Device rendering in Python and JavaScript

Both scripts upload an EML file, start one render run, poll for completion, fetch the results, and download the screenshot assets. They call the REST API directly and need no third-party packages. Use Python 3.9+ or Node.js 22+.

Set MAILSLURP_API_KEY (or API_KEY) to a MailSlurp account key with device rendering access, then run either example from this directory:

export MAILSLURP_API_KEY=your-api-key
python3 python-device-render.py
# Or run the same flow in JavaScript:
node javascript-device-render.mjs

The Makefile loads API_KEY or MAILSLURP_API_KEY from the repository’s ../.env file. Run either example, or both:

make test-python
make test-javascript
make test # Runs both examples, creating two render runs.

With no arguments, each script uploads sample.eml and renders the gmail_chromium_web target. To use your own EML and select one or more targets:

python3 python-device-render.py message.eml gmail_chromium_web outlook_android_emulator_dark
node javascript-device-render.mjs message.eml gmail_chromium_web outlook_android_emulator_dark

With Make, pass those arguments as make test-python ARGS="message.eml gmail_chromium_web" (or test-javascript).

Use target keys from the current target catalog. Each invocation creates a new render and uses the account’s rendering allowance.

The scripts follow the same sequence:

1. POST the original EML bytes to /emails/device-previews/import/raw with Content-Type: message/rfc822 and the selected targets. This imports the message and starts its render in one request.
2. Poll /emails/device-previews/{runId}/wait in waits of up to 60 seconds, for up to ten minutes overall.
3. Fetch /emails/device-previews/{runId}/results and save results.json to render-artifacts/<run-id>/.
4. Download every screenshot’s signed accessUrl into the same directory as a PNG, without sending the account API key to the asset host.

Incomplete runs retain their metadata and available screenshots, then exit with an error. A successful run requires every selected target’s primary screenshot. The printed run ID identifies the remote run if polling times out; the timeout does not cancel it.

See the source/import guide and wait/results API guide.