可扩展项目的 GraphQL 代码组织

Question

认为我们有一个大项目，其中包含许多应用程序，这会导致大量查询和突变。对于这样的项目，你们如何处理 graphql 代码架构。让我们举个例子。

我有一个名为帐户的应用程序。它将具有与用户和配置文件相关的查询和更改。我使用的文件夹结构是每个应用程序都有 graphql 文件夹，然后会有 schema.py 和 mutations.py。代码现在安排成这样

schema.py

class User(DjangoObjectType):
class Meta:
model = CustomUser
filter_fields = {
'email': ['exact', ],
}
exclude_fields = ('password', 'is_superuser', )
interfaces = (relay.Node, )

class ProfileNode(DjangoObjectType):
class Meta:
model = Profile
interfaces = (relay.Node, )

class UserQuery(object):
user = relay.Node.Field(User)
users = DjangoFilterConnectionField(User)  # resolve_users is not needed now

class ProfileQuery(object):
profile = relay.Node.Field(ProfileNode)
profiles = DjangoFilterConnectionField(ProfileNode)

class UserProfile(ObjectType):
profile = Field(ProfileNode)
def resolve_profile(self, info, **kwargs):
if id is not None and info.context.user.is_authenticated:
profile = Profile.objects.get(user=info.context.user)
return profile
return None

class Viewer(ObjectType):
user = Field(User)
def resolve_user(self, info, **kwargs):
if info.context.user.is_authenticated:
return info.context.user
return None 

mutations.py

class Register(graphene.Mutation):
"""
Mutation to register a user
"""
class Arguments:
first_name = graphene.String(required=True)
last_name = graphene.String(required=True)
email = graphene.String(required=True)
password = graphene.String(required=True)
password_repeat = graphene.String(required=True)
success = graphene.Boolean()
errors = graphene.List(graphene.String)
def mutate(self, info, first_name, last_name, email, password, password_repeat):
# console.log('info', info, first_name, last_name, email, password)
if password == password_repeat:
try:
user = CustomUser.objects.create(
first_name=first_name,
last_name=last_name,
email=email,
is_active=False
)
print ('user', user)
user.set_password(password)
user.save()
if djoser_settings.get('SEND_ACTIVATION_EMAIL'):
send_activation_email(user, info.context)
return Register(success=bool(user.id))
# TODO: specify exception
except Exception:
errors = ["email", "Email already registered."]
return Register(success=False, errors=errors)
errors = ["password", "Passwords don't match."]
return Register(success=False, errors=errors)

根架构

// just to show the number of mutations just for account apps.
from accounts.graphql.mutations import (
Activate,
DeleteAccount,
Login,
RefreshToken,
Register,
ResetPassword,
ResetPasswordConfirm,
)
from accounts.graphql.schema import Viewer, UserProfile

class Mutation(company_mutation.Mutation, graphene.ObjectType):
activate = Activate.Field()
debug = graphene.Field(DjangoDebug, name='__debug')

class Query(company_schema.Query, graphene.ObjectType):
viewer = graphene.Field(Viewer)
user_profile = graphene.Field(UserProfile)
debug = graphene.Field(DjangoDebug, name='__debug')
@staticmethod
def resolve_viewer(self, info, **kwargs):
if info.context.user.is_authenticated:
return info.context.user
return None
@staticmethod
def resolve_user_profile(self, info, **kwargs):
if info.context.user.is_authenticated and id:
return info.context.user
return None

schema = graphene.Schema(query=Query, mutation=Mutation)

你可以看到只有帐户应用程序的突变。有很多突变，在考虑所有应用程序时会有更多突变。你们这些人是怎么处理的？

Answer 1

我认为仅基于schema.py，queries.py，mutations.py等文件的项目组织对于大型项目非常不利。

这就像组织您的模型零件，使用models.py、fields.py、utils.py等文件......

特别是，在queries.py文件中某处定义的 GraphQL ObjectType 可以返回或用作突变的输入。

所以我更喜欢更基于对象及其逻辑关系的结构

schema/
__init__.py : contains your root schema actual code
viewer.py : contains the ObjectType Viewer
user/
user.py : contains the ObjectType User and UserQuery
profile.py : contains the ObjectType Profil
account : contains the account management mutations
login.py : contains the login / logout mutations
...

注意：您可以像查询一样嵌入突变，以便能够获取突变查询，例如：

mutation {
account {
delete(id: 12) {
status
}
}
}