Verify bundled PCRE2 in CI

Closes GH-20354

Author: mvorisek

.github/actions/verify-generated-files/action.yml

@@ -13,5 +13,4 @@ runs:
         ext/tokenizer/tokenizer_data_gen.php
         build/gen_stub.php -f --generate-optimizer-info --verify
         ext/phar/makestub.php
-        # Use the -a flag for a bug in git 2.46.0, which doesn't consider changed -diff files.
-        git add . -N && git diff -a --exit-code
+        .github/scripts/test-directory-unchanged.sh .

.github/scripts/download-bundled/pcre2.sh

@@ -0,0 +1,32 @@
+#!/bin/sh
+set -ex
+cd "$(dirname "$0")/../../.."
+
+revision=refs/tags/pcre2-10.44
+
+git clone --depth 1 --recurse-submodules --revision="$revision" https://github.com/PCRE2Project/pcre2.git /tmp/php-src-bundled/pcre2
+
+rm -rf ext/pcre/pcre2lib
+cp -R /tmp/php-src-bundled/pcre2/src ext/pcre/pcre2lib
+
+cd ext/pcre/pcre2lib
+
+# remove unneeded files
+rm config.h.generic
+rm pcre2.h.in
+rm pcre2_dftables.c
+rm pcre2_fuzzsupport.c
+rm pcre2_jit_test.c
+rm pcre2demo.c
+rm pcre2grep.c
+rm pcre2posix.c
+rm pcre2posix.h
+rm pcre2posix_test.c
+rm pcre2test.c
+
+# move renamed files
+mv pcre2.h.generic pcre2.h
+mv pcre2_chartables.c.dist pcre2_chartables.c
+
+# add extra files
+git restore config.h # based on config.h.generic but with many changes

.github/scripts/test-directory-unchanged.sh

@@ -0,0 +1,13 @@
+#!/bin/sh
+set -ex
+
+cd "$(dirname "$0")/../../$1"
+
+# notify git about untracked (except ignored) files
+git add -N .
+
+# display overview of changed files
+git status .
+
+# display diff of working directory vs HEAD commit and set exit code
+git diff -a --exit-code HEAD .

.github/workflows/verify-bundled-files.yml

@@ -0,0 +1,43 @@
+name: Verify Bundled Files
+
+on:
+  push:
+    paths: &paths
+      - '.github/scripts/download-bundled/pcre2.sh'
+      - 'ext/pcre/pcre2lib/**'
+  pull_request:
+    paths: *paths
+  schedule:
+    - cron: "0 1 * * *"
+  workflow_dispatch: ~
+
+permissions:
+  contents: read
+
+jobs:
+  VERIFY_BUNDLED_FILES:
+    name: Verify Bundled Files
+    runs-on: ubuntu-24.04
+    steps:
+      - name: git checkout
+        uses: actions/checkout@v5
+
+      - name: Detect changed files
+        uses: dorny/paths-filter@v3
+        id: changes
+        with:
+          base: master
+          filters: |
+            pcre2:
+              - '.github/scripts/download-bundled/pcre2.sh'
+              - 'ext/pcre/pcre2lib/**'
+
+      - name: PCRE2
+        if: ${{ !cancelled() && (steps.changes.outputs.pcre2 == 'true' || github.event_name == 'schedule' || github.event_name == 'workflow_dispatch') }}
+        run: |
+          echo "::group::Download"
+          .github/scripts/download-bundled/pcre2.sh
+          echo "::endgroup::"
+          echo "::group::Verify files"
+          .github/scripts/test-directory-unchanged.sh ext/pcre/pcre2lib
+          echo "::endgroup::"

ext/pcre/pcre2lib/pcre2_chartables.c

@@ -5,7 +5,8 @@
 /* This file was automatically written by the pcre2_dftables auxiliary
 program. It contains character tables that are used when no external
 tables are passed to PCRE2 by the application that calls it. The tables
-are used only for characters whose code values are less than 256. */
+are used only for characters whose code values are less than 256, and
+only relevant if not in UCP mode. */
 
 /* This set of tables was written in the C locale. */
 
@@ -18,13 +19,6 @@ PCRE2 is configured with --enable-rebuild-chartables. However, you can run
 pcre2_dftables manually with the -L option to build tables using the LC_ALL
 locale. */
 
-/* The following #include is present because without it gcc 4.x may remove
-the array definition from the final binary if PCRE2 is built into a static
-library and dead code stripping is activated. This leads to link errors.
-Pulling in the header ensures that the array gets flagged as "someone
-outside this compilation unit might reference this" and so it will always
-be supplied to the linker. */
-
 #ifdef HAVE_CONFIG_H
 #include "config.h"
 #endif
@@ -163,7 +157,7 @@ graph, print, punct, and cntrl. Other classes are built from combinations. */
   0x02   letter
   0x04   lower case letter
   0x08   decimal digit
-  0x10   alphanumeric or '_'
+  0x10   word (alphanumeric or '_')
 */
 
   0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */