Instead of doing the query like this
query {
allObjectC {
id # self ID()
nestedB_set {
id
nestedA_set {
id # ID() we want to get
}
}
}
}
New Graphql Query
query {
allObjectC {
id # self ID()
nestedA_id # nested nested ID() we want to get
}
}
Any best pratice or ideas for this ?
Thanks in advance .
We change the model like this // Models.py
class ObjectA(models.Model):
...
class ObjectB(models.Model):
propertyB = models.ForeignKey(ObjectA, on_delete=models.CASCADE, related_name='nestedB')
class ObjectC(models.Model):
propertyC = models.ForeignKey(ObjectB, on_delete=models.CASCADE, related_name='nestedC')
#property # new property created
def nestedB_nestedA_id(self):
return self.propertyC.nested_objectB.nested_objectA.id
We change the graphql schema like this // schema.py
class ObjectCNode(DjangoObjectType):
nestedB_nestedA_id = graphene.Int(source='nestedB_nestedA_id') # as proxy
nestedA_id = graphene.ID() # globalID of ObjectANode we want to get
def resolve_nestedB_nestedA_id(self, info, **kwargs):
return self.nestedB_nestedA_id
def resolve_nestedA_id(self, info, **kwargs):
return relay.Node.to_global_id(ObjectANode._meta.name, self.nestedB_nestedA_id)
And now query
query {
allObjectC {
id # self ID()
nestedA_id # nested nested ID() we want to get
}
}
Others suggestions ?
Related
Using Graphene in Django to create Gql schema, now trying to filter Foreign Keys with list of strings. It kinda works, but not exactly.
schema.py
class CharInFilter(BaseInFilter, CharFilter):
pass
class ProductFilter(FilterSet):
softwares__name = CharInFilter(field_name="softwares__name", lookup_expr="in")
class Meta:
model = Product
fields = {"name": ["exact", "icontains"]}
class ProductType(DjangoObjectType):
class Meta:
model = Product
filterset_class = ProductFilter
interfaces = (graphene.relay.Node,)
query
query authorPageProducts {
user(slug: "john") {
productSet(softwares_Name: "Blender") {
edges {
node {
name
softwares {
name
}
}
}
}
}
}
Here is what works and what not:
softwares_Name: "Blender" -> correct
softwares_Name: "Houdini" -> correct
softwares_Name: "Blender,Houdini" -> empty result, not correct
I am passing string separated with comma. Can/should I pass list of strings in Gql query? Im not sure if its possible/necessary.
I do have Products that have both Foreign Keys with values "Houdini" and "Blender", so query with "Blender,Houdini" shouldn't be empty.
I tried this query in shell, and its correct. Here I used list of strings.
u = User.objects.get(id=2)
p = u.product_set.filter(softwares__name__in=["Blender", "Houdini"])
Here is some info from Django Debug Toolbar, to see SQL expression for third case.
SELECT COUNT(*) AS "__count"
FROM "shop_product"
INNER JOIN "shop_product_softwares"
ON ("shop_product"."id" = "shop_product_softwares"."product_id")
INNER JOIN "shop_software"
ON ("shop_product_softwares"."software_id" = "shop_software"."id")
WHERE ("shop_product"."author_id" = 2 AND "shop_software"."name" IN ('Blender,Houdini'))
I cannot figure out where is the issue, while querying with comma separated string, as docs suggest.
Thank you
I'm new with GraphQL schemas and I would like to do a mutation using the gql client. The query below works like a charme in the graphql web interface after replacing the 5 variables with the corresponding strings and integers.
But when I put a $ before every variables in the query, as mentionned in the documentation, it throws an error saying Variable '$w' is not defined by operation 'createMutation'.
What am'I missing ?
transport = AIOHTTPTransport(url="http://x.x.x.x:8000/graphql")
client = Client(transport=transport, fetch_schema_from_transport=True)
query = gql(
"""
mutation createMutation {
createTarget(targetData: {weight: $w, dt: $dt,
exchangeId: $exchangeId,
strategyId: $strategyId,
marketId:$marketId
}) {
target {
dt,
weight,
market,
exchange,
strategy
}
}
}
"""
)
params = {"w": self.weight,
"dt": self.dt,
"exchangeId": self.exchange.pk,
"strategyId": self.strategy.pk,
"marketId": self.market.pk
}
result = client.execute(query, variable_values=params)
When I remove the $ it says Float cannot represent non numeric value: w.
And this is how the graphene code looks like at the server side :
class TargetInput(graphene.InputObjectType):
weight = graphene.Float()
dt = graphene.DateTime()
strategy_id = graphene.Int()
exchange_id = graphene.Int()
market_id = graphene.Int()
class CreateTarget(graphene.Mutation):
class Arguments:
target_data = TargetInput(required=True)
target = graphene.Field(CustomObject)
#staticmethod
def mutate(root, info, target_data):
target = Target.objects.create(**target_data)
return CreateTarget(target=target)
class Mutation(graphene.ObjectType):
create_target = CreateTarget.Field()
schema = graphene.Schema(query=Query, mutation=Mutation)
There is also another question related to gql variables but it doesn't solve my problem.
I have found the answer to my own question. When using variables it's necessary to declare each of them between ( and ) at the beggining of the query, as stipulated here.
So in my case the correct query was:
query = gql(
"""
mutation createMutation ($w: Float, $dt: DateTime, $exchangeId: Int, $strategyId: Int, $marketId: Int){
createTarget(targetData: {weight: $w, dt: $dt,
exchangeId: $exchangeId,
strategyId: $strategyId,
marketId: $marketId
}) {
target {
dt
}
}
}
"""
)
In Graphene-Django and GraphQL I am trying to create a resolve_or_create method for nested data creation inside my mutations.
I'm trying to pass a dictionary with the input from the user as a **kwarg to my resolve_or_create function, and though I can see "location" in the variable watcher (in VSCode), I continuously am getting the error 'dict' object has no attribute 'location'
Here's my resolve_or_create method:
def resolve_or_create(*args, **kwargs):
input = {}
result = {}
input.location = kwargs.get('location', None)
if input.location is not None:
input.location = Location.objects.filter(pk=input.location.id).first()
if input.location is None:
location = Location.objects.create(
location_city = input.location.location_city,
location_state = input.location.location_state,
location_sales_tax_rate = input.location.location_sales_tax_rate
)
if location is None:
return None
result.location = location
return result
and my CreateCustomer definition, where this method is being called
class CreateCustomer(graphene.Mutation):
class Arguments:
input = CustomerInput(required=True)
ok = graphene.Boolean()
customer = graphene.Field(CustomerType)
#staticmethod
def mutate(root, info, input=None):
ok = True
resolved = resolve_or_create(**{'location':input.customer_city})
customer_instance = Customer(
customer_name = input.customer_name,
customer_address = input.customer_address,
customer_city = resolved.location,
customer_state = input.customer_state,
customer_zip = input.customer_zip,
customer_email = input.customer_email,
customer_cell_phone = input.customer_cell_phone,
customer_home_phone = input.customer_home_phone,
referred_from = input.referred_from
)
customer_instance.save()
return CreateCustomer(ok=ok, customer=customer_instance)
Here is an example mutation that would create a new customer with an existing location
mutation createCustomer {
createCustomer(input: {
customerName: "Ricky Bobby",
customerAddress: "1050 Airport Drive",
customerCity: {id:1},
customerState: "TX",
customerZip: "75222",
customerEmail: "mem",
customerCellPhone: "124567894",
referredFrom: "g"
}) {
ok,
customer{
id,
customerName,
customerAddress,
customerCity {
id
},
customerState,
customerZip,
customerEmail,
customerCellPhone,
referredFrom
}
}
}
and here is an example mutation that would create a customer with a new location
mutation createCustomer {
createCustomer(input: {
customerName: "Ricky Bobby",
customerAddress: "1050 Airport Drive",
customerCity: {locationCity: "Dallas", locationState: "TX", locationSalesTaxRate:7.77},
customerState: "TX",
customerZip: "75222",
customerEmail: "mem",
customerCellPhone: "124567894",
referredFrom: "g"
}) {
ok,
customer{
id,
customerName,
customerAddress,
customerCity {
id
},
customerState,
customerZip,
customerEmail,
customerCellPhone,
referredFrom
}
}
}
So my question is two-fold.
First, how can I retrieve my passed in location dict from kwargs?
Second, is there a better way to be resolving and creating nested data than this? Would this be expected behavior in a best-practice GraphQL API?
I have also tried resolve_or_create(location=input.customer_city})
I realized my syntax for dictionary assignment and access was wrong.
Corrected function:
def resolve_or_create(*args, **kwargs):
input = {}
result = {}
candidate = None
input['location'] = kwargs['location']
input['customer'] = kwargs.get('customer', None)
input['technician'] = kwargs.get('tech', None)
if input['location'] is not None:
if 'id' in input['location']:
candidate = Location.objects.filter(pk=input['location']['id']).first()
result['location'] = candidate
if candidate is None:
result['location'] = Location.objects.create(
location_city = input['location']['location_city'],
location_state = input['location']['location_state'],
location_sales_tax_rate = input['location']['location_sales_tax_rate']
)
if result['location'] is None:
return None
return result
I would still like to discuss the most effective way to accomplish created and mutating nested data in graphene-django. Is there a DRYer way to achieve what I'm looking to do or something I'm missing?
I have successfully created an all graphene query that responds to
query {
person (id: "Mary") {
id
name
}
}
I now want to extend this to be able to loop through all people and return similar data for each.
query {
people {
count
allPersons {
name
}
}
}
How do I get the resolve_allPersons resolver in people to call the person resolver for each person?
Second query you've described can be done with custom type, for example:
class AllPeopleType(graphene.ObjectType):
count = graphene.Int()
all_persons = graphene.List(YourPersonType)
def resolve_count(self, info, **kwargs):
# assumed that django used on backend
return Person.objects.count()
def resolve_all_persons(self, info, **kwargs):
return Person.objects.all()
and query:
class YourQuery(object):
# person = ...
people = graphene.Field(AllPeopleType)
def resolve_people(self, info):
return AllPeopleType()
I'm using python 3.5 with
Django==1.11.6
graphene==2.0.dev20170802065539
graphene-django==2.0.dev2017083101
graphql-core==2.0.dev20171009101843
graphql-relay==0.4.5
I have a schema that fetch's single objects like this:
class Query(graphene.AbstractType):
story = graphene.Field(storyType, category=graphene.String(), id=graphene.Int())
def resolve_story(self, info, **kwargs):
category = kwargs.get('category')
id = kwargs.get('id')
if category is not None:
return models.story.objects.get(category=models.category.objects.get(name=category))
if id is not None:
return models.story.objects.get(pk=id)
return None
My problem is that I can't use both story(category:"category") and (id:"id") in one query. I read here that I should use aliases but I don't know how.
Thank's for any help.
Is as simple as the example just:
{
storyId: story(id: 1) {
name
}
storyCategory: story(category: "category_name") {
name
}
}
but you can use filter if that is what you mean doing something like:
if category and id is not None:
return models.story.objects.filter(id=id, category=models.category.objects.get(name=category))
and in the query something like:
story(id: 1, name: "category_name") {
name
}