CFB path separators aren't properly handled

[tim-weis]

PR #21 surfaced an issue with path separators as interpreted by the cfb crate, that isn't currently honored by this implementation.

@JanMarvin Could you verify that the update on the feature/path_sep branch works for you? The following entry in your Cargo.toml file should pull in the latest changes:

[dependencies]
ovba = { git = "https://github.com/tim-weis/ovba", branch = "feature/path_sep" }

[JanMarvin]

Thanks, works fine! Tested in my tiny R package locally and in the CI.

[tim-weis]

Great, thank you!

I also tested this on Windows and Ubuntu with an Excel VBA project created on Windows, and things work as expected. I'm not sure how to incorporate that into a CI pipeline (your link gets me to a 404, unfortunately). Ideas are welcome.

Version 0.7.0 incorporates the fixes in path separator handling. Version 0.7.1, published shortly after, addresses an (inconsequential) safety issue. Just letting you know in case RUSTSEC-2023-0086 surfaces at your end (it probably will since your Cargo.lock file references the affected crate). Updating to 0.7.1 silences this warning; previous versions aren't vulnerable either since the problematic code is never exercised.

[JanMarvin]

Thanks for the heads up!

Regarding the tests: I also see the 404, odd. I have a basic functionality test in my R package (the R package simply wraps the two functions shown in your README and makes them available in R). My test reads a file that I have once stored in another github repository and compares the output against the expectation. I assume something similar should be doable in rust as well? Maybe add a vbaProject.bin file in a tests folder with both tiny programs and evaluate the output?

For reference my test and the gh-actions runs, these might not be publicly available.