Truncate output on narrow terminals

[triarius]

When the terminal is narrower than the width of the output, currently, the default mode experiences line wrapping, which does not look great:

In contrast, neofetch truncates each line to the width of the terminal:

This PR causes zeitfetch to have the same behaviour:

To do this, I had to implement a truncate function that ignores but preserves ANSI SGR escape codes. I couldn't find one that behaves precisely the way I wanted, in particular, my implementation will preserve all ANSI SGRs in the input, even if they are past the truncation point.

This lead me to notice that some of the logos don't reset the SGR \x1b[0m on each line, which I think is important because now the logo lines and the sysinfo lines are combined, so if the logo does not reset the graphics, the sysinfo line could be affected. So I've terminated each logo with \x1b[0m where necessary.

Note: the width detection may not work on all terminals. I've noticed kitty on hyprland has this issue: kovidgoyal/kitty#6530 which I've seen affecting the code in this PR. However, I still think this PR is worth merging because:

1. Not all terminals have this issue on hyprland, I've tested alacritty and wezterm, and they seem to work fine.
2. The solution for kitty (add a signal handler for SIGWINCH and wait for at least one before displaying the output) seems overly complex for an application like zeitfetch, is likely to lengthen the runtime, and there's no guarantee a SIGWINCH will be sent or caught in time.

I can have a go at implementing with some heuristic on how long to wait for a SIGWINCH if there is sufficient demand, but without it, it seems we only loose truncation for kitty on some window managers. The output will still be displayed without truncation, so nothing has been taken away from kitty users.

[nidnogg]

Hi @triarius! Happy New Year's and thanks for chipping in with another great PR! I might test these changes out later today, but if I don't merge it in right away, I'll possibly check them in after New Year's as I'm away from home at the moment.

I currently use zetifetch on iTerm and Powershell, and I know a few users who have it on macOS's default terminal too. So I think as long as those work OK these changes should be fine.

Regarding this change:

This lead me to notice that some of the logo don't reset the SGR \x1b[0m on each line, which I think is important because now the logo lines and the sysinfo lines are combined, so if the logo does not reset the graphics, the sysinfo line could be affected. So I've terminated each logo with \x1b[0m where necessary.

I want to test out how logos look without those SGR \x1b[0m completions a bit, as I can see myself or other contributors forgetting to add those later down the road, but they might not be too much of an extra step after all.

[triarius]

@nidnogg Happy New Year to you too!

No rush to merge, I may be able to test it on a macOS environment soon, I'll post some screenshots if I do.

As for future devs forgetting the SGR \x1b[0m, how about a test that fails if they aren't present? I think that's a good idea anyway. I'll do that now.

[triarius]

The logos are now in a static map, with the help of the phf dependency. I made the implicit map (i.e. a match statement) explicit, so I could write a test that checks the logos terminate their SGR codes. Ideally, I would create a map that takes enum keys and &'static str values, but I could not get phf to work like this. Instead, I created an enum and added a test that the variants of Logo enum are present in the static LOGOS map. I used the sturm and strum_derive crates for this.

While there is a small amount of duplication needed to add a new logo, it will be checked by the test_logos_keys_match_enum test.

[triarius]

@nidnogg I'm back from travels, and I've got screenshots showing the truncation working on macOS:

Terminal.app

iTerm 2

[nidnogg]

@nidnogg I'm back from travels, and I've got screenshots showing the truncation working on macOS:

Terminal.app

## iTerm 2

Heya again @triarius! I'm finally back from travels and I've reviewed the PR and it should be good to merge soon. I'll do some final testing and wrap it in the next release.

Thanks again!

[triarius]

Thanks, @nidnogg. I realised I had tweaks to make, so I push a few more commits. Apologies.

The main change is b2a9999 which falls back to not truncating if the terminal width is not detected. It should make this PR safer to merge.

[nidnogg]

Merged - should be live in latest release soon. Thanks for another great contribution!