英文版http://books.agiliq.com/projects/django-orm-cookbook/en/latest/
中文版https://django-orm-cookbook-zh-cn.readthedocs.io/zh_CN/latest/query.html

查询和筛选

1. 如何查看Django ORM查询集的原生SQL？

>>> queryset = Event.objects.all()
>>> str(queryset.query)
SELECT "events_event"."id", "events_event"."epic_id",
    "events_event"."details", "events_event"."years_ago"
    FROM "events_event"

2. 如何在Django ORM中使用 OR 查询？

queryset = User.objects.filter(
        first_name__startswith='R'
    ) | User.objects.filter(
    last_name__startswith='D'
)
queryset
<QuerySet [<User: Ricky>, <User: Ritesh>, <User: Radha>, <User: Raghu>, <User: rishab>]>

from django.db.models import Q
qs = User.objects.filter(Q(first_name__startswith='R')|Q(last_name__startswith='D'))

3. 如何在Django ORM中使用 AND 查询？

queryset_1 = User.objects.filter(
    first_name__startswith='R',
    last_name__startswith='D'
)

queryset_2 = User.objects.filter(
    first_name__startswith='R'
) & User.objects.filter(
    last_name__startswith='D'
)

queryset_3 = User.objects.filter(
    Q(first_name__startswith='R') &
    Q(last_name__startswith='D')
)

4. 如何在Django ORM中使用 NOT 查询？

>>> from django.db.models import Q
>>> queryset = User.objects.filter(~Q(id__lt=5))

5. 如何从相同或不同的模型类中联合两个查询集结果？

相同的模型类

>>> q1 = User.objects.filter(id__gte=5)
>>> q1
<QuerySet [<User: Ritesh>, <User: Billy>, <User: Radha>, <User: sohan>, <User: Raghu>, <User: rishab>]>
>>> q2 = User.objects.filter(id__lte=9)
>>> q2
<QuerySet [<User: yash>, <User: John>, <User: Ricky>, <User: sharukh>, <User: Ritesh>, <User: Billy>, <User: Radha>, <User: sohan>, <User: Raghu>]>
>>> q1.union(q2)
<QuerySet [<User: yash>, <User: John>, <User: Ricky>, <User: sharukh>, <User: Ritesh>, <User: Billy>, <User: Radha>, <User: sohan>, <User: Raghu>, <User: rishab>]>
>>> q2.union(q1)
<QuerySet [<User: yash>, <User: John>, <User: Ricky>, <User: sharukh>, <User: Ritesh>, <User: Billy>, <User: Radha>, <User: sohan>, <User: Raghu>, <User: rishab>]>

不同的模型类

Hero.objects.all().values_list(
    "name", "gender"
).union(
Villain.objects.all().values_list(
    "name", "gender"
))

6. 如何选择同一查询集中的某些字段？

>>> queryset = User.objects.filter(
    first_name__startswith='R'
).values('first_name', 'last_name')
>>> queryset
<QuerySet [{'first_name': 'Ricky', 'last_name': 'Dayal'}, {'first_name': 'Ritesh', 'last_name': 'Deshmukh'}, {'first_name': 'Radha', 'last_name': 'George'}, {'first_name': 'Raghu', 'last_name': 'Khan'}, {'first_name': 'Rishabh', 'last_name': 'Deol'}]

>>> queryset = User.objects.filter(
    first_name__startswith='R'
).only("first_name", "last_name")

7. 如何在Django中使用子查询(内连接)？

>>> from django.db.models import Subquery
>>> users = User.objects.all()
>>> UserParent.objects.filter(user_id__in=Subquery(users.values('id')))
<QuerySet [<UserParent: UserParent object (2)>, <UserParent: UserParent object (5)>, <UserParent: UserParent object (8)>]>

8. 如何基于字段值比较标准来筛选字符集？

In [29]: User.objects.filter(last_name=F("first_name"))
Out[29]: <QuerySet [<User: Guido>]>

9. 如何筛选没有任何文件的FileField字段？

no_files_objects = MyModel.objects.filter(
    Q(file='')|Q(file=None)
)

10. 如何在Django ORM中执行JOIN操作？

reporter__username=‘John’

>>> a1 = Article.objects.select_related('reporter') // Using select_related
>>> a1
<QuerySet [<Article: International News>, <Article: Local News>, <Article: Morning news>, <Article: Prime time>, <Article: Test Article>, <Article: Weather Report>]>
>>> print(a1.query)
SELECT "events_article"."id", "events_article"."headline", "events_article"."pub_date", "events_article"."reporter_id", "events_article"."slug", "auth_user"."id", "auth_user"."password", "auth_user"."last_login", "auth_user"."is_superuser", "auth_user"."username", "auth_user"."first_name", "auth_user"."last_name", "auth_user"."email", "auth_user"."is_staff", "auth_user"."is_active", "auth_user"."date_joined" FROM "events_article" INNER JOIN "auth_user" ON ("events_article"."reporter_id" = "auth_user"."id") ORDER BY "events_article"."headline" ASC
>>> a2 = Article.objects.filter(reporter__username='John')
>>> a2
<QuerySet [<Article: International News>, <Article: Local News>, <Article: Prime time>, <Article: Test Article>, <Article: Weather Report>]>
>>> print(a2.query)
SELECT "events_article"."id", "events_article"."headline", "events_article"."pub_date", "events_article"."reporter_id", "events_article"."slug" FROM "events_article" INNER JOIN "auth_user" ON ("events_article"."reporter_id" = "auth_user"."id") WHERE "auth_user"."username" = John ORDER BY "events_article"."headline" ASC

11. 如何使用Django ORM找到排行第二的记录？

>>> user = User.objects.order_by('-last_login')[1] // Second Highest record w.r.t 'last_login'
>>> user.first_name
'Raghu'
>>> user = User.objects.order_by('-last_login')[2] // Third Highest record w.r.t 'last_login'
>>> user.first_name
'Sohan'

12. 如何查找具有重复字段值的列？

>>> duplicates = User.objects.values(
    'first_name'
    ).annotate(name_count=Count('first_name')).filter(name_count__gt=1)
>>> duplicates
<QuerySet [{'first_name': 'John', 'name_count': 3}]>

13. 如何从查询集中找到独一无二的字段值？

distinct = User.objects.values(
    'first_name'
).annotate(
    name_count=Count('first_name')
).filter(name_count=1)
records = User.objects.filter(first_name__in=[item['first_name'] for item in distinct])

这个和 User.objects.distinct(“first_name”).all() 不同， User.objects.distinct(“first_name”).all() 会获取遇到的不同的first_name时的第一条记录。

14. 如何使用Q对象进行复杂查询？

>>> queryset = User.objects.filter(
    Q(first_name__startswith='R') & ~Q(last_name__startswith='Z')
)

15. 如何在Django ORM中使用聚合查询？

>>> from django.db.models import Avg, Max, Min, Sum, Count
>>> User.objects.all().aggregate(Avg('id'))
{'id__avg': 7.571428571428571}
>>> User.objects.all().aggregate(Max('id'))
{'id__max': 15}
>>> User.objects.all().aggregate(Min('id'))
{'id__min': 1}
>>> User.objects.all().aggregate(Sum('id'))
{'id__sum': 106}

16. 如何从模型中有效地选取一个随机对象？

def get_random():
    return Category.objects.order_by("?").first()

In [8]: def get_random3():
   ...:     max_id = Category.objects.all().aggregate(max_id=Max("id"))['max_id']
   ...:     while True:
   ...:         pk = random.randint(1, max_id)
   ...:         category = Category.objects.filter(pk=pk).first()
   ...:         if category:
   ...:             return category

17. 如何在查询集中使用任意的数据库函数？（不懂）

from django.db.models import Func, F
Hero.objects.annotate(like_zeus=Func(F('name'), function='levenshtein', template="%(function)s(%(expressions)s, 'Zeus')"))

增删改

1. 如何一次创建多个对象？

>>> Category.objects.all().count()
2
>>> Category.objects.bulk_create(
    [Category(name="God"),
     Category(name="Demi God"),
     Category(name="Mortal")]
)
[<Category: God>, <Category: Demi God>, <Category: Mortal>]
>>> Category.objects.all().count()
5

2. 如何复制一个现有的模型对象？

In [2]: Hero.objects.all().count()
Out[2]: 4

In [3]: hero = Hero.objects.first()

In [4]: hero.pk = None

In [5]: hero.save()

In [6]: Hero.objects.all().count()
Out[6]: 5

3. 如何确保只有一个对象被创建？

这常被用来做应用程序配置存储。

class Origin(models.Model):
    name = models.CharField(max_length=100)

    def save(self, *args, **kwargs):
        if self.__class__.objects.count():
            self.pk = self.__class__.objects.first().pk
        super().save(*args, **kwargs)

4. 如何在保存时更新其他模型中的非规范化字段？

只介绍save用法，signals没有介绍。

class Hero(models.Model):
    # ...

    def save(self, *args, **kwargs):
        if not self.pk:
            Category.objects.filter(pk=self.category_id).update(hero_count=F('hero_count')+1)
        super().save(*args, **kwargs)


class Villain(models.Model):
    # ...

    def save(self, *args, **kwargs):
        if not self.pk:
            Category.objects.filter(pk=self.category_id).update(villain_count=F('villain_count')+1)
        super().save(*args, **kwargs)

5. 如何使用Django ORM执行类似truncate操作？（不懂）

class Category(models.Model):
    # ...

    @classmethod
    def truncate(cls):
        with connection.cursor() as cursor:
            cursor.execute('TRUNCATE TABLE "{0}" CASCADE'.format(cls._meta.db_table))

6. Django在对象创建和更新时，会引发哪些信号？（不懂）

7. 如何将字符串转化为datetime并存入数据库？

>>> user = User.objects.get(id=1)
>>> date_str = "2018-03-11"
>>> from django.utils.dateparse import parse_date // Way 1
>>> temp_date = parse_date(date_str)
>>> a1 = Article(headline="String converted to date", pub_date=temp_date, reporter=user)
>>> a1.save()
>>> a1.pub_date
datetime.date(2018, 3, 11)
>>> from datetime import datetime // Way 2
>>> temp_date = datetime.strptime(date_str, "%Y-%m-%d").date()
>>> a2 = Article(headline="String converted to date way 2", pub_date=temp_date, reporter=user)
>>> a2.save()
>>> a2.pub_date
datetime.date(2018, 3, 11)

排序

1. 如果以升序或降序方式给查询集排序？

User.objects.all().order_by('date_joined', '-last_login')

2. 如何以不区分大小写的方式排序查询集？

>>> from django.db.models.functions import Lower
>>> User.objects.all().order_by(Lower('username')).values_list('username', flat=True)
<QuerySet ['Billy', 'John', 'johny', 'johny1', 'paul', 'Radha', 'Raghu', 'Ricky', 'rishab', 'Ritesh', 'sharukh', 'sohan', 'yash']>

3. 如何给两个字段排序？

User.objects.all().order_by(“is_active”, “-last_login”, “first_name”)

4. 如何给关联模型(使用外键)的一个字段排序？

class Category(models.Model):
    name = models.CharField(max_length=100)


class Hero(models.Model):
    # ...
    name = models.CharField(max_length=100)
    category = models.ForeignKey(Category, on_delete=models.CASCADE)

Hero.objects.all().order_by(
    'category__name', 'name'
)

5. 如何给注解字段排序？

Category.objects.annotate(
    hero_count=Count("hero")
).order_by(
    "-hero_count"
)

数据库模型

1. 如何建立一对一的关系模型？

from django.contrib.auth.models import User

class UserParent(models.Model):
    user = models.OneToOneField(
        User,
        on_delete=models.CASCADE,
        primary_key=True,
    )
    father_name = models.CharField(max_length=100)
    mother_name = models.CharField(max_length=100)

>>> u1 = User.objects.get(first_name='Ritesh', last_name='Deshmukh')
>>> u2 = User.objects.get(first_name='Sohan', last_name='Upadhyay')
>>> p1 = UserParent(user=u1, father_name='Vilasrao Deshmukh', mother_name='Vaishali Deshmukh')
>>> p1.save()
>>> p1.user.first_name
'Ritesh'
>>> p2 = UserParent(user=u2, father_name='Mr R S Upadhyay', mother_name='Mrs S K Upadhyay')
>>> p2.save()
>>> p2.user.last_name
'Upadhyay'

2. 如何建立一对多的关系模型？

class Article(models.Model):
    headline = models.CharField(max_length=100)
    pub_date = models.DateField()
    reporter = models.ForeignKey(User, on_delete=models.CASCADE, related_name='reporter')

    def __str__(self):
        return self.headline

    class Meta:
        ordering = ('headline',)

>>> u1 = User(username='johny1', first_name='Johny', last_name='Smith', email='johny@example.com')
>>> u1.save()
>>> u2 = User(username='alien', first_name='Alien', last_name='Mars', email='alien@example.com')
>>> u2.save()
>>> from datetime import date
>>> a1 = Article(headline="This is a test", pub_date=date(2018, 3, 6), reporter=u1)
>>> a1.save()
>>> a1.reporter.id
13
>>> a1.reporter
<User: johny1>

3. 如何建立多对多的模型关系？（不懂）

class User(AbstractUser):
    tweet = models.ManyToManyField(Tweet, blank=True)
    follower = models.ManyToManyField(settings.AUTH_USER_MODEL, blank=True)
    pass

class Tweet(models.Model):
    tweet = models.TextField()
    favourite = models.ManyToManyField(settings.AUTH_USER_MODEL, blank=True, related_name='user_favourite')

    def __unicode__(self):
        return self.tweet

以上的模型能做些什么呢？

1）用户可以关注或取关其他用户。

2）用户可以他关注的用户的推文。

3）用户可以喜欢/不喜欢其他人的推文。

4. 如何在模型中包含自引用外键？

自引用外键可以有两种方法实现。

class Employee(models.Model):
    manager = models.ForeignKey('self', on_delete=models.CASCADE)

# OR

class Employee(models.Model):
    manager = models.ForeignKey("app.Employee", on_delete=models.CASCADE)

5. 如何转化已有的数据表到Django模型？

$ python manage.py inspectdb
$ python manage.py inspectdb > models.py

6. 如何为数据库视图添加模型？

create view temp_user as select id, first_name from auth_user;

class TempUser(models.Model):
    first_name = models.CharField(max_length=100)

    class Meta:
        managed = False
        db_table = "temp_user"

7. 如何创建可与任何实体相关联的通用模型（如类别或评论）？（不懂）

太难，略

8. 如何指定模型的表名？

class TempUser(models.Model):
    first_name = models.CharField(max_length=100)
    . . .
    class Meta:
        db_table = "temp_user"

9. 如何给模型字段指定列名？

class ColumnName(models.Model):
    a = models.CharField(max_length=40,db_column='column1')
    column2 = models.CharField(max_length=50)

    def __str__(self):
        return self.a

10. null=True 和 blank=True 的区别？

1. null 和 blank 的默认值都是 False 。他们两者的值都在字段级别工作，即，我们是否要保持字段null还是blank。

2.null=True 将会设置字段的值为空，即没有数据。这基本上用于数据库列值。

date = models.DateTimeField(null=True)

3.blank=True 决定在表格中是否需要该字段。这包含了admin表单和你自己的表单。

title = models.CharField(blank=True) // title can be kept blank. In the database ("") will be stored.

4.null=True blank=True 这意味着这个字段在所有情况下是可选的。

epic = models.ForeignKey(null=True, blank=True) // The exception is CharFields() and TextFields(), which in Django are never saved as NULL. Blank values are stored in the DB as an empty string (‘’).

但也有个特殊例子，但你需要接受一个 BooleanField 的Null值时，请使用 NullBooleanField 。

11. 如何使用UUID替代ID作为主键？

import uuid
from django.db import models

class Event(models.Model):
    id = models.UUIDField(primary_key=True, default=uuid.uuid4, editable=False)
    details = models.TextField()
    years_ago = models.PositiveIntegerField()

>>> eventobject = Event.objects.all()
>>> eventobject.first().id
'3cd2b4b0c36f43488a93b3bb72029f46'

12. 如何使用Django的slug字段提高代码可读性？

from django.utils.text import slugify
class Article(models.Model):
    headline = models.CharField(max_length=100)
    . . .
    slug = models.SlugField(unique=True)

    def save(self, *args, **kwargs):
        self.slug = slugify(self.headline)
        super(Article, self).save(*args, **kwargs)
    . . .

>>> u1 = User.objects.get(id=1)
>>> from datetime import date
>>> a1 = Article.objects.create(headline="todays market report", pub_date=date(2018, 3, 6), reporter=u1)
>>> a1.save()
// slug here is auto-generated, we haven't created it in the above create method.
>>> a1.slug
'todays-market-report'

13. 如果添加多个数据到Django应用？（难）

1.数据库配置相关的东西基本上在 settings.py 中被完成。所以，要给Django应用添加多个数据库，我们需要把他们添加到 DATABASES 的字典中。

DATABASE_ROUTERS = ['path.to.DemoRouter']
DATABASE_APPS_MAPPING = {'user_data': 'users_db',
                        'customer_data':'customers_db'}

DATABASES = {
    'default': {
        'ENGINE': 'django.db.backends.sqlite3',
        'NAME': os.path.join(BASE_DIR, 'db.sqlite3'),
    },
    'users_db': {
        'NAME': 'user_data',
        'ENGINE': 'django.db.backends.postgresql',
        'USER': 'postgres_user',
        'PASSWORD': 'password'
    },
    'customers_db': {
        'NAME': 'customer_data',
        'ENGINE': 'django.db.backends.mysql',
        'USER': 'mysql_cust',
        'PASSWORD': 'root'
    }
}

2.对于多个数据库，我们最好聊聊 Database Router 。默认的路由方案如果一个数据库未指明，所有的查询都会返回到默认数据库。 Database Router 默认是 [] 。

class DemoRouter:
    """
    A router to control all database operations on models in the
    user application.
    """
    def db_for_read(self, model, **hints):
        """
        Attempts to read user models go to users_db.
        """
        if model._meta.app_label == 'user_data':
            return 'users_db'
        return None

    def db_for_write(self, model, **hints):
        """
        Attempts to write user models go to users_db.
        """
        if model._meta.app_label == 'user_data':
            return 'users_db'
        return None

    def allow_relation(self, obj1, obj2, **hints):
        """
        Allow relations if a model in the user app is involved.
        """
        if obj1._meta.app_label == 'user_data' or \
           obj2._meta.app_label == 'user_data':
           return True
        return None

    def allow_migrate(self, db, app_label, model_name=None, **hints):
        """
        Make sure the auth app only appears in the 'users_db'
        database.
        """
        if app_label == 'user_data':
            return db == 'users_db'
        return None

3.各个模型将被修改为：

class User(models.Model):
    username = models.Charfield(ax_length=100)
    . . .
        class Meta:
        app_label = 'user_data'

class Customer(models.Model):
    name = models.TextField(max_length=100)
    . . .
        class Meta:
        app_label = 'customer_data'

4.使用多个数据库是的有用的命令:

$ ./manage.py migrate --database=users_db

测试

1. 如何断言一个使用了固定查询数量的函数？（没用过）

def test_number_of_queries(self):
    User.objects.create(username='testuser1', first_name='Test', last_name='user1')
    # Above ORM create will run only one query.
    self.assertNumQueries(1)
    User.objects.filter(username='testuser').update(username='test1user')
    # One more query added.
    self.assertNumQueries(2)

2. 如何通过在测试运行中重用数据库来加速测试？（没用过）

当我们执行命令 python manage.py test ，一个新的数据库每次会被创建。如果我们没有太多迁移，这并没有什么问题。 但是当我们有许多迁移时，这将耗费大量的时间在测试运行时来重新创建数据库。为了避免这种清况，我们可以重用旧数据库。

你可以通过添加 --keepdb 标识到测试命令后以防止数据库被破坏。这将保存在测试过程中保存测试数据库。如果数据库不存在，它首先会被创建。如果有任何迁移在最后一次测试运行后被添加，为保证同步，这些迁移会被应用。

$ python manage.py test --keepdb

3. 如何从数据库中重载一个模型对象？（没用过）

通过使用 refresh_from_db() 方法，模型就可以从数据库中被从新加载。这在测试中非常有用。例如：

class TestORM(TestCase):
    def test_update_result(self):
        userobject = User.objects.create(username='testuser', first_name='Test', last_name='user')
        User.objects.filter(username='testuser').update(username='test1user')
        # At this point userobject.val is still testuser, but the value in the database
        # was updated to test1user. The object's updated value needs to be reloaded
        # from the database.
        userobject.refresh_from_db()
        self.assertEqual(userobject.username, 'test1user')