image processing adaptor

[mtuchi]

Summary

New @openfn/language-image-utils adaptor providing image processing utilities for OpenFn job code.

Details

Using joe's implementation i was able to do the following

Five exported operations, each accepting a base64 string, data URL, or Buffer as input:
- resize(input, { width, height }) — resizes to target dimensions using jimp; outputs { buffer, width, height }
- compress(input, { maxBytes, minQuality }) — quality-loop compression to a byte target; outputs { buffer, size, quality }
- stripMetadata(input) — strips all EXIF metadata (including GPS) by re-encoding through Jimp; outputs { buffer }
- embedMetadata(input, { comment }) — embeds a string in the JPEG EXIF UserComment field; outputs { buffer }
- metadata(input) — reads image dimensions and orientation without modifying the image; outputs { width, height, orientation, size }

• All image-returning functions (resize, compress, strip, annotate) accept { parseAs: 'base64' } to return a base64 string instead of a Buffer, similar to parseAs convention in language-common
• Image processing functions use jimp package. The piexifjs package is used only for EXIF UserComment injection in annotate function

Here is sample job code

const url = 'https://tinyurl.com/rucpr6tk';

http.get(url, { parseAs: 'base64', maxRedirections: 1 });
resize($.data, { width: 1200, height: 1600 });
embedMetadata($.data.buffer, { UserComment: 'patient-id=42' });
compress($.data.buffer, { maxBytes: 700 * 1024 });

AI Usage

Please disclose how you've used AI in this work (it's cool, we just want to
know!):

• I have used Claude Code
• I have used another model
• I have not used AI

You can read more details in our
Responsible AI Policy

Review Checklist

Before merging, the reviewer should check the following items:

• Does the PR do what it claims to do?
• If this is a new adaptor, added the adaptor on marketing website ?
• If this PR includes breaking changes, do we need to update any jobs in
  production? Is it safe to release?
• Are there any unit tests?
• Is there a changeset associated with this PR? Should there be? Note that
  dev only changes don't need a changeset.
• Have you ticked a box under AI Usage?

[mtuchi]

image-fn.mp4

[mtuchi]

Hiya @josephjclark i have added processJsquash which uses @jsquash library.

jimp vs @jsquash/jpeg

	jimp	@jsquash
Output size (q80)	~389KB	~312KB (–20%)
Input formats	JPEG, PNG, BMP, GIF, TIFF	JPEG only
Setup cost	None	~200ms WASM init
Compression loop	4170ms	2512ms (40% faster)

For this use case (single CommCare JPEG per job): differences are minor. @jsquash wins on compression efficiency; jimp wins on format flexibility.

⚠️ If we ever need to accept PNGs or screenshots, @jsquash throws at runtime

I think the right choice depends on use case. But i will pick jimp over @jsquash because of multiple image support and simple implementation. @jsquash wasmBinary feels a bit complicated to tbh

[josephjclark]

@mtuchi jimp is fine by me - let's optimise when we need to. I'm a little anxious about the wasm in the work (although it should be fine)

Please see my earlier comment about the design of the adaptor. Need to get that right before I can look closer

[mtuchi]

Lets call it image-utils

[mtuchi]

Hiya @josephjclark i have addressed you're early feedback and cleanup .avi and .gif images, i will wait for you're feedback before cleaning up more images

[josephjclark]

Why resize(image, options) and not resize(image, width, height)? What do we gain with the options object?

[mtuchi]

@josephjclark in options we have parseAs option which is by default set to buffer.

But there are couple of other benefits like if we decide to add other options like,fit for managing fit image aspect ration, or quality to control image quality

Should we consider
resize(image, width, height, options)
OR
resize(image, dimensions ={width, height}, options= {parseAs})

[mtuchi]

Hiya @josephjclark i have added some integration test based on recent feedback from @jfeldman-openfn, See justin comment here. I would love you're feedback when you get back