demux: heif: send extradata with avif
[vlc.git] / modules / codec / cvdsub.c
blob9037d6ebf9b706752ff52a4ec3a0f805af4283e3
1 /*****************************************************************************
2 * cvdsub.c : CVD Subtitle decoder
3 *****************************************************************************
4 * Copyright (C) 2003, 2004 VLC authors and VideoLAN
5 * $Id$
7 * Authors: Rocky Bernstein
8 * Gildas Bazin <gbazin@videolan.org>
9 * Julio Sanchez Fernandez (http://subhandler.sourceforge.net)
10 * Laurent Aimar <fenrir@via.ecp.fr>
12 * This program is free software; you can redistribute it and/or modify it
13 * under the terms of the GNU Lesser General Public License as published by
14 * the Free Software Foundation; either version 2.1 of the License, or
15 * (at your option) any later version.
17 * This program is distributed in the hope that it will be useful,
18 * but WITHOUT ANY WARRANTY; without even the implied warranty of
19 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
20 * GNU Lesser General Public License for more details.
22 * You should have received a copy of the GNU Lesser General Public License
23 * along with this program; if not, write to the Free Software Foundation,
24 * Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
25 *****************************************************************************/
27 /*****************************************************************************
28 * Preamble
29 *****************************************************************************/
30 #ifdef HAVE_CONFIG_H
31 # include "config.h"
32 #endif
34 #include <vlc_common.h>
35 #include <vlc_plugin.h>
36 #include <vlc_codec.h>
38 #include <vlc_bits.h>
40 #include "../demux/mpeg/timestamps.h"
42 #define DEBUG_CVDSUB 1
44 /*****************************************************************************
45 * Module descriptor.
46 *****************************************************************************/
47 static int DecoderOpen ( vlc_object_t * );
48 static int PacketizerOpen( vlc_object_t * );
49 static void DecoderClose ( vlc_object_t * );
51 vlc_module_begin ()
52 set_description( N_("CVD subtitle decoder") )
53 set_capability( "spu decoder", 50 )
54 set_callbacks( DecoderOpen, DecoderClose )
56 add_submodule ()
57 set_description( N_("Chaoji VCD subtitle packetizer") )
58 set_capability( "packetizer", 50 )
59 set_callbacks( PacketizerOpen, DecoderClose )
60 vlc_module_end ()
62 /*****************************************************************************
63 * Local prototypes
64 *****************************************************************************/
65 static int Decode( decoder_t *, block_t * );
66 static block_t *Packetize ( decoder_t *, block_t ** );
67 static block_t *Reassemble ( decoder_t *, block_t * );
68 static void ParseMetaInfo ( decoder_t *, block_t * );
69 static void ParseHeader ( decoder_t *, block_t * );
70 static subpicture_t *DecodePacket( decoder_t *, block_t * );
71 static void RenderImage( decoder_t *, block_t *, subpicture_region_t * );
73 #define SUBTITLE_BLOCK_EMPTY 0
74 #define SUBTITLE_BLOCK_PARTIAL 1
75 #define SUBTITLE_BLOCK_COMPLETE 2
77 typedef struct
79 int b_packetizer;
81 int i_state; /* data-gathering state for this subtitle */
83 block_t *p_spu; /* Bytes of the packet. */
85 size_t i_spu_size; /* goal for subtitle_data_pos while gathering,
86 size of used subtitle_data later */
88 uint16_t i_image_offset; /* offset from subtitle_data to compressed
89 image data */
90 size_t i_image_length; /* size of the compressed image data */
91 size_t first_field_offset; /* offset of even raster lines */
92 size_t second_field_offset; /* offset of odd raster lines */
93 size_t metadata_offset; /* offset to data describing the image */
94 size_t metadata_length; /* length of metadata */
96 vlc_tick_t i_duration; /* how long to display the image, 0 stands
97 for "until next subtitle" */
99 uint16_t i_x_start, i_y_start; /* position of top leftmost pixel of
100 image when displayed */
101 uint16_t i_width, i_height; /* dimensions in pixels of image */
103 uint8_t p_palette[4][4]; /* Palette of colors used in subtitle */
104 uint8_t p_palette_highlight[4][4];
105 } decoder_sys_t;
107 static int OpenCommon( vlc_object_t *p_this, bool b_packetizer )
109 decoder_t *p_dec = (decoder_t*)p_this;
110 decoder_sys_t *p_sys;
112 if( p_dec->fmt_in.i_codec != VLC_CODEC_CVD )
113 return VLC_EGENERIC;
115 p_dec->p_sys = p_sys = malloc( sizeof( decoder_sys_t ) );
116 if( !p_sys )
117 return VLC_ENOMEM;
119 p_sys->b_packetizer = b_packetizer;
121 p_sys->i_state = SUBTITLE_BLOCK_EMPTY;
122 p_sys->p_spu = NULL;
124 if( b_packetizer )
126 p_dec->pf_packetize = Packetize;
127 p_dec->fmt_out.i_codec = VLC_CODEC_CVD;
129 else
131 p_dec->pf_decode = Decode;
132 p_dec->fmt_out.i_codec = VLC_CODEC_YUVP;
135 return VLC_SUCCESS;
137 /*****************************************************************************
138 * DecoderOpen: open/initialize the cvdsub decoder.
139 *****************************************************************************/
140 static int DecoderOpen( vlc_object_t *p_this )
142 return OpenCommon( p_this, false );
145 /*****************************************************************************
146 * PacketizerOpen: open/initialize the cvdsub packetizer.
147 *****************************************************************************/
148 static int PacketizerOpen( vlc_object_t *p_this )
150 return OpenCommon( p_this, true );
153 /*****************************************************************************
154 * DecoderClose: closes the cvdsub decoder/packetizer.
155 *****************************************************************************/
156 void DecoderClose( vlc_object_t *p_this )
158 decoder_t *p_dec = (decoder_t*)p_this;
159 decoder_sys_t *p_sys = p_dec->p_sys;
161 if( p_sys->p_spu ) block_ChainRelease( p_sys->p_spu );
162 free( p_sys );
165 /*****************************************************************************
166 * Decode:
167 *****************************************************************************/
168 static int Decode( decoder_t *p_dec, block_t *p_block )
170 block_t *p_data;
172 if( p_block == NULL ) /* No Drain */
173 return VLCDEC_SUCCESS;
175 if( p_block->i_flags & BLOCK_FLAG_CORRUPTED )
177 block_Release( p_block );
178 return VLCDEC_SUCCESS;
181 if( !(p_data = Reassemble( p_dec, p_block )) )
182 return VLCDEC_SUCCESS;
184 /* Parse and decode */
185 subpicture_t *p_spu = DecodePacket( p_dec, p_data );
186 block_Release( p_data );
187 if( p_spu != NULL )
188 decoder_QueueSub( p_dec, p_spu );
189 return VLCDEC_SUCCESS;
192 /*****************************************************************************
193 * Packetize:
194 *****************************************************************************/
195 static block_t *Packetize( decoder_t *p_dec, block_t **pp_block )
197 block_t *p_block, *p_spu;
199 if( pp_block == NULL || *pp_block == NULL ) return NULL;
201 p_block = *pp_block;
202 *pp_block = NULL;
204 if( !(p_spu = Reassemble( p_dec, p_block )) ) return NULL;
206 p_spu->i_dts = p_spu->i_pts;
207 p_spu->i_length = VLC_TICK_INVALID;
209 return p_spu;
213 /*****************************************************************************
214 Reassemble:
216 Data for single screen subtitle may come in several non-contiguous
217 packets of a stream. This routine is called when the next packet in
218 the stream comes in. The job of this routine is to parse the header,
219 if this is the beginning, and combine the packets into one complete
220 subtitle unit.
222 If everything is complete, we will return a block. Otherwise return
223 NULL.
225 *****************************************************************************/
226 #define SPU_HEADER_LEN 1
228 static block_t *Reassemble( decoder_t *p_dec, block_t *p_block )
230 decoder_sys_t *p_sys = p_dec->p_sys;
232 if( p_block->i_buffer < SPU_HEADER_LEN )
234 msg_Dbg( p_dec, "invalid packet header (size %zu < %u)" ,
235 p_block->i_buffer, SPU_HEADER_LEN );
236 block_Release( p_block );
237 return NULL;
240 /* From the scant data on the format, there is only only way known
241 * to detect the first packet in a subtitle. The first packet
242 * seems to have a valid PTS while later packets for the same
243 * image don't. */
244 if( p_sys->i_state == SUBTITLE_BLOCK_EMPTY && p_block->i_pts == VLC_TICK_INVALID )
246 msg_Warn( p_dec, "first packet expected but no PTS present");
247 return NULL;
250 p_block->p_buffer += SPU_HEADER_LEN;
251 p_block->i_buffer -= SPU_HEADER_LEN;
253 /* First packet in the subtitle block */
254 if( p_sys->i_state == SUBTITLE_BLOCK_EMPTY ) ParseHeader( p_dec, p_block );
256 block_ChainAppend( &p_sys->p_spu, p_block );
257 p_sys->p_spu = block_ChainGather( p_sys->p_spu );
259 if( p_sys->p_spu->i_buffer >= p_sys->i_spu_size )
261 block_t *p_spu = p_sys->p_spu;
263 if( p_spu->i_buffer != p_sys->i_spu_size )
265 msg_Warn( p_dec, "SPU packets size=%zu should be %zu",
266 p_spu->i_buffer, p_sys->i_spu_size );
269 msg_Dbg( p_dec, "subtitle packet complete, size=%zuu", p_spu->i_buffer);
271 ParseMetaInfo( p_dec, p_spu );
273 p_sys->i_state = SUBTITLE_BLOCK_EMPTY;
274 p_sys->p_spu = 0;
275 return p_spu;
277 else
279 /* Not last block in subtitle, so wait for another. */
280 p_sys->i_state = SUBTITLE_BLOCK_PARTIAL;
283 return NULL;
287 We do not have information on the subtitle format used on CVD's
288 except the submux sample code and a couple of samples of dubious
289 origin. Thus, this is the result of reading some code whose
290 correctness is not known and some experimentation.
292 CVD subtitles are different in several ways from SVCD OGT subtitles.
293 Image comes first and metadata is at the end. So that the metadata
294 can be found easily, the subtitle packet starts with two bytes
295 (everything is big-endian again) that give the total size of the
296 subtitle data and the offset to the metadata - i.e. size of the
297 image data plus the four bytes at the beginning.
299 Image data comes interlaced is run-length encoded. Each field is a
300 four-bit nibble. Each nibble contains a two-bit repeat count and a
301 two-bit color number so that up to three pixels can be described in
302 four bits. The function of a 0 repeat count is unknown; it might be
303 used for RLE extension. However when the full nibble is zero, the
304 rest of the line is filled with the color value in the next nibble.
305 It is unknown what happens if the color value is greater than three.
306 The rest seems to use a 4-entries palette. It is not impossible
307 that the fill-line complete case above is not as described and the
308 zero repeat count means fill line. The sample code never produces
309 this, so it may be untested.
312 static void ParseHeader( decoder_t *p_dec, block_t *p_block )
314 decoder_sys_t *p_sys = p_dec->p_sys;
315 uint8_t *p = p_block->p_buffer;
317 p_sys->i_spu_size = (p[0] << 8) + p[1] + 4; p += 2;
319 /* FIXME: check data sanity */
320 p_sys->metadata_offset = (p[0] << 8) + p[1]; p +=2;
321 p_sys->metadata_length = p_sys->i_spu_size - p_sys->metadata_offset;
323 p_sys->i_image_offset = 4;
324 p_sys->i_image_length = p_sys->metadata_offset - p_sys->i_image_offset;
326 #ifdef DEBUG_CVDSUB
327 msg_Dbg( p_dec, "total size: %zu image size: %zu",
328 p_sys->i_spu_size, p_sys->i_image_length );
329 #endif
333 We parse the metadata information here.
335 Although metadata information does not have to come in a fixed field
336 order, every metadata field consists of a tag byte followed by
337 parameters. In all cases known, the size including tag byte is
338 exactly four bytes in length.
341 #define ExtractXY(x, y) x = ((p[1]&0x0f)<<6) + (p[2]>>2); \
342 y = ((p[2]&0x03)<<8) + p[3];
344 static void ParseMetaInfo( decoder_t *p_dec, block_t *p_spu )
346 /* Last packet in subtitle block. */
348 decoder_sys_t *p_sys = p_dec->p_sys;
349 uint8_t *p = p_spu->p_buffer + p_sys->metadata_offset;
350 uint8_t *p_end = p + p_sys->metadata_length;
352 for( ; p < p_end; p += 4 )
354 switch( p[0] )
356 case 0x04: /* subtitle duration in 1/90000ths of a second */
357 p_sys->i_duration = FROM_SCALE_NZ( (p[1]<<16) + (p[2]<<8) + p[3] );
359 #ifdef DEBUG_CVDSUB
360 msg_Dbg( p_dec, "subtitle display duration %"PRIu64" ms",
361 MS_FROM_VLC_TICK(p_sys->i_duration) );
362 #endif
363 break;
365 case 0x0c: /* unknown */
366 #ifdef DEBUG_CVDSUB
367 msg_Dbg( p_dec, "subtitle command unknown "
368 "0x%02"PRIx8" 0x%02"PRIx8" 0x%02"PRIx8" 0x%02"PRIx8,
369 p[0], p[1], p[2], p[3] );
370 #endif
371 break;
373 case 0x17: /* coordinates of subtitle upper left x, y position */
374 ExtractXY(p_sys->i_x_start, p_sys->i_y_start);
376 #ifdef DEBUG_CVDSUB
377 msg_Dbg( p_dec, "start position (%"PRIu16",%"PRIu16")",
378 p_sys->i_x_start, p_sys->i_y_start );
379 #endif
380 break;
382 case 0x1f: /* coordinates of subtitle bottom right x, y position */
384 int lastx;
385 int lasty;
386 ExtractXY(lastx, lasty);
387 p_sys->i_width = lastx - p_sys->i_x_start + 1;
388 p_sys->i_height = lasty - p_sys->i_y_start + 1;
390 #ifdef DEBUG_CVDSUB
391 msg_Dbg( p_dec, "end position (%d,%d), w x h: %"PRIu16"x%"PRIu16,
392 lastx, lasty, p_sys->i_width, p_sys->i_height );
393 #endif
394 break;
397 case 0x24:
398 case 0x25:
399 case 0x26:
400 case 0x27:
402 uint8_t v = p[0] - 0x24;
404 #ifdef DEBUG_CVDSUB
405 /* Primary Palette */
406 msg_Dbg( p_dec, "primary palette %"PRIu8" (y,u,v): "
407 "(0x%02"PRIx8",0x%02"PRIx8",0x%02"PRIx8")",
408 v, p[1], p[2], p[3] );
409 #endif
411 p_sys->p_palette[v][0] = p[1]; /* Y */
412 p_sys->p_palette[v][1] = p[3]; /* Cr / V */
413 p_sys->p_palette[v][2] = p[2]; /* Cb / U */
414 break;
417 case 0x2c:
418 case 0x2d:
419 case 0x2e:
420 case 0x2f:
422 uint8_t v = p[0] - 0x2c;
424 #ifdef DEBUG_CVDSUB
425 msg_Dbg( p_dec,"highlight palette %"PRIu8" (y,u,v): "
426 "(0x%02"PRIx8",0x%02"PRIx8",0x%02"PRIx8")",
427 v, p[1], p[2], p[3] );
428 #endif
430 /* Highlight Palette */
431 p_sys->p_palette_highlight[v][0] = p[1]; /* Y */
432 p_sys->p_palette_highlight[v][1] = p[3]; /* Cr / V */
433 p_sys->p_palette_highlight[v][2] = p[2]; /* Cb / U */
434 break;
437 case 0x37:
438 /* transparency for primary palette */
439 p_sys->p_palette[0][3] = (p[3] & 0x0f) << 4;
440 p_sys->p_palette[1][3] = (p[3] >> 4) << 4;
441 p_sys->p_palette[2][3] = (p[2] & 0x0f) << 4;
442 p_sys->p_palette[3][3] = (p[2] >> 4) << 4;
444 #ifdef DEBUG_CVDSUB
445 msg_Dbg( p_dec, "transparency for primary palette 0..3: "
446 "0x%02"PRIx8" 0x%02"PRIx8" 0x%02"PRIx8" 0x%02"PRIx8,
447 p_sys->p_palette[0][3], p_sys->p_palette[1][3],
448 p_sys->p_palette[2][3], p_sys->p_palette[3][3]);
449 #endif
450 break;
452 case 0x3f:
453 /* transparency for highlight palette */
454 p_sys->p_palette_highlight[0][3] = (p[2] & 0x0f) << 4;
455 p_sys->p_palette_highlight[1][3] = (p[2] >> 4) << 4;
456 p_sys->p_palette_highlight[2][3] = (p[1] & 0x0f) << 4;
457 p_sys->p_palette_highlight[3][3] = (p[1] >> 4) << 4;
459 #ifdef DEBUG_CVDSUB
460 msg_Dbg( p_dec, "transparency for highlight palette 0..3: "
461 "0x%02"PRIx8" 0x%02"PRIx8" 0x%02"PRIx8" 0x%02"PRIx8,
462 p_sys->p_palette_highlight[0][3],
463 p_sys->p_palette_highlight[1][3],
464 p_sys->p_palette_highlight[2][3],
465 p_sys->p_palette_highlight[3][3] );
466 #endif
467 break;
469 case 0x47:
470 /* offset to start of even rows of interlaced image, we correct
471 * to make it relative to i_image_offset (usually 4) */
472 p_sys->first_field_offset =
473 (p[2] << 8) + p[3] - p_sys->i_image_offset;
474 #ifdef DEBUG_CVDSUB
475 msg_Dbg( p_dec, "1st_field_offset %zu",
476 p_sys->first_field_offset );
477 #endif
478 break;
480 case 0x4f:
481 /* offset to start of odd rows of interlaced image, we correct
482 * to make it relative to i_image_offset (usually 4) */
483 p_sys->second_field_offset =
484 (p[2] << 8) + p[3] - p_sys->i_image_offset;
485 #ifdef DEBUG_CVDSUB
486 msg_Dbg( p_dec, "2nd_field_offset %zu",
487 p_sys->second_field_offset);
488 #endif
489 break;
491 default:
492 #ifdef DEBUG_CVDSUB
493 msg_Warn( p_dec, "unknown sequence in control header "
494 "0x%02"PRIx8" 0x%02"PRIx8" 0x%02"PRIx8" 0x%02"PRIx8,
495 p[0], p[1], p[2], p[3]);
496 #endif
501 /*****************************************************************************
502 * DecodePacket: parse and decode an SPU packet
503 *****************************************************************************
504 * This function parses and decodes an SPU packet and, if valid, returns a
505 * subpicture.
506 *****************************************************************************/
507 static subpicture_t *DecodePacket( decoder_t *p_dec, block_t *p_data )
509 decoder_sys_t *p_sys = p_dec->p_sys;
510 subpicture_t *p_spu;
511 subpicture_region_t *p_region;
512 video_format_t fmt;
513 video_palette_t palette;
514 int i;
516 /* Allocate the subpicture internal data. */
517 p_spu = decoder_NewSubpicture( p_dec, NULL );
518 if( !p_spu ) return NULL;
520 p_spu->i_start = p_data->i_pts;
521 p_spu->i_stop = p_data->i_pts + p_sys->i_duration;
522 p_spu->b_ephemer = true;
524 /* Create new SPU region */
525 video_format_Init( &fmt, VLC_CODEC_YUVP );
526 fmt.i_sar_num = 1;
527 fmt.i_sar_den = 1;
528 fmt.i_width = fmt.i_visible_width = p_sys->i_width;
529 fmt.i_height = fmt.i_visible_height = p_sys->i_height;
530 fmt.i_x_offset = fmt.i_y_offset = 0;
531 fmt.p_palette = &palette;
532 fmt.p_palette->i_entries = 4;
533 for( i = 0; i < fmt.p_palette->i_entries; i++ )
535 fmt.p_palette->palette[i][0] = p_sys->p_palette[i][0];
536 fmt.p_palette->palette[i][1] = p_sys->p_palette[i][1];
537 fmt.p_palette->palette[i][2] = p_sys->p_palette[i][2];
538 fmt.p_palette->palette[i][3] = p_sys->p_palette[i][3];
541 p_region = subpicture_region_New( &fmt );
542 if( !p_region )
544 msg_Err( p_dec, "cannot allocate SPU region" );
545 subpicture_Delete( p_spu );
546 return NULL;
549 p_spu->p_region = p_region;
550 p_region->i_x = p_sys->i_x_start;
551 p_region->i_x = p_region->i_x * 3 / 4; /* FIXME: use aspect ratio for x? */
552 p_region->i_y = p_sys->i_y_start;
554 RenderImage( p_dec, p_data, p_region );
556 return p_spu;
559 /*****************************************************************************
560 * ParseImage: parse and render the image part of the subtitle
561 *****************************************************************************
562 This part parses the subtitle graphical data and renders it.
564 Image data comes interlaced and is run-length encoded (RLE). Each
565 field is a four-bit nibbles that is further subdivided in a two-bit
566 repeat count and a two-bit color number - up to three pixels can be
567 described in four bits. What a 0 repeat count means is unknown. It
568 might be used for RLE extension. There is a special case of a 0
569 repeat count though. When the full nibble is zero, the rest of the
570 line is filled with the color value in the next nibble. It is
571 unknown what happens if the color value is greater than three. The
572 rest seems to use a 4-entries palette. It is not impossible that the
573 fill-line complete case above is not as described and the zero repeat
574 count means fill line. The sample code never produces this, so it
575 may be untested.
577 However we'll transform this so that that the RLE is expanded and
578 interlacing will also be removed. On output each pixel entry will by
579 a 4-bit alpha (filling 8 bits), and 8-bit y, u, and v entry.
581 *****************************************************************************/
582 static void RenderImage( decoder_t *p_dec, block_t *p_data,
583 subpicture_region_t *p_region )
585 decoder_sys_t *p_sys = p_dec->p_sys;
586 uint8_t *p_dest = p_region->p_picture->Y_PIXELS;
587 int i_field; /* The subtitles are interlaced */
588 int i_row, i_column; /* scanline row/column number */
589 uint8_t i_color, i_count;
590 bs_t bs;
592 bs_init( &bs, p_data->p_buffer + p_sys->i_image_offset,
593 p_data->i_buffer - p_sys->i_image_offset );
595 for( i_field = 0; i_field < 2; i_field++ )
597 for( i_row = i_field; i_row < p_sys->i_height; i_row += 2 )
599 for( i_column = 0; i_column < p_sys->i_width; i_column++ )
601 uint8_t i_val = bs_read( &bs, 4 );
603 if( i_val == 0 )
605 /* Fill the rest of the line with next color */
606 i_color = bs_read( &bs, 4 );
608 memset( &p_dest[i_row * p_region->p_picture->Y_PITCH +
609 i_column], i_color,
610 p_sys->i_width - i_column );
611 i_column = p_sys->i_width;
612 continue;
614 else
616 /* Normal case: get color and repeat count */
617 i_count = (i_val >> 2);
618 i_color = i_val & 0x3;
620 i_count = __MIN( i_count, p_sys->i_width - i_column );
622 memset( &p_dest[i_row * p_region->p_picture->Y_PITCH +
623 i_column], i_color, i_count );
624 i_column += i_count - 1;
625 continue;
629 bs_align( &bs );