AVIF support for getimagesize() and imagecreatefromstring()

Thanks to Joe Drago for help with the AVIF detection code.

Co-authored-by: Nikita Popov <nikita.ppv@googlemail.com>
Co-authored-by: Christoph M. Becker <cmbecker69@gmx.de>

Closes GH-7091.

ext/gd/gd.c

@@ -31,8 +31,10 @@
 #include <math.h>
 #include "SAPI.h"
 #include "php_gd.h"
+#include "ext/standard/php_image.h"
 #include "ext/standard/info.h"
 #include "php_open_temporary_file.h"
+#include "php_memory_streams.h"
 #include "zend_object_handlers.h"
 #include "zend_interfaces.h"
 
@@ -145,7 +147,7 @@ static void _php_image_output(INTERNAL_FUNCTION_PARAMETERS, int image_type, char
 static gdIOCtx *create_stream_context_from_zval(zval *to_zval);
 static gdIOCtx *create_stream_context(php_stream *stream, int close_stream);
 static gdIOCtx *create_output_context(void);
-static int _php_image_type(char data[12]);
+static int _php_image_type(zend_string *data);
 
 /* output streaming (formerly gd_ctx.c) */
 static void _php_image_output_ctx(INTERNAL_FUNCTION_PARAMETERS, int image_type, char *tn, void (*func_p)());
@@ -1469,42 +1471,54 @@ static int _php_ctx_getmbi(gdIOCtx *ctx)
 }
 /* }}} */
 
-/* {{{ _php_image_type */
+/* {{{ _php_image_type
+ * Based on ext/standard/image.c
+ */
 static const char php_sig_gd2[3] = {'g', 'd', '2'};
 
-static int _php_image_type (char data[12])
+static int _php_image_type(zend_string *data)
 {
-	/* Based on ext/standard/image.c */
-
-	if (data == NULL) {
+	if (ZSTR_LEN(data) < 12) {
+		/* Handle this the same way as an unknown image type. */
 		return -1;
 	}
 
-	if (!memcmp(data, php_sig_gd2, sizeof(php_sig_gd2))) {
+	if (!memcmp(ZSTR_VAL(data), php_sig_gd2, sizeof(php_sig_gd2))) {
 		return PHP_GDIMG_TYPE_GD2;
-	} else if (!memcmp(data, php_sig_jpg, sizeof(php_sig_jpg))) {
+	} else if (!memcmp(ZSTR_VAL(data), php_sig_jpg, sizeof(php_sig_jpg))) {
 		return PHP_GDIMG_TYPE_JPG;
-	} else if (!memcmp(data, php_sig_png, sizeof(php_sig_png))) {
+	} else if (!memcmp(ZSTR_VAL(data), php_sig_png, sizeof(php_sig_png))) {
 		return PHP_GDIMG_TYPE_PNG;
-	} else if (!memcmp(data, php_sig_gif, sizeof(php_sig_gif))) {
+	} else if (!memcmp(ZSTR_VAL(data), php_sig_gif, sizeof(php_sig_gif))) {
 		return PHP_GDIMG_TYPE_GIF;
-	} else if (!memcmp(data, php_sig_bmp, sizeof(php_sig_bmp))) {
+	} else if (!memcmp(ZSTR_VAL(data), php_sig_bmp, sizeof(php_sig_bmp))) {
 		return PHP_GDIMG_TYPE_BMP;
-	} else if(!memcmp(data, php_sig_riff, sizeof(php_sig_riff)) && !memcmp(data + sizeof(php_sig_riff) + sizeof(uint32_t), php_sig_webp, sizeof(php_sig_webp))) {
+	} else if(!memcmp(ZSTR_VAL(data), php_sig_riff, sizeof(php_sig_riff)) && !memcmp(ZSTR_VAL(data) + sizeof(php_sig_riff) + sizeof(uint32_t), php_sig_webp, sizeof(php_sig_webp))) {
 		return PHP_GDIMG_TYPE_WEBP;
 	}
-	else {
-		gdIOCtx *io_ctx;
-		io_ctx = gdNewDynamicCtxEx(8, data, 0);
-		if (io_ctx) {
-			if (_php_ctx_getmbi(io_ctx) == 0 && _php_ctx_getmbi(io_ctx) >= 0) {
-				io_ctx->gd_free(io_ctx);
-				return PHP_GDIMG_TYPE_WBM;
-			} else {
-				io_ctx->gd_free(io_ctx);
-			}
+
+	php_stream *image_stream = php_stream_memory_open(TEMP_STREAM_READONLY, data);
+
+	if (image_stream != NULL) {
+		bool is_avif = php_is_image_avif(image_stream);
+		php_stream_close(image_stream);
+
+		if (is_avif) {
+			return PHP_GDIMG_TYPE_AVIF;
 		}
 	}
+
+	gdIOCtx *io_ctx;
+	io_ctx = gdNewDynamicCtxEx(8, ZSTR_VAL(data), 0);
+	if (io_ctx) {
+		if (_php_ctx_getmbi(io_ctx) == 0 && _php_ctx_getmbi(io_ctx) >= 0) {
+			io_ctx->gd_free(io_ctx);
+			return PHP_GDIMG_TYPE_WBM;
+		} else {
+			io_ctx->gd_free(io_ctx);
+		}
+	}
+
 	return -1;
 }
 /* }}} */
@@ -1540,21 +1554,12 @@ PHP_FUNCTION(imagecreatefromstring)
 	zend_string *data;
 	gdImagePtr im;
 	int imtype;
-	char sig[12];
 
 	if (zend_parse_parameters(ZEND_NUM_ARGS(), "S", &data) == FAILURE) {
 		RETURN_THROWS();
 	}
 
-	if (ZSTR_LEN(data) < sizeof(sig)) {
-		/* Handle this the same way as an unknown image type. */
-		php_error_docref(NULL, E_WARNING, "Data is not in a recognized format");
-		RETURN_FALSE;
-	}
-
-	memcpy(sig, ZSTR_VAL(data), sizeof(sig));
-
-	imtype = _php_image_type(sig);
+	imtype = _php_image_type(data);
 
 	switch (imtype) {
 		case PHP_GDIMG_TYPE_JPG:

ext/gd/libgd/gd.h

@@ -628,7 +628,6 @@ gdImagePtr gdImageCreateFromGif(FILE *fd);
 gdImagePtr gdImageCreateFromGifCtx(gdIOCtxPtr in);
 gdImagePtr gdImageCreateFromGifSource(gdSourcePtr in);
 
-//TODO: we may not need all of these
 void gdImageAvif(gdImagePtr im, FILE *outfile);
 void gdImageAvifEx(gdImagePtr im, FILE *outfile, int quality, int speed);
 void *gdImageAvifPtr(gdImagePtr im, int *size);

ext/gd/php_gd.h

@@ -18,6 +18,9 @@
 #ifndef PHP_GD_H
 #define PHP_GD_H
 
+#include "zend_string.h"
+#include "php_streams.h"
+
 #if defined(HAVE_LIBGD) || defined(HAVE_GD_BUNDLED)
 
 /* open_basedir and safe_mode checks */

ext/gd/tests/imagecreatefromstring_avif.phpt

@@ -0,0 +1,29 @@
+--TEST--
+imagecreatefromstring() - AVIF format
+--EXTENSIONS--
+gd
+--SKIPIF--
+<?php
+  if (!(imagetypes() & IMG_AVIF)) {
+    die('skip AVIF support required');
+  }
+?>
+--FILE--
+<?php
+  echo "Reading image whose major brand is 'avif':\n";
+  $im = imagecreatefromstring(file_get_contents(__DIR__ . '/imagecreatefromstring_major_brand.avif'));
+  var_dump(imagesx($im));
+  var_dump(imagesy($im));
+
+  echo "Reading image with a compatible brand that's 'avif':\n";
+  $im = imagecreatefromstring(file_get_contents(__DIR__ . '/imagecreatefromstring_compatible_brand.avif'));
+  var_dump(imagesx($im));
+  var_dump(imagesy($im));
+?>
+--EXPECT--
+Reading image whose major brand is 'avif':
+int(250)
+int(375)
+Reading image with a compatible brand that's 'avif':
+int(480)
+int(270)

ext/standard/image.c

@@ -87,7 +87,8 @@ PHP_MINIT_FUNCTION(imagetypes)
 	REGISTER_LONG_CONSTANT("IMAGETYPE_JPEG2000",IMAGE_FILETYPE_JPC,     CONST_CS | CONST_PERSISTENT); /* keep alias */
 	REGISTER_LONG_CONSTANT("IMAGETYPE_XBM",     IMAGE_FILETYPE_XBM,     CONST_CS | CONST_PERSISTENT);
 	REGISTER_LONG_CONSTANT("IMAGETYPE_ICO",     IMAGE_FILETYPE_ICO,     CONST_CS | CONST_PERSISTENT);
-	REGISTER_LONG_CONSTANT("IMAGETYPE_WEBP",	IMAGE_FILETYPE_WEBP,	CONST_CS | CONST_PERSISTENT);
+	REGISTER_LONG_CONSTANT("IMAGETYPE_WEBP",    IMAGE_FILETYPE_WEBP,    CONST_CS | CONST_PERSISTENT);
+	REGISTER_LONG_CONSTANT("IMAGETYPE_AVIF",    IMAGE_FILETYPE_AVIF,    CONST_CS | CONST_PERSISTENT);
 	REGISTER_LONG_CONSTANT("IMAGETYPE_UNKNOWN", IMAGE_FILETYPE_UNKNOWN, CONST_CS | CONST_PERSISTENT);
 	REGISTER_LONG_CONSTANT("IMAGETYPE_COUNT",   IMAGE_FILETYPE_COUNT,   CONST_CS | CONST_PERSISTENT);
 	return SUCCESS;
@@ -1148,6 +1149,99 @@ static struct gfxinfo *php_handle_webp(php_stream * stream)
 }
 /* }}} */
 
+/* {{{ php_handle_avif
+ * There's no simple way to get this information - so, for now, this is unsupported.
+ * Simply return 0 for everything.
+ */
+static struct gfxinfo *php_handle_avif(php_stream * stream) {
+	return ecalloc(1, sizeof(struct gfxinfo));
+}
+/* }}} */
+
+/* {{{ php_ntohl
+ * Convert a big-endian network uint32 to host order - 
+ * which may be either little-endian or big-endian.
+ * Thanks to Rob Pike via Joe Drago:
+ * https://commandcenter.blogspot.nl/2012/04/byte-order-fallacy.html
+ */
+static uint32_t php_ntohl(uint32_t val) {
+	uint8_t data[4];
+
+	memcpy(&data, &val, sizeof(data));
+	return ((uint32_t)data[3] << 0) |
+		((uint32_t)data[2] << 8) |
+		((uint32_t)data[1] << 16) |
+		((uint32_t)data[0] << 24);
+}
+/* }}} */
+
+/* {{{ php_is_image_avif
+ * detect whether an image is of type AVIF
+ * 
+ * An AVIF image will start off a header "box".
+ * This starts with with a four-byte integer containing the number of bytes in the filetype box.
+ * This must be followed by the string "ftyp".
+ * Next comes a four-byte string indicating the "major brand".
+ * If that's "avif" or "avis", this is an AVIF image.
+ * Next, there's a four-byte "minor version" field, which we can ignore.
+ * Next comes an array of four-byte strings containing "compatible brands".
+ * These extend to the end of the box.
+ * If any of the compatible brands is "avif" or "avis", then this is an AVIF image.
+ * Otherwise, well, it's not.
+ * For more, see https://mpeg.chiariglione.org/standards/mpeg-4/iso-base-media-file-format/text-isoiec-14496-12-5th-edition
+ */
+bool php_is_image_avif(php_stream * stream) {
+	uint32_t header_size_reversed, header_size, i;
+	char box_type[4], brand[4];
+
+	ZEND_ASSERT(stream != NULL);
+
+	if (php_stream_read(stream, (char *) &header_size_reversed, 4) != 4) {
+		return 0;
+	}
+
+	header_size = php_ntohl(header_size_reversed);
+
+	/* If the box type isn't "ftyp", it can't be an AVIF image. */
+	if (php_stream_read(stream, box_type, 4) != 4) {
+		return 0;
+	}
+
+	if (memcmp(box_type, "ftyp", 4)) {
+		return 0;
+	}
+
+	/* If the major brand is "avif" or "avis", it's an AVIF image. */
+	if (php_stream_read(stream, brand, 4) != 4) {
+		return 0;
+	}
+
+	if (!memcmp(brand, "avif", 4) || !memcmp(brand, "avis", 4)) {
+		return 1;
+	}
+
+	/* Skip the next four bytes, which are the "minor version". */
+	if (php_stream_read(stream, brand, 4) != 4) {
+		return 0;
+	}
+
+	/* Look for "avif" or "avis" in any member of compatible_brands[], to the end of the header.
+	   Note we've already read four groups of four bytes. */
+
+	for (i = 16; i < header_size; i += 4) {
+		if (php_stream_read(stream, brand, 4) != 4) {
+			return 0;
+		}
+
+		if (!memcmp(brand, "avif", 4) || !memcmp(brand, "avis", 4)) {
+			return 1;
+		}
+	}
+
+	return 0;
+}
+/* }}} */
+
 /* {{{ php_image_type_to_mime_type
  * Convert internal image_type to mime type */
 PHPAPI char * php_image_type_to_mime_type(int image_type)
@@ -1183,6 +1277,8 @@ PHPAPI char * php_image_type_to_mime_type(int image_type)
 			return "image/vnd.microsoft.icon";
 		case IMAGE_FILETYPE_WEBP:
 			return "image/webp";
+		case IMAGE_FILETYPE_AVIF:
+			return "image/avif";
 		default:
 		case IMAGE_FILETYPE_UNKNOWN:
 			return "application/octet-stream"; /* suppose binary format */
@@ -1265,6 +1361,9 @@ PHP_FUNCTION(image_type_to_extension)
 		case IMAGE_FILETYPE_WEBP:
 			imgext = ".webp";
 			break;
+		case IMAGE_FILETYPE_AVIF:
+			imgext = ".avif";
+			break;
 	}
 
 	if (imgext) {
@@ -1277,7 +1376,7 @@ PHP_FUNCTION(image_type_to_extension)
 
 /* {{{ php_imagetype
    detect filetype from first bytes */
-PHPAPI int php_getimagetype(php_stream * stream, const char *input, char *filetype)
+PHPAPI int php_getimagetype(php_stream *stream, const char *input, char *filetype)
 {
 	char tmp[12];
 	int twelve_bytes_read;
@@ -1349,17 +1448,24 @@ PHPAPI int php_getimagetype(php_stream * stream, const char *input, char *filety
 		return IMAGE_FILETYPE_JP2;
 	}
 
+	if (!php_stream_rewind(stream) && php_is_image_avif(stream)) {
+		return IMAGE_FILETYPE_AVIF;
+	}
+
 /* AFTER ALL ABOVE FAILED */
 	if (php_get_wbmp(stream, NULL, 1)) {
 		return IMAGE_FILETYPE_WBMP;
 	}
+
 	if (!twelve_bytes_read) {
 		php_error_docref(NULL, E_NOTICE, "Error reading from %s!", input);
 		return IMAGE_FILETYPE_UNKNOWN;
 	}
+
 	if (php_get_xbm(stream, NULL)) {
 		return IMAGE_FILETYPE_XBM;
 	}
+
 	return IMAGE_FILETYPE_UNKNOWN;
 }
 /* }}} */
@@ -1431,6 +1537,9 @@ static void php_getimagesize_from_stream(php_stream *stream, char *input, zval *
 		case IMAGE_FILETYPE_WEBP:
 			result = php_handle_webp(stream);
 			break;
+		case IMAGE_FILETYPE_AVIF:
+			result = php_handle_avif(stream);
+			break;
 		default:
 		case IMAGE_FILETYPE_UNKNOWN:
 			break;

ext/standard/php_image.h

@@ -43,6 +43,7 @@ typedef enum
   IMAGE_FILETYPE_XBM,
   IMAGE_FILETYPE_ICO,
   IMAGE_FILETYPE_WEBP,
+  IMAGE_FILETYPE_AVIF,
 /* WHEN EXTENDING: PLEASE ALSO REGISTER IN image.c:PHP_MINIT_FUNCTION(imagetypes) */
   IMAGE_FILETYPE_COUNT
 } image_filetype;
@@ -54,4 +55,6 @@ PHPAPI int php_getimagetype(php_stream *stream, const char *input, char *filetyp
 
 PHPAPI char * php_image_type_to_mime_type(int image_type);
 
+PHPAPI bool php_is_image_avif(php_stream *stream);
+
 #endif /* PHP_IMAGE_H */

ext/standard/tests/image/getimagesize.phpt

@@ -23,7 +23,7 @@ GetImageSize()
     var_dump($result);
 ?>
 --EXPECT--
-array(16) {
+array(17) {
   ["test-1pix.bmp"]=>
   array(6) {
     [0]=>
@@ -69,6 +69,19 @@ array(16) {
     ["mime"]=>
     string(9) "image/bmp"
   }
+  ["test1pix.avif"]=>
+  array(5) {
+    [0]=>
+    int(0)
+    [1]=>
+    int(0)
+    [2]=>
+    int(19)
+    [3]=>
+    string(20) "width="0" height="0""
+    ["mime"]=>
+    string(10) "image/avif"
+  }
   ["test1pix.bmp"]=>
   array(6) {
     [0]=>