| Line | Branch | Exec | Source |
|---|---|---|---|
| 1 | /* | ||
| 2 | * KTX 1.0 demuxer for ASTC | ||
| 3 | * Copyright (c) 2026 Jun Zhao | ||
| 4 | * | ||
| 5 | * This file is part of FFmpeg. | ||
| 6 | * | ||
| 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. | ||
| 11 | * | ||
| 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. | ||
| 16 | * | ||
| 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 | ||
| 20 | */ | ||
| 21 | |||
| 22 | /** | ||
| 23 | * @file | ||
| 24 | * KTX 1.0 demuxer for ASTC textures. | ||
| 25 | * | ||
| 26 | * Parses the KTX 1.0 header, recovers the ASTC block size from the GL | ||
| 27 | * internal format enum, and synthesizes a 16-byte .astc-style extradata | ||
| 28 | * for the decoder. The raw ASTC bitstream is emitted as a single packet. | ||
| 29 | */ | ||
| 30 | |||
| 31 | #include "config_components.h" | ||
| 32 | |||
| 33 | #include <inttypes.h> | ||
| 34 | #include <limits.h> | ||
| 35 | |||
| 36 | #include "libavcodec/defs.h" | ||
| 37 | #include "avformat.h" | ||
| 38 | #include "avio.h" | ||
| 39 | #include "avio_internal.h" | ||
| 40 | #include "demux.h" | ||
| 41 | #include "internal.h" | ||
| 42 | #include "libavutil/display.h" | ||
| 43 | #include "libavutil/intreadwrite.h" | ||
| 44 | |||
| 45 | #define ASTC_HEADER_SIZE 16 | ||
| 46 | static const uint8_t astc_magic[4] = { 0x13, 0xAB, 0xA1, 0x5C }; | ||
| 47 | static const uint8_t ktx_magic[12] = { | ||
| 48 | 0xAB, 'K', 'T', 'X', ' ', '1', '1', 0xBB, 0x0D, 0x0A, 0x1A, 0x0A | ||
| 49 | }; | ||
| 50 | |||
| 51 | /* ASTC GL internal format enums (2D), linear base 0x93B0, sRGB = +0x20. */ | ||
| 52 | static const int astc_gl_linear[14] = { | ||
| 53 | 0x93B0, 0x93B1, 0x93B2, 0x93B3, 0x93B4, 0x93B5, 0x93B6, 0x93B7, | ||
| 54 | 0x93B8, 0x93B9, 0x93BA, 0x93BB, 0x93BC, 0x93BD | ||
| 55 | }; | ||
| 56 | static const int astc_bx[14] = { 4, 5, 5, 6, 6, 8, 8, 8, 10, 10, 10, 10, 12, 12 }; | ||
| 57 | static const int astc_by[14] = { 4, 4, 5, 5, 6, 5, 6, 8, 5, 6, 8, 10, 10, 12 }; | ||
| 58 | |||
| 59 | typedef struct KTXDemuxerContext { | ||
| 60 | uint32_t image_size; | ||
| 61 | int truncated; | ||
| 62 | } KTXDemuxerContext; | ||
| 63 | |||
| 64 | /* Map a GL internal format enum to a 2D ASTC block size. Returns 1 on success. */ | ||
| 65 | ✗ | static int gl_enum_to_block(uint32_t e, int *bx, int *by) | |
| 66 | { | ||
| 67 | ✗ | int base = (e >= 0x93D0 && e <= 0x93DD) ? (e - 0x20) : e; | |
| 68 | ✗ | for (int i = 0; i < 14; i++) { | |
| 69 | ✗ | if (astc_gl_linear[i] == (int)base) { | |
| 70 | ✗ | *bx = astc_bx[i]; | |
| 71 | ✗ | *by = astc_by[i]; | |
| 72 | ✗ | return 1; | |
| 73 | } | ||
| 74 | } | ||
| 75 | ✗ | return 0; | |
| 76 | } | ||
| 77 | |||
| 78 | 8043 | static int ktx_probe(const AVProbeData *p) | |
| 79 | { | ||
| 80 | int bx, by; | ||
| 81 | |||
| 82 |
3/4✓ Branch 0 taken 8042 times.
✓ Branch 1 taken 1 times.
✓ Branch 2 taken 8042 times.
✗ Branch 3 not taken.
|
8043 | if (p->buf_size < 32 || memcmp(p->buf, ktx_magic, sizeof(ktx_magic))) |
| 83 | 8043 | return 0; | |
| 84 | |||
| 85 | /* KTX 1.0 carries ETC, uncompressed and other payloads this demuxer cannot | ||
| 86 | * read, so only claim the file once glInternalFormat is an ASTC format we | ||
| 87 | * actually support. */ | ||
| 88 | ✗ | if (!gl_enum_to_block(AV_RL32(p->buf + 28), &bx, &by)) | |
| 89 | ✗ | return 0; | |
| 90 | |||
| 91 | /* Just below the maximum: this is a KTX file, but only the ASTC subset of | ||
| 92 | * the container is implemented here. */ | ||
| 93 | ✗ | return AVPROBE_SCORE_MAX - 1; | |
| 94 | } | ||
| 95 | |||
| 96 | /* Map a KTXorientation value such as "S=r,T=d" onto the flips needed to | ||
| 97 | * bring the stored rows back to the top-down, left-to-right order the | ||
| 98 | * decoder produces. The R component is irrelevant for 2D textures. */ | ||
| 99 | ✗ | static void ktx_orientation_flips(const char *value, int *hflip, int *vflip) | |
| 100 | { | ||
| 101 | ✗ | const char *p = value; | |
| 102 | |||
| 103 | ✗ | while (p && *p) { | |
| 104 | ✗ | if (p[0] == 'S' && p[1] == '=' && p[2]) | |
| 105 | ✗ | *hflip = p[2] == 'l'; | |
| 106 | ✗ | else if (p[0] == 'T' && p[1] == '=' && p[2]) | |
| 107 | ✗ | *vflip = p[2] == 'u'; | |
| 108 | ✗ | p = strchr(p, ','); | |
| 109 | ✗ | if (p) | |
| 110 | ✗ | p++; | |
| 111 | } | ||
| 112 | ✗ | } | |
| 113 | |||
| 114 | /* Walk the key/value block, picking up KTXorientation. Unknown or oversized | ||
| 115 | * entries are skipped rather than buffered wholesale. */ | ||
| 116 | ✗ | static int ktx_read_orientation(AVFormatContext *s, uint32_t kvdata, | |
| 117 | int *hflip, int *vflip) | ||
| 118 | { | ||
| 119 | ✗ | AVIOContext *pb = s->pb; | |
| 120 | uint8_t buf[256]; | ||
| 121 | ✗ | uint32_t left = kvdata; | |
| 122 | |||
| 123 | ✗ | *hflip = *vflip = 0; | |
| 124 | |||
| 125 | ✗ | while (left >= 4) { | |
| 126 | ✗ | uint32_t entry = avio_rl32(pb); | |
| 127 | ✗ | uint32_t padded = (entry + 3) & ~3u; | |
| 128 | |||
| 129 | ✗ | left -= 4; | |
| 130 | ✗ | if (!entry || entry > left || padded > left) { | |
| 131 | ✗ | av_log(s, AV_LOG_ERROR, "Invalid KTX key/value data.\n"); | |
| 132 | ✗ | return AVERROR_INVALIDDATA; | |
| 133 | } | ||
| 134 | |||
| 135 | ✗ | if (entry >= sizeof(buf)) { | |
| 136 | /* Nothing interesting is this large. */ | ||
| 137 | ✗ | if (avio_skip(pb, padded) < 0) | |
| 138 | ✗ | return AVERROR_INVALIDDATA; | |
| 139 | } else { | ||
| 140 | const char *key, *value; | ||
| 141 | |||
| 142 | ✗ | if (ffio_read_size(pb, buf, entry) < 0) | |
| 143 | ✗ | return AVERROR_INVALIDDATA; | |
| 144 | ✗ | buf[entry] = 0; | |
| 145 | |||
| 146 | ✗ | key = (const char *)buf; | |
| 147 | ✗ | value = key + strlen(key) + 1; | |
| 148 | ✗ | if (value < (const char *)buf + entry) { | |
| 149 | ✗ | if (!strcmp(key, "KTXorientation")) | |
| 150 | ✗ | ktx_orientation_flips(value, hflip, vflip); | |
| 151 | else | ||
| 152 | ✗ | av_log(s, AV_LOG_VERBOSE, "Ignoring KTX key '%s'.\n", key); | |
| 153 | } | ||
| 154 | ✗ | if (avio_skip(pb, padded - entry) < 0) | |
| 155 | ✗ | return AVERROR_INVALIDDATA; | |
| 156 | } | ||
| 157 | ✗ | left -= padded; | |
| 158 | } | ||
| 159 | |||
| 160 | /* Every entry occupies a padded, four byte aligned slot, so a leftover | ||
| 161 | * tail means the declared block size is not aligned and the image payload | ||
| 162 | * does not start where the header claims it does. */ | ||
| 163 | ✗ | if (left) { | |
| 164 | ✗ | av_log(s, AV_LOG_ERROR, "Invalid KTX key/value data.\n"); | |
| 165 | ✗ | return AVERROR_INVALIDDATA; | |
| 166 | } | ||
| 167 | |||
| 168 | ✗ | return 0; | |
| 169 | } | ||
| 170 | |||
| 171 | ✗ | static int ktx_read_header(AVFormatContext *s) | |
| 172 | { | ||
| 173 | ✗ | AVIOContext *pb = s->pb; | |
| 174 | ✗ | KTXDemuxerContext *ktx = s->priv_data; | |
| 175 | AVStream *st; | ||
| 176 | uint8_t hdr[64]; | ||
| 177 | uint32_t endian, gl_type, gl_type_size, gl_format, gl_internal; | ||
| 178 | uint32_t gl_base_internal, w, h, depth, array, faces, mips; | ||
| 179 | uint32_t kvdata, image_size; | ||
| 180 | uint8_t image_size_buf[4]; | ||
| 181 | ✗ | int bx = 0, by = 0; | |
| 182 | ✗ | int hflip = 0, vflip = 0; | |
| 183 | uint8_t extra[16]; | ||
| 184 | |||
| 185 | ✗ | if (ffio_read_size(pb, hdr, sizeof(hdr)) < 0) | |
| 186 | ✗ | return AVERROR_INVALIDDATA; | |
| 187 | ✗ | if (memcmp(hdr, ktx_magic, sizeof(ktx_magic))) { | |
| 188 | ✗ | av_log(s, AV_LOG_ERROR, "Not a KTX file (bad magic).\n"); | |
| 189 | ✗ | return AVERROR_INVALIDDATA; | |
| 190 | } | ||
| 191 | |||
| 192 | ✗ | endian = AV_RL32(hdr + 12); | |
| 193 | ✗ | gl_type = AV_RL32(hdr + 16); | |
| 194 | ✗ | gl_type_size = AV_RL32(hdr + 20); | |
| 195 | ✗ | gl_format = AV_RL32(hdr + 24); | |
| 196 | ✗ | gl_internal = AV_RL32(hdr + 28); | |
| 197 | ✗ | gl_base_internal = AV_RL32(hdr + 32); | |
| 198 | ✗ | w = AV_RL32(hdr + 36); | |
| 199 | ✗ | h = AV_RL32(hdr + 40); | |
| 200 | ✗ | depth = AV_RL32(hdr + 44); | |
| 201 | ✗ | array = AV_RL32(hdr + 48); | |
| 202 | ✗ | faces = AV_RL32(hdr + 52); | |
| 203 | ✗ | mips = AV_RL32(hdr + 56); | |
| 204 | ✗ | kvdata = AV_RL32(hdr + 60); | |
| 205 | |||
| 206 | ✗ | if (endian != 0x04030201) { | |
| 207 | ✗ | av_log(s, AV_LOG_ERROR, "KTX big-endian / unknown endianness not supported.\n"); | |
| 208 | ✗ | return AVERROR_INVALIDDATA; | |
| 209 | } | ||
| 210 | ✗ | if (gl_type || gl_type_size != 1 || gl_format || gl_base_internal != 0x1908 || | |
| 211 | ✗ | !w || !h || w > 0xFFFFFF || h > 0xFFFFFF || | |
| 212 | ✗ | depth || array || faces != 1 || mips != 1) { | |
| 213 | ✗ | av_log(s, AV_LOG_ERROR, "Unsupported KTX texture layout.\n"); | |
| 214 | ✗ | return AVERROR_INVALIDDATA; | |
| 215 | } | ||
| 216 | |||
| 217 | ✗ | if (!gl_enum_to_block(gl_internal, &bx, &by)) { | |
| 218 | ✗ | av_log(s, AV_LOG_ERROR, "Unsupported ASTC GL format 0x%X in KTX.\n", gl_internal); | |
| 219 | ✗ | return AVERROR_INVALIDDATA; | |
| 220 | } | ||
| 221 | |||
| 222 | ✗ | if (ktx_read_orientation(s, kvdata, &hflip, &vflip) < 0) | |
| 223 | ✗ | return AVERROR_INVALIDDATA; | |
| 224 | ✗ | if (ffio_read_size(pb, image_size_buf, sizeof(image_size_buf)) < 0) | |
| 225 | ✗ | return AVERROR_INVALIDDATA; | |
| 226 | ✗ | image_size = AV_RL32(image_size_buf); | |
| 227 | ✗ | if (!image_size || image_size > INT_MAX) | |
| 228 | ✗ | return AVERROR_INVALIDDATA; | |
| 229 | /* The payload must hold exactly the blocks the texture geometry needs; | ||
| 230 | * anything else would either truncate the image or smuggle in trailing | ||
| 231 | * data that the decoder would ignore. */ | ||
| 232 | { | ||
| 233 | ✗ | uint64_t expected = (((uint64_t)w + bx - 1) / bx) * | |
| 234 | ✗ | (((uint64_t)h + by - 1) / by) * 16; | |
| 235 | ✗ | if (image_size != expected) { | |
| 236 | ✗ | av_log(s, AV_LOG_ERROR, | |
| 237 | "KTX image size %u does not match the %"PRIu64" bytes " | ||
| 238 | "required for a %ux%u texture with %dx%d blocks.\n", | ||
| 239 | image_size, expected, w, h, bx, by); | ||
| 240 | ✗ | return AVERROR_INVALIDDATA; | |
| 241 | } | ||
| 242 | } | ||
| 243 | ✗ | ktx->image_size = image_size; | |
| 244 | |||
| 245 | ✗ | st = avformat_new_stream(s, NULL); | |
| 246 | ✗ | if (!st) | |
| 247 | ✗ | return AVERROR(ENOMEM); | |
| 248 | |||
| 249 | ✗ | st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO; | |
| 250 | ✗ | st->codecpar->codec_id = AV_CODEC_ID_ASTC; | |
| 251 | ✗ | st->codecpar->width = w; | |
| 252 | ✗ | st->codecpar->height = h; | |
| 253 | /* The GL internal format fixes the colour space but not the endpoint | ||
| 254 | * format: an sRGB texture is LDR, while a linear one may hold LDR, HDR RGB | ||
| 255 | * with LDR alpha, or full HDR endpoints. Report that distinction so the | ||
| 256 | * decoder does not assume LDR for a linear texture; a linear texture is | ||
| 257 | * sampled with HDR precision by default (LINEAR_ANY), so the stream format | ||
| 258 | * is the matching half-float one rather than 8-bit RGBA. */ | ||
| 259 | ✗ | st->codecpar->profile = gl_internal >= 0x93D0 && gl_internal <= 0x93DD ? | |
| 260 | ✗ | AV_PROFILE_ASTC_LDR_SRGB : AV_PROFILE_ASTC_LINEAR_ANY; | |
| 261 | ✗ | st->codecpar->format = st->codecpar->profile == AV_PROFILE_ASTC_LDR_SRGB ? | |
| 262 | ✗ | AV_PIX_FMT_RGBA : AV_PIX_FMT_RGBAF16; | |
| 263 | |||
| 264 | /* Synthesize the .astc-style extradata for the decoder. Bytes [13-15] hold | ||
| 265 | * the image depth (dim_z), which is always 1 here: KTX 1.0 2D only. */ | ||
| 266 | ✗ | memset(extra, 0, sizeof(extra)); | |
| 267 | ✗ | memcpy(extra, astc_magic, sizeof(astc_magic)); | |
| 268 | ✗ | extra[4] = (uint8_t)bx; | |
| 269 | ✗ | extra[5] = (uint8_t)by; | |
| 270 | ✗ | extra[6] = 1; | |
| 271 | ✗ | AV_WL24(extra + 7, w); | |
| 272 | ✗ | AV_WL24(extra + 10, h); | |
| 273 | ✗ | extra[13] = 1; | |
| 274 | ✗ | if (ff_alloc_extradata(st->codecpar, ASTC_HEADER_SIZE) < 0) | |
| 275 | ✗ | return AVERROR(ENOMEM); | |
| 276 | ✗ | memcpy(st->codecpar->extradata, extra, ASTC_HEADER_SIZE); | |
| 277 | |||
| 278 | /* Carry the stored row order to the caller instead of silently returning | ||
| 279 | * a mirrored image. */ | ||
| 280 | ✗ | if (hflip || vflip) { | |
| 281 | AVPacketSideData *sd; | ||
| 282 | int32_t *matrix; | ||
| 283 | |||
| 284 | ✗ | sd = av_packet_side_data_new(&st->codecpar->coded_side_data, | |
| 285 | ✗ | &st->codecpar->nb_coded_side_data, | |
| 286 | AV_PKT_DATA_DISPLAYMATRIX, | ||
| 287 | 9 * sizeof(*matrix), 0); | ||
| 288 | ✗ | if (!sd) | |
| 289 | ✗ | return AVERROR(ENOMEM); | |
| 290 | ✗ | matrix = (int32_t *)sd->data; | |
| 291 | ✗ | av_display_rotation_set(matrix, 0); | |
| 292 | ✗ | av_display_matrix_flip(matrix, hflip, vflip); | |
| 293 | ✗ | av_log(s, AV_LOG_VERBOSE, | |
| 294 | "Applying KTXorientation (hflip %d, vflip %d).\n", hflip, vflip); | ||
| 295 | } | ||
| 296 | |||
| 297 | ✗ | avpriv_set_pts_info(st, 64, 1, 1); | |
| 298 | ✗ | return 0; | |
| 299 | } | ||
| 300 | |||
| 301 | ✗ | static int ktx_read_packet(AVFormatContext *s, AVPacket *pkt) | |
| 302 | { | ||
| 303 | ✗ | KTXDemuxerContext *ktx = s->priv_data; | |
| 304 | int ret; | ||
| 305 | |||
| 306 | /* Keep reporting a detected truncation instead of falling through to EOF: | ||
| 307 | * a caller that stops at the first error must not see the broken image as | ||
| 308 | * a clean end of stream. */ | ||
| 309 | ✗ | if (ktx->truncated) | |
| 310 | ✗ | return AVERROR_INVALIDDATA; | |
| 311 | |||
| 312 | ✗ | if (!ktx->image_size) | |
| 313 | ✗ | return AVERROR_EOF; | |
| 314 | |||
| 315 | /* A payload that is missing entirely reads as EOF, which has to be | ||
| 316 | * reported as the truncation it is rather than as the end of a valid | ||
| 317 | * stream. */ | ||
| 318 | ✗ | ret = av_get_packet(s->pb, pkt, ktx->image_size); | |
| 319 | ✗ | if (ret == AVERROR_EOF) | |
| 320 | ✗ | ret = 0; | |
| 321 | ✗ | if (ret < 0) | |
| 322 | ✗ | return ret; | |
| 323 | ✗ | if (ret != ktx->image_size) { | |
| 324 | ✗ | av_log(s, AV_LOG_ERROR, | |
| 325 | "Truncated KTX image: got %d of %u bytes.\n", | ||
| 326 | ret, ktx->image_size); | ||
| 327 | ✗ | av_packet_unref(pkt); | |
| 328 | ✗ | ktx->truncated = 1; | |
| 329 | ✗ | return AVERROR_INVALIDDATA; | |
| 330 | } | ||
| 331 | |||
| 332 | ✗ | pkt->stream_index = 0; | |
| 333 | ✗ | pkt->flags |= AV_PKT_FLAG_KEY; | |
| 334 | ✗ | ktx->image_size = 0; /* single mip */ | |
| 335 | ✗ | return 0; | |
| 336 | } | ||
| 337 | |||
| 338 | const FFInputFormat ff_ktx_demuxer = { | ||
| 339 | .p.name = "ktx", | ||
| 340 | .p.long_name = NULL_IF_CONFIG_SMALL("KTX 1.0 (Khronos Texture) for ASTC"), | ||
| 341 | .p.mime_type = "image/ktx", | ||
| 342 | .p.extensions = "ktx", | ||
| 343 | .p.flags = AVFMT_NOTIMESTAMPS, | ||
| 344 | .priv_data_size = sizeof(KTXDemuxerContext), | ||
| 345 | .read_probe = ktx_probe, | ||
| 346 | .read_header = ktx_read_header, | ||
| 347 | .read_packet = ktx_read_packet, | ||
| 348 | }; | ||
| 349 |