Run basic GraphQL server with GraphiQL support

Run a basic GraphQL server using plumber. This server is provided to show basic interaction with GraphQL. The server will run until the function execution is canceled.

Usage

server(
  schema,
  port = 8000L,
  ...,
  graphiql = interactive(),
  log = TRUE,
  initial_value = NULL
)

Arguments

schema: Schema object to use execute requests
port: web port to serve the server from. Set port to NULL to not run the plumber server and return it.
...: ignored for paramter expansion
graphiql: logical to determine if the GraphiQL interface should be enabled. By default, this route is only available when running the server interactively.
log: boolean that determines if server logging is done. Defaults to TRUE
initial_value: default value to use in execute_request()

Details

To view the GraphiQL user interface, navigate to the URL provided when the server is started. The default location is http://localhost:8000/graphiql/. By default, this route is only available when running the server interactively (graphiql = rlang::is_interactive()).

server() implements the basic necessities described in https://graphql.org/learn/serving-over-http/. There are four routes implemented:

'/'

GET. If run interactively, forwards to /graphiql for user interaction with the GraphQL server. This route is diabled if graphiql = rlang::is_interactive() is not TRUE.

'/graphiql/'

GET. Returns a GraphiQL formatted schema definition interface to manually interact with the GraphQL server. By default this route is disabled if graphiql = rlang::is_interactive() is not TRUE.

'/graphql'

GET. Executes a query. The parameter 'query' (which contains a GraphQL formatted query string) must be included. Optional parameters include: 'variables' a JSON string containing a dictionary of variables (defaults to an empty named list), 'operationName' name of the particular query operation to execute (defaults to NULL), and 'pretty' boolean to determine if the response should be compact (FALSE, default) or expanded (TRUE)

'/graphql'

POST. Executes a query. Must provide Content-Type of either 'application/json' or 'application/graphql'.

If 'application/json' is provided, a named JSON list containing 'query', 'operationName' (optional, default = NULL), 'variables' (optional, default = list()) and 'pretty' (optional, default = TRUE). The information will used just the same as the GET-'/graphql' route.

If 'application/graphql' is provided, the POST body will be interpreted as the query string. All other possible parameters will take on their default value.

Using bash's curl, we can ask the server questions:

 #R
  # load Star Wars schema from 'execute_request' example
  example(gqlr_schema)
  # run server
  server(star_wars_schema, port = 8000)

 #bash
  # GET Schema definition
  curl '127.0.0.1:8000/'

  ## POST for R2-D2 and his friends' names
  # defaults to parse as JSON
  curl --data '{"query":"{hero{name, friends { name }}}", "pretty": true}' '127.0.0.1:8000/graphql'
  # send json header
  curl --data '{"query":"{hero{name, friends { name }}}"}' '127.0.0.1:8000/graphql' --header "Content-Type:application/json"
  # send graphql header
  curl --data '{hero{name, friends { name }}}' '127.0.0.1:8000/graphql' --header "Content-Type:application/graphql"
  # use variables
  curl --data '{"query":"query Droid($someId: String!) {droid(id: $someId) {name, friends { name }}}", "variables": {"someId": "2001"}}' '127.0.0.1:8000/graphql'

  # GET R2-D2 and his friends' names
  curl '127.0.0.1:8000/graphql?query=
  # ... using a variable
  curl '127.0.0.1:8000/graphql?query=query