Hey guys, been playing around with an obj-c target on my fork of this repo (https://github.com/kieranlafferty/apollo-codegen)

Still not full working yet but thought it may be of interest to some people who are also interested in contributing to getting the codegen to work for objc.

Interested to get some early feedback. Thanks again for making such an awesome project!

Bug:
The schema.ts changes depending on which env it was last generated on. Without any code change.

Steps:
apollo-codegen generate $PATH/**/*.gql --schema $PATH/doc.json --target typescript --output $PATH/schema.ts

1. Run the above command on macOS Sierra (node v6.9.5), and save the schema.ts and submit it to your repository.

2. Check out your code on Ubuntu (node v6.9.5) and run the above command again.

3. Diff the schema.ts

You'd see that the schema.ts has changed. Nothing is removed and nothing is added. Just the order of some interfaces has changed.

Cause:
It is possible that $PATH/**/*.gql produces a different sorted list on OSX and Linux as sort may be different going from one OS to another.

Side effect
This doesn't break anything and everything works fine. However, in a group development environment, every time the codegen is executed in different env, it will produce a change which may require an unnecessary repo submission.

Currently in our JSON schema we have an input_obect that contains the following field

{
                            "name": "deviceID",
                            "description": null,
                            "type": {
                                "kind": "SCALAR",
                                "name": "String",
                                "ofType": null
                            },
                            "defaultValue": null
}

In the generated swift code, however, this is becoming deviceId and this same capitalization is being used in the endpoint request, causing it to fail as an unrecognized field.

Generated swift looks like:

public struct addUserInput: GraphQLMapConvertible {
  public var graphQLMap: GraphQLMap

  public init(firstName: String, lastName: String, identifier: String, identifierType: String, deviceId: String? = nil, clientMutationId: String? = nil) {
    graphQLMap = ["firstName": firstName, "lastName": lastName, "identifier": identifier, "identifierType": identifierType, "deviceId": deviceId, "clientMutationId": clientMutationId]
  }
}

I'm using version 0.10.9 and ran into the problem that some properties are generated more than one time into a Fragment interface.

I've assembled a very trivial example for this.
Schema:
zoo.graphql

interface Feedable {
    _id: ID!,
    kind: String!
    feedWith: String!
}


interface Reptile {
    scaleSkin: String!
}

type Crocodile implements Feedable, Reptile {
    _id: ID!,
    kind: String!
    feedWith: String!
    scaleSkin: String!

    victims: Int!
}

type Iguana implements Feedable, Reptile {
    _id: ID!,
    kind: String!
    feedWith: String!
    scaleSkin: String!

    age: Int!
}

root.graphql

type Query {
    getZoo: [Feedable!]!
}

schema {
    query: Query
}

And now I setup a very simple query with some fragments:
zooQuery.graphql

query getZoo {
    getZoo {
        ...FeedableFragment
    }
}

FeedableFragment

fragment FeedableFragment on Feedable {
    _id
    kind,
    feedWith

    ... on Reptile {
        scaleSkin
    }

    ... on Crocodile {
        victims
    }

    ... on Iguana {
        age
    }
}

And now I check the generated interfaces and I have this:

export interface FeedableFragment {
  _id: string;
  kind: string;
  feedWith: string;
  _id: string;
  kind: string;
  feedWith: string;
  scaleSkin: string;
  victims: number;
  _id: string;
  kind: string;
  feedWith: string;
  scaleSkin: string;
  age: number;
}

As you can see, the _id, kind, feedWith fields are repeated three times, instead of one.
What's more, I think that the scaleSkin and age properties should be optional properties.

Followup to #24 (comment)

We need to support extracting queries from template strings.

• Extract from gql tags
• Extract and infer a fragment name for Relay.QL tags

The following line
https://github.com/apollographql/apollo-codegen/blob/master/src/flow/codeGeneration.js#L81
optimistically PascalCases all type names even if they might in fact not be.

I agree PascalCase is conventional for naming GraphQL types but I can't find that it's required.

I was trying to generate flow annotations against a schema that contained types with camelCase names. This caused fields referencing the type to remain in camelCase whereas the definition of the type was in PascalCase

I can use objects as inline fragments, but for interface fragments no code is generated. Shouldn't this be possible?

{
... on ClosedEvent {
        # object, works
    }  
... on IssueEvent {
        # interface, no code is generated
    }
}

(Using Swift, but saw similar results with the other targets.)

Android GraphQL requires custom scalar types to be listed in typesUsed array in IR.json file.
Custom scalar type should be defined in the same way as enum types. This definitions will be later used by android code generator and response adapters

For example:

"typesUsed": [
    {
      "kind": "???",
      "name": "Date",
      "description": "DateTime value"
    },
    {
      "kind": "EnumType",
      "name": "Episode",
      "description": "The episodes in the Star Wars trilogy",
      "values": [
        {
          "name": "NEWHOPE",
          "description": "Star Wars Episode IV: A New Hope, released in 1977."
        },
        {
          "name": "EMPIRE",
          "description": "Star Wars Episode V: The Empire Strikes Back, released in 1980."
        },
        {
          "name": "JEDI",
          "description": "Star Wars Episode VI: Return of the Jedi, released in 1983."
        }
      ]
    }

Currently String Enum types are not being converted to its rawValue before being used as a parameter. This causes a crash as Swift Enum types are not allowed in JSON serialization and the following error is thrown:

Terminating app due to uncaught exception 'NSInvalidArgumentException', reason: 'Invalid type in JSON write (_SwiftValue)'

Here is some relevant code demonstrating the issue:

API.swift

public enum Size: String {
  case small = "small" /// A Small Size
  case large = "large" /// A Large Size
}

public final class UpdateSizeMutation: GraphQLMutation {
   ...
  public var variables: GraphQLMap? {
    return ["size": size]
  }
}

GraphQLNetworkTransport

    func send<Operation: GraphQLOperation>(operation: Operation, completionHandler: @escaping (GraphQLResponse<Operation>?, Error?) -> Void) -> Cancellable {
        let params: [String: Any] = ["query": type(of: operation).queryDocument as Any, "variables": operation.variables as Any]
        let request = manager.request(baseURLString, method: .post, parameters: params, encoding: JSONEncoding.default, headers: headers)
        ...
        return request
    }

The Swift Enum is therefore included inside of the params variable when sending the request and causes an error when encoding to JSON.

A potential fix is to use .rawValue for string enum types inside of variables during code generation.

Potential Fix

  public var variables: GraphQLMap? {
    return ["size": size.rawValue]
  }

This is a problem that can also be addressed within apollo-ios.

Ive installed apollo-codegen globally, which is at version .10.4:

/usr/local/lib
└─┬ [email protected]

When running the iOS sample apps, I get the following error:

/usr/local/lib/node_modules/apollo-codegen/lib/cli.js:33
_process2.default.on('unhandledRejection', error => {
                                                 ^^
SyntaxError: Unexpected token =>
    at exports.runInThisContext (vm.js:53:16)
    at Module._compile (module.js:413:25)
    at Object.Module._extensions..js (module.js:448:10)
    at Module.load (module.js:355:32)
    at Function.Module._load (module.js:310:12)
    at Function.Module.runMain (module.js:471:10)
    at startup (node.js:117:18)
    at node.js:953:3
error: The version of Apollo.framework in your project requires the use of version 0.9 of apollo-codegen, but an unknown older version seems to be installed.

• It seems to work from command line:
  apollo-codegen download-schema http://localhost:8080/graphql --output schema.json
• So it may be how its executed from Xcode.
• Seems to work fine on 0.10.0

If we try to spread a fragment on a list type, the generated ts type won't be a list.

@dotansimha this feels very similar to the issue you raised about inline fragments and for which you created a PR (#17). Care to take a look? Don't worry, if you don't have time, I'll do it, but I'm interested in bringing contributors like you to this!

When I generate Swift-code from a query that include __typename there is duplication in the Query class.

query Foo($id: ID!) {
    bar {
        __typename
        ... on bazType {
            ..
        }
    }
}

Generate code like:

public static let operationDefinition =
"query Foo($id: ID!) {" +
"  bar {" +
"    __typename" +
"    __typename" +
"    ... on bazType {" +
"      .." +
"    }" +
"  }" +
"}"

While it is easy to avoid including it in the query, it seem reasonable to handle this.

apollo-codegen takes all of a project's GraphQL files at once to produce a schema, in the form **/*.graphql (or similar).

Whilst validation isn't this tool's primary job (I assume), when it displays errors, it currently doesn't display from which files those errors are sourced.

Whether embedded in something else or used directly, it seems like the resulting validation errors would be more useful if they displayed the file in which they occurred?

I see 0.8.0 is installed, but it still gives me errors

npm install -g apollo-codegen
/usr/local/bin/apollo-codegen -> /usr/local/lib/node_modules/apollo-codegen/lib/cli.js
[email protected] /usr/local/lib/node_modules/apollo-codegen

In Xcode:

Error: Cannot find module 'process'
    at Function.Module._resolveFilename (module.js:336:15)
    at Function.Module._load (module.js:278:25)
    at Module.require (module.js:365:17)
    at require (module.js:384:17)
    at Object.<anonymous> (/usr/local/lib/node_modules/apollo-codegen/lib/cli.js:20:16)
    at Module._compile (module.js:460:26)
    at Object.Module._extensions..js (module.js:478:10)
    at Module.load (module.js:355:32)
    at Function.Module._load (module.js:310:12)
    at Function.Module.runMain (module.js:501:10)
error: The version of Apollo.framework in your project requires the use of version 0.8 of apollo-codegen, but an unknown older version seems to be installed.

Many variants of apollo-codegen -version fail with the same error as above.
Please advise.

Cheers

Hi there,

apollo-codegen seems to generate an empty init() on a mutation struct in API.swift, my server implementation for the mutation is pretty straightforward with one input object (UserInput) and one return type (UserType)

Xcode won't compile with an empty init because the graphQLMap property is not being initialised. The workaround currently is modifying API.swift and initialising graphQLMap inside the empty init before running the project, like:

public init() {
  graphQLMap = [:]
}

Steps to reproduce:

1. Define a Mutation on the server that takes an input object
2. Use apollo-codegen to download the schema and copy it to your target
3. Build your project
4. The generated API.swift has an empty public init()

Can anyone else reproduce this?

This is my API.swift for reference: API.swift

Thanks for the wonderful set of tools, otherwise. :)

In some cases, the generated interfaces can contain duplicate fields,
coming from multiple fragments.

This will cause TS errors.

Assume schema/type:

type Query {
  episodes: [Episode]! //non-null list
}

Typescript output for query:

query LoadEpisodes {
  episodes {
     ...EpisodeInfo
  }
}

fragment EpisodeInfo on Episode {
  name
}

Results in the following output:

export interface LoadEpisodesQuery {
  episodes: EpisodeInfoFragment;
}

Expected:

export interface LoadEpisodesQuery {
  episodes: Array<EpisodeInfoFragment>;
}

Am I doing something wrong or are these errors able to be recreated by anyone else?

Environment

• nvm 0.32.1
• node 7.4.0
• yarn 0.18.1
• Darwin Kernel Version 16.3.0: Thu Nov 17 20:23:58 PST 2016; root:xnu-3789.31.2~1/RELEASE_X86_64 x86_64

Steps To Reproduce

• Install apollo-codegen via yarn global add apollo-codegen
• Run apollo-codegen download-schema ...
• Run apollo-codegen generate **/*.graphql ...

I was originally trying to generate flow types and wanted to verify undesired behavior so I attempted the same command so i tried with --target typescript instead of --target flow. The inputs and outputs of both commands are below.

Errors

Flow Scenario

Command:

apollo-codegen generate **/*.graphql --schema schema.graphql.json --target flow --output schema.flow.js`

Response:

Invalid values:
  Argument: target, Given: "flow", Choices: "swift", "json", "ts", "typescript"

TypeScript Scenario

Command:

apollo-codegen generate **/*.graphql --schema schema.graphql.json --target typescript --output schema.graphql.ts

Response:

Error: ENOENT: no such file or directory, open '/Users/Gabe/Sites/foresight-client/**/*.graphql'
    at Object.fs.openSync (fs.js:558:18)
    at Object.fs.readFileSync (fs.js:468:33)
    at /Users/Gabe/.nvm/versions/node/v7.4.0/lib/node_modules/apollo-codegen/src/loading.js:28:21
    at Array.map (native)
    at loadAndMergeQueryDocuments (/Users/Gabe/.nvm/versions/node/v7.4.0/lib/node_modules/apollo-codegen/src/loading.js:27:30)
    at generate (/Users/Gabe/.nvm/versions/node/v7.4.0/lib/node_modules/apollo-codegen/src/generate.js:14:20)
    at Object.handler (/Users/Gabe/.nvm/versions/node/v7.4.0/lib/node_modules/apollo-codegen/src/cli.js:91:9)
    at Object.self.runCommand (/Users/Gabe/.nvm/versions/node/v7.4.0/lib/node_modules/apollo-codegen/node_modules/yargs/lib/command.js:170:22)
    at parseArgs (/Users/Gabe/.nvm/versions/node/v7.4.0/lib/node_modules/apollo-codegen/node_modules/yargs/yargs.js:920:28)
    at Object.get [as argv] (/Users/Gabe/.nvm/versions/node/v7.4.0/lib/node_modules/apollo-codegen/node_modules/yargs/yargs.js:860:16)

We could generate Flow types for the GraphQL queries in the project in parallel of something like graphql-document-collector to get the query docs globally.

I'll work on this in a folder parallel to the swift/ folder.

I've been messing around and converting the GitHunt-React repo to use TypeScript along with codegen from this module. Really cool stuff, and it almost worked except for one error in the generated Typescript file.

I got an error on FeedEntryFragment due to incorrect extension of RepoInfoFragment.

export interface FeedEntryFragment extends VoteButtonsFragment, RepoInfoFragment {
  id: number;
  commentCount: number;
  repository: {
    fullName: string,
    htmlUrl: string,
    owner: {
      avatarUrl: string,
    } | null,
  };
}

export interface RepoInfoFragment {
  createdAt: number;
  repository: {
    description: string | null,
    stargazersCount: number,
    openIssuesCount: number | null,
  };
  postedBy: {
    htmlUrl: string,
    login: string,
  };
}

export interface VoteButtonsFragment {
  score: number;
  vote: {
    voteValue: number,
  };
}

Unfortunately typescript can't merge nested interface properties. What seems to be needed is to extract the nested type information into their own interfaces which could then be either merged or extended. I can't think of an exact example right now, but I'll try to put something together

In my scheme there is a non optional array called genres.
genres: [String!]

In the API.swift file it is not optional
public let genres: [String]

Error:

▿ Optional<Error>
  ▿ some : GraphQLResultError
    ▿ path : 7 elements
      - 0 : "viewer"
      - 1 : "market"
      - 2 : "channels"
      - 3 : "edges"
      - 4 : "0"
      - 5 : "node"
      - 6 : "genres"
    - underlying : Apollo.JSONDecodingError.nullValue

Looking through the tests for TypeScript codegen, the only non-nullable fields are scalars. Is this a limitation of the codegen or of GraphQL?

export interface HeroAndFriendsNamesQuery {
  hero: {
    name: string,
    friends: Array< {
      name: string,
    } > | null,
  } | null;
}

I bumped to version 0.10.9 and get an error I didn't get with version 0.9.4 for the same query.

The GraphQLResultReader doesn't have a variables member.

API.swift with my query and the xCode error:

I get the following error:

> @ generate-types /Users/clayne/workspace/casalova-web
> apollo-codegen generate **/*.graphql --schema graphql-schema.json --target typescript --output ./packages/casalova-web-client/src/universal/graphql-schema.d.ts

Apollo iOS does not support anonymous operations
This anonymous operation must be the only defined operation.
Cannot query field "hero" on type "Query".
Cannot query field "hero" on type "Query".
Apollo iOS inserts __typename automatically when needed, please do not include it explicitly
Unknown type "Episode".
Cannot query field "hero" on type "Query".
Unknown type "Character".
Unknown type "Droid".
Unknown type "Human".
Unknown type "Episode".
Cannot query field "hero" on type "Query".
Cannot query field "hero" on type "Query".
There can only be one operation named "HeroName".
There can only be one operation named "HeroName".
Cannot query field "hero" on type "Query".
Cannot query field "hero" on type "Query".
Cannot query field "hero" on type "Query".
There can only be one operation named "HeroName".
There can only be one operation named "HeroName".
Cannot query field "hero" on type "Query".
Apollo iOS needs to be able to insert __typename when needed, please do not use it as an alias
Validation of GraphQL query document failed

I'm generating swift using the default startwars scheme.json. Its failing to produce output.
scheme.json is attached.

apollo-codegen generate **/*.graphql --schema schema.json --output API.swift

produces the error:

Apollo iOS does not support anonymous operations
This anonymous operation must be the only defined operation.
Apollo iOS inserts __typename automatically when needed, please do not include it explicitly
Cannot query field "height" on type "Human".
There can only be one operation named "HeroName".
There can only be one operation named "HeroName".
There can only be one operation named "HeroName".
There can only be one operation named "HeroName".
Apollo iOS needs to be able to insert __typename when needed, please do not use it as an alias
Validation of GraphQL query document failed

schema.json.txt

I'm following the fragment colocation suggestion as outlined here but apollo-codegen doesn't seem to be following the import.

The current flow type generation only generates specific flow types for queries/mutations detected.

I'm curious about if there are any benefits to this approach compared to just generating the flow equivalent of the entire schema. It should be a pretty simple mapping, but maybe I'm missing something.

E.g. if I have a couple of mutations that all resolve to SomeResponseInterface and a specific mutation resolves to SomeImplementationOfSomeResponseInterface I can find use for both the generic and specific types in my code (shared error handling code can be annotate with interface and code dealing with the special fields can be annotated with the other type etc.).

This should also enable any traversals of returned data automatically being type checked even when a query is changed, independent of regenerating flow types (as long as schema stays the same).

The obvious drawback with not-query-specific-annotations is that the flow type-checking doesn't guarantee that I've actually queried for a specific field (but this is currently a little sketchy depending on how diligent(/automated) you are in regenerating the flow annotations since that need to happen for every change in a query).

Maybe the current style of annotations should be accompanied by a complete set of type declarations for the entire schema too?

We could generate TS types for the GraphQL queries in the project in parallel of something like graphql-document-collector to get the query docs globally.

I'll work on this in a folder parallel to the swift/ folder.

Question/Feature request regarding the output of the generated code for swift when fragments are used.

We have a fragment called "FormattedText" to encapsulate complex text fields that have nested bold/italic/link content.
And example quest would be

...
...
  headingText {
        ...formattedText
      }
...

The returned data is as follows

"headingText": {
            "spans": [
              {
                "text": "Credit card utilization"
              }
            ]
          },

The issue arrises with the gerenated code:

...
public let headingText: HeadingText?
...


  public struct HeadingText: GraphQLMappable {
                    public let __typename = "FormattedText"

                    public let fragments: Fragments

                    public init(reader: GraphQLResultReader) throws {
                        let formattedText = try FormattedText(reader: reader)
                        fragments = Fragments(formattedText: formattedText)
                    }

                    public struct Fragments {
                        public let formattedText: FormattedText
                    }
                }
...
...
//and finally Fragmented Text

public struct FormattedText: GraphQLNamedFragment {
    public static let fragmentDefinition =
        "fragment formattedText on FormattedText {" +
        "  spans {" +
        "    text" +
        "  }" +
        "}"

    public static let possibleTypes = ["FormattedText"]

    public let __typename = "FormattedText"
    public let spans: [Span?]?

    public init(reader: GraphQLResultReader) throws {
        spans = try reader.optionalList(for: Field(responseName: "spans"))
    }

    public struct Span: GraphQLMappable {
        public let __typename = "FormattedSpan"
        public let text: String?

        public init(reader: GraphQLResultReader) throws {
            text = try reader.optionalValue(for: Field(responseName: "text"))
        }
    }
}

In order to access this object you would write:

data.headingText?.fragments.formattedText

The feedback is the fragments contatiner necessary? Should the HeadingText class itself not be a FormattedText or more closely implement the FormattedText fragment. This indirection strikes me as unnecessary.

Even the headingText?.__typename == "FormattedText" so this type has been fully specified.

eg:

headingText	CKGraphQL.TestCreditFactorsQuery.Data.CreditFactor.HeadingText?	some
	__typename	String	"FormattedText"			
	fragments	CKGraphQL.TestCreditFactorsQuery.Data.CreditFactor.HeadingText.Fragments	
		formattedText	CKGraphQL.FormattedText	
			__typename	String	"FormattedText"	
			spans	[CKGraphQL.FormattedText.Span?]?	1 value	some
				[0]	CKGraphQL.FormattedText.Span?	some
					__typename	String	"FormattedSpan"	
					text	String?	"Credit card utilization"	some

So my suggestion is, can we make HeadingText more directly inherit from FormattedText?

 public struct HeadingText: FormattedText {

What are the conditions under which this fragment structure becomes more useful?

Version: 0.11.0

Part of my schema.json looks as follows:

"fields": null,
          "enumValues": null,
          "description": "This is the first line of the comment.\n\nThis is the second line.\n"

Intended outcome of generated code

// This is the first line of the comment.
//
// This is the second line.
//

public struct ResponseInput: GraphQLMapConvertible {
  public var graphQLMap: GraphQLMap
...

Actual outcome

// This is the first line of the comment.

This is the second line. 

public struct ResponseInput: GraphQLMapConvertible {
  public var graphQLMap: GraphQLMap
...

Because This is the second line. is not in comment-block, it's treated as part of the swift code. This obviously leads to compiler error.

Flow types that represent response values (rather than input values) could probably be defined as “exact” objects, using {| |} instead of { }. See: https://flowtype.org/docs/objects.html#exact-object-types

Per the above link, “Exact object types are a very useful tool for helping Flow to refine unions of object types and notice typos on property names and refinements.”

This has been discussed on apollographql/apollo-kotlin#83
On Android, the generated classes need to be organized by package, so that requires us to be able to determine the original file path for each of the .graphql files so we can use that to determine which package name to use. Organizing classes in packages is a very standard practice on Android and this solution is inspired by how Sqldelight handles .sq files.

@martijnwalraven

I've tested on Mac, Windows, and Bash on Ubuntu on Windows. It works great on Mac and Windows Bash, but when I run in a command prompt or powershell:

apollo-codegen generate src/**/*.graphql --schema graphql.schema.json --target typescript --output src/schema.d.ts

I get errors like the following:

Error: ENOENT: no such file or directory, open 'C:\Projects\src\**\*.graphql'

It's interpreting the wildcard path as a string

Using version 0.4.3, performing a mutation that returns the following content:

{
  "errors": [
    {
      "message":"In field \"updatePassword\": Invalid password",
      "locations":[{"line":1,"column":0}]
    }
  ],
  "data":{}
}

the completionHandler returns no error (GraphQLResultError), but the result contains errors (result.errors: [GraphQLError]?).

Now using 0.5.1, the resultHandler returns an error (GraphQLResultError) with the following message:

GraphQLResultError(path: ["updatePassword"], underlying: Apollo.JSONDecodingError.missingValue)

For information, here is the client definition for that mutation (apollo-cogen 0.10):

public final class UpdatePasswordMutation: GraphQLMutation {
  public static let operationDefinition =
    "mutation UpdatePassword($oldPassword: String!, $newPassword: String!) {" +
    "  updatePassword(oldPassword: $oldPassword, newPassword: $newPassword) {" +
    "    email" +
    "    timezone" +
    "  }" +
    "}"

  public let oldPassword: String
  public let newPassword: String

  public init(oldPassword: String, newPassword: String) {
    self.oldPassword = oldPassword
    self.newPassword = newPassword
  }

  public var variables: GraphQLMap? {
    return ["oldPassword": oldPassword, "newPassword": newPassword]
  }

  public struct Data: GraphQLMappable {
    public let updatePassword: UpdatePassword?

    public init(reader: GraphQLResultReader) throws {
      updatePassword = try reader.optionalValue(for: Field(responseName: "updatePassword", arguments: ["oldPassword": reader.variables["oldPassword"], "newPassword": reader.variables["newPassword"]]))
    }

    public struct UpdatePassword: GraphQLMappable {
      public let __typename = "Account"
      public let email: String?
      public let timezone: String?

      public init(reader: GraphQLResultReader) throws {
        email = try reader.optionalValue(for: Field(responseName: "email"))
        timezone = try reader.optionalValue(for: Field(responseName: "timezone"))
      }
    }
  }
}

and how it was with apollo-codegen 0.9:

public final class UpdatePasswordMutation: GraphQLMutation {
  public static let operationDefinition =
    "mutation UpdatePassword($oldPassword: String!, $newPassword: String!) {" +
    "  updatePassword(oldPassword: $oldPassword, newPassword: $newPassword) {" +
    "    email" +
    "    timezone" +
    "  }" +
    "}"

  public let oldPassword: String
  public let newPassword: String

  public init(oldPassword: String, newPassword: String) {
    self.oldPassword = oldPassword
    self.newPassword = newPassword
  }

  public var variables: GraphQLMap? {
    return ["oldPassword": oldPassword, "newPassword": newPassword]
  }

  public struct Data: GraphQLMappable {
    public let updatePassword: UpdatePassword?

    public init(reader: GraphQLResultReader) throws {
      updatePassword = try reader.optionalValue(for: Field(responseName: "updatePassword"))
    }

    public struct UpdatePassword: GraphQLMappable {
      public let __typename = "Account"
      public let email: String?
      public let timezone: String?

      public init(reader: GraphQLResultReader) throws {
        email = try reader.optionalValue(for: Field(responseName: "email"))
        timezone = try reader.optionalValue(for: Field(responseName: "timezone"))
      }
    }
  }
}

It looks like something's wrong.. Can you help me there?

Hi, I have been using this library for the past 2 months and it is fantastic (app for an animal shelter), but I added a package to Carthage this evening and upgraded, now the generated API file is not being generated correctly. I have a rather large query which fetches the bulk of the data and I have structured my GraphQL schema with nested inputs I have not changed the schema and it was previously working.

Here is the query (I have left out the fragments):

query AllData($facilityId:ID, $squareImageWidth:Float!, $aspectImageWidth:Float!, $cityId:ID) { viewer { allAnimals(order:NAME filters: { facilityId: $facilityId, cityId: $cityId }) { list { ...AnimalFragment } count } allFacilities(order: NAME, filters: { cityId: $cityId }) { list { ...FacilityFragment } count } allLocations(order: NAME, filters: { locationType:[ADOPTION_LOCATION], cityId: $cityId }) { list { ...LocationFragment } count } allCountries(order: NAME) { list { ...CountryFragment } count } allCities(order: NAME) { list { ...CityFragment } count } } }

but I now get undefined for all those variables in the generated code:

`
public struct Viewer: GraphQLMappable {
public let __typename: String
public let allAnimals: AllAnimal?
public let allFacilities: AllFacility
public let allLocations: AllLocation
public let allCountries: AllCountry
public let allCities: AllCity

  public init(reader: GraphQLResultReader) throws {
    __typename = try reader.value(for: Field(responseName: "__typename"))
    allAnimals = try reader.optionalValue(for: Field(responseName: "allAnimals", arguments: ["order": "NAME", "filters": ["facilityId": undefined, "cityId": undefined]]))
    allFacilities = try reader.value(for: Field(responseName: "allFacilities", arguments: ["order": "NAME", "filters": ["cityId": undefined]]))
    allLocations = try reader.value(for: Field(responseName: "allLocations", arguments: ["order": "NAME", "filters": ["locationType": ["ADOPTION_LOCATION"], "cityId": undefined]]))
    allCountries = try reader.value(for: Field(responseName: "allCountries", arguments: ["order": "NAME"]))
    allCities = try reader.value(for: Field(responseName: "allCities", arguments: ["order": "NAME"]))
  }

etc...

`

the shelter I am developing this for is hosting a huge event in a few weeks and I need to get this complete, please any help would be much appreciated.

Thanks.

Hi,

It seems that the apollo-codegen is generating an unnecessary objects off the typeName.

I think the typeName should define the type and not the instance name.

For example:

        public struct PointsLastYear: GraphQLMappable {
          public let __typename = "point"
          public let time: String?
          public let value: Double?

          public init(reader: GraphQLResultReader) throws {
            time = try reader.optionalValue(for: Field(responseName: "time"))
            value = try reader.optionalValue(for: Field(responseName: "value"))
          }
        }

        public struct Point: GraphQLMappable {
          public let __typename = "point"
          public let time: String?
          public let value: Double?

          public init(reader: GraphQLResultReader) throws {
            time = try reader.optionalValue(for: Field(responseName: "time"))
            value = try reader.optionalValue(for: Field(responseName: "value"))
          }
        }
      }

The typeName for PointsLastYear and Point is the same i.e point.

Based on the schema below, it does seem like we are defining the typeName as point, but the generator ends up creating an additional object of the same instance type, which deems unnecessary.

    {
              "type": {
                "ofType": {
                  "ofType": null,
                  "name": "point",
                  "kind": "OBJECT"
                },
                "name": null,
                "kind": "LIST"
              },
              "name": "points",
              "isDeprecated": false,
              "description": null,
              "deprecationReason": null,
              "args": []
            },
            {
              "type": {
                "ofType": {
                  "ofType": null,
                  "name": "point",
                  "kind": "OBJECT"
                },
                "name": null,
                "kind": "LIST"
              },
              "name": "pointsLastYear",
              "isDeprecated": false,
              "description": null,
              "deprecationReason": null,
              "args": []
            },

This is causing unnecessary abstractions and objects which are really not needed as they are of the same type.

Is this the kind of expected behavior and by design ?

Please advice.

Hi,

When a query does not have any variables, the generated GraphQLQuery class does not contain a public init.
This prevents initialising an instance outside of the module.
You can see this in the StarWars Playground when using HeroNameQuery():

Playground execution failed: error: StarWarsAPI.playground:4:21: error: 'HeroNameQuery' initializer is inaccessible due to 'internal' protection level
client.fetch(query: HeroNameQuery()) { (result, error) in

The ParseQueryResponseTests however don’t catch this issue since there, the StarWars API is local, inside the ApolloTests module.
Is this by design, or should the queries be available outside of the module as well?

Possible fix is emitting an empty init for operations without variables :

diff --git a/src/swift/codeGeneration.js b/src/swift/codeGeneration.js
index cc415fd..ea0b2a9 100644
--- a/src/swift/codeGeneration.js
+++ b/src/swift/codeGeneration.js
@@ -126,6 +126,8 @@ export function classDeclarationForOperation(
           `]`
         ));
       });
+    } else {
+      initializerDeclarationForProperties(generator, []);
     }

     structDeclarationForSelectionSet(

I'm using version 0.10.9 , and if I declare a fragment like this:

fragment ICQMessageFragment on RandomTest {
    _id
}

the name of the generated ts interface will be:

export interface IcqMessageFragment  

Notice that ICQ turned into Icq

however when I use this fragment on a query, like:

getMessages {
  ...ICQMessageFragment 
} 

The generated ts interface inside the query will use the original form of the fragment, like this:

 Array<ICQMessageFragment>

And Typescript immediately throws an error that it doesn't find: ICQMessageFragment because it is case-sensitive.

If you want to reuse fragments and import them with the

#import "../common-fragments.graphql

syntax, then validation will fall as the common-fragments.graphql file will fail with these errors:

Fragment "Foo" is never used.
Validation of GraphQL query document failed

This can be "hacked" by changing line 16 of validations.js from:

  const rules = [NoAnonymousQueries, NoExplicitTypename, NoTypenameAlias].concat(_graphql.specifiedRules)

to

  const rules = [NoAnonymousQueries, NoExplicitTypename, NoTypenameAlias].concat(_graphql.specifiedRules).filter((rule) => rule.name !== "NoUnusedFragments")

However, the default should probably remain as is but this be set via CLI option.

If this would be an acceptable feature, I'll add a PR.

[email protected] prepublish /Users/moniquehaas/apollo-codegen
rm -rf lib && npm run compile

[email protected] compile /Users/moniquehaas/apollo-codegen
babel src --out-dir lib --source-maps

ReferenceError: Unknown plugin "transform-runtime" specified in "/Users/moniquehaas/apollo-codegen/.babelrc" at 0, attempted to resolve relative to "/Users/moniquehaas/apollo-codegen"
at /usr/local/lib/node_modules/babel-cli/node_modules/babel-core/lib/transformation/file/options/option-manager.js:193:17
at Array.map (native)
at Function.normalisePlugins (/usr/local/lib/node_modules/babel-cli/node_modules/babel-core/lib/transformation/file/options/option-manager.js:173:20)
at OptionManager.mergeOptions (/usr/local/lib/node_modules/babel-cli/node_modules/babel-core/lib/transformation/file/options/option-manager.js:271:36)
at OptionManager.addConfig (/usr/local/lib/node_modules/babel-cli/node_modules/babel-core/lib/transformation/file/options/option-manager.js:221:10)
at OptionManager.findConfigs (/usr/local/lib/node_modules/babel-cli/node_modules/babel-core/lib/transformation/file/options/option-manager.js:364:16)
at OptionManager.init (/usr/local/lib/node_modules/babel-cli/node_modules/babel-core/lib/transformation/file/options/option-manager.js:412:12)
at File.initOptions (/usr/local/lib/node_modules/babel-cli/node_modules/babel-core/lib/transformation/file/index.js:191:75)
at new File (/usr/local/lib/node_modules/babel-cli/node_modules/babel-core/lib/transformation/file/index.js:122:22)
at Pipeline.transform (/usr/local/lib/node_modules/babel-cli/node_modules/babel-core/lib/transformation/pipeline.js:42:16)

npm ERR! Darwin 16.5.0
npm ERR! argv "/usr/local/bin/node" "/usr/local/bin/npm" "run" "compile"
npm ERR! node v6.10.2
npm ERR! npm v3.10.10
npm ERR! code ELIFECYCLE
npm ERR! [email protected] compile: babel src --out-dir lib --source-maps
npm ERR! Exit status 1
npm ERR!
npm ERR! Failed at the [email protected] compile script 'babel src --out-dir lib --source-maps'.
npm ERR! Make sure you have the latest version of node.js and npm installed.
npm ERR! If you do, this is most likely a problem with the apollo-codegen package,
npm ERR! not with npm itself.
npm ERR! Tell the author that this fails on your system:
npm ERR! babel src --out-dir lib --source-maps
npm ERR! You can get information on how to open an issue for this project with:
npm ERR! npm bugs apollo-codegen
npm ERR! Or if that isn't available, you can get their info via:
npm ERR! npm owner ls apollo-codegen
npm ERR! There is likely additional logging output above.

npm ERR! Please include the following file with any support request:
npm ERR! /Users/moniquehaas/apollo-codegen/npm-debug.log

npm ERR! addLocal Could not install /Users/moniquehaas/apollo-codegen
npm ERR! Darwin 16.5.0
npm ERR! argv "/usr/local/bin/node" "/usr/local/bin/npm" "install" "-g" "apollo-codegen"
npm ERR! node v6.10.2
npm ERR! npm v3.10.10
npm ERR! code ELIFECYCLE
npm ERR! [email protected] prepublish: rm -rf lib && npm run compile
npm ERR! Exit status 1
npm ERR!
npm ERR! Failed at the [email protected] prepublish script 'rm -rf lib && npm run compile'.
npm ERR! Make sure you have the latest version of node.js and npm installed.
npm ERR! If you do, this is most likely a problem with the apollo-codegen package,
npm ERR! not with npm itself.
npm ERR! Tell the author that this fails on your system:
npm ERR! rm -rf lib && npm run compile
npm ERR! You can get information on how to open an issue for this project with:
npm ERR! npm bugs apollo-codegen
npm ERR! Or if that isn't available, you can get their info via:
npm ERR! npm owner ls apollo-codegen
npm ERR! There is likely additional logging output above.

npm ERR! Please include the following file with any support request:
npm ERR! /Users/moniquehaas/npm-debug.log

Is there currently a mechanism by which the Swift model structs can be mutated?

My two use cases are as follows:

1. Optimistic updates -- aka the user submits a change, and we want to update the UI before receiving the response from the server
2. A source other than the GraphQL API indicates that data returned from the GraphQL API has changed -- for example, we have unreadCount as a property of a GraphQL model struct A, but push notifications also contain an unreadCount and when we receive a push notification, we want to mutate the A struct to update it's unreadCount.

Currently, it appears that structs are generated with properties as constants (let), and the simple solution would be to change these to var. Would this be something that would be considered to add to the apollo-codegen CLI as an option?

Most (if not all) of the schema.json files provided online (such as Relay for example) have the data field as the top-level field.

When downloading a schema through an introspection query using apollo-codegen, it only stores the __schema object in the json file.

IMO, keeping the data field should provide more flexibility for users, particularly in cases when someone doesn't go through the download-schema process.

This would also require minor modification to the checks in loading.js to support both cases.

I can push a fix if that's of interest.

We were relying on the GraphQL type names coming in as described in GraphQL, so we could add our own typealiases and conversion code in Swift. For example we had:

public extension OurGeneratedQueryType {
    public typealias ID = String
    public typealias ResourceURL = URL
}

What's more, we have certain types notated as Doubles in our Schema, which used to arrive in Swift as Doubles. Since the change these are all Strings. Using GraphQL's built-in Float type is not a good compromise here, since we want to keep the precision of the Double type (from what I've read, Javascript / MongoDB floating point numbers have double precision).

By changing everything to a String at the codegen stage we're losing valuable type information – type information which in my opinion is the whole point of using Apollo-iOS / codegen.

If the aim of this change was to simplify the developer experience for beginners, maybe it'd be a good compromise to generate a block that the user can then override themselves if needed? Like this:

public extension GeneratedQueryType {
    public typealias ID = String
    public typealias CustomScalarType = String
    etc...
}

That way users can create their own conversion code for specific Scalars without coupling the conversion code to specific queries.

Original issue described here apollographql/apollo-kotlin#222

Query:

query getAllTrips {
  allTrips {
    ...commonFields
  }
}

fragment commonFields on Plan {
  id
  title
}

Where Plan extends Trip, response:

{
  "data": {
    "allTrips": [
      {
        "__typename": "Trip",
        "id": "2c494055-78bc-430c-9ab7-19817f3fc060",
        "title": "Vacation!"
      }
    ]
  }
}

Codegen generates IR with only one typeCondition =Plan.

"fragments": [
		{
			"fragmentName": "commonFields",
			"source": "fragment commonFields on Plan {\n  __typename\n  id\n  title\n}",
			"typeCondition": "Plan",
			"fields": [
				{
					"responseName": "id",
					"fieldName": "id",
					"type": "ID!"
				},
				{
					"responseName": "title",
					"fieldName": "title",
					"type": "String!"
				}
			],
		}
	],

So to be able correctly parse response on Android, we need to get access to all possibleTypes = ["Tree", "Plan"]

Many generators or config editors will include a commented warning at the top of generated files to indicate that the file should be edited directly since it was generated.

Perhaps it would be a good idea to either add a flag to insert a standard warning or an argument to specify the warning text.

If you have a query like:

query GetSimpleQuery {
    member {
        profile {
            private
        }
    }
}

This'll generate

public final class GetSimpleQueryQuery: GraphQLQuery {
  public static let operationDefinition =
    "query GetSimpleQuery {" +
    "  member {" +
    "    profile {" +
    "      private" +
    "    }" +
    "  }" +
    "}"
  public init() {
  }

  public struct Data: GraphQLMappable {
    public let member: Member?

    public init(reader: GraphQLResultReader) throws {
      member = try reader.optionalValue(for: Field(responseName: "member"))
    }

    public struct Member: GraphQLMappable {
      public let __typename = "Member"
      public let profile: Profile

      public init(reader: GraphQLResultReader) throws {
        profile = try reader.value(for: Field(responseName: "profile"))
      }

      public struct Profile: GraphQLMappable {
        public let __typename = "MemberProfile"
        public let private: Bool?

        public init(reader: GraphQLResultReader) throws {
          private = try reader.optionalValue(for: Field(responseName: "private"))
        }
      }
    }
  }
}

Which produces the error:

Keyword private cannot be used as an identifier here

When generating .swift files can we escape these?

Just a suggestion, but would it make sense to split out different targets ala plugins rather than having them all exist in the same repo?

When we use react-apollo like this :

let Component = ({data:GetSomethingQuery}) => {
  ...
}
export let ComponentWithApollo = graphql(getSomethingQuery)(Component)

with GetSomethingQuery as the result of the apollo-codegen generate, react-apollo add to data additional properties like loading, refetch or subscribeToMore.

How could we add those fields to the type definition ?

The graphql spec at http://facebook.github.io/graphql/#sec-Required-Non-Null-Arguments says

If it is not non‐null, the argument is optional.

And I have been happily omitting them in my JavaScript/react-apollo projects for weeks. But in transitioning to TypeScript I want to avoid explicitly supplying each nullable argument, with a null value.

So the following is just a hacky excursion to make such nullable arguments optional in their typescript interface declarations:

For example:

export interface AddPassengerMutationVariables {
  bookingId: number;
  name: string;
  email: string | null;
  mobile: string | null;
  phone: string | null;
}

becomes

export interface AddPassengerMutationVariables {
  bookingId: number;
  name: string;
  email?: string | null;
  mobile?: string | null;
  phone?: string | null;
}

Frankly I just started with typescript yesterday so this might not even be a sound idea but I just wanted to leave this here for any discussion...

diff --git a/src/typescript/codeGeneration.js b/src/typescript/codeGeneration.js
index fb5dfad..d5b5b89 100644
--- a/src/typescript/codeGeneration.js
+++ b/src/typescript/codeGeneration.js
@@ -126,7 +126,7 @@ export function interfaceVariablesDeclarationForOperation(
   interfaceDeclaration(generator, {
     interfaceName,
   }, () => {
-    const properties = propertiesFromFields(generator.context, variables);
+    const properties = propertiesFromFields(generator.context, variables, false, true);
     propertyDeclarations(generator, properties, true);
   });
 }
@@ -179,17 +179,17 @@ export function interfaceDeclarationForFragment(
   });
 }
 
-export function propertiesFromFields(context, fields, forceNullable) {
-  return fields.map(field => propertyFromField(context, field, forceNullable));
+export function propertiesFromFields(context, fields, forceNullable, isVariable) {
+  return fields.map(field => propertyFromField(context, field, forceNullable, isVariable));
 }
 
-export function propertyFromField(context, field, forceNullable) {
+export function propertyFromField(context, field, forceNullable, isVariable) {
   let { name: fieldName, type: fieldType, description, fragmentSpreads, inlineFragments } = field;
   fieldName = fieldName || field.responseName;
 
   const propertyName = fieldName;
 
-  let property = { fieldName, fieldType, propertyName, description };
+  let property = { fieldName, fieldType, propertyName, description, isVariable };
 
   const namedType = getNamedType(fieldType);
 
@@ -212,8 +212,9 @@ export function propertyFromField(context, field, forceNullable) {
       isArray, isNullable,
     };
   } else {
+    const isNullable = !(fieldType instanceof GraphQLNonNull)
     const typeName = typeNameFromGraphQLType(context, fieldType);
-    return { ...property, typeName, isComposite: false, fieldType };
+    return { ...property, typeName, isNullable, isComposite: false, fieldType };
   }
 }
 
diff --git a/src/typescript/language.js b/src/typescript/language.js
index 0ab5440..53e3a61 100644
--- a/src/typescript/language.js
+++ b/src/typescript/language.js
@@ -15,10 +15,11 @@ export function interfaceDeclaration(generator, { interfaceName, extendTypes },
   generator.popScope();
 }
 
-export function propertyDeclaration(generator, { propertyName, typeName, description, isArray, isNullable, inInterface, fragmentSpreads }, closure) {
+export function propertyDeclaration(generator, { propertyName, typeName, description, isArray, isNullable, isVariable, inInterface, fragmentSpreads }, closure) {
   generator.printOnNewline(description && `// ${description}`);
+  const optional = isVariable && isNullable ? '?' : ''
   if (closure) {
-    generator.printOnNewline(`${propertyName}:`);
+    generator.printOnNewline(`${propertyName}${optional}:`);
     if (isArray) {
       generator.print(' Array<');
     }
@@ -37,7 +38,7 @@ export function propertyDeclaration(generator, { propertyName, typeName, descrip
   } else if (fragmentSpreads && fragmentSpreads.length > 0) {
     generator.printOnNewline(`${propertyName}: ${isArray ? 'Array<' : ''}${fragmentSpreads.map(n => `${n}Fragment`).join(' & ')}${isArray ? '>' : ''}`);
   } else {
-    generator.printOnNewline(`${propertyName}: ${typeName}`);
+    generator.printOnNewline(`${propertyName}${optional}: ${typeName}`);
   }
   generator.print(inInterface ? ';' : ',');
 }

I just update my apollo-ios from 0.3.1 to 0.4.2 and apollo-codegen to 0.9.6.

I have 19 *.graphql files with approximate 3k lines. It's quite good on old version, but they have some issue so I have to update.

Then now, I get this error:

apollo-codegen generate **/*.graphql --schema schema.json --output API.swift
RangeError: Maximum call stack size exceeded
at _loop (/usr/local/lib/node_modules/apollo-codegen/src/swift/codeGeneration.js:443:22)
at /usr/local/lib/node_modules/apollo-codegen/src/swift/codeGeneration.js:442:40
at CodeGenerator.withIndent (/usr/local/lib/node_modules/apollo-codegen/src/utilities/CodeGenerator.js:63:5)
at CodeGenerator.withinBlock (/usr/local/lib/node_modules/apollo-codegen/src/utilities/CodeGenerator.js:69:10)
at structDeclaration (/usr/local/lib/node_modules/apollo-codegen/src/swift/language.js:23:13)
at structDeclarationForInputObjectType (/usr/local/lib/node_modules/apollo-codegen/src/swift/codeGeneration.js:437:3)
at typeDeclarationForGraphQLType (/usr/local/lib/node_modules/apollo-codegen/src/swift/codeGeneration.js:412:5)
at /usr/local/lib/node_modules/apollo-codegen/src/swift/codeGeneration.js:48:5
at Array.forEach (native)
at generateSource (/usr/local/lib/node_modules/apollo-codegen/src/swift/codeGeneration.js:47:21)

Did anyone get this isssue and solved it ?

P/s: I'm not english speaking, so sorry for my bad english