Commit dc929937558 for php
commit dc929937558a23fca49ad5b0c12fe103b0c4f248
Merge: bd28fb2ad58 0fcec730405
Author: David Carlier <devnexen@gmail.com>
Date: Sat Nov 21 13:04:16 2026 +0000
Merge branch 'PHP-8.5' into PHP-8.6
* PHP-8.5:
ext/dom: querySelector() rejects a non-ASCII leading identifier.
# Conflicts:
# ext/lexbor/patches/0001-Expose-line-and-column-information-for-use-in-PHP.patch
# ext/lexbor/patches/0002-Track-implied-added-nodes-for-options-use-in-PHP.patch
# ext/lexbor/patches/0003-Patch-utilities-and-data-structure-to-be-able-to-gen.patch
# ext/lexbor/patches/0004-Remove-unused-upper-case-tag-static-data.patch
# ext/lexbor/patches/0005-Shrink-size-of-static-binary-search-tree.patch
# ext/lexbor/patches/0006-Patch-out-unused-CSS-style-code.patch
# ext/lexbor/patches/0008-URL-fixed-setters-for-empty-hosts.patch
# ext/lexbor/patches/0009-URL-fixed-uninitialized-memory-in-the-path-buffer-gr.patch
# ext/lexbor/patches/0010-Fix-parsing-for-URL-containing-empty-host-and-userin.patch
# ext/lexbor/patches/0011-Percent-encode-the-caret-in-the-path.patch
# ext/lexbor/patches/0014-CSS-fixed-heap-buffer-overflow-in-lexbor-contains-pa.patch
# ext/lexbor/patches/0015-Encoding-fixed-buffer-overflows-in-malformed-decode-.patch
# ext/lexbor/patches/0016-URL-fixed-tail-pointer-in-URLSearchParams-for-delimi.patch
# ext/lexbor/patches/0017-URL-Fix-parsing-of-usernames-containing.patch
# ext/lexbor/patches/0018-Unicode-fixed-uninitialized-memory-in-IDNA-buffer-gr.patch
# ext/lexbor/patches/0019-URL-encode-opaque-path-spaces-before-query-and-fragm.patch
# ext/lexbor/patches/0020-URL-Fix-lxb_url_serialize_fragment-without-a-query-4.patch
# ext/lexbor/patches/0021-URL-expose-component-reset-functions.-415.patch
# ext/lexbor/patches/0022-URL-Fix-parsing-of-query-and-fragment-after-a-dot-co.patch
# ext/lexbor/patches/0023-URL-Keep-replacement-file-drive-paths-hierarchical-4.patch
# ext/lexbor/patches/0024-URL-normalize-output-encoding-for-percent-encoding.patch
# ext/lexbor/patches/0025-HTML-fixed-crash-use-after-free-and-leaks-on-allocat.patch
diff --cc ext/lexbor/patches/0001-Expose-line-and-column-information-for-use-in-PHP.patch
index 3a4a2072682,8a5a13166a7..a3ae53aeb4c
--- a/ext/lexbor/patches/0001-Expose-line-and-column-information-for-use-in-PHP.patch
+++ b/ext/lexbor/patches/0001-Expose-line-and-column-information-for-use-in-PHP.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Niels Dossche <7771979+nielsdos@users.noreply.github.com>
Date: Sat, 26 Aug 2023 15:08:59 +0200
- Subject: [PATCH 01/25] Expose line and column information for use in PHP
-Subject: [PATCH 01/26] Expose line and column information for use in PHP
++Subject: [PATCH 01/29] Expose line and column information for use in PHP
---
source/lexbor/dom/interfaces/node.h | 2 ++
diff --cc ext/lexbor/patches/0002-Track-implied-added-nodes-for-options-use-in-PHP.patch
index 6cfa2fe4060,c6f567e9954..9015cb14a52
--- a/ext/lexbor/patches/0002-Track-implied-added-nodes-for-options-use-in-PHP.patch
+++ b/ext/lexbor/patches/0002-Track-implied-added-nodes-for-options-use-in-PHP.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Niels Dossche <7771979+nielsdos@users.noreply.github.com>
Date: Mon, 14 Aug 2023 20:18:51 +0200
- Subject: [PATCH 02/25] Track implied added nodes for options use in PHP
-Subject: [PATCH 02/26] Track implied added nodes for options use in PHP
++Subject: [PATCH 02/29] Track implied added nodes for options use in PHP
---
source/lexbor/html/tree.h | 3 +++
diff --cc ext/lexbor/patches/0003-Patch-utilities-and-data-structure-to-be-able-to-gen.patch
index df0778c5c8c,e2bddc62b70..1137f3c75d9
--- a/ext/lexbor/patches/0003-Patch-utilities-and-data-structure-to-be-able-to-gen.patch
+++ b/ext/lexbor/patches/0003-Patch-utilities-and-data-structure-to-be-able-to-gen.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Niels Dossche <7771979+nielsdos@users.noreply.github.com>
Date: Thu, 24 Aug 2023 22:57:48 +0200
- Subject: [PATCH 03/25] Patch utilities and data structure to be able to
-Subject: [PATCH 03/26] Patch utilities and data structure to be able to
++Subject: [PATCH 03/29] Patch utilities and data structure to be able to
generate smaller lookup tables
Changed the generation script to check if everything fits in 32-bits.
diff --cc ext/lexbor/patches/0004-Remove-unused-upper-case-tag-static-data.patch
index aab66f18bab,2a4dcdc4e4d..450d29327d1
--- a/ext/lexbor/patches/0004-Remove-unused-upper-case-tag-static-data.patch
+++ b/ext/lexbor/patches/0004-Remove-unused-upper-case-tag-static-data.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Niels Dossche <7771979+nielsdos@users.noreply.github.com>
Date: Wed, 29 Nov 2023 21:26:47 +0100
- Subject: [PATCH 04/25] Remove unused upper case tag static data
-Subject: [PATCH 04/26] Remove unused upper case tag static data
++Subject: [PATCH 04/29] Remove unused upper case tag static data
---
source/lexbor/tag/res.h | 2 ++
diff --cc ext/lexbor/patches/0005-Shrink-size-of-static-binary-search-tree.patch
index 441e3f5d155,5ed745307ad..6bcb36d6f6e
--- a/ext/lexbor/patches/0005-Shrink-size-of-static-binary-search-tree.patch
+++ b/ext/lexbor/patches/0005-Shrink-size-of-static-binary-search-tree.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Niels Dossche <7771979+nielsdos@users.noreply.github.com>
Date: Wed, 29 Nov 2023 21:29:31 +0100
- Subject: [PATCH 05/25] Shrink size of static binary search tree
-Subject: [PATCH 05/26] Shrink size of static binary search tree
++Subject: [PATCH 05/29] Shrink size of static binary search tree
This also makes it more efficient on the data cache.
---
diff --cc ext/lexbor/patches/0006-Patch-out-unused-CSS-style-code.patch
index b9466577c05,58e64a0441c..f9236836703
--- a/ext/lexbor/patches/0006-Patch-out-unused-CSS-style-code.patch
+++ b/ext/lexbor/patches/0006-Patch-out-unused-CSS-style-code.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Niels Dossche <7771979+nielsdos@users.noreply.github.com>
Date: Sun, 7 Jan 2024 21:59:28 +0100
- Subject: [PATCH 06/25] Patch out unused CSS style code
-Subject: [PATCH 06/26] Patch out unused CSS style code
++Subject: [PATCH 06/29] Patch out unused CSS style code
---
source/lexbor/css/rule.h | 2 ++
diff --cc ext/lexbor/patches/0007-Add-lxb_url_is_special-to-the-public-API-362.patch
index 22ab3caea1f,00000000000..54b48ecceb6
mode 100644,000000..100644
--- a/ext/lexbor/patches/0007-Add-lxb_url_is_special-to-the-public-API-362.patch
+++ b/ext/lexbor/patches/0007-Add-lxb_url_is_special-to-the-public-API-362.patch
@@@ -1,44 -1,0 +1,44 @@@
+From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
+From: =?UTF-8?q?M=C3=A1t=C3=A9=20Kocsis?= <kocsismate@woohoolabs.com>
+Date: Sun, 17 May 2026 22:17:14 +0200
- Subject: [PATCH 07/25] Add lxb_url_is_special() to the public API (#362)
++Subject: [PATCH 07/29] Add lxb_url_is_special() to the public API (#362)
+
+As https://wiki.php.net/rfc/uri_followup#uri_type_detection relies on this information.
+---
+ source/lexbor/url/url.c | 2 +-
+ source/lexbor/url/url.h | 9 +++++++++
+ 2 files changed, 10 insertions(+), 1 deletion(-)
+
+diff --git a/source/lexbor/url/url.c b/source/lexbor/url/url.c
+index 5a11434..a5b323f 100644
+--- a/source/lexbor/url/url.c
++++ b/source/lexbor/url/url.c
+@@ -860,7 +860,7 @@ lxb_url_is_url_codepoint(lxb_codepoint_t cp)
+ return lxb_url_codepoint_alphanumeric[(lxb_char_t) cp] != 0xFF;
+ }
+
+-lxb_inline bool
++bool
+ lxb_url_is_special(const lxb_url_t *url)
+ {
+ return url->scheme.type != LXB_URL_SCHEMEL_TYPE__UNKNOWN;
+diff --git a/source/lexbor/url/url.h b/source/lexbor/url/url.h
+index 4ed3f32..6cc6f10 100644
+--- a/source/lexbor/url/url.h
++++ b/source/lexbor/url/url.h
+@@ -763,6 +763,15 @@ LXB_API lxb_status_t
+ lxb_url_search_params_serialize(lxb_url_search_params_t *search_params,
+ lexbor_callback_f cb, void *ctx);
+
++/**
++ * Returns whether the URL is special.
++ *
++ * @param[in] lxb_url_t *. Cannot be NULL.
++ * @return true if URL is special, false otherwise.
++ */
++LXB_API bool
++lxb_url_is_special(const lxb_url_t *url);
++
+ /*
+ * Inline functions.
+ */
diff --cc ext/lexbor/patches/0008-URL-fixed-setters-for-empty-hosts.patch
index 6ffd2f11f1c,ea8a498ef59..e994f62a6a7
--- a/ext/lexbor/patches/0008-URL-fixed-setters-for-empty-hosts.patch
+++ b/ext/lexbor/patches/0008-URL-fixed-setters-for-empty-hosts.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Alexander Borisov <lex.borisov@gmail.com>
Date: Fri, 26 Jun 2026 18:55:56 +0300
- Subject: [PATCH 08/25] URL: fixed setters for empty hosts.
-Subject: [PATCH 07/26] URL: fixed setters for empty hosts.
++Subject: [PATCH 08/29] URL: fixed setters for empty hosts.
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
diff --cc ext/lexbor/patches/0009-URL-fixed-uninitialized-memory-in-the-path-buffer-gr.patch
index 8a6c4a9d2f4,745e49983da..4815378bb26
--- a/ext/lexbor/patches/0009-URL-fixed-uninitialized-memory-in-the-path-buffer-gr.patch
+++ b/ext/lexbor/patches/0009-URL-fixed-uninitialized-memory-in-the-path-buffer-gr.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Alexander Borisov <lex.borisov@gmail.com>
Date: Fri, 5 Jun 2026 22:13:32 +0300
- Subject: [PATCH 09/25] URL: fixed uninitialized memory in the path buffer
-Subject: [PATCH 08/26] URL: fixed uninitialized memory in the path buffer
++Subject: [PATCH 09/29] URL: fixed uninitialized memory in the path buffer
growth.
When a path was long enough to outgrow the on-stack buffer, the first
diff --cc ext/lexbor/patches/0010-Fix-parsing-for-URL-containing-empty-host-and-userin.patch
index 8f3ed0c5b05,3a9b367139e..3760e28b8e3
--- a/ext/lexbor/patches/0010-Fix-parsing-for-URL-containing-empty-host-and-userin.patch
+++ b/ext/lexbor/patches/0010-Fix-parsing-for-URL-containing-empty-host-and-userin.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=C3=A1t=C3=A9=20Kocsis?= <kocsismate@woohoolabs.com>
Date: Thu, 9 Jul 2026 21:51:05 +0200
- Subject: [PATCH 10/25] Fix parsing for URL containing empty host and userinfo
-Subject: [PATCH 09/26] Fix parsing for URL containing empty host and userinfo
++Subject: [PATCH 10/29] Fix parsing for URL containing empty host and userinfo
The returned error code (LXB_URL_ERROR_TYPE_INVALID_CREDENTIALS) apparently contradicts the specification:
diff --cc ext/lexbor/patches/0011-Percent-encode-the-caret-in-the-path.patch
index 542488d1049,bfd69c9a3f9..41d54b1b015
--- a/ext/lexbor/patches/0011-Percent-encode-the-caret-in-the-path.patch
+++ b/ext/lexbor/patches/0011-Percent-encode-the-caret-in-the-path.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=C3=A1t=C3=A9=20Kocsis?= <kocsismate@woohoolabs.com>
Date: Fri, 10 Jul 2026 22:31:16 +0200
- Subject: [PATCH 11/25] Percent-encode the caret in the path
-Subject: [PATCH 10/26] Percent-encode the caret in the path
++Subject: [PATCH 11/29] Percent-encode the caret in the path
The caret (^) is part of the path percent-encode set:
diff --cc ext/lexbor/patches/0012-URL-added-public-IPv6-parser.patch
index d4dccb9b898,00000000000..bb5846d63de
mode 100644,000000..100644
--- a/ext/lexbor/patches/0012-URL-added-public-IPv6-parser.patch
+++ b/ext/lexbor/patches/0012-URL-added-public-IPv6-parser.patch
@@@ -1,320 -1,0 +1,320 @@@
+From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
+From: Alexander Borisov <lex.borisov@gmail.com>
+Date: Wed, 12 Aug 2026 23:29:20 +0300
- Subject: [PATCH 12/25] URL: added public IPv6 parser.
++Subject: [PATCH 12/29] URL: added public IPv6 parser.
+MIME-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 8bit
+
+Added lxb_url_parse_host_ipv6() — a public entry point to the IPv6
+parser from the WHATWG specification:
+https://url.spec.whatwg.org/#concept-ipv6-parser
+
+The address is accepted both with and without the surrounding square
+brackets: "::1" and "[::1]" give the same result.
+
+https://github.com/lexbor/lexbor/pull/402
+
+The API was requested in #402 for use by php/php-src#22268.
+
+Suggested-by: Máté Kocsis (@kocsismate)
+---
+ source/lexbor/url/url.c | 42 +++++++
+ source/lexbor/url/url.h | 26 ++++
+ test/lexbor/url/parse_host_ipv6.c | 190 ++++++++++++++++++++++++++++++
+ 3 files changed, 258 insertions(+)
+ create mode 100644 test/lexbor/url/parse_host_ipv6.c
+
+diff --git a/source/lexbor/url/url.c b/source/lexbor/url/url.c
+index 8099c12..7487762 100644
+--- a/source/lexbor/url/url.c
++++ b/source/lexbor/url/url.c
+@@ -3752,6 +3752,46 @@ lxb_url_is_ipv4(lxb_url_parser_t *parser, const lxb_char_t *data,
+ return status != LXB_STATUS_ERROR;
+ }
+
++lxb_status_t
++lxb_url_parse_host_ipv6(lxb_url_parser_t *parser, const lxb_char_t *data,
++ size_t length, uint16_t *ipv6)
++{
++ lxb_status_t status;
++ lxb_url_parser_t self_parser;
++
++ if (parser == NULL) {
++ parser = &self_parser;
++
++ parser->log = NULL;
++ parser->idna = NULL;
++ parser->buffer = NULL;
++ }
++
++ if (data < data + length && *data == '[') {
++ if (data[length - 1] != ']') {
++ (void) lxb_url_log_append(parser, &data[length - 1],
++ LXB_URL_ERROR_TYPE_IPV6_UNCLOSED);
++
++ status = LXB_STATUS_ERROR_UNEXPECTED_DATA;
++
++ goto done;
++ }
++
++ data += 1;
++ length -= 2;
++ }
++
++ status = lxb_url_ipv6_parse(parser, data, data + length, ipv6);
++
++done:
++
++ if (parser == &self_parser) {
++ lxb_url_parser_destroy(parser, false);
++ }
++
++ return status;
++}
++
+ static lxb_status_t
+ lxb_url_ipv6_parse(lxb_url_parser_t *parser, const lxb_char_t *data,
+ const lxb_char_t *end, uint16_t *ipv6)
+@@ -3763,6 +3803,8 @@ lxb_url_ipv6_parse(lxb_url_parser_t *parser, const lxb_char_t *data,
+ const lxb_char_t *p;
+ lxb_url_error_type_t err_type;
+
++ memset(ipv6, 0x00, sizeof(uint16_t) * 8);
++
+ piece = ipv6;
+ compress = NULL;
+ p = data;
+diff --git a/source/lexbor/url/url.h b/source/lexbor/url/url.h
+index 6cc6f10..aa50485 100644
+--- a/source/lexbor/url/url.h
++++ b/source/lexbor/url/url.h
+@@ -321,6 +321,32 @@ lxb_url_parse_basic(lxb_url_parser_t *parser, lxb_url_t *url,
+ const lxb_char_t *data, size_t length,
+ lxb_url_state_t override_state, lxb_encoding_t encoding);
+
++/*
++ * IPv6 parser.
++ *
++ * This function is an implementation of IPv6 parsing according to the WHATWG
++ * specification.
++ * https://url.spec.whatwg.org/#concept-ipv6-parser
++ *
++ * The address can be passed both with and without the surrounding square
++ * brackets: "::1" and "[::1]" give the same result. If the opening bracket is
++ * present, the closing one is required.
++ *
++ * The output buffer is zeroed by the function, there is no need to prepare it.
++ * Use the lxb_url_serialize_host_ipv6() function to serialize the result.
++ *
++ * @param[in] lxb_url_parser_t *. Can be NULL.
++ * @param[in] Pointer to the beginning of the data. Not NULL.
++ * @param[in] Length of the data. Can be 0.
++ * @param[out] Buffer for eight (uint16_t[8]) IPv6 pieces. Not NULL. The value
++ * is meaningful only if LXB_STATUS_OK is returned.
++ *
++ * @return LXB_STATUS_OK if successful, otherwise an error status value.
++ */
++LXB_API lxb_status_t
++lxb_url_parse_host_ipv6(lxb_url_parser_t *parser, const lxb_char_t *data,
++ size_t length, uint16_t *ipv6);
++
+ /*
+ * Erase URL.
+ *
+diff --git a/test/lexbor/url/parse_host_ipv6.c b/test/lexbor/url/parse_host_ipv6.c
+new file mode 100644
+index 0000000..bbf5293
+--- /dev/null
++++ b/test/lexbor/url/parse_host_ipv6.c
+@@ -0,0 +1,190 @@
++/*
++ * Copyright (C) 2026 Alexander Borisov
++ *
++ * Author: Alexander Borisov <borisov@lexbor.com>
++ */
++
++#include <unit/test.h>
++#include <lexbor/url/url.h>
++
++
++typedef struct {
++ const lxb_char_t *input;
++ size_t length;
++ uint16_t ipv6[8];
++}
++ipv6_success_t;
++
++typedef struct {
++ const lxb_char_t *input;
++ lxb_url_error_type_t error;
++}
++ipv6_failure_t;
++
++
++static const ipv6_success_t success_entries[] = {
++ {
++ (const lxb_char_t *) "::",
++ sizeof("::") - 1,
++ {0, 0, 0, 0, 0, 0, 0, 0}
++ },
++ {
++ (const lxb_char_t *) "::1",
++ sizeof("::1") - 1,
++ {0, 0, 0, 0, 0, 0, 0, 1}
++ },
++ {
++ (const lxb_char_t *) "[::1]",
++ sizeof("[::1]") - 1,
++ {0, 0, 0, 0, 0, 0, 0, 1}
++ },
++ {
++ (const lxb_char_t *) "1:2:3:4:5:6:7:8",
++ sizeof("1:2:3:4:5:6:7:8") - 1,
++ {1, 2, 3, 4, 5, 6, 7, 8}
++ },
++ {
++ (const lxb_char_t *) "2001:db8::ff00:42:8329",
++ sizeof("2001:db8::ff00:42:8329") - 1,
++ {0x2001, 0x0db8, 0, 0, 0, 0xff00, 0x0042, 0x8329}
++ },
++ {
++ (const lxb_char_t *) "::ffff:192.0.2.1",
++ sizeof("::ffff:192.0.2.1") - 1,
++ {0, 0, 0, 0, 0, 0xffff, 0xc000, 0x0201}
++ },
++ {
++ (const lxb_char_t *) "[::1]ignored",
++ 5,
++ {0, 0, 0, 0, 0, 0, 0, 1}
++ }
++};
++
++static const ipv6_failure_t failure_entries[] = {
++ {
++ (const lxb_char_t *) "",
++ LXB_URL_ERROR_TYPE_IPV6_TOO_FEW_PIECES
++ },
++ {
++ (const lxb_char_t *) "[::1",
++ LXB_URL_ERROR_TYPE_IPV6_UNCLOSED
++ },
++ {
++ (const lxb_char_t *) ":",
++ LXB_URL_ERROR_TYPE_IPV6_INVALID_COMPRESSION
++ },
++ {
++ (const lxb_char_t *) "1::2::3",
++ LXB_URL_ERROR_TYPE_IPV6_MULTIPLE_COMPRESSION
++ },
++ {
++ (const lxb_char_t *) "1:2:3:4:5:6:7:8:9",
++ LXB_URL_ERROR_TYPE_IPV6_TOO_MANY_PIECES
++ },
++ {
++ (const lxb_char_t *) "1:2:3:4:5:6:7",
++ LXB_URL_ERROR_TYPE_IPV6_TOO_FEW_PIECES
++ },
++ {
++ (const lxb_char_t *) "1:2:3:4:5:6:7:g",
++ LXB_URL_ERROR_TYPE_IPV6_INVALID_CODE_POINT
++ },
++ {
++ (const lxb_char_t *) "1:2:3:4:5:6:7:1.2.3.4",
++ LXB_URL_ERROR_TYPE_IPV4_IN_IPV6_TOO_MANY_PIECES
++ },
++ {
++ (const lxb_char_t *) "::ffff:.1.2.3",
++ LXB_URL_ERROR_TYPE_IPV4_IN_IPV6_INVALID_CODE_POINT
++ },
++ {
++ (const lxb_char_t *) "::ffff:192.0.2.256",
++ LXB_URL_ERROR_TYPE_IPV4_IN_IPV6_OUT_OF_RANGE_PART
++ },
++ {
++ (const lxb_char_t *) "::ffff:192.0.2",
++ LXB_URL_ERROR_TYPE_IPV4_IN_IPV6_TOO_FEW_PARTS
++ }
++};
++
++
++TEST_BEGIN(parse_success)
++{
++ size_t length;
++ lxb_status_t status;
++ uint16_t ipv6[8];
++
++ length = sizeof(success_entries) / sizeof(ipv6_success_t);
++
++ for (size_t i = 0; i < length; i++) {
++ memset(ipv6, 0xff, sizeof(ipv6));
++
++ status = lxb_url_parse_host_ipv6(NULL, success_entries[i].input,
++ success_entries[i].length, ipv6);
++ test_eq(status, LXB_STATUS_OK);
++
++ for (size_t j = 0; j < 8; j++) {
++ test_eq_u_short(ipv6[j], success_entries[i].ipv6[j]);
++ }
++ }
++}
++TEST_END
++
++TEST_BEGIN(parse_failure)
++{
++ size_t length;
++ lxb_status_t status;
++ lxb_url_parser_t parser;
++ lexbor_plog_entry_t *error;
++
++ status = lxb_url_parser_init(&parser, NULL);
++ test_eq(status, LXB_STATUS_OK);
++
++ length = sizeof(failure_entries) / sizeof(ipv6_failure_t);
++
++ for (size_t i = 0; i < length; i++) {
++ status = lxb_url_parse_host_ipv6(
++ &parser, failure_entries[i].input,
++ strlen((const char *) failure_entries[i].input),
++ (uint16_t[8]) {0});
++
++ test_eq(status, LXB_STATUS_ERROR_UNEXPECTED_DATA);
++ test_ne(parser.log, NULL);
++ test_eq_size(lexbor_plog_length(parser.log), 1UL);
++
++ error = lexbor_array_obj_get(&parser.log->list, 0);
++ test_ne(error, NULL);
++ test_eq(error->id, failure_entries[i].error);
++
++ lxb_url_parser_clean(&parser);
++ }
++
++ lxb_url_parser_memory_destroy(&parser);
++ lxb_url_parser_destroy(&parser, false);
++}
++TEST_END
++
++TEST_BEGIN(parse_failure_without_parser)
++{
++ lxb_status_t status;
++ uint16_t ipv6[8];
++
++ static const lexbor_str_t input = lexbor_str("::ffff:192.0.2.256");
++
++ status = lxb_url_parse_host_ipv6(NULL, input.data, input.length, ipv6);
++ test_eq(status, LXB_STATUS_ERROR_UNEXPECTED_DATA);
++}
++TEST_END
++
++int
++main(int argc, const char *argv[])
++{
++ TEST_INIT();
++
++ TEST_ADD(parse_success);
++ TEST_ADD(parse_failure);
++ TEST_ADD(parse_failure_without_parser);
++
++ TEST_RUN("lexbor/url/parse_host_ipv6");
++ TEST_RELEASE();
++}
diff --cc ext/lexbor/patches/0013-URL-added-public-percent-encoder-API.patch
index 9f44f2cce2f,00000000000..7b0f95cc9ea
mode 100644,000000..100644
--- a/ext/lexbor/patches/0013-URL-added-public-percent-encoder-API.patch
+++ b/ext/lexbor/patches/0013-URL-added-public-percent-encoder-API.patch
@@@ -1,626 -1,0 +1,626 @@@
+From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
+From: Alexander Borisov <lex.borisov@gmail.com>
+Date: Thu, 13 Aug 2026 23:29:16 +0300
- Subject: [PATCH 13/25] URL: added public percent-encoder API.
++Subject: [PATCH 13/29] URL: added public percent-encoder API.
+MIME-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 8bit
+
+Added public entry points to the percent-encoder from the WHATWG
+specification:
+https://url.spec.whatwg.org/#percent-encoded-bytes
+
+ lxb_url_percent_encode_utf_8()
+ lxb_url_percent_encode_encoding()
+ lxb_url_get_percent_encoding_map()
+
+Both encoders take a caller-supplied table of 256 entries indexed by byte
+value, where each entry is a bit mask of lxb_url_map_type_t values, so the
+percent-encode sets can be adjusted without patching the library.
+lxb_url_get_percent_encoding_map() returns the built-in table for callers
+that need only the sets defined by the specification.
+
+https://github.com/lexbor/lexbor/pull/404
+
+The API was requested in #404 for use by PHP:
+https://wiki.php.net/rfc/uri_followup#percent-encoding_support
+
+Based-on-patch-by: Máté Kocsis (@kocsismate)
+---
+ source/lexbor/url/url.c | 93 ++++++++-----
+ source/lexbor/url/url.h | 96 +++++++++++++
+ test/lexbor/url/percent_encode.c | 228 +++++++++++++++++++++++++++++++
+ 3 files changed, 380 insertions(+), 37 deletions(-)
+ create mode 100644 test/lexbor/url/percent_encode.c
+
+diff --git a/source/lexbor/url/url.c b/source/lexbor/url/url.c
+index 7487762..69d9196 100644
+--- a/source/lexbor/url/url.c
++++ b/source/lexbor/url/url.c
+@@ -27,20 +27,6 @@
+ #define LXB_URL_BUFFER_NUM_SIZE 128
+
+
+-typedef enum {
+- LXB_URL_MAP_UNDEF = 0x00,
+- LXB_URL_MAP_C0 = 0x01,
+- LXB_URL_MAP_FRAGMENT = 0x02,
+- LXB_URL_MAP_QUERY = 0x04,
+- LXB_URL_MAP_SPECIAL_QUERY = 0x08,
+- LXB_URL_MAP_PATH = 0x10,
+- LXB_URL_MAP_USERINFO = 0x20,
+- LXB_URL_MAP_COMPONENT = 0x40,
+- LXB_URL_MAP_X_WWW_FORM = 0x80,
+- LXB_URL_MAP_ALL = 0xff
+-}
+-lxb_url_map_type_t;
+-
+ typedef enum {
+ LXB_URL_HOST_OPT_UNDEF = 0 << 0,
+ LXB_URL_HOST_OPT_NOT_SPECIAL = 1 << 0,
+@@ -563,7 +549,7 @@ lxb_url_path_fix_windows_drive(lxb_url_t *url, lxb_char_t *sbuf,
+ static lxb_status_t
+ lxb_url_percent_encode_after_encoding(const lxb_char_t *data,
+ const lxb_char_t *end, lexbor_str_t *str,
+- lexbor_mraw_t *mraw,
++ lexbor_mraw_t *mraw, const uint8_t *url_map,
+ const lxb_encoding_data_t *encoding,
+ lxb_url_map_type_t enmap,
+ bool space_as_plus);
+@@ -571,7 +557,7 @@ lxb_url_percent_encode_after_encoding(const lxb_char_t *data,
+ static lxb_status_t
+ lxb_url_percent_encode_after_utf_8(const lxb_char_t *data,
+ const lxb_char_t *end, lexbor_str_t *str,
+- lexbor_mraw_t *mraw,
++ lexbor_mraw_t *mraw, const uint8_t *url_map,
+ lxb_url_map_type_t enmap,
+ bool space_as_plus);
+
+@@ -1757,9 +1743,9 @@ again:
+ tmp = (pswd != NULL) ? pswd - 1 : p;
+
+ if (tmp > begin) {
+- status = lxb_url_percent_encode_after_utf_8(begin, tmp,
+- &url->username, url->mraw,
+- LXB_URL_MAP_USERINFO, false);
++ status = lxb_url_percent_encode_after_utf_8(begin,
++ tmp, &url->username, url->mraw, lxb_url_map,
++ LXB_URL_MAP_USERINFO, false);
+ if (status != LXB_STATUS_OK) {
+ lxb_url_parse_return(orig_data, buf, status);
+ }
+@@ -1768,8 +1754,8 @@ again:
+
+ if (pswd != NULL && p > pswd) {
+ status = lxb_url_percent_encode_after_utf_8(pswd, p,
+- &url->password, url->mraw,
+- LXB_URL_MAP_USERINFO, false);
++ &url->password, url->mraw, lxb_url_map,
++ LXB_URL_MAP_USERINFO, false);
+ if (status != LXB_STATUS_OK) {
+ lxb_url_parse_return(orig_data, buf, status);
+ }
+@@ -2319,8 +2305,8 @@ again:
+ if (p >= end) {
+ tmp_str.data = NULL;
+
+- status = lxb_url_percent_encode_after_utf_8(begin, p,
+- &tmp_str, url->mraw,
++ status = lxb_url_percent_encode_after_utf_8(begin, p, &tmp_str,
++ url->mraw, lxb_url_map,
+ LXB_URL_MAP_C0, false);
+ if (status != LXB_STATUS_OK) {
+ lxb_url_parse_return(orig_data, buf, status);
+@@ -2336,8 +2322,8 @@ again:
+ if (c == '#' || c == '?') {
+ tmp_str.data = NULL;
+
+- status = lxb_url_percent_encode_after_utf_8(begin, p,
+- &tmp_str, url->mraw,
++ status = lxb_url_percent_encode_after_utf_8(begin, p, &tmp_str,
++ url->mraw, lxb_url_map,
+ LXB_URL_MAP_C0, false);
+ if (status != LXB_STATUS_OK) {
+ lxb_url_parse_return(orig_data, buf, status);
+@@ -2407,7 +2393,8 @@ again:
+
+ status = lxb_url_percent_encode_after_encoding(begin, p,
+ &url->query,
+- url->mraw, enc,
++ url->mraw,
++ lxb_url_map, enc,
+ map_type, false);
+ if (status != LXB_STATUS_OK) {
+ lxb_url_parse_return(orig_data, buf, status);
+@@ -2461,7 +2448,7 @@ again:
+ }
+
+ status = lxb_url_percent_encode_after_utf_8(begin, p, &url->fragment,
+- url->mraw,
++ url->mraw, lxb_url_map,
+ LXB_URL_MAP_FRAGMENT, false);
+ lxb_url_parse_return(orig_data, buf, status);
+
+@@ -3161,10 +3148,23 @@ lxb_url_scheme_find(const lxb_char_t *data, size_t length)
+ return &lxb_url_scheme_res[LXB_URL_SCHEMEL_TYPE__UNKNOWN];
+ }
+
++lxb_status_t
++lxb_url_percent_encode_encoding(const lxb_char_t *data, size_t length,
++ lexbor_str_t *str, lexbor_mraw_t *mraw,
++ const uint8_t *url_map,
++ const lxb_encoding_data_t *encoding,
++ lxb_url_map_type_t enmap,
++ bool space_as_plus)
++{
++ return lxb_url_percent_encode_after_encoding(data, data + length, str, mraw,
++ url_map, encoding, enmap,
++ space_as_plus);
++}
++
+ static lxb_status_t
+ lxb_url_percent_encode_after_encoding(const lxb_char_t *data,
+ const lxb_char_t *end, lexbor_str_t *str,
+- lexbor_mraw_t *mraw,
++ lexbor_mraw_t *mraw, const uint8_t *url_map,
+ const lxb_encoding_data_t *encoding,
+ lxb_url_map_type_t enmap,
+ bool space_as_plus)
+@@ -3182,7 +3182,8 @@ lxb_url_percent_encode_after_encoding(const lxb_char_t *data,
+
+ if (encoding->encoding == LXB_ENCODING_UTF_8) {
+ return lxb_url_percent_encode_after_utf_8(data, end, str, mraw,
+- enmap, space_as_plus);
++ url_map, enmap,
++ space_as_plus);
+ }
+
+ lxb_url_encoding_init(encoding, &encode);
+@@ -3193,7 +3194,7 @@ lxb_url_percent_encode_after_encoding(const lxb_char_t *data,
+ /* Only valid for UTF-8. */
+
+ while (p < end) {
+- if (lxb_url_map[*p++] & enmap) {
++ if (url_map[*p++] & enmap) {
+ length += 2;
+ }
+ }
+@@ -3249,7 +3250,7 @@ lxb_url_percent_encode_after_encoding(const lxb_char_t *data,
+ return LXB_STATUS_ERROR_MEMORY_ALLOCATION;
+ }
+ }
+- else if (lxb_url_map[c] & enmap) {
++ else if (url_map[c] & enmap) {
+ percent[1] = lexbor_str_res_char_to_two_hex_value[c][0];
+ percent[2] = lexbor_str_res_char_to_two_hex_value[c][1];
+
+@@ -3280,10 +3281,20 @@ lxb_url_percent_encode_after_encoding(const lxb_char_t *data,
+ return LXB_STATUS_OK;
+ }
+
++lxb_status_t
++lxb_url_percent_encode_utf_8(const lxb_char_t *data, size_t length,
++ lexbor_str_t *str, lexbor_mraw_t *mraw,
++ const uint8_t *url_map, lxb_url_map_type_t enmap,
++ bool space_as_plus)
++{
++ return lxb_url_percent_encode_after_utf_8(data, data + length, str, mraw,
++ url_map, enmap, space_as_plus);
++}
++
+ static lxb_status_t
+ lxb_url_percent_encode_after_utf_8(const lxb_char_t *data,
+ const lxb_char_t *end, lexbor_str_t *str,
+- lexbor_mraw_t *mraw,
++ lexbor_mraw_t *mraw, const uint8_t *url_map,
+ lxb_url_map_type_t enmap,
+ bool space_as_plus)
+ {
+@@ -3298,7 +3309,7 @@ lxb_url_percent_encode_after_utf_8(const lxb_char_t *data,
+ /* Only valid for UTF-8. */
+
+ while (p < end) {
+- if (lxb_url_map[*p++] & enmap) {
++ if (url_map[*p++] & enmap) {
+ length += 2;
+ }
+ }
+@@ -3317,7 +3328,7 @@ lxb_url_percent_encode_after_utf_8(const lxb_char_t *data,
+ if (space_as_plus && c == ' ') {
+ *pd++ = '+';
+ }
+- else if (lxb_url_map[c] & enmap) {
++ else if (url_map[c] & enmap) {
+ *pd++ = '%';
+ *pd++ = lexbor_str_res_char_to_two_hex_value[c][0];
+ *pd++ = lexbor_str_res_char_to_two_hex_value[c][1];
+@@ -3335,6 +3346,12 @@ lxb_url_percent_encode_after_utf_8(const lxb_char_t *data,
+ return LXB_STATUS_OK;
+ }
+
++const uint8_t *
++lxb_url_get_percent_encoding_map(void)
++{
++ return lxb_url_map;
++}
++
+ static lxb_status_t
+ lxb_url_host_parse(lxb_url_parser_t *parser, const lxb_char_t *data,
+ const lxb_char_t *end, lxb_url_host_t *host,
+@@ -4065,7 +4082,7 @@ lxb_url_opaque_host_parse(lxb_url_parser_t *parser, const lxb_char_t *data,
+ host->type = LXB_URL_HOST_TYPE_OPAQUE;
+
+ return lxb_url_percent_encode_after_utf_8(data, end, &host->u.opaque, mraw,
+- LXB_URL_MAP_C0, false);
++ lxb_url_map, LXB_URL_MAP_C0, false);
+ }
+
+ static lxb_status_t
+@@ -4344,7 +4361,8 @@ lxb_url_api_username_set(lxb_url_t *url,
+
+ return lxb_url_percent_encode_after_utf_8(username, username + length,
+ &url->username, url->mraw,
+- LXB_URL_MAP_USERINFO, false);
++ lxb_url_map, LXB_URL_MAP_USERINFO,
++ false);
+ }
+
+ lxb_status_t
+@@ -4364,7 +4382,8 @@ lxb_url_api_password_set(lxb_url_t *url,
+
+ return lxb_url_percent_encode_after_utf_8(password, password + length,
+ &url->password, url->mraw,
+- LXB_URL_MAP_USERINFO, false);
++ lxb_url_map, LXB_URL_MAP_USERINFO,
++ false);
+ }
+
+ lxb_status_t
+diff --git a/source/lexbor/url/url.h b/source/lexbor/url/url.h
+index aa50485..d2c9308 100644
+--- a/source/lexbor/url/url.h
++++ b/source/lexbor/url/url.h
+@@ -81,6 +81,20 @@ typedef enum {
+ }
+ lxb_url_state_t;
+
++typedef enum {
++ LXB_URL_MAP_UNDEF = 0x00,
++ LXB_URL_MAP_C0 = 0x01,
++ LXB_URL_MAP_FRAGMENT = 0x02,
++ LXB_URL_MAP_QUERY = 0x04,
++ LXB_URL_MAP_SPECIAL_QUERY = 0x08,
++ LXB_URL_MAP_PATH = 0x10,
++ LXB_URL_MAP_USERINFO = 0x20,
++ LXB_URL_MAP_COMPONENT = 0x40,
++ LXB_URL_MAP_X_WWW_FORM = 0x80,
++ LXB_URL_MAP_ALL = 0xff
++}
++lxb_url_map_type_t;
++
+ /*
+ * New values can only be added downwards.
+ * Before LXB_URL_SCHEMEL_TYPE__LAST_ENTRY.
+@@ -347,6 +361,88 @@ LXB_API lxb_status_t
+ lxb_url_parse_host_ipv6(lxb_url_parser_t *parser, const lxb_char_t *data,
+ size_t length, uint16_t *ipv6);
+
++/*
++ * UTF-8 percent-encoder.
++ *
++ * Percent-encodes bytes from data according to url_map and appends the result
++ * to str. A byte is encoded as "%HH" when the result of
++ * (url_map[byte] & enmap) is non-zero. Uppercase hexadecimal digits are used.
++ * If space_as_plus is true, U+0020 SPACE is encoded as '+' before the map is
++ * checked.
++ *
++ * The input is expected to be valid UTF-8; the function does not validate it.
++ *
++ * @param[in] Pointer to UTF-8 data. Not NULL.
++ * @param[in] Length of data. Can be 0.
++ * @param[in, out] Output string. Can be uninitialized (data = NULL). Encoded
++ * data is appended to any existing content. Not NULL.
++ * @param[in] Memory object used to allocate or resize the output string. Not
++ * NULL.
++ * @param[in] Table of 256 entries indexed by input byte, each entry is a bit
++ * mask of lxb_url_map_type_t values. Not NULL.
++ * @param[in] Mask selecting the bytes to percent-encode.
++ * @param[in] Replace U+0020 SPACE with '+' if true.
++ *
++ * @return LXB_STATUS_OK if successful, otherwise an error status value.
++ */
++LXB_API lxb_status_t
++lxb_url_percent_encode_utf_8(const lxb_char_t *data, size_t length,
++ lexbor_str_t *str, lexbor_mraw_t *mraw,
++ const uint8_t *url_map, lxb_url_map_type_t enmap,
++ bool space_as_plus);
++
++/*
++ * Percent-encode after encoding.
++ *
++ * Converts valid UTF-8 data to the specified encoding and appends the
++ * percent-encoded result to str. Each encoded byte for which
++ * (url_map[byte] & enmap) is non-zero is written as "%HH" using uppercase
++ * hexadecimal digits. If a code point cannot be represented in the target
++ * encoding, its percent-encoded numeric character reference is appended.
++ *
++ * If encoding is UTF-8, no conversion is performed. If space_as_plus is true,
++ * an encoded U+0020 SPACE is replaced with '+' before the map is checked.
++ * The input is expected to be valid UTF-8; the function does not validate it.
++ *
++ * @param[in] Pointer to UTF-8 data. Not NULL.
++ * @param[in] Length of data. Can be 0.
++ * @param[in, out] Output string. Can be uninitialized (data = NULL). Encoded
++ * data is appended to any existing content. Not NULL.
++ * @param[in] Memory object used to allocate or resize the output string. Not
++ * NULL.
++ * @param[in] Table of 256 entries indexed by encoded byte, each entry is a bit
++ * mask of lxb_url_map_type_t values. Not NULL.
++ * @param[in] Target encoding. Not NULL.
++ * @param[in] Mask selecting the bytes to percent-encode.
++ * @param[in] Replace an encoded U+0020 SPACE with '+' if true.
++ *
++ * @return LXB_STATUS_OK if successful, otherwise an error status value.
++ */
++LXB_API lxb_status_t
++lxb_url_percent_encode_encoding(const lxb_char_t *data, size_t length,
++ lexbor_str_t *str, lexbor_mraw_t *mraw,
++ const uint8_t *url_map,
++ const lxb_encoding_data_t *encoding,
++ lxb_url_map_type_t enmap,
++ bool space_as_plus);
++
++/*
++ * Get the URL percent-encoding map.
++ *
++ * Returns the built-in lookup table for the percent-encode sets defined by the
++ * URL specification. The table contains 256 entries indexed by byte value.
++ * Each entry is a bit mask of the lxb_url_map_type_t sets in which the byte
++ * must be percent-encoded.
++ *
++ * The returned map can be passed to lxb_url_percent_encode_utf_8() or
++ * lxb_url_percent_encode_encoding(). It has static storage duration and must
++ * not be modified or freed.
++ *
++ * @return Pointer to a read-only table of 256 entries. Never NULL.
++ */
++LXB_API const uint8_t *
++lxb_url_get_percent_encoding_map(void);
++
+ /*
+ * Erase URL.
+ *
+diff --git a/test/lexbor/url/percent_encode.c b/test/lexbor/url/percent_encode.c
+new file mode 100644
+index 0000000..361e22e
+--- /dev/null
++++ b/test/lexbor/url/percent_encode.c
+@@ -0,0 +1,228 @@
++/*
++ * Copyright (C) 2026 Alexander Borisov
++ *
++ * Author: Alexander Borisov <borisov@lexbor.com>
++ */
++
++#include <unit/test.h>
++#include <lexbor/url/url.h>
++
++
++typedef struct {
++ const lexbor_str_t input;
++ const lexbor_str_t output;
++ lxb_encoding_t encoding;
++ lxb_url_map_type_t enmap;
++ bool space_as_plus;
++ const uint8_t *url_map;
++ const lexbor_str_t initial;
++}
++percent_encode_entry_t;
++
++
++static const uint8_t custom_url_map[256] = {
++ ['a'] = LXB_URL_MAP_QUERY,
++ ['b'] = LXB_URL_MAP_PATH
++};
++
++static const percent_encode_entry_t entries[] = {
++ {
++ lexbor_str(""),
++ lexbor_str(""),
++ LXB_ENCODING_UTF_8,
++ LXB_URL_MAP_COMPONENT,
++ false,
++ NULL,
++ lexbor_str("")
++ },
++ {
++ lexbor_str("AZaz09-._~!*'()"),
++ lexbor_str("AZaz09-._~!*'()"),
++ LXB_ENCODING_UTF_8,
++ LXB_URL_MAP_COMPONENT,
++ false,
++ NULL,
++ lexbor_str("")
++ },
++ {
++ lexbor_str("\x00" "A /?\xC3\xA9"),
++ lexbor_str("%00A%20%2F%3F%C3%A9"),
++ LXB_ENCODING_UTF_8,
++ LXB_URL_MAP_COMPONENT,
++ false,
++ NULL,
++ lexbor_str("")
++ },
++ {
++ lexbor_str("A b"),
++ lexbor_str("prefix:A+b"),
++ LXB_ENCODING_UTF_8,
++ LXB_URL_MAP_COMPONENT,
++ true,
++ NULL,
++ lexbor_str("prefix:")
++ },
++ {
++ lexbor_str("\xE2\x89\xA1\xE2\x80\xBD"),
++ lexbor_str("%E2%89%A1%E2%80%BD"),
++ LXB_ENCODING_UTF_8,
++ LXB_URL_MAP_SPECIAL_QUERY,
++ false,
++ NULL,
++ lexbor_str("")
++ },
++ {
++ lexbor_str("\xE2\x89\xA1"),
++ lexbor_str("%81%DF"),
++ LXB_ENCODING_SHIFT_JIS,
++ LXB_URL_MAP_SPECIAL_QUERY,
++ false,
++ NULL,
++ lexbor_str("")
++ },
++ {
++ lexbor_str("\xE2\x80\xBD"),
++ lexbor_str("%26%238253%3B"),
++ LXB_ENCODING_SHIFT_JIS,
++ LXB_URL_MAP_SPECIAL_QUERY,
++ false,
++ NULL,
++ lexbor_str("")
++ },
++ {
++ lexbor_str("1+1 \xE2\x89\xA1 2%20\xE2\x80\xBD"),
++ lexbor_str("1+1%20%81%DF%202%20%26%238253%3B"),
++ LXB_ENCODING_SHIFT_JIS,
++ LXB_URL_MAP_SPECIAL_QUERY,
++ false,
++ NULL,
++ lexbor_str("")
++ },
++ {
++ lexbor_str("\xC2\xA5"),
++ lexbor_str("%1B(J\\%1B(B"),
++ LXB_ENCODING_ISO_2022_JP,
++ LXB_URL_MAP_SPECIAL_QUERY,
++ false,
++ NULL,
++ lexbor_str("")
++ },
++ {
++ lexbor_str("caf\xC3\xA9"),
++ lexbor_str("caf%E9"),
++ LXB_ENCODING_WINDOWS_1252,
++ LXB_URL_MAP_COMPONENT,
++ false,
++ NULL,
++ lexbor_str("")
++ },
++ {
++ lexbor_str("\xD0\xAF"),
++ lexbor_str("%DF"),
++ LXB_ENCODING_WINDOWS_1251,
++ LXB_URL_MAP_COMPONENT,
++ false,
++ NULL,
++ lexbor_str("")
++ },
++ {
++ lexbor_str("\xE4\xB8\xAD\xE6\x96\x87"),
++ lexbor_str("%A4%A4%A4%E5"),
++ LXB_ENCODING_BIG5,
++ LXB_URL_MAP_COMPONENT,
++ false,
++ NULL,
++ lexbor_str("")
++ },
++ {
++ lexbor_str("a b+c~"),
++ lexbor_str("a+b%2Bc%7E"),
++ LXB_ENCODING_UTF_8,
++ LXB_URL_MAP_X_WWW_FORM,
++ true,
++ NULL,
++ lexbor_str("")
++ },
++ {
++ lexbor_str("abc"),
++ lexbor_str("%61bc"),
++ LXB_ENCODING_UTF_8,
++ LXB_URL_MAP_QUERY,
++ false,
++ custom_url_map,
++ lexbor_str("")
++ }
++};
++
++
++TEST_BEGIN(percent_encode)
++{
++ size_t length;
++ lxb_char_t *data;
++ lxb_status_t status;
++ lexbor_mraw_t mraw;
++ lexbor_str_t str;
++ const uint8_t *url_map, *default_url_map;
++ const lxb_encoding_data_t *encoding;
++ const percent_encode_entry_t *entry;
++
++ status = lexbor_mraw_init(&mraw, 1024);
++ test_eq(status, LXB_STATUS_OK);
++
++ default_url_map = lxb_url_get_percent_encoding_map();
++ test_ne(default_url_map, NULL);
++
++ length = sizeof(entries) / sizeof(percent_encode_entry_t);
++
++ for (size_t i = 0; i < length; i++) {
++ entry = &entries[i];
++ encoding = lxb_encoding_data(entry->encoding);
++ test_ne(encoding, NULL);
++
++ str = (lexbor_str_t) {0};
++
++ if (entry->initial.length != 0) {
++ data = lexbor_str_init_append(&str, &mraw, entry->initial.data,
++ entry->initial.length);
++ test_ne(data, NULL);
++ }
++
++ url_map = entry->url_map;
++ if (url_map == NULL) {
++ url_map = default_url_map;
++ }
++
++ status = lxb_url_percent_encode_encoding(entry->input.data,
++ entry->input.length,
++ &str, &mraw, url_map, encoding,
++ entry->enmap,
++ entry->space_as_plus);
++ test_eq(status, LXB_STATUS_OK);
++
++ if (str.length != entry->output.length
++ || memcmp(str.data, entry->output.data, str.length) != 0)
++ {
++ TEST_PRINTLN("Percent-encode entry %zu (%s)", i + 1,
++ encoding->name);
++ }
++
++ test_eq_str_n(str.data, str.length, entry->output.data,
++ entry->output.length);
++
++ lexbor_str_destroy(&str, &mraw, false);
++ }
++
++ lexbor_mraw_destroy(&mraw, false);
++}
++TEST_END
++
++int
++main(int argc, const char *argv[])
++{
++ TEST_INIT();
++
++ TEST_ADD(percent_encode);
++
++ TEST_RUN("lexbor/url/percent_encode");
++ TEST_RELEASE();
++}
diff --cc ext/lexbor/patches/0014-CSS-fixed-heap-buffer-overflow-in-lexbor-contains-pa.patch
index f6fa34ec060,8d32202c968..6a4688e9537
--- a/ext/lexbor/patches/0014-CSS-fixed-heap-buffer-overflow-in-lexbor-contains-pa.patch
+++ b/ext/lexbor/patches/0014-CSS-fixed-heap-buffer-overflow-in-lexbor-contains-pa.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Alexander Borisov <lex.borisov@gmail.com>
Date: Fri, 5 Jun 2026 22:34:23 +0300
- Subject: [PATCH 14/25] CSS: fixed heap buffer overflow in :lexbor-contains()
-Subject: [PATCH 11/26] CSS: fixed heap buffer overflow in :lexbor-contains()
++Subject: [PATCH 14/29] CSS: fixed heap buffer overflow in :lexbor-contains()
parsing.
The contains string buffer was allocated by the size of the string
diff --cc ext/lexbor/patches/0015-Encoding-fixed-buffer-overflows-in-malformed-decode-.patch
index 3d2abe02550,b7c87caf18f..78a8038a752
--- a/ext/lexbor/patches/0015-Encoding-fixed-buffer-overflows-in-malformed-decode-.patch
+++ b/ext/lexbor/patches/0015-Encoding-fixed-buffer-overflows-in-malformed-decode-.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Alexander Borisov <lex.borisov@gmail.com>
Date: Wed, 10 Jun 2026 19:50:10 +0300
- Subject: [PATCH 15/25] Encoding: fixed buffer overflows in malformed decode
-Subject: [PATCH 12/26] Encoding: fixed buffer overflows in malformed decode
++Subject: [PATCH 15/29] Encoding: fixed buffer overflows in malformed decode
replay.
Fixed out-of-bounds writes in buffering decoders when replacement output
diff --cc ext/lexbor/patches/0016-URL-fixed-tail-pointer-in-URLSearchParams-for-delimi.patch
index f8e9a9bf1ff,de4d33a4ea6..9f876e4ace9
--- a/ext/lexbor/patches/0016-URL-fixed-tail-pointer-in-URLSearchParams-for-delimi.patch
+++ b/ext/lexbor/patches/0016-URL-fixed-tail-pointer-in-URLSearchParams-for-delimi.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Alexander Borisov <lex.borisov@gmail.com>
Date: Fri, 5 Jun 2026 21:46:26 +0300
- Subject: [PATCH 16/25] URL: fixed tail pointer in URLSearchParams for
-Subject: [PATCH 13/26] URL: fixed tail pointer in URLSearchParams for
++Subject: [PATCH 16/29] URL: fixed tail pointer in URLSearchParams for
delimiter-free query.
When a query had a single token without '=' or '&' (e.g. "?abc"), the
diff --cc ext/lexbor/patches/0017-URL-Fix-parsing-of-usernames-containing.patch
index d5719b2abad,aa557cb10f1..c6b4a6015ee
--- a/ext/lexbor/patches/0017-URL-Fix-parsing-of-usernames-containing.patch
+++ b/ext/lexbor/patches/0017-URL-Fix-parsing-of-usernames-containing.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tim=20D=C3=BCsterhus?= <tim@bastelstu.be>
Date: Fri, 31 Jul 2026 20:56:18 +0200
- Subject: [PATCH 17/25] URL: Fix parsing of usernames containing `@`
-Subject: [PATCH 14/26] URL: Fix parsing of usernames containing `@`
++Subject: [PATCH 17/29] URL: Fix parsing of usernames containing `@`
Fixes lexbor/lexbor#399.
---
diff --cc ext/lexbor/patches/0018-Unicode-fixed-uninitialized-memory-in-IDNA-buffer-gr.patch
index 34029e75d97,313d0cda436..4336482563b
--- a/ext/lexbor/patches/0018-Unicode-fixed-uninitialized-memory-in-IDNA-buffer-gr.patch
+++ b/ext/lexbor/patches/0018-Unicode-fixed-uninitialized-memory-in-IDNA-buffer-gr.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Alexander Borisov <lex.borisov@gmail.com>
Date: Mon, 7 Sep 2026 22:11:39 +0300
- Subject: [PATCH 18/25] Unicode: fixed uninitialized memory in IDNA buffer
-Subject: [PATCH 15/26] Unicode: fixed uninitialized memory in IDNA buffer
++Subject: [PATCH 18/29] Unicode: fixed uninitialized memory in IDNA buffer
growth.
When an IDNA buffer outgrew the stack allocation, the move to the heap
diff --cc ext/lexbor/patches/0019-URL-encode-opaque-path-spaces-before-query-and-fragm.patch
index 8ee829a14b7,e6bc91549d4..dfb3f38c5f1
--- a/ext/lexbor/patches/0019-URL-encode-opaque-path-spaces-before-query-and-fragm.patch
+++ b/ext/lexbor/patches/0019-URL-encode-opaque-path-spaces-before-query-and-fragm.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=C3=A1t=C3=A9=20Kocsis?= <kocsismate@woohoolabs.com>
Date: Fri, 11 Sep 2026 21:58:45 +0200
- Subject: [PATCH 19/25] URL: encode opaque path spaces before query and
-Subject: [PATCH 16/26] URL: encode opaque path spaces before query and
++Subject: [PATCH 19/29] URL: encode opaque path spaces before query and
fragment. (#405)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
diff --cc ext/lexbor/patches/0020-URL-Fix-lxb_url_serialize_fragment-without-a-query-4.patch
index 07f44d7a3fe,1fa964f2120..4efb0b51bb6
--- a/ext/lexbor/patches/0020-URL-Fix-lxb_url_serialize_fragment-without-a-query-4.patch
+++ b/ext/lexbor/patches/0020-URL-Fix-lxb_url_serialize_fragment-without-a-query-4.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tim=20D=C3=BCsterhus?= <tim@tideways-gmbh.com>
Date: Fri, 11 Sep 2026 22:45:33 +0200
- Subject: [PATCH 20/25] URL: Fix `lxb_url_serialize_fragment()` without a query
-Subject: [PATCH 17/26] URL: Fix `lxb_url_serialize_fragment()` without a query
++Subject: [PATCH 20/29] URL: Fix `lxb_url_serialize_fragment()` without a query
(#410)
The `lxb_url_serialize_fragment()` function checked if the query contained data
diff --cc ext/lexbor/patches/0021-URL-expose-component-reset-functions.-415.patch
index caed16aba8b,1a972b600bc..b4fb0ded012
--- a/ext/lexbor/patches/0021-URL-expose-component-reset-functions.-415.patch
+++ b/ext/lexbor/patches/0021-URL-expose-component-reset-functions.-415.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=C3=A1t=C3=A9=20Kocsis?= <kocsismate@woohoolabs.com>
Date: Tue, 22 Sep 2026 22:20:28 +0200
- Subject: [PATCH 21/25] URL: expose component reset functions. (#415)
-Subject: [PATCH 18/26] URL: expose component reset functions. (#415)
++Subject: [PATCH 21/29] URL: expose component reset functions. (#415)
* URL: expose component reset functions.
diff --cc ext/lexbor/patches/0022-URL-Fix-parsing-of-query-and-fragment-after-a-dot-co.patch
index 62b5607e709,9f16a867ec8..b6fd62e7ffd
--- a/ext/lexbor/patches/0022-URL-Fix-parsing-of-query-and-fragment-after-a-dot-co.patch
+++ b/ext/lexbor/patches/0022-URL-Fix-parsing-of-query-and-fragment-after-a-dot-co.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tim=20D=C3=BCsterhus?= <tim@tideways-gmbh.com>
Date: Sat, 26 Sep 2026 18:41:05 +0200
- Subject: [PATCH 22/25] URL: Fix parsing of query and fragment after a
-Subject: [PATCH 19/26] URL: Fix parsing of query and fragment after a
++Subject: [PATCH 22/29] URL: Fix parsing of query and fragment after a
dot-component in path (#411)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
diff --cc ext/lexbor/patches/0023-URL-Keep-replacement-file-drive-paths-hierarchical-4.patch
index e8562311f16,f4a861c022a..7cb64b3cf67
--- a/ext/lexbor/patches/0023-URL-Keep-replacement-file-drive-paths-hierarchical-4.patch
+++ b/ext/lexbor/patches/0023-URL-Keep-replacement-file-drive-paths-hierarchical-4.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?M=C3=A1t=C3=A9=20Kocsis?= <kocsismate@woohoolabs.com>
Date: Mon, 21 Sep 2026 20:01:43 +0200
- Subject: [PATCH 23/25] URL: Keep replacement file drive paths hierarchical
-Subject: [PATCH 20/26] URL: Keep replacement file drive paths hierarchical
++Subject: [PATCH 23/29] URL: Keep replacement file drive paths hierarchical
(#423)
WHATWG file state (https://url.spec.whatwg.org/#file-state) step 4.4.3.2 resets the path to an empty list. Do not mark it opaque, as that makes subsequent pathname updates silently do nothing.
diff --cc ext/lexbor/patches/0024-URL-normalize-output-encoding-for-percent-encoding.patch
index 3f18aada984,2e2f4331a51..22f100b4e7f
--- a/ext/lexbor/patches/0024-URL-normalize-output-encoding-for-percent-encoding.patch
+++ b/ext/lexbor/patches/0024-URL-normalize-output-encoding-for-percent-encoding.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Alexander Borisov <lex.borisov@gmail.com>
Date: Thu, 24 Sep 2026 15:05:51 +0300
- Subject: [PATCH 24/25] URL: normalize output encoding for percent-encoding.
-Subject: [PATCH 21/26] URL: normalize output encoding for percent-encoding.
++Subject: [PATCH 24/29] URL: normalize output encoding for percent-encoding.
---
source/lexbor/url/url.c | 34 +++++++++++++++++++++++++---------
diff --cc ext/lexbor/patches/0025-HTML-fixed-crash-use-after-free-and-leaks-on-allocat.patch
index b8070ae1dce,46c06e28840..1498d3ae38a
--- a/ext/lexbor/patches/0025-HTML-fixed-crash-use-after-free-and-leaks-on-allocat.patch
+++ b/ext/lexbor/patches/0025-HTML-fixed-crash-use-after-free-and-leaks-on-allocat.patch
@@@ -1,7 -1,7 +1,7 @@@
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: sjanel <67502353+sjanel@users.noreply.github.com>
Date: Wed, 11 Nov 2026 20:26:42 +0100
- Subject: [PATCH 25/25] HTML: fixed crash, use-after-free and leaks on
-Subject: [PATCH 22/26] HTML: fixed crash, use-after-free and leaks on
++Subject: [PATCH 25/29] HTML: fixed crash, use-after-free and leaks on
allocation failure in document create and parse (#448)
HTML: fixed crash, use-after-free and leaks on allocation failure.
diff --cc ext/lexbor/patches/0026-CSS-Syntax-fixed-non-ASCII-ident-code-point-ranges.patch
index 00000000000,280e7839582..8cb05c63905
mode 000000,100644..100644
--- a/ext/lexbor/patches/0026-CSS-Syntax-fixed-non-ASCII-ident-code-point-ranges.patch
+++ b/ext/lexbor/patches/0026-CSS-Syntax-fixed-non-ASCII-ident-code-point-ranges.patch
@@@ -1,0 -1,78 +1,78 @@@
+ From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
+ From: Alexander Borisov <lex.borisov@gmail.com>
+ Date: Sat, 20 Jun 2026 15:39:44 +0400
-Subject: [PATCH 23/26] CSS::Syntax: fixed non-ASCII ident code point ranges.
++Subject: [PATCH 26/29] CSS::Syntax: fixed non-ASCII ident code point ranges.
+
+ By specification https://drafts.csswg.org/css-syntax/#non-ascii-ident-code-point.
+
+ Per report from Jon Surrell (@sirreal).
+
+ This relates to #382 issue on GitHub.
+ ---
+ source/lexbor/css/syntax/state.c | 41 ++++++++++++++++++++++----------
+ 1 file changed, 28 insertions(+), 13 deletions(-)
+
+ diff --git a/source/lexbor/css/syntax/state.c b/source/lexbor/css/syntax/state.c
+ index a93731e..a011c5a 100644
+ --- a/source/lexbor/css/syntax/state.c
+ +++ b/source/lexbor/css/syntax/state.c
+ @@ -165,31 +165,46 @@ lxb_css_syntax_state_non_ascii(lxb_codepoint_t cp)
+ /* This code is generated by the file utils/lexbor/css/syntax/non_ascii.pl */
+
+ /* Begin */
+ - if (cp <= 0x2FEF) {
+ - if (cp <= 0x1FFF) {
+ - if (cp >= 0x00F8 && cp < 0x037D) {
+ + if (cp <= 0x218F) {
+ + if (cp <= 0x00F6) {
+ + if (cp >= 0x00B7 && cp <= 0x00B7) {
+ return true;
+ }
+ - else if (cp >= 0x037F) {
+ - return true;
+ + else if (cp >= 0x00C0) {
+ + if (cp <= 0x00D6) {
+ + return true;
+ + }
+ + else if (cp >= 0x00D8) {
+ + return true;
+ + }
+ }
+ }
+ - else if (cp >= 0x2070) {
+ - if (cp <= 0x218F) {
+ + else if (cp >= 0x00F8) {
+ + if (cp <= 0x037D) {
+ return true;
+ }
+ - else if (cp >= 0x2C00) {
+ - return true;
+ + else if (cp >= 0x037F) {
+ + if (cp <= 0x1FFF) {
+ + return true;
+ + }
+ + else if (cp >= 0x2070) {
+ + return true;
+ + }
+ }
+ }
+ }
+ - else if (cp >= 0x3001) {
+ + else if (cp >= 0x2C00) {
+ if (cp <= 0xFDCF) {
+ - if (cp <= 0xDFFF) {
+ + if (cp <= 0x2FEF) {
+ return true;
+ }
+ - else if (cp >= 0xF900) {
+ - return true;
+ + else if (cp >= 0x3001) {
+ + if (cp <= 0xDFFF) {
+ + return true;
+ + }
+ + else if (cp >= 0xF900) {
+ + return true;
+ + }
+ }
+ }
+ else if (cp >= 0xFDF0) {
diff --cc ext/lexbor/patches/0027-iCSS-Syntax-fixed-U-FFFD-replacement-boundaries-for-.patch
index 00000000000,dbafbc7b0cc..fff0427e08c
mode 000000,100644..100644
--- a/ext/lexbor/patches/0027-iCSS-Syntax-fixed-U-FFFD-replacement-boundaries-for-.patch
+++ b/ext/lexbor/patches/0027-iCSS-Syntax-fixed-U-FFFD-replacement-boundaries-for-.patch
@@@ -1,0 -1,201 +1,201 @@@
+ From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
+ From: Alexander Borisov <lex.borisov@gmail.com>
+ Date: Wed, 8 Jul 2026 20:10:47 +0300
-Subject: [PATCH 24/26] iCSS::Syntax: fixed U+FFFD replacement boundaries for
++Subject: [PATCH 27/29] iCSS::Syntax: fixed U+FFFD replacement boundaries for
+ UTF-8.
+
+ The tokenizer replaced ill-formed UTF-8 with U+FFFD at boundaries that do
+ not match the specification and browsers. A truncated sequence with valid
+ continuation bytes (E2 8C) produced two U+FFFD instead of one, and a UTF-8
+ encoded surrogate (ED A0 80) produced one U+FFFD instead of three.
+
+ By specification, each maximal subpart of an ill-formed subsequence must
+ be replaced with a single U+FFFD:
+ https://drafts.csswg.org/css-syntax/#input-byte-stream
+ https://encoding.spec.whatwg.org/#utf-8-decoder
+
+ Per report from Jon Surrell (@sirreal).
+
+ This closes #390 issue on GitHub.
+ ---
+ source/lexbor/css/syntax/state.c | 129 +++++++++----------------------
+ 1 file changed, 38 insertions(+), 91 deletions(-)
+
+ diff --git a/source/lexbor/css/syntax/state.c b/source/lexbor/css/syntax/state.c
+ index a011c5a..abcf930 100644
+ --- a/source/lexbor/css/syntax/state.c
+ +++ b/source/lexbor/css/syntax/state.c
+ @@ -67,21 +67,13 @@
+ }
+
+ #define LXB_CSS_SYNTAX_UTF_8_UP_80_BLOCK(tkz, begin, data, end) \
+ + const lxb_char_t *_bad = data; \
+ lxb_codepoint_t cp = lxb_css_syntax_state_decode_utf_8_up_80(&data, \
+ end); \
+ \
+ if (cp == LXB_CSS_SYNTAX_ERROR_CODEPOINT) { \
+ - lxb_css_syntax_buffer_append_m(tkz, begin, data - begin); \
+ - /* U+FFFD REPLACEMENT CHARACTER */ \
+ - lxb_css_syntax_buffer_append_m(tkz, "\uFFFD", \
+ - lxb_css_syntax_replacement_length);\
+ - data += 1; \
+ - begin = data; \
+ - } \
+ - /* Surrogate U+D800 to U+DFFF. */ \
+ - else if (cp >= 0xD800 && cp <= 0xDFFF) { \
+ - lxb_css_syntax_buffer_append_m(tkz, begin, \
+ - (data - begin) - 3); \
+ + /* Flush the run before the ill-formed sequence, then U+FFFD. */ \
+ + lxb_css_syntax_buffer_append_m(tkz, begin, _bad - begin); \
+ /* U+FFFD REPLACEMENT CHARACTER */ \
+ lxb_css_syntax_buffer_append_m(tkz, "\uFFFD", \
+ lxb_css_syntax_replacement_length);\
+ @@ -580,14 +572,6 @@ lxb_css_syntax_state_string(lxb_css_syntax_tokenizer_t *tkz, lxb_css_syntax_toke
+ cp = lxb_css_syntax_state_decode_utf_8_up_80(&data, end);
+
+ if (cp == LXB_CSS_SYNTAX_ERROR_CODEPOINT) {
+ - /* U+FFFD REPLACEMENT CHARACTER */
+ - lxb_css_syntax_buffer_append_m(tkz, "\uFFFD",
+ - lxb_css_syntax_replacement_length);
+ - data += 1;
+ - begin = data;
+ - }
+ - /* Surrogate U+D800 to U+DFFF. */
+ - else if (cp >= 0xD800 && cp <= 0xDFFF) {
+ /* U+FFFD REPLACEMENT CHARACTER */
+ lxb_css_syntax_buffer_append_m(tkz, "\uFFFD",
+ lxb_css_syntax_replacement_length);
+ @@ -1785,98 +1769,61 @@ static lxb_codepoint_t
+ lxb_css_syntax_state_decode_utf_8_up_80(const lxb_char_t **data,
+ const lxb_char_t *end)
+ {
+ - lxb_char_t ch;
+ + lxb_char_t ch, lower, upper;
+ + unsigned needed;
+ const lxb_char_t *p;
+ lxb_codepoint_t cp;
+
+ + /*
+ + * WHATWG UTF-8 decoder: https://encoding.spec.whatwg.org/#utf-8-decoder
+ + */
+ +
+ p = *data;
+ - ch = *p;
+ + ch = *p++;
+
+ if (ch <= 0xDF) {
+ - if (ch < 0xC2 || p + 1 >= end) {
+ + if (ch < 0xC2) {
+ + *data = p;
+ return LXB_CSS_SYNTAX_ERROR_CODEPOINT;
+ }
+
+ + needed = 1;
+ + lower = 0x80;
+ + upper = 0xBF;
+ cp = ch & 0x1F;
+ -
+ - if (p[1] < 0x80 || p[1] > 0xBF) {
+ - return LXB_CSS_SYNTAX_ERROR_CODEPOINT;
+ - }
+ -
+ - cp = (cp << 6) | (p[1] & 0x3F);
+ -
+ - *data += 2;
+ }
+ - else if (ch < 0xF0) {
+ + else if (ch <= 0xEF) {
+ + needed = 2;
+ + lower = (ch == 0xE0) ? 0xA0 : 0x80;
+ + upper = (ch == 0xED) ? 0x9F : 0xBF;
+ cp = ch & 0x0F;
+ -
+ - if (p + 2 >= end) {
+ - return LXB_CSS_SYNTAX_ERROR_CODEPOINT;
+ - }
+ -
+ - if (ch == 0xE0) {
+ - if (p[1] < 0xA0 || p[1] > 0xBF) {
+ - return LXB_CSS_SYNTAX_ERROR_CODEPOINT;
+ - }
+ - }
+ -
+ - /*
+ - * We returning surrogate pair!
+ - */
+ -/*
+ - else if (ch == 0xED) {
+ - if (p[1] < 0x80 || p[1] > 0x9F) {
+ - return LXB_CSS_SYNTAX_ERROR_CODEPOINT;
+ - }
+ - }
+ - */
+ - else if (p[1] < 0x80 || p[1] > 0xBF) {
+ - return LXB_CSS_SYNTAX_ERROR_CODEPOINT;
+ - }
+ -
+ - if (p[2] < 0x80 || p[2] > 0xBF) {
+ - return LXB_CSS_SYNTAX_ERROR_CODEPOINT;
+ - }
+ -
+ - cp = (cp << 6) | (p[1] & 0x3F);
+ - cp = (cp << 6) | (p[2] & 0x3F);
+ -
+ - *data += 3;
+ }
+ - else if (ch < 0xF5) {
+ + else if (ch <= 0xF4) {
+ + needed = 3;
+ + lower = (ch == 0xF0) ? 0x90 : 0x80;
+ + upper = (ch == 0xF4) ? 0x8F : 0xBF;
+ cp = ch & 0x07;
+ + }
+ + else {
+ + /* 0xF5..0xFF: an invalid lead byte. */
+ + *data = p;
+ + return LXB_CSS_SYNTAX_ERROR_CODEPOINT;
+ + }
+
+ - if (p + 3 >= end) {
+ - return LXB_CSS_SYNTAX_ERROR_CODEPOINT;
+ - }
+ -
+ - if (ch == 0xF0) {
+ - if (p[1] < 0x90 || p[1] > 0xBF) {
+ - return LXB_CSS_SYNTAX_ERROR_CODEPOINT;
+ - }
+ - }
+ - else if (ch == 0xF4) {
+ - if (p[1] < 0x80 || p[1] > 0x8F) {
+ - return LXB_CSS_SYNTAX_ERROR_CODEPOINT;
+ - }
+ - }
+ - else if (p[1] < 0x80 || p[1] > 0xBF) {
+ - return LXB_CSS_SYNTAX_ERROR_CODEPOINT;
+ - }
+ -
+ - if ((p[2] < 0x80 || p[2] > 0xBF) || (p[3] < 0x80 || p[3] > 0xBF)) {
+ + do {
+ + if (p >= end || *p < lower || *p > upper) {
+ + *data = p;
+ return LXB_CSS_SYNTAX_ERROR_CODEPOINT;
+ }
+
+ - cp = (cp << 6) | (p[1] & 0x3F);
+ - cp = (cp << 6) | (p[2] & 0x3F);
+ - cp = (cp << 6) | (p[3] & 0x3F);
+ + cp = (cp << 6) | (*p++ & 0x3F);
+
+ - *data += 4;
+ - }
+ - else {
+ - return LXB_CSS_SYNTAX_ERROR_CODEPOINT;
+ + lower = 0x80;
+ + upper = 0xBF;
+ }
+ + while (--needed != 0);
+ +
+ + *data = p;
+
+ return cp;
+ }
diff --cc ext/lexbor/patches/0028-CSS-Syntax-fixed-UTF-8-escapes-in-ident-sequences.patch
index 00000000000,8f90693abc4..7c57e923a23
mode 000000,100644..100644
--- a/ext/lexbor/patches/0028-CSS-Syntax-fixed-UTF-8-escapes-in-ident-sequences.patch
+++ b/ext/lexbor/patches/0028-CSS-Syntax-fixed-UTF-8-escapes-in-ident-sequences.patch
@@@ -1,0 -1,60 +1,60 @@@
+ From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
+ From: Alexander Borisov <lex.borisov@gmail.com>
+ Date: Thu, 9 Jul 2026 23:24:05 +0300
-Subject: [PATCH 25/26] CSS::Syntax: fixed UTF-8 escapes in ident sequences.
++Subject: [PATCH 28/29] CSS::Syntax: fixed UTF-8 escapes in ident sequences.
+ MIME-Version: 1.0
+ Content-Type: text/plain; charset=UTF-8
+ Content-Transfer-Encoding: 8bit
+
+ Non-hex escapes consumed only the first byte of a UTF-8 sequence and used it
+ as a code point. Because of this, "a\é" became "aÃ�" instead of "aé".
+ The same issue affected unquoted URL tokens.
+ ---
+ source/lexbor/css/syntax/state.c | 20 +++++++++++++++++++-
+ 1 file changed, 19 insertions(+), 1 deletion(-)
+
+ diff --git a/source/lexbor/css/syntax/state.c b/source/lexbor/css/syntax/state.c
+ index abcf930..a1db76a 100644
+ --- a/source/lexbor/css/syntax/state.c
+ +++ b/source/lexbor/css/syntax/state.c
+ @@ -628,6 +628,17 @@ lxb_css_syntax_state_hash(lxb_css_syntax_tokenizer_t *tkz,
+ return lxb_css_syntax_state_delim_set(token, data - 1, '#');
+ }
+
+ + if (*data >= 0x80) {
+ + begin = data;
+ +
+ + if (lxb_css_syntax_state_start_ident_utf_8_80(&data, end)) {
+ + data = begin;
+ + goto hash;
+ + }
+ +
+ + return lxb_css_syntax_state_delim_set(token, begin - 1, '#');
+ + }
+ +
+ if (lxb_css_syntax_res_name_map[*data] == 0x00) {
+ if (*data == 0x00) {
+ goto hash;
+ @@ -1201,6 +1212,7 @@ lxb_css_syntax_state_consume_ident(lxb_css_syntax_tokenizer_t *tkz,
+ else {
+ LXB_CSS_SYNTAX_UTF_8_UP_80_BLOCK(tkz, begin, data, end)
+ else if (!lxb_css_syntax_state_non_ascii(cp)) {
+ + data = _bad;
+ break;
+ }
+ }
+ @@ -1629,7 +1641,13 @@ lxb_css_syntax_state_escaped(lxb_css_syntax_tokenizer_t *tkz,
+
+ if (lexbor_str_res_map_hex[*data] == 0xFF) {
+ if (count == 0) {
+ - cp = (lxb_codepoint_t) *data++;
+ + if (*data < 0x80) {
+ + cp = (lxb_codepoint_t) *data++;
+ + }
+ + else {
+ + cp = lxb_css_syntax_state_decode_utf_8_up_80(&data, end);
+ + }
+ +
+ goto done;
+ }
+
diff --cc ext/lexbor/patches/0029-CSS-Syntax-fixed-non-ASCII-ident-code-points.patch
index 00000000000,66046794646..d979c149f14
mode 000000,100644..100644
--- a/ext/lexbor/patches/0029-CSS-Syntax-fixed-non-ASCII-ident-code-points.patch
+++ b/ext/lexbor/patches/0029-CSS-Syntax-fixed-non-ASCII-ident-code-points.patch
@@@ -1,0 -1,102 +1,102 @@@
+ From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
+ From: Alexander Borisov <lex.borisov@gmail.com>
+ Date: Fri, 10 Jul 2026 00:39:54 +0300
-Subject: [PATCH 26/26] CSS::Syntax: fixed non-ASCII ident code points.
++Subject: [PATCH 29/29] CSS::Syntax: fixed non-ASCII ident code points.
+
+ Updated generated non-ASCII ident ranges to match CSS Syntax:
+ included U+200C/U+200D and U+203F/U+2040, and stopped the
+ U+3001 range at U+D7FF to exclude surrogate code points.
+ ---
+ source/lexbor/css/syntax/state.c | 45 ++++++++++++++++++--------------
+ 1 file changed, 25 insertions(+), 20 deletions(-)
+
+ diff --git a/source/lexbor/css/syntax/state.c b/source/lexbor/css/syntax/state.c
+ index a1db76a..5b694e9 100644
+ --- a/source/lexbor/css/syntax/state.c
+ +++ b/source/lexbor/css/syntax/state.c
+ @@ -149,15 +149,10 @@ lxb_css_syntax_state_valid_escape(const lxb_char_t *data,
+ lxb_inline bool
+ lxb_css_syntax_state_non_ascii(lxb_codepoint_t cp)
+ {
+ - /*
+ - * Actually here the comparison should be to U+D7FF.
+ - * But we capture surrogate pairs, they will be further converted to 0xFFFD.
+ - */
+ -
+ /* This code is generated by the file utils/lexbor/css/syntax/non_ascii.pl */
+
+ /* Begin */
+ - if (cp <= 0x218F) {
+ + if (cp <= 0x2040) {
+ if (cp <= 0x00F6) {
+ if (cp >= 0x00B7 && cp <= 0x00B7) {
+ return true;
+ @@ -172,36 +167,46 @@ lxb_css_syntax_state_non_ascii(lxb_codepoint_t cp)
+ }
+ }
+ else if (cp >= 0x00F8) {
+ - if (cp <= 0x037D) {
+ - return true;
+ + if (cp <= 0x1FFF) {
+ + if (cp <= 0x037D) {
+ + return true;
+ + }
+ + else if (cp >= 0x037F) {
+ + return true;
+ + }
+ }
+ - else if (cp >= 0x037F) {
+ - if (cp <= 0x1FFF) {
+ + else if (cp >= 0x200C) {
+ + if (cp <= 0x200D) {
+ return true;
+ }
+ - else if (cp >= 0x2070) {
+ + else if (cp >= 0x203F) {
+ return true;
+ }
+ }
+ }
+ }
+ - else if (cp >= 0x2C00) {
+ - if (cp <= 0xFDCF) {
+ - if (cp <= 0x2FEF) {
+ + else if (cp >= 0x2070) {
+ + if (cp <= 0xD7FF) {
+ + if (cp <= 0x218F) {
+ return true;
+ }
+ - else if (cp >= 0x3001) {
+ - if (cp <= 0xDFFF) {
+ + else if (cp >= 0x2C00) {
+ + if (cp <= 0x2FEF) {
+ return true;
+ }
+ - else if (cp >= 0xF900) {
+ + else if (cp >= 0x3001) {
+ return true;
+ }
+ }
+ }
+ - else if (cp >= 0xFDF0) {
+ + else if (cp >= 0xF900) {
+ if (cp <= 0xFFFD) {
+ - return true;
+ + if (cp <= 0xFDCF) {
+ + return true;
+ + }
+ + else if (cp >= 0xFDF0) {
+ + return true;
+ + }
+ }
+ else if (cp >= 0x10000) {
+ if (cp <= 0x10FFFF) {
+ @@ -212,7 +217,7 @@ lxb_css_syntax_state_non_ascii(lxb_codepoint_t cp)
+ }
+ /* End */
+
+ - return cp == 0x200C || cp == 0x200D || cp == 0x203F || cp == 0x2040;
+ + return false;
+ }
+
+ lxb_inline bool