Display binary file content for bat -A

[pjsier]

Closes #623. When the -A flag is supplied, attempts to print hex-escaped lines for binary files. The one caveat is it escapes non-ASCII characters, not just non-UTF-8.

From what I could tell printing the hex-escaped version of UTF-8 text would be a lot more involved, but if there's an option I missed here I'm happy to make changes

[sharkdp]

Thank you very much for your contribution!

There are a few things that could be improved.

For binary files, the grid styling is not quite right:

───────┬─────────────────────────────────────────────────
       │ File: test.png   <BINARY>
───────┴─────────────────────────────────────────────────
   1   │ \x89PNG\r\n
   2   │ \x1a\n
   3   │ \x00\x00\x00\rIHDR\x00\x00\x00\x80\x00\x00\x00D\
       │ x08\x02\x00\x00\x00\xc6%\xaa>\x00\x00\x00\xc2IDA
       │ Tx^\xed\xd4\x81\x06\xc30\x14@\xd1\xb74\xdd\xff\x
       │ ffo\xb3tV\xea\x89\x12l(s\xe2\xaa4I\x03\x87\xd6\x
       │ fe\xd8{\x89\xbbR\x8d;\x87\xfe\x01\x00\x80\x00\x0
       │ 0\x10\x00\x00\x02\x00@\x00\x00\x08\x00\x00\x01\x
       │ 00•\x00\x00\x04\x00\x80\x00\x00\x10\x00\x00\x02\
       │ x00@\x00\x00\x08\x00\x00\x01\x00•\x00\x00\x00\xd
       │ 4^jdK\x94\xf5\x98|\xd1\xf4\x92\\\\>\xcf\x9c?sqX_
       │ \xaf\x8by[\xee\x96\xb6G\xeb\xf1\xea\xd1\xce\xb6\
       │ xe3u;\xe6\xb9\x95\x8d\xc7\xce\x039\xc9\xaf\xc63\
       │ x93{f7\xcf\xab\xbf\xf9\xc9/\x08\x80\x00\x00\x10\
       │ x00\x00\x02\x00@\x00\x00\x08\x00\x00\x01\x00•\x0
       │ 0\x00\x04\x00\x80\x00\x00\x10\x00\x00\x02\x00@\x
       │ 00\x00\x08\x00\x00\x01\x00•\x00\x00\x8c7\xdbh\x0
       │ 3•\xfb\xed\x96e\x00\x00\x00\x00IEND\xaeB`\x82

Notice that the grid in the header-section is not joined with the lower part (┴ instead of ┼). Also, there is no final separator line at the end. For comparison, here is a text file:

───────┬─────────────────────────────────────────────────
       │ File: example
───────┼─────────────────────────────────────────────────
   1   │ hello␊
───────┴─────────────────────────────────────────────────

Implementation-wise, I think that the preprocessor module would be a much better place to make the non-ascii => hex conversion:

bat/src/preprocessor.rs

Lines 36 to 73 in c64ab29

	pub fn replace_nonprintable(input: &str, tab_width: usize) -> String {
	let mut output = String::new();
	let tab_width = if tab_width == 0 { 4 } else { tab_width };
	for chr in input.chars() {
	match chr {
	// space
	' ' => output.push('•'),
	// tab
	'\t' => {
	if tab_width == 1 {
	output.push('↹');
	} else {
	output.push('├');
	output.push_str(&"─".repeat(tab_width - 2));
	output.push('┤');
	}
	}
	// line feed
	'\x0A' => output.push('␊'),
	// carriage return
	'\x0D' => output.push('␍'),
	// null
	'\x00' => output.push('␀'),
	// bell
	'\x07' => output.push('␇'),
	// backspace
	'\x08' => output.push('␈'),
	// escape
	'\x1B' => output.push('␛'),
	// anything else
	_ => output.push(chr),
	}
	}
	output
	}

This would have the added advantage that we could highlight the \x9a tokens in a nice way.

[sharkdp]

A more serious issue:
This destroys some of the preprocessing that we did before where we replaced some characters like \n with Unicode characters. So this should definitely be done in the preprocessing module.

I'm working on something...

[sharkdp]

Some impressions: