2 * JPEG XL Header Parser
3 * Copyright (c) 2023 Leo Izen <leo.izen@gmail.com>
5 * This file is part of FFmpeg.
7 * FFmpeg is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
12 * FFmpeg is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
24 #include "bytestream.h"
25 #define UNCHECKED_BITSTREAM_READER 0
26 #define BITSTREAM_READER_LE
29 #include "jpegxl_parse.h"
31 /* read a U32(c_i + u(u_i)) */
32 static av_always_inline uint32_t jxl_u32(GetBitContext *gb,
33 uint32_t c0, uint32_t c1, uint32_t c2, uint32_t c3,
34 uint32_t u0, uint32_t u1, uint32_t u2, uint32_t u3)
36 const uint32_t constants[4] = {c0, c1, c2, c3};
37 const uint32_t ubits [4] = {u0, u1, u2, u3};
38 uint32_t ret, choice = get_bits(gb, 2);
40 ret = constants[choice];
42 ret += get_bits_long(gb, ubits[choice]);
47 static av_always_inline uint32_t jxl_enum(GetBitContext *gb)
49 return jxl_u32(gb, 0, 1, 2, 18, 0, 0, 4, 6);
53 static uint64_t jxl_u64(GetBitContext *gb)
55 uint64_t shift = 12, ret;
57 switch (get_bits(gb, 2)) {
59 ret = 1 + get_bits(gb, 4);
62 ret = 17 + get_bits(gb, 8);
65 ret = get_bits(gb, 12);
66 while (get_bits1(gb)) {
68 ret |= (uint64_t)get_bits(gb, 8) << shift;
71 ret |= (uint64_t)get_bits(gb, 4) << shift;
83 static uint32_t jpegxl_width_from_ratio(uint32_t height, int ratio)
85 uint64_t height64 = height; /* avoid integer overflow */
90 return (uint32_t)((height64 * 12) / 10);
92 return (uint32_t)((height64 * 4) / 3);
94 return (uint32_t)((height64 * 3) / 2);
96 return (uint32_t)((height64 * 16) / 9);
98 return (uint32_t)((height64 * 5) / 4);
100 return (uint32_t)(height64 * 2);
105 return 0; /* manual width */
109 * validate a Jpeg XL Size Header
110 * @return >= 0 upon valid size, < 0 upon invalid size found
112 static int jpegxl_read_size_header(GetBitContext *gb, FFJXLMetadata *meta, int validate)
114 uint32_t width, height;
117 /* small size header */
118 height = (get_bits(gb, 5) + 1) << 3;
119 width = jpegxl_width_from_ratio(height, get_bits(gb, 3));
121 width = (get_bits(gb, 5) + 1) << 3;
123 /* large size header */
124 height = 1 + jxl_u32(gb, 0, 0, 0, 0, 9, 13, 18, 30);
125 width = jpegxl_width_from_ratio(height, get_bits(gb, 3));
127 width = 1 + jxl_u32(gb, 0, 0, 0, 0, 9, 13, 18, 30);
129 if (validate && (width > (1 << 18) || height > (1 << 18)
130 || (width >> 4) * (height >> 4) > (1 << 20)))
131 return AVERROR_INVALIDDATA;
134 meta->width = meta->coded_width = width;
135 meta->height = meta->coded_height = height;
142 * validate a Jpeg XL Preview Header
143 * @return >= 0 upon valid size, < 0 upon invalid size found
145 static int jpegxl_read_preview_header(GetBitContext *gb, int validate)
147 uint32_t width, height;
150 /* coded height and width divided by eight */
151 height = jxl_u32(gb, 16, 32, 1, 33, 0, 0, 5, 9) << 3;
152 width = jpegxl_width_from_ratio(height, get_bits(gb, 3));
154 width = jxl_u32(gb, 16, 32, 1, 33, 0, 0, 5, 9) << 3;
156 /* full height and width coded */
157 height = jxl_u32(gb, 1, 65, 321, 1345, 6, 8, 10, 12);
158 width = jpegxl_width_from_ratio(height, get_bits(gb, 3));
160 width = jxl_u32(gb, 1, 65, 321, 1345, 6, 8, 10, 12);
162 if (validate && (width > 4096 || height > 4096))
163 return AVERROR_INVALIDDATA;
169 * get a Jpeg XL BitDepth Header. These cannot be invalid.
171 static void jpegxl_get_bit_depth(GetBitContext *gb, FFJXLMetadata *meta)
176 bit_depth = jxl_u32(gb, 32, 16, 24, 1, 0, 0, 0, 6); /* mantissa */
177 skip_bits_long(gb, 4); /* exponent */
179 /* integer samples */
180 bit_depth = jxl_u32(gb, 8, 10, 12, 1, 0, 0, 0, 6);
183 meta->bit_depth = bit_depth;
187 * validate a Jpeg XL Extra Channel Info bundle
188 * @return >= 0 upon valid, < 0 upon invalid
190 static int jpegxl_read_extra_channel_info(GetBitContext *gb, FFJXLMetadata *meta, int validate)
192 int default_alpha = get_bits1(gb);
193 uint32_t type, name_len = 0;
195 if (!default_alpha) {
197 if (validate && type > 63)
198 return AVERROR_INVALIDDATA; /* enum types cannot be 64+ */
199 if (validate && validate < 10 && type == JPEGXL_CT_BLACK)
200 return AVERROR_INVALIDDATA;
201 jpegxl_get_bit_depth(gb, NULL);
202 jxl_u32(gb, 0, 3, 4, 1, 0, 0, 0, 3); /* dim-shift */
203 /* max of name_len is 1071 = 48 + 2^10 - 1 */
204 name_len = 8 * jxl_u32(gb, 0, 0, 16, 48, 0, 4, 5, 10);
206 type = JPEGXL_CT_ALPHA;
209 if (get_bits_left(gb) < name_len)
210 return AVERROR_BUFFER_TOO_SMALL;
212 /* skip over the name */
213 skip_bits_long(gb, name_len);
215 if (!default_alpha && type == JPEGXL_CT_ALPHA)
218 if (type == JPEGXL_CT_SPOT_COLOR)
219 skip_bits_long(gb, 16 * 4);
221 if (type == JPEGXL_CT_CFA)
222 jxl_u32(gb, 1, 0, 3, 19, 0, 2, 4, 8);
224 if (meta && type == JPEGXL_CT_ALPHA)
225 meta->have_alpha = 1;
230 static int jpegxl_skip_extensions(GetBitContext *gb)
232 uint64_t extensions = jxl_u64(gb), extensions_len = 0;
234 if (get_bits_left(gb) <= 0)
235 return AVERROR_BUFFER_TOO_SMALL;
240 for (int i = 0; i < 64; i++) {
241 if (extensions & (UINT64_C(1) << i))
242 extensions_len += jxl_u64(gb);
243 if (get_bits_left(gb) <= 0)
244 return AVERROR_BUFFER_TOO_SMALL;
247 if (extensions_len > INT_MAX || get_bits_left(gb) <= extensions_len)
248 return AVERROR_BUFFER_TOO_SMALL;
250 skip_bits_long(gb, extensions_len);
255 int ff_jpegxl_parse_codestream_header(const uint8_t *buf, int buflen, FFJXLMetadata *meta, int validate)
257 GetBitContext gbi, *gb = &gbi;
259 int all_default, extra_fields = 0;
260 int xyb_encoded = 1, have_icc_profile = 0;
261 int animation_offset = 0, have_timecodes = 0;
263 FFJXLPrimaries primaries = JPEGXL_PR_SRGB;
264 FFJXLTransferCharacteristic trc = JPEGXL_TR_SRGB + (1U << 24);
265 FFJXLWhitePoint white_point = JPEGXL_WP_D65;
266 FFJXLColorSpace color_space = JPEGXL_CS_RGB;
269 uint32_t num_extra_channels = 0;
272 ret = init_get_bits8(gb, buf, buflen);
276 if (get_bits(gb, 16) != FF_JPEGXL_CODESTREAM_SIGNATURE_LE && validate)
277 return AVERROR_INVALIDDATA;
279 ret = jpegxl_read_size_header(gb, meta, validate);
283 all_default = get_bits1(gb);
285 extra_fields = get_bits1(gb);
288 int orientation = get_bits(gb, 3);
289 if (orientation > 3 && meta)
290 FFSWAP(uint32_t, meta->width, meta->height);
294 * any size header here is valid, but as it
295 * is variable length we have to read it
298 jpegxl_read_size_header(gb, NULL, 0);
302 ret = jpegxl_read_preview_header(gb, 0);
307 /* animation header */
309 animation_offset = get_bits_count(gb);
310 tb.den = jxl_u32(gb, 100, 1000, 1, 1, 0, 0, 10, 30);
311 tb.num = jxl_u32(gb, 1, 1001, 1, 1, 0, 0, 8, 10);
312 jxl_u32(gb, 0, 0, 0, 0, 0, 3, 16, 32);
313 have_timecodes = get_bits1(gb);
317 if (animation_offset && meta) {
318 meta->animation_offset = animation_offset;
320 meta->have_timecodes = have_timecodes;
323 if (get_bits_left(gb) <= 0)
324 return AVERROR_BUFFER_TOO_SMALL;
327 jpegxl_get_bit_depth(gb, meta);
329 /* modular_16bit_buffers must equal 1 */
330 if (!get_bits1(gb) && validate && validate < 10)
331 return AVERROR_INVALIDDATA;
333 num_extra_channels = jxl_u32(gb, 0, 1, 2, 1, 0, 0, 4, 12);
334 if (num_extra_channels > 4 && validate && validate < 10)
335 return AVERROR_INVALIDDATA;
336 for (uint32_t i = 0; i < num_extra_channels; i++) {
337 ret = jpegxl_read_extra_channel_info(gb, meta, validate);
340 if (get_bits_left(gb) <= 0)
341 return AVERROR_BUFFER_TOO_SMALL;
344 xyb_encoded = get_bits1(gb);
346 /* color encoding bundle */
347 if (!get_bits1(gb)) {
348 have_icc_profile = get_bits1(gb);
349 color_space = jxl_enum(gb);
350 if (color_space > 63 && validate)
351 return AVERROR_INVALIDDATA;
352 if (!have_icc_profile) {
353 if (color_space != JPEGXL_CS_XYB) {
354 white_point = jxl_enum(gb);
355 if (white_point > 63 && validate)
356 return AVERROR_INVALIDDATA;
357 if (white_point == JPEGXL_WP_CUSTOM) {
358 /* ux and uy values */
359 jxl_u32(gb, 0, 524288, 1048576, 2097152, 19, 19, 20, 21);
360 jxl_u32(gb, 0, 524288, 1048576, 2097152, 19, 19, 20, 21);
362 if (color_space != JPEGXL_CS_GRAY) {
364 primaries = jxl_enum(gb);
365 if (primaries > 63 && validate)
366 return AVERROR_INVALIDDATA;
367 if (primaries == JPEGXL_PR_CUSTOM) {
368 /* ux/uy values for r,g,b */
369 for (int i = 0; i < 6; i++) {
370 jxl_u32(gb, 0, 524288, 1048576, 2097152, 19, 19, 20, 21);
371 if (get_bits_left(gb) <= 0)
372 return AVERROR_BUFFER_TOO_SMALL;
378 /* transfer characteristics */
381 trc = get_bits(gb, 24);
383 /* transfer function */
385 if (trc > 63 && validate)
386 return AVERROR_INVALIDDATA;
390 /* rendering intent */
391 if (jxl_enum(gb) > 63 && validate)
392 return AVERROR_INVALIDDATA;
396 /* tone mapping bundle */
397 if (extra_fields && !get_bits1(gb))
398 skip_bits_long(gb, 16 + 16 + 1 + 16);
400 ret = jpegxl_skip_extensions(gb);
406 meta->xyb_encoded = xyb_encoded;
407 meta->have_icc_profile = have_icc_profile;
408 meta->csp = color_space;
409 meta->primaries = primaries;
410 meta->wp = white_point;
412 if (!meta->bit_depth)
414 meta->num_extra_channels = num_extra_channels;
417 /* default transform */
418 if (!get_bits1(gb)) {
419 /* opsin inverse matrix */
420 if (xyb_encoded && !get_bits1(gb))
421 skip_bits_long(gb, 16 * 16);
422 /* cw_mask and default weights */
424 skip_bits_long(gb, 16 * 15);
426 skip_bits_long(gb, 16 * 55);
428 skip_bits_long(gb, 16 * 210);
431 if (!have_icc_profile) {
432 int bits_remaining = 7 - ((get_bits_count(gb) - 1) & 0x7);
433 if (bits_remaining && get_bits(gb, bits_remaining))
434 return AVERROR_INVALIDDATA;
437 if (get_bits_left(gb) < 0)
438 return AVERROR_BUFFER_TOO_SMALL;
440 return get_bits_count(gb);
444 * copies as much of the codestream into the buffer as possible
445 * pass a shorter buflen to request less
446 * returns the number of bytes consumed from input, may be greater than input_len
447 * if the input doesn't end on an ISOBMFF-box boundary
449 int ff_jpegxl_collect_codestream_header(const uint8_t *input_buffer, int input_len,
450 uint8_t *buffer, int buflen, int *copied)
453 int pos = 0, last_box = 0;
454 bytestream2_init(&gb, input_buffer, input_len);
461 if (bytestream2_get_bytes_left(&gb) < 8)
462 return AVERROR_BUFFER_TOO_SMALL;
464 size = bytestream2_get_be32(&gb);
466 if (bytestream2_get_bytes_left(&gb) < 12)
467 return AVERROR_BUFFER_TOO_SMALL;
468 size = bytestream2_get_be64(&gb);
471 /* invalid ISOBMFF size */
472 if (size && size <= head_size)
473 return AVERROR_INVALIDDATA;
477 tag = bytestream2_get_le32(&gb);
479 if (tag == MKTAG('j','x','l','p')) {
481 if (bytestream2_get_bytes_left(&gb) < 4)
482 return AVERROR_BUFFER_TOO_SMALL;
483 idx = bytestream2_get_be32(&gb);
484 if (idx >= UINT32_C(0x80000000))
488 return AVERROR_INVALIDDATA;
492 if (tag == MKTAG('j','x','l','c'))
496 * size = 0 means "until EOF". this is legal but uncommon
497 * here we just set it to the remaining size of the probe buffer
500 size = bytestream2_get_bytes_left(&gb);
502 pos += size + head_size;
504 if (tag == MKTAG('j','x','l','c') || tag == MKTAG('j','x','l','p')) {
505 if (size > buflen - *copied)
506 size = buflen - *copied;
508 * arbitrary chunking of the payload makes this memcpy hard to avoid
509 * in practice this will only be performed one or two times at most
511 *copied += bytestream2_get_buffer(&gb, buffer + *copied, size);
513 bytestream2_skip(&gb, size);
515 if (last_box || bytestream2_get_bytes_left(&gb) <= 0 || *copied >= buflen)