Yeah, this could definitely be added. Feel free to open a PR, although I'm not sure I would be able to review it because my C knowledge isn't great, but I'm sure someone else would be able to.

I'm interested in opening a PR to add this feature (here's the PR #873). In trying to do so, I found something to be a bit confusing. As you can see, the command line argument below requests a path to be provided to inject a stylesheet.

So, I compiled resvg into a binary, provided the path to the stylesheet, and ran the command. Unsuprisingly, it worked great.

$ target/debug/resvg --stylesheet test.css some.svg some.png

However, I came across some issues when trying to replicate those results when using resvg as a library. I started from the miminal example (provided here: resvg/examples/minimal.rs). For some reason, it didn't work as intended. After some digging, it turned out that resvg as a binary expects a path to the stylesheet; however, resvg as a library expects the content of the stylesheet provided as a string. For more detail about the inconsistency, take a look a the dropdown.

I anticipate the difference when using resvg as a library and binary would cause confusion. In my opinion, the best course of action would be to treat the stylesheet as a PathBuf throughout the entire codebase. In addition to establishing consistency, when using the c-api, the additional layer of issues associated with managing memory to read from the file and save to a string would be eliminated.

If this is unintentional, how would like to handle this issue? I'm willing to open to creating another issue and creating a PR for it, but I wanted to get your opinion on it before I do that. On the other hand, if it's intentional, are there any plans to add a function or method that can accept the path to a stylesheet?

This is intentional, resvg as a library should be pure and not have any interaction with the filesystem whatsoever, so I'm afraid we can't change that.

Why can't we just ask the user of the C api to just load the file from their side and then pass the contents to resvg? (EDIT: Nevermind, didn't read the whole thing. :p But I'm afraid there really isn't much we can do, reading the file in resvg itself is both, bad from a security standpoint as well as a portability one.)

Understood. Security and portability are important considerations. I've updated my PR for this issue (#873) to account for this fact.