uucore: treat empty locale env vars as unset (fixes #13964)

[MadeNavaneeth]

Fixes #13964

Problem

When LC_ALL is set to empty string (which POSIX says means "use the next variable in the cascade"), get_locale_from_env parsed it as the C/POSIX locale with ASCII encoding. This caused ls to escape non-ASCII filenames into octal sequences even when LC_CTYPE or LANG specified UTF-8.

Fix

Treat empty environment variables the same as unset in get_locale_from_env by filtering out empty values with .filter(|v| !v.is_empty()).

Verification

Before (with LC_ALL=, LANG=uk_UA.UTF-8, LC_CTYPE=uk_UA.UTF-8):

$ ls
'\303\266 \303\246 \303\251 latin umlauts UTF-8'
\320\272\320\270\321\200\320\270\320\273\321\226\321\207\320\275\320\260 ...

After:

$ ls
ö æ é latin umlauts UTF-8
кирилічна назва в UTF-8

This matches GNU ls behavior.

[sylvestre]

sorry but it needs a test

[github-actions]

GNU testsuite comparison:

Skip an intermittent issue tests/tail/tail-n0f (fails in this run but passes in the 'main' branch)
Skipping an intermittent issue tests/date/date-locale-hour (passes in this run but fails in the 'main' branch)
Congrats! The gnu test tests/seq/seq-epipe is no longer failing!

[codspeed-hq]

Merging this PR will not alter performance

✅ 3 untouched benchmarks
⏩ 409 skipped benchmarks¹

---

_{Comparing MadeNavaneeth:fix/ls-non-ascii-locale (9307cb6) with main (30de660)}

Footnotes

1. 409 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[xtqqczze]

Suggested change

	.find_map(|&key| std::env::var(key).ok().filter(|v| !v.is_empty()));
	.find_map(|&key| std::env::var(key).ok())
	.filter(|v| !v.is_empty());

this is a bit clearer?

[xtqqczze]

The filter needs to stay inside find_map rather than after it — moving it outside would change the semantics:

In that case, we may have incorrect behavior here:

coreutils/src/uu/uniq/src/uniq.rs

Lines 199 to 205 in 5477f1e

	fn is_c_locale() -> bool {
	["LC_ALL", "LC_CTYPE", "LANG"]
	.iter()
	.find_map(|&key| std::env::var_os(key))
	.filter(|v| !v.is_empty())
	.is_none_or(|v| v == "C" || v == "POSIX")
	}

[MadeNavaneeth]

Thanks for the suggestion! The filter needs to stay inside find_map rather than after it — moving it outside would change the semantics:

// Current (correct): skips empty values, continues to next env var
.find_map(|&key| std::env::var(key).ok().filter(|v| !v.is_empty()))

// Suggested: would match `LC_ALL=""` as the first value, then drop it,
// returning None (POSIX default) instead of falling through to LANG
.find_map(|&key| std::env::var(key).ok())
.filter(|v| !v.is_empty());

The filter inside the closure ensures LC_ALL="" is treated as "not set" and the cascade continues to LC_CTYPE/LANG — which is the whole point of the fix.

Added a test (test_empty_lc_all_falls_through_to_lang) to lock in this behavior.