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 *****************************************************************************/
36 #include <vlc_codec.h>
40 #define DEBUG_CVDSUB 1
42 /*****************************************************************************
44 *****************************************************************************/
45 static int DecoderOpen ( vlc_object_t * );
46 static int PacketizerOpen( vlc_object_t * );
47 static void DecoderClose ( vlc_object_t * );
50 set_description( _("CVD subtitle decoder") );
51 set_capability( "decoder", 50 );
52 set_callbacks( DecoderOpen, DecoderClose );
55 set_description( _("Chaoji VCD subtitle packetizer") );
56 set_capability( "packetizer", 50 );
57 set_callbacks( PacketizerOpen, DecoderClose );
60 /*****************************************************************************
62 *****************************************************************************/
63 static subpicture_t *Decode( decoder_t *, block_t ** );
64 static block_t *Packetize ( decoder_t *, block_t ** );
65 static block_t *Reassemble ( decoder_t *, block_t * );
66 static void ParseMetaInfo ( decoder_t *, block_t * );
67 static void ParseHeader ( decoder_t *, block_t * );
68 static subpicture_t *DecodePacket( decoder_t *, block_t * );
69 static void RenderImage( decoder_t *, block_t *, subpicture_region_t * );
71 #define SUBTITLE_BLOCK_EMPTY 0
72 #define SUBTITLE_BLOCK_PARTIAL 1
73 #define SUBTITLE_BLOCK_COMPLETE 2
79 int i_state; /* data-gathering state for this subtitle */
81 block_t *p_spu; /* Bytes of the packet. */
83 size_t i_spu_size; /* goal for subtitle_data_pos while gathering,
84 size of used subtitle_data later */
86 uint16_t i_image_offset; /* offset from subtitle_data to compressed
88 size_t i_image_length; /* size of the compressed image data */
89 size_t first_field_offset; /* offset of even raster lines */
90 size_t second_field_offset; /* offset of odd raster lines */
91 size_t metadata_offset; /* offset to data describing the image */
92 size_t metadata_length; /* length of metadata */
94 mtime_t i_duration; /* how long to display the image, 0 stands
95 for "until next subtitle" */
97 uint16_t i_x_start, i_y_start; /* position of top leftmost pixel of
98 image when displayed */
99 uint16_t i_width, i_height; /* dimensions in pixels of image */
101 uint8_t p_palette[4][4]; /* Palette of colors used in subtitle */
102 uint8_t p_palette_highlight[4][4];
105 /*****************************************************************************
106 * DecoderOpen: open/initialize the cvdsub decoder.
107 *****************************************************************************/
108 static int DecoderOpen( vlc_object_t *p_this )
110 decoder_t *p_dec = (decoder_t*)p_this;
111 decoder_sys_t *p_sys;
113 if( p_dec->fmt_in.i_codec != VLC_FOURCC( 'c','v','d',' ' ) )
118 p_dec->p_sys = p_sys = malloc( sizeof( decoder_sys_t ) );
120 p_sys->b_packetizer = false;
122 p_sys->i_state = SUBTITLE_BLOCK_EMPTY;
125 es_format_Init( &p_dec->fmt_out, SPU_ES, VLC_FOURCC( 'c','v','d',' ' ) );
127 p_dec->pf_decode_sub = Decode;
128 p_dec->pf_packetize = Packetize;
133 /*****************************************************************************
134 * PacketizerOpen: open/initialize the cvdsub packetizer.
135 *****************************************************************************/
136 static int PacketizerOpen( vlc_object_t *p_this )
138 decoder_t *p_dec = (decoder_t*)p_this;
140 if( DecoderOpen( p_this ) != VLC_SUCCESS ) return VLC_EGENERIC;
142 p_dec->p_sys->b_packetizer = true;
147 /*****************************************************************************
148 * DecoderClose: closes the cvdsub decoder/packetizer.
149 *****************************************************************************/
150 void DecoderClose( vlc_object_t *p_this )
152 decoder_t *p_dec = (decoder_t*)p_this;
153 decoder_sys_t *p_sys = p_dec->p_sys;
155 if( p_sys->p_spu ) block_ChainRelease( p_sys->p_spu );
159 /*****************************************************************************
161 *****************************************************************************/
162 static subpicture_t *Decode( decoder_t *p_dec, block_t **pp_block )
164 block_t *p_block, *p_spu;
166 if( pp_block == NULL || *pp_block == NULL ) return NULL;
171 if( !(p_spu = Reassemble( p_dec, p_block )) ) return NULL;
173 /* Parse and decode */
174 return DecodePacket( p_dec, p_spu );
177 /*****************************************************************************
179 *****************************************************************************/
180 static block_t *Packetize( decoder_t *p_dec, block_t **pp_block )
182 block_t *p_block, *p_spu;
184 if( pp_block == NULL || *pp_block == NULL ) return NULL;
189 if( !(p_spu = Reassemble( p_dec, p_block )) ) return NULL;
191 p_spu->i_dts = p_spu->i_pts;
198 /*****************************************************************************
201 Data for single screen subtitle may come in several non-contiguous
202 packets of a stream. This routine is called when the next packet in
203 the stream comes in. The job of this routine is to parse the header,
204 if this is the beginning, and combine the packets into one complete
207 If everything is complete, we will return a block. Otherwise return
210 *****************************************************************************/
211 #define SPU_HEADER_LEN 1
213 static block_t *Reassemble( decoder_t *p_dec, block_t *p_block )
215 decoder_sys_t *p_sys = p_dec->p_sys;
218 if( p_block->i_buffer < SPU_HEADER_LEN )
220 msg_Dbg( p_dec, "invalid packet header (size %d < %d)" ,
221 p_block->i_buffer, SPU_HEADER_LEN );
222 block_Release( p_block );
226 p_buffer = p_block->p_buffer;
228 /* From the scant data on the format, there is only only way known
229 * to detect the first packet in a subtitle. The first packet
230 * seems to have a valid PTS while later packets for the same
232 if( p_sys->i_state == SUBTITLE_BLOCK_EMPTY && p_block->i_pts == 0 )
234 msg_Warn( p_dec, "first packet expected but no PTS present");
238 p_block->p_buffer += SPU_HEADER_LEN;
239 p_block->i_buffer -= SPU_HEADER_LEN;
241 /* First packet in the subtitle block */
242 if( p_sys->i_state == SUBTITLE_BLOCK_EMPTY ) ParseHeader( p_dec, p_block );
244 block_ChainAppend( &p_sys->p_spu, p_block );
245 p_sys->p_spu = block_ChainGather( p_sys->p_spu );
247 if( p_sys->p_spu->i_buffer >= p_sys->i_spu_size )
249 block_t *p_spu = p_sys->p_spu;
251 if( p_spu->i_buffer != p_sys->i_spu_size )
253 msg_Warn( p_dec, "SPU packets size=%d should be %d",
254 p_spu->i_buffer, p_sys->i_spu_size );
257 msg_Dbg( p_dec, "subtitle packet complete, size=%d", p_spu->i_buffer);
259 ParseMetaInfo( p_dec, p_spu );
261 p_sys->i_state = SUBTITLE_BLOCK_EMPTY;
267 /* Not last block in subtitle, so wait for another. */
268 p_sys->i_state = SUBTITLE_BLOCK_PARTIAL;
275 We do not have information on the subtitle format used on CVD's
276 except the submux sample code and a couple of samples of dubious
277 origin. Thus, this is the result of reading some code whose
278 correctness is not known and some experimentation.
280 CVD subtitles are different in several ways from SVCD OGT subtitles.
281 Image comes first and metadata is at the end. So that the metadata
282 can be found easily, the subtitle packet starts with two bytes
283 (everything is big-endian again) that give the total size of the
284 subtitle data and the offset to the metadata - i.e. size of the
285 image data plus the four bytes at the beginning.
287 Image data comes interlaced is run-length encoded. Each field is a
288 four-bit nibble. Each nibble contains a two-bit repeat count and a
289 two-bit color number so that up to three pixels can be described in
290 four bits. The function of a 0 repeat count is unknown; it might be
291 used for RLE extension. However when the full nibble is zero, the
292 rest of the line is filled with the color value in the next nibble.
293 It is unknown what happens if the color value is greater than three.
294 The rest seems to use a 4-entries palette. It is not impossible
295 that the fill-line complete case above is not as described and the
296 zero repeat count means fill line. The sample code never produces
297 this, so it may be untested.
300 static void ParseHeader( decoder_t *p_dec, block_t *p_block )
302 decoder_sys_t *p_sys = p_dec->p_sys;
303 uint8_t *p = p_block->p_buffer;
305 p_sys->i_spu_size = (p[0] << 8) + p[1] + 4; p += 2;
307 /* FIXME: check data sanity */
308 p_sys->metadata_offset = (p[0] << 8) + p[1]; p +=2;
309 p_sys->metadata_length = p_sys->i_spu_size - p_sys->metadata_offset;
311 p_sys->i_image_offset = 4;
312 p_sys->i_image_length = p_sys->metadata_offset - p_sys->i_image_offset;
315 msg_Dbg( p_dec, "total size: %d image size: %d",
316 p_sys->i_spu_size, p_sys->i_image_length );
321 We parse the metadata information here.
323 Although metadata information does not have to come in a fixed field
324 order, every metadata field consists of a tag byte followed by
325 parameters. In all cases known, the size including tag byte is
326 exactly four bytes in length.
329 #define ExtractXY(x, y) x = ((p[1]&0x0f)<<6) + (p[2]>>2); \
330 y = ((p[2]&0x03)<<8) + p[3];
332 static void ParseMetaInfo( decoder_t *p_dec, block_t *p_spu )
334 /* Last packet in subtitle block. */
336 decoder_sys_t *p_sys = p_dec->p_sys;
337 uint8_t *p = p_spu->p_buffer + p_sys->metadata_offset;
338 uint8_t *p_end = p + p_sys->metadata_length;
340 for( ; p < p_end; p += 4 )
344 case 0x04: /* subtitle duration in 1/90000ths of a second */
345 p_sys->i_duration = (p[1]<<16) + (p[2]<<8) + p[3];
348 msg_Dbg( p_dec, "subtitle display duration %lu secs",
349 (long unsigned int)(p_sys->i_duration / 90000) );
351 p_sys->i_duration *= 100 / 9;
354 case 0x0c: /* unknown */
356 msg_Dbg( p_dec, "subtitle command unknown 0x%0x 0x%0x 0x%0x 0x%0x",
357 (int)p[0], (int)p[1], (int)p[2], (int)p[3] );
361 case 0x17: /* coordinates of subtitle upper left x, y position */
362 ExtractXY(p_sys->i_x_start, p_sys->i_y_start);
365 msg_Dbg( p_dec, "start position (%d,%d)",
366 p_sys->i_x_start, p_sys->i_y_start );
370 case 0x1f: /* coordinates of subtitle bottom right x, y position */
374 ExtractXY(lastx, lasty);
375 p_sys->i_width = lastx - p_sys->i_x_start + 1;
376 p_sys->i_height = lasty - p_sys->i_y_start + 1;
379 msg_Dbg( p_dec, "end position (%d,%d), w x h: %dx%d",
380 lastx, lasty, p_sys->i_width, p_sys->i_height );
390 uint8_t v = p[0] - 0x24;
393 /* Primary Palette */
394 msg_Dbg( p_dec, "primary palette %d (y,u,v): (0x%0x,0x%0x,0x%0x)",
395 (int)v, (int)p[1], (int)p[2], (int)p[3] );
398 p_sys->p_palette[v][0] = p[1]; /* Y */
399 p_sys->p_palette[v][1] = p[3]; /* Cr / V */
400 p_sys->p_palette[v][2] = p[2]; /* Cb / U */
409 uint8_t v = p[0] - 0x2c;
412 msg_Dbg( p_dec,"highlight palette %d (y,u,v): (0x%0x,0x%0x,0x%0x)",
413 (int)v, (int)p[1], (int)p[2], (int)p[3] );
416 /* Highlight Palette */
417 p_sys->p_palette_highlight[v][0] = p[1]; /* Y */
418 p_sys->p_palette_highlight[v][1] = p[3]; /* Cr / V */
419 p_sys->p_palette_highlight[v][2] = p[2]; /* Cb / U */
424 /* transparency for primary palette */
425 p_sys->p_palette[0][3] = (p[3] & 0x0f) << 4;
426 p_sys->p_palette[1][3] = (p[3] >> 4) << 4;
427 p_sys->p_palette[2][3] = (p[2] & 0x0f) << 4;
428 p_sys->p_palette[3][3] = (p[2] >> 4) << 4;
431 msg_Dbg( p_dec, "transparency for primary palette 0..3: "
432 "0x%0x 0x%0x 0x%0x 0x%0x",
433 (int)p_sys->p_palette[0][3], (int)p_sys->p_palette[1][3],
434 (int)p_sys->p_palette[2][3], (int)p_sys->p_palette[3][3]);
439 /* transparency for highlight palette */
440 p_sys->p_palette_highlight[0][3] = (p[2] & 0x0f) << 4;
441 p_sys->p_palette_highlight[1][3] = (p[2] >> 4) << 4;
442 p_sys->p_palette_highlight[2][3] = (p[1] & 0x0f) << 4;
443 p_sys->p_palette_highlight[3][3] = (p[1] >> 4) << 4;
446 msg_Dbg( p_dec, "transparency for highlight palette 0..3: "
447 "0x%0x 0x%0x 0x%0x 0x%0x",
448 (int)p_sys->p_palette_highlight[0][3],
449 (int)p_sys->p_palette_highlight[1][3],
450 (int)p_sys->p_palette_highlight[2][3],
451 (int)p_sys->p_palette_highlight[3][3] );
456 /* offset to start of even rows of interlaced image, we correct
457 * to make it relative to i_image_offset (usually 4) */
458 p_sys->first_field_offset =
459 (p[2] << 8) + p[3] - p_sys->i_image_offset;
461 msg_Dbg( p_dec, "1st_field_offset %d", p_sys->first_field_offset );
466 /* offset to start of odd rows of interlaced image, we correct
467 * to make it relative to i_image_offset (usually 4) */
468 p_sys->second_field_offset =
469 (p[2] << 8) + p[3] - p_sys->i_image_offset;
471 msg_Dbg( p_dec, "2nd_field_offset %d", p_sys->second_field_offset);
477 msg_Warn( p_dec, "unknown sequence in control header "
478 "0x%0x 0x%0x 0x%0x 0x%0x", p[0], p[1], p[2], p[3]);
484 /*****************************************************************************
485 * DecodePacket: parse and decode an SPU packet
486 *****************************************************************************
487 * This function parses and decodes an SPU packet and, if valid, returns a
489 *****************************************************************************/
490 static subpicture_t *DecodePacket( decoder_t *p_dec, block_t *p_data )
492 decoder_sys_t *p_sys = p_dec->p_sys;
494 subpicture_region_t *p_region;
498 /* Allocate the subpicture internal data. */
499 p_spu = p_dec->pf_spu_buffer_new( p_dec );
500 if( !p_spu ) return NULL;
502 p_spu->b_pausable = true;
504 p_spu->i_x = p_sys->i_x_start;
505 p_spu->i_x = p_spu->i_x * 3 / 4; /* FIXME: use aspect ratio for x? */
506 p_spu->i_y = p_sys->i_y_start;
507 p_spu->i_start = p_data->i_pts;
508 p_spu->i_stop = p_data->i_pts + p_sys->i_duration;
509 p_spu->b_ephemer = true;
511 /* Create new SPU region */
512 memset( &fmt, 0, sizeof(video_format_t) );
513 fmt.i_chroma = VLC_FOURCC('Y','U','V','P');
514 fmt.i_aspect = VOUT_ASPECT_FACTOR;
515 fmt.i_width = fmt.i_visible_width = p_sys->i_width;
516 fmt.i_height = fmt.i_visible_height = p_sys->i_height;
517 fmt.i_x_offset = fmt.i_y_offset = 0;
518 p_region = p_spu->pf_create_region( VLC_OBJECT(p_dec), &fmt );
521 msg_Err( p_dec, "cannot allocate SPU region" );
525 p_spu->p_region = p_region;
526 p_region->i_x = p_region->i_y = 0;
529 fmt.p_palette->i_entries = 4;
530 for( i = 0; i < fmt.p_palette->i_entries; i++ )
532 fmt.p_palette->palette[i][0] = p_sys->p_palette[i][0];
533 fmt.p_palette->palette[i][1] = p_sys->p_palette[i][1];
534 fmt.p_palette->palette[i][2] = p_sys->p_palette[i][2];
535 fmt.p_palette->palette[i][3] = p_sys->p_palette[i][3];
538 RenderImage( p_dec, p_data, p_region );
543 /*****************************************************************************
544 * ParseImage: parse and render the image part of the subtitle
545 *****************************************************************************
546 This part parses the subtitle graphical data and renders it.
548 Image data comes interlaced and is run-length encoded (RLE). Each
549 field is a four-bit nibbles that is further subdivided in a two-bit
550 repeat count and a two-bit color number - up to three pixels can be
551 described in four bits. What a 0 repeat count means is unknown. It
552 might be used for RLE extension. There is a special case of a 0
553 repeat count though. When the full nibble is zero, the rest of the
554 line is filled with the color value in the next nibble. It is
555 unknown what happens if the color value is greater than three. The
556 rest seems to use a 4-entries palette. It is not impossible that the
557 fill-line complete case above is not as described and the zero repeat
558 count means fill line. The sample code never produces this, so it
561 However we'll transform this so that that the RLE is expanded and
562 interlacing will also be removed. On output each pixel entry will by
563 a 4-bit alpha (filling 8 bits), and 8-bit y, u, and v entry.
565 *****************************************************************************/
566 static void RenderImage( decoder_t *p_dec, block_t *p_data,
567 subpicture_region_t *p_region )
569 decoder_sys_t *p_sys = p_dec->p_sys;
570 uint8_t *p_dest = p_region->picture.Y_PIXELS;
571 int i_field; /* The subtitles are interlaced */
572 int i_row, i_column; /* scanline row/column number */
573 uint8_t i_color, i_count;
576 bs_init( &bs, p_data->p_buffer + p_sys->i_image_offset,
577 p_data->i_buffer - p_sys->i_image_offset );
579 for( i_field = 0; i_field < 2; i_field++ )
581 for( i_row = i_field; i_row < p_sys->i_height; i_row += 2 )
583 for( i_column = 0; i_column < p_sys->i_width; i_column++ )
585 uint8_t i_val = bs_read( &bs, 4 );
589 /* Fill the rest of the line with next color */
590 i_color = bs_read( &bs, 4 );
592 memset( &p_dest[i_row * p_region->picture.Y_PITCH +
594 p_sys->i_width - i_column );
595 i_column = p_sys->i_width;
600 /* Normal case: get color and repeat count */
601 i_count = (i_val >> 2);
602 i_color = i_val & 0x3;
604 i_count = __MIN( i_count, p_sys->i_width - i_column );
606 memset( &p_dest[i_row * p_region->picture.Y_PITCH +
607 i_column], i_color, i_count );
608 i_column += i_count - 1;