1
0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-11-26 19:01:44 +02:00
FFmpeg/libavcodec/pgssubdec.c
Andreas Rheinhardt 6834c79636 avcodec/pgssubdec: Mark decoder as init-threadsafe
Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@gmail.com>
2021-05-02 02:34:46 +02:00

718 lines
22 KiB
C

/*
* PGS subtitle decoder
* Copyright (c) 2009 Stephen Backway
*
* This file is part of FFmpeg.
*
* FFmpeg is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* FFmpeg is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
/**
* @file
* PGS subtitle decoder
*/
#include "avcodec.h"
#include "bytestream.h"
#include "internal.h"
#include "mathops.h"
#include "libavutil/colorspace.h"
#include "libavutil/imgutils.h"
#include "libavutil/opt.h"
#define RGBA(r,g,b,a) (((unsigned)(a) << 24) | ((r) << 16) | ((g) << 8) | (b))
#define MAX_EPOCH_PALETTES 8 // Max 8 allowed per PGS epoch
#define MAX_EPOCH_OBJECTS 64 // Max 64 allowed per PGS epoch
#define MAX_OBJECT_REFS 2 // Max objects per display set
enum SegmentType {
PALETTE_SEGMENT = 0x14,
OBJECT_SEGMENT = 0x15,
PRESENTATION_SEGMENT = 0x16,
WINDOW_SEGMENT = 0x17,
DISPLAY_SEGMENT = 0x80,
};
typedef struct PGSSubObjectRef {
int id;
int window_id;
uint8_t composition_flag;
int x;
int y;
int crop_x;
int crop_y;
int crop_w;
int crop_h;
} PGSSubObjectRef;
typedef struct PGSSubPresentation {
int id_number;
int palette_id;
int object_count;
PGSSubObjectRef objects[MAX_OBJECT_REFS];
int64_t pts;
} PGSSubPresentation;
typedef struct PGSSubObject {
int id;
int w;
int h;
uint8_t *rle;
unsigned int rle_buffer_size, rle_data_len;
unsigned int rle_remaining_len;
} PGSSubObject;
typedef struct PGSSubObjects {
int count;
PGSSubObject object[MAX_EPOCH_OBJECTS];
} PGSSubObjects;
typedef struct PGSSubPalette {
int id;
uint32_t clut[256];
} PGSSubPalette;
typedef struct PGSSubPalettes {
int count;
PGSSubPalette palette[MAX_EPOCH_PALETTES];
} PGSSubPalettes;
typedef struct PGSSubContext {
AVClass *class;
PGSSubPresentation presentation;
PGSSubPalettes palettes;
PGSSubObjects objects;
int forced_subs_only;
} PGSSubContext;
static void flush_cache(AVCodecContext *avctx)
{
PGSSubContext *ctx = avctx->priv_data;
int i;
for (i = 0; i < ctx->objects.count; i++) {
av_freep(&ctx->objects.object[i].rle);
ctx->objects.object[i].rle_buffer_size = 0;
ctx->objects.object[i].rle_remaining_len = 0;
}
ctx->objects.count = 0;
ctx->palettes.count = 0;
}
static PGSSubObject * find_object(int id, PGSSubObjects *objects)
{
int i;
for (i = 0; i < objects->count; i++) {
if (objects->object[i].id == id)
return &objects->object[i];
}
return NULL;
}
static PGSSubPalette * find_palette(int id, PGSSubPalettes *palettes)
{
int i;
for (i = 0; i < palettes->count; i++) {
if (palettes->palette[i].id == id)
return &palettes->palette[i];
}
return NULL;
}
static av_cold int init_decoder(AVCodecContext *avctx)
{
avctx->pix_fmt = AV_PIX_FMT_PAL8;
return 0;
}
static av_cold int close_decoder(AVCodecContext *avctx)
{
flush_cache(avctx);
return 0;
}
/**
* Decode the RLE data.
*
* The subtitle is stored as a Run Length Encoded image.
*
* @param avctx contains the current codec context
* @param sub pointer to the processed subtitle data
* @param buf pointer to the RLE data to process
* @param buf_size size of the RLE data to process
*/
static int decode_rle(AVCodecContext *avctx, AVSubtitleRect *rect,
const uint8_t *buf, unsigned int buf_size)
{
const uint8_t *rle_bitmap_end;
int pixel_count, line_count;
rle_bitmap_end = buf + buf_size;
rect->data[0] = av_malloc_array(rect->w, rect->h);
if (!rect->data[0])
return AVERROR(ENOMEM);
pixel_count = 0;
line_count = 0;
while (buf < rle_bitmap_end && line_count < rect->h) {
uint8_t flags, color;
int run;
color = bytestream_get_byte(&buf);
run = 1;
if (color == 0x00) {
flags = bytestream_get_byte(&buf);
run = flags & 0x3f;
if (flags & 0x40)
run = (run << 8) + bytestream_get_byte(&buf);
color = flags & 0x80 ? bytestream_get_byte(&buf) : 0;
}
if (run > 0 && pixel_count + run <= rect->w * rect->h) {
memset(rect->data[0] + pixel_count, color, run);
pixel_count += run;
} else if (!run) {
/*
* New Line. Check if correct pixels decoded, if not display warning
* and adjust bitmap pointer to correct new line position.
*/
if (pixel_count % rect->w > 0) {
av_log(avctx, AV_LOG_ERROR, "Decoded %d pixels, when line should be %d pixels\n",
pixel_count % rect->w, rect->w);
if (avctx->err_recognition & AV_EF_EXPLODE) {
return AVERROR_INVALIDDATA;
}
}
line_count++;
}
}
if (pixel_count < rect->w * rect->h) {
av_log(avctx, AV_LOG_ERROR, "Insufficient RLE data for subtitle\n");
return AVERROR_INVALIDDATA;
}
ff_dlog(avctx, "Pixel Count = %d, Area = %d\n", pixel_count, rect->w * rect->h);
return 0;
}
/**
* Parse the picture segment packet.
*
* The picture segment contains details on the sequence id,
* width, height and Run Length Encoded (RLE) bitmap data.
*
* @param avctx contains the current codec context
* @param buf pointer to the packet to process
* @param buf_size size of packet to process
*/
static int parse_object_segment(AVCodecContext *avctx,
const uint8_t *buf, int buf_size)
{
PGSSubContext *ctx = avctx->priv_data;
PGSSubObject *object;
uint8_t sequence_desc;
unsigned int rle_bitmap_len, width, height;
int id;
if (buf_size <= 4)
return AVERROR_INVALIDDATA;
buf_size -= 4;
id = bytestream_get_be16(&buf);
object = find_object(id, &ctx->objects);
if (!object) {
if (ctx->objects.count >= MAX_EPOCH_OBJECTS) {
av_log(avctx, AV_LOG_ERROR, "Too many objects in epoch\n");
return AVERROR_INVALIDDATA;
}
object = &ctx->objects.object[ctx->objects.count++];
object->id = id;
}
/* skip object version number */
buf += 1;
/* Read the Sequence Description to determine if start of RLE data or appended to previous RLE */
sequence_desc = bytestream_get_byte(&buf);
if (!(sequence_desc & 0x80)) {
/* Additional RLE data */
if (buf_size > object->rle_remaining_len)
return AVERROR_INVALIDDATA;
memcpy(object->rle + object->rle_data_len, buf, buf_size);
object->rle_data_len += buf_size;
object->rle_remaining_len -= buf_size;
return 0;
}
if (buf_size <= 7)
return AVERROR_INVALIDDATA;
buf_size -= 7;
/* Decode rle bitmap length, stored size includes width/height data */
rle_bitmap_len = bytestream_get_be24(&buf) - 2*2;
if (buf_size > rle_bitmap_len) {
av_log(avctx, AV_LOG_ERROR,
"Buffer dimension %d larger than the expected RLE data %d\n",
buf_size, rle_bitmap_len);
return AVERROR_INVALIDDATA;
}
/* Get bitmap dimensions from data */
width = bytestream_get_be16(&buf);
height = bytestream_get_be16(&buf);
/* Make sure the bitmap is not too large */
if (avctx->width < width || avctx->height < height || !width || !height) {
av_log(avctx, AV_LOG_ERROR, "Bitmap dimensions (%dx%d) invalid.\n", width, height);
return AVERROR_INVALIDDATA;
}
object->w = width;
object->h = height;
av_fast_padded_malloc(&object->rle, &object->rle_buffer_size, rle_bitmap_len);
if (!object->rle) {
object->rle_data_len = 0;
object->rle_remaining_len = 0;
return AVERROR(ENOMEM);
}
memcpy(object->rle, buf, buf_size);
object->rle_data_len = buf_size;
object->rle_remaining_len = rle_bitmap_len - buf_size;
return 0;
}
/**
* Parse the palette segment packet.
*
* The palette segment contains details of the palette,
* a maximum of 256 colors can be defined.
*
* @param avctx contains the current codec context
* @param buf pointer to the packet to process
* @param buf_size size of packet to process
*/
static int parse_palette_segment(AVCodecContext *avctx,
const uint8_t *buf, int buf_size)
{
PGSSubContext *ctx = avctx->priv_data;
PGSSubPalette *palette;
const uint8_t *buf_end = buf + buf_size;
const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP;
int color_id;
int y, cb, cr, alpha;
int r, g, b, r_add, g_add, b_add;
int id;
id = bytestream_get_byte(&buf);
palette = find_palette(id, &ctx->palettes);
if (!palette) {
if (ctx->palettes.count >= MAX_EPOCH_PALETTES) {
av_log(avctx, AV_LOG_ERROR, "Too many palettes in epoch\n");
return AVERROR_INVALIDDATA;
}
palette = &ctx->palettes.palette[ctx->palettes.count++];
palette->id = id;
}
/* Skip palette version */
buf += 1;
while (buf < buf_end) {
color_id = bytestream_get_byte(&buf);
y = bytestream_get_byte(&buf);
cr = bytestream_get_byte(&buf);
cb = bytestream_get_byte(&buf);
alpha = bytestream_get_byte(&buf);
/* Default to BT.709 colorspace. In case of <= 576 height use BT.601 */
if (avctx->height <= 0 || avctx->height > 576) {
YUV_TO_RGB1_CCIR_BT709(cb, cr);
} else {
YUV_TO_RGB1_CCIR(cb, cr);
}
YUV_TO_RGB2_CCIR(r, g, b, y);
ff_dlog(avctx, "Color %d := (%d,%d,%d,%d)\n", color_id, r, g, b, alpha);
/* Store color in palette */
palette->clut[color_id] = RGBA(r,g,b,alpha);
}
return 0;
}
/**
* Parse the presentation segment packet.
*
* The presentation segment contains details on the video
* width, video height, x & y subtitle position.
*
* @param avctx contains the current codec context
* @param buf pointer to the packet to process
* @param buf_size size of packet to process
* @todo TODO: Implement cropping
*/
static int parse_presentation_segment(AVCodecContext *avctx,
const uint8_t *buf, int buf_size,
int64_t pts)
{
PGSSubContext *ctx = avctx->priv_data;
int i, state, ret;
const uint8_t *buf_end = buf + buf_size;
// Video descriptor
int w = bytestream_get_be16(&buf);
int h = bytestream_get_be16(&buf);
ctx->presentation.pts = pts;
ff_dlog(avctx, "Video Dimensions %dx%d\n",
w, h);
ret = ff_set_dimensions(avctx, w, h);
if (ret < 0)
return ret;
/* Skip 1 bytes of unknown, frame rate */
buf++;
// Composition descriptor
ctx->presentation.id_number = bytestream_get_be16(&buf);
/*
* state is a 2 bit field that defines pgs epoch boundaries
* 00 - Normal, previously defined objects and palettes are still valid
* 01 - Acquisition point, previous objects and palettes can be released
* 10 - Epoch start, previous objects and palettes can be released
* 11 - Epoch continue, previous objects and palettes can be released
*
* reserved 6 bits discarded
*/
state = bytestream_get_byte(&buf) >> 6;
if (state != 0) {
flush_cache(avctx);
}
/*
* skip palette_update_flag (0x80),
*/
buf += 1;
ctx->presentation.palette_id = bytestream_get_byte(&buf);
ctx->presentation.object_count = bytestream_get_byte(&buf);
if (ctx->presentation.object_count > MAX_OBJECT_REFS) {
av_log(avctx, AV_LOG_ERROR,
"Invalid number of presentation objects %d\n",
ctx->presentation.object_count);
ctx->presentation.object_count = 2;
if (avctx->err_recognition & AV_EF_EXPLODE) {
return AVERROR_INVALIDDATA;
}
}
for (i = 0; i < ctx->presentation.object_count; i++)
{
if (buf_end - buf < 8) {
av_log(avctx, AV_LOG_ERROR, "Insufficent space for object\n");
ctx->presentation.object_count = i;
return AVERROR_INVALIDDATA;
}
ctx->presentation.objects[i].id = bytestream_get_be16(&buf);
ctx->presentation.objects[i].window_id = bytestream_get_byte(&buf);
ctx->presentation.objects[i].composition_flag = bytestream_get_byte(&buf);
ctx->presentation.objects[i].x = bytestream_get_be16(&buf);
ctx->presentation.objects[i].y = bytestream_get_be16(&buf);
// If cropping
if (ctx->presentation.objects[i].composition_flag & 0x80) {
ctx->presentation.objects[i].crop_x = bytestream_get_be16(&buf);
ctx->presentation.objects[i].crop_y = bytestream_get_be16(&buf);
ctx->presentation.objects[i].crop_w = bytestream_get_be16(&buf);
ctx->presentation.objects[i].crop_h = bytestream_get_be16(&buf);
}
ff_dlog(avctx, "Subtitle Placement x=%d, y=%d\n",
ctx->presentation.objects[i].x, ctx->presentation.objects[i].y);
if (ctx->presentation.objects[i].x > avctx->width ||
ctx->presentation.objects[i].y > avctx->height) {
av_log(avctx, AV_LOG_ERROR, "Subtitle out of video bounds. x = %d, y = %d, video width = %d, video height = %d.\n",
ctx->presentation.objects[i].x,
ctx->presentation.objects[i].y,
avctx->width, avctx->height);
ctx->presentation.objects[i].x = 0;
ctx->presentation.objects[i].y = 0;
if (avctx->err_recognition & AV_EF_EXPLODE) {
return AVERROR_INVALIDDATA;
}
}
}
return 0;
}
/**
* Parse the display segment packet.
*
* The display segment controls the updating of the display.
*
* @param avctx contains the current codec context
* @param data pointer to the data pertaining the subtitle to display
* @param buf pointer to the packet to process
* @param buf_size size of packet to process
*/
static int display_end_segment(AVCodecContext *avctx, void *data,
const uint8_t *buf, int buf_size)
{
AVSubtitle *sub = data;
PGSSubContext *ctx = avctx->priv_data;
int64_t pts;
PGSSubPalette *palette;
int i, ret;
pts = ctx->presentation.pts != AV_NOPTS_VALUE ? ctx->presentation.pts : sub->pts;
memset(sub, 0, sizeof(*sub));
sub->pts = pts;
ctx->presentation.pts = AV_NOPTS_VALUE;
sub->start_display_time = 0;
// There is no explicit end time for PGS subtitles. The end time
// is defined by the start of the next sub which may contain no
// objects (i.e. clears the previous sub)
sub->end_display_time = UINT32_MAX;
sub->format = 0;
// Blank if last object_count was 0.
if (!ctx->presentation.object_count)
return 1;
sub->rects = av_mallocz_array(ctx->presentation.object_count, sizeof(*sub->rects));
if (!sub->rects) {
return AVERROR(ENOMEM);
}
palette = find_palette(ctx->presentation.palette_id, &ctx->palettes);
if (!palette) {
// Missing palette. Should only happen with damaged streams.
av_log(avctx, AV_LOG_ERROR, "Invalid palette id %d\n",
ctx->presentation.palette_id);
avsubtitle_free(sub);
return AVERROR_INVALIDDATA;
}
for (i = 0; i < ctx->presentation.object_count; i++) {
PGSSubObject *object;
sub->rects[i] = av_mallocz(sizeof(*sub->rects[0]));
if (!sub->rects[i]) {
avsubtitle_free(sub);
return AVERROR(ENOMEM);
}
sub->num_rects++;
sub->rects[i]->type = SUBTITLE_BITMAP;
/* Process bitmap */
object = find_object(ctx->presentation.objects[i].id, &ctx->objects);
if (!object) {
// Missing object. Should only happen with damaged streams.
av_log(avctx, AV_LOG_ERROR, "Invalid object id %d\n",
ctx->presentation.objects[i].id);
if (avctx->err_recognition & AV_EF_EXPLODE) {
avsubtitle_free(sub);
return AVERROR_INVALIDDATA;
}
// Leaves rect empty with 0 width and height.
continue;
}
if (ctx->presentation.objects[i].composition_flag & 0x40)
sub->rects[i]->flags |= AV_SUBTITLE_FLAG_FORCED;
sub->rects[i]->x = ctx->presentation.objects[i].x;
sub->rects[i]->y = ctx->presentation.objects[i].y;
if (object->rle) {
sub->rects[i]->w = object->w;
sub->rects[i]->h = object->h;
sub->rects[i]->linesize[0] = object->w;
if (object->rle_remaining_len) {
av_log(avctx, AV_LOG_ERROR, "RLE data length %u is %u bytes shorter than expected\n",
object->rle_data_len, object->rle_remaining_len);
if (avctx->err_recognition & AV_EF_EXPLODE) {
avsubtitle_free(sub);
return AVERROR_INVALIDDATA;
}
}
ret = decode_rle(avctx, sub->rects[i], object->rle, object->rle_data_len);
if (ret < 0) {
if ((avctx->err_recognition & AV_EF_EXPLODE) ||
ret == AVERROR(ENOMEM)) {
avsubtitle_free(sub);
return ret;
}
sub->rects[i]->w = 0;
sub->rects[i]->h = 0;
continue;
}
}
/* Allocate memory for colors */
sub->rects[i]->nb_colors = 256;
sub->rects[i]->data[1] = av_mallocz(AVPALETTE_SIZE);
if (!sub->rects[i]->data[1]) {
avsubtitle_free(sub);
return AVERROR(ENOMEM);
}
if (!ctx->forced_subs_only || ctx->presentation.objects[i].composition_flag & 0x40)
memcpy(sub->rects[i]->data[1], palette->clut, sub->rects[i]->nb_colors * sizeof(uint32_t));
}
return 1;
}
static int decode(AVCodecContext *avctx, void *data, int *got_sub_ptr,
AVPacket *avpkt)
{
const uint8_t *buf = avpkt->data;
int buf_size = avpkt->size;
const uint8_t *buf_end;
uint8_t segment_type;
int segment_length;
int i, ret;
ff_dlog(avctx, "PGS sub packet:\n");
for (i = 0; i < buf_size; i++) {
ff_dlog(avctx, "%02x ", buf[i]);
if (i % 16 == 15)
ff_dlog(avctx, "\n");
}
if (i & 15)
ff_dlog(avctx, "\n");
*got_sub_ptr = 0;
/* Ensure that we have received at a least a segment code and segment length */
if (buf_size < 3)
return -1;
buf_end = buf + buf_size;
/* Step through buffer to identify segments */
while (buf < buf_end) {
segment_type = bytestream_get_byte(&buf);
segment_length = bytestream_get_be16(&buf);
ff_dlog(avctx, "Segment Length %d, Segment Type %x\n", segment_length, segment_type);
if (segment_type != DISPLAY_SEGMENT && segment_length > buf_end - buf)
break;
ret = 0;
switch (segment_type) {
case PALETTE_SEGMENT:
ret = parse_palette_segment(avctx, buf, segment_length);
break;
case OBJECT_SEGMENT:
ret = parse_object_segment(avctx, buf, segment_length);
break;
case PRESENTATION_SEGMENT:
ret = parse_presentation_segment(avctx, buf, segment_length, ((AVSubtitle*)(data))->pts);
break;
case WINDOW_SEGMENT:
/*
* Window Segment Structure (No new information provided):
* 2 bytes: Unknown,
* 2 bytes: X position of subtitle,
* 2 bytes: Y position of subtitle,
* 2 bytes: Width of subtitle,
* 2 bytes: Height of subtitle.
*/
break;
case DISPLAY_SEGMENT:
if (*got_sub_ptr) {
av_log(avctx, AV_LOG_ERROR, "Duplicate display segment\n");
ret = AVERROR_INVALIDDATA;
break;
}
ret = display_end_segment(avctx, data, buf, segment_length);
if (ret >= 0)
*got_sub_ptr = ret;
break;
default:
av_log(avctx, AV_LOG_ERROR, "Unknown subtitle segment type 0x%x, length %d\n",
segment_type, segment_length);
ret = AVERROR_INVALIDDATA;
break;
}
if (ret < 0 && (avctx->err_recognition & AV_EF_EXPLODE)) {
avsubtitle_free(data);
*got_sub_ptr = 0;
return ret;
}
buf += segment_length;
}
return buf_size;
}
#define OFFSET(x) offsetof(PGSSubContext, x)
#define SD AV_OPT_FLAG_SUBTITLE_PARAM | AV_OPT_FLAG_DECODING_PARAM
static const AVOption options[] = {
{"forced_subs_only", "Only show forced subtitles", OFFSET(forced_subs_only), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, SD},
{ NULL },
};
static const AVClass pgsdec_class = {
.class_name = "PGS subtitle decoder",
.item_name = av_default_item_name,
.option = options,
.version = LIBAVUTIL_VERSION_INT,
};
const AVCodec ff_pgssub_decoder = {
.name = "pgssub",
.long_name = NULL_IF_CONFIG_SMALL("HDMV Presentation Graphic Stream subtitles"),
.type = AVMEDIA_TYPE_SUBTITLE,
.id = AV_CODEC_ID_HDMV_PGS_SUBTITLE,
.priv_data_size = sizeof(PGSSubContext),
.init = init_decoder,
.close = close_decoder,
.decode = decode,
.priv_class = &pgsdec_class,
.caps_internal = FF_CODEC_CAP_INIT_THREADSAFE,
};