This is a huge refactoring based on an idea to model the schema with a proper normalized representation. The results so far are promising in terms of code clarity and maintainability. At the moment, this is far from complete.

Previously, name resolution, query validation and code generation all happened in one pass. In this PR, these three are cleanly separated, leading to much more understandable code in my opinion.

So, this is the most naïve implementation I could fathom. Current problems:

1. to_snake_case won't take care of numbers. In the existing example, Query1 is snake-cased as query1 but the file name was query_1. I suppose this is not a big issue.
2. implicitly_discover_schema_name is extremely verbose, I suppose you can come up with some nicer way of writing this.
3. This assumes there's src/. There should probably be another approach, and I'd like to store those magic strings somewhere nicer.
4. I'm probably using some archaic way of manipulating strings.

https://rust-lang-nursery.github.io/edition-guide/editions/transitioning-an-existing-project-to-a-new-edition.html#writing-idiomatic-code-in-a-new-edition

after merge these PR, #194, #191.

I've built a Juniper-based GraphQL server this week and now I'd like to execute tests against it. This project seems like the best way to do so from Rust, and so I'm trying to generate a schema for my server. The built-in GraphiQL instance can read the introspection data just fine, but when I run graphql-client introspect-schema http://localhost:3000/graphql I just get {"data":{"schema":null},"errors":null}.

(Just to be sure, I get the same result if I run introspect-schema against the GitHub GraphQL endpoint.)

What, if anything, am I getting wrong?

I have a schema generated by a hasura instance that is causing the rust compiler to panic.

thread 'rustc' has overflowed its stack

Schema and query causing the issue: rustc_panic.zip

I had some code working successfully with a previous schema, but I added some tables and now I'm getting a stack overflow.

I'm wondering if there's some recursive relationships in my schema that are causing issues, or just the size of the schema itself (hasura generates a ton of schema)... mine is currently 732kb of JSON

For example if we have this:

query SelectName { person { name } }

query SelectAge { person { age } }

And our code looks like this:

query_path = "..", schema_path = "...", )] struct SomeQuery;

then the operation we generate code for will default to the first one, and operationName should be SelectName.

It would be useful for people who want to generate the code in advance to get better editor support (RLS for example does not take into account generated code from derives yet).

Linking proc_macro2 could be a challenge, from my past experience.

Currently, the code generation doesn't allow for recursive types and compilation will fail with an error.

Eg:

type Node {
  left: Node
  right: Node
}

This should be supportable relatively easily by detecting the recursion and boxing the type where necessary (serde supports Box<> ed types).

At the moment the CLI generates one module for one file. The general use-case however is probably projects with more than one query, and having to issue more than one command seems a bit too complicated.

The most use similar CLI tool at the moment is apollo-cli and I think we could take a similar approach.

Invoking graphql-client-cli could look something like this:

graphql-client-cli generate --schema-path=./schema.json --deprecations=warn 'src/**/*.graphql', and it would generate one file per .graphql file, with one module per query.

The shared code could live in one module for the whole project - we can achieve a lot more reuse with this approach than with the derive-based interface.

Let's discuss what interface we want for the CLI in this isue.

This was introduced in #350. Besides this, my thanks for upgrading reqwest. The old version cost me some time yesterday.

I do not know why CI did not catch this. Maybe something for the maintainers to take a look at. Or I made a mistake in my setup.

Below is the error I got:

---- /Users/user/.cargo/registry/src/github.com-1ecc6299db9ec823/doc-comment-0.3.3/src/lib.rs - (line 228) stdout ----
error[E0433]: failed to resolve: use of undeclared type or module `reqwest`
  --> /Users/user/.cargo/registry/src/github.com-1ecc6299db9ec823/doc-comment-0.3.3/src/lib.rs:245:27
   |
20 |     let client = reqwest::Client::new();
   |                           ^^^^^^ not found in `reqwest`
   |
help: consider importing this struct
   |
3  | use graphql_client::web::Client;
   |

error: aborting due to previous error

This probably isn't the final solution to this issue but I made this change so that I could continue with what I was implementing. At a minimum, the readme would need to be updated to describe the feature.

I have added the normalize_query_types that normalises types (structs, type aliases, etc) generated by the code-gen to use CamelCase as it is my understanding that this is idiomatic rust. I have done this as a feature as it's potentially a breaking change.

I was having issues with scalars as described in #254 but then noticed that other types were non-idiomatic.

Fixes #254 and fixes #255.

I have a type like so:

type AuthenticationOutput {
  access_token: String!
  admin: Boolean!
  refresh_token: String!
}

and I'd like to generate a Rust struct for it. The use case is that Hasura allows you to define web hooks, and defines the inputs / outputs of the web hooks as GraphQL types. Is there a way to generate the struct for this type? or does the type have to be attached to a query? Thanks!

Given an query

query HomeQuery {
  viewer {
    ...home_viewer
    id
  }
}

fragment home_viewer on Customer {
  id
}

and a response

{
  "data": {
    "viewer": {
      "id": "Q3VzdG9tZXI6MA=="
    }
  }
}

An error is is encountered.

response_body = Err(
    reqwest::Error {
        kind: Decode,
        source: Error("missing field `id`", line: 1, column: 43),
    },
)

If id is added along side with the the fragment, i.e.

    ...home_viewer
    id

somehow it cannot decode it. If id is removed then everything is okay.

I can reproduce the same problem in the github example by adding a fragment to the query

query RepoView($owner: String!, $name: String!) {
  repository(owner: $owner, name: $name) {
    ...home_viewer             
    id
    ...
  }
}

fragment home_viewer on Repository {
  id
}

This spec proposes an alternate request format to allow uploading files in graphql requests. It's supported by a lot of client/servers, e.g. async-graphql.

I've got as far as producing a schema.json file by sending an introspection query to my backend with graphql-client, but I want to produce a schema.graphql file; ~or anything that will allow me to traverse the GraphQL schema easily client-side~. Is there a simple way to do this? graphql_client_codegen seems to have the functionality but it's all private.

Basically my end objective is to generate a schema.graphql file and Typescript types from a GraphQL schema. I'm aware of graphql-codegen, I just want to make my own implementation.

(Edit: I'm a fool, I've realised IntrospectionResponse is good enough for traversing)

I'm trying to setup my queries and for the most part its being accepted by the rust compiler. However the Variables type is acting a bit weird. Basically I have to use the fully qualified syntax or I get an error "ambiguous associated type". So then I switch to the fully qualified syntax but then later in code the compiler will not allow me to instantiate an instance of the fully qualified type. What am I supposed to do?

` #[derive(GraphQLQuery)] #[graphql( query_path = "src/common/gql/transactions.graphql", schema_path = "src/common/gql/schema.json", response_derives = "Debug" )] pub struct TransactionsQuery; #[allow(unused)] pub async fn transactions(config: &Config, variables: ::Variables) -> Result<(), Box> { let request_body = TransactionsQuery::build_query(variables); let mut res = client.post(&config.arweave_gql_url) .json(&request_body) .send() .await?; let response_body: Response<::ResponseData> = res.json().await?; println!("{:#?}", response_body);

Ok(())

}

async fn get(&self, tx_id: &str) -> Result<(), Box> { let variables = ::Variables { // Fails here! ids: Some(vec!["r5_3_wqF4KGDzLnpRv70sI_Z1pE5JAl71-Bi9yJA5-8".to_string()]) }; transactions(&self.config, variables).await?; Ok(()) } `