S3 upload custom endpoint url

[peppelauro]

Description

Allow to use also a non-Amazon S3 provider like OVH, DigitalOcean and so on

Contributor Checklist

Make sure to do the following steps if they are applicable to your PR:

• Update e2e tests: If you changed the /template/app, then make sure to do any neccessary updates to /template/e2e-tests also.
• Update demo app: If you changed the /template/app, then make sure to do any neccessary updates to /opensaas-sh/app_diff also. Check /opensaas-sh/README.md for details.
• Update docs: If needed, update the /opensaas-sh/blog/src/content/docs.

[vincanger]

Thanks for the contribution here @peppelauro. What happens if the user wants to use Amazon S3 and not a custom endpoint or 3rd-party provider? Could this possible be confusing for them? Shouldn't we add some comments in the files or docs to explain what's going on? What do you think?

[peppelauro]

@vincanger good observation, what if we check the existence of the env AWS_S3_ENDPOINT_URL?

If it doesn't exists then use S3Client as usual (amazon default, no endpoint required), else use it in the client, something like follow:

let conf = {
  region: process.env.AWS_S3_REGION,
  credentials: {
    accessKeyId: process.env.AWS_S3_IAM_ACCESS_KEY!,
    secretAccessKey: process.env.AWS_S3_IAM_SECRET_KEY!,
  },
}

if ("AWS_S3_ENDPOINT_URL" in process.env) {
    conf.endpoint = process.env.AWS_S3_ENDPOINT_URL,
}

const s3Client = new S3Client(conf);

[peppelauro]

And of course, in case you validate the previous snippet, we will add a commented line in the .env.server:

`

(OPTIONAL) get your aws s3 credentials at https://console.aws.amazon.com and create a new IAM user with S3 access

#If you want to use a non-amazon S3 decomment the following line:
#AWS_S3_ENDPOINT_URL=https://...
AWS_S3_IAM_ACCESS_KEY=ACK...
AWS_S3_IAM_SECRET_KEY=t+33a...
AWS_S3_FILES_BUCKET=your-bucket-name
AWS_S3_REGION=your-region
`

[vincanger]

@peppelauro on second thought, I think it's too confusing to add the conditional code and optional env variable to the template. I think it's best if we don't clutter it too much. We can only offer so many integrations and examples for the user, so if the user wants to use something else besides AWS S3, that's up to them to figure out how to integrate it.

With that said, what do you think about adding a section to the "File Upload" docs, e.g. "Using a non-Amazon S3 provider (OVH, DigitalOcean)" that includes this information and code snippet?

[peppelauro]

@peppelauro on second thought, I think it's too confusing to add the conditional code and optional env variable to the template. I think it's best if we don't clutter it too much. We can only offer so many integrations and examples for the user, so if the user wants to use something else besides AWS S3, that's up to them to figure out how to integrate it.

With that said, what do you think about adding a section to the "File Upload" docs, e.g. "Using a non-Amazon S3 provider (OVH, DigitalOcean)" that includes this information and code snippet?

I understand but the user will have to modify the source code to enable that feature is it?

No problem, I will specify with an "how to" into the doc :)