From: Lucas C.S. <lucascs@protonmail.com>
Subject: [PATCH] v8: gen-regexp-special-case: ignore supplementary code points

With USE=system-icu (bentoo's default) against dev-libs/icu-79 (Unicode 18),
the run_gen-regexp-special-case action dies: the host tool that generates
special-case.cc traps.

That is SIGTRAP from CHECK_EQ(canonical, Canonicalize(c2)) in
PrintSpecial(): Unicode 18 puts U+1DF95 (LATIN SMALL LIGATURE LONG S WITH
DESCENDER S) in the case closure of U+00DF, and the generator compares a
supplementary code point's canonical value with a BMP one's.

Backport of the generator half of upstream V8's fix (in Chromium 157's V8):
drop supplementary code points from the closure before classifying.
Non-unicode ignoreCase matching works on UTF-16 code units, so they can
never match a BMP character anyway. Upstream's runtime half (a new
SupplementaryEquivalentsSet plus a namespace refactor) is not taken: the
affected BMP characters land in IgnoreSet, and GetCaseIndependentLetters()
returns before closeOver() for those, so no supplementary code point reaches
the non-unicode path. The file is identical in node 24.22.2 and 26.13.0.
Drop once node's V8 carries upstream's RemoveSupplementary().
--- a/deps/v8/src/regexp/gen-regexp-special-case.cc
+++ b/deps/v8/src/regexp/gen-regexp-special-case.cc
@@ -17,6 +17,16 @@
 static const base::uc32 kSurrogateEnd = 0xdfff;
 static const base::uc32 kNonBmpStart = 0x10000;
 
+// Removes supplementary code points from |set|. Since Unicode 18 (ICU 79),
+// the case closure of a BMP character can contain supplementary code points
+// (U+1DF95 LATIN SMALL LIGATURE LONG S WITH DESCENDER S case-folds to U+00DF
+// LATIN SMALL LETTER SHARP S). Non-unicode ignoreCase matching compares UTF-16
+// code units, so a supplementary code point never matches a BMP character and
+// does not take part in Canonicalize.
+static void RemoveSupplementary(icu::UnicodeSet& set) {
+  set.remove(static_cast<UChar32>(kNonBmpStart), UCHAR_MAX_VALUE);
+}
+
 // The following code generates "src/regexp/special-case.cc".
 void PrintSet(std::ofstream& out, const char* name,
               const icu::UnicodeSet& set) {
@@ -63,6 +73,7 @@
     }
     current.set(i, i);
     current.closeOver(USET_CASE_INSENSITIVE);
+    RemoveSupplementary(current);
 
     // Check to see if all characters in the case-folding equivalence
     // class as defined by UnicodeSet::closeOver all map to the same
@@ -112,6 +123,7 @@
       UChar32 canonical = RegExpCaseFolding::Canonicalize(c);
       current.set(c, c);
       current.closeOver(USET_CASE_INSENSITIVE);
+      RemoveSupplementary(current);
       current.removeAll(ignore);
       for (int32_t j = 0; j < current.getRangeCount(); j++) {
         for (UChar32 c2 = current.getRangeStart(j);