pub trait GraphQLType<S = DefaultScalarValue>: GraphQLValue<S>where
S: ScalarValue,{
// Required methods
fn name(info: &Self::TypeInfo) -> Option<&str>;
fn meta<'r>(
info: &Self::TypeInfo,
registry: &mut Registry<'r, S>,
) -> MetaType<'r, S>
where S: 'r;
}Expand description
Primary trait used to expose Rust types in a GraphQL schema.
All of the convenience macros ultimately expand into an implementation of this trait for the given type. This can all be done manually.
§Example
Manually deriving an object is straightforward, but tedious. This is the equivalent of the
User object as shown in the example in the documentation root:
use juniper::{
meta::MetaType, Arguments, Context, DefaultScalarValue, Executor, ExecutionResult,
FieldResult, GraphQLType, GraphQLValue, Registry,
};
#[derive(Debug)]
struct Database { users: HashMap<String, User> }
impl Context for Database {}
#[derive(Debug)]
struct User { id: String, name: String, friend_ids: Vec<String> }
impl GraphQLType<DefaultScalarValue> for User {
fn name(_: &()) -> Option<&'static str> {
Some("User")
}
fn meta<'r>(_: &(), registry: &mut Registry<'r>) -> MetaType<'r>
where DefaultScalarValue: 'r,
{
// First, we need to define all fields and their types on this type.
//
// If we need arguments, want to implement interfaces, or want to add documentation
// strings, we can do it here.
let fields = &[
registry.field::<&String>("id", &()),
registry.field::<&String>("name", &()),
registry.field::<Vec<&User>>("friends", &()),
];
registry.build_object_type::<User>(&(), fields).into_meta()
}
}
impl GraphQLValue<DefaultScalarValue> for User {
type Context = Database;
type TypeInfo = ();
fn type_name(&self, _: &()) -> Option<&'static str> {
<User as GraphQLType>::name(&())
}
fn resolve_field(
&self,
info: &(),
field_name: &str,
args: &Arguments,
executor: &Executor<Database>
) -> ExecutionResult
{
// Next, we need to match the queried field name. All arms of this match statement
// return `ExecutionResult`, which makes it hard to statically verify that the type you
// pass on to `executor.resolve*` actually matches the one that you defined in `meta()`
// above.
let database = executor.context();
match field_name {
// Because scalars are defined with another `Context` associated type, you must use
// `resolve_with_ctx` here to make the `executor` perform automatic type conversion
// of its argument.
"id" => executor.resolve_with_ctx(info, &self.id),
"name" => executor.resolve_with_ctx(info, &self.name),
// You pass a vector of `User` objects to `executor.resolve`, and it will determine
// which fields of the sub-objects to actually resolve based on the query.
// The `executor` instance keeps track of its current position in the query.
"friends" => executor.resolve(info,
&self.friend_ids.iter()
.filter_map(|id| database.users.get(id))
.collect::<Vec<_>>()
),
// We can only reach this panic in two cases: either a mismatch between the defined
// schema in `meta()` above, or a validation failed because of a this library bug.
//
// In either of those two cases, the only reasonable way out is to panic the thread.
_ => panic!("Field {field_name} not found on type User"),
}
}
}Required Methods§
Dyn Compatibility§
This trait is not dyn compatible.
In older versions of Rust, dyn compatibility was called "object safety", so this trait is not object safe.