Add logging to your GraphQL resolvers so you know what's going on in your app.

Everytime a resolver is executed it logs the full path to the resolver. The output in your terminal (when using the popular debug module for logging) will be something like this:

server:resolvers Message.channel +0ms
server:resolvers User.avatar +0ms
server:resolvers User.username +0ms
server:resolvers User.avatar +4ms
server:resolvers User.coverPhoto +0ms
server:resolvers User.username +0ms

Note: The "server:resolvers" and "+xms" parts are added by debug, this module only logs the path on execution.

This is the simplest example:

// Require the module
const createGraphQLLogger = require('graphql-log');

// Create a logger
const logExecutions = createGraphQLLogger();

// Wrap your resolvers
logExecutions(resolvers);

Note that your resolvers need to be a plain object of resolvers. (like you'd create for graphql-tools makeExecutableSchema function)

graphql-log has some options:

• logger (function, default: console.log): Use a custom logger like debug, pino or any other one.
• prefix (string): Prefix all logs with a certain string

const logExecutions = createGraphQLLogger({
  // Prefix all logs with resolvers.
  prefix: 'resolvers.',
});

Let's say you want to use the popular debug module for logging. Doing so would be as easy as passing it into the logger option:

const debug = require('debug')('server:resolvers');

const logExecutions = createGraphQLLogger({
  logger: debug,
});

Logging every execution of every resolver will have a performance impact on your app, so we recommend only enabling this module in development. A common way to do so (depending on your setup) would be something like this:

const createGraphQLLogger = require('graphql-log');

const resolvers = {/*...*/};

if (process.env.NODE_ENV === 'development') {
  const logExecutions = createGraphQLLogger();
  logExecutions(resolvers);
}

The main thing I want to change is that instead of wrapping the resolvers this module would wrap a finished schema instead. This would make it compatible not only with folks who keep their resolvers separate, but with everybody.

That'd probably look something like this:

// I'd love to be able to pass a schema
logExecutions(schema);

Good first PR?

Licensed under the MIT License, Copyright ©️ 2017 Maximilian Stoiber. See LICENSE.md for more information.