Merge branch 'PHP-8.3' into PHP-8.4

* PHP-8.3:
  NEWS for GH-17168
  ext/gettext/config.m4: symlink en_US.UTF-8 test bits to en_US for musl
  ext/gettext/tests: fix libintl return values under musl
  ext/gettext/gettext.c: handle NULLs from bindtextdomain()

.gitignore

@@ -177,6 +177,9 @@ php
 /ext/*/configure.ac
 /ext/*/run-tests.php
 
+# Generated by ./configure if libc might be musl
+/ext/gettext/tests/locale/en_US
+
 # ------------------------------------------------------------------------------
 # Generated by Windows build system
 # ------------------------------------------------------------------------------

NEWS

@@ -44,6 +44,10 @@ PHP                                                                        NEWS
   . Fixed bug GH-16255 (Unexpected nan value in ext/gd/libgd/gd_filter.c).
     (nielsdos, cmb)
 
+- Gettext:
+  . Fixed bug GH-17202 (Segmentation fault ext/gettext/gettext.c
+    bindtextdomain()). (Michael Orlitzky)
+
 - Iconv:
   . Fixed bug GH-17047 (UAF on iconv filter failure). (nielsdos)
 

ext/gettext/config.m4

@@ -23,6 +23,18 @@ if test "$PHP_GETTEXT" != "no"; then
     [AC_CHECK_LIB([c], [bindtextdomain], [
       GETTEXT_LIBS=
       GETTEXT_CHECK_IN_LIB=c
+
+      dnl If libintl.h is provided by libc, it's possible that libc is musl.
+      dnl The gettext family of functions under musl ignores the codeset
+      dnl suffix on directories like "en_US.UTF-8"; instead they look only
+      dnl in "en_US". To accomodate that, we symlink some test data from one
+      dnl to the other.
+      AC_MSG_NOTICE([symlinking en_US.UTF-8 messages to en_US in case you are on musl])
+      _linkdest="${srcdir%/}"/ext/gettext/tests/locale/en_US
+      AS_IF([test ! -e "${_linkdest}"],[
+        ln -s en_US.UTF-8 "${_linkdest}"
+      ])
+
       ],
       [AC_MSG_FAILURE([Unable to find required intl library for gettext.])])])
 

ext/gettext/gettext.c

@@ -180,7 +180,7 @@ PHP_FUNCTION(dcgettext)
 PHP_FUNCTION(bindtextdomain)
 {
 	zend_string *domain, *dir = NULL;
-	char *retval, dir_name[MAXPATHLEN];
+	char *retval, dir_name[MAXPATHLEN], *btd_result;
 
 	ZEND_PARSE_PARAMETERS_START(1, 2)
 		Z_PARAM_STR(domain)
@@ -191,7 +191,16 @@ PHP_FUNCTION(bindtextdomain)
 	PHP_GETTEXT_DOMAIN_LENGTH_CHECK(1, ZSTR_LEN(domain))
 
 	if (dir == NULL) {
-		RETURN_STRING(bindtextdomain(ZSTR_VAL(domain), NULL));
+		btd_result = bindtextdomain(ZSTR_VAL(domain), NULL);
+		if (btd_result == NULL) {
+			/* POSIX-compliant implementations can return
+			 * NULL if an error occured. On musl you will
+			 * also get NULL if the domain is not yet
+			 * bound, because musl has no default directory
+			 * to return in that case. */
+			 RETURN_FALSE;
+		}
+		RETURN_STRING(btd_result);
 	}
 
 	if (ZSTR_LEN(dir) != 0 && !zend_string_equals_literal(dir, "0")) {

ext/gettext/tests/bug53251.phpt

@@ -8,18 +8,28 @@ if (getenv('SKIP_REPEAT')) die('skip gettext leaks global state across requests'
 ?>
 --FILE--
 <?php
-var_dump(is_string(bindtextdomain('foo', null)));
+$results[] = is_string(bindtextdomain('foo', null));
 $dir = bindtextdomain('foo', '.');
-var_dump(bindtextdomain('foo', null) === $dir);
+$results[] = bindtextdomain('foo', null) === $dir;
+$results[] = bind_textdomain_codeset('foo', null);
+$results[] = bind_textdomain_codeset('foo', 'UTF-8');
+$results[] = bind_textdomain_codeset('foo', null);
 
-var_dump(bind_textdomain_codeset('foo', null));
+$expected = [true, true, false, "UTF-8", "UTF-8"];
-var_dump(bind_textdomain_codeset('foo', 'UTF-8'));
-var_dump(bind_textdomain_codeset('foo', null));
 
+// musl's bindtextdomain() has no default directory to return when
+// "foo" is unbound, so in the first call, you will get false instead
+// of a string.
+//
+// bind_textdomain_codeset() always returns false on musl
+// because musl only supports UTF-8. For more information:
+//
+//   * https://github.com/php/doc-en/issues/4311,
+//   * https://github.com/php/php-src/issues/17163
+//
+$expected_musl = [false, true, false, false, false];
+
+var_dump($results === $expected || $results === $expected_musl);
 ?>
 --EXPECT--
 bool(true)
-bool(true)
-bool(false)
-string(5) "UTF-8"
-string(5) "UTF-8"

ext/gettext/tests/gettext_bind_textdomain_codeset-retval.phpt

@@ -15,14 +15,22 @@ gettext
     } catch (ValueError $e) {
 	    echo $e->getMessage() . PHP_EOL;
     }
-    var_dump(bind_textdomain_codeset('messages', "UTF-8"));
+
+    // bind_textdomain_codeset() always returns false on musl
+    // because musl only supports UTF-8. For more information:
+    //
+    //   * https://github.com/php/doc-en/issues/4311,
+    //   * https://github.com/php/php-src/issues/17163
+    //
+    $result = bind_textdomain_codeset('messages', "UTF-8");
+    var_dump($result === false || $result === "UTF-8");
 
     echo "Done\n";
 ?>
 --EXPECT--
 bind_textdomain_codeset(): Argument #1 ($domain) must not be empty
 bind_textdomain_codeset(): Argument #1 ($domain) must not be empty
-string(5) "UTF-8"
+bool(true)
 Done
 --CREDITS--
 Florian Holzhauer fh-pt@fholzhauer.de