Debugging Options for a blank PDF?

[nbyloff]

I have an issue where a couple (not all) PDFs I am generating only work when local generating, but not using this AWS Lambda package. The largest, most complete of my 3 PDFs works fine with sidecar-browsershot. The other two just return single blank PDFs.

I tried editing the browsershot.js in the lambda layer to print some console.logs to the cloudwatch logger, but I am not seeing anything written. I test writing the HTML to my local drive first before submitting. I have verified everything is a public, reachable URL and loading it in my browser locally produces no console errors.

$html = View::make($view, $data);
file_put_contents('summary_test.html', $html); // loads fine with no errors in browser & produces PDF fine locally
$pdf = BrowsershotLambda::html($html); // generates 1 page blank PDF using sidecar-browsershot

I have found the 2 broken reports will generate (poorly formatted) with this package if I remove a specific JS file, a beta version of paged.js.

It's weird because the same paged.js file works fine in my browser (with no console errors), generating all 3 reports locally, and has no problem generating 700+ page PDF using this lambda package. The failing reports are much smaller, and again, have no browser errors and are using all publicly available URLs.

I am not sure how to find the answer without more verbose logging, like if Chrome is dealing with unhandled errors, issues loading files, or just being able to add console.log entries. I am wondering if it has to do with some version mismatch with Chrome on my Mac vs. this package.

What are my options as far as debugging this package inside Lambda?

[stefanzweifel]

This is such a tough question I haven't been able to anwer myself.

When I read you render 700+ page PDFs, I first wanted to point to an issue which a similar problem.

Pupppeteer on Lambda seems to have massive issues rendering huge PDFs.
I remember testing what is the critical file size where Lambda/Puppeteer gives up and I think it was somewhere around 20MB.

---

I don't have a good answer yet, how to debug this. I honestly haven't used this package or Lambda in a while and don't have the knowlege in front of me how to best tackle this.

Some suggestions:

• Store the HTML on disk and check the size of the HTML. Is it huge (+10 MB)?
• Can you maybe use readHtmlFromS3()?