AWS Supports node.js v14!!!! https://aws.amazon.com/about-aws/whats-new/2021/02/aws-lambda-now-supports-node-js-14/
Voxel Group has just announced a port of middy to the .net runtime: https://twitter.com/vgaltes/status/1366371605337825284
Middy passed 100k downloads per week. Crazyness. https://www.npmtrends.com/@middy/core-vs-middy
Oh yeah, v2 beta is now releaseed :)
Oh yeah, v2 beta is now releaseed :)
really cool news, a good example here: https://eoins.medium.com/using-s3-object-lambdas-to-generate-and-transform-on-the-fly-874b0f27fb84
I was wondering how they allowed "streaming" file changes since lambda doesn't support streamed responses. Not a fan of the pattern their using, I think we could make this much better. I was hoping they would pass in a read stream and we could return a transform stream. I'll for sure use this feature once terraform supports it. I see a lot of value in having a middleware for this to make it way easier.
I had the exact same thought and I was a bit disappointed too :D However, from what i have seen in the api to create a custom lambda executor, I think doing a streaming implementation for Node.js could be feasible. I had a look sometime ago but if I remember correclty in that abstraction you have access to raw http requests, therefore you should be able to access streaming requests and provide streaming responses...
and see if my instincts are actually correct :D
@/all Looking for some feedback. I'm looking into adding in support for parsing nested aws events, ex: s3 -> sns -> sqs. In this example would you prefer using; a) single middlware for the source event (s3 in this case) that parses the wrapper events (sns/sqs) as well; b) have a parser for each event, and each would need to be applied in the correct order of use; c) single middleware for any event with wrapper event parsing.
They all have their pros/cons. Let me know what you think, which would you prefer to use?
They all have their pros/cons. Let me know what you think, which would you prefer to use?
Guys , i have a very simple payload via api-0gateway to lambda
{"email":"abc@gmail.com"}. and my Schema looks like this .const inputRequestSchema = {
type: 'object',
required: [
'email',
],
properties: {
email: {
type: 'string',
},
},
additionalProperties: true,
}; when i fire up the request , i get error message: 'must have required property email'les/@middy/core/index.js?:80:5) {ex.js?:51:21) Event object failed validation. any help here ? i am using following dependency "@middy/core": "^2.2.0",
"@middy/http-error-handler": "^2.2.0",
"@middy/http-header-normalizer": "^2.2.0",
"@middy/http-json-body-parser": "^2.2.0",
"@middy/validator": "^2.2.0". . use of middleware code \ export const handler = middy(handlerLambda)
.use(httpHeaderNormalizer())
.use(httpJsonBodyParser())
.use(validator({ inputSchema: inputRequestSchema }))
.use(httpErrorHandler());
@khirodAsurion your input schema doesn't match what API Gateway sends in. Should look something like:
{ "queryStringParameters":{...}, "body":{...} }. https://docs.aws.amazon.com/lambda/latest/dg/services-apigateway.html
@willfarrell thanks for the quick response. but I am only interested to validating body . in that case should I always have to provide other details also in schema ? or is there any other way ? See that I am adding
additionalProperties: true, just to make sure other added by APIGateway do not create issue .but still it says message: 'must have required property email'les/@middy/core/index.js?:80:5) {ex.js?:51:21) Event object failed validation
If you want to return errors, you need to handler ajv errors yourself. There isn't a standard everyone follows so we can't do it for you.
httpErrorHandler is a catch all if an error hasn't already been handled. It returns a generic message (which can be overridden), because you don't want stack traces returned to the consumer, leading to leaking too much information (security best practice).
Has anyone here built an API with middy? If so, I'd love to take a look at it (assuming it's in a public repo)
I just answered the questions on StackOverflow - https://stackoverflow.com/questions/69766960/how-to-use-middy-with-swagger/69956826
Hi, I have a question regarding the sqs-partial-batch-failure package. When I use it the lambda functions is not processed as fast as when I don't use it. Here's a sample of the sqs queue.
The moment I switched to not using the sqs-partial batch the queue processes super fast. I'm not sure what is causing that. The processEvent is the same it just changes the handler function
module.exports.handler = async ({ Records }) => {
const failedMessageIds = []
let batch = {}
for (const record of Records) {
try {
console.log(`Processing ${record.messageId}`)
const bodyData = JSON.parse(record.body)
await processEvent(bodyData, batch)
} catch (e) {
console.log(e)
failedMessageIds.push(record.messageId)
}
}
await table.batchWrite(batch)
await flushQueue()
return {
batchItemFailures: failedMessageIds.map(id => ({ itemIdentifier: id }))
}
}That's what I'm using now and goes fast.
const middy = require("@middy/core")
const sqsBatch = require("@middy/sqs-partial-batch-failure")
const sqsJsonBodyParser = require("@middy/sqs-json-body-parser")
const originalHandler = ({ Records }) => {
const recordPromises = Records.map(async ({ body }) => processEvent(body))
return Promise.allSettled(recordPromises)
}
module.exports = {
handler: middy(originalHandler).use([sqsJsonBodyParser(), sqsBatch()])
}This is the slow code.
Any ideas?
Maybe I should post it on the issues aswell
AWS has released some of their own Middy middlewares! https://github.com/awslabs/aws-lambda-powertools-typescript
hey folks, quick question.
I want to create a logger in a middleware that always logs the context information and then pass that logger to the middleware. Is my only option to attach the logger to the event or context?
Something like this:
function attachLoggerMiddleware(){
const attachLoggerMiddlewareBefore = (
request
) => {
request.event.logger = new Logger(request.context);
};
return {
before: attachLoggerMiddlewareBefore,
};
}And then in the handler I can do:
function (event) {
event.logger.info('yo');
}
2 replies
@nem035_gitlab Attaching to the context is the recommended way for passing things from middleware to the handler. However you could initialize in the global scope, then inject the context during a before middleware. Have you seen: https://awslabs.github.io/aws-lambda-powertools-typescript/latest/core/logger/
14 replies
A new redesigned website has just been publishet at https://middy.js.org .
We'd really love your feedback on the new design, documentation and all over the new DX.
HI all,
I'm having an issue using middyjs, I'll glad if anyone here can help.
Is there a particular format one must follow to make use of the onError function? I've caught the error and passed it into the onError, but I've not been able to return this error back to the handler. It just keeps throwing a 502. The success case works perfectly.
const onError = async (request: any): Promise<any> => {
console.log(request.error)
if (request.error.statusCode == 403) return forbiddenError('Unauthorized Request');
// throw createHttpError(401, 'Unauthorized Request')
return;
};
const before = async (request: any): Promise<any> => {
const integratorPk = '';
if (integratorPk) {
console.log('Success');
return;
}
return forbiddenError('Unauthorized Request');
};
export function domainValidationMiddleWare(): MiddlewareObject<
APIGatewayEvent,
APIGatewayProxyResult
> {
return {before, after, onError
};
}
The
onError function successfully logs request.error, but throws a 502 instead of returning 403
Created an issue here also. Thanking you in advance.