File 0152-Make-sure-to-keep-tail-in-string-next_grapheme-1.patch of Package erlang
From 2f0bf4fbd45acc9f6815b4acfb99d64fadca9fa8 Mon Sep 17 00:00:00 2001
From: Christian Rasmussen <christian.rasmussen@me.com>
Date: Fri, 11 Mar 2022 23:44:19 +0100
Subject: [PATCH 1/2] Make sure to keep tail in string:next_grapheme/1
---
lib/stdlib/test/string_SUITE.erl | 14 ++++++++++----
lib/stdlib/uc_spec/gen_unicode_mod.escript | 6 +++---
2 files changed, 13 insertions(+), 7 deletions(-)
diff --git a/lib/stdlib/test/string_SUITE.erl b/lib/stdlib/test/string_SUITE.erl
index 4475d7c06e..2539bf8692 100644
--- a/lib/stdlib/test/string_SUITE.erl
+++ b/lib/stdlib/test/string_SUITE.erl
@@ -643,19 +643,25 @@ cd_gc(_) ->
[] = string:next_codepoint(""),
[] = string:next_codepoint(<<>>),
[] = string:next_codepoint([<<>>]),
+ [$a|""] = string:next_codepoint("a"),
+ [$a|<<>>] = string:next_codepoint(<<"a">>),
+ [$a|[<<>>,$b]] = string:next_codepoint([<<"a">>,$b]),
"abcd" = string:next_codepoint("abcd"),
- [$e,778] = string:next_codepoint([$e,778]),
+ [$e|[778]] = string:next_codepoint([$e,778]),
[$e|<<204,138>>] = string:next_codepoint(<<$e,778/utf8>>),
- [778|_] = string:next_codepoint(tl(string:next_codepoint(<<$e,778/utf8>>))),
+ [778|<<>>] = string:next_codepoint(tl(string:next_codepoint(<<$e,778/utf8>>))),
[0|<<128,1>>] = string:next_codepoint(<<0,128,1>>),
{error,<<128,1>>} = string:next_codepoint(<<128,1>>),
[] = string:next_grapheme(""),
[] = string:next_grapheme(<<>>),
[] = string:next_grapheme([<<>>]),
+ [$a|""] = string:next_grapheme("a"),
+ [$a|<<>>] = string:next_grapheme(<<"a">>),
+ [$a|[<<>>,$b]] = string:next_grapheme([<<"a">>,$b]),
"abcd" = string:next_grapheme("abcd"),
- [[$e,778]] = string:next_grapheme([$e,778]),
- [[$e,778]] = string:next_grapheme(<<$e,778/utf8>>),
+ [[$e,778]|""] = string:next_grapheme([$e,778]),
+ [[$e,778]|<<>>] = string:next_grapheme(<<$e,778/utf8>>),
[0|<<128,1>>] = string:next_grapheme(<<0,128,1>>),
{error,<<128,1>>} = string:next_grapheme(<<128,1>>),
diff --git a/lib/stdlib/uc_spec/gen_unicode_mod.escript b/lib/stdlib/uc_spec/gen_unicode_mod.escript
index 8239b0d5a1..af27e47337 100644
--- a/lib/stdlib/uc_spec/gen_unicode_mod.escript
+++ b/lib/stdlib/uc_spec/gen_unicode_mod.escript
@@ -671,7 +671,7 @@ gen_gc(Fd, GBP) ->
" false -> [CP0|T0]; % losing work done on T\n"
" _TrueOrZWJ -> gc_extend2(cp(T), T, [CP,CP0])\n"
" end;\n"
- "gc_extend([], _, CP) -> [CP];\n"
+ "gc_extend([], T0, CP) -> [CP|T0];\n"
"gc_extend({error,R}, _, CP) -> [CP|R].\n\n"),
io:put_chars(Fd,
"gc_extend2([CP|T], T0, Acc) ->\n"
@@ -679,8 +679,8 @@ gen_gc(Fd, GBP) ->
" false -> [lists:reverse(Acc)|T0]; % losing work done on T\n"
" _TrueOrZWJ -> gc_extend2(cp(T), T, [CP|Acc])\n"
" end;\n"
- "gc_extend2([], _, Acc) ->\n"
- " [lists:reverse(Acc)];\n"
+ "gc_extend2([], T0, Acc) ->\n"
+ " [lists:reverse(Acc)|T0];\n"
"gc_extend2({error,R}, _, Acc) ->\n"
" [lists:reverse(Acc)] ++ [R].\n\n"
),
--
2.34.1