Cache d_type in directory entry

Author: nielsdos

UPGRADING.INTERNALS

@@ -67,6 +67,9 @@ PHP 8.3 INTERNALS UPGRADE NOTES
   - zend_set_user_opcode_handler
   - zend_ssa_inference
 * Removed unused macros PHP_FNV1_32A_INIT and PHP_FNV1A_64_INIT. See GH-11114.
+* _php_stream_dirent now has an extra d_type field that is used to store the
+  directory entry type. This can be used to avoid additional stat calls for
+  types when the type is already known.
 
 ========================
 2. Build system changes

Zend/zend_virtual_cwd.h

@@ -87,6 +87,16 @@ typedef unsigned short mode_t;
 #else
 #ifdef HAVE_DIRENT_H
 #include <dirent.h>
+
+#ifndef DT_UNKNOWN
+# define DT_UNKNOWN 0
+#endif
+#ifndef DT_DIR
+# define DT_DIR 4
+#endif
+#ifndef DT_REG
+# define DT_REG 8
+#endif
 #endif
 
 #define DEFAULT_SLASH '/'

ext/phar/dirstream.c

@@ -108,6 +108,7 @@ static ssize_t phar_dir_read(php_stream *stream, char *buf, size_t count) /* {{{
 	memset(buf, 0, sizeof(php_stream_dirent));
 	memcpy(((php_stream_dirent *) buf)->d_name, ZSTR_VAL(str_key), to_read);
 	((php_stream_dirent *) buf)->d_name[to_read + 1] = '\0';
+	((php_stream_dirent *) buf)->d_type = DT_UNKNOWN;
 
 	return sizeof(php_stream_dirent);
 }

ext/spl/spl_directory.c

@@ -1485,6 +1485,11 @@ PHP_METHOD(RecursiveDirectoryIterator, hasChildren)
 	if (spl_filesystem_is_invalid_or_dot(intern->u.dir.entry.d_name)) {
 		RETURN_FALSE;
 	} else {
+		if (intern->u.dir.entry.d_type == DT_DIR) {
+			RETURN_TRUE;
+		} else if (intern->u.dir.entry.d_type == DT_REG) {
+			RETURN_FALSE;
+		}
 		if (spl_filesystem_object_get_file_name(intern) == FAILURE) {
 			RETURN_THROWS();
 		}

ext/standard/ftp_fopen_wrapper.c

@@ -629,6 +629,7 @@ static ssize_t php_ftp_dirstream_read(php_stream *stream, char *buf, size_t coun
 	memcpy(ent->d_name, ZSTR_VAL(basename), tmp_len);
 	ent->d_name[tmp_len - 1] = '\0';
 	zend_string_release_ex(basename, 0);
+	ent->d_type = DT_UNKNOWN;
 
 	/* Trim off trailing whitespace characters */
 	while (tmp_len > 0 &&

main/php_streams.h

@@ -112,6 +112,7 @@ typedef struct _php_stream_dirent {
 #else
 	char d_name[MAXPATHLEN];
 #endif
+	unsigned char d_type;
 } php_stream_dirent;
 
 /* operations on streams that are file-handles */

main/streams/glob_wrapper.c

@@ -150,6 +150,7 @@ static ssize_t php_glob_stream_read(php_stream *stream, char *buf, size_t count)
 			php_glob_stream_path_split(pglob, pglob->glob.gl_pathv[index], pglob->flags & GLOB_APPEND, &path);
 			++pglob->index;
 			PHP_STRLCPY(ent->d_name, path, sizeof(ent->d_name), strlen(path));
+			ent->d_type = DT_UNKNOWN;
 			return sizeof(php_stream_dirent);
 		}
 		pglob->index = glob_result_count;

main/streams/plain_wrapper.c

@@ -1032,6 +1032,11 @@ static ssize_t php_plain_files_dirstream_read(php_stream *stream, char *buf, siz
 	result = readdir(dir);
 	if (result) {
 		PHP_STRLCPY(ent->d_name, result->d_name, sizeof(ent->d_name), strlen(result->d_name));
+#ifdef _DIRENT_HAVE_D_TYPE
+		ent->d_type = result->d_type;
+#else
+		ent->d_type = DT_UNKNOWN;
+#endif
 		return sizeof(php_stream_dirent);
 	}
 	return 0;

main/streams/userspace.c

@@ -1320,6 +1320,7 @@ static ssize_t php_userstreamop_readdir(php_stream *stream, char *buf, size_t co
 	if (call_result == SUCCESS && Z_TYPE(retval) != IS_FALSE && Z_TYPE(retval) != IS_TRUE) {
 		convert_to_string(&retval);
 		PHP_STRLCPY(ent->d_name, Z_STRVAL(retval), sizeof(ent->d_name), Z_STRLEN(retval));
+		ent->d_type = DT_UNKNOWN;
 
 		didread = sizeof(php_stream_dirent);
 	} else if (call_result == FAILURE) {

win32/readdir.c

@@ -120,6 +120,13 @@ struct dirent *readdir(DIR *dp)
 
 	dp->dent.d_ino = 1;
 	dp->dent.d_off = dp->offset;
+	if (dp->fileinfo.dwFileAttributes & (FILE_ATTRIBUTE_REPARSE_POINT | FILE_ATTRIBUTE_DEVICE)) {
+		dp->dent.d_type = DT_UNKNOWN; /* conservative */
+	} else if (dp->fileinfo.dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY) {
+		dp->dent.d_type = DT_DIR;
+	} else {
+		dp->dent.d_type = DT_REG;
+	}
 
 	return &(dp->dent);
 }/*}}}*/

win32/readdir.h

@@ -15,11 +15,17 @@ extern "C" {
 
 #include "ioutil.h"
 
+#define _DIRENT_HAVE_D_TYPE
+#define DT_UNKNOWN 0
+#define DT_DIR 4
+#define DT_REG 8
+
 /* struct dirent - same as Unix */
 struct dirent {
 	long d_ino;					/* inode (always 1 in WIN32) */
 	off_t d_off;					/* offset to this dirent */
 	unsigned short d_reclen;			/* length of d_name */
+	unsigned char d_type;
 	char d_name[1];	/* null terminated filename in the current encoding, glyph number <= 255 wchar_t's + \0 byte */
 };