Bat: Wrapping causes problems with ANSI codes

[sharkdp]

The current implementation of character-level wrapping counts incoming characters and adds a new line if the length exceeds the terminal width.

This causes problems if the incoming data stream contains ANSI characters, for example:

There are several problems:

• The lines are wrapped early due to the hidden ANSI chararacters
• Some line-wraps occur in ANSI codes, causing artifacts like 8;2;255;255;255m] to be displayed.

@eth-p: I am mentioning you just for your information.

[eth-p]

@sharkdp Good catch, I never considered that. I can think of two ways to fix this:

1. Strip the ANSI codes with console::strip_ansi_codes.
2. Attempt to parse the ANSI codes, making sure to not add them as part of the width counter.

[sharkdp]

I would definitely like to support incoming ANSI codes, just like cat or less. This will be useful if people try to alias cat=bat or use bat as a general-purpose pager, etc. - so stripping ANSI codes is not really an option for me.

Also, I would really like to avoid having to implement a ANSI-code parser within bat. The console trait seems to have a few useful functions, for example: measure_text_width. Maybe we could utilize those... but I'm not sure if this is enough.

[eth-p]

Perhaps console::AnsiCodeIterator would work? I'll play around with it and see what I can do.

[sharkdp]

There is also truncate_str, which seems really helpful!

[eth-p]

Unfortunately, truncate_str doesn't allow me to get the text that was truncated. The source code for it used AnsiCodeIterator under the hood, so I ended up just borrowing the logic from that.