Libraries supporting GraphQL in Dart

I just wanted to start a discussion about the possibility of building data as built_values rather than storing the Json object and using getters.

I find the immutability and serializability of built_value to be compelling, although it would add some complexity.

Thoughts?

Extends the set of code builders available.

ast_builder

Generates an AST representation of a GraphQL Document

data_builder

Generates a typed view on top of data JSON. The generated classes are not supposed to be used directly as that would introduce tight coupling between your GraphQL documents and your application code.

class $ListPokemon {
  const $ListPokemon(this.data);

  final Map<String, dynamic> data;

  List<$ListPokemon$pokemons> get pokemons => (data["pokemons"] as List)
      .map((dynamic e) => $ListPokemon$pokemons(e as Map<String, dynamic>))
      .toList();
}

class $ListPokemon$pokemons {
  const $ListPokemon$pokemons(this.data);

  final Map<String, dynamic> data;

  String get id => data["id"] as String;
  String get name => data["name"] as String;
}

op_builder

Wraps AST and operation name into an Operation.

import 'package:gql_exec/gql_exec.dart';
import 'find_pokemon.ast.gql.dart' as find_pokemon;

const FindPokemon = Operation(
  document: find_pokemon.document,
  operationName: 'FindPokemon',
);

req_builder

Extend Request class to use specific Operation and provide ability to build variables. Builder pattern let's you handle nullable variables correctly (GraphQL spec pretty much requires both undefined and null values).

import 'package:gql_exec/gql_exec.dart' as _i1;
import 'find_pokemon.op.gql.dart' as _i2;

class FindPokemon extends _i1.Request {
  FindPokemon()
      : super(operation: _i2.FindPokemon, variables: <String, dynamic>{});

  set name(String value) => variables["name"] = value;
}

enum_builder

Generates an enum-like class per GraphQL enum type. Defines known enum values to be used in your code, and allows unknown enum values to be used without causing runtime error when handling response data.

class ReleaseType {
  const ReleaseType(this.value);

  final String value;

  static const ReleaseType ALPHA = ReleaseType('ALPHA');

  static const ReleaseType BETA = ReleaseType('BETA');

  static const ReleaseType GAMMA = ReleaseType('GAMMA');

  @override
  int get hashCode => value.hashCode;
  @override
  bool operator ==(Object o) => o is ReleaseType && o.value == value;
}

scalar_builder

Generates a container for a scalar value to be used when viewing the response data and building request variables.

class ID {
  const ID(this.value);

  final String value;

  @override
  int get hashCode => value.hashCode;
  @override
  bool operator ==(Object o) => o is ID && o.value == value;
}

input_builder

Generates an input builder to be used to build request variables.

class MutationInput {
  final Map<String, dynamic> input = <String, dynamic>{};

  set a(String value) => input['a'] = value;
}

schema_builder

Combines enum_builder, input_builder and scalar_builder.

Usage example

final result = await link.request(
  FindPokemon()..name = "Charizard",
);
final data = $FindPokemon(result.data);

final pokemon = data.pokemon;

print("Found ${pokemon.name}");
print("ID: ${pokemon.id}");

final weight = pokemon.weight;
final height = pokemon.height;

print(
  "Weight: ${weight.minimum} – ${weight.maximum}",
);
print(
  "Height: ${height.minimum} – ${height.maximum}",
);

Hi @smkhalsa I'm pingin you cause you was the last who contributed to gql_code_builder I'm helping @comigor to maintain artemis

We would like to update our deps to latest version but we cannot due to analyzer: 1.2.0 dep in gql_code_builder

I made this PR to update ananlyzer to latest version.

I also added some fixes for failing actions - mostly dependency reordering.

I would appreciate if you merge and release my PR.

Thanks

Hello, recently when running dart run build_runner build in order to get graphql query and mutation methods generated by Artemis, I am receiving this error on my build server (Codemagic) but not on my local machine. I have verified the package versions used are the same. Is it possible this package content changed without a bump in version number?

[INFO] Generating build script...
[INFO] Generating build script completed, took 803ms

[INFO] Precompiling build script......
[WARNING] ../../../.pub-cache/hosted/pub.dartlang.org/gql_code_builder-0.3.0-alpha+1635885531709/lib/src/ast.dart:618:12: Error: A non-null value must be returned since the return type 'Expression' doesn't allow null.
 - 'Expression' is from 'package:code_builder/src/specs/expression.dart' ('../../../.pub-cache/hosted/pub.dartlang.org/code_builder-4.1.0/lib/src/specs/expression.dart').
Expression _directiveLocation(DirectiveLocation location) {
           ^
[INFO] Precompiling build script... completed, took 11.6s

[SEVERE] Failed to precompile build script .dart_tool/build/entrypoint/build.dart.
This is likely caused by a misconfigured builder definition.

Hey @klavs! I'm getting an error while using the language SDL parser (parseString) on some of my examples. It say it can't find the Entity interface while parsing the Area object even though it's declared on the same schema:

Error on line 32, column 28 of lib/graphbrainz.schema.graphql: Unknown definition type 'Entity'
   ╷
32 │ type Area implements Node, Entity {
   │                            ^^^^^^
   ╵
package:gql/src/language/parser.dart 175:5   _Parser._parseDefinition
package:gql/src/language/parser.dart 117:22  _Parser._parseMany
package:gql/src/language/parser.dart 142:22  _Parser._parseDocument
package:gql/src/language/parser.dart 41:27   _Parser.parse
package:gql/src/language/parser.dart 17:17   parse
package:gql/src/language/parser.dart 27:5    parseString

The full schema, for referece: graphbrainz.schema.graphql.

Supersedes #193 as the nullsafety tracking PR. Will remain a draft and reference point for prereleases until dart has a stable null safe release. @klavs am planning on publishing *-nullsafety.0 versions for migrated packages soon, so that downstream packages can begin migrating.

Null safety migrations run thus far:

Libraries

• [X] gql: 0.13.0-nullsafety.1
• [X] gql_exec: 0.3.0-nullsafety.1
• [X] gql_link: 0.4.0-nullsafety.1
• [x] gql_http_link: 0.4.0-nullsafety.1
• [x] gql_dedupe_link: 2.0.0-nullsafety.1
• [x] gql_error_link: 0.2.0-nullsafety.1
• [x] gql_transform_link: 0.2.0-nullsafety.1
• [x] gql_websocket_link: 0.3.0-nullsafety.1
• [x] gql_dio_link: 0.1.0-nullsafety.1

~~blocked by https://github.com/dart-lang/build/issues/2920~~

• [x] gql_code_builder
• [x] gql_build

Examples

• [X] cats
• [x] gql_example_cli
• [x] gql_example_cli_github
• [x] gql_example_build
• [x] gql_example_http_auth_link
• [x] gql_example_dio_link
• [x] gql_example_flutter

stable release TODOs

• [x] ~~Hand-verify gql ASTs against the current spec~~

This WIP-PR that aims to solve the issue discussed in #154.

The old API will remain almost the same. HttpLink default constructor will create a new HttpGqlClient. The HttpLinkParserException & HttpLinkServerException will now take the abstracted GqlClientResponse type instead of the http lib type.

All tests are passing, but the test that depend on the http.Response type in the exceptions.

Suggestions for how to improve the new API are more than welcome.

Update the analyzer package from ^0.39.14 to ^0.40.4.

Advantages:

• Quite some open issues are fixed
• gql_code_builder would be now compatible with the latest version of build_runner (^1.10.3)

EDIT: This PR is blocked, waiting for the new release of build_value_generator. More info here

• [x] Example of Auth link functionality
• [x] Error Link
• [x] Websocket Link
• [x] File upload
• [x] http_link should have something like the useGETForQueries option as well
• [ ] persisted queries were recently contributed

Hi there! I left my app open overnight, and ended up with over four thousand (4000!) errors originating from this package with the error message type '(dynamic) => Map<String, dynamic>?' is not a subtype of type '(dynamic) => FutureOr<Map<String, dynamic>>?' of 'function result'.

I've seen this error before but haven't thought much of it until I began getting messages from sentry about over use. Here's a small stack trace that sentry gives me:

link.dart in WebSocketLink._parseSocketMessage within gql_websocket_link
link.dart in WebSocketLink._connect.<fn> at line 206:37 within gql_websocket_link
link.dart in WebSocketLink._connect.<fn> at line 201:31 within gql_websocket_link

and here are my versions:

  gql: 0.13.0
  gql_link: ^0.4.0
  gql_http_link: ^0.4.0
  ferry_flutter: ^0.5.4
  gql_exec: ^0.3.0
  gql_websocket_link: ^0.3.1

is this seen anywhere? how can I fix this error? Thanks!

This extracts the logic for multipart file upload and useGETForQueries from graphql/client.dart, with accompanying tests

Major changes:

• useGetForQueries argument to HttpLink
• using http.Client.send instead of post
• HttpLink stores uri as a Uri
• I reworked HttpLink._serializeRequest into an exception handling HOF to handle the different paths, but really I think RequestSerializer should have a generic Serialized type which is http.BaseRequest for HttpLink. Otherwise it can't really handle the serialization concern IMO

This addresses some of #57

When looking at the docs for HttpLinkServerException it has the following text:

/// Exception occurring when network fails
/// or parsed response is missing both `data` and `errors`. <--- parsedResponse should have missing data and errors

Is this true? I get exceptions thrown with HttpLinkServerException that contains the parsedResponse.

In this case it has the following information:

// the below code is thrown when we catch for `HttpLinkServerException`
ServerException(
  originalException: null, 
  parsedResponse: Response(
    data: null, errors: [GraphQLError(message: unauthenticated, locations: null, path: null, extensions: {code: AUTHENTICATION_ERROR, statusCode: 401})], context: Context({ResponseExtensions: Instance of 'ResponseExtensions'}), response {errors: [{message: unauthenticated, extensions: {code: AUTHENTICATION_ERROR, statusCode: 401}}]}
  )
)

I don't have a good way currently to reproduce this more than seeing it logged to Sentry. I could also misunderstand this, so would love some input!

Currently, there's no easy way to check whether a Request is a query or a mutation.

This leads to code duplication, as seen below:

https://github.com/gql-dart/gql/blob/c13e2fe2e6ca234ff5bc17263e3536ed2c907474/links/gql_dio_link/lib/src/_utils.dart#L85-L101

https://github.com/gql-dart/gql/blob/c13e2fe2e6ca234ff5bc17263e3536ed2c907474/links/gql_http_link/lib/src/_utils.dart#L7-L23

I can also see this being very useful for other developers for doing conditional checks based on queries/mutations.

Closes https://github.com/gql-dart/gql/issues/305.

Hi this PR implements the graphql-transport-ws subprotocol. The code is a direct port of https://github.com/enisdenjo/graphql-ws/blob/master/src/client.ts with some changes to implement the Link interface and replace some JavaScript patterns to what one would expect in Dart. There is still some work to be done, at the moment there are some TODO comments and print statements in the code, some of the comments reference the JavaScript API and the tests for the apollo part are failing. I think some timer is being executed multiple times when the expectAsyncN function only expects a single execution. When trying to reconnect to the server, for example. However, the core logic works and is fully tested most of the changes that need to be done are about the external API or cleaning up comments.

I created a _testLinks function in the test file to test both subprotocols using the same logic. However, that makes it harder to test it in vscode since the extension does not recognize the test prefix associated with each subprotocol. One can use the dart test --name "transport ws sub-protocol <test-name>" cli, but I am not sure if we should try to split it so it is easier to debug.

At the moment, there are two separate Link implementations, with somewhat different parameters. Not sure what could be the best approach here, maybe we could try to unify some of the constructor parameters.

When a message of type "error" is encountered, the request stream throws an error with the payload (a list of GraphQLError). The behavior is different from the apollo protocol where multiple errors can be sent as regular events within the stream. We could do the same, the exact line is here. However, I believe there would be no easy way to distinguish between a message of type "error" and type "next" with errors payload as an end user. According to the spec, the type "error" ends the request, no more messages should be sent.

I have tested it end to end with a https://github.com/juancastillo0/leto server.

Thanks!

Right now, there is onelist of reserved names which get escaped when used as an identifier. This list includes reserved keywords in dart (like class) and names which are reserved by built_value (like 'toJson') and cannot be used as name for a field in built_value.

Since we also generate top-level constants, this single list is too restrictive and might lead to escaping of names when it is not needed.

Example: It is valid to define a top-level constant with the name 'rebuild', but we would still escape the name of the top-level constants which is generated from a directive like

directive @rebuild on FIELD_DEFINITION

to Grebuild

repeat 20 multipack info | dot -Tsvg -x | md5

outputs

2db5aa6b11b6b4f52e8283f2435d5212
d6156c52000e690a7c3e72e52553b9db
2db5aa6b11b6b4f52e8283f2435d5212
2db5aa6b11b6b4f52e8283f2435d5212
1979929664b2fbc1bce4975000f37ac3
5201338dd5c626e72645bbcedf9ea4bf
5201338dd5c626e72645bbcedf9ea4bf
2db5aa6b11b6b4f52e8283f2435d5212
5201338dd5c626e72645bbcedf9ea4bf
2db5aa6b11b6b4f52e8283f2435d5212
2db5aa6b11b6b4f52e8283f2435d5212
2db5aa6b11b6b4f52e8283f2435d5212
5201338dd5c626e72645bbcedf9ea4bf
5201338dd5c626e72645bbcedf9ea4bf
5201338dd5c626e72645bbcedf9ea4bf
5201338dd5c626e72645bbcedf9ea4bf
5201338dd5c626e72645bbcedf9ea4bf
2db5aa6b11b6b4f52e8283f2435d5212
5201338dd5c626e72645bbcedf9ea4bf
1979929664b2fbc1bce4975000f37ac3

on my machine. It should output the same hash every time if the algorithm was deterministic. The output of multipack info is deterministic, so it must be dot which is the culprit here. The actual difference in the SVG is tiny, it seems to come down to some rounding differences.

I believe this is a known issue of dot discussed here: https://gitlab.com/graphviz/graphviz/-/issues/1767

Because of this, the check SVG step in the CI does not really make sense in my opinion, since it would often fail even though the SVG image is up to date.