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 DEBUG_PACKET_CONTENTS
35 #define RGBA(r,g,b,a) (((a) << 24) | ((r) << 16) | ((g) << 8) | (b))
38 PALETTE_SEGMENT = 0x14,
39 PICTURE_SEGMENT = 0x15,
40 PRESENTATION_SEGMENT = 0x16,
41 WINDOW_SEGMENT = 0x17,
42 DISPLAY_SEGMENT = 0x80,
45 typedef struct PGSSubPresentation {
52 typedef struct PGSSubPicture {
56 unsigned int rle_buffer_size, rle_data_len;
57 unsigned int rle_remaining_len;
60 typedef struct PGSSubContext {
61 PGSSubPresentation presentation;
63 PGSSubPicture picture;
66 static av_cold int init_decoder(AVCodecContext *avctx)
68 avctx->pix_fmt = PIX_FMT_PAL8;
73 static av_cold int close_decoder(AVCodecContext *avctx)
75 PGSSubContext *ctx = avctx->priv_data;
77 av_freep(&ctx->picture.rle);
78 ctx->picture.rle_buffer_size = 0;
84 * Decode the RLE data.
86 * The subtitle is stored as an Run Length Encoded image.
88 * @param avctx contains the current codec context
89 * @param sub pointer to the processed subtitle data
90 * @param buf pointer to the RLE data to process
91 * @param buf_size size of the RLE data to process
93 static int decode_rle(AVCodecContext *avctx, AVSubtitle *sub,
94 const uint8_t *buf, unsigned int buf_size)
96 const uint8_t *rle_bitmap_end;
97 int pixel_count, line_count;
99 rle_bitmap_end = buf + buf_size;
101 sub->rects[0]->pict.data[0] = av_malloc(sub->rects[0]->w * sub->rects[0]->h);
103 if (!sub->rects[0]->pict.data[0])
109 while (buf < rle_bitmap_end && line_count < sub->rects[0]->h) {
110 uint8_t flags, color;
113 color = bytestream_get_byte(&buf);
117 flags = bytestream_get_byte(&buf);
120 run = (run << 8) + bytestream_get_byte(&buf);
121 color = flags & 0x80 ? bytestream_get_byte(&buf) : 0;
124 if (run > 0 && pixel_count + run <= sub->rects[0]->w * sub->rects[0]->h) {
125 memset(sub->rects[0]->pict.data[0] + pixel_count, color, run);
129 * New Line. Check if correct pixels decoded, if not display warning
130 * and adjust bitmap pointer to correct new line position.
132 if (pixel_count % sub->rects[0]->w > 0)
133 av_log(avctx, AV_LOG_ERROR, "Decoded %d pixels, when line should be %d pixels\n",
134 pixel_count % sub->rects[0]->w, sub->rects[0]->w);
139 if (pixel_count < sub->rects[0]->w * sub->rects[0]->h) {
140 av_log(avctx, AV_LOG_ERROR, "Insufficient RLE data for subtitle\n");
144 av_dlog(avctx, "Pixel Count = %d, Area = %d\n", pixel_count, sub->rects[0]->w * sub->rects[0]->h);
150 * Parse the picture segment packet.
152 * The picture segment contains details on the sequence id,
153 * width, height and Run Length Encoded (RLE) bitmap data.
155 * @param avctx contains the current codec context
156 * @param buf pointer to the packet to process
157 * @param buf_size size of packet to process
158 * @todo TODO: Enable support for RLE data over multiple packets
160 static int parse_picture_segment(AVCodecContext *avctx,
161 const uint8_t *buf, int buf_size)
163 PGSSubContext *ctx = avctx->priv_data;
165 uint8_t sequence_desc;
166 unsigned int rle_bitmap_len, width, height;
172 /* skip 3 unknown bytes: Object ID (2 bytes), Version Number */
175 /* Read the Sequence Description to determine if start of RLE data or appended to previous RLE */
176 sequence_desc = bytestream_get_byte(&buf);
178 if (!(sequence_desc & 0x80)) {
179 /* Additional RLE data */
180 if (buf_size > ctx->picture.rle_remaining_len)
183 memcpy(ctx->picture.rle + ctx->picture.rle_data_len, buf, buf_size);
184 ctx->picture.rle_data_len += buf_size;
185 ctx->picture.rle_remaining_len -= buf_size;
194 /* Decode rle bitmap length, stored size includes width/height data */
195 rle_bitmap_len = bytestream_get_be24(&buf) - 2*2;
197 /* Get bitmap dimensions from data */
198 width = bytestream_get_be16(&buf);
199 height = bytestream_get_be16(&buf);
201 /* Make sure the bitmap is not too large */
202 if (avctx->width < width || avctx->height < height) {
203 av_log(avctx, AV_LOG_ERROR, "Bitmap dimensions larger then video.\n");
207 ctx->picture.w = width;
208 ctx->picture.h = height;
210 av_fast_malloc(&ctx->picture.rle, &ctx->picture.rle_buffer_size, rle_bitmap_len);
212 if (!ctx->picture.rle)
215 memcpy(ctx->picture.rle, buf, buf_size);
216 ctx->picture.rle_data_len = buf_size;
217 ctx->picture.rle_remaining_len = rle_bitmap_len - buf_size;
223 * Parse the palette segment packet.
225 * The palette segment contains details of the palette,
226 * a maximum of 256 colors can be defined.
228 * @param avctx contains the current codec context
229 * @param buf pointer to the packet to process
230 * @param buf_size size of packet to process
232 static void parse_palette_segment(AVCodecContext *avctx,
233 const uint8_t *buf, int buf_size)
235 PGSSubContext *ctx = avctx->priv_data;
237 const uint8_t *buf_end = buf + buf_size;
238 const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
240 int y, cb, cr, alpha;
241 int r, g, b, r_add, g_add, b_add;
243 /* Skip two null bytes */
246 while (buf < buf_end) {
247 color_id = bytestream_get_byte(&buf);
248 y = bytestream_get_byte(&buf);
249 cr = bytestream_get_byte(&buf);
250 cb = bytestream_get_byte(&buf);
251 alpha = bytestream_get_byte(&buf);
254 YUV_TO_RGB2(r, g, b, y);
256 av_dlog(avctx, "Color %d := (%d,%d,%d,%d)\n", color_id, r, g, b, alpha);
258 /* Store color in palette */
259 ctx->clut[color_id] = RGBA(r,g,b,alpha);
264 * Parse the presentation segment packet.
266 * The presentation segment contains details on the video
267 * width, video height, x & y subtitle position.
269 * @param avctx contains the current codec context
270 * @param buf pointer to the packet to process
271 * @param buf_size size of packet to process
272 * @todo TODO: Implement cropping
273 * @todo TODO: Implement forcing of subtitles
275 static void parse_presentation_segment(AVCodecContext *avctx,
276 const uint8_t *buf, int buf_size)
278 PGSSubContext *ctx = avctx->priv_data;
282 int w = bytestream_get_be16(&buf);
283 int h = bytestream_get_be16(&buf);
285 av_dlog(avctx, "Video Dimensions %dx%d\n",
287 if (av_image_check_size(w, h, 0, avctx) >= 0)
288 avcodec_set_dimensions(avctx, w, h);
290 /* Skip 1 bytes of unknown, frame rate? */
293 ctx->presentation.id_number = bytestream_get_be16(&buf);
296 * Skip 3 bytes of unknown:
298 * palette_update_flag (0x80),
303 ctx->presentation.object_number = bytestream_get_byte(&buf);
304 if (!ctx->presentation.object_number)
308 * Skip 4 bytes of unknown:
309 * object_id_ref (2 bytes),
311 * composition_flag (0x80 - object cropped, 0x40 - object forced)
315 x = bytestream_get_be16(&buf);
316 y = bytestream_get_be16(&buf);
318 /* TODO If cropping, cropping_x, cropping_y, cropping_width, cropping_height (all 2 bytes).*/
320 av_dlog(avctx, "Subtitle Placement x=%d, y=%d\n", x, y);
322 if (x > avctx->width || y > avctx->height) {
323 av_log(avctx, AV_LOG_ERROR, "Subtitle out of video bounds. x = %d, y = %d, video width = %d, video height = %d.\n",
324 x, y, avctx->width, avctx->height);
328 /* Fill in dimensions */
329 ctx->presentation.x = x;
330 ctx->presentation.y = y;
334 * Parse the display segment packet.
336 * The display segment controls the updating of the display.
338 * @param avctx contains the current codec context
339 * @param data pointer to the data pertaining the subtitle to display
340 * @param buf pointer to the packet to process
341 * @param buf_size size of packet to process
342 * @todo TODO: Fix start time, relies on correct PTS, currently too late
344 * @todo TODO: Fix end time, normally cleared by a second display
345 * @todo segment, which is currently ignored as it clears
346 * @todo the subtitle too early.
348 static int display_end_segment(AVCodecContext *avctx, void *data,
349 const uint8_t *buf, int buf_size)
351 AVSubtitle *sub = data;
352 PGSSubContext *ctx = avctx->priv_data;
355 * The end display time is a timeout value and is only reached
356 * if the next subtitle is later then timeout or subtitle has
357 * not been cleared by a subsequent empty display command.
360 memset(sub, 0, sizeof(*sub));
361 // Blank if last object_number was 0.
362 // Note that this may be wrong for more complex subtitles.
363 if (!ctx->presentation.object_number)
365 sub->start_display_time = 0;
366 sub->end_display_time = 20000;
369 sub->rects = av_mallocz(sizeof(*sub->rects));
370 sub->rects[0] = av_mallocz(sizeof(*sub->rects[0]));
373 sub->rects[0]->x = ctx->presentation.x;
374 sub->rects[0]->y = ctx->presentation.y;
375 sub->rects[0]->w = ctx->picture.w;
376 sub->rects[0]->h = ctx->picture.h;
377 sub->rects[0]->type = SUBTITLE_BITMAP;
380 sub->rects[0]->pict.linesize[0] = ctx->picture.w;
382 if (ctx->picture.rle) {
383 if (ctx->picture.rle_remaining_len)
384 av_log(avctx, AV_LOG_ERROR, "RLE data length %u is %u bytes shorter than expected\n",
385 ctx->picture.rle_data_len, ctx->picture.rle_remaining_len);
386 if(decode_rle(avctx, sub, ctx->picture.rle, ctx->picture.rle_data_len) < 0)
389 /* Allocate memory for colors */
390 sub->rects[0]->nb_colors = 256;
391 sub->rects[0]->pict.data[1] = av_mallocz(AVPALETTE_SIZE);
393 memcpy(sub->rects[0]->pict.data[1], ctx->clut, sub->rects[0]->nb_colors * sizeof(uint32_t));
398 static int decode(AVCodecContext *avctx, void *data, int *data_size,
401 const uint8_t *buf = avpkt->data;
402 int buf_size = avpkt->size;
404 const uint8_t *buf_end;
405 uint8_t segment_type;
408 #ifdef DEBUG_PACKET_CONTENTS
411 av_log(avctx, AV_LOG_INFO, "PGS sub packet:\n");
413 for (i = 0; i < buf_size; i++) {
414 av_log(avctx, AV_LOG_INFO, "%02x ", buf[i]);
416 av_log(avctx, AV_LOG_INFO, "\n");
420 av_log(avctx, AV_LOG_INFO, "\n");
425 /* Ensure that we have received at a least a segment code and segment length */
429 buf_end = buf + buf_size;
431 /* Step through buffer to identify segments */
432 while (buf < buf_end) {
433 segment_type = bytestream_get_byte(&buf);
434 segment_length = bytestream_get_be16(&buf);
436 av_dlog(avctx, "Segment Length %d, Segment Type %x\n", segment_length, segment_type);
438 if (segment_type != DISPLAY_SEGMENT && segment_length > buf_end - buf)
441 switch (segment_type) {
442 case PALETTE_SEGMENT:
443 parse_palette_segment(avctx, buf, segment_length);
445 case PICTURE_SEGMENT:
446 parse_picture_segment(avctx, buf, segment_length);
448 case PRESENTATION_SEGMENT:
449 parse_presentation_segment(avctx, buf, segment_length);
453 * Window Segment Structure (No new information provided):
455 * 2 bytes: X position of subtitle,
456 * 2 bytes: Y position of subtitle,
457 * 2 bytes: Width of subtitle,
458 * 2 bytes: Height of subtitle.
461 case DISPLAY_SEGMENT:
462 *data_size = display_end_segment(avctx, data, buf, segment_length);
465 av_log(avctx, AV_LOG_ERROR, "Unknown subtitle segment type 0x%x, length %d\n",
466 segment_type, segment_length);
470 buf += segment_length;
476 AVCodec ff_pgssub_decoder = {
478 AVMEDIA_TYPE_SUBTITLE,
479 CODEC_ID_HDMV_PGS_SUBTITLE,
480 sizeof(PGSSubContext),
485 .long_name = NULL_IF_CONFIG_SMALL("HDMV Presentation Graphic Stream subtitles"),