3 * Copyright (c) 2009 Stephen Backway
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 * PGS subtitle decoder
29 #include "bytestream.h"
30 #include "libavutil/colorspace.h"
31 #include "libavutil/imgutils.h"
33 #define RGBA(r,g,b,a) (((a) << 24) | ((r) << 16) | ((g) << 8) | (b))
36 PALETTE_SEGMENT = 0x14,
37 PICTURE_SEGMENT = 0x15,
38 PRESENTATION_SEGMENT = 0x16,
39 WINDOW_SEGMENT = 0x17,
40 DISPLAY_SEGMENT = 0x80,
43 typedef struct PGSSubPictureReference {
47 } PGSSubPictureReference;
49 typedef struct PGSSubPresentation {
52 PGSSubPictureReference *objects;
55 typedef struct PGSSubPicture {
59 unsigned int rle_buffer_size, rle_data_len;
60 unsigned int rle_remaining_len;
63 typedef struct PGSSubContext {
64 PGSSubPresentation presentation;
66 PGSSubPicture pictures[UINT16_MAX];
69 static av_cold int init_decoder(AVCodecContext *avctx)
71 avctx->pix_fmt = PIX_FMT_PAL8;
76 static av_cold int close_decoder(AVCodecContext *avctx)
80 PGSSubContext *ctx = avctx->priv_data;
82 av_freep(&ctx->presentation.objects);
83 ctx->presentation.object_count = 0;
85 for (picture = 0; picture < UINT16_MAX; ++picture) {
86 av_freep(&ctx->pictures[picture].rle);
87 ctx->pictures[picture].rle_buffer_size = 0;
94 * Decode the RLE data.
96 * The subtitle is stored as an Run Length Encoded image.
98 * @param avctx contains the current codec context
99 * @param sub pointer to the processed subtitle data
100 * @param buf pointer to the RLE data to process
101 * @param buf_size size of the RLE data to process
103 static int decode_rle(AVCodecContext *avctx, AVSubtitle *sub, int rect,
104 const uint8_t *buf, unsigned int buf_size)
106 const uint8_t *rle_bitmap_end;
107 int pixel_count, line_count;
109 rle_bitmap_end = buf + buf_size;
111 sub->rects[rect]->pict.data[0] = av_malloc(sub->rects[rect]->w * sub->rects[rect]->h);
113 if (!sub->rects[rect]->pict.data[0])
119 while (buf < rle_bitmap_end && line_count < sub->rects[rect]->h) {
120 uint8_t flags, color;
123 color = bytestream_get_byte(&buf);
127 flags = bytestream_get_byte(&buf);
130 run = (run << 8) + bytestream_get_byte(&buf);
131 color = flags & 0x80 ? bytestream_get_byte(&buf) : 0;
134 if (run > 0 && pixel_count + run <= sub->rects[rect]->w * sub->rects[rect]->h) {
135 memset(sub->rects[rect]->pict.data[0] + pixel_count, color, run);
139 * New Line. Check if correct pixels decoded, if not display warning
140 * and adjust bitmap pointer to correct new line position.
142 if (pixel_count % sub->rects[rect]->w > 0)
143 av_log(avctx, AV_LOG_ERROR, "Decoded %d pixels, when line should be %d pixels\n",
144 pixel_count % sub->rects[rect]->w, sub->rects[rect]->w);
149 if (pixel_count < sub->rects[rect]->w * sub->rects[rect]->h) {
150 av_log(avctx, AV_LOG_ERROR, "Insufficient RLE data for subtitle\n");
154 av_dlog(avctx, "Pixel Count = %d, Area = %d\n", pixel_count, sub->rects[rect]->w * sub->rects[rect]->h);
160 * Parse the picture segment packet.
162 * The picture segment contains details on the sequence id,
163 * width, height and Run Length Encoded (RLE) bitmap data.
165 * @param avctx contains the current codec context
166 * @param buf pointer to the packet to process
167 * @param buf_size size of packet to process
168 * @todo TODO: Enable support for RLE data over multiple packets
170 static int parse_picture_segment(AVCodecContext *avctx,
171 const uint8_t *buf, int buf_size)
173 PGSSubContext *ctx = avctx->priv_data;
175 uint8_t sequence_desc;
176 unsigned int rle_bitmap_len, width, height;
183 picture_id = bytestream_get_be16(&buf);
185 /* skip 1 unknown byte: Version Number */
188 /* Read the Sequence Description to determine if start of RLE data or appended to previous RLE */
189 sequence_desc = bytestream_get_byte(&buf);
191 if (!(sequence_desc & 0x80)) {
192 /* Additional RLE data */
193 if (buf_size > ctx->pictures[picture_id].rle_remaining_len)
196 memcpy(ctx->pictures[picture_id].rle + ctx->pictures[picture_id].rle_data_len, buf, buf_size);
197 ctx->pictures[picture_id].rle_data_len += buf_size;
198 ctx->pictures[picture_id].rle_remaining_len -= buf_size;
207 /* Decode rle bitmap length, stored size includes width/height data */
208 rle_bitmap_len = bytestream_get_be24(&buf) - 2*2;
210 /* Get bitmap dimensions from data */
211 width = bytestream_get_be16(&buf);
212 height = bytestream_get_be16(&buf);
214 /* Make sure the bitmap is not too large */
215 if (avctx->width < width || avctx->height < height) {
216 av_log(avctx, AV_LOG_ERROR, "Bitmap dimensions larger than video.\n");
220 ctx->pictures[picture_id].w = width;
221 ctx->pictures[picture_id].h = height;
223 av_fast_malloc(&ctx->pictures[picture_id].rle, &ctx->pictures[picture_id].rle_buffer_size, rle_bitmap_len);
225 if (!ctx->pictures[picture_id].rle)
228 memcpy(ctx->pictures[picture_id].rle, buf, buf_size);
229 ctx->pictures[picture_id].rle_data_len = buf_size;
230 ctx->pictures[picture_id].rle_remaining_len = rle_bitmap_len - buf_size;
236 * Parse the palette segment packet.
238 * The palette segment contains details of the palette,
239 * a maximum of 256 colors can be defined.
241 * @param avctx contains the current codec context
242 * @param buf pointer to the packet to process
243 * @param buf_size size of packet to process
245 static void parse_palette_segment(AVCodecContext *avctx,
246 const uint8_t *buf, int buf_size)
248 PGSSubContext *ctx = avctx->priv_data;
250 const uint8_t *buf_end = buf + buf_size;
251 const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
253 int y, cb, cr, alpha;
254 int r, g, b, r_add, g_add, b_add;
256 /* Skip two null bytes */
259 while (buf < buf_end) {
260 color_id = bytestream_get_byte(&buf);
261 y = bytestream_get_byte(&buf);
262 cr = bytestream_get_byte(&buf);
263 cb = bytestream_get_byte(&buf);
264 alpha = bytestream_get_byte(&buf);
267 YUV_TO_RGB2(r, g, b, y);
269 av_dlog(avctx, "Color %d := (%d,%d,%d,%d)\n", color_id, r, g, b, alpha);
271 /* Store color in palette */
272 ctx->clut[color_id] = RGBA(r,g,b,alpha);
277 * Parse the presentation segment packet.
279 * The presentation segment contains details on the video
280 * width, video height, x & y subtitle position.
282 * @param avctx contains the current codec context
283 * @param buf pointer to the packet to process
284 * @param buf_size size of packet to process
285 * @todo TODO: Implement cropping
286 * @todo TODO: Implement forcing of subtitles
288 static void parse_presentation_segment(AVCodecContext *avctx,
289 const uint8_t *buf, int buf_size)
291 PGSSubContext *ctx = avctx->priv_data;
293 int w = bytestream_get_be16(&buf);
294 int h = bytestream_get_be16(&buf);
296 uint16_t object_index;
298 av_dlog(avctx, "Video Dimensions %dx%d\n",
300 if (av_image_check_size(w, h, 0, avctx) >= 0)
301 avcodec_set_dimensions(avctx, w, h);
303 /* Skip 1 bytes of unknown, frame rate? */
306 ctx->presentation.id_number = bytestream_get_be16(&buf);
309 * Skip 3 bytes of unknown:
311 * palette_update_flag (0x80),
316 ctx->presentation.object_count = bytestream_get_byte(&buf);
317 if (!ctx->presentation.object_count)
320 /* Verify that enough bytes are remaining for all of the objects. */
322 if (buf_size < ctx->presentation.object_count * 8) {
323 ctx->presentation.object_count = 0;
327 av_freep(&ctx->presentation.objects);
328 ctx->presentation.objects = av_malloc(sizeof(PGSSubPictureReference) * ctx->presentation.object_count);
329 if (!ctx->presentation.objects) {
330 ctx->presentation.object_count = 0;
334 for (object_index = 0; object_index < ctx->presentation.object_count; ++object_index) {
335 PGSSubPictureReference *reference = &ctx->presentation.objects[object_index];
336 reference->picture_id = bytestream_get_be16(&buf);
339 * Skip 2 bytes of unknown:
341 * composition_flag (0x80 - object cropped, 0x40 - object forced)
345 reference->x = bytestream_get_be16(&buf);
346 reference->y = bytestream_get_be16(&buf);
348 /* TODO If cropping, cropping_x, cropping_y, cropping_width, cropping_height (all 2 bytes).*/
349 av_dlog(avctx, "Subtitle Placement ID=%d, x=%d, y=%d\n", reference->picture_id, reference->x, reference->y);
351 if (reference->x > avctx->width || reference->y > avctx->height) {
352 av_log(avctx, AV_LOG_ERROR, "Subtitle out of video bounds. x = %d, y = %d, video width = %d, video height = %d.\n",
353 reference->x, reference->y, avctx->width, avctx->height);
361 * Parse the display segment packet.
363 * The display segment controls the updating of the display.
365 * @param avctx contains the current codec context
366 * @param data pointer to the data pertaining the subtitle to display
367 * @param buf pointer to the packet to process
368 * @param buf_size size of packet to process
369 * @todo TODO: Fix start time, relies on correct PTS, currently too late
371 * @todo TODO: Fix end time, normally cleared by a second display
372 * @todo segment, which is currently ignored as it clears
373 * @todo the subtitle too early.
375 static int display_end_segment(AVCodecContext *avctx, void *data,
376 const uint8_t *buf, int buf_size)
378 AVSubtitle *sub = data;
379 PGSSubContext *ctx = avctx->priv_data;
384 * The end display time is a timeout value and is only reached
385 * if the next subtitle is later than timeout or subtitle has
386 * not been cleared by a subsequent empty display command.
389 memset(sub, 0, sizeof(*sub));
391 // Blank if last object_count was 0.
392 if (!ctx->presentation.object_count)
395 sub->start_display_time = 0;
396 sub->end_display_time = 20000;
399 sub->num_rects = ctx->presentation.object_count;
400 sub->rects = av_mallocz(sizeof(*sub->rects) * sub->num_rects);
402 for (rect = 0; rect < sub->num_rects; ++rect) {
403 uint16_t picture_id = ctx->presentation.objects[rect].picture_id;
404 sub->rects[rect] = av_mallocz(sizeof(*sub->rects[rect]));
405 sub->rects[rect]->x = ctx->presentation.objects[rect].x;
406 sub->rects[rect]->y = ctx->presentation.objects[rect].y;
407 sub->rects[rect]->w = ctx->pictures[picture_id].w;
408 sub->rects[rect]->h = ctx->pictures[picture_id].h;
409 sub->rects[rect]->type = SUBTITLE_BITMAP;
412 sub->rects[rect]->pict.linesize[0] = ctx->pictures[picture_id].w;
413 if (ctx->pictures[picture_id].rle) {
414 if (ctx->pictures[picture_id].rle_remaining_len)
415 av_log(avctx, AV_LOG_ERROR, "RLE data length %u is %u bytes shorter than expected\n",
416 ctx->pictures[picture_id].rle_data_len, ctx->pictures[picture_id].rle_remaining_len);
417 if (decode_rle(avctx, sub, rect, ctx->pictures[picture_id].rle, ctx->pictures[picture_id].rle_data_len) < 0)
421 /* Allocate memory for colors */
422 sub->rects[rect]->nb_colors = 256;
423 sub->rects[rect]->pict.data[1] = av_mallocz(AVPALETTE_SIZE);
425 memcpy(sub->rects[rect]->pict.data[1], ctx->clut, sub->rects[rect]->nb_colors * sizeof(uint32_t));
431 static int decode(AVCodecContext *avctx, void *data, int *data_size,
434 const uint8_t *buf = avpkt->data;
435 int buf_size = avpkt->size;
437 const uint8_t *buf_end;
438 uint8_t segment_type;
442 av_dlog(avctx, "PGS sub packet:\n");
444 for (i = 0; i < buf_size; i++) {
445 av_dlog(avctx, "%02x ", buf[i]);
447 av_dlog(avctx, "\n");
451 av_dlog(avctx, "\n");
455 /* Ensure that we have received at a least a segment code and segment length */
459 buf_end = buf + buf_size;
461 /* Step through buffer to identify segments */
462 while (buf < buf_end) {
463 segment_type = bytestream_get_byte(&buf);
464 segment_length = bytestream_get_be16(&buf);
466 av_dlog(avctx, "Segment Length %d, Segment Type %x\n", segment_length, segment_type);
468 if (segment_type != DISPLAY_SEGMENT && segment_length > buf_end - buf)
471 switch (segment_type) {
472 case PALETTE_SEGMENT:
473 parse_palette_segment(avctx, buf, segment_length);
475 case PICTURE_SEGMENT:
476 parse_picture_segment(avctx, buf, segment_length);
478 case PRESENTATION_SEGMENT:
479 parse_presentation_segment(avctx, buf, segment_length);
483 * Window Segment Structure (No new information provided):
485 * 2 bytes: X position of subtitle,
486 * 2 bytes: Y position of subtitle,
487 * 2 bytes: Width of subtitle,
488 * 2 bytes: Height of subtitle.
491 case DISPLAY_SEGMENT:
492 *data_size = display_end_segment(avctx, data, buf, segment_length);
495 av_log(avctx, AV_LOG_ERROR, "Unknown subtitle segment type 0x%x, length %d\n",
496 segment_type, segment_length);
500 buf += segment_length;
506 AVCodec ff_pgssub_decoder = {
508 .type = AVMEDIA_TYPE_SUBTITLE,
509 .id = CODEC_ID_HDMV_PGS_SUBTITLE,
510 .priv_data_size = sizeof(PGSSubContext),
511 .init = init_decoder,
512 .close = close_decoder,
514 .long_name = NULL_IF_CONFIG_SMALL("HDMV Presentation Graphic Stream subtitles"),