esbenp/pdf-bot

Support for posting the html instead of adding a url

Hagbarth opened this issue · 6 comments

Hagbarth commented

I think it would be reasonable to make it possible to generate the pdf's directly from html instead of from visiting a url.

Not sure how html-pdf-chrome handles this, but Puppeteer has support out of the box, so it should be possible.

TomSeldon commented

I'm yet to play around with this, but it could potentially work as-is if you converted the HTML to a data URI string and used that as the URL.

This might be a terrible suggestion, but occurred to me that it could work. :)

esbenp commented

Actually if you paste a non-url to html-pdf-chrome it will just render it as html. However, pdf-bot validates the input and throws an error if you post a non-url. Maybe a new endpoint for non-urls or a meta-type key would be the way to go here.

So either POST / for urls POST /html for html, or something like

POST /
{"url":"http://google.com"}

POST /
{"html":"<b>html</b>"}
Hagbarth commented

@esbenp okay

POST /html is probably best, so we don't run into the issue of having to handle both a url and an html string in the same request. Although I guess the two could be pushed to the queue separately?

esbenp commented

I think either is a valid option. Personally I like a fewer, configurable endpoints. I will look into this soon

Dexus commented

You need only check

const url = /^(https?|file|data):/i.test(html) ? html : `data:text/html,${html}`;

and open the URL... I think it would not be difficult to implement it.

Exiliot commented

const url = data:text/html,${ html }

Actually if the html is really vast "PayloadTooLargeError: request entity too large" error can happen. Probably it might be solved by putting some configs for large requests within pdf-bot api express app.