api-discovery.md

phenocam notes

There will be a load¹ of images dumped into S3. We need a way of indexing and filtering them by site and date, generating pre-signed URLs and presenting them to the user.

Use metadata api? Not 100% clear what it can provide.

• Is it fast enough just to grab bucket contents as an API call, generate pre-signed URLs for each, and return them to the frontend?

Pre-signed URLs

https://docs.aws.amazon.com/AmazonS3/latest/userguide/using-presigned-url.html

Validity up to 7 days, though depends on the credential used to create it.

Autogenerating thumbnails/resizing images

Dom mentioned that it might be possible to have S3 automatically resize the images on the fly as they are requested. I can't see whether this is possible without using Lambda, as in this tutorial. Not sure how necessary this is really - depends on how big the images are and how many the client requests at once.

Sketch API design

/v1/phenocam/images?site=<site>&from=<from>&to=<to> returns

{
    "images": [
        {
            "site": "ALIC1",
            "date": "1970-01-01T00:00:00Z", // or 1739976978
            "url": "<pre-signed url>",
            "metadata": {
                "reviewed": false,
                // ...
            }
        },
        // ...
    ],
    // ...
}

Eventually, perhaps also /v1/phenocam/masks/<site>?from=<>&to=<> returns

{
    "masks": [
        {
            "site": "ALIC1",
            "dateFrom": "1970-01-01T00:00:00Z",
            "dateTo": "1970-01-01T00:00:00Z",
            "url": "<pre-signed url>",
            "metadata": {
                //...
            }
        },
        //...
    ],
    //...
}

Footnotes

1. something like 5 images per site per day ↩