From c51611a26c6d711c29e88300f98be4d73e74a7fb Mon Sep 17 00:00:00 2001 From: Weilin Du Date: Wed, 30 Sep 2026 17:41:07 +0800 Subject: [PATCH 1/2] ext/intl: Fix grapheme_strstr() and grapheme_stristr() match offsets after supplementary characters --- ext/intl/grapheme/grapheme_util.c | 2 +- .../grapheme_strstr_supplementary_prefix.phpt | 27 +++++++++++++++++++ 2 files changed, 28 insertions(+), 1 deletion(-) create mode 100644 ext/intl/tests/grapheme_strstr_supplementary_prefix.phpt diff --git a/ext/intl/grapheme/grapheme_util.c b/ext/intl/grapheme/grapheme_util.c index f0c79785483f..66c3876a5bd2 100644 --- a/ext/intl/grapheme/grapheme_util.c +++ b/ext/intl/grapheme/grapheme_util.c @@ -190,7 +190,7 @@ int32_t grapheme_strpos_utf16(char *haystack, size_t haystack_len, char *needle, if(char_pos != USEARCH_DONE && ubrk_isBoundary(bi, char_pos)) { ret_pos = grapheme_count_graphemes(bi, uhaystack,char_pos); if(puchar_pos) { - *puchar_pos = char_pos; + *puchar_pos = u_countChar32(uhaystack, char_pos); } } else { ret_pos = -1; diff --git a/ext/intl/tests/grapheme_strstr_supplementary_prefix.phpt b/ext/intl/tests/grapheme_strstr_supplementary_prefix.phpt new file mode 100644 index 000000000000..f9f85880604a --- /dev/null +++ b/ext/intl/tests/grapheme_strstr_supplementary_prefix.phpt @@ -0,0 +1,27 @@ +--TEST-- +grapheme_strstr() and grapheme_stristr() preserve match boundaries after supplementary characters +--EXTENSIONS-- +intl +--FILE-- + +--EXPECT-- +string(3) "abc" +string(4) "šŸ˜€" +string(3) "abc" +string(4) "šŸ˜€" +string(3) "abc" +string(8) "šŸ˜€šŸ˜€" +string(3) "abc" +string(8) "šŸ˜€šŸ˜€" +string(3) "abc" +string(7) "šŸ˜€é" +string(3) "abc" +string(7) "šŸ˜€é" From 94231988f21e775903cf3a7cd44868b726a0f89a Mon Sep 17 00:00:00 2001 From: Weilin Du Date: Thu, 1 Oct 2026 00:27:27 +0800 Subject: [PATCH 2/2] [skip ci] NEWS --- NEWS | 2 ++ 1 file changed, 2 insertions(+) diff --git a/NEWS b/NEWS index 4ffc2b8c0bcc..b5d863271cb0 100644 --- a/NEWS +++ b/NEWS @@ -57,6 +57,8 @@ PHP NEWS (Ilia Alshanetsky) . Fixed Collator attribute and strength methods not rejecting an unconstructed Collator. (Ilia Alshanetsky) + . Fixed grapheme_strstr() and grapheme_stristr() returning incorrect substrings + when supplementary characters precede the match. (Weilin Du) - Lexbor: . Merge patches lexbor/lexbor@8a14bc0 and lexbor/lexbor@f67ce4b, fixing a