内为空
IntegerField()

# 时间格式,常用参数有两个:
DateField()
    auto_now:每次对对象进行操作都会更新当前时间
    auto_now_add:在创建或添加对象的时候会添加这个时间,然后不会改变
# 小数格式的数字,max_digits为数字的总位数,decimal_places为数字的小数的位数
DecimalField(max_digits=8,decimal_places=2)
# 布尔类型,存储到数据库中的时候是0/1,并不是True和False
BooleanField()

orm transaction operations

We know that there is a transaction database, the transaction is ACID four characteristics

Atomicity
consistency
Isolation
Endurance

So we ORM also requires the presence of affairs. To do the following

from django.db import transaction
with transaction.atomic()
    #这里写的所有ORM语句都属于同一个事务
# 事务的结束标志就是这一个缩进