* Authors: Gildas Bazin <gbazin@videolan.org>
* Samuel Hocevar <sam@zoy.org>
* Derk-Jan Hartman <hartman at videolan dot org>
+ * Bernie Purcell <bitmap@videolan.org>
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
/*****************************************************************************
* Preamble
*****************************************************************************/
-#include <vlc/vlc.h>
-#include <vlc_vout.h>
-#include <vlc_codec.h>
+#ifdef HAVE_CONFIG_H
+# include "config.h"
+#endif
-#include <vlc_osd.h>
-#include <vlc_filter.h>
-#include <vlc_charset.h>
-#include <errno.h>
-#include <string.h>
-
-typedef struct
-{
- char * psz_stylename; /* The name of the style, no comma's allowed */
- text_style_t font_style;
- int i_align;
- int i_margin_h;
- int i_margin_v;
-} ssa_style_t;
-
-/*****************************************************************************
- * decoder_sys_t : decoder descriptor
- *****************************************************************************/
-struct decoder_sys_t
-{
- vlc_bool_t b_ass; /* The subs are ASS */
- int i_original_height;
- int i_original_width;
- int i_align; /* Subtitles alignment on the vout */
- vlc_iconv_t iconv_handle; /* handle to iconv instance */
- vlc_bool_t b_autodetect_utf8;
-
- ssa_style_t **pp_ssa_styles;
- int i_ssa_styles;
-};
+#include "subsdec.h"
+#include <vlc_plugin.h>
/*****************************************************************************
* Local prototypes
static int OpenDecoder ( vlc_object_t * );
static void CloseDecoder ( vlc_object_t * );
-static subpicture_t *DecodeBlock ( decoder_t *, block_t ** );
-static subpicture_t *ParseText ( decoder_t *, block_t * );
-static void ParseSSAHeader ( decoder_t * );
-static void ParseSSAString ( decoder_t *, char *, subpicture_t * );
-static void ParseColor ( decoder_t *, char *, int *, int * );
-static void StripTags ( char * );
+static subpicture_t *DecodeBlock ( decoder_t *, block_t ** );
+static subpicture_t *ParseText ( decoder_t *, block_t * );
+static char *StripTags ( char * );
+static char *CreateHtmlSubtitle( int *pi_align, char * );
-#define DEFAULT_NAME "Default"
-#define MAX_LINE 8192
/*****************************************************************************
* Module descriptor.
*****************************************************************************/
-static const char *ppsz_encodings[] = { DEFAULT_NAME, "ASCII", "UTF-8", "",
- "ISO-8859-1", "CP1252", "MacRoman", "MacIceland","ISO-8859-15", "",
- "ISO-8859-2", "CP1250", "MacCentralEurope", "MacCroatian", "MacRomania", "",
- "ISO-8859-5", "CP1251", "MacCyrillic", "MacUkraine", "KOI8-R", "KOI8-U", "KOI8-RU", "",
- "ISO-8859-6", "CP1256", "MacArabic", "",
- "ISO-8859-7", "CP1253", "MacGreek", "",
- "ISO-8859-8", "CP1255", "MacHebrew", "",
- "ISO-8859-9", "CP1254", "MacTurkish", "",
- "ISO-8859-13", "CP1257", "",
- "ISO-2022-JP", "ISO-2022-JP-1", "ISO-2022-JP-2", "EUC-JP", "SHIFT_JIS", "",
- "ISO-2022-CN", "ISO-2022-CN-EXT", "EUC-CN", "EUC-TW", "BIG5", "BIG5-HKSCS", "",
- "ISO-2022-KR", "EUC-KR", "",
- "MacThai", "KOI8-T", "",
- "ISO-8859-3", "ISO-8859-4", "ISO-8859-10", "ISO-8859-14", "ISO-8859-16", "",
- "CP850", "CP862", "CP866", "CP874", "CP932", "CP949", "CP950", "CP1133", "CP1258", "",
- "Macintosh", "",
- "UTF-7", "UTF-16", "UTF-16BE", "UTF-16LE", "UTF-32", "UTF-32BE", "UTF-32LE",
- "C99", "JAVA", "UCS-2", "UCS-2BE", "UCS-2LE", "UCS-4", "UCS-4BE", "UCS-4LE", "",
- "HZ", "GBK", "GB18030", "JOHAB", "ARMSCII-8",
- "Georgian-Academy", "Georgian-PS", "TIS-620", "MuleLao-1", "VISCII", "TCVN",
- "HPROMAN8", "NEXTSTEP" };
+static const char *const ppsz_encodings[] = {
+ "",
+ "system",
+ "UTF-8",
+ "UTF-16",
+ "UTF-16BE",
+ "UTF-16LE",
+ "GB18030",
+ "ISO-8859-15",
+ "Windows-1252",
+ "ISO-8859-2",
+ "Windows-1250",
+ "ISO-8859-3",
+ "ISO-8859-10",
+ "Windows-1251",
+ "KOI8-R",
+ "KOI8-U",
+ "ISO-8859-6",
+ "Windows-1256",
+ "ISO-8859-7",
+ "Windows-1253",
+ "ISO-8859-8",
+ "Windows-1255",
+ "ISO-8859-9",
+ "Windows-1254",
+ "ISO-8859-11",
+ "Windows-874",
+ "ISO-8859-13",
+ "Windows-1257",
+ "ISO-8859-14",
+ "ISO-8859-16",
+ "ISO-2022-CN-EXT",
+ "EUC-CN",
+ "ISO-2022-JP-2",
+ "EUC-JP",
+ "Shift_JIS",
+ "CP949",
+ "ISO-2022-KR",
+ "Big5",
+ "ISO-2022-TW",
+ "Big5-HKSCS",
+ "VISCII",
+ "Windows-1258",
+};
+
+static const char *const ppsz_encoding_names[] = {
+ /* xgettext:
+ The character encoding name in parenthesis corresponds to that used for
+ the GetACP translation. "Windows-1252" applies to Western European
+ languages using the Latin alphabet. */
+ N_("Default (Windows-1252)"),
+ N_("System codeset"),
+ N_("Universal (UTF-8)"),
+ N_("Universal (UTF-16)"),
+ N_("Universal (big endian UTF-16)"),
+ N_("Universal (little endian UTF-16)"),
+ N_("Universal, Chinese (GB18030)"),
+
+ /* ISO 8859 and the likes */
+ /* 1 */
+ N_("Western European (Latin-9)"), /* mostly superset of Latin-1 */
+ N_("Western European (Windows-1252)"),
+ /* 2 */
+ N_("Eastern European (Latin-2)"),
+ N_("Eastern European (Windows-1250)"),
+ /* 3 */
+ N_("Esperanto (Latin-3)"),
+ /* 4 */
+ N_("Nordic (Latin-6)"), /* Latin 6 supersedes Latin 4 */
+ /* 5 */
+ N_("Cyrillic (Windows-1251)"), /* ISO 8859-5 is not practically used */
+ N_("Russian (KOI8-R)"),
+ N_("Ukrainian (KOI8-U)"),
+ /* 6 */
+ N_("Arabic (ISO 8859-6)"),
+ N_("Arabic (Windows-1256)"),
+ /* 7 */
+ N_("Greek (ISO 8859-7)"),
+ N_("Greek (Windows-1253)"),
+ /* 8 */
+ N_("Hebrew (ISO 8859-8)"),
+ N_("Hebrew (Windows-1255)"),
+ /* 9 */
+ N_("Turkish (ISO 8859-9)"),
+ N_("Turkish (Windows-1254)"),
+ /* 10 -> 4 */
+ /* 11 */
+ N_("Thai (TIS 620-2533/ISO 8859-11)"),
+ N_("Thai (Windows-874)"),
+ /* 13 */
+ N_("Baltic (Latin-7)"),
+ N_("Baltic (Windows-1257)"),
+ /* 12 -> /dev/null */
+ /* 14 */
+ N_("Celtic (Latin-8)"),
+ /* 15 -> 1 */
+ /* 16 */
+ N_("South-Eastern European (Latin-10)"),
+ /* CJK families */
+ N_("Simplified Chinese (ISO-2022-CN-EXT)"),
+ N_("Simplified Chinese Unix (EUC-CN)"),
+ N_("Japanese (7-bits JIS/ISO-2022-JP-2)"),
+ N_("Japanese Unix (EUC-JP)"),
+ N_("Japanese (Shift JIS)"),
+ N_("Korean (EUC-KR/CP949)"),
+ N_("Korean (ISO-2022-KR)"),
+ N_("Traditional Chinese (Big5)"),
+ N_("Traditional Chinese Unix (EUC-TW)"),
+ N_("Hong-Kong Supplementary (HKSCS)"),
+ /* Other */
+ N_("Vietnamese (VISCII)"),
+ N_("Vietnamese (Windows-1258)"),
+};
/*
SSA supports charset selection.
The following known charsets are used:
254 = PC 437
*/
-static int pi_justification[] = { 0, 1, 2 };
-static const char *ppsz_justification_text[] = {N_("Center"),N_("Left"),N_("Right")};
+static const int pi_justification[] = { 0, 1, 2 };
+static const char *const ppsz_justification_text[] = {
+ N_("Center"),N_("Left"),N_("Right")};
#define ENCODING_TEXT N_("Subtitles text encoding")
#define ENCODING_LONGTEXT N_("Set the encoding used in text subtitles")
"VLC partly implements this, but you can choose to disable all formatting.")
-vlc_module_begin();
- set_shortname( _("Subtitles"));
- set_description( _("Text subtitles decoder") );
- set_capability( "decoder", 50 );
- set_callbacks( OpenDecoder, CloseDecoder );
- set_category( CAT_INPUT );
- set_subcategory( SUBCAT_INPUT_SCODEC );
+vlc_module_begin ()
+ set_shortname( N_("Subtitles"))
+ set_description( N_("Text subtitles decoder") )
+ set_capability( "decoder", 50 )
+ set_callbacks( OpenDecoder, CloseDecoder )
+ set_category( CAT_INPUT )
+ set_subcategory( SUBCAT_INPUT_SCODEC )
add_integer( "subsdec-align", 0, NULL, ALIGN_TEXT, ALIGN_LONGTEXT,
- VLC_FALSE );
- change_integer_list( pi_justification, ppsz_justification_text, 0 );
- add_string( "subsdec-encoding", DEFAULT_NAME, NULL,
- ENCODING_TEXT, ENCODING_LONGTEXT, VLC_FALSE );
- change_string_list( ppsz_encodings, 0, 0 );
- add_bool( "subsdec-autodetect-utf8", VLC_TRUE, NULL,
- AUTODETECT_UTF8_TEXT, AUTODETECT_UTF8_LONGTEXT, VLC_FALSE );
- add_bool( "subsdec-formatted", VLC_TRUE, NULL, FORMAT_TEXT, FORMAT_LONGTEXT,
- VLC_FALSE );
-vlc_module_end();
+ false )
+ change_integer_list( pi_justification, ppsz_justification_text, NULL )
+ add_string( "subsdec-encoding", "", NULL,
+ ENCODING_TEXT, ENCODING_LONGTEXT, false )
+ change_string_list( ppsz_encodings, ppsz_encoding_names, 0 )
+ add_bool( "subsdec-autodetect-utf8", true, NULL,
+ AUTODETECT_UTF8_TEXT, AUTODETECT_UTF8_LONGTEXT, false )
+ add_bool( "subsdec-formatted", true, NULL, FORMAT_TEXT, FORMAT_LONGTEXT,
+ false )
+vlc_module_end ()
/*****************************************************************************
* OpenDecoder: probe the decoder and return score
{
decoder_t *p_dec = (decoder_t*)p_this;
decoder_sys_t *p_sys;
- vlc_value_t val;
- if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','u','b','t') &&
- p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
+ switch( p_dec->fmt_in.i_codec )
{
- return VLC_EGENERIC;
+ case VLC_CODEC_SUBT:
+ case VLC_CODEC_SSA:
+ case VLC_CODEC_ITU_T140:
+ break;
+ default:
+ return VLC_EGENERIC;
}
p_dec->pf_decode_sub = DecodeBlock;
+ p_dec->fmt_out.i_cat = SPU_ES;
+ p_dec->fmt_out.i_codec = 0;
/* Allocate the memory needed to store the decoder's structure */
- if( ( p_dec->p_sys = p_sys =
- (decoder_sys_t *)calloc(1, sizeof(decoder_sys_t)) ) == NULL )
- {
- msg_Err( p_dec, "out of memory" );
+ p_dec->p_sys = p_sys = calloc( 1, sizeof( *p_sys ) );
+ if( p_sys == NULL )
return VLC_ENOMEM;
- }
/* init of p_sys */
p_sys->i_align = 0;
p_sys->iconv_handle = (vlc_iconv_t)-1;
- p_sys->b_autodetect_utf8 = VLC_FALSE;
- p_sys->b_ass = VLC_FALSE;
+ p_sys->b_autodetect_utf8 = false;
+ p_sys->b_ass = false;
p_sys->i_original_height = -1;
p_sys->i_original_width = -1;
- p_sys->pp_ssa_styles = NULL;
- p_sys->i_ssa_styles = 0;
+ TAB_INIT( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
+ TAB_INIT( p_sys->i_images, p_sys->pp_images );
char *psz_charset = NULL;
+
/* First try demux-specified encoding */
+ if( p_dec->fmt_in.i_codec == VLC_CODEC_ITU_T140 )
+ psz_charset = strdup( "UTF-8" ); /* IUT T.140 is always using UTF-8 */
+ else
if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
{
psz_charset = strdup (p_dec->fmt_in.subs.psz_encoding);
msg_Dbg (p_dec, "trying demuxer-specified character encoding: %s",
- p_dec->fmt_in.subs.psz_encoding ?: "not specified");
+ p_dec->fmt_in.subs.psz_encoding ?
+ p_dec->fmt_in.subs.psz_encoding : "not specified");
}
/* Second, try configured encoding */
if (psz_charset == NULL)
{
- psz_charset = var_CreateGetNonEmptyString (p_dec, "subsdec-encoding");
- if ((psz_charset != NULL) && !strcasecmp (psz_charset, DEFAULT_NAME))
+ psz_charset = var_InheritString (p_dec, "subsdec-encoding");
+ msg_Dbg (p_dec, "trying configured character encoding: %s",
+ psz_charset ? psz_charset : "not specified");
+ if (psz_charset != NULL && !strcmp (psz_charset, "system"))
{
free (psz_charset);
- psz_charset = NULL;
+ psz_charset = strdup ("");
+ /* ^ iconv() treats "" as nl_langinfo(CODESET) */
}
-
- msg_Dbg (p_dec, "trying configured character encoding: %s",
- psz_charset ?: "not specified");
}
/* Third, try "local" encoding with optional UTF-8 autodetection */
if (psz_charset == NULL)
{
- psz_charset = strdup (GetFallbackEncoding ());
+ /* xgettext:
+ The Windows ANSI code page most commonly used for this language.
+ VLC uses this as a guess of the subtitle files character set
+ (if UTF-8 and UTF-16 autodetection fails).
+ Western European languages normally use "CP1252", which is a
+ Microsoft-variant of ISO 8859-1. That suits the Latin alphabet.
+ Other scripts use other code pages.
+
+ This MUST be a valid iconv character set. If unsure, please refer
+ the VideoLAN translators mailing list. */
+ const char *acp = vlc_pgettext("GetACP", "CP1252");
+
+ psz_charset = strdup (acp);
msg_Dbg (p_dec, "trying default character encoding: %s",
- psz_charset ?: "not specified");
+ psz_charset ? psz_charset : "not specified");
- if (var_CreateGetBool (p_dec, "subsdec-autodetect-utf8"))
+ if (var_InheritBool (p_dec, "subsdec-autodetect-utf8"))
{
msg_Dbg (p_dec, "using automatic UTF-8 detection");
- p_sys->b_autodetect_utf8 = VLC_TRUE;
+ p_sys->b_autodetect_utf8 = true;
}
}
+ /* Forth, don't do character decoding, i.e. assume UTF-8 */
if (psz_charset == NULL)
{
psz_charset = strdup ("UTF-8");
- msg_Dbg (p_dec, "trying hard-coded character encoding: %s",
- psz_charset ?: "error");
+ msg_Dbg (p_dec, "using UTF-8 character encoding" );
}
- if (psz_charset == NULL)
- {
- free (p_sys);
- return VLC_ENOMEM;
- }
-
- if (strcasecmp (psz_charset, "UTF-8") && strcasecmp (psz_charset, "utf8"))
+ if ((psz_charset != NULL)
+ && strcasecmp (psz_charset, "UTF-8")
+ && strcasecmp (psz_charset, "utf8"))
{
p_sys->iconv_handle = vlc_iconv_open ("UTF-8", psz_charset);
if (p_sys->iconv_handle == (vlc_iconv_t)(-1))
- msg_Err (p_dec, "cannot convert from %s: %s", psz_charset,
- strerror (errno));
+ msg_Err (p_dec, "cannot convert from %s: %m", psz_charset);
}
free (psz_charset);
- var_Create( p_dec, "subsdec-align", VLC_VAR_INTEGER | VLC_VAR_DOINHERIT );
- var_Get( p_dec, "subsdec-align", &val );
- p_sys->i_align = val.i_int;
+ p_sys->i_align = var_InheritInteger( p_dec, "subsdec-align" );
- if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ') && var_CreateGetBool( p_dec, "subsdec-formatted" ) )
+ if( p_dec->fmt_in.i_codec == VLC_CODEC_SSA
+ && var_InheritBool( p_dec, "subsdec-formatted" ) )
{
if( p_dec->fmt_in.i_extra > 0 )
ParseSSAHeader( p_dec );
****************************************************************************/
static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
{
- subpicture_t *p_spu = NULL;
+ subpicture_t *p_spu;
+ block_t *p_block;
- if( !pp_block || *pp_block == NULL ) return NULL;
+ if( !pp_block || *pp_block == NULL )
+ return NULL;
- p_spu = ParseText( p_dec, *pp_block );
+ p_block = *pp_block;
+ if( p_block->i_flags & (BLOCK_FLAG_DISCONTINUITY|BLOCK_FLAG_CORRUPTED) )
+ {
+ block_Release( p_block );
+ return NULL;
+ }
- block_Release( *pp_block );
+ p_spu = ParseText( p_dec, p_block );
+
+ block_Release( p_block );
*pp_block = NULL;
return p_spu;
decoder_sys_t *p_sys = p_dec->p_sys;
if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
- {
vlc_iconv_close( p_sys->iconv_handle );
- }
if( p_sys->pp_ssa_styles )
{
int i;
for( i = 0; i < p_sys->i_ssa_styles; i++ )
{
- if( p_sys->pp_ssa_styles[i]->psz_stylename ) free( p_sys->pp_ssa_styles[i]->psz_stylename );
- p_sys->pp_ssa_styles[i]->psz_stylename = NULL;
- if( p_sys->pp_ssa_styles[i]->font_style.psz_fontname ) free( p_sys->pp_ssa_styles[i]->font_style.psz_fontname );
- p_sys->pp_ssa_styles[i]->font_style.psz_fontname = NULL;
- if( p_sys->pp_ssa_styles[i] ) free( p_sys->pp_ssa_styles[i] ); p_sys->pp_ssa_styles[i] = NULL;
+ if( !p_sys->pp_ssa_styles[i] )
+ continue;
+
+ free( p_sys->pp_ssa_styles[i]->psz_stylename );
+ free( p_sys->pp_ssa_styles[i]->font_style.psz_fontname );
+ free( p_sys->pp_ssa_styles[i] );
+ }
+ TAB_CLEAN( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
+ }
+ if( p_sys->pp_images )
+ {
+ int i;
+ for( i = 0; i < p_sys->i_images; i++ )
+ {
+ if( !p_sys->pp_images[i] )
+ continue;
+
+ if( p_sys->pp_images[i]->p_pic )
+ picture_Release( p_sys->pp_images[i]->p_pic );
+ free( p_sys->pp_images[i]->psz_filename );
+
+ free( p_sys->pp_images[i] );
}
- free( p_sys->pp_ssa_styles ); p_sys->pp_ssa_styles = NULL;
+ TAB_CLEAN( p_sys->i_images, p_sys->pp_images );
}
free( p_sys );
video_format_t fmt;
/* We cannot display a subpicture with no date */
- if( p_block->i_pts == 0 )
+ if( p_block->i_pts <= VLC_TS_INVALID )
{
msg_Warn( p_dec, "subtitle without a date" );
return NULL;
}
/* Should be resiliant against bad subtitles */
- psz_subtitle = strndup( (const char *)p_block->p_buffer,
- p_block->i_buffer );
+ psz_subtitle = malloc( p_block->i_buffer + 1 );
if( psz_subtitle == NULL )
return NULL;
+ memcpy( psz_subtitle, p_block->p_buffer, p_block->i_buffer );
+ psz_subtitle[p_block->i_buffer] = '\0';
if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
{
if (EnsureUTF8( psz_subtitle ) == NULL)
{
- msg_Err( p_dec, _("failed to convert subtitle encoding.\n"
+ msg_Err( p_dec, "failed to convert subtitle encoding.\n"
"Try manually setting a character-encoding "
- "before you open the file.") );
+ "before you open the file." );
}
}
else
{
msg_Dbg( p_dec, "invalid UTF-8 sequence: "
"disabling UTF-8 subtitles autodetection" );
- p_sys->b_autodetect_utf8 = VLC_FALSE;
+ p_sys->b_autodetect_utf8 = false;
}
}
{
size_t inbytes_left = strlen( psz_subtitle );
size_t outbytes_left = 6 * inbytes_left;
- char *psz_new_subtitle = malloc( outbytes_left + 1 );
+ char *psz_new_subtitle = xmalloc( outbytes_left + 1 );
char *psz_convert_buffer_out = psz_new_subtitle;
const char *psz_convert_buffer_in = psz_subtitle;
if( ( ret == (size_t)(-1) ) || inbytes_left )
{
free( psz_new_subtitle );
- msg_Err( p_dec, _("failed to convert subtitle encoding.\n"
+ msg_Err( p_dec, "failed to convert subtitle encoding.\n"
"Try manually setting a character-encoding "
- "before you open the file.") );
+ "before you open the file." );
return NULL;
}
psz_subtitle = realloc( psz_new_subtitle,
psz_convert_buffer_out - psz_new_subtitle );
+ if( !psz_subtitle )
+ psz_subtitle = psz_new_subtitle;
}
}
/* Create the subpicture unit */
- p_spu = p_dec->pf_spu_buffer_new( p_dec );
+ p_spu = decoder_NewSubpicture( p_dec, NULL );
if( !p_spu )
{
msg_Warn( p_dec, "can't get spu buffer" );
- if( psz_subtitle ) free( psz_subtitle );
+ free( psz_subtitle );
return NULL;
}
- p_spu->b_pausable = VLC_TRUE;
-
/* Create a new subpicture region */
memset( &fmt, 0, sizeof(video_format_t) );
- fmt.i_chroma = VLC_FOURCC('T','E','X','T');
- fmt.i_aspect = 0;
+ fmt.i_chroma = VLC_CODEC_TEXT;
fmt.i_width = fmt.i_height = 0;
fmt.i_x_offset = fmt.i_y_offset = 0;
- p_spu->p_region = p_spu->pf_create_region( VLC_OBJECT(p_dec), &fmt );
+ p_spu->p_region = subpicture_region_New( &fmt );
if( !p_spu->p_region )
{
msg_Err( p_dec, "cannot allocate SPU region" );
- if( psz_subtitle ) free( psz_subtitle );
- p_dec->pf_spu_buffer_del( p_dec, p_spu );
+ free( psz_subtitle );
+ decoder_DeleteSubpicture( p_dec, p_spu );
return NULL;
}
/* Decode and format the subpicture unit */
- if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
+ if( p_dec->fmt_in.i_codec != VLC_CODEC_SSA )
{
/* Normal text subs, easy markup */
- p_spu->i_flags = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
- p_spu->i_x = p_sys->i_align ? 20 : 0;
- p_spu->i_y = 10;
+ p_spu->p_region->i_align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
+ p_spu->p_region->i_x = p_sys->i_align ? 20 : 0;
+ p_spu->p_region->i_y = 10;
/* Remove formatting from string */
- StripTags( psz_subtitle );
- p_spu->p_region->psz_text = psz_subtitle;
+ p_spu->p_region->psz_text = StripTags( psz_subtitle );
+ if( var_InheritBool( p_dec, "subsdec-formatted" ) )
+ {
+ p_spu->p_region->psz_html = CreateHtmlSubtitle( &p_spu->p_region->i_align, psz_subtitle );
+ }
+
p_spu->i_start = p_block->i_pts;
p_spu->i_stop = p_block->i_pts + p_block->i_length;
p_spu->b_ephemer = (p_block->i_length == 0);
- p_spu->b_absolute = VLC_FALSE;
+ p_spu->b_absolute = false;
}
else
{
- /* Decode SSA strings */
+ /* Decode SSA/USF strings */
ParseSSAString( p_dec, psz_subtitle, p_spu );
+
p_spu->i_start = p_block->i_pts;
p_spu->i_stop = p_block->i_pts + p_block->i_length;
p_spu->b_ephemer = (p_block->i_length == 0);
- p_spu->b_absolute = VLC_FALSE;
+ p_spu->b_absolute = false;
p_spu->i_original_picture_width = p_sys->i_original_width;
p_spu->i_original_picture_height = p_sys->i_original_height;
- if( psz_subtitle ) free( psz_subtitle );
}
+ free( psz_subtitle );
+
return p_spu;
}
-static void ParseSSAString( decoder_t *p_dec, char *psz_subtitle, subpicture_t *p_spu_in )
+char* GotoNextLine( char *psz_text )
{
- /* We expect MKV formatted SSA:
- * ReadOrder, Layer, Style, CharacterName, MarginL, MarginR,
- * MarginV, Effect, Text */
- decoder_sys_t *p_sys = p_dec->p_sys;
- subpicture_t *p_spu = p_spu_in;
- ssa_style_t *p_style = NULL;
- char *psz_new_subtitle = NULL;
- char *psz_buffer_sub = NULL;
- char *psz_style = NULL;
- char *psz_style_start = NULL;
- char *psz_style_end = NULL;
- int i_text = 0, i_comma = 0, i_strlen = 0, i;
- int i_margin_l = 0, i_margin_r = 0, i_margin_v = 0;
-
- psz_buffer_sub = psz_subtitle;
-
- i_comma = 0;
- while( i_comma < 8 && *psz_buffer_sub != '\0' )
+ char *p_newline = psz_text;
+
+ while( p_newline[0] != '\0' )
{
- if( *psz_buffer_sub == ',' )
+ if( p_newline[0] == '\n' || p_newline[0] == '\r' )
{
- i_comma++;
- if( i_comma == 2 ) psz_style_start = &psz_buffer_sub[1];
- if( i_comma == 3 ) psz_style_end = &psz_buffer_sub[0];
- if( i_comma == 4 ) i_margin_l = (int)strtol( psz_buffer_sub+1, NULL, 10 );
- if( i_comma == 5 ) i_margin_r = (int)strtol( psz_buffer_sub+1, NULL, 10 );
- if( i_comma == 6 ) i_margin_v = (int)strtol( psz_buffer_sub+1, NULL, 10 );
+ p_newline++;
+ while( p_newline[0] == '\n' || p_newline[0] == '\r' )
+ p_newline++;
+ break;
}
- psz_buffer_sub++;
+ else p_newline++;
}
+ return p_newline;
+}
- if( *psz_buffer_sub == '\0' && i_comma == 8 )
- {
- msg_Dbg( p_dec, "couldn't find all fields in this SSA line" );
- return;
- }
+/* Function now handles tags with attribute values, and tries
+ * to deal with &' commands too. It no longer modifies the string
+ * in place, so that the original text can be reused
+ */
+static char *StripTags( char *psz_subtitle )
+{
+ char *psz_text_start;
+ char *psz_text;
+
+ psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
+ if( !psz_text_start )
+ return NULL;
- psz_new_subtitle = malloc( strlen( psz_buffer_sub ) + 1);
- i_text = 0;
- while( psz_buffer_sub[0] != '\0' )
+ while( *psz_subtitle )
{
- if( psz_buffer_sub[0] == '\\' && psz_buffer_sub[1] == 'n' )
+ if( *psz_subtitle == '<' )
{
- psz_new_subtitle[i_text] = ' ';
- i_text++;
- psz_buffer_sub += 2;
- }
- else if( psz_buffer_sub[0] == '\\' && psz_buffer_sub[1] == 'N' )
- {
- psz_new_subtitle[i_text] = '\n';
- i_text++;
- psz_buffer_sub += 2;
+ if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
+ *psz_text++ = '\n';
+
+ psz_subtitle += strcspn( psz_subtitle, ">" );
}
- else if( psz_buffer_sub[0] == '{' &&
- psz_buffer_sub[1] == '\\' )
+ else if( *psz_subtitle == '&' )
{
- /* SSA control code */
- while( psz_buffer_sub[0] != '\0' &&
- psz_buffer_sub[0] != '}' )
+ if( !strncasecmp( psz_subtitle, "<", 4 ))
{
- psz_buffer_sub++;
+ *psz_text++ = '<';
+ psz_subtitle += strcspn( psz_subtitle, ";" );
+ }
+ else if( !strncasecmp( psz_subtitle, ">", 4 ))
+ {
+ *psz_text++ = '>';
+ psz_subtitle += strcspn( psz_subtitle, ";" );
+ }
+ else if( !strncasecmp( psz_subtitle, "&", 5 ))
+ {
+ *psz_text++ = '&';
+ psz_subtitle += strcspn( psz_subtitle, ";" );
+ }
+ else if( !strncasecmp( psz_subtitle, """, 6 ))
+ {
+ *psz_text++ = '\"';
+ psz_subtitle += strcspn( psz_subtitle, ";" );
+ }
+ else
+ {
+ /* Assume it is just a normal ampersand */
+ *psz_text++ = '&';
}
- psz_buffer_sub++;
}
else
{
- psz_new_subtitle[i_text] = psz_buffer_sub[0];
- i_text++;
- psz_buffer_sub++;
+ *psz_text++ = *psz_subtitle;
}
- }
- psz_new_subtitle[i_text] = '\0';
-
- i_strlen = __MAX( psz_style_end - psz_style_start, 0);
- psz_style = (char *)malloc( i_strlen + 1);
- psz_style = memcpy( psz_style, psz_style_start, i_strlen );
- psz_style[i_strlen] = '\0';
- for( i = 0; i < p_sys->i_ssa_styles; i++ )
- {
- if( !strcmp( p_sys->pp_ssa_styles[i]->psz_stylename, psz_style ) )
- p_style = p_sys->pp_ssa_styles[i];
+ psz_subtitle++;
}
- if( psz_style ) free( psz_style );
+ *psz_text = '\0';
+ char *psz = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
+ if( psz ) psz_text_start = psz;
- p_spu->p_region->psz_text = psz_new_subtitle;
- if( p_style == NULL )
- {
- p_spu->i_flags = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
- p_spu->i_x = p_sys->i_align ? 20 : 0;
- p_spu->i_y = 10;
- }
- else
- {
- msg_Dbg( p_dec, "style is: %s", p_style->psz_stylename);
- p_spu->p_region->p_style = &p_style->font_style;
- p_spu->i_flags = p_style->i_align;
- if( p_style->i_align & SUBPICTURE_ALIGN_LEFT )
- {
- p_spu->i_x = (i_margin_l) ? i_margin_l : p_style->i_margin_h;
- }
- else if( p_style->i_align & SUBPICTURE_ALIGN_RIGHT )
- {
- p_spu->i_x = (i_margin_r) ? i_margin_r : p_style->i_margin_h;
- }
- p_spu->i_y = (i_margin_v) ? i_margin_v : p_style->i_margin_v;
- }
+ return psz_text_start;
}
-static char* GotoNextLine( char *psz_text )
+/* Try to respect any style tags present in the subtitle string. The main
+ * problem here is a lack of adequate specs for the subtitle formats.
+ * SSA/ASS and USF are both detail spec'ed -- but they are handled elsewhere.
+ * SAMI has a detailed spec, but extensive rework is needed in the demux
+ * code to prevent all this style information being excised, as it presently
+ * does.
+ * That leaves the others - none of which were (I guess) originally intended
+ * to be carrying style information. Over time people have used them that way.
+ * In the absence of specifications from which to work, the tags supported
+ * have been restricted to the simple set permitted by the USF DTD, ie. :
+ * Basic: <br>, <i>, <b>, <u>, <s>
+ * Extended: <font>
+ * Attributes: face
+ * family
+ * size
+ * color
+ * outline-color
+ * shadow-color
+ * outline-level
+ * shadow-level
+ * back-color
+ * alpha
+ * There is also the further restriction that the subtitle be well-formed
+ * as an XML entity, ie. the HTML sentence:
+ * <b><i>Bold and Italics</b></i>
+ * doesn't qualify because the tags aren't nested one inside the other.
+ * <text> tags are automatically added to the output to ensure
+ * well-formedness.
+ * If the text doesn't qualify for any reason, a NULL string is
+ * returned, and the rendering engine will fall back to the
+ * plain text version of the subtitle.
+ */
+static void HtmlNPut( char **ppsz_html, const char *psz_text, int i_max )
{
- char *p_newline = psz_text;
+ const int i_len = strlen(psz_text);
- while( p_newline[0] != '\0' )
- {
- if( p_newline[0] == '\n' || p_newline[0] == '\r' )
- {
- p_newline++;
- while( p_newline[0] == '\n' || p_newline[0] == '\r' )
- p_newline++;
- break;
- }
- else p_newline++;
- }
- return p_newline;
+ strncpy( *ppsz_html, psz_text, i_max );
+ *ppsz_html += __MIN(i_max,i_len);
}
-/*****************************************************************************
- * ParseColor: SSA stores color in BBGGRR, in ASS it uses AABBGGRR
- * The string value in the string can be a pure integer, or hexadecimal &HBBGGRR
- *****************************************************************************/
-static void ParseColor( decoder_t *p_dec, char *psz_color, int *pi_color, int *pi_alpha )
+static void HtmlPut( char **ppsz_html, const char *psz_text )
{
- int i_color = 0;
- if( !strncasecmp( psz_color, "&H", 2 ) )
- {
- /* textual HEX representation */
- i_color = (int) strtol( psz_color+2, NULL, 16 );
- }
- else i_color = (int) strtol( psz_color, NULL, 0 );
-
- *pi_color = 0;
- *pi_color |= ( ( i_color & 0x000000FF ) << 16 ); /* Red */
- *pi_color |= ( ( i_color & 0x0000FF00 ) ); /* Green */
- *pi_color |= ( ( i_color & 0x00FF0000 ) >> 16 ); /* Blue */
-
- if( pi_alpha != NULL )
- *pi_alpha = ( i_color & 0xFF000000 ) >> 24;
+ strcpy( *ppsz_html, psz_text );
+ *ppsz_html += strlen(psz_text);
}
-
-/*****************************************************************************
- * ParseSSAHeader: Retrieve global formatting information etc
- *****************************************************************************/
-static void ParseSSAHeader( decoder_t *p_dec )
+static void HtmlCopy( char **ppsz_html, char **ppsz_subtitle, const char *psz_text )
{
- decoder_sys_t *p_sys = p_dec->p_sys;
- char *psz_parser = NULL;
- char *psz_header = malloc( p_dec->fmt_in.i_extra+1 );
- int i_section_type = 1;
+ HtmlPut( ppsz_html, psz_text );
+ *ppsz_subtitle += strlen(psz_text);
+}
- memcpy( psz_header, p_dec->fmt_in.p_extra, p_dec->fmt_in.i_extra );
- psz_header[ p_dec->fmt_in.i_extra] = '\0';
+static char *CreateHtmlSubtitle( int *pi_align, char *psz_subtitle )
+{
+ /* */
+ char *psz_tag = malloc( ( strlen( psz_subtitle ) / 3 ) + 1 );
+ if( !psz_tag )
+ return NULL;
+ psz_tag[ 0 ] = '\0';
+
+ /* */
+ //Oo + 100 ???
+ size_t i_buf_size = strlen( psz_subtitle ) + 100;
+ char *psz_html_start = malloc( i_buf_size );
+ char *psz_html = psz_html_start;
+ if( psz_html_start == NULL )
+ {
+ free( psz_tag );
+ return NULL;
+ }
+ psz_html[0] = '\0';
- /* Handle [Script Info] section */
- psz_parser = strcasestr( psz_header, "[Script Info]" );
- if( psz_parser == NULL ) goto eof;
+ bool b_has_align = false;
- psz_parser = GotoNextLine( psz_parser );
+ HtmlPut( &psz_html, "<text>" );
- while( psz_parser[0] != '\0' )
+ /* */
+ while( *psz_subtitle )
{
- int temp;
- char buffer_text[MAX_LINE + 1];
-
- if( psz_parser[0] == '!' || psz_parser[0] == ';' ) /* comment */;
- else if( sscanf( psz_parser, "PlayResX: %d", &temp ) == 1 )
- p_sys->i_original_width = ( temp > 0 ) ? temp : -1;
- else if( sscanf( psz_parser, "PlayResY: %d", &temp ) == 1 )
- p_sys->i_original_height = ( temp > 0 ) ? temp : -1;
- else if( sscanf( psz_parser, "Script Type: %8192s", buffer_text ) == 1 )
+ if( *psz_subtitle == '\n' )
{
- if( !strcasecmp( buffer_text, "V4.00+" ) ) p_sys->b_ass = VLC_TRUE;
+ HtmlPut( &psz_html, "<br/>" );
+ psz_subtitle++;
}
- else if( !strncasecmp( psz_parser, "[V4 Styles]", 11 ) )
- i_section_type = 1;
- else if( !strncasecmp( psz_parser, "[V4+ Styles]", 12) )
+ else if( *psz_subtitle == '<' )
{
- i_section_type = 2;
- p_sys->b_ass = VLC_TRUE;
- }
- else if( !strncasecmp( psz_parser, "[Events]", 8 ) )
- i_section_type = 4;
- else if( !strncasecmp( psz_parser, "Style:", 6 ) )
- {
- int i_font_size, i_bold, i_italic, i_border, i_outline, i_shadow, i_underline,
- i_strikeout, i_scale_x, i_scale_y, i_spacing, i_align, i_margin_l, i_margin_r, i_margin_v;
+ if( !strncasecmp( psz_subtitle, "<br/>", 5 ))
+ {
+ HtmlCopy( &psz_html, &psz_subtitle, "<br/>" );
+ }
+ else if( !strncasecmp( psz_subtitle, "<b>", 3 ) )
+ {
+ HtmlCopy( &psz_html, &psz_subtitle, "<b>" );
+ strcat( psz_tag, "b" );
+ }
+ else if( !strncasecmp( psz_subtitle, "<i>", 3 ) )
+ {
+ HtmlCopy( &psz_html, &psz_subtitle, "<i>" );
+ strcat( psz_tag, "i" );
+ }
+ else if( !strncasecmp( psz_subtitle, "<u>", 3 ) )
+ {
+ HtmlCopy( &psz_html, &psz_subtitle, "<u>" );
+ strcat( psz_tag, "u" );
+ }
+ else if( !strncasecmp( psz_subtitle, "<s>", 3 ) )
+ {
+ HtmlCopy( &psz_html, &psz_subtitle, "<s>" );
+ strcat( psz_tag, "s" );
+ }
+ else if( !strncasecmp( psz_subtitle, "<font ", 6 ))
+ {
+ const char *psz_attribs[] = { "face=", "family=", "size=",
+ "color=", "outline-color=", "shadow-color=",
+ "outline-level=", "shadow-level=", "back-color=",
+ "alpha=", NULL };
- char psz_temp_stylename[MAX_LINE+1];
- char psz_temp_fontname[MAX_LINE+1];
- char psz_temp_color1[MAX_LINE+1];
- char psz_temp_color2[MAX_LINE+1];
- char psz_temp_color3[MAX_LINE+1];
- char psz_temp_color4[MAX_LINE+1];
+ HtmlCopy( &psz_html, &psz_subtitle, "<font " );
+ strcat( psz_tag, "f" );
- if( i_section_type == 1 ) /* V4 */
- {
- if( sscanf( psz_parser, "Style: %8192[^,],%8192[^,],%d,%8192[^,],%8192[^,],%8192[^,],%8192[^,],%d,%d,%d,%d,%d,%d,%d,%d,%d%*[^\r\n]",
- psz_temp_stylename, psz_temp_fontname, &i_font_size,
- psz_temp_color1, psz_temp_color2, psz_temp_color3, psz_temp_color4, &i_bold, &i_italic,
- &i_border, &i_outline, &i_shadow, &i_align, &i_margin_l, &i_margin_r, &i_margin_v ) == 16 )
+ while( *psz_subtitle != '>' )
{
- ssa_style_t *p_style = malloc( sizeof(ssa_style_t) );
-
- p_style->psz_stylename = strdup( psz_temp_stylename );
- p_style->font_style.psz_fontname = strdup( psz_temp_fontname );
- p_style->font_style.i_font_size = i_font_size;
-
- ParseColor( p_dec, psz_temp_color1, &p_style->font_style.i_font_color, NULL );
- ParseColor( p_dec, psz_temp_color4, &p_style->font_style.i_shadow_color, NULL );
- p_style->font_style.i_outline_color = p_style->font_style.i_shadow_color;
- p_style->font_style.i_font_alpha = p_style->font_style.i_outline_alpha = p_style->font_style.i_shadow_alpha = 0x00;
- p_style->font_style.i_style_flags = 0;
- if( i_bold ) p_style->font_style.i_style_flags |= STYLE_BOLD;
- if( i_italic ) p_style->font_style.i_style_flags |= STYLE_ITALIC;
-
- if( i_border == 1 ) p_style->font_style.i_style_flags |= (STYLE_ITALIC | STYLE_OUTLINE);
- else if( i_border == 3 )
+ int k;
+
+ for( k=0; psz_attribs[ k ]; k++ )
{
- p_style->font_style.i_style_flags |= STYLE_BACKGROUND;
- p_style->font_style.i_background_color = p_style->font_style.i_shadow_color;
- p_style->font_style.i_background_alpha = p_style->font_style.i_shadow_alpha;
+ int i_len = strlen( psz_attribs[ k ] );
+
+ if( !strncasecmp( psz_subtitle, psz_attribs[k], i_len ) )
+ {
+ /* */
+ HtmlPut( &psz_html, psz_attribs[k] );
+ psz_subtitle += i_len;
+
+ /* */
+ if( *psz_subtitle == '"' )
+ {
+ psz_subtitle++;
+ i_len = strcspn( psz_subtitle, "\"" );
+ }
+ else
+ {
+ i_len = strcspn( psz_subtitle, " \t>" );
+ }
+ HtmlPut( &psz_html, "\"" );
+ if( !strcmp( psz_attribs[ k ], "color=" ) && *psz_subtitle >= '0' && *psz_subtitle <= '9' )
+ HtmlPut( &psz_html, "#" );
+ HtmlNPut( &psz_html, psz_subtitle, i_len );
+ HtmlPut( &psz_html, "\"" );
+
+ psz_subtitle += i_len;
+ if( *psz_subtitle == '\"' )
+ psz_subtitle++;
+ break;
+ }
}
- p_style->font_style.i_shadow_width = i_shadow;
- p_style->font_style.i_outline_width = i_outline;
-
- p_style->i_align = 0;
- if( i_align == 1 || i_align == 5 || i_align == 9 ) p_style->i_align |= SUBPICTURE_ALIGN_LEFT;
- if( i_align == 3 || i_align == 7 || i_align == 11 ) p_style->i_align |= SUBPICTURE_ALIGN_RIGHT;
- if( i_align < 4 ) p_style->i_align |= SUBPICTURE_ALIGN_BOTTOM;
- else if( i_align < 8 ) p_style->i_align |= SUBPICTURE_ALIGN_TOP;
+ if( psz_attribs[ k ] == NULL )
+ {
+ /* Jump over unrecognised tag */
+ int i_len = strcspn( psz_subtitle, "\"" );
+ if( psz_subtitle[i_len] == '\"' )
+ {
+ i_len += 1 + strcspn( &psz_subtitle[i_len + 1], "\"" );
+ if( psz_subtitle[i_len] == '\"' )
+ i_len++;
+ }
+ psz_subtitle += i_len;
+ }
+ while (*psz_subtitle == ' ')
+ *psz_html++ = *psz_subtitle++;
+ }
+ *psz_html++ = *psz_subtitle++;
+ }
+ else if( !strncmp( psz_subtitle, "</", 2 ))
+ {
+ bool b_match = false;
+ bool b_ignore = false;
+ int i_len = strlen( psz_tag ) - 1;
+ char *psz_lastTag = NULL;
- p_style->i_margin_h = ( p_style->i_align & SUBPICTURE_ALIGN_RIGHT ) ? i_margin_r : i_margin_l;
- p_style->i_margin_v = i_margin_v;
+ if( i_len >= 0 )
+ {
+ psz_lastTag = psz_tag + i_len;
+ i_len = 0;
- TAB_APPEND( p_sys->i_ssa_styles, p_sys->pp_ssa_styles, p_style );
+ switch( *psz_lastTag )
+ {
+ case 'b':
+ b_match = !strncasecmp( psz_subtitle, "</b>", 4 );
+ i_len = 4;
+ break;
+ case 'i':
+ b_match = !strncasecmp( psz_subtitle, "</i>", 4 );
+ i_len = 4;
+ break;
+ case 'u':
+ b_match = !strncasecmp( psz_subtitle, "</u>", 4 );
+ i_len = 4;
+ break;
+ case 's':
+ b_match = !strncasecmp( psz_subtitle, "</s>", 4 );
+ i_len = 4;
+ break;
+ case 'f':
+ b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
+ i_len = 7;
+ break;
+ case 'I':
+ i_len = strcspn( psz_subtitle, ">" );
+ b_match = psz_subtitle[i_len] == '>';
+ b_ignore = true;
+ if( b_match )
+ i_len++;
+ break;
+ }
+ }
+ if( !b_match )
+ {
+ /* Not well formed -- kill everything */
+ free( psz_html_start );
+ psz_html_start = NULL;
+ break;
}
- else msg_Warn( p_dec, "SSA v4 styleline parsing failed" );
+ *psz_lastTag = '\0';
+ if( !b_ignore )
+ HtmlNPut( &psz_html, psz_subtitle, i_len );
+
+ psz_subtitle += i_len;
+ }
+ else if( ( psz_subtitle[1] < 'a' || psz_subtitle[1] > 'z' ) &&
+ ( psz_subtitle[1] < 'A' || psz_subtitle[1] > 'Z' ) )
+ {
+ /* We have a single < */
+ HtmlPut( &psz_html, "<" );
+ psz_subtitle++;
}
- else if( i_section_type == 2 ) /* V4+ */
+ else
{
- /* Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour,
- Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline,
- Shadow, Alignment, MarginL, MarginR, MarginV, Encoding
- */
- if( sscanf( psz_parser, "Style: %8192[^,],%8192[^,],%d,%8192[^,],%8192[^,],%8192[^,],%8192[^,],%d,%d,%d,%d,%d,%d,%d,%*f,%d,%d,%d,%d,%d,%d,%d%*[^\r\n]",
- psz_temp_stylename, psz_temp_fontname, &i_font_size,
- psz_temp_color1, psz_temp_color2, psz_temp_color3, psz_temp_color4, &i_bold, &i_italic,
- &i_underline, &i_strikeout, &i_scale_x, &i_scale_y, &i_spacing, &i_border, &i_outline,
- &i_shadow, &i_align, &i_margin_l, &i_margin_r, &i_margin_v ) == 21 )
+ /* We have an unknown tag or a single < */
+
+ /* Search for the next tag or end of tag or end of string */
+ char *psz_stop = psz_subtitle + 1 + strcspn( &psz_subtitle[1], "<>" );
+ char *psz_closing = strstr( psz_subtitle, "/>" );
+
+ if( psz_closing && psz_closing < psz_stop )
{
- ssa_style_t *p_style = malloc( sizeof(ssa_style_t) );
-
- p_style->psz_stylename = strdup( psz_temp_stylename );
- p_style->font_style.psz_fontname = strdup( psz_temp_fontname );
- p_style->font_style.i_font_size = i_font_size;
- msg_Dbg( p_dec, psz_temp_color1 );
- ParseColor( p_dec, psz_temp_color1, &p_style->font_style.i_font_color, &p_style->font_style.i_font_alpha );
- ParseColor( p_dec, psz_temp_color3, &p_style->font_style.i_outline_color, &p_style->font_style.i_outline_alpha );
- ParseColor( p_dec, psz_temp_color4, &p_style->font_style.i_shadow_color, &p_style->font_style.i_shadow_alpha );
-
- p_style->font_style.i_style_flags = 0;
- if( i_bold ) p_style->font_style.i_style_flags |= STYLE_BOLD;
- if( i_italic ) p_style->font_style.i_style_flags |= STYLE_ITALIC;
- if( i_underline ) p_style->font_style.i_style_flags |= STYLE_UNDERLINE;
- if( i_strikeout ) p_style->font_style.i_style_flags |= STYLE_STRIKEOUT;
- if( i_border == 1 ) p_style->font_style.i_style_flags |= (STYLE_ITALIC | STYLE_OUTLINE);
- else if( i_border == 3 )
+ /* We have a self closed tag, remove it */
+ psz_subtitle = &psz_closing[2];
+ }
+ else if( *psz_stop == '>' )
+ {
+ char psz_match[256];
+
+ snprintf( psz_match, sizeof(psz_match), "</%s", &psz_subtitle[1] );
+ psz_match[strcspn( psz_match, " \t>" )] = '\0';
+
+ if( strstr( psz_subtitle, psz_match ) )
+ {
+ /* We have the closing tag, ignore it TODO */
+ psz_subtitle = &psz_stop[1];
+ strcat( psz_tag, "I" );
+ }
+ else
{
- p_style->font_style.i_style_flags |= STYLE_BACKGROUND;
- p_style->font_style.i_background_color = p_style->font_style.i_shadow_color;
- p_style->font_style.i_background_alpha = p_style->font_style.i_shadow_alpha;
+ int i_len = psz_stop + 1 - psz_subtitle;
+
+ /* Copy the whole data */
+ for( ; i_len > 0; i_len--, psz_subtitle++ )
+ {
+ if( *psz_subtitle == '<' )
+ HtmlPut( &psz_html, "<" );
+ else if( *psz_subtitle == '>' )
+ HtmlPut( &psz_html, ">" );
+ else
+ *psz_html++ = *psz_subtitle;
+ }
}
- p_style->font_style.i_shadow_width = ( i_border == 1 ) ? i_shadow : 0;
- p_style->font_style.i_outline_width = ( i_border == 1 ) ? i_outline : 0;
- p_style->font_style.i_spacing = i_spacing;
- //p_style->font_style.f_angle = f_angle;
-
- p_style->i_align = 0;
- if( i_align == 0x1 || i_align == 0x4 || i_align == 0x1 ) p_style->i_align |= SUBPICTURE_ALIGN_LEFT;
- if( i_align == 0x3 || i_align == 0x6 || i_align == 0x9 ) p_style->i_align |= SUBPICTURE_ALIGN_RIGHT;
- if( i_align == 0x7 || i_align == 0x8 || i_align == 0x9 ) p_style->i_align |= SUBPICTURE_ALIGN_TOP;
- if( i_align == 0x1 || i_align == 0x2 || i_align == 0x3 ) p_style->i_align |= SUBPICTURE_ALIGN_BOTTOM;
- p_style->i_margin_h = ( p_style->i_align & SUBPICTURE_ALIGN_RIGHT ) ? i_margin_r : i_margin_l;
- p_style->i_margin_v = i_margin_v;
-
- /*TODO: Ignored: angle i_scale_x|y (fontscaling), i_encoding */
- TAB_APPEND( p_sys->i_ssa_styles, p_sys->pp_ssa_styles, p_style );
}
- else msg_Dbg( p_dec, "SSA V4+ styleline parsing failed" );
+ else
+ {
+ /* We have a single < */
+ HtmlPut( &psz_html, "<" );
+ psz_subtitle++;
+ }
}
}
- psz_parser = GotoNextLine( psz_parser );
- }
-
-eof:
- if( psz_header ) free( psz_header );
- return;
-}
-
-static void StripTags( char *psz_text )
-{
- int i_left_moves = 0;
- vlc_bool_t b_inside_tag = VLC_FALSE;
- int i = 0;
- int i_tag_start = -1;
- while( psz_text[ i ] )
- {
- if( !b_inside_tag )
+ else if( *psz_subtitle == '&' )
{
- if( psz_text[ i ] == '<' )
+ if( !strncasecmp( psz_subtitle, "<", 4 ))
+ {
+ HtmlCopy( &psz_html, &psz_subtitle, "<" );
+ }
+ else if( !strncasecmp( psz_subtitle, ">", 4 ))
{
- b_inside_tag = VLC_TRUE;
- i_tag_start = i;
+ HtmlCopy( &psz_html, &psz_subtitle, ">" );
+ }
+ else if( !strncasecmp( psz_subtitle, "&", 5 ))
+ {
+ HtmlCopy( &psz_html, &psz_subtitle, "&" );
+ }
+ else
+ {
+ HtmlPut( &psz_html, "&" );
+ psz_subtitle++;
}
- psz_text[ i - i_left_moves ] = psz_text[ i ];
}
- else
+ else if( *psz_subtitle == '>' )
+ {
+ HtmlPut( &psz_html, ">" );
+ psz_subtitle++;
+ }
+ else if( psz_subtitle[0] == '{' && psz_subtitle[1] == '\\' &&
+ strchr( psz_subtitle, '}' ) )
+ {
+ /* Check for forced alignment */
+ if( !b_has_align &&
+ !strncmp( psz_subtitle, "{\\an", 4 ) && psz_subtitle[4] >= '1' && psz_subtitle[4] <= '9' && psz_subtitle[5] == '}' )
+ {
+ static const int pi_vertical[3] = { SUBPICTURE_ALIGN_BOTTOM, 0, SUBPICTURE_ALIGN_TOP };
+ static const int pi_horizontal[3] = { SUBPICTURE_ALIGN_LEFT, 0, SUBPICTURE_ALIGN_RIGHT };
+ const int i_id = psz_subtitle[4] - '1';
+
+ b_has_align = true;
+ *pi_align = pi_vertical[i_id/3] | pi_horizontal[i_id%3];
+ }
+ /* TODO fr -> rotation */
+
+ /* Hide {\stupidity} */
+ psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
+ }
+ else if( psz_subtitle[0] == '{' && psz_subtitle[1] == 'Y'
+ && psz_subtitle[2] == ':' && strchr( psz_subtitle, '}' ) )
+ {
+ /* Hide {Y:stupidity} */
+ psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
+ }
+ else if( psz_subtitle[0] == '\\' && psz_subtitle[1] )
{
- if( ( psz_text[ i ] == ' ' ) ||
- ( psz_text[ i ] == '\t' ) ||
- ( psz_text[ i ] == '\n' ) ||
- ( psz_text[ i ] == '\r' ) )
+ if( psz_subtitle[1] == 'N' || psz_subtitle[1] == 'n' )
{
- b_inside_tag = VLC_FALSE;
- i_tag_start = -1;
+ HtmlPut( &psz_html, "<br/>" );
+ psz_subtitle += 2;
}
- else if( psz_text[ i ] == '>' )
+ else if( psz_subtitle[1] == 'h' )
{
- i_left_moves += i - i_tag_start + 1;
- i_tag_start = -1;
- b_inside_tag = VLC_FALSE;
+ /* Non breakable space */
+ HtmlPut( &psz_html, NO_BREAKING_SPACE );
+ psz_subtitle += 2;
}
else
{
- psz_text[ i - i_left_moves ] = psz_text[ i ];
+ HtmlPut( &psz_html, "\\" );
+ psz_subtitle++;
}
}
- i++;
+ else
+ {
+ *psz_html = *psz_subtitle;
+ if( psz_html > psz_html_start )
+ {
+ /* Check for double whitespace */
+ if( ( *psz_html == ' ' || *psz_html == '\t' ) &&
+ ( *(psz_html-1) == ' ' || *(psz_html-1) == '\t' ) )
+ {
+ HtmlPut( &psz_html, NO_BREAKING_SPACE );
+ psz_html--;
+ }
+ }
+ psz_html++;
+ psz_subtitle++;
+ }
+
+ if( ( size_t )( psz_html - psz_html_start ) > i_buf_size - 50 )
+ {
+ const int i_len = psz_html - psz_html_start;
+
+ i_buf_size += 200;
+ char *psz_new = realloc( psz_html_start, i_buf_size );
+ if( !psz_new )
+ break;
+ psz_html_start = psz_new;
+ psz_html = &psz_new[i_len];
+ }
}
- psz_text[ i - i_left_moves ] = '\0';
+ if( psz_html_start )
+ {
+ static const char *psz_text_close = "</text>";
+ static const char *psz_tag_long = "/font>";
+
+ /* Realloc for closing tags and shrink memory */
+ const size_t i_length = (size_t)( psz_html - psz_html_start );
+
+ const size_t i_size = i_length + strlen(psz_tag_long) * strlen(psz_tag) + strlen(psz_text_close) + 1;
+ char *psz_new = realloc( psz_html_start, i_size );
+ if( psz_new )
+ {
+ psz_html_start = psz_new;
+ psz_html = &psz_new[i_length];
+
+ /* Close not well formed subtitle */
+ while( *psz_tag )
+ {
+ /* */
+ char *psz_last = &psz_tag[strlen(psz_tag)-1];
+ switch( *psz_last )
+ {
+ case 'b':
+ HtmlPut( &psz_html, "</b>" );
+ break;
+ case 'i':
+ HtmlPut( &psz_html, "</i>" );
+ break;
+ case 'u':
+ HtmlPut( &psz_html, "</u>" );
+ break;
+ case 's':
+ HtmlPut( &psz_html, "</s>" );
+ break;
+ case 'f':
+ HtmlPut( &psz_html, "/font>" );
+ break;
+ case 'I':
+ break;
+ }
+
+ *psz_last = '\0';
+ }
+ HtmlPut( &psz_html, psz_text_close );
+ }
+ }
+ free( psz_tag );
+
+ return psz_html_start;
}
+