Kiedy stworzyłeś już swoje modele danych, Django automatycznie daje ci API abstrakcji nad bazą danych, które pozwala dodawać, pobierać, aktualizować i usuwać obiekty. Ten dokument wyjaśnia jak używać tego API. Przejdź do dokumentacji modelu danych po pełne szczegóły wszystkich różnych opcji lookupów modeli.
\nThroughout this guide (and in the reference), we’ll refer to the following\nmodels, which comprise a blog application:
\nfrom datetime import date\n\nfrom django.db import models\n\nclass Blog(models.Model):\n name = models.CharField(max_length=100)\n tagline = models.TextField()\n\n def __str__(self):\n return self.name\n\nclass Author(models.Model):\n name = models.CharField(max_length=200)\n email = models.EmailField()\n\n def __str__(self):\n return self.name\n\nclass Entry(models.Model):\n blog = models.ForeignKey(Blog, on_delete=models.CASCADE)\n headline = models.CharField(max_length=255)\n body_text = models.TextField()\n pub_date = models.DateField()\n mod_date = models.DateField(default=date.today)\n authors = models.ManyToManyField(Author)\n number_of_comments = models.IntegerField(default=0)\n number_of_pingbacks = models.IntegerField(default=0)\n rating = models.IntegerField(default=5)\n\n def __str__(self):\n return self.headline\nAby przełożyć dane tabel bazy danych na pythonowe obiekty, Django używa intuicyjnego systemu: Klasa modelu reprezentuje tabelę w bazie danych a instancja tej klasy reprezentuje pojedynczy rekord w tabeli bazy danych.
\nAby stworzyć obiekt, zinstancjonuj go używając keyword argumentów na klasie modelu, następnie wywołaj save(), aby zapisać go w bazie danych.
Zakładając, że modele są w pliku mysite/blog/models.py, oto przykład:
>>> from blog.models import Blog\n>>> b = Blog(name='Beatles Blog', tagline='All the latest Beatles news.')\n>>> b.save()\nWykonuje on „pod maską” instrukcję SQL INSERT. Django nie łączy się z bazą danych dopóki nie wywołasz wprost save().
Metoda save() nie zwraca wartości.
Aby zapisać zmiany w obiekcie, który jest już w bazie danych, użyj save().
Mając instancję klasy Blog b5, która już została zapisana w bazie danych, ten przykład zmienia jej nazwę i aktualizuje jej rekord w bazie danych:
>>> b5.name = 'New name'\n>>> b5.save()\nWykonuje on „pod maską” instrukcję SQL UPDATE. Django nie łączy się z bazą danych dopóki nie wywołasz wprost save().
ForeignKey i ManyToManyFieldLink to this heading#Updating a ForeignKey field works exactly the same\nway as saving a normal field – assign an object of the right type to the field\nin question. This example updates the blog attribute of an Entry\ninstance entry, assuming appropriate instances of Entry and Blog\nare already saved to the database (so we can retrieve them below):
>>> from blog.models import Blog, Entry\n>>> entry = Entry.objects.get(pk=1)\n>>> cheese_blog = Blog.objects.get(name="Cheddar Talk")\n>>> entry.blog = cheese_blog\n>>> entry.save()\nUpdating a ManyToManyField works a little\ndifferently – use the\nadd() method on the field\nto add a record to the relation. This example adds the Author instance\njoe to the entry object:
>>> from blog.models import Author\n>>> joe = Author.objects.create(name="Joe")\n>>> entry.authors.add(joe)\nTo add multiple records to a ManyToManyField in one\ngo, include multiple arguments in the call to\nadd(), like this:
>>> john = Author.objects.create(name="John")\n>>> paul = Author.objects.create(name="Paul")\n>>> george = Author.objects.create(name="George")\n>>> ringo = Author.objects.create(name="Ringo")\n>>> entry.authors.add(john, paul, george, ringo)\nDjango will complain if you try to assign or add an object of the wrong type.
\nTo retrieve objects from your database, construct a\nQuerySet via a\nManager on your model class.
A QuerySet represents a collection of objects\nfrom your database. It can have zero, one or many filters. Filters narrow\ndown the query results based on the given parameters. In SQL terms, a\nQuerySet equates to a SELECT statement,\nand a filter is a limiting clause such as WHERE or LIMIT.
You get a QuerySet by using your model’s\nManager. Each model has at least one\nManager, and it’s called\nobjects by default. Access it directly via the\nmodel class, like so:
>>> Blog.objects\n<django.db.models.manager.Manager object at ...>\n>>> b = Blog(name='Foo', tagline='Bar')\n>>> b.objects\nTraceback:\n ...\nAttributeError: "Manager isn't accessible via Blog instances."\nThe Manager is the main source of QuerySets for\na model. For example, Blog.objects.all() returns a\nQuerySet that contains all Blog objects in\nthe database.
The simplest way to retrieve objects from a table is to get all of them. To do\nthis, use the all() method on a\nManager:
>>> all_entries = Entry.objects.all()\nThe all() method returns a\nQuerySet of all the objects in the database.
The QuerySet returned by\nall() describes all objects in the\ndatabase table. Usually, though, you’ll need to select only a subset of the\ncomplete set of objects.
To create such a subset, you refine the initial\nQuerySet, adding filter conditions. The two\nmost common ways to refine a QuerySet are:
filter(**kwargs)Returns a new QuerySet containing objects\nthat match the given lookup parameters.
exclude(**kwargs)Returns a new QuerySet containing objects\nthat do not match the given lookup parameters.
The lookup parameters (**kwargs in the above function definitions) should\nbe in the format described in Field lookups below.
For example, to get a QuerySet of blog entries\nfrom the year 2006, use filter() like\nso:
Entry.objects.filter(pub_date__year=2006)\nWith the default manager class, it is the same as:
\nEntry.objects.all().filter(pub_date__year=2006)\nThe result of refining a QuerySet is itself a\nQuerySet, so it’s possible to chain\nrefinements together. For example:
>>> Entry.objects.filter(\n... headline__startswith='What'\n... ).exclude(\n... pub_date__gte=datetime.date.today()\n... ).filter(\n... pub_date__gte=datetime.date(2005, 1, 30)\n... )\nThis takes the initial QuerySet of all entries\nin the database, adds a filter, then an exclusion, then another filter. The\nfinal result is a QuerySet containing all\nentries with a headline that starts with „What”, that were published between\nJanuary 30, 2005, and the current day.
QuerySets are uniqueLink to this heading#Each time you refine a QuerySet, you get a\nbrand-new QuerySet that is in no way bound to\nthe previous QuerySet. Each refinement creates\na separate and distinct QuerySet that can be\nstored, used and reused.
Przykład:
\n>>> q1 = Entry.objects.filter(headline__startswith="What")\n>>> q2 = q1.exclude(pub_date__gte=datetime.date.today())\n>>> q3 = q1.filter(pub_date__gte=datetime.date.today())\nThese three QuerySets are separate. The first is a base\nQuerySet containing all entries that contain a\nheadline starting with „What”. The second is a subset of the first, with an\nadditional criteria that excludes records whose pub_date is today or in the\nfuture. The third is a subset of the first, with an additional criteria that\nselects only the records whose pub_date is today or in the future. The\ninitial QuerySet (q1) is unaffected by the\nrefinement process.
QuerySets are lazyLink to this heading#QuerySets are lazy – the act of creating a\nQuerySet doesn’t involve any database\nactivity. You can stack filters together all day long, and Django won’t\nactually run the query until the QuerySet is\nevaluated. Take a look at this example:
>>> q = Entry.objects.filter(headline__startswith="What")\n>>> q = q.filter(pub_date__lte=datetime.date.today())\n>>> q = q.exclude(body_text__icontains="food")\n>>> print(q)\nThough this looks like three database hits, in fact it hits the database only\nonce, at the last line (print(q)). In general, the results of a\nQuerySet aren’t fetched from the database\nuntil you „ask” for them. When you do, the\nQuerySet is evaluated by accessing the\ndatabase. For more details on exactly when evaluation takes place, see\nWhen QuerySets are evaluated.
get()Link to this heading#filter() will always give you a\nQuerySet, even if only a single object matches\nthe query - in this case, it will be a\nQuerySet containing a single element.
If you know there is only one object that matches your query, you can use the\nget() method on a\nManager which returns the object directly:
>>> one_entry = Entry.objects.get(pk=1)\nYou can use any query expression with\nget(), just like with\nfilter() - again, see Field lookups\nbelow.
Note that there is a difference between using\nget(), and using\nfilter() with a slice of [0]. If\nthere are no results that match the query,\nget() will raise a DoesNotExist\nexception. This exception is an attribute of the model class that the query is\nbeing performed on - so in the code above, if there is no Entry object with\na primary key of 1, Django will raise Entry.DoesNotExist.
Similarly, Django will complain if more than one item matches the\nget() query. In this case, it will raise\nMultipleObjectsReturned, which again is an\nattribute of the model class itself.
QuerySet methodsLink to this heading#Most of the time you’ll use all(),\nget(),\nfilter() and\nexclude() when you need to look up\nobjects from the database. However, that’s far from all there is; see the\nQuerySet API Reference for a complete list of all the\nvarious QuerySet methods.
QuerySetsLink to this heading#Use a subset of Python’s array-slicing syntax to limit your\nQuerySet to a certain number of results. This\nis the equivalent of SQL’s LIMIT and OFFSET clauses.
For example, this returns the first 5 objects (LIMIT 5):
>>> Entry.objects.all()[:5]\nThis returns the sixth through tenth objects (OFFSET 5 LIMIT 5):
>>> Entry.objects.all()[5:10]\nNegative indexing (i.e. Entry.objects.all()[-1]) is not supported.
Generally, slicing a QuerySet returns a new\nQuerySet – it doesn’t evaluate the query. An\nexception is if you use the „step” parameter of Python slice syntax. For\nexample, this would actually execute the query in order to return a list of\nevery second object of the first 10:
>>> Entry.objects.all()[:10:2]\nFurther filtering or ordering of a sliced queryset is prohibited due to the\nambiguous nature of how that might work.
\nTo retrieve a single object rather than a list\n(e.g. SELECT foo FROM bar LIMIT 1), use an index instead of a slice. For\nexample, this returns the first Entry in the database, after ordering\nentries alphabetically by headline:
>>> Entry.objects.order_by('headline')[0]\nThis is roughly equivalent to:
\n>>> Entry.objects.order_by('headline')[0:1].get()\nNote, however, that the first of these will raise IndexError while the\nsecond will raise DoesNotExist if no objects match the given criteria. See\nget() for more details.
Field lookups are how you specify the meat of an SQL WHERE clause. They’re\nspecified as keyword arguments to the QuerySet\nmethods filter(),\nexclude() and\nget().
Basic lookups keyword arguments take the form field__lookuptype=value.\n(That’s a double-underscore). For example:
>>> Entry.objects.filter(pub_date__lte='2006-01-01')\ntranslates (roughly) into the following SQL:
\nSELECT * FROM blog_entry WHERE pub_date <= '2006-01-01';\nThe field specified in a lookup has to be the name of a model field. There’s\none exception though, in case of a ForeignKey you\ncan specify the field name suffixed with _id. In this case, the value\nparameter is expected to contain the raw value of the foreign model’s primary\nkey. For example:
>>> Entry.objects.filter(blog_id=4)\nIf you pass an invalid keyword argument, a lookup function will raise\nTypeError.
The database API supports about two dozen lookup types; a complete reference\ncan be found in the field lookup reference. To give you\na taste of what’s available, here’s some of the more common lookups you’ll\nprobably use:
\nexactAn „exact” match. For example:
\n>>> Entry.objects.get(headline__exact="Cat bites dog")\nWould generate SQL along these lines:
\nSELECT ... WHERE headline = 'Cat bites dog';\nIf you don’t provide a lookup type – that is, if your keyword argument\ndoesn’t contain a double underscore – the lookup type is assumed to be\nexact.
For example, the following two statements are equivalent:
\n>>> Blog.objects.get(id__exact=14) # Explicit form\n>>> Blog.objects.get(id=14) # __exact is implied\nThis is for convenience, because exact lookups are the common case.
iexactA case-insensitive match. So, the query:
\n>>> Blog.objects.get(name__iexact="beatles blog")\nWould match a Blog titled "Beatles Blog", "beatles blog", or\neven "BeAtlES blOG".
containsCase-sensitive containment test. For example:
\nEntry.objects.get(headline__contains='Lennon')\nRoughly translates to this SQL:
\nSELECT ... WHERE headline LIKE '%Lennon%';\nNote this will match the headline 'Today Lennon honored' but not\n'today lennon honored'.
There’s also a case-insensitive version, icontains.
startswith, endswithStarts-with and ends-with search, respectively. There are also\ncase-insensitive versions called istartswith and\niendswith.
Again, this only scratches the surface. A complete reference can be found in the\nfield lookup reference.
\nDjango offers a powerful and intuitive way to „follow” relationships in\nlookups, taking care of the SQL JOINs for you automatically, behind the\nscenes. To span a relationship, use the field name of related fields\nacross models, separated by double underscores, until you get to the field you\nwant.
This example retrieves all Entry objects with a Blog whose name\nis 'Beatles Blog':
>>> Entry.objects.filter(blog__name='Beatles Blog')\nThis spanning can be as deep as you’d like.
\nIt works backwards, too. While it can be customized, by default you refer to a „reverse”\nrelationship in a lookup using the lowercase name of the model.
This example retrieves all Blog objects which have at least one Entry\nwhose headline contains 'Lennon':
>>> Blog.objects.filter(entry__headline__contains='Lennon')\nIf you are filtering across multiple relationships and one of the intermediate\nmodels doesn’t have a value that meets the filter condition, Django will treat\nit as if there is an empty (all values are NULL), but valid, object there.\nAll this means is that no error will be raised. For example, in this filter:
Blog.objects.filter(entry__authors__name='Lennon')\n(if there was a related Author model), if there was no author\nassociated with an entry, it would be treated as if there was also no name\nattached, rather than raising an error because of the missing author.\nUsually this is exactly what you want to have happen. The only case where it\nmight be confusing is if you are using isnull. Thus:
Blog.objects.filter(entry__authors__name__isnull=True)\nwill return Blog objects that have an empty name on the author and\nalso those which have an empty author on the entry. If you don’t want\nthose latter objects, you could write:
Blog.objects.filter(entry__authors__isnull=False, entry__authors__name__isnull=True)\nWhen spanning a ManyToManyField or a reverse\nForeignKey (such as from Blog to Entry),\nfiltering on multiple attributes raises the question of whether to require each\nattribute to coincide in the same related object. We might seek blogs that have\nan entry from 2008 with “Lennon” in its headline, or we might seek blogs that\nmerely have any entry from 2008 as well as some newer or older entry with\n“Lennon” in its headline.
To select all blogs containing at least one entry from 2008 having „Lennon”\nin its headline (the same entry satisfying both conditions), we would write:
\nBlog.objects.filter(entry__headline__contains='Lennon', entry__pub_date__year=2008)\nOtherwise, to perform a more permissive query selecting any blogs with merely\nsome entry with „Lennon” in its headline and some entry from 2008, we\nwould write:
\nBlog.objects.filter(entry__headline__contains='Lennon').filter(entry__pub_date__year=2008)\nSuppose there is only one blog that has both entries containing „Lennon” and\nentries from 2008, but that none of the entries from 2008 contained „Lennon”.\nThe first query would not return any blogs, but the second query would return\nthat one blog. (This is because the entries selected by the second filter may\nor may not be the same as the entries in the first filter. We are filtering the\nBlog items with each filter statement, not the Entry items.) In short,\nif each condition needs to match the same related object, then each should be\ncontained in a single filter() call.
In the examples given so far, we have constructed filters that compare\nthe value of a model field with a constant. But what if you want to compare\nthe value of a model field with another field on the same model?
\nDjango provides F expressions to allow such\ncomparisons. Instances of F() act as a reference to a model field within a\nquery. These references can then be used in query filters to compare the values\nof two different fields on the same model instance.
For example, to find a list of all blog entries that have had more comments\nthan pingbacks, we construct an F() object to reference the pingback count,\nand use that F() object in the query:
>>> from django.db.models import F\n>>> Entry.objects.filter(number_of_comments__gt=F('number_of_pingbacks'))\nDjango supports the use of addition, subtraction, multiplication,\ndivision, modulo, and power arithmetic with F() objects, both with constants\nand with other F() objects. To find all the blog entries with more than\ntwice as many comments as pingbacks, we modify the query:
>>> Entry.objects.filter(number_of_comments__gt=F('number_of_pingbacks') * 2)\nTo find all the entries where the rating of the entry is less than the\nsum of the pingback count and comment count, we would issue the\nquery:
\n>>> Entry.objects.filter(rating__lt=F('number_of_comments') + F('number_of_pingbacks'))\nYou can also use the double underscore notation to span relationships in\nan F() object. An F() object with a double underscore will introduce\nany joins needed to access the related object. For example, to retrieve all\nthe entries where the author’s name is the same as the blog name, we could\nissue the query:
>>> Entry.objects.filter(authors__name=F('blog__name'))\nFor date and date/time fields, you can add or subtract a\ntimedelta object. The following would return all entries\nthat were modified more than 3 days after they were published:
>>> from datetime import timedelta\n>>> Entry.objects.filter(mod_date__gt=F('pub_date') + timedelta(days=3))\nThe F() objects support bitwise operations by .bitand(), .bitor(),\n.bitxor(), .bitrightshift(), and .bitleftshift(). For example:
>>> F('somefield').bitand(16)\nDjango supports using transforms in expressions.
\nFor example, to find all Entry objects published in the same year as they\nwere last modified:
>>> from django.db.models import F\n>>> Entry.objects.filter(pub_date__year=F('mod_date__year'))\nTo find the earliest year an entry was published, we can issue the query:
\n>>> from django.db.models import Min\n>>> Entry.objects.aggregate(first_published_year=Min('pub_date__year'))\nThis example finds the value of the highest rated entry and the total number\nof comments on all entries for each year:
\n>>> from django.db.models import OuterRef, Subquery, Sum\n>>> Entry.objects.values('pub_date__year').annotate(\n... top_rating=Subquery(\n... Entry.objects.filter(\n... pub_date__year=OuterRef('pub_date__year'),\n... ).order_by('-rating').values('rating')[:1]\n... ),\n... total_comments=Sum('number_of_comments'),\n... )\npk lookup shortcutLink to this heading#For convenience, Django provides a pk lookup shortcut, which stands for\n„primary key”.
In the example Blog model, the primary key is the id field, so these\nthree statements are equivalent:
>>> Blog.objects.get(id__exact=14) # Explicit form\n>>> Blog.objects.get(id=14) # __exact is implied\n>>> Blog.objects.get(pk=14) # pk implies id__exact\nThe use of pk isn’t limited to __exact queries – any query term\ncan be combined with pk to perform a query on the primary key of a model:
# Get blogs entries with id 1, 4 and 7\n>>> Blog.objects.filter(pk__in=[1,4,7])\n\n# Get all blog entries with id > 14\n>>> Blog.objects.filter(pk__gt=14)\npk lookups also work across joins. For example, these three statements are\nequivalent:
>>> Entry.objects.filter(blog__id__exact=3) # Explicit form\n>>> Entry.objects.filter(blog__id=3) # __exact is implied\n>>> Entry.objects.filter(blog__pk=3) # __pk implies __id__exact\nLIKE statementsLink to this heading#The field lookups that equate to LIKE SQL statements (iexact,\ncontains, icontains, startswith, istartswith, endswith\nand iendswith) will automatically escape the two special characters used in\nLIKE statements – the percent sign and the underscore. (In a LIKE\nstatement, the percent sign signifies a multiple-character wildcard and the\nunderscore signifies a single-character wildcard.)
This means things should work intuitively, so the abstraction doesn’t leak.\nFor example, to retrieve all the entries that contain a percent sign, use the\npercent sign as any other character:
\n>>> Entry.objects.filter(headline__contains='%')\nDjango takes care of the quoting for you; the resulting SQL will look something\nlike this:
\nSELECT ... WHERE headline LIKE '%\\%%';\nSame goes for underscores. Both percentage signs and underscores are handled\nfor you transparently.
\nQuerySetsLink to this heading#Each QuerySet contains a cache to minimize\ndatabase access. Understanding how it works will allow you to write the most\nefficient code.
In a newly created QuerySet, the cache is\nempty. The first time a QuerySet is evaluated\n– and, hence, a database query happens – Django saves the query results in\nthe QuerySet’s cache and returns the results\nthat have been explicitly requested (e.g., the next element, if the\nQuerySet is being iterated over). Subsequent\nevaluations of the QuerySet reuse the cached\nresults.
Keep this caching behavior in mind, because it may bite you if you don’t use\nyour QuerySets correctly. For example, the\nfollowing will create two QuerySets, evaluate\nthem, and throw them away:
>>> print([e.headline for e in Entry.objects.all()])\n>>> print([e.pub_date for e in Entry.objects.all()])\nThat means the same database query will be executed twice, effectively doubling\nyour database load. Also, there’s a possibility the two lists may not include\nthe same database records, because an Entry may have been added or deleted\nin the split second between the two requests.
To avoid this problem, save the QuerySet and\nreuse it:
>>> queryset = Entry.objects.all()\n>>> print([p.headline for p in queryset]) # Evaluate the query set.\n>>> print([p.pub_date for p in queryset]) # Reuse the cache from the evaluation.\nQuerySets are not cachedLink to this heading#Querysets do not always cache their results. When evaluating only part of\nthe queryset, the cache is checked, but if it is not populated then the items\nreturned by the subsequent query are not cached. Specifically, this means that\nlimiting the queryset using an array slice or an\nindex will not populate the cache.
\nFor example, repeatedly getting a certain index in a queryset object will query\nthe database each time:
\n>>> queryset = Entry.objects.all()\n>>> print(queryset[5]) # Queries the database\n>>> print(queryset[5]) # Queries the database again\nHowever, if the entire queryset has already been evaluated, the cache will be\nchecked instead:
\n>>> queryset = Entry.objects.all()\n>>> [entry for entry in queryset] # Queries the database\n>>> print(queryset[5]) # Uses cache\n>>> print(queryset[5]) # Uses cache\nHere are some examples of other actions that will result in the entire queryset\nbeing evaluated and therefore populate the cache:
\n>>> [entry for entry in queryset]\n>>> bool(queryset)\n>>> entry in queryset\n>>> list(queryset)\nIf you are writing asynchronous views or code, you cannot use the ORM for\nqueries in quite the way we have described above, as you cannot call blocking\nsynchronous code from asynchronous code - it will block up the event loop\n(or, more likely, Django will notice and raise a SynchronousOnlyOperation\nto stop that from happening).
Fortunately, you can do many queries using Django’s asynchronous query APIs.\nEvery method that might block - such as get() or delete() - has an\nasynchronous variant (aget() or adelete()), and when you iterate over\nresults, you can use asynchronous iteration (async for) instead.
The default way of iterating over a query - with for - will result in a\nblocking database query behind the scenes as Django loads the results at\niteration time. To fix this, you can swap to async for:
async for entry in Authors.objects.filter(name__startswith="A"):\n ...\nBe aware that you also can’t do other things that might iterate over the\nqueryset, such as wrapping list() around it to force its evaluation (you\ncan use async for in a comprehension, if you want it).
Because QuerySet methods like filter() and exclude() do not\nactually run the query - they set up the queryset to run when it’s iterated\nover - you can use those freely in asynchronous code. For a guide to which\nmethods can keep being used like this, and which have asynchronous versions,\nread the next section.
QuerySet and manager methodsLink to this heading#Some methods on managers and querysets - like get() and first() - force\nexecution of the queryset and are blocking. Some, like filter() and\nexclude(), don’t force execution and so are safe to run from asynchronous\ncode. But how are you supposed to tell the difference?
While you could poke around and see if there is an a-prefixed version of\nthe method (for example, we have aget() but not afilter()), there is a\nmore logical way - look up what kind of method it is in the\nQuerySet reference.
In there, you’ll find the methods on QuerySets grouped into two sections:
\nMethods that return new querysets: These are the non-blocking ones,\nand don’t have asynchronous versions. You’re free to use these in any\nsituation, though read the notes on defer() and only() before you use\nthem.
Methods that do not return querysets: These are the blocking ones, and\nhave asynchronous versions - the asynchronous name for each is noted in its\ndocumentation, though our standard pattern is to add an a prefix.
Using this distinction, you can work out when you need to use asynchronous\nversions, and when you don’t. For example, here’s a valid asynchronous query:
\nuser = await User.objects.filter(username=my_input).afirst()\nfilter() returns a queryset, and so it’s fine to keep chaining it inside an\nasynchronous environment, whereas first() evaluates and returns a model\ninstance - thus, we change to afirst(), and use await at the front of\nthe whole expression in order to call it in an asynchronous-friendly way.
Transactions are not currently supported with asynchronous queries and\nupdates. You will find that trying to use one raises\nSynchronousOnlyOperation.
If you wish to use a transaction, we suggest you write your ORM code inside a\nseparate, synchronous function and then call that using sync_to_async - see\nAsynchronous support for more.
JSONFieldLink to this heading#Lookups implementation is different in JSONField,\nmainly due to the existence of key transformations. To demonstrate, we will use\nthe following example model:
from django.db import models\n\nclass Dog(models.Model):\n name = models.CharField(max_length=200)\n data = models.JSONField(null=True)\n\n def __str__(self):\n return self.name\nNoneLink to this heading#As with other fields, storing None as the field’s value will store it as\nSQL NULL. While not recommended, it is possible to store JSON scalar\nnull instead of SQL NULL by using Value('null').
Whichever of the values is stored, when retrieved from the database, the Python\nrepresentation of the JSON scalar null is the same as SQL NULL, i.e.\nNone. Therefore, it can be hard to distinguish between them.
This only applies to None as the top-level value of the field. If None\nis inside a list or dict, it will always be interpreted\nas JSON null.
When querying, None value will always be interpreted as JSON null. To\nquery for SQL NULL, use isnull:
>>> Dog.objects.create(name='Max', data=None) # SQL NULL.\n<Dog: Max>\n>>> Dog.objects.create(name='Archie', data=Value('null')) # JSON null.\n<Dog: Archie>\n>>> Dog.objects.filter(data=None)\n<QuerySet [<Dog: Archie>]>\n>>> Dog.objects.filter(data=Value('null'))\n<QuerySet [<Dog: Archie>]>\n>>> Dog.objects.filter(data__isnull=True)\n<QuerySet [<Dog: Max>]>\n>>> Dog.objects.filter(data__isnull=False)\n<QuerySet [<Dog: Archie>]>\nUnless you are sure you wish to work with SQL NULL values, consider setting\nnull=False and providing a suitable default for empty values, such as\ndefault=dict.
To query based on a given dictionary key, use that key as the lookup name:
\n>>> Dog.objects.create(name='Rufus', data={\n... 'breed': 'labrador',\n... 'owner': {\n... 'name': 'Bob',\n... 'other_pets': [{\n... 'name': 'Fishy',\n... }],\n... },\n... })\n<Dog: Rufus>\n>>> Dog.objects.create(name='Meg', data={'breed': 'collie', 'owner': None})\n<Dog: Meg>\n>>> Dog.objects.filter(data__breed='collie')\n<QuerySet [<Dog: Meg>]>\nMultiple keys can be chained together to form a path lookup:
\n>>> Dog.objects.filter(data__owner__name='Bob')\n<QuerySet [<Dog: Rufus>]>\nIf the key is an integer, it will be interpreted as an index transform in an\narray:
\n>>> Dog.objects.filter(data__owner__other_pets__0__name='Fishy')\n<QuerySet [<Dog: Rufus>]>\nIf the key you wish to query by clashes with the name of another lookup, use\nthe contains lookup instead.
To query for missing keys, use the isnull lookup:
>>> Dog.objects.create(name='Shep', data={'breed': 'collie'})\n<Dog: Shep>\n>>> Dog.objects.filter(data__owner__isnull=True)\n<QuerySet [<Dog: Shep>]>\ncontainsLink to this heading#The contains lookup is overridden on JSONField. The returned\nobjects are those where the given dict of key-value pairs are all\ncontained in the top-level of the field. For example:
>>> Dog.objects.create(name='Rufus', data={'breed': 'labrador', 'owner': 'Bob'})\n<Dog: Rufus>\n>>> Dog.objects.create(name='Meg', data={'breed': 'collie', 'owner': 'Bob'})\n<Dog: Meg>\n>>> Dog.objects.create(name='Fred', data={})\n<Dog: Fred>\n>>> Dog.objects.filter(data__contains={'owner': 'Bob'})\n<QuerySet [<Dog: Rufus>, <Dog: Meg>]>\n>>> Dog.objects.filter(data__contains={'breed': 'collie'})\n<QuerySet [<Dog: Meg>]>\ncontained_byLink to this heading#This is the inverse of the contains lookup - the\nobjects returned will be those where the key-value pairs on the object are a\nsubset of those in the value passed. For example:
>>> Dog.objects.create(name='Rufus', data={'breed': 'labrador', 'owner': 'Bob'})\n<Dog: Rufus>\n>>> Dog.objects.create(name='Meg', data={'breed': 'collie', 'owner': 'Bob'})\n<Dog: Meg>\n>>> Dog.objects.create(name='Fred', data={})\n<Dog: Fred>\n>>> Dog.objects.filter(data__contained_by={'breed': 'collie', 'owner': 'Bob'})\n<QuerySet [<Dog: Meg>, <Dog: Fred>]>\n>>> Dog.objects.filter(data__contained_by={'breed': 'collie'})\n<QuerySet [<Dog: Fred>]>\nhas_keyLink to this heading#Returns objects where the given key is in the top-level of the data. For\nexample:
\n>>> Dog.objects.create(name='Rufus', data={'breed': 'labrador'})\n<Dog: Rufus>\n>>> Dog.objects.create(name='Meg', data={'breed': 'collie', 'owner': 'Bob'})\n<Dog: Meg>\n>>> Dog.objects.filter(data__has_key='owner')\n<QuerySet [<Dog: Meg>]>\nhas_keysLink to this heading#Returns objects where all of the given keys are in the top-level of the data.\nFor example:
\n>>> Dog.objects.create(name='Rufus', data={'breed': 'labrador'})\n<Dog: Rufus>\n>>> Dog.objects.create(name='Meg', data={'breed': 'collie', 'owner': 'Bob'})\n<Dog: Meg>\n>>> Dog.objects.filter(data__has_keys=['breed', 'owner'])\n<QuerySet [<Dog: Meg>]>\nhas_any_keysLink to this heading#Returns objects where any of the given keys are in the top-level of the data.\nFor example:
\n>>> Dog.objects.create(name='Rufus', data={'breed': 'labrador'})\n<Dog: Rufus>\n>>> Dog.objects.create(name='Meg', data={'owner': 'Bob'})\n<Dog: Meg>\n>>> Dog.objects.filter(data__has_any_keys=['owner', 'breed'])\n<QuerySet [<Dog: Rufus>, <Dog: Meg>]>\nQ objectsLink to this heading#Keyword argument queries – in filter(),\netc. – are „AND”ed together. If you need to execute more complex queries (for\nexample, queries with OR statements), you can use Q objects.
A Q object (django.db.models.Q) is an object\nused to encapsulate a collection of keyword arguments. These keyword arguments\nare specified as in „Field lookups” above.
For example, this Q object encapsulates a single LIKE query:
from django.db.models import Q\nQ(question__startswith='What')\nQ objects can be combined using the &, |, and ^ operators. When\nan operator is used on two Q objects, it yields a new Q object.
For example, this statement yields a single Q object that represents the\n„OR” of two "question__startswith" queries:
Q(question__startswith='Who') | Q(question__startswith='What')\nThis is equivalent to the following SQL WHERE clause:
WHERE question LIKE 'Who%' OR question LIKE 'What%'\nYou can compose statements of arbitrary complexity by combining Q objects\nwith the &, |, and ^ operators and use parenthetical grouping.\nAlso, Q objects can be negated using the ~ operator, allowing for\ncombined lookups that combine both a normal query and a negated (NOT)\nquery:
Q(question__startswith='Who') | ~Q(pub_date__year=2005)\nEach lookup function that takes keyword-arguments\n(e.g. filter(),\nexclude(),\nget()) can also be passed one or more\nQ objects as positional (not-named) arguments. If you provide multiple\nQ object arguments to a lookup function, the arguments will be „AND”ed\ntogether. For example:
Poll.objects.get(\n Q(question__startswith='Who'),\n Q(pub_date=date(2005, 5, 2)) | Q(pub_date=date(2005, 5, 6))\n)\n… roughly translates into the SQL:
\nSELECT * from polls WHERE question LIKE 'Who%'\n AND (pub_date = '2005-05-02' OR pub_date = '2005-05-06')\nLookup functions can mix the use of Q objects and keyword arguments. All\narguments provided to a lookup function (be they keyword arguments or Q\nobjects) are „AND”ed together. However, if a Q object is provided, it must\nprecede the definition of any keyword arguments. For example:
Poll.objects.get(\n Q(pub_date=date(2005, 5, 2)) | Q(pub_date=date(2005, 5, 6)),\n question__startswith='Who',\n)\n… would be a valid query, equivalent to the previous example; but:
\n# INVALID QUERY\nPoll.objects.get(\n question__startswith='Who',\n Q(pub_date=date(2005, 5, 2)) | Q(pub_date=date(2005, 5, 6))\n)\n… would not be valid.
\n\n\nTo compare two model instances, use the standard Python comparison operator,\nthe double equals sign: ==. Behind the scenes, that compares the primary\nkey values of two models.
Using the Entry example above, the following two statements are equivalent:
>>> some_entry == other_entry\n>>> some_entry.id == other_entry.id\nIf a model’s primary key isn’t called id, no problem. Comparisons will\nalways use the primary key, whatever it’s called. For example, if a model’s\nprimary key field is called name, these two statements are equivalent:
>>> some_obj == other_obj\n>>> some_obj.name == other_obj.name\nThe delete method, conveniently, is named\ndelete(). This method immediately deletes the\nobject and returns the number of objects deleted and a dictionary with\nthe number of deletions per object type. Example:
>>> e.delete()\n(1, {'blog.Entry': 1})\nYou can also delete objects in bulk. Every\nQuerySet has a\ndelete() method, which deletes all\nmembers of that QuerySet.
For example, this deletes all Entry objects with a pub_date year of\n2005:
>>> Entry.objects.filter(pub_date__year=2005).delete()\n(5, {'webapp.Entry': 5})\nKeep in mind that this will, whenever possible, be executed purely in SQL, and\nso the delete() methods of individual object instances will not necessarily\nbe called during the process. If you’ve provided a custom delete() method\non a model class and want to ensure that it is called, you will need to\n„manually” delete instances of that model (e.g., by iterating over a\nQuerySet and calling delete() on each\nobject individually) rather than using the bulk\ndelete() method of a\nQuerySet.
When Django deletes an object, by default it emulates the behavior of the SQL\nconstraint ON DELETE CASCADE – in other words, any objects which had\nforeign keys pointing at the object to be deleted will be deleted along with\nit. For example:
b = Blog.objects.get(pk=1)\n# This will delete the Blog and all of its Entry objects.\nb.delete()\nThis cascade behavior is customizable via the\non_delete argument to the\nForeignKey.
Note that delete() is the only\nQuerySet method that is not exposed on a\nManager itself. This is a safety mechanism to\nprevent you from accidentally requesting Entry.objects.delete(), and\ndeleting all the entries. If you do want to delete all the objects, then\nyou have to explicitly request a complete query set:
Entry.objects.all().delete()\nAlthough there is no built-in method for copying model instances, it is\npossible to easily create new instance with all fields» values copied. In the\nsimplest case, you can set pk to None and\n_state.adding to True. Using our\nblog example:
blog = Blog(name='My blog', tagline='Blogging is easy')\nblog.save() # blog.pk == 1\n\nblog.pk = None\nblog._state.adding = True\nblog.save() # blog.pk == 2\nThings get more complicated if you use inheritance. Consider a subclass of\nBlog:
class ThemeBlog(Blog):\n theme = models.CharField(max_length=200)\n\ndjango_blog = ThemeBlog(name='Django', tagline='Django is easy', theme='python')\ndjango_blog.save() # django_blog.pk == 3\nDue to how inheritance works, you have to set both pk and id to\nNone, and _state.adding to True:
django_blog.pk = None\ndjango_blog.id = None\ndjango_blog._state.adding = True\ndjango_blog.save() # django_blog.pk == 4\nThis process doesn’t copy relations that aren’t part of the model’s database\ntable. For example, Entry has a ManyToManyField to Author. After\nduplicating an entry, you must set the many-to-many relations for the new\nentry:
entry = Entry.objects.all()[0] # some previous entry\nold_authors = entry.authors.all()\nentry.pk = None\nentry._state.adding = True\nentry.save()\nentry.authors.set(old_authors)\nFor a OneToOneField, you must duplicate the related object and assign it\nto the new object’s field to avoid violating the one-to-one unique constraint.\nFor example, assuming entry is already duplicated as above:
detail = EntryDetail.objects.all()[0]\ndetail.pk = None\ndetail._state.adding = True\ndetail.entry = entry\ndetail.save()\nSometimes you want to set a field to a particular value for all the objects in\na QuerySet. You can do this with the\nupdate() method. For example:
# Update all the headlines with pub_date in 2007.\nEntry.objects.filter(pub_date__year=2007).update(headline='Everything is the same')\nYou can only set non-relation fields and ForeignKey\nfields using this method. To update a non-relation field, provide the new value\nas a constant. To update ForeignKey fields, set the\nnew value to be the new model instance you want to point to. For example:
>>> b = Blog.objects.get(pk=1)\n\n# Change every Entry so that it belongs to this Blog.\n>>> Entry.objects.update(blog=b)\nThe update() method is applied instantly and returns the number of rows\nmatched by the query (which may not be equal to the number of rows updated if\nsome rows already have the new value). The only restriction on the\nQuerySet being updated is that it can only\naccess one database table: the model’s main table. You can filter based on\nrelated fields, but you can only update columns in the model’s main\ntable. Example:
>>> b = Blog.objects.get(pk=1)\n\n# Update all the headlines belonging to this Blog.\n>>> Entry.objects.filter(blog=b).update(headline='Everything is the same')\nBe aware that the update() method is converted directly to an SQL\nstatement. It is a bulk operation for direct updates. It doesn’t run any\nsave() methods on your models, or emit the\npre_save or post_save signals (which are a consequence of calling\nsave()), or honor the\nauto_now field option.\nIf you want to save every item in a QuerySet\nand make sure that the save() method is called on\neach instance, you don’t need any special function to handle that. Loop over\nthem and call save():
for item in my_queryset:\n item.save()\nCalls to update can also use F expressions to\nupdate one field based on the value of another field in the model. This is\nespecially useful for incrementing counters based upon their current value. For\nexample, to increment the pingback count for every entry in the blog:
>>> Entry.objects.update(number_of_pingbacks=F('number_of_pingbacks') + 1)\nHowever, unlike F() objects in filter and exclude clauses, you can’t\nintroduce joins when you use F() objects in an update – you can only\nreference fields local to the model being updated. If you attempt to introduce\na join with an F() object, a FieldError will be raised:
# This will raise a FieldError\n>>> Entry.objects.update(headline=F('blog__name'))\nIf you find yourself needing to write an SQL query that is too complex for\nDjango’s database-mapper to handle, you can fall back on writing SQL by hand.\nDjango has a couple of options for writing raw SQL queries; see\nPerforming raw SQL queries.
\nFinally, it’s important to note that the Django database layer is merely an\ninterface to your database. You can access your database via other tools,\nprogramming languages or database frameworks; there’s nothing Django-specific\nabout your database.
\n