changed generator.py to allow internal dispatching

[Hedde]

It would be convenient to be able to set the domain via settings so we can set it to localhost for example. This makes sure we can dispatch calls to the pdf from the system instead of going outside, e.g. through load balancers or reverse proxies (if any).

[tzenderman]

Hey @Hedde Thanks for making this PR, but we've actually moved on to using wkhtmltopdf which I recommend over using phantom for generating PDFs. Because of that, this repo and package has gone unmaintained by us. If you'd like, I can give you push priveleges to the repo so that you can keep it updated if you are interested? Otherwise, I would suggest working off your forks, or using a different package as we won't be maintaining this package any longer. Sorry and thanks for understanding.

[kevingu1003]

Hi @tzenderman I have also being doing some headless HTML to PDF work for some time and I actually migrated from wkhtmltopdf to phantomjs. The reason why we moved is that:

1. wkhtmltopdf is buggy(unfortunately phantomjs is buggy too) and gives few options;
2. at the time being wkhtmltopdf is distributed less friendly, I have to compile the entire project plus library(that's Gigabytes I'm talking about) to make it work on our ubuntu server, and it also have various dependencies that's hard to manage, while phantomjs is distributed more friendly;
3. phantomjs gives more debugging possibilities because of it's JS interface.

To be clear I'm not questioning that your choice of tool at all, I'm sure wkhtmltopdf must have improved a lot over the years and I also admit phantomjs's rendering bugs are super annoying. I'm just curious what made you made the decision to move to wkhtmltopdf, because your decision process can potentially help with my work too. Could you elaborate on that a little bit? Thanks a lot!

[tzenderman]

Hey @kevingu1003 ! Thanks for the message. It's funny because most of your reasons are the same ones we had for switching the other way! haha It really came down to:

• wkhtmltopdf was easier to install than the fast moving node and phantomjs and its dependencies so things would break often and issues were harder to reproduce on different machines.
• phantomjs is buggy and hard to deal with sometimes...