From 5428a8e6d9d51169de571b5e3883dea3ca31c27e Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Michael=20Vo=C5=99=C3=AD=C5=A1ek?= <mvorisek@mvorisek.cz>
Date: Thu, 12 Feb 2026 10:41:40 +0100
Subject: [PATCH] impl UCD download

---
 .../unicode-character-database.sh             | 19 +++++++++++++++++++
 .github/workflows/verify-bundled-files.yml    | 13 +++++++++++++
 2 files changed, 32 insertions(+)
 create mode 100755 .github/scripts/download-bundled/unicode-character-database.sh

diff --git a/.github/scripts/download-bundled/unicode-character-database.sh b/.github/scripts/download-bundled/unicode-character-database.sh
new file mode 100755
index 000000000000..805c0495c636
--- /dev/null
+++ b/.github/scripts/download-bundled/unicode-character-database.sh
@@ -0,0 +1,19 @@
+#!/bin/sh
+set -ex
+cd "$(dirname "$0")/../../.."
+
+tmp_dir=/tmp/php-src-download-bundled/unicode-character-database
+rm -rf "$tmp_dir"
+
+version=17.0.0
+
+mkdir -p "$tmp_dir"
+curl --fail "https://www.unicode.org/Public/$version/ucd/UCD.zip" -o "$tmp_dir/data.zip"
+unzip "$tmp_dir/data.zip" -d "$tmp_dir"
+
+cd ext/mbstring
+
+rm libmbfl/mbfl/eaw_table.h
+rm unicode_data.h
+
+./ucgendat/ucgendat.php "$tmp_dir"
diff --git a/.github/workflows/verify-bundled-files.yml b/.github/workflows/verify-bundled-files.yml
index 473d6e2d8f4d..96a4e24d2940 100644
--- a/.github/workflows/verify-bundled-files.yml
+++ b/.github/workflows/verify-bundled-files.yml
@@ -5,6 +5,7 @@ on:
     paths: &paths
       - '.github/scripts/download-bundled/**'
       - 'Zend/asm/**'
+      - 'ext/mbstring/**'
       - 'ext/pcre/pcre2lib/**'
       - 'ext/uri/uriparser/**'
   pull_request:
@@ -34,6 +35,9 @@ jobs:
             'boost-context':
               - '.github/scripts/download-bundled/boost-context.*'
               - 'Zend/asm/**'
+            'unicode-character-database':
+              - '.github/scripts/download-bundled/unicode-character-database.*'
+              - 'ext/mbstring/**'
             pcre2:
               - '.github/scripts/download-bundled/pcre2.*'
               - 'ext/pcre/pcre2lib/**'
@@ -50,6 +54,15 @@ jobs:
           echo "::group::Verify files"
           .github/scripts/test-directory-unchanged.sh Zend/asm
           echo "::endgroup::"
+      - name: 'Unicode Character Database'
+        if: ${{ !cancelled() && (steps.changes.outputs.unicode-character-database == 'true' || github.event_name == 'schedule' || github.event_name == 'workflow_dispatch') }}
+        run: |
+          echo "::group::Download"
+          .github/scripts/download-bundled/unicode-character-database.sh
+          echo "::endgroup::"
+          echo "::group::Verify files"
+          .github/scripts/test-directory-unchanged.sh 'ext/mbstring'
+          echo "::endgroup::"
       - name: PCRE2
         if: ${{ !cancelled() && (steps.changes.outputs.pcre2 == 'true' || github.event_name == 'schedule' || github.event_name == 'workflow_dispatch') }}
         run: |