ffmpeg: targetos must now be in lower case.
[vlc.git] / modules / codec / cvdsub.c
blob84b42748f8e64850bb27a53101faa17a7c284051
1 /*****************************************************************************
2 * cvd.c : CVD Subtitle decoder
3 *****************************************************************************
4 * Copyright (C) 2003, 2004 the VideoLAN team
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
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 /*****************************************************************************
28 * Preamble
29 *****************************************************************************/
30 #include <vlc/vlc.h>
31 #include <vlc_vout.h>
32 #include <vlc_codec.h>
34 #include "vlc_bits.h"
36 #define DEBUG_CVDSUB 1
38 /*****************************************************************************
39 * Module descriptor.
40 *****************************************************************************/
41 static int DecoderOpen ( vlc_object_t * );
42 static int PacketizerOpen( vlc_object_t * );
43 static void DecoderClose ( vlc_object_t * );
45 vlc_module_begin();
46 set_description( _("CVD subtitle decoder") );
47 set_capability( "decoder", 50 );
48 set_callbacks( DecoderOpen, DecoderClose );
50 add_submodule();
51 set_description( _("Chaoji VCD subtitle packetizer") );
52 set_capability( "packetizer", 50 );
53 set_callbacks( PacketizerOpen, DecoderClose );
54 vlc_module_end();
56 /*****************************************************************************
57 * Local prototypes
58 *****************************************************************************/
59 static subpicture_t *Decode( decoder_t *, block_t ** );
60 static block_t *Packetize ( decoder_t *, block_t ** );
61 static block_t *Reassemble ( decoder_t *, block_t * );
62 static void ParseMetaInfo ( decoder_t *, block_t * );
63 static void ParseHeader ( decoder_t *, block_t * );
64 static subpicture_t *DecodePacket( decoder_t *, block_t * );
65 static void RenderImage( decoder_t *, block_t *, subpicture_region_t * );
67 #define SUBTITLE_BLOCK_EMPTY 0
68 #define SUBTITLE_BLOCK_PARTIAL 1
69 #define SUBTITLE_BLOCK_COMPLETE 2
71 struct decoder_sys_t
73 int b_packetizer;
75 int i_state; /* data-gathering state for this subtitle */
77 block_t *p_spu; /* Bytes of the packet. */
79 int i_spu_size; /* goal for subtitle_data_pos while gathering,
80 size of used subtitle_data later */
82 uint16_t i_image_offset; /* offset from subtitle_data to compressed
83 image data */
84 int i_image_length; /* size of the compressed image data */
85 int first_field_offset; /* offset of even raster lines */
86 int second_field_offset; /* offset of odd raster lines */
87 int metadata_offset; /* offset to data describing the image */
88 int metadata_length; /* length of metadata */
90 mtime_t i_duration; /* how long to display the image, 0 stands
91 for "until next subtitle" */
93 uint16_t i_x_start, i_y_start; /* position of top leftmost pixel of
94 image when displayed */
95 uint16_t i_width, i_height; /* dimensions in pixels of image */
97 uint8_t p_palette[4][4]; /* Palette of colors used in subtitle */
98 uint8_t p_palette_highlight[4][4];
101 /*****************************************************************************
102 * DecoderOpen: open/initialize the cvdsub decoder.
103 *****************************************************************************/
104 static int DecoderOpen( vlc_object_t *p_this )
106 decoder_t *p_dec = (decoder_t*)p_this;
107 decoder_sys_t *p_sys;
109 if( p_dec->fmt_in.i_codec != VLC_FOURCC( 'c','v','d',' ' ) )
111 return VLC_EGENERIC;
114 p_dec->p_sys = p_sys = malloc( sizeof( decoder_sys_t ) );
116 p_sys->b_packetizer = VLC_FALSE;
118 p_sys->i_state = SUBTITLE_BLOCK_EMPTY;
119 p_sys->p_spu = NULL;
121 es_format_Init( &p_dec->fmt_out, SPU_ES, VLC_FOURCC( 'c','v','d',' ' ) );
123 p_dec->pf_decode_sub = Decode;
124 p_dec->pf_packetize = Packetize;
126 return VLC_SUCCESS;
129 /*****************************************************************************
130 * PacketizerOpen: open/initialize the cvdsub packetizer.
131 *****************************************************************************/
132 static int PacketizerOpen( vlc_object_t *p_this )
134 decoder_t *p_dec = (decoder_t*)p_this;
136 if( DecoderOpen( p_this ) != VLC_SUCCESS ) return VLC_EGENERIC;
138 p_dec->p_sys->b_packetizer = VLC_TRUE;
140 return VLC_SUCCESS;
143 /*****************************************************************************
144 * DecoderClose: closes the cvdsub decoder/packetizer.
145 *****************************************************************************/
146 void DecoderClose( vlc_object_t *p_this )
148 decoder_t *p_dec = (decoder_t*)p_this;
149 decoder_sys_t *p_sys = p_dec->p_sys;
151 if( p_sys->p_spu ) block_ChainRelease( p_sys->p_spu );
152 free( p_sys );
155 /*****************************************************************************
156 * Decode:
157 *****************************************************************************/
158 static subpicture_t *Decode( decoder_t *p_dec, block_t **pp_block )
160 block_t *p_block, *p_spu;
162 if( pp_block == NULL || *pp_block == NULL ) return NULL;
164 p_block = *pp_block;
165 *pp_block = NULL;
167 if( !(p_spu = Reassemble( p_dec, p_block )) ) return NULL;
169 /* Parse and decode */
170 return DecodePacket( p_dec, p_spu );
173 /*****************************************************************************
174 * Packetize:
175 *****************************************************************************/
176 static block_t *Packetize( decoder_t *p_dec, block_t **pp_block )
178 block_t *p_block, *p_spu;
180 if( pp_block == NULL || *pp_block == NULL ) return NULL;
182 p_block = *pp_block;
183 *pp_block = NULL;
185 if( !(p_spu = Reassemble( p_dec, p_block )) ) return NULL;
187 p_spu->i_dts = p_spu->i_pts;
188 p_spu->i_length = 0;
190 return p_spu;
194 /*****************************************************************************
195 Reassemble:
197 Data for single screen subtitle may come in several non-contiguous
198 packets of a stream. This routine is called when the next packet in
199 the stream comes in. The job of this routine is to parse the header,
200 if this is the beginning, and combine the packets into one complete
201 subtitle unit.
203 If everything is complete, we will return a block. Otherwise return
204 NULL.
206 *****************************************************************************/
207 #define SPU_HEADER_LEN 1
209 static block_t *Reassemble( decoder_t *p_dec, block_t *p_block )
211 decoder_sys_t *p_sys = p_dec->p_sys;
212 uint8_t *p_buffer;
214 if( p_block->i_buffer < SPU_HEADER_LEN )
216 msg_Dbg( p_dec, "invalid packet header (size %d < %d)" ,
217 p_block->i_buffer, SPU_HEADER_LEN );
218 block_Release( p_block );
219 return NULL;
222 p_buffer = p_block->p_buffer;
224 /* From the scant data on the format, there is only only way known
225 * to detect the first packet in a subtitle. The first packet
226 * seems to have a valid PTS while later packets for the same
227 * image don't. */
228 if( p_sys->i_state == SUBTITLE_BLOCK_EMPTY && p_block->i_pts == 0 )
230 msg_Warn( p_dec, "first packet expected but no PTS present");
231 return NULL;
234 p_block->p_buffer += SPU_HEADER_LEN;
235 p_block->i_buffer -= SPU_HEADER_LEN;
237 /* First packet in the subtitle block */
238 if( p_sys->i_state == SUBTITLE_BLOCK_EMPTY ) ParseHeader( p_dec, p_block );
240 block_ChainAppend( &p_sys->p_spu, p_block );
241 p_sys->p_spu = block_ChainGather( p_sys->p_spu );
243 if( p_sys->p_spu->i_buffer >= p_sys->i_spu_size )
245 block_t *p_spu = p_sys->p_spu;
247 if( p_spu->i_buffer != p_sys->i_spu_size )
249 msg_Warn( p_dec, "SPU packets size=%d should be %d",
250 p_spu->i_buffer, p_sys->i_spu_size );
253 msg_Dbg( p_dec, "subtitle packet complete, size=%d", p_spu->i_buffer);
255 ParseMetaInfo( p_dec, p_spu );
257 p_sys->i_state = SUBTITLE_BLOCK_EMPTY;
258 p_sys->p_spu = 0;
259 return p_spu;
261 else
263 /* Not last block in subtitle, so wait for another. */
264 p_sys->i_state = SUBTITLE_BLOCK_PARTIAL;
267 return NULL;
271 We do not have information on the subtitle format used on CVD's
272 except the submux sample code and a couple of samples of dubious
273 origin. Thus, this is the result of reading some code whose
274 correctness is not known and some experimentation.
276 CVD subtitles are different in several ways from SVCD OGT subtitles.
277 Image comes first and metadata is at the end. So that the metadata
278 can be found easily, the subtitle packet starts with two bytes
279 (everything is big-endian again) that give the total size of the
280 subtitle data and the offset to the metadata - i.e. size of the
281 image data plus the four bytes at the beginning.
283 Image data comes interlaced is run-length encoded. Each field is a
284 four-bit nibble. Each nibble contains a two-bit repeat count and a
285 two-bit color number so that up to three pixels can be described in
286 four bits. The function of a 0 repeat count is unknown; it might be
287 used for RLE extension. However when the full nibble is zero, the
288 rest of the line is filled with the color value in the next nibble.
289 It is unknown what happens if the color value is greater than three.
290 The rest seems to use a 4-entries palette. It is not impossible
291 that the fill-line complete case above is not as described and the
292 zero repeat count means fill line. The sample code never produces
293 this, so it may be untested.
296 static void ParseHeader( decoder_t *p_dec, block_t *p_block )
298 decoder_sys_t *p_sys = p_dec->p_sys;
299 uint8_t *p = p_block->p_buffer;
301 p_sys->i_spu_size = (p[0] << 8) + p[1] + 4; p += 2;
303 /* FIXME: check data sanity */
304 p_sys->metadata_offset = (p[0] << 8) + p[1]; p +=2;
305 p_sys->metadata_length = p_sys->i_spu_size - p_sys->metadata_offset;
307 p_sys->i_image_offset = 4;
308 p_sys->i_image_length = p_sys->metadata_offset - p_sys->i_image_offset;
310 #ifdef DEBUG_CVDSUB
311 msg_Dbg( p_dec, "total size: %d image size: %d",
312 p_sys->i_spu_size, p_sys->i_image_length );
313 #endif
318 We parse the metadata information here.
320 Although metadata information does not have to come in a fixed field
321 order, every metadata field consists of a tag byte followed by
322 parameters. In all cases known, the size including tag byte is
323 exactly four bytes in length.
326 #define ExtractXY(x, y) x = ((p[1]&0x0f)<<6) + (p[2]>>2); \
327 y = ((p[2]&0x03)<<8) + p[3];
329 static void ParseMetaInfo( decoder_t *p_dec, block_t *p_spu )
331 /* Last packet in subtitle block. */
333 decoder_sys_t *p_sys = p_dec->p_sys;
334 uint8_t *p = p_spu->p_buffer + p_sys->metadata_offset;
335 uint8_t *p_end = p + p_sys->metadata_length;
337 for( ; p < p_end; p += 4 )
339 switch( p[0] )
341 case 0x04: /* subtitle duration in 1/90000ths of a second */
342 p_sys->i_duration = (p[1]<<16) + (p[2]<<8) + p[3];
344 #ifdef DEBUG_CVDSUB
345 msg_Dbg( p_dec, "subtitle display duration %lu secs",
346 (long unsigned int)(p_sys->i_duration / 90000) );
347 #endif
348 p_sys->i_duration *= 100 / 9;
349 break;
351 case 0x0c: /* unknown */
352 #ifdef DEBUG_CVDSUB
353 msg_Dbg( p_dec, "subtitle command unknown 0x%0x 0x%0x 0x%0x 0x%0x",
354 (int)p[0], (int)p[1], (int)p[2], (int)p[3] );
355 #endif
356 break;
358 case 0x17: /* coordinates of subtitle upper left x, y position */
359 ExtractXY(p_sys->i_x_start, p_sys->i_y_start);
361 #ifdef DEBUG_CVDSUB
362 msg_Dbg( p_dec, "start position (%d,%d)",
363 p_sys->i_x_start, p_sys->i_y_start );
364 #endif
365 break;
367 case 0x1f: /* coordinates of subtitle bottom right x, y position */
369 int lastx;
370 int lasty;
371 ExtractXY(lastx, lasty);
372 p_sys->i_width = lastx - p_sys->i_x_start + 1;
373 p_sys->i_height = lasty - p_sys->i_y_start + 1;
375 #ifdef DEBUG_CVDSUB
376 msg_Dbg( p_dec, "end position (%d,%d), w x h: %dx%d",
377 lastx, lasty, p_sys->i_width, p_sys->i_height );
378 #endif
379 break;
382 case 0x24:
383 case 0x25:
384 case 0x26:
385 case 0x27:
387 uint8_t v = p[0] - 0x24;
389 #ifdef DEBUG_CVDSUB
390 /* Primary Palette */
391 msg_Dbg( p_dec, "primary palette %d (y,u,v): (0x%0x,0x%0x,0x%0x)",
392 (int)v, (int)p[1], (int)p[2], (int)p[3] );
393 #endif
395 p_sys->p_palette[v][0] = p[1]; /* Y */
396 p_sys->p_palette[v][1] = p[3]; /* Cr / V */
397 p_sys->p_palette[v][2] = p[2]; /* Cb / U */
398 break;
401 case 0x2c:
402 case 0x2d:
403 case 0x2e:
404 case 0x2f:
406 uint8_t v = p[0] - 0x2c;
408 #ifdef DEBUG_CVDSUB
409 msg_Dbg( p_dec,"highlight palette %d (y,u,v): (0x%0x,0x%0x,0x%0x)",
410 (int)v, (int)p[1], (int)p[2], (int)p[3] );
411 #endif
413 /* Highlight Palette */
414 p_sys->p_palette_highlight[v][0] = p[1]; /* Y */
415 p_sys->p_palette_highlight[v][1] = p[3]; /* Cr / V */
416 p_sys->p_palette_highlight[v][2] = p[2]; /* Cb / U */
417 break;
420 case 0x37:
421 /* transparency for primary palette */
422 p_sys->p_palette[0][3] = (p[3] & 0x0f) << 4;
423 p_sys->p_palette[1][3] = (p[3] >> 4) << 4;
424 p_sys->p_palette[2][3] = (p[2] & 0x0f) << 4;
425 p_sys->p_palette[3][3] = (p[2] >> 4) << 4;
427 #ifdef DEBUG_CVDSUB
428 msg_Dbg( p_dec, "transparency for primary palette 0..3: "
429 "0x%0x 0x%0x 0x%0x 0x%0x",
430 (int)p_sys->p_palette[0][3], (int)p_sys->p_palette[1][3],
431 (int)p_sys->p_palette[2][3], (int)p_sys->p_palette[3][3]);
432 #endif
433 break;
435 case 0x3f:
436 /* transparency for highlight palette */
437 p_sys->p_palette_highlight[0][3] = (p[2] & 0x0f) << 4;
438 p_sys->p_palette_highlight[1][3] = (p[2] >> 4) << 4;
439 p_sys->p_palette_highlight[2][3] = (p[1] & 0x0f) << 4;
440 p_sys->p_palette_highlight[3][3] = (p[1] >> 4) << 4;
442 #ifdef DEBUG_CVDSUB
443 msg_Dbg( p_dec, "transparency for highlight palette 0..3: "
444 "0x%0x 0x%0x 0x%0x 0x%0x",
445 (int)p_sys->p_palette_highlight[0][3],
446 (int)p_sys->p_palette_highlight[1][3],
447 (int)p_sys->p_palette_highlight[2][3],
448 (int)p_sys->p_palette_highlight[3][3] );
449 #endif
450 break;
452 case 0x47:
453 /* offset to start of even rows of interlaced image, we correct
454 * to make it relative to i_image_offset (usually 4) */
455 p_sys->first_field_offset =
456 (p[2] << 8) + p[3] - p_sys->i_image_offset;
457 #ifdef DEBUG_CVDSUB
458 msg_Dbg( p_dec, "1st_field_offset %d", p_sys->first_field_offset );
459 #endif
460 break;
462 case 0x4f:
463 /* offset to start of odd rows of interlaced image, we correct
464 * to make it relative to i_image_offset (usually 4) */
465 p_sys->second_field_offset =
466 (p[2] << 8) + p[3] - p_sys->i_image_offset;
467 #ifdef DEBUG_CVDSUB
468 msg_Dbg( p_dec, "2nd_field_offset %d", p_sys->second_field_offset);
469 #endif
470 break;
472 default:
473 #ifdef DEBUG_CVDSUB
474 msg_Warn( p_dec, "unknown sequence in control header "
475 "0x%0x 0x%0x 0x%0x 0x%0x", p[0], p[1], p[2], p[3]);
476 #endif
481 /*****************************************************************************
482 * DecodePacket: parse and decode an SPU packet
483 *****************************************************************************
484 * This function parses and decodes an SPU packet and, if valid, returns a
485 * subpicture.
486 *****************************************************************************/
487 static subpicture_t *DecodePacket( decoder_t *p_dec, block_t *p_data )
489 decoder_sys_t *p_sys = p_dec->p_sys;
490 subpicture_t *p_spu;
491 subpicture_region_t *p_region;
492 video_format_t fmt;
493 int i;
495 /* Allocate the subpicture internal data. */
496 p_spu = p_dec->pf_spu_buffer_new( p_dec );
497 if( !p_spu ) return NULL;
499 p_spu->b_pausable = VLC_TRUE;
501 p_spu->i_x = p_sys->i_x_start;
502 p_spu->i_x = p_spu->i_x * 3 / 4; /* FIXME: use aspect ratio for x? */
503 p_spu->i_y = p_sys->i_y_start;
504 p_spu->i_start = p_data->i_pts;
505 p_spu->i_stop = p_data->i_pts + p_sys->i_duration;
506 p_spu->b_ephemer = VLC_TRUE;
508 /* Create new SPU region */
509 memset( &fmt, 0, sizeof(video_format_t) );
510 fmt.i_chroma = VLC_FOURCC('Y','U','V','P');
511 fmt.i_aspect = VOUT_ASPECT_FACTOR;
512 fmt.i_width = fmt.i_visible_width = p_sys->i_width;
513 fmt.i_height = fmt.i_visible_height = p_sys->i_height;
514 fmt.i_x_offset = fmt.i_y_offset = 0;
515 p_region = p_spu->pf_create_region( VLC_OBJECT(p_dec), &fmt );
516 if( !p_region )
518 msg_Err( p_dec, "cannot allocate SPU region" );
519 //goto error;
522 p_spu->p_region = p_region;
523 p_region->i_x = p_region->i_y = 0;
525 /* Build palette */
526 fmt.p_palette->i_entries = 4;
527 for( i = 0; i < fmt.p_palette->i_entries; i++ )
529 fmt.p_palette->palette[i][0] = p_sys->p_palette[i][0];
530 fmt.p_palette->palette[i][1] = p_sys->p_palette[i][1];
531 fmt.p_palette->palette[i][2] = p_sys->p_palette[i][2];
532 fmt.p_palette->palette[i][3] = p_sys->p_palette[i][3];
535 RenderImage( p_dec, p_data, p_region );
537 return p_spu;
540 /*****************************************************************************
541 * ParseImage: parse and render the image part of the subtitle
542 *****************************************************************************
543 This part parses the subtitle graphical data and renders it.
545 Image data comes interlaced and is run-length encoded (RLE). Each
546 field is a four-bit nibbles that is further subdivided in a two-bit
547 repeat count and a two-bit color number - up to three pixels can be
548 described in four bits. What a 0 repeat count means is unknown. It
549 might be used for RLE extension. There is a special case of a 0
550 repeat count though. When the full nibble is zero, the rest of the
551 line is filled with the color value in the next nibble. It is
552 unknown what happens if the color value is greater than three. The
553 rest seems to use a 4-entries palette. It is not impossible that the
554 fill-line complete case above is not as described and the zero repeat
555 count means fill line. The sample code never produces this, so it
556 may be untested.
558 However we'll transform this so that that the RLE is expanded and
559 interlacing will also be removed. On output each pixel entry will by
560 a 4-bit alpha (filling 8 bits), and 8-bit y, u, and v entry.
562 *****************************************************************************/
563 static void RenderImage( decoder_t *p_dec, block_t *p_data,
564 subpicture_region_t *p_region )
566 decoder_sys_t *p_sys = p_dec->p_sys;
567 uint8_t *p_dest = p_region->picture.Y_PIXELS;
568 int i_field; /* The subtitles are interlaced */
569 int i_row, i_column; /* scanline row/column number */
570 uint8_t i_color, i_count;
571 bs_t bs;
573 bs_init( &bs, p_data->p_buffer + p_sys->i_image_offset,
574 p_data->i_buffer - p_sys->i_image_offset );
576 for( i_field = 0; i_field < 2; i_field++ )
578 for( i_row = i_field; i_row < p_sys->i_height; i_row += 2 )
580 for( i_column = 0; i_column < p_sys->i_width; i_column++ )
582 uint8_t i_val = bs_read( &bs, 4 );
584 if( i_val == 0 )
586 /* Fill the rest of the line with next color */
587 i_color = bs_read( &bs, 4 );
589 memset( &p_dest[i_row * p_region->picture.Y_PITCH +
590 i_column], i_color,
591 p_sys->i_width - i_column );
592 i_column = p_sys->i_width;
593 continue;
595 else
597 /* Normal case: get color and repeat count */
598 i_count = (i_val >> 2);
599 i_color = i_val & 0x3;
601 i_count = __MIN( i_count, p_sys->i_width - i_column );
603 memset( &p_dest[i_row * p_region->picture.Y_PITCH +
604 i_column], i_color, i_count );
605 i_column += i_count - 1;
606 continue;
610 bs_align( &bs );