Match String case folding in StrBuilder.equalsIgnoreCase

[alhudz]

Repro: new StrBuilder("K").equalsIgnoreCase(new StrBuilder("K")) (KELVIN SIGN vs K) returns false, and so does the å/ANGSTROM pair new StrBuilder("å").equalsIgnoreCase(new StrBuilder("Å")). String.equalsIgnoreCase returns true for both.

Cause: the per-character loop only rules a differing pair equal when Character.toUpperCase(c1) == Character.toUpperCase(c2). Characters that fold together only through their lower-case mapping (U+212A → k, U+212B → å) fail that test, so the whole compare reports not-equal.

Fix: also accept the pair when Character.toLowerCase of the upper-cased chars agrees, the same fold String#regionMatches and this library's own CharSequenceUtils.equalsIgnoreCase already use. The case-sensitive equals sibling is untouched.

[garydgregory]

Thank you @alhudz , merged 🚀
Since this class is deprecated in Commons Lang (but bugs should still be fixed) in favor of Commons Text's TextStringBuilder, would you please verify parity there?

[alhudz]

Checked, parity is not there. TextStringBuilder.equalsIgnoreCase has the same upper-case-only fold (line 1911 on current master).

Repro against commons-text 1.15.0: new TextStringBuilder("K").equalsIgnoreCase(new TextStringBuilder("K")) (KELVIN SIGN vs K) -> false while String -> true; same for å vs ANGSTROM SIGN Å -> false vs true. The same one-line toLowerCase fold from this PR fixes it there too.

Thanks for merging.

[garydgregory]

Would you please create a PR for Commons Text?

[alhudz]

Done: apache/commons-text#762, same toLowerCase fold plus the Kelvin/Angstrom regression assertions in TextStringBuilderTest.