Support --quoting-style like ls (and quote filenames with spaces by default)

[tspiteri]

When a filename ends in a space, there is no indication. Compare this with GNU ls, which encloses filenames with spaces in quotes.

$ touch hello
$ ls
hello
$ exa
hello
$ touch 'hello '
$ ls
 hello	'hello '
$ exa
hello   hello

As an aside on the example above, GNU ls sorts 'hello ' after 'hello' while exa sorts them the other way round.

[FSMaxB]

This is a special case of #211 in a way.

[ddnexus]

#211 looks more complex to implement, maybe that's the reason it's 3+ years old?

Simply wrapping in quotes the file names when they contain a space would be a lot easier to implement and REALLY useful.

exa -l

lrwxrwxrwx 10 dd 29 Oct 11:59 foo -> bar -> baz -> foo -> bar

which is the link and which is the target?

ls -l

lrwxrwxrwx 1 dd dd 10 Oct 29 11:59 'foo -> bar -> baz' -> 'foo -> bar'

That looks quite fundamental to me.

[ogham]

This is something I still want to have!

[ariasuni]

What does ls do?

There are more details about the feature in info ls, section 10.1.7. Reproducing here the content (as of coreutils 8.32):

Documentation about --quoting-style

These options change how file names themselves are printed.

‘-b’
‘--escape’
‘--quoting-style=escape’
     Quote nongraphic characters in file names using alphabetic and
     octal backslash sequences like those used in C.

‘-N’
‘--literal’
‘--quoting-style=literal’
     Do not quote file names.  However, with ‘ls’ nongraphic characters
     are still printed as question marks if the output is a terminal and
     you do not specify the ‘--show-control-chars’ option.

‘-q’
‘--hide-control-chars’
     Print question marks instead of nongraphic characters in file
     names.  This is the default if the output is a terminal and the
     program is ‘ls’.

‘-Q’
‘--quote-name’
‘--quoting-style=c’
     Enclose file names in double quotes and quote nongraphic characters
     as in C.

‘--quoting-style=WORD’
     Use style WORD to quote file names and other strings that may
     contain arbitrary characters.  The WORD should be one of the
     following:

     ‘literal’
          Output strings as-is; this is the same as the ‘-N’ or
          ‘--literal’ option.
     ‘shell’
          Quote strings for the shell if they contain shell
          metacharacters or would cause ambiguous output.  The quoting
          is suitable for POSIX-compatible shells like ‘bash’, but it
          does not always work for incompatible shells like ‘csh’.
     ‘shell-always’
          Quote strings for the shell, even if they would normally not
          require quoting.
     ‘shell-escape’
          Like ‘shell’, but also quoting non-printable characters using
          the POSIX proposed ‘$''’ syntax suitable for most shells.
     ‘shell-escape-always’
          Like ‘shell-escape’, but quote strings even if they would
          normally not require quoting.
     ‘c’
          Quote strings as for C character string literals, including
          the surrounding double-quote characters; this is the same as
          the ‘-Q’ or ‘--quote-name’ option.
     ‘escape’
          Quote strings as for C character string literals, except omit
          the surrounding double-quote characters; this is the same as
          the ‘-b’ or ‘--escape’ option.
     ‘clocale’
          Quote strings as for C character string literals, except use
          surrounding quotation marks appropriate for the locale.
     ‘locale’
          Quote strings as for C character string literals, except use
          surrounding quotation marks appropriate for the locale, and
          quote 'like this' instead of "like this" in the default C
          locale.  This looks nicer on many displays.

     You can specify the default value of the ‘--quoting-style’ option
     with the environment variable ‘QUOTING_STYLE’.  If that environment
     variable is not set, the default value is ‘shell-escape’ when the
     output is a terminal, and ‘literal’ otherwise.

A few notes:

• literal, shell and shell-always: display some characters it can’t display, like new line (\n), as ?
• shell-escape and shell-escape-always: “POSIX proposed $'' syntax” is hard to read IMHO (the C/Rust string "first\nsecond" is printed as 'first'$'\n''second' by ls — but first$'\n'second is totally valid) and doesn’t work with fish.
• c and escape: May be easy to implement, a Rust mode might be relevant too.
• clocale and locale: I doesn’t seem very useful, and it needs support for knowing which quotes are used for which language.

What does exa do now?

• When exa displays a character by its escape code it uses a different color (it’s good!), but different filenames can be represented by the same string, only with different colors (e.g. in test\n, \n can represent a new line character or the two characters \ and n).
• Some escapes used are not valid in terminals, e.g. what ls prints as 'test'$'\b' is printed as test\u{8} in exa.

What do we want exa to do?

Several possible goals:

• Default output should be unambiguous (be able to differentiate between test\\n and test\n even without color)
• Be able to copy-paste the result and reuse it in a POSIX shell
• Support an output style for fish (see fish quotes and escaping characters, function to escape strings in fish code)

What details should be taken care of

GNU Coreutils have a page about the feature. It’s interesting because it tells about a few bugs they had to fix, especially regarding alignment.

---

Here’s the code used by GNU ls to do it:
https://github.com/coreutils/gnulib/blob/master/lib/quotearg.c

But really it sounds like a nightmare to implement all these quoting styles, and I don’t think it’s really useful. We could probably agree on three or four quoting styles.

Also, I’d argue it could be good to at least quote the filename if there’s a space in it as a first step for better handling of filenames, since I believe it’s the more common use case.