Plugin server action timed out

Need help Plugins
1

Hello,
I am creating an open source plugin where i use aws textextract to transform img scan to text,
when i trying to use it the plugin is taking a few secunds and that’s making the app faild, and also i can’t get the text of the scan images.
how i could make that the app will wait until it’s get the text ?

1 Like
2

Well, you’ve only got 30 seconds. That’s hard coded.

You probably are looking to use the context.async function provided by the plugin builder. Have you explored that context yet?

3

thats what i am using, i do an await to get the text then i return the text.
the code dosent take 30 secunds, far maybe 5 not more, so i don’t know why it’s trowing me this error.
Here the code
function(properties, context) {
const aws = require(‘aws-sdk’)
aws.config.update({
accessKeyId: context.accessKeyId,
secretAccessKey: context.secretAccessKey,
region: context.region
});
const extract = new aws.Textract()

	var options = {
    uri: "https:"+properties.image_to_text,
    method: "GET",
    encoding: null,
    headers: { "Content-type": "application/octet-stream" }
	};

    const buffer = context.request(options).body;
    
    const text = context.async(async callback => {
        try {
      		const parse = await extract.detectDocumentText({ Document: { Bytes: buffer } }).promise()
    		return parse.Blocks.filter(b => b.BlockType === 'LINE').map(w => w.Text).join('\n')
       } catch (error) {
			console.log('error with Textract', error)
       }
    })

	return {imageText: text}
}
4

Instead of return

Try wrapping your return like this callback(<VALUEtoRETURN>) in the callback()

1 Like
5

Like this ? i tried i still have the same problem
const buffer = context.request(options).body;

    context.async(async callback => {
        try {
      		const parse = await extract.detectDocumentText({ Document: { Bytes: buffer } }).promise()
    		const text = await parse.Blocks.filter(b => b.BlockType === 'LINE').map(w => w.Text).join('\n')
            console.log('ouech text',text)
            return {imageText: text}
       } catch (error) {
			console.log('error with Textract', error)
       }
    })
6

Based on the AWS documentation I think the preferred architecture is to use three server-side actions:

  1. The first action posts the job to AWS and returns the job identifier.
  2. The second action queries the job status.
  3. The third action retrieves the completed job results.

The design pattern is then:

  1. Post the job using the first action
  2. Schedule a recursive workflow to run every “N” many seconds.
  3. This workflow queries the job status with the second action.
  4. Once success or fail returns end the workflow.
  5. On success retrieve the results with the third action.

Maybe AWS can accept a web-hook? In which case you could build a Bubble endpoint to conform to the AWS web-hook specification.