Easy GraphQL Server
easy_graphql_server provides an easy way to expose a database in GraphQL, using ORM models and web frameworks (so far only Django is supported, but SQLAlchemy & Flask will soon come).
easy_graphql_server can be installed from PyPI using the built-in pip command:
pip install easy-graphql-server
Usage
Expose methods
There are three ways to expose a method in the GraphQL schema.
Calling Schema.expose_query()
and Schema.expose_mutation()
import easy_graphql_server as egs
schema = egs.Schema()
schema.expose_query(
name = 'foo',
input_format = {
'input_string': egs.Required(str),
'input_integer': int,
},
output_format = {
'output_string': str,
'output_integer': int,
},
method = lambda input_string, input_integer=None: {
'output_string': 2 * input_string,
'output_integer': None if input_integer is None else 2 * input_integer,
},
)
_internal_value = 0
def bar_mutation_method(value=None, increment_value=None):
if value is not None:
_internal_value = value
if increment_value is not None:
_internal_value += increment_value
return {
'value': _internal_value,
}
schema.expose_mutation(
name = 'bar',
input_format = {
'input_string': egs.Required(str),
'input_integer': int,
},
output_format = {
'output_string': str,
'output_integer': int,
},
method = bar_mutation_method,
)
Subclassing Schema.ExposedQuery
and Schema.ExposedMutation
import easy_graphql_server as egs
schema = egs.Schema()
class FooQuery(schema.ExposedQuery):
name = 'foo'
input_format = {
'input_string': egs.Required(str),
'input_integer': int,
}
output_format = {
'output_string': str,
'output_integer': int,
}
@staticmethod
def method(input_string, input_integer=None):
return {
'output_string': 2 * input_string,
'output_integer': None if input_integer is None else 2 * input_integer,
}
class BarMutation(schema.ExposedMutation):
name = 'bar'
_internal_value = 0
input_format = {
'value': int,
'increment_value': int,
}
output_format = {
'value': int,
}
@classmethod
def method(cls, value=None, increment_value=None):
if value is not None:
cls._internal_value = value
if increment_value is not None:
cls._internal_value += increment_value
return {
'value': cls._internal_value,
}
Subclassing easy_graphql_server.ExposedQuery
and easy_graphql_server.ExposedMutation
This is very similar to the previous way.
import easy_graphql_server as egs
class FooQuery(schema.ExposedQuery):
name = 'foo'
input_format = {
'input_string': egs.Required(str),
'input_integer': int,
}
output_format = {
'output_string': str,
'output_integer': int,
}
@staticmethod
def method(input_string, input_integer=None):
return {
'output_string': 2 * input_string,
'output_integer': None if input_integer is None else 2 * input_integer,
}
class BarMutation(schema.ExposedMutation):
name = 'bar'
_internal_value = 0
input_format = {
'value': int,
'increment_value': int,
}
output_format = {
'value': int,
}
@classmethod
def method(cls, value=None, increment_value=None):
if value is not None:
cls._internal_value = value
if increment_value is not None:
cls._internal_value += increment_value
return {
'value': cls._internal_value,
}
schema = egs.Schema()
schema.expose(FooQuery)
schema.expose(BarMutation)
Available options for exposing methods
The same options can be passed either as class attributes for subclasses of Schema.ExposedQuery
and Schema.ExposedMutation
, or as keyword arguments to Schema.expose_query()
and Schema.expose_mutation()
methods.
Options for queries and mutations are the same.
-
name
is the name under which the method shall be exposed
-
method
is the callback function of your choice
-
input_format
is the input format for the GraphQL method, passed as a (possibly recursive) mapping; if unspecified or None
, the defined GraphQL method will take no input; the mapping keys are
-
output_format
is the output format of the GraphQL method, passed as a (possibly recursive) mapping or as a list
containing one mapping
-
pass_graphql_selection
can either be a bool
or a str
; if set to True
, the graphql_selection
parameter will be passed to the callback method, indicating which fields are requested for output; if set to a str
, the given string will be the name of the keyword parameter passed to the callback method instead of graphql_selection
-
pass_graphql_path
can either be a bool
or a str
; if set to True
, the graphql_path
parameter will be passed to the callback method, indicating as a list[str]
the GraphQL path in which the method is being executed; if set to a str
, the given string will be the name of the keyword parameter passed to the callback method instead of graphql_path
-
pass_authenticated_user
can either be a bool
or a str
; if set to True
, the authenticated_user
parameter will be passed to the callback method, indicating the user authenticated in the source HTTP request (or None
if the request was unauthenticated); if set to a str
, the given string will be the name of the keyword parameter passed to the callback method instead of authenticated_user
-
require_authenticated_user
is a bool
indicating whether or not authentication is required for the exposed method
Expose ORM models
What does it do?
For instance, exposing a model called Thing
will expose the following queries...
thing
: fetch a single instance of the model given its unique identifierthings
: fetch a collection of model instances, given filtering criteria, or unfiltered, paginated or not
...and mutations:
create_thing
: create a single instance of the model given the data to be insertedupdate_thing
: update a single instance of the model given its unique identifier and a mapping the new data to applydelete_thing
: delete a single instance of the model given its unique identifier
Calling Schema.expose_model()
import easy_graphql_server
from my_django_application.models import Person
schema = easy_graphql_server.Schema()
schema.expose_model(
orm_model=Person,
plural_name='people',
can_expose=('id', 'first_name', 'last_name'),
cannot_write=('id',),
)
Subclassing Schema.ExposedModel
import easy_graphql_server
from my_django_application.models import Person
schema = easy_graphql_server.Schema()
class ExposedPerson(schema.ExposedModel):
orm_model = Person
plural_name = 'people'
can_expose = ('id', 'first_name', 'last_name')
cannot_write = ('id',)
Subclassing easy_graphql_server.ExposedModel
This is very similar to the previous way.
import easy_graphql_server
from my_django_application.models import Person
schema = easy_graphql_server.Schema()
class ExposedPerson(easy_graphql_server.ExposedModel):
orm_model = Person
plural_name = 'people'
can_expose = ('id', 'first_name', 'last_name')
cannot_write = ('id',)
schema = easy_graphql_server.Schema()
schema.expose(ExposedPerson)
Available options for exposing models
The same options can be passed either as class attributes for subclasses of ExposedModel
, or as keyword arguments to Schema.expose_model()
method.
-
cannot_expose
is either a bool
, or a tuple[str]
; defaults to False
; if set to True
, no query or mutation method will be exposed for the model; if set to a list of field names, those fields will be excluded from every query and mutation
-
can_create
is either a bool
, or a tuple[str]
; defaults to True
; if set to False
, no creation mutation method will be exposed for the model; if set to a list of field names, only those fields will be possible field values passed at insertion time to create_...
-
cannot_create
is either a bool
, or a tuple[str]
; defaults to False
; if set to True
, no creation mutation method will be exposed for the model; if set to a list of field names, those fields will be excluded from possible field values passed at insertion time to create_...
-
can_read
is either a bool
, or a tuple[str]
; defaults to True
; if set to False
, no query method will be exposed for the model; if set to a list of field names, only those fields will be exposed for the ...
(show one instance) and ...s
(show a collection of instances) queries (it also defines which fields are available as mutations results)
-
cannot_read
is either a bool
, or a tuple[str]
; defaults to False
; if set to True
, no query method will be exposed for the model; if set to a list of field names, only those fields will be exposed for the ...
(show one instance) and ...s
(show a collection of instances) queries (it also defines which fields are not available as mutations results)
-
can_update
is either a bool
, or a tuple[str]
; defaults to True
; if set to True
, no delete_...
mutation method will be exposed for the model; if set to a list of field names, those fields will be the only possible keys for the _
parameter (new data for instance fields) of the update_...
mutation
-
cannot_update
is either a bool
, or a tuple[str]
; defaults to False
; if set to True
, no update_...
mutation method will be exposed for the model; if set to a list of field names, those fields will be excluded from possible keys for the _
parameter (new data for instance fields) of the update_...
mutation
-
can_delete
is a bool
; defaults to True
; if set to False
, no delete_...
mutation method will be exposed for the model
-
cannot_delete
is a bool
; defaults to False
; if set to True
, no delete_...
mutation method will be exposed for the model
-
only_when_child_of
is either None
(model does not have to be nested to be exposed), True
(the model is not exposed if not nested), an ORM model class, or a tuple/list/set of ORM model classes (the defined model can only be exposed when nested directly under one of models passed as only_when_child_of
parameter)
-
require_authenticated_user
is a bool
indicating whether or not authentication is required for the exposed method
-
has_permission
is either None
, or a callback method returning a bool
(True
if operation is authorized, False
otherwise), and taking as parameters authenticated_user
(self-explanatory), operation
(a value of the easy_graphql_server.Operation
enum: CREATE
, READ
, UPDATE
or DELETE
) and data
(new data, only applies to CREATE
and UPDATE
)
-
filter_for_user
is either None
, or a callback method returning a queryset
, and taking as parameters queryset
and authenticated_user
Perform GraphQL queries
If you want to perform GraphQL queries on the schema without going through a schema, you can use Schema.execute()
. This method can take the following parameters:
query
: the GraphQL query, in the form of a str
variables
: variables to go along with the query (optional), as a dict[str,Any]
operation_name
: name of the operation to be executed within the query (optional)authenticated_user
: parameter that will be passed to the callback functions of GraphQL methods that require it (optional)serializable_output
: the output will be rendered as JSON-serializable dict
, instead of a graphql.execution.execute.ExecutionResult
instance
Credits and history
The easy_graphql_server library was originally a subproject within the Bridger development
team, to provide an easy way to expose database models with GraphQL using
Graphene.
The project was then rewritten with graphq-core
and Graphene was dropped.
License
easy_graphql_server is under MIT license.