ffmpeg/libavcodec/targa.c

337 lines
11 KiB
C

/*
* Targa (.tga) image decoder
* Copyright (c) 2006 Konstantin Shishkov
*
* 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
*/
#include "libavutil/intreadwrite.h"
#include "libavutil/imgutils.h"
#include "avcodec.h"
#include "bytestream.h"
#include "internal.h"
#include "targa.h"
typedef struct TargaContext {
AVFrame picture;
GetByteContext gb;
} TargaContext;
static uint8_t *advance_line(uint8_t *start, uint8_t *line,
int stride, int *y, int h, int interleave)
{
*y += interleave;
if (*y < h) {
return line + interleave * stride;
} else {
*y = (*y + 1) & (interleave - 1);
if (*y) {
return start + *y * stride;
} else {
return NULL;
}
}
}
static int targa_decode_rle(AVCodecContext *avctx, TargaContext *s,
uint8_t *start, int w, int h, int stride,
int bpp, int interleave)
{
int x, y;
int depth = (bpp + 1) >> 3;
int type, count;
uint8_t *line = start;
uint8_t *dst = line;
x = y = count = 0;
while (dst) {
if (bytestream2_get_bytes_left(&s->gb) <= 0) {
av_log(avctx, AV_LOG_ERROR,
"Ran ouf of data before end-of-image\n");
return AVERROR_INVALIDDATA;
}
type = bytestream2_get_byteu(&s->gb);
count = (type & 0x7F) + 1;
type &= 0x80;
if (!type) {
do {
int n = FFMIN(count, w - x);
bytestream2_get_buffer(&s->gb, dst, n * depth);
count -= n;
dst += n * depth;
x += n;
if (x == w) {
x = 0;
dst = line = advance_line(start, line, stride, &y, h, interleave);
}
} while (dst && count > 0);
} else {
uint8_t tmp[4];
bytestream2_get_buffer(&s->gb, tmp, depth);
do {
int n = FFMIN(count, w - x);
count -= n;
x += n;
do {
memcpy(dst, tmp, depth);
dst += depth;
} while (--n);
if (x == w) {
x = 0;
dst = line = advance_line(start, line, stride, &y, h, interleave);
}
} while (dst && count > 0);
}
}
if (count) {
av_log(avctx, AV_LOG_ERROR, "Packet went out of bounds\n");
return AVERROR_INVALIDDATA;
}
return 0;
}
static int decode_frame(AVCodecContext *avctx,
void *data, int *got_frame,
AVPacket *avpkt)
{
TargaContext * const s = avctx->priv_data;
AVFrame *picture = data;
AVFrame * const p = &s->picture;
uint8_t *dst;
int stride;
int idlen, pal, compr, y, w, h, bpp, flags, ret;
int first_clr, colors, csize;
int interleave;
bytestream2_init(&s->gb, avpkt->data, avpkt->size);
/* parse image header */
idlen = bytestream2_get_byte(&s->gb);
pal = bytestream2_get_byte(&s->gb);
compr = bytestream2_get_byte(&s->gb);
first_clr = bytestream2_get_le16(&s->gb);
colors = bytestream2_get_le16(&s->gb);
csize = bytestream2_get_byte(&s->gb);
bytestream2_skip(&s->gb, 4); /* 2: x, 2: y */
w = bytestream2_get_le16(&s->gb);
h = bytestream2_get_le16(&s->gb);
bpp = bytestream2_get_byte(&s->gb);
if (bytestream2_get_bytes_left(&s->gb) <= idlen) {
av_log(avctx, AV_LOG_ERROR,
"Not enough data to read header\n");
return AVERROR_INVALIDDATA;
}
flags = bytestream2_get_byte(&s->gb);
if (!pal && (first_clr || colors || csize)) {
av_log(avctx, AV_LOG_WARNING, "File without colormap has colormap information set.\n");
// specification says we should ignore those value in this case
first_clr = colors = csize = 0;
}
// skip identifier if any
bytestream2_skip(&s->gb, idlen);
switch (bpp) {
case 8:
avctx->pix_fmt = ((compr & (~TGA_RLE)) == TGA_BW) ? AV_PIX_FMT_GRAY8 : AV_PIX_FMT_PAL8;
break;
case 15:
case 16:
avctx->pix_fmt = AV_PIX_FMT_RGB555LE;
break;
case 24:
avctx->pix_fmt = AV_PIX_FMT_BGR24;
break;
case 32:
avctx->pix_fmt = AV_PIX_FMT_BGRA;
break;
default:
av_log(avctx, AV_LOG_ERROR, "Bit depth %i is not supported\n", bpp);
return AVERROR_INVALIDDATA;
}
if (s->picture.data[0])
avctx->release_buffer(avctx, &s->picture);
if (colors && (colors + first_clr) > 256) {
av_log(avctx, AV_LOG_ERROR, "Incorrect palette: %i colors with offset %i\n", colors, first_clr);
return AVERROR_INVALIDDATA;
}
if ((ret = av_image_check_size(w, h, 0, avctx)))
return ret;
if (w != avctx->width || h != avctx->height)
avcodec_set_dimensions(avctx, w, h);
if ((ret = ff_get_buffer(avctx, p)) < 0) {
av_log(avctx, AV_LOG_ERROR, "get_buffer() failed\n");
return ret;
}
if (flags & TGA_TOPTOBOTTOM) {
dst = p->data[0];
stride = p->linesize[0];
} else { //image is upside-down
dst = p->data[0] + p->linesize[0] * (h - 1);
stride = -p->linesize[0];
}
interleave = flags & TGA_INTERLEAVE2 ? 2 :
flags & TGA_INTERLEAVE4 ? 4 : 1;
if (colors) {
int pal_size, pal_sample_size;
switch (csize) {
case 32: pal_sample_size = 4; break;
case 24: pal_sample_size = 3; break;
case 16:
case 15: pal_sample_size = 2; break;
default:
av_log(avctx, AV_LOG_ERROR, "Palette entry size %i bits is not supported\n", csize);
return AVERROR_INVALIDDATA;
}
pal_size = colors * pal_sample_size;
if (avctx->pix_fmt != AV_PIX_FMT_PAL8) //should not occur but skip palette anyway
bytestream2_skip(&s->gb, pal_size);
else {
int t;
uint32_t *pal = ((uint32_t *)p->data[1]) + first_clr;
if (bytestream2_get_bytes_left(&s->gb) < pal_size) {
av_log(avctx, AV_LOG_ERROR,
"Not enough data to read palette\n");
return AVERROR_INVALIDDATA;
}
switch (pal_sample_size) {
case 4:
for (t = 0; t < colors; t++)
*pal++ = bytestream2_get_le32u(&s->gb);
break;
case 3:
/* RGB24 */
for (t = 0; t < colors; t++)
*pal++ = (0xffU<<24) | bytestream2_get_le24u(&s->gb);
break;
case 2:
/* RGB555 */
for (t = 0; t < colors; t++) {
uint32_t v = bytestream2_get_le16u(&s->gb);
v = ((v & 0x7C00) << 9) |
((v & 0x03E0) << 6) |
((v & 0x001F) << 3);
/* left bit replication */
v |= (v & 0xE0E0E0U) >> 5;
*pal++ = (0xffU<<24) | v;
}
break;
}
p->palette_has_changed = 1;
}
}
if ((compr & (~TGA_RLE)) == TGA_NODATA) {
memset(p->data[0], 0, p->linesize[0] * h);
} else {
if (compr & TGA_RLE) {
int res = targa_decode_rle(avctx, s, dst, w, h, stride, bpp, interleave);
if (res < 0)
return res;
} else {
size_t img_size = w * ((bpp + 1) >> 3);
uint8_t *line;
if (bytestream2_get_bytes_left(&s->gb) < img_size * h) {
av_log(avctx, AV_LOG_ERROR,
"Not enough data available for image\n");
return AVERROR_INVALIDDATA;
}
line = dst;
y = 0;
do {
bytestream2_get_bufferu(&s->gb, line, img_size);
line = advance_line(dst, line, stride, &y, h, interleave);
} while (line);
}
}
if (flags & TGA_RIGHTTOLEFT) { // right-to-left, needs horizontal flip
int x;
for (y = 0; y < h; y++) {
void *line = &p->data[0][y * p->linesize[0]];
for (x = 0; x < w >> 1; x++) {
switch (bpp) {
case 32:
FFSWAP(uint32_t, ((uint32_t *)line)[x], ((uint32_t *)line)[w - x - 1]);
break;
case 24:
FFSWAP(uint8_t, ((uint8_t *)line)[3 * x ], ((uint8_t *)line)[3 * w - 3 * x - 3]);
FFSWAP(uint8_t, ((uint8_t *)line)[3 * x + 1], ((uint8_t *)line)[3 * w - 3 * x - 2]);
FFSWAP(uint8_t, ((uint8_t *)line)[3 * x + 2], ((uint8_t *)line)[3 * w - 3 * x - 1]);
break;
case 16:
FFSWAP(uint16_t, ((uint16_t *)line)[x], ((uint16_t *)line)[w - x - 1]);
break;
case 8:
FFSWAP(uint8_t, ((uint8_t *)line)[x], ((uint8_t *)line)[w - x - 1]);
}
}
}
}
*picture = s->picture;
*got_frame = 1;
return avpkt->size;
}
static av_cold int targa_init(AVCodecContext *avctx)
{
TargaContext *s = avctx->priv_data;
avcodec_get_frame_defaults(&s->picture);
avctx->coded_frame = &s->picture;
return 0;
}
static av_cold int targa_end(AVCodecContext *avctx)
{
TargaContext *s = avctx->priv_data;
if (s->picture.data[0])
avctx->release_buffer(avctx, &s->picture);
return 0;
}
AVCodec ff_targa_decoder = {
.name = "targa",
.type = AVMEDIA_TYPE_VIDEO,
.id = AV_CODEC_ID_TARGA,
.priv_data_size = sizeof(TargaContext),
.init = targa_init,
.close = targa_end,
.decode = decode_frame,
.capabilities = CODEC_CAP_DR1,
.long_name = NULL_IF_CONFIG_SMALL("Truevision Targa image"),
};