1 /*****************************************************************************
2 * cvd.c : CVD Subtitle decoder
3 *****************************************************************************
4 * Copyright (C) 2003, 2004 the VideoLAN team
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
13 * it under the terms of the GNU General Public License as published by
14 * the Free Software Foundation; either version 2 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 General Public License for more details.
22 * You should have received a copy of the GNU General Public License
23 * along with this program; if not, write to the Free Software
24 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
25 *****************************************************************************/
27 /*****************************************************************************
29 *****************************************************************************/
35 #include <vlc_plugin.h>
37 #include <vlc_codec.h>
41 #define DEBUG_CVDSUB 1
43 /*****************************************************************************
45 *****************************************************************************/
46 static int DecoderOpen ( vlc_object_t * );
47 static int PacketizerOpen( vlc_object_t * );
48 static void DecoderClose ( vlc_object_t * );
51 set_description( _("CVD subtitle decoder") );
52 set_capability( "decoder", 50 );
53 set_callbacks( DecoderOpen, DecoderClose );
56 set_description( _("Chaoji VCD subtitle packetizer") );
57 set_capability( "packetizer", 50 );
58 set_callbacks( PacketizerOpen, DecoderClose );
61 /*****************************************************************************
63 *****************************************************************************/
64 static subpicture_t *Decode( decoder_t *, block_t ** );
65 static block_t *Packetize ( decoder_t *, block_t ** );
66 static block_t *Reassemble ( decoder_t *, block_t * );
67 static void ParseMetaInfo ( decoder_t *, block_t * );
68 static void ParseHeader ( decoder_t *, block_t * );
69 static subpicture_t *DecodePacket( decoder_t *, block_t * );
70 static void RenderImage( decoder_t *, block_t *, subpicture_region_t * );
72 #define SUBTITLE_BLOCK_EMPTY 0
73 #define SUBTITLE_BLOCK_PARTIAL 1
74 #define SUBTITLE_BLOCK_COMPLETE 2
80 int i_state; /* data-gathering state for this subtitle */
82 block_t *p_spu; /* Bytes of the packet. */
84 size_t i_spu_size; /* goal for subtitle_data_pos while gathering,
85 size of used subtitle_data later */
87 uint16_t i_image_offset; /* offset from subtitle_data to compressed
89 size_t i_image_length; /* size of the compressed image data */
90 size_t first_field_offset; /* offset of even raster lines */
91 size_t second_field_offset; /* offset of odd raster lines */
92 size_t metadata_offset; /* offset to data describing the image */
93 size_t metadata_length; /* length of metadata */
95 mtime_t i_duration; /* how long to display the image, 0 stands
96 for "until next subtitle" */
98 uint16_t i_x_start, i_y_start; /* position of top leftmost pixel of
99 image when displayed */
100 uint16_t i_width, i_height; /* dimensions in pixels of image */
102 uint8_t p_palette[4][4]; /* Palette of colors used in subtitle */
103 uint8_t p_palette_highlight[4][4];
106 /*****************************************************************************
107 * DecoderOpen: open/initialize the cvdsub decoder.
108 *****************************************************************************/
109 static int DecoderOpen( vlc_object_t *p_this )
111 decoder_t *p_dec = (decoder_t*)p_this;
112 decoder_sys_t *p_sys;
114 if( p_dec->fmt_in.i_codec != VLC_FOURCC( 'c','v','d',' ' ) )
119 p_dec->p_sys = p_sys = malloc( sizeof( decoder_sys_t ) );
121 p_sys->b_packetizer = false;
123 p_sys->i_state = SUBTITLE_BLOCK_EMPTY;
126 es_format_Init( &p_dec->fmt_out, SPU_ES, VLC_FOURCC( 'c','v','d',' ' ) );
128 p_dec->pf_decode_sub = Decode;
129 p_dec->pf_packetize = Packetize;
134 /*****************************************************************************
135 * PacketizerOpen: open/initialize the cvdsub packetizer.
136 *****************************************************************************/
137 static int PacketizerOpen( vlc_object_t *p_this )
139 decoder_t *p_dec = (decoder_t*)p_this;
141 if( DecoderOpen( p_this ) != VLC_SUCCESS ) return VLC_EGENERIC;
143 p_dec->p_sys->b_packetizer = true;
148 /*****************************************************************************
149 * DecoderClose: closes the cvdsub decoder/packetizer.
150 *****************************************************************************/
151 void DecoderClose( vlc_object_t *p_this )
153 decoder_t *p_dec = (decoder_t*)p_this;
154 decoder_sys_t *p_sys = p_dec->p_sys;
156 if( p_sys->p_spu ) block_ChainRelease( p_sys->p_spu );
160 /*****************************************************************************
162 *****************************************************************************/
163 static subpicture_t *Decode( decoder_t *p_dec, block_t **pp_block )
165 block_t *p_block, *p_spu;
167 if( pp_block == NULL || *pp_block == NULL ) return NULL;
172 if( !(p_spu = Reassemble( p_dec, p_block )) ) return NULL;
174 /* Parse and decode */
175 return DecodePacket( p_dec, p_spu );
178 /*****************************************************************************
180 *****************************************************************************/
181 static block_t *Packetize( decoder_t *p_dec, block_t **pp_block )
183 block_t *p_block, *p_spu;
185 if( pp_block == NULL || *pp_block == NULL ) return NULL;
190 if( !(p_spu = Reassemble( p_dec, p_block )) ) return NULL;
192 p_spu->i_dts = p_spu->i_pts;
199 /*****************************************************************************
202 Data for single screen subtitle may come in several non-contiguous
203 packets of a stream. This routine is called when the next packet in
204 the stream comes in. The job of this routine is to parse the header,
205 if this is the beginning, and combine the packets into one complete
208 If everything is complete, we will return a block. Otherwise return
211 *****************************************************************************/
212 #define SPU_HEADER_LEN 1
214 static block_t *Reassemble( decoder_t *p_dec, block_t *p_block )
216 decoder_sys_t *p_sys = p_dec->p_sys;
219 if( p_block->i_buffer < SPU_HEADER_LEN )
221 msg_Dbg( p_dec, "invalid packet header (size %d < %d)" ,
222 p_block->i_buffer, SPU_HEADER_LEN );
223 block_Release( p_block );
227 p_buffer = p_block->p_buffer;
229 /* From the scant data on the format, there is only only way known
230 * to detect the first packet in a subtitle. The first packet
231 * seems to have a valid PTS while later packets for the same
233 if( p_sys->i_state == SUBTITLE_BLOCK_EMPTY && p_block->i_pts == 0 )
235 msg_Warn( p_dec, "first packet expected but no PTS present");
239 p_block->p_buffer += SPU_HEADER_LEN;
240 p_block->i_buffer -= SPU_HEADER_LEN;
242 /* First packet in the subtitle block */
243 if( p_sys->i_state == SUBTITLE_BLOCK_EMPTY ) ParseHeader( p_dec, p_block );
245 block_ChainAppend( &p_sys->p_spu, p_block );
246 p_sys->p_spu = block_ChainGather( p_sys->p_spu );
248 if( p_sys->p_spu->i_buffer >= p_sys->i_spu_size )
250 block_t *p_spu = p_sys->p_spu;
252 if( p_spu->i_buffer != p_sys->i_spu_size )
254 msg_Warn( p_dec, "SPU packets size=%d should be %d",
255 p_spu->i_buffer, p_sys->i_spu_size );
258 msg_Dbg( p_dec, "subtitle packet complete, size=%d", p_spu->i_buffer);
260 ParseMetaInfo( p_dec, p_spu );
262 p_sys->i_state = SUBTITLE_BLOCK_EMPTY;
268 /* Not last block in subtitle, so wait for another. */
269 p_sys->i_state = SUBTITLE_BLOCK_PARTIAL;
276 We do not have information on the subtitle format used on CVD's
277 except the submux sample code and a couple of samples of dubious
278 origin. Thus, this is the result of reading some code whose
279 correctness is not known and some experimentation.
281 CVD subtitles are different in several ways from SVCD OGT subtitles.
282 Image comes first and metadata is at the end. So that the metadata
283 can be found easily, the subtitle packet starts with two bytes
284 (everything is big-endian again) that give the total size of the
285 subtitle data and the offset to the metadata - i.e. size of the
286 image data plus the four bytes at the beginning.
288 Image data comes interlaced is run-length encoded. Each field is a
289 four-bit nibble. Each nibble contains a two-bit repeat count and a
290 two-bit color number so that up to three pixels can be described in
291 four bits. The function of a 0 repeat count is unknown; it might be
292 used for RLE extension. However when the full nibble is zero, the
293 rest of the line is filled with the color value in the next nibble.
294 It is unknown what happens if the color value is greater than three.
295 The rest seems to use a 4-entries palette. It is not impossible
296 that the fill-line complete case above is not as described and the
297 zero repeat count means fill line. The sample code never produces
298 this, so it may be untested.
301 static void ParseHeader( decoder_t *p_dec, block_t *p_block )
303 decoder_sys_t *p_sys = p_dec->p_sys;
304 uint8_t *p = p_block->p_buffer;
306 p_sys->i_spu_size = (p[0] << 8) + p[1] + 4; p += 2;
308 /* FIXME: check data sanity */
309 p_sys->metadata_offset = (p[0] << 8) + p[1]; p +=2;
310 p_sys->metadata_length = p_sys->i_spu_size - p_sys->metadata_offset;
312 p_sys->i_image_offset = 4;
313 p_sys->i_image_length = p_sys->metadata_offset - p_sys->i_image_offset;
316 msg_Dbg( p_dec, "total size: %d image size: %d",
317 p_sys->i_spu_size, p_sys->i_image_length );
322 We parse the metadata information here.
324 Although metadata information does not have to come in a fixed field
325 order, every metadata field consists of a tag byte followed by
326 parameters. In all cases known, the size including tag byte is
327 exactly four bytes in length.
330 #define ExtractXY(x, y) x = ((p[1]&0x0f)<<6) + (p[2]>>2); \
331 y = ((p[2]&0x03)<<8) + p[3];
333 static void ParseMetaInfo( decoder_t *p_dec, block_t *p_spu )
335 /* Last packet in subtitle block. */
337 decoder_sys_t *p_sys = p_dec->p_sys;
338 uint8_t *p = p_spu->p_buffer + p_sys->metadata_offset;
339 uint8_t *p_end = p + p_sys->metadata_length;
341 for( ; p < p_end; p += 4 )
345 case 0x04: /* subtitle duration in 1/90000ths of a second */
346 p_sys->i_duration = (p[1]<<16) + (p[2]<<8) + p[3];
349 msg_Dbg( p_dec, "subtitle display duration %lu secs",
350 (long unsigned int)(p_sys->i_duration / 90000) );
352 p_sys->i_duration *= 100 / 9;
355 case 0x0c: /* unknown */
357 msg_Dbg( p_dec, "subtitle command unknown 0x%0x 0x%0x 0x%0x 0x%0x",
358 (int)p[0], (int)p[1], (int)p[2], (int)p[3] );
362 case 0x17: /* coordinates of subtitle upper left x, y position */
363 ExtractXY(p_sys->i_x_start, p_sys->i_y_start);
366 msg_Dbg( p_dec, "start position (%d,%d)",
367 p_sys->i_x_start, p_sys->i_y_start );
371 case 0x1f: /* coordinates of subtitle bottom right x, y position */
375 ExtractXY(lastx, lasty);
376 p_sys->i_width = lastx - p_sys->i_x_start + 1;
377 p_sys->i_height = lasty - p_sys->i_y_start + 1;
380 msg_Dbg( p_dec, "end position (%d,%d), w x h: %dx%d",
381 lastx, lasty, p_sys->i_width, p_sys->i_height );
391 uint8_t v = p[0] - 0x24;
394 /* Primary Palette */
395 msg_Dbg( p_dec, "primary palette %d (y,u,v): (0x%0x,0x%0x,0x%0x)",
396 (int)v, (int)p[1], (int)p[2], (int)p[3] );
399 p_sys->p_palette[v][0] = p[1]; /* Y */
400 p_sys->p_palette[v][1] = p[3]; /* Cr / V */
401 p_sys->p_palette[v][2] = p[2]; /* Cb / U */
410 uint8_t v = p[0] - 0x2c;
413 msg_Dbg( p_dec,"highlight palette %d (y,u,v): (0x%0x,0x%0x,0x%0x)",
414 (int)v, (int)p[1], (int)p[2], (int)p[3] );
417 /* Highlight Palette */
418 p_sys->p_palette_highlight[v][0] = p[1]; /* Y */
419 p_sys->p_palette_highlight[v][1] = p[3]; /* Cr / V */
420 p_sys->p_palette_highlight[v][2] = p[2]; /* Cb / U */
425 /* transparency for primary palette */
426 p_sys->p_palette[0][3] = (p[3] & 0x0f) << 4;
427 p_sys->p_palette[1][3] = (p[3] >> 4) << 4;
428 p_sys->p_palette[2][3] = (p[2] & 0x0f) << 4;
429 p_sys->p_palette[3][3] = (p[2] >> 4) << 4;
432 msg_Dbg( p_dec, "transparency for primary palette 0..3: "
433 "0x%0x 0x%0x 0x%0x 0x%0x",
434 (int)p_sys->p_palette[0][3], (int)p_sys->p_palette[1][3],
435 (int)p_sys->p_palette[2][3], (int)p_sys->p_palette[3][3]);
440 /* transparency for highlight palette */
441 p_sys->p_palette_highlight[0][3] = (p[2] & 0x0f) << 4;
442 p_sys->p_palette_highlight[1][3] = (p[2] >> 4) << 4;
443 p_sys->p_palette_highlight[2][3] = (p[1] & 0x0f) << 4;
444 p_sys->p_palette_highlight[3][3] = (p[1] >> 4) << 4;
447 msg_Dbg( p_dec, "transparency for highlight palette 0..3: "
448 "0x%0x 0x%0x 0x%0x 0x%0x",
449 (int)p_sys->p_palette_highlight[0][3],
450 (int)p_sys->p_palette_highlight[1][3],
451 (int)p_sys->p_palette_highlight[2][3],
452 (int)p_sys->p_palette_highlight[3][3] );
457 /* offset to start of even rows of interlaced image, we correct
458 * to make it relative to i_image_offset (usually 4) */
459 p_sys->first_field_offset =
460 (p[2] << 8) + p[3] - p_sys->i_image_offset;
462 msg_Dbg( p_dec, "1st_field_offset %d", p_sys->first_field_offset );
467 /* offset to start of odd rows of interlaced image, we correct
468 * to make it relative to i_image_offset (usually 4) */
469 p_sys->second_field_offset =
470 (p[2] << 8) + p[3] - p_sys->i_image_offset;
472 msg_Dbg( p_dec, "2nd_field_offset %d", p_sys->second_field_offset);
478 msg_Warn( p_dec, "unknown sequence in control header "
479 "0x%0x 0x%0x 0x%0x 0x%0x", p[0], p[1], p[2], p[3]);
485 /*****************************************************************************
486 * DecodePacket: parse and decode an SPU packet
487 *****************************************************************************
488 * This function parses and decodes an SPU packet and, if valid, returns a
490 *****************************************************************************/
491 static subpicture_t *DecodePacket( decoder_t *p_dec, block_t *p_data )
493 decoder_sys_t *p_sys = p_dec->p_sys;
495 subpicture_region_t *p_region;
499 /* Allocate the subpicture internal data. */
500 p_spu = p_dec->pf_spu_buffer_new( p_dec );
501 if( !p_spu ) return NULL;
503 p_spu->b_pausable = true;
505 p_spu->i_x = p_sys->i_x_start;
506 p_spu->i_x = p_spu->i_x * 3 / 4; /* FIXME: use aspect ratio for x? */
507 p_spu->i_y = p_sys->i_y_start;
508 p_spu->i_start = p_data->i_pts;
509 p_spu->i_stop = p_data->i_pts + p_sys->i_duration;
510 p_spu->b_ephemer = true;
512 /* Create new SPU region */
513 memset( &fmt, 0, sizeof(video_format_t) );
514 fmt.i_chroma = VLC_FOURCC('Y','U','V','P');
515 fmt.i_aspect = VOUT_ASPECT_FACTOR;
516 fmt.i_width = fmt.i_visible_width = p_sys->i_width;
517 fmt.i_height = fmt.i_visible_height = p_sys->i_height;
518 fmt.i_x_offset = fmt.i_y_offset = 0;
519 p_region = p_spu->pf_create_region( VLC_OBJECT(p_dec), &fmt );
522 msg_Err( p_dec, "cannot allocate SPU region" );
526 p_spu->p_region = p_region;
527 p_region->i_x = p_region->i_y = 0;
530 fmt.p_palette->i_entries = 4;
531 for( i = 0; i < fmt.p_palette->i_entries; i++ )
533 fmt.p_palette->palette[i][0] = p_sys->p_palette[i][0];
534 fmt.p_palette->palette[i][1] = p_sys->p_palette[i][1];
535 fmt.p_palette->palette[i][2] = p_sys->p_palette[i][2];
536 fmt.p_palette->palette[i][3] = p_sys->p_palette[i][3];
539 RenderImage( p_dec, p_data, p_region );
544 /*****************************************************************************
545 * ParseImage: parse and render the image part of the subtitle
546 *****************************************************************************
547 This part parses the subtitle graphical data and renders it.
549 Image data comes interlaced and is run-length encoded (RLE). Each
550 field is a four-bit nibbles that is further subdivided in a two-bit
551 repeat count and a two-bit color number - up to three pixels can be
552 described in four bits. What a 0 repeat count means is unknown. It
553 might be used for RLE extension. There is a special case of a 0
554 repeat count though. When the full nibble is zero, the rest of the
555 line is filled with the color value in the next nibble. It is
556 unknown what happens if the color value is greater than three. The
557 rest seems to use a 4-entries palette. It is not impossible that the
558 fill-line complete case above is not as described and the zero repeat
559 count means fill line. The sample code never produces this, so it
562 However we'll transform this so that that the RLE is expanded and
563 interlacing will also be removed. On output each pixel entry will by
564 a 4-bit alpha (filling 8 bits), and 8-bit y, u, and v entry.
566 *****************************************************************************/
567 static void RenderImage( decoder_t *p_dec, block_t *p_data,
568 subpicture_region_t *p_region )
570 decoder_sys_t *p_sys = p_dec->p_sys;
571 uint8_t *p_dest = p_region->picture.Y_PIXELS;
572 int i_field; /* The subtitles are interlaced */
573 int i_row, i_column; /* scanline row/column number */
574 uint8_t i_color, i_count;
577 bs_init( &bs, p_data->p_buffer + p_sys->i_image_offset,
578 p_data->i_buffer - p_sys->i_image_offset );
580 for( i_field = 0; i_field < 2; i_field++ )
582 for( i_row = i_field; i_row < p_sys->i_height; i_row += 2 )
584 for( i_column = 0; i_column < p_sys->i_width; i_column++ )
586 uint8_t i_val = bs_read( &bs, 4 );
590 /* Fill the rest of the line with next color */
591 i_color = bs_read( &bs, 4 );
593 memset( &p_dest[i_row * p_region->picture.Y_PITCH +
595 p_sys->i_width - i_column );
596 i_column = p_sys->i_width;
601 /* Normal case: get color and repeat count */
602 i_count = (i_val >> 2);
603 i_color = i_val & 0x3;
605 i_count = __MIN( i_count, p_sys->i_width - i_column );
607 memset( &p_dest[i_row * p_region->picture.Y_PITCH +
608 i_column], i_color, i_count );
609 i_column += i_count - 1;