DEV Community

loading...
Cover image for Direct Lambda Resolvers with AWS Amplify and AppSync
AWS Community Builders

Direct Lambda Resolvers with AWS Amplify and AppSync

andthensumm profile image Matt Marks 🐣 ・Updated on ・4 min read

How I feel about VTL on any given day

woman raising fist saying "curse youuuu!"

Haha okay, okay. I'm just kiddin' (mostly).

AWS recently announced Direct Lambda Resolvers support, which means you can write AppSync resolvers in your favorite Lambda runtimes without relying on the Velocity Template Language.

I wanted to take advantage of this, as I had some complicated logic that that I didn't know how to write with VTL and it was much easier to do with javascript. I'll share with you how I added Direct Lambda Resolvers to an Amplify project with a little magic from the @function directive.

I'm gonna assume you've already created a GraphQL API for
your Amplify project. If you haven't, follow the steps here on how to Create a GraphQL API

Step 1: Create a Lambda with Amplify cli

amplify add function
? Provide a friendly name for your resource to be used as a label for this category in the project: DirectLambdaResolver
? Provide the AWS Lambda function name: DirectLambdaResolver
? Choose the function runtime that you want to use: NodeJS
? Choose the function template that you want to use: Hello World
? Do you want to access other resources created in this project from your Lambda function? No
? Do you want to invoke this function on a recurring schedule? No
? Do you want to edit the local lambda function now? Yes

Go to amplify/backend/function/DirectLambdaResolver/src/index.js and update it with the following:

let response;

exports.handler = async (event, context) => {
    console.log({event, context})
    try {
        response = "this lambda is super direct"
    } catch (err) {
        console.log({err})
        console.log(err);
        return err;
    }

    return response
};

Alright, save the file and press enter in the terminal to continue.

Step 2: Auto-generate code with @function and copy code to CustomResources.json

Now we're gonna go to amplify/backend/api/[your-api-name]/schema.graphql and take advantage of the @function directive

### schema.graphql
type Todo @model {
  id: ID
  task: String @function(name: "DirectLambdaResolver-${env}")
}

then

$: amplify api gql-compile

This will create all the files inside amplify/backend/api/[your-api-name]/build/. What we're looking for is the file FunctionDirectiveStack.json. Open it up and you'll see the following:

{
    "AWSTemplateFormatVersion": "2010-09-09",
    "Description": "An auto-generated nested stack.",
    "Metadata": {},
    "Parameters": {
      ...
    },
    "Resources": {
      // IAM Role
      "DirectLambdaResolverLambdaDataSourceRole": {
         ...
      },
      // DataSource
      "DirectLambdaResolverLambdaDataSource": {
         ...
      },
      // Direct Lambda Resolver
      "InvokeDirectLambdaResolverLambdaDataSource": {
         ...
      },
      // Pipeline Resolver
      "TodotaskResolver": {
         ...
      }
    }

When amplify auto-generates with @function, it creates a Lambda DataSource, an IAM Role for that DataSource, the Direct Lambda Resolver and a Pipeline Resolver.

We're going to copy everything from the Resources object and paste it in our CustomResources.json file located at amplify/backend/api/[your-api-name]/stacks/. It should now look like the following:

{
  "AWSTemplateFormatVersion": "2010-09-09",
  "Description": "An auto-generated nested stack.",
  "Metadata": {},
  "Parameters": {
     ...
  },
  "Resources": 
    "DirectLambdaResolverLambdaDataSourceRole": {
      ...
    },
    "DirectLambdaResolverLambdaDataSource": {
      ...
    },
    "InvokeDirectLambdaResolverLambdaDataSource": {
      ...
    },
    "TodotaskResolver": {
      ...
    }
  },
  "Conditions": {
    ...
  },
  "Outputs": {
    ...
  }
}

We also need to update our DirectLambdaResolverLambdaDataSourceRole to replace GetAttGraphQLAPIApiId with AppSyncApiId

"DirectLambdaResolverLambdaDataSourceRole": {
      "Type": "AWS::IAM::Role",
      "Properties": {
        "RoleName": {
          "Fn::If": [
            "HasEnvironmentParameter",
            {
              "Fn::Join": [
                "-",
                [
                  "DirectLambdaResolver774c",
                  {
                    "Ref": "AppSyncApiId" // Used to be GetAttGraphQLAPIApiId
                  },
                  {
                    "Ref": "env"
                  }
                ]
              ]
            },
            {
              "Fn::Join": [
                "-",
                [
                  "DirectLambdaResolver774c",
                  {
                     "Ref": "AppSyncApiId" // Used to be GetAttGraphQLAPIApiId
                  }
                ]
              ]
            }
          ]
        },
        "AssumeRolePolicyDocument": {
         ...
        },
        "Policies": [
          ...
        ]
      }
    },

Next up, in order to make our new Lambda a Direct Lambda Resolver with AppSync, we'll need to remove the request and response mapping templates from our function InvokeDirectLambdaResolverLambdaDataSource. They're nested inside the Properties object

"InvokeDirectLambdaResolverLambdaDataSource": {
  "Type": "AWS::AppSync::FunctionConfiguration",
  "Properties": {
    "ApiId": {
        "Ref": "AppSyncApiId"
    },
    "Name": "InvokeDirectLambdaResolverLambdaDataSource",
    "DataSourceName": "DirectLambdaResolverLambdaDataSource",
    "FunctionVersion": "2018-05-29",
    // REMOVE ME 
    "RequestMappingTemplateS3Location": {
      ...
    },
    // REMOVE ME 
    "ResponseMappingTemplateS3Location": {
      ...
    }
  },

  "DependsOn": "DirectLambdaResolverLambdaDataSource"
}

We'll also need to copy Todo.task.req.vtl and Todo.task.res.vtl from amplify/backend/api/[your-api-name]/build/resolvers to amplify/backend/api/[your-api-name]/resolvers.

Wait a minute. I thought we said no more VTL? What gives?
TodotaskResolver is a Pipeline Resolver. InvokeDirectLambdaResolverLambdaDataSource is our Direct Lambda Resolver.

Don't wan't a Pipeline resolver? Don't worry, I got you covered towards the end πŸ˜‰

Step 3: Remove autogenerated code

Remember how we added @function(name: "DirectLambdaResolver-${env} directive in our task: String field within our schema.graphql? We need to remove all the generated code, including FunctionDirectiveStack.json.

Luckily that's pretty easy. Just remove the @function and run:

$: amplify api gql-compile

If you look inside amplify/backend/api/[your-api-name]/build/, you'll see our updated CustomResources.json file and that the FunctionDirectiveStack.json has been removed.

Puh-push it real good

$: amplify push

If you go to your AppSync console, click on schema and search Todo inside the Resolvers search field, and scroll down to task and it should say Pipeline. Click on it and it will take you to the Todo.task Pipeline Resolver with your DirectLambdaResolver listed. Notice the mapping templates disabled?

Alt Text

Yeaaa buddy, no more VTL (kind of) πŸ˜† πŸ˜† πŸ˜†

Bonus Round: Remove Pipeline Resolver

As I promised earlier, if you want to get rid of the Todo.task Pipeline Resolver, we need to make two changes inside amplify/backend/api/[your-api-name]/stacks/CustomResources.json.

First, delete TodotaskResolver.
Second, replace your InvokeDirectLambdaResolverLambdaDataSource with the following:

"TodotaskDirectLambdaResolver": {
  "Type": "AWS::AppSync::Resolver",
  "Properties": {
    "ApiId": {
        "Ref": "AppSyncApiId"
    },
    "DataSourceName": "DirectLambdaResolverLambdaDataSource",
    "TypeName": "Todo",
    "FieldName": "task"
  },

  "DependsOn": "DirectLambdaResolverLambdaDataSource"
}
$: amplify api push

Now, go forth and banish VTL templates at will!

I hope you enjoyed this post and if you did, follow me on twitter for all things AWS Amplify, AppSync, and React-Native πŸ™Œ πŸ™Œ πŸ™Œ
https://twitter.com/andthensumm

Resources:
Add a custom resolver that targets an aws lambda
How to add pipelineResolvers - Github Issue

Discussion (15)

pic
Editor guide
Collapse
swyx profile image
swyx
Syntax Error: Expected Name, found @

GraphQL request:4:9
3 |   id: ID
4 |   task: @function(name: "DirectLambdaResolver-${env}")
  |         ^
5 | }
Enter fullscreen mode Exit fullscreen mode
Collapse
andthensumm profile image
Matt Marks 🐣 Author

Thank you Shawn for catching this, it's a treat just to have you actually use this code!

Collapse
swyx profile image
swyx

not at all lol, i barely know what i'm doing and was following along just to try to learn about this new feature πŸ˜…

i think the struggle i had was how to use direct lambda resolvers in mutations not just queries. i couldnt figure out how to do it. maybe worth a followup post?

Thread Thread
andthensumm profile image
Matt Marks 🐣 Author

Pfffh, you got it. I'll whip something up for this week.

Thread Thread
nxtra profile image
Nxtra

Did you guys manage to figure this out?

Thread Thread
andthensumm profile image
Matt Marks 🐣 Author

@nxtra , in order to do this for a mutation, you'd set the typeName: Mutation and fieldName: yourMutationName. Query, Mutation and Subscription are really just a Type like Todo.

The one gotcha is that you can't create a resolver on a mutation that already has one. Example being createToDo. Amplify has already created a resolver for it inside CloudFormation. You'll need to either to add @model(mutations: null) so you can override it or create a custom mutation. I prefer the custom mutation because then I still have access to the autogenerated vtl from Amplify. There are times where the authorization rules that are generated in the response.vtl can still be useful.

  "MutationyourMutationNameLambdaResolver": {
  "Type": "AWS::AppSync::Resolver",
  "Properties": {
    "ApiId": {
        "Ref": "AppSyncApiId"
    },
    "DataSourceName": "DirectLambdaResolverLambdaDataSource",
    "TypeName": "Mutation",
    "FieldName": "yourMutationName"
  },

  "DependsOn": "DirectLambdaResolverLambdaDataSource"
}
Enter fullscreen mode Exit fullscreen mode
Thread Thread
nxtra profile image
Nxtra

Thanks for the info! I didn't realize yet that indeed a resolver with that name already exists if you name it like that.
Until now I've been making custom Resolvers with lambda functions as pipeline resolvers. I'll try to convert one to a "no-more-vtl" version using your guide.

Collapse
michaeldbrant profile image
Michael Brant

I'm confused, I thought the whole point of the Amplify cli was so I wasn't modifying cft's directly?

Collapse
andthensumm profile image
Matt Marks 🐣 Author

I'd also say the intent of Amplify is to offer you a core set of services that includes authentication, database, file storage, analytics, managed hosting, deployments, with more and counting.

The services are supposed to make it more accessible and approachable for backend and front-end devs, which on the whole, they absolutely do. They're still growing and figuring things out though and in the meantime, they've given the option of customizing your CFTs when you need to do something off the beaten bath.

Collapse
andthensumm profile image
Matt Marks 🐣 Author

It's almost inevitable that you will modify CFTs in the future. Amplify offers a lot out of the box but there are several solutions that they don't offer from the cli.

Collapse
radams0x profile image
Rich Adams

I followed the above, but now the lambda is no longer invoked and the field that previously had the @function now comes back with a null value; any suggestions as to what may be wrong? Thanks!!

Collapse
andthensumm profile image
Matt Marks 🐣 Author

Hey Rich, I’d have to know more about your project and see your CustomResources.json file

Collapse
nxtra profile image
Nxtra

Is there any risk in breaking this when you regenerate code or add other resources with Amplify? Or once you've done this it's stable?

Collapse
andthensumm profile image
Matt Marks 🐣 Author

As long as you keep the code in the CustomResources file (or another custom stack) and you make sure to remove the @directive, you’re good.

These days I mostly use amplify for its api/auth/appsync integration and the cli for generating as much code as possible because I’m lazy ;)

Collapse
nxtra profile image
Nxtra

Awesome, thanks!