1/* 2 * PGS subtitle decoder 3 * Copyright (c) 2009 Stephen Backway 4 * 5 * This file is part of Libav. 6 * 7 * Libav 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 * Libav 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 Libav; 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 * PGS subtitle decoder 25 */ 26 27#include "avcodec.h" 28#include "dsputil.h" 29#include "bytestream.h" 30#include "libavutil/colorspace.h" 31#include "libavutil/imgutils.h" 32 33#define RGBA(r,g,b,a) (((a) << 24) | ((r) << 16) | ((g) << 8) | (b)) 34 35enum SegmentType { 36 PALETTE_SEGMENT = 0x14, 37 PICTURE_SEGMENT = 0x15, 38 PRESENTATION_SEGMENT = 0x16, 39 WINDOW_SEGMENT = 0x17, 40 DISPLAY_SEGMENT = 0x80, 41}; 42 43typedef struct PGSSubPresentation { 44 int x; 45 int y; 46 int id_number; 47 int object_number; 48} PGSSubPresentation; 49 50typedef struct PGSSubPicture { 51 int w; 52 int h; 53 uint8_t *rle; 54 unsigned int rle_buffer_size, rle_data_len; 55 unsigned int rle_remaining_len; 56} PGSSubPicture; 57 58typedef struct PGSSubContext { 59 PGSSubPresentation presentation; 60 uint32_t clut[256]; 61 PGSSubPicture picture; 62} PGSSubContext; 63 64static av_cold int init_decoder(AVCodecContext *avctx) 65{ 66 avctx->pix_fmt = PIX_FMT_PAL8; 67 68 return 0; 69} 70 71static av_cold int close_decoder(AVCodecContext *avctx) 72{ 73 PGSSubContext *ctx = avctx->priv_data; 74 75 av_freep(&ctx->picture.rle); 76 ctx->picture.rle_buffer_size = 0; 77 78 return 0; 79} 80 81/** 82 * Decode the RLE data. 83 * 84 * The subtitle is stored as an Run Length Encoded image. 85 * 86 * @param avctx contains the current codec context 87 * @param sub pointer to the processed subtitle data 88 * @param buf pointer to the RLE data to process 89 * @param buf_size size of the RLE data to process 90 */ 91static int decode_rle(AVCodecContext *avctx, AVSubtitle *sub, 92 const uint8_t *buf, unsigned int buf_size) 93{ 94 const uint8_t *rle_bitmap_end; 95 int pixel_count, line_count; 96 97 rle_bitmap_end = buf + buf_size; 98 99 sub->rects[0]->pict.data[0] = av_malloc(sub->rects[0]->w * sub->rects[0]->h); 100 101 if (!sub->rects[0]->pict.data[0]) 102 return -1; 103 104 pixel_count = 0; 105 line_count = 0; 106 107 while (buf < rle_bitmap_end && line_count < sub->rects[0]->h) { 108 uint8_t flags, color; 109 int run; 110 111 color = bytestream_get_byte(&buf); 112 run = 1; 113 114 if (color == 0x00) { 115 flags = bytestream_get_byte(&buf); 116 run = flags & 0x3f; 117 if (flags & 0x40) 118 run = (run << 8) + bytestream_get_byte(&buf); 119 color = flags & 0x80 ? bytestream_get_byte(&buf) : 0; 120 } 121 122 if (run > 0 && pixel_count + run <= sub->rects[0]->w * sub->rects[0]->h) { 123 memset(sub->rects[0]->pict.data[0] + pixel_count, color, run); 124 pixel_count += run; 125 } else if (!run) { 126 /* 127 * New Line. Check if correct pixels decoded, if not display warning 128 * and adjust bitmap pointer to correct new line position. 129 */ 130 if (pixel_count % sub->rects[0]->w > 0) 131 av_log(avctx, AV_LOG_ERROR, "Decoded %d pixels, when line should be %d pixels\n", 132 pixel_count % sub->rects[0]->w, sub->rects[0]->w); 133 line_count++; 134 } 135 } 136 137 if (pixel_count < sub->rects[0]->w * sub->rects[0]->h) { 138 av_log(avctx, AV_LOG_ERROR, "Insufficient RLE data for subtitle\n"); 139 return -1; 140 } 141 142 av_dlog(avctx, "Pixel Count = %d, Area = %d\n", pixel_count, sub->rects[0]->w * sub->rects[0]->h); 143 144 return 0; 145} 146 147/** 148 * Parse the picture segment packet. 149 * 150 * The picture segment contains details on the sequence id, 151 * width, height and Run Length Encoded (RLE) bitmap data. 152 * 153 * @param avctx contains the current codec context 154 * @param buf pointer to the packet to process 155 * @param buf_size size of packet to process 156 * @todo TODO: Enable support for RLE data over multiple packets 157 */ 158static int parse_picture_segment(AVCodecContext *avctx, 159 const uint8_t *buf, int buf_size) 160{ 161 PGSSubContext *ctx = avctx->priv_data; 162 163 uint8_t sequence_desc; 164 unsigned int rle_bitmap_len, width, height; 165 166 if (buf_size <= 4) 167 return -1; 168 buf_size -= 4; 169 170 /* skip 3 unknown bytes: Object ID (2 bytes), Version Number */ 171 buf += 3; 172 173 /* Read the Sequence Description to determine if start of RLE data or appended to previous RLE */ 174 sequence_desc = bytestream_get_byte(&buf); 175 176 if (!(sequence_desc & 0x80)) { 177 /* Additional RLE data */ 178 if (buf_size > ctx->picture.rle_remaining_len) 179 return -1; 180 181 memcpy(ctx->picture.rle + ctx->picture.rle_data_len, buf, buf_size); 182 ctx->picture.rle_data_len += buf_size; 183 ctx->picture.rle_remaining_len -= buf_size; 184 185 return 0; 186 } 187 188 if (buf_size <= 7) 189 return -1; 190 buf_size -= 7; 191 192 /* Decode rle bitmap length, stored size includes width/height data */ 193 rle_bitmap_len = bytestream_get_be24(&buf) - 2*2; 194 195 /* Get bitmap dimensions from data */ 196 width = bytestream_get_be16(&buf); 197 height = bytestream_get_be16(&buf); 198 199 /* Make sure the bitmap is not too large */ 200 if (avctx->width < width || avctx->height < height) { 201 av_log(avctx, AV_LOG_ERROR, "Bitmap dimensions larger than video.\n"); 202 return -1; 203 } 204 205 ctx->picture.w = width; 206 ctx->picture.h = height; 207 208 av_fast_malloc(&ctx->picture.rle, &ctx->picture.rle_buffer_size, rle_bitmap_len); 209 210 if (!ctx->picture.rle) 211 return -1; 212 213 memcpy(ctx->picture.rle, buf, buf_size); 214 ctx->picture.rle_data_len = buf_size; 215 ctx->picture.rle_remaining_len = rle_bitmap_len - buf_size; 216 217 return 0; 218} 219 220/** 221 * Parse the palette segment packet. 222 * 223 * The palette segment contains details of the palette, 224 * a maximum of 256 colors can be defined. 225 * 226 * @param avctx contains the current codec context 227 * @param buf pointer to the packet to process 228 * @param buf_size size of packet to process 229 */ 230static void parse_palette_segment(AVCodecContext *avctx, 231 const uint8_t *buf, int buf_size) 232{ 233 PGSSubContext *ctx = avctx->priv_data; 234 235 const uint8_t *buf_end = buf + buf_size; 236 const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; 237 int color_id; 238 int y, cb, cr, alpha; 239 int r, g, b, r_add, g_add, b_add; 240 241 /* Skip two null bytes */ 242 buf += 2; 243 244 while (buf < buf_end) { 245 color_id = bytestream_get_byte(&buf); 246 y = bytestream_get_byte(&buf); 247 cr = bytestream_get_byte(&buf); 248 cb = bytestream_get_byte(&buf); 249 alpha = bytestream_get_byte(&buf); 250 251 YUV_TO_RGB1(cb, cr); 252 YUV_TO_RGB2(r, g, b, y); 253 254 av_dlog(avctx, "Color %d := (%d,%d,%d,%d)\n", color_id, r, g, b, alpha); 255 256 /* Store color in palette */ 257 ctx->clut[color_id] = RGBA(r,g,b,alpha); 258 } 259} 260 261/** 262 * Parse the presentation segment packet. 263 * 264 * The presentation segment contains details on the video 265 * width, video height, x & y subtitle position. 266 * 267 * @param avctx contains the current codec context 268 * @param buf pointer to the packet to process 269 * @param buf_size size of packet to process 270 * @todo TODO: Implement cropping 271 * @todo TODO: Implement forcing of subtitles 272 */ 273static void parse_presentation_segment(AVCodecContext *avctx, 274 const uint8_t *buf, int buf_size) 275{ 276 PGSSubContext *ctx = avctx->priv_data; 277 278 int x, y; 279 280 int w = bytestream_get_be16(&buf); 281 int h = bytestream_get_be16(&buf); 282 283 av_dlog(avctx, "Video Dimensions %dx%d\n", 284 w, h); 285 if (av_image_check_size(w, h, 0, avctx) >= 0) 286 avcodec_set_dimensions(avctx, w, h); 287 288 /* Skip 1 bytes of unknown, frame rate? */ 289 buf++; 290 291 ctx->presentation.id_number = bytestream_get_be16(&buf); 292 293 /* 294 * Skip 3 bytes of unknown: 295 * state 296 * palette_update_flag (0x80), 297 * palette_id_to_use, 298 */ 299 buf += 3; 300 301 ctx->presentation.object_number = bytestream_get_byte(&buf); 302 if (!ctx->presentation.object_number) 303 return; 304 305 /* 306 * Skip 4 bytes of unknown: 307 * object_id_ref (2 bytes), 308 * window_id_ref, 309 * composition_flag (0x80 - object cropped, 0x40 - object forced) 310 */ 311 buf += 4; 312 313 x = bytestream_get_be16(&buf); 314 y = bytestream_get_be16(&buf); 315 316 /* TODO If cropping, cropping_x, cropping_y, cropping_width, cropping_height (all 2 bytes).*/ 317 318 av_dlog(avctx, "Subtitle Placement x=%d, y=%d\n", x, y); 319 320 if (x > avctx->width || y > avctx->height) { 321 av_log(avctx, AV_LOG_ERROR, "Subtitle out of video bounds. x = %d, y = %d, video width = %d, video height = %d.\n", 322 x, y, avctx->width, avctx->height); 323 x = 0; y = 0; 324 } 325 326 /* Fill in dimensions */ 327 ctx->presentation.x = x; 328 ctx->presentation.y = y; 329} 330 331/** 332 * Parse the display segment packet. 333 * 334 * The display segment controls the updating of the display. 335 * 336 * @param avctx contains the current codec context 337 * @param data pointer to the data pertaining the subtitle to display 338 * @param buf pointer to the packet to process 339 * @param buf_size size of packet to process 340 * @todo TODO: Fix start time, relies on correct PTS, currently too late 341 * 342 * @todo TODO: Fix end time, normally cleared by a second display 343 * @todo segment, which is currently ignored as it clears 344 * @todo the subtitle too early. 345 */ 346static int display_end_segment(AVCodecContext *avctx, void *data, 347 const uint8_t *buf, int buf_size) 348{ 349 AVSubtitle *sub = data; 350 PGSSubContext *ctx = avctx->priv_data; 351 352 /* 353 * The end display time is a timeout value and is only reached 354 * if the next subtitle is later then timeout or subtitle has 355 * not been cleared by a subsequent empty display command. 356 */ 357 358 memset(sub, 0, sizeof(*sub)); 359 // Blank if last object_number was 0. 360 // Note that this may be wrong for more complex subtitles. 361 if (!ctx->presentation.object_number) 362 return 1; 363 sub->start_display_time = 0; 364 sub->end_display_time = 20000; 365 sub->format = 0; 366 367 sub->rects = av_mallocz(sizeof(*sub->rects)); 368 sub->rects[0] = av_mallocz(sizeof(*sub->rects[0])); 369 sub->num_rects = 1; 370 371 sub->rects[0]->x = ctx->presentation.x; 372 sub->rects[0]->y = ctx->presentation.y; 373 sub->rects[0]->w = ctx->picture.w; 374 sub->rects[0]->h = ctx->picture.h; 375 sub->rects[0]->type = SUBTITLE_BITMAP; 376 377 /* Process bitmap */ 378 sub->rects[0]->pict.linesize[0] = ctx->picture.w; 379 380 if (ctx->picture.rle) { 381 if (ctx->picture.rle_remaining_len) 382 av_log(avctx, AV_LOG_ERROR, "RLE data length %u is %u bytes shorter than expected\n", 383 ctx->picture.rle_data_len, ctx->picture.rle_remaining_len); 384 if(decode_rle(avctx, sub, ctx->picture.rle, ctx->picture.rle_data_len) < 0) 385 return 0; 386 } 387 /* Allocate memory for colors */ 388 sub->rects[0]->nb_colors = 256; 389 sub->rects[0]->pict.data[1] = av_mallocz(AVPALETTE_SIZE); 390 391 memcpy(sub->rects[0]->pict.data[1], ctx->clut, sub->rects[0]->nb_colors * sizeof(uint32_t)); 392 393 return 1; 394} 395 396static int decode(AVCodecContext *avctx, void *data, int *data_size, 397 AVPacket *avpkt) 398{ 399 const uint8_t *buf = avpkt->data; 400 int buf_size = avpkt->size; 401 402 const uint8_t *buf_end; 403 uint8_t segment_type; 404 int segment_length; 405 int i; 406 407 av_dlog(avctx, "PGS sub packet:\n"); 408 409 for (i = 0; i < buf_size; i++) { 410 av_dlog(avctx, "%02x ", buf[i]); 411 if (i % 16 == 15) 412 av_dlog(avctx, "\n"); 413 } 414 415 if (i & 15) 416 av_dlog(avctx, "\n"); 417 418 *data_size = 0; 419 420 /* Ensure that we have received at a least a segment code and segment length */ 421 if (buf_size < 3) 422 return -1; 423 424 buf_end = buf + buf_size; 425 426 /* Step through buffer to identify segments */ 427 while (buf < buf_end) { 428 segment_type = bytestream_get_byte(&buf); 429 segment_length = bytestream_get_be16(&buf); 430 431 av_dlog(avctx, "Segment Length %d, Segment Type %x\n", segment_length, segment_type); 432 433 if (segment_type != DISPLAY_SEGMENT && segment_length > buf_end - buf) 434 break; 435 436 switch (segment_type) { 437 case PALETTE_SEGMENT: 438 parse_palette_segment(avctx, buf, segment_length); 439 break; 440 case PICTURE_SEGMENT: 441 parse_picture_segment(avctx, buf, segment_length); 442 break; 443 case PRESENTATION_SEGMENT: 444 parse_presentation_segment(avctx, buf, segment_length); 445 break; 446 case WINDOW_SEGMENT: 447 /* 448 * Window Segment Structure (No new information provided): 449 * 2 bytes: Unkown, 450 * 2 bytes: X position of subtitle, 451 * 2 bytes: Y position of subtitle, 452 * 2 bytes: Width of subtitle, 453 * 2 bytes: Height of subtitle. 454 */ 455 break; 456 case DISPLAY_SEGMENT: 457 *data_size = display_end_segment(avctx, data, buf, segment_length); 458 break; 459 default: 460 av_log(avctx, AV_LOG_ERROR, "Unknown subtitle segment type 0x%x, length %d\n", 461 segment_type, segment_length); 462 break; 463 } 464 465 buf += segment_length; 466 } 467 468 return buf_size; 469} 470 471AVCodec ff_pgssub_decoder = { 472 .name = "pgssub", 473 .type = AVMEDIA_TYPE_SUBTITLE, 474 .id = CODEC_ID_HDMV_PGS_SUBTITLE, 475 .priv_data_size = sizeof(PGSSubContext), 476 .init = init_decoder, 477 .close = close_decoder, 478 .decode = decode, 479 .long_name = NULL_IF_CONFIG_SMALL("HDMV Presentation Graphic Stream subtitles"), 480}; 481