From 04fa12be2eae71d8a0a3989f852369f185b556d0 Mon Sep 17 00:00:00 2001
From: Dan Gudmundsson <dgud@erlang.org>
Date: Tue, 11 Aug 2026 13:13:05 +0200
Subject: [PATCH] stdlib: Fix unicode_util:gc/1 breaking binary continuations
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The gc/1 fast-path for $\r was calling cp(R0) and returning
[CP|T] when \r was not followed by \n. This "exploded" the
binary tail into mixed chardata (integers + binary fragments),
which downstream functions in string.erl (bin_search_loop,
bin_search_inv_n) could not handle — they expect
[BinR|Cont] when is_binary(BinR).

This caused string:trim/3 to return wrong results or crash
with {case_clause,[]} when trimming strings containing binaries
followed by another list element, and the binary contained $\r
not followed by $\n.

Fix: return the original input (Str) instead of [CP|cp(R0)],
consistent with what gc_1/1 already does.

Closes #11380
---
 lib/stdlib/test/string_SUITE.erl           | 1 +
 lib/stdlib/test/unicode_util_SUITE.erl     | 6 ++++++
 lib/stdlib/uc_spec/gen_unicode_mod.escript | 4 ++--
 3 files changed, 9 insertions(+), 2 deletions(-)

diff --git a/lib/stdlib/test/string_SUITE.erl b/lib/stdlib/test/string_SUITE.erl
index dc7afaaa441d..cd553c5cc567 100644
--- a/lib/stdlib/test/string_SUITE.erl
+++ b/lib/stdlib/test/string_SUITE.erl
@@ -327,6 +327,7 @@ trim(_) ->
     ?TEST(["..h", ".e", <<"j..">>], [both, ". "], "h.ej"),
     ?TEST(["..h", <<".ejsa"/utf8>>, "n.."], [both, ". "], "h.ejsan"),
     %% Test that it behaves with graphemes (i.e. nfd tests are the hard part)
+    ?TEST([<<"\r \r">>, []], [trailing, [$\r, [$\r, $\n]]], "\r "),
     ?TEST([1013,101,778,101,101], [trailing, [101]], [1013,101,778]),
     ?TEST("aaåaa", [both, "a"], "å"),
     ?TEST(["aaa",778,"äöoo"], [both, "ao"], "åäö"),
diff --git a/lib/stdlib/test/unicode_util_SUITE.erl b/lib/stdlib/test/unicode_util_SUITE.erl
index 42a119fba997..2fca23963196 100644
--- a/lib/stdlib/test/unicode_util_SUITE.erl
+++ b/lib/stdlib/test/unicode_util_SUITE.erl
@@ -123,6 +123,12 @@ gc(Config) ->
     "hejsan" = fetch(["hej"|<<"san">>], Get),
     {error, <<128>>} = Get(<<128>>),
     {error, [<<128>>, 0]} = Get([<<128>>, 0]),
+    [$\r, <<>>, []] = Get([<<$\r>>, []]),
+    [$\r|<<$a>>] = Get(<<$\r,$a>>),
+    [$\r|<<$a>>] = Get([<<$\r,$a>>]),
+    [$\r, <<$a>>, []] = Get([<<$\r,$a>>, []]),
+    [$\r, <<$a>>, <<>>] = Get([<<$\r,$a>>, <<>>]),
+
 
     {'EXIT', _} = catch Get([-1]),
     {'EXIT', _} = catch Get([-1, $a]),
diff --git a/lib/stdlib/uc_spec/gen_unicode_mod.escript b/lib/stdlib/uc_spec/gen_unicode_mod.escript
index 0278197e985f..ee23e335ea6d 100644
--- a/lib/stdlib/uc_spec/gen_unicode_mod.escript
+++ b/lib/stdlib/uc_spec/gen_unicode_mod.escript
@@ -666,10 +666,10 @@ gen_gc(Fd, GBP) ->
     io:put_chars(Fd,
                  "gc([]=R) -> R;\n"
                  "gc([CP]=R) when ?IS_CP(CP) -> R;\n"
-                 "gc([$\\r=CP|R0]) ->\n"
+                 "gc([$\\r=CP|R0] = Str) ->\n"
                  "    case cp(R0) of % Don't break CRLF\n"
                  "        [$\\n|R1] -> [[$\\r,$\\n]|R1];\n"
-                 "        T -> [CP|T]\n"
+                 "        _ -> Str  % Keep the tail binary\n"
                  "    end;\n"
                  "gc([CP1|T1]=T) when ?IS_CP(CP1), CP1 < 256 ->\n"
                  "    case T1 of\n"