S3 URL support for image-service and pipeline

[sat01a]

Time to time we get datasets with images that they are not on the web and need to be ingested into ALA.
The easy way can be: upload them into subfolder of the DR on a S3 bucket and put the paths in the DwCA files.
This may need a translation of s3:// to https:// or support of s3;// in image-service.

The other workaround (for limited number of images) can be, uploading them into the image-service first and then linking them back in the DwCA.

Needs more discussion to find the best solution.

[sadeghim]

Hi @sat01a do we have any update on this? The process of loading actual image files are tedious now and involves Database update and image-reindex. It will be very good if we can give it some priority.

[sbearcsiro]

@sadeghim You should already be able to address s3 objects via a HTTP URL. If you have the s3 client library available you should be able to use s3Client.getUrl(bucket, path) or the equivalent if it's a public object. For private objects you can supply presigned URLs, like so:

            GeneratePresignedUrlRequest generatePresignedUrlRequest =
                    new GeneratePresignedUrlRequest(bucket, path)
                            .withMethod(HttpMethod.GET)
                            .withExpiration(expiration)
            s3Client.generatePresignedUrl(generatePresignedUrlRequest)

[peggynewman]

@sbearcsiro We understand that we can't have public buckets/objects. I don't want to add temporary presigned URLs to objects in the data. It is messy and unmanageable. Is there another option?