E2E tests for torrent routes

[josecelano]

E2E tests for API torrent routes.

[josecelano]

Hi @WarmBeer @da2ce7,

Create random torrent files

I needed to generate random torrents for the E2E tests. I found this repo: https://github.com/casey/intermodal. It's a command-line tool to generate torrents.

I think it could be pretty helpful for testing all our BitTorrent-related projects.

Unfortunately, they have not published a library yet (only the binary), but in this case, the wrapper is straightforward:

    fn create_torrent_file(dir: &Path, file_name: &str) -> PathBuf {
        let input_file_path = Path::new(dir).join(file_name);
        let output_file_path = Path::new(dir).join(format!("{file_name}.torrent"));

        let output = Command::new("imdl")
            .args(["torrent", "create", "--show"])
            .args(["--input", &format!("{}", input_file_path.to_string_lossy())])
            .args(["--output", &format!("{}", output_file_path.to_string_lossy())])
            .output()
            .expect("failed to create torrent file: {output_file_path}");

        //io::stdout().write_all(&output.stdout).unwrap();
        io::stderr().write_all(&output.stderr).unwrap();

        output_file_path
    }

Since we use it for E2E testing, it's not a big problem. But for unit testing, having a system dependency would not be a good option.

Extracting logic for torrent edition

I think mods src/utils/parse_torrent.rs and src/models/torrent_file.rs are good candidates to be extracted into a primitives crate for torrent files edition. It could become a library like imdl.

Autoincrement IDs

IDs generated by the database are sometimes a pain because it makes it harder to write the asserts. For example, when you upload a torrent the response contains the ID. I'm going to add the infohash, so I can be sure the torrent was added to the index. In general, I think maybe using the infohash instead of those IDs would improve:

• Testability
• DB portability (exports/imports)

If we have cases where the infohash is not a real ID in our domain we could generate a UUID.
In general, I prefer UUID over DB autoincrement.

Finally, I still have to add more tests to finish this PR.

[josecelano]

ACK e3ed344

[josecelano]

Hi @da2ce7 @WarmBeer, this is ready to review.