Hacker News Reader

Comment by mikegioia | Hacker News Reader

This is pretty awesome, but I think something that could PDF an html page would be much more valuable. Currently I have to use some wkhtmltopdf server-side trickery.

View on HN

Use PhantomJS to Create PDFs from HTML: https://coderwall.com/p/5vmo1g

Thank you for the lead - looks good. I found this, which goes into a bit more details for the Rails folks (like me):

http://big-elephants.com/2012-12/pdf-rendering-with-phantomj...

Looks very promising. Open to suggestions if others have better PDF-generating solutions!

What they don't tell you is that this method takes a long time and generates huge files unless you used a patched version of qt. I've been meaning to write a blog post on this.

Please do.

PhantomJS uses the same mechanism that wkhtmltopdf uses (headless webkit) and thus requires the 'server-side trickery' mentioned by the parent.

Also, the correct link is http://phantomjs.org/

I've used PrinceXML in the past and it worked well: http://www.princexml.com/

For use on the web you'll need a server license, though, which is $3800 for one machine. Fine if it's a business with revenue, but pretty steep for side projects or free services.

(There's also a free version for non-commercial users, but it's really only usable for evaluation, because it adds a PrinceXML logo to the generated PDFs.)

We offer our database publishing software as open source (http://speedata.github.io/publisher/index.html). It is used in many commercial environments as a better alternative to XSL-FO. Our software allows much more flexible layouts. That said: english documentation is still behind and we are lacking nice show cases. If you need more info, just contact me.

IIRC it's one of the few big applications written with Mercury functional logic language.

The version of this I used had basic HTML Rendering in it. I believe that the author (@mrRio) was going to do more on that. It looks, for the moment, that they've stopped mentioning the HTML rendering though, so maybe that was too tricky.

Maybe the prior HTML rendering functions are still in the source. I think it's worth downloading and checking.

HTMLtoPDF is so convenient until you need clean page breaks, page numbers, consistent header/footers etc..getting that right is extremely tedious with HTML so until now i always needed to revert back to a full blown (often expensive) PDF lib. Any1 has a better solution ?

Last time I had to generate pdf's, I just generated latex markup and then rendered it to pdf.

You could look into something like http://prawn.majesticseacreature.com/ though.

The nice thing about HTML is that it does layout for you. If you have to generate PDF yourself you have to do the layout, in particular sizing tables is a huge pain...

Latex also does the layout for you. Thats the entire point :).

Also, it is surprisingly fast.

It's a new layout, though.

Yes and one better suited for PDFs I would think: PDF files are a sequence of pages, and LaTeX pours a little effort into choosing page breaks, when I print HTML pages from a browser, the division into pages seems to done with the greedy algorithm.

The only thing that bothers be about LaTeX (or PDF readers?) is that I can't copy and paste cleanly from the document. Many glyphs get tied together, white space is lost, code doesn't come out cleanly, etc.

Thats mostly a PDF/viewer problem. PDF is not really meant for C&P.

Concerning quality of the PDF document, LaTeX very good in relation to other creators.

I absolutely agree in terms of quality.

For being a universal document format, it's very frustrating that PDF readers don't make using them easier.

Looking at http://en.wikibooks.org/wiki/LaTeX/Tables I see that the column width needs to be specified to avoid text overflow. HTML does much better job, automatically fitting a table to desired width.

It's not as easy to lay things out nicely when you have things of an unknown length that could spread out to multiple pages with HTML as with pretty much any other method.

I was using LaTeX => PDF generation, but because I don't do it all the time I found it cumbersome and confusing at times.

Prawn seems to have come a long way since I used it last - and has addressed all the limitations that pushed me towards LaTeX in the first place.

Definitely worth looking at Prawn again in my opinion.

> Currently I have to use some wkhtmltopdf server-side trickery.

I've used wkhtmltopdf, but lately I've discovered that Chrome does a pretty good job if you print a page, then specify print-to-file to a PDF destination.

That's probably the same rendering engine as wkhtmltopdf; the wk stands for webkit.

Yes, that occurred to me, but the outcomes are different. The Chrome printing route produces better results.

wkhtmltopdf to me is very unstable and crashes quite often.

How have you found it?

It's been pretty flawless for me. I'm running the 64 bit 0.11.0 static RC1. Like another post said though, it's really a pain to handle page breaks, page numbers, headers/footers etc. But for straight HTML page -> PDF it's really good.

Another option is WeasyPrint — although it is rather slow.

Reply on news.ycombinator.com