create embed test, verify argv always passed as valid utf8

Author: henderkes

sapi/cli/php_cli.c

@@ -1207,6 +1207,18 @@ PHP_CLI_API int do_php_cli(int argc, char *argv[])
 	 */
 	argv = save_ps_args(argc, argv);
 
+#ifdef PHP_WIN32
+	if (argv_save != __argv) {
+		for (int i = 0; i < argc; i++) {
+			if (!MultiByteToWideChar(CP_UTF8, MB_ERR_INVALID_CHARS, argv[i], -1, NULL, 0)) {
+				fprintf(stderr, "Invalid UTF-8 in command line argument %d.\n", i);
+				cleanup_ps_args(argv);
+				return 1;
+			}
+		}
+	}
+#endif
+
 #if defined(PHP_WIN32) && !defined(PHP_CLI_WIN32_NO_CONSOLE)
 	php_win32_console_fileno_set_vt100(STDOUT_FILENO, TRUE);
 	php_win32_console_fileno_set_vt100(STDERR_FILENO, TRUE);

sapi/embed/Makefile.frag.w32

@@ -0,0 +1,8 @@
+test_helpers: $(BUILD_DIR)\php_embed_cli_test.exe
+
+$(BUILD_DIR)\sapi\embed\cli_argv.obj: $(PHP_SRC_DIR)\sapi\embed\tests\cli_argv.c
+	cd $(PHP_SRC_DIR)
+	$(CC) $(CFLAGS_EMBED_OBJ) $(CFLAGS) $(CFLAGS_BD_SAPI_EMBED) /Fo$@ /c $(PHP_SRC_DIR)\sapi\embed\tests\cli_argv.c
+
+$(BUILD_DIR)\php_embed_cli_test.exe: $(BUILD_DIR)\sapi\embed\cli_argv.obj $(BUILD_DIR)\$(EMBED_LIB)
+	"$(LINK)" /nologo $(BUILD_DIR)\sapi\embed\cli_argv.obj $(BUILD_DIR)\$(EMBED_LIB) /out:$@ $(LDFLAGS) $(LDFLAGS_EMBED) /stack:67108864

sapi/embed/config.w32

@@ -10,4 +10,6 @@ if (PHP_EMBED != "no") {
 	ADD_FLAG("LIBS_EMBED", "ws2_32.lib");
 	ADD_FLAG("LIBS_EMBED", "shell32.lib");
 	PHP_INSTALL_HEADERS("sapi/embed", "php_embed.h");
+	DEFINE('EMBED_LIB', 'php' + PHP_VERSION + 'embed.lib');
+	ADD_MAKEFILE_FRAGMENT();
 }

sapi/embed/tests/cli_argv.c

@@ -0,0 +1,26 @@
+#include "php.h"
+#include "sapi/cli/cli.h"
+
+int main(int argc, char *argv[])
+{
+	char *php_argv[] = {
+		"embedded-php", "-n", "-d", "default_charset=UTF-8", "-r",
+		"echo json_encode([$argc, array_map('bin2hex', array_slice($argv, 1))]), PHP_EOL; exit(23);",
+		"--", "caf\xc3\xa9", "argument with spaces", "", "\xf0\x9f\x98\x80", NULL
+	};
+	if (argc > 1 && strcmp(argv[1], "default-charset") == 0) {
+		php_argv[3] = "default_charset=Windows-1252";
+		php_argv[10] = "\xe2\x82\xac";
+	} else if (argc > 1 && strcmp(argv[1], "internal-encoding") == 0) {
+		php_argv[3] = "internal_encoding=Windows-1252";
+		php_argv[10] = "\xe2\x82\xac";
+	}
+	if (argc > 2 && strcmp(argv[2], "file") == 0) {
+		php_argv[4] = "-f";
+		php_argv[5] = "cli_argv_caf\xc3\xa9.php";
+	} else if (argc > 2 && strcmp(argv[2], "invalid-utf8") == 0) {
+		/* Latin-1 cafe with an accented e, which is not valid UTF-8. */
+		php_argv[7] = "caf\xe9";
+	}
+	return do_php_cli((int) (sizeof(php_argv) / sizeof(*php_argv)) - 1, php_argv);
+}

sapi/embed/tests/cli_argv.phpt

@@ -0,0 +1,42 @@
+--TEST--
+do_php_cli() validates and converts custom UTF-8 argv on Windows
+--SKIPIF--
+<?php
+if (PHP_OS_FAMILY !== 'Windows') die('skip Windows only');
+if (!file_exists(dirname(PHP_BINARY) . '/php_embed_cli_test.exe')) {
+    die('skip embed SAPI test host is not built (requires --enable-embed)');
+}
+?>
+--FILE--
+<?php
+$host = dirname(PHP_BINARY) . '/php_embed_cli_test.exe';
+foreach (['ignored-host-argument', 'default-charset', 'internal-encoding'] as $mode) {
+    passthru(escapeshellarg($host) . ' ' . $mode, $status);
+    var_dump($status);
+    passthru(escapeshellarg($host) . ' ' . $mode . ' invalid-utf8 2>&1', $status);
+    var_dump($status);
+}
+chdir(__DIR__);
+file_put_contents("cli_argv_caf\xc3\xa9.php", '<?php echo "Unicode filename works\n";');
+passthru(escapeshellarg($host) . ' default-charset file', $status);
+var_dump($status);
+?>
+--CLEAN--
+<?php
+unlink(__DIR__ . "/cli_argv_caf\xc3\xa9.php");
+?>
+--EXPECT--
+[5,["636166c3a9","617267756d656e74207769746820737061636573","","f09f9880"]]
+int(23)
+Invalid UTF-8 in command line argument 7.
+int(1)
+[5,["636166e9","617267756d656e74207769746820737061636573","","80"]]
+int(23)
+Invalid UTF-8 in command line argument 7.
+int(1)
+[5,["636166e9","617267756d656e74207769746820737061636573","","80"]]
+int(23)
+Invalid UTF-8 in command line argument 7.
+int(1)
+Unicode filename works
+int(0)