# include "config.h"
#endif
-#include <vlc/vlc.h>
+#include <vlc_common.h>
#include <vlc_plugin.h>
#include <vlc_input.h>
"This will only work with MicroDVD and SubRIP (SRT) subtitles.")
#define SUB_TYPE_LONGTEXT \
N_("Force the subtiles format. Valid values are : \"microdvd\", " \
- "\"subrip\", \"ssa1\", \"ssa2-4\", \"ass\", \"vplayer\" " \
- "\"sami\", \"dvdsubtitle\", \"mpl2\", \"aqt\", \"pjs\" "\
- "\"mpsub\" \"jacosub\" and \"auto\" (meaning autodetection, this " \
- "should always work).")
-static const char *ppsz_sub_type[] =
+ "\"subrip\", \"subviewer\", \"ssa1\", \"ssa2-4\", \"ass\", \"vplayer\", " \
+ "\"sami\", \"dvdsubtitle\", \"mpl2\", \"aqt\", \"pjs\", "\
+ "\"mpsub\", \"jacosub\", \"psb\", \"realtext\", \"dks\", \"subviewer1\", " \
+ " and \"auto\" (meaning autodetection, this should always work).")
+
+static const char *const ppsz_sub_type[] =
{
"auto", "microdvd", "subrip", "subviewer", "ssa1",
"ssa2-4", "ass", "vplayer", "sami", "dvdsubtitle", "mpl2",
- "aqt", "pjs", "mpsub", "jacosub"
+ "aqt", "pjs", "mpsub", "jacosub", "psb", "realtext", "dks",
+ "subviewer1"
};
vlc_module_begin();
- set_shortname( _("Subtitles"));
- set_description( _("Text subtitles parser") );
+ set_shortname( N_("Subtitles"));
+ set_description( N_("Text subtitles parser") );
set_capability( "demux", 0 );
set_category( CAT_INPUT );
set_subcategory( SUBCAT_INPUT_DEMUX );
SUB_TYPE_ASS,
SUB_TYPE_VPLAYER,
SUB_TYPE_SAMI,
- SUB_TYPE_SUBVIEWER, //SUBVIEWER 2!
- SUB_TYPE_DVDSUBTITLE,
+ SUB_TYPE_SUBVIEWER, /* SUBVIEWER 2 */
+ SUB_TYPE_DVDSUBTITLE, /* Mplayer calls it subviewer2 */
SUB_TYPE_MPL2,
SUB_TYPE_AQT,
SUB_TYPE_PJS,
SUB_TYPE_MPSUB,
SUB_TYPE_JACOSUB,
- SUB_TYPE_PSB
+ SUB_TYPE_PSB,
+ SUB_TYPE_RT,
+ SUB_TYPE_DKS,
+ SUB_TYPE_SUBVIEW1 /* SUBVIEWER 1 - mplayer calls it subrip09,
+ and Gnome subtitles SubViewer 1.0 */
};
typedef struct
static int ParseMPSub ( demux_t *, subtitle_t *, int );
static int ParseJSS ( demux_t *, subtitle_t *, int );
static int ParsePSB ( demux_t *, subtitle_t *, int );
+static int ParseRealText ( demux_t *, subtitle_t *, int );
+static int ParseDKS ( demux_t *, subtitle_t *, int );
+static int ParseSubViewer1 ( demux_t *, subtitle_t *, int );
static struct
{
{ "mpsub", SUB_TYPE_MPSUB, "MPSub", ParseMPSub },
{ "jacosub", SUB_TYPE_JACOSUB, "JacoSub", ParseJSS },
{ "psb", SUB_TYPE_PSB, "PowerDivx", ParsePSB },
+ { "realtext", SUB_TYPE_RT, "RealText", ParseRealText },
+ { "dks", SUB_TYPE_DKS, "DKS", ParseDKS },
+ { "subviewer1", SUB_TYPE_SUBVIEW1, "Subviewer 1", ParseSubViewer1 },
{ NULL, SUB_TYPE_UNKNOWN, "Unknown", NULL }
};
-/* Missing Detect
- SubViewer 1
- JSS
- RealText
- Subrip09
- */
-
-
static int Demux( demux_t * );
static int Control( demux_t *, int, va_list );
p_sys->i_microsecperframe = 40000;
/* Get the FPS */
- f_fps = var_CreateGetFloat( p_demux, "sub-original-fps" );
+ f_fps = var_CreateGetFloat( p_demux, "sub-original-fps" ); /* FIXME */
if( f_fps >= 1.0 )
p_sys->i_microsecperframe = (int64_t)( (float)1000000 / f_fps );
p_sys->i_type = SUB_TYPE_SUBVIEWER; /* I hope this will work */
break;
}
- else if( sscanf( s, "%d:%d:%d.%d %d:%d:%d", &i_dummy, &i_dummy, &i_dummy, &i_dummy, &i_dummy, &i_dummy, &i_dummy ) == 7 ||
- sscanf( s, "@%d @%d", &i_dummy, &i_dummy) == 2)
+ else if( sscanf( s, "%d:%d:%d.%d %d:%d:%d",
+ &i_dummy, &i_dummy, &i_dummy, &i_dummy,
+ &i_dummy, &i_dummy, &i_dummy ) == 7 ||
+ sscanf( s, "@%d @%d", &i_dummy, &i_dummy) == 2)
{
p_sys->i_type = SUB_TYPE_JACOSUB;
+ break;
}
else if( sscanf( s, "%d:%d:%d:", &i_dummy, &i_dummy, &i_dummy ) == 3 ||
sscanf( s, "%d:%d:%d ", &i_dummy, &i_dummy, &i_dummy ) == 3 )
p_sys->i_type = SUB_TYPE_DVDSUBTITLE;
break;
}
+ else if( sscanf( s, "[%d:%d:%d]%c",
+ &i_dummy, &i_dummy, &i_dummy, &p_dummy ) == 4 )
+ {
+ p_sys->i_type = SUB_TYPE_DKS;
+ break;
+ }
+ else if( strstr( s, "*** START SCRIPT" ) )
+ {
+ p_sys->i_type = SUB_TYPE_SUBVIEW1;
+ break;
+ }
else if( sscanf( s, "[%d][%d]", &i_dummy, &i_dummy ) == 2 ||
sscanf( s, "[%d][]", &i_dummy ) == 1)
{
&& p_dummy =='E' ) )
{
p_sys->i_type = SUB_TYPE_MPSUB;
+ break;
}
else if( sscanf( s, "-->> %d", &i_dummy) == 1 )
{
p_sys->i_type = SUB_TYPE_AQT;
+ break;
}
else if( sscanf( s, "%d,%d,", &i_dummy, &i_dummy ) == 2 )
{
p_sys->i_type = SUB_TYPE_PJS;
+ break;
}
- else if( sscanf( s, "{%d:%d:%d}", &i_dummy, &i_dummy, &i_dummy ) == 3 )
+ else if( sscanf( s, "{%d:%d:%d}",
+ &i_dummy, &i_dummy, &i_dummy ) == 3 )
{
p_sys->i_type = SUB_TYPE_PSB;
+ break;
+ }
+ else if( strcasestr( s, "<time" ) )
+ {
+ p_sys->i_type = SUB_TYPE_RT;
+ break;
}
free( s );
msg_Warn( p_demux, "failed to rewind" );
}
}
+
+ /* Quit on unknown subtitles */
if( p_sys->i_type == SUB_TYPE_UNKNOWN )
{
msg_Err( p_demux, "failed to recognize subtitle type" );
if( !( p_sys->subtitle = realloc( p_sys->subtitle,
sizeof(subtitle_t) * i_max ) ) )
{
- msg_Err( p_demux, "out of memory");
free( p_sys->subtitle );
TextUnload( &p_sys->txt );
free( p_sys );
if( b_replace_br )
{
char *p;
-
+
while( ( p = strstr( psz_text, "[br]" ) ) )
{
*p++ = '\n';
/* We have been too far: end of the subtitle, begin of next */
else
{
- txt->i_line--;
+ TextPreviousLine( txt );
break;
}
}
demux_sys_t *p_sys = p_demux->p_sys;
text_t *txt = &p_sys->txt;
char *psz_text;
+ int i;
for( ;; )
{
}
free( psz_text );
}
+
+ /* replace | by \n */
+ for( i = 0; psz_text[i] != '\0'; i++ )
+ {
+ if( psz_text[i] == '|' )
+ psz_text[i] = '\n';
+ }
+
p_subtitle->psz_text = psz_text;
msg_Dbg( p_demux, "%s", psz_text );
return VLC_SUCCESS;
free( psz_temp );
}
/* Data Lines */
- if( sscanf (s, "%f %f", &f1, &f2 ) == 2 )
+ f1 = us_strtod( s, &psz_temp );
+ if( *psz_temp )
{
+ f2 = us_strtod( psz_temp, NULL );
mpsub_total += f1 * mpsub_factor;
p_subtitle->i_start = (int64_t)(10000.0 * mpsub_total);
mpsub_total += f2 * mpsub_factor;
char *psz_text, *psz_orig;
char *psz_text2, *psz_orig2;
int h1, h2, m1, m2, s1, s2, f1, f2;
+
static int i_comment = 0;
static int jss_time_resolution = 30;
if( !s )
return VLC_EGENERIC;
- psz_text = malloc( strlen( s ) + 1 );
- if( !psz_text )
+ psz_orig = malloc( strlen( s ) + 1 );
+ if( !psz_orig )
return VLC_ENOMEM;
- psz_orig = psz_text;
+ psz_text = psz_orig;
/* Complete time lines */
if( sscanf( s, "%d:%d:%d.%d %d:%d:%d.%d %[^\n\r]",
p_subtitle->i_stop = ( (int64_t)( h2 *3600 + m2 * 60 + s2 ) +
(int64_t)( ( f2 + jss_time_shift ) / jss_time_resolution ) )
* 1000000;
+ break;
}
/* Short time lines */
else if( sscanf( s, "@%d @%d %[^\n\r]", &f1, &f2, psz_text ) == 3 )
( f1 + jss_time_shift ) / jss_time_resolution * 1000000.0 );
p_subtitle->i_stop = (int64_t)(
( f2 + jss_time_shift ) / jss_time_resolution * 1000000.0 );
+ break;
}
/* General Directive lines */
/* Only TIME and SHIFT are supported so far */
sscanf( &psz_text[shift], "%d", &jss_time_resolution );
break;
}
- free( psz_text );
+ free( psz_orig );
continue;
}
else
/* Unkown type line, probably a comment */
{
- free( psz_text );
+ free( psz_orig );
continue;
}
+ }
+
+ while( psz_text[ strlen( psz_text ) - 1 ] == '\\' )
+ {
+ const char *s2 = TextGetLine( txt );
- /* Skip the blanks */
- while( *psz_text == ' ' || *psz_text == '\t' ) psz_text++;
+ if( !s2 )
+ return VLC_EGENERIC;
- /* Parse the directives */
- if( isalpha( *psz_text ) || *psz_text == '[' )
- {
- while( *psz_text != ' ' )
- { psz_text++ ;};
+ int i_len = strlen( s2 );
+ if( i_len == 0 )
+ break;
- /* Directives are NOT parsed yet */
- /* This has probably a better place in a decoder ? */
- /* directive = malloc( strlen( psz_text ) + 1 );
- if( sscanf( psz_text, "%s %[^\n\r]", directive, psz_text2 ) == 2 )*/
- }
+ int i_old = strlen( psz_text );
+
+ psz_text = realloc( psz_text, i_old + i_len + 1 );
+ if( !psz_text )
+ return VLC_ENOMEM;
- /* Skip the blanks after directives */
- while( *psz_text == ' ' || *psz_text == '\t' ) psz_text++;
+ psz_orig = psz_text;
+ strcat( psz_text, s2 );
+ }
+ /* Skip the blanks */
+ while( *psz_text == ' ' || *psz_text == '\t' ) psz_text++;
- /* Clean all the lines from inline comments and other stuffs */
- psz_text2 = calloc( strlen( psz_text) + 1, 1 );
- psz_orig2 = psz_text2;
+ /* Parse the directives */
+ if( isalpha( *psz_text ) || *psz_text == '[' )
+ {
+ while( *psz_text != ' ' )
+ { psz_text++ ;};
+
+ /* Directives are NOT parsed yet */
+ /* This has probably a better place in a decoder ? */
+ /* directive = malloc( strlen( psz_text ) + 1 );
+ if( sscanf( psz_text, "%s %[^\n\r]", directive, psz_text2 ) == 2 )*/
+ }
- for( ; *psz_text != '\0' && *psz_text != '\n' && *psz_text != '\r'; )
+ /* Skip the blanks after directives */
+ while( *psz_text == ' ' || *psz_text == '\t' ) psz_text++;
+
+ /* Clean all the lines from inline comments and other stuffs */
+ psz_orig2 = calloc( strlen( psz_text) + 1, 1 );
+ psz_text2 = psz_orig2;
+
+ for( ; *psz_text != '\0' && *psz_text != '\n' && *psz_text != '\r'; )
+ {
+ switch( *psz_text )
{
- switch( *psz_text )
+ case '{':
+ i_comment++;
+ break;
+ case '}':
+ if( i_comment )
{
- case '{':
- i_comment++;
+ i_comment = 0;
+ if( (*(psz_text + 1 ) ) == ' ' ) psz_text++;
+ }
+ break;
+ case '~':
+ if( !i_comment )
+ {
+ *psz_text2 = ' ';
+ psz_text2++;
+ }
+ break;
+ case ' ':
+ case '\t':
+ if( (*(psz_text + 1 ) ) == ' ' || (*(psz_text + 1 ) ) == '\t' )
break;
- case '}':
- if( i_comment )
- {
- i_comment = 0;
- if( (*(psz_text + 1 ) ) == ' ' ) psz_text++;
- }
+ if( !i_comment )
+ {
+ *psz_text2 = ' ';
+ psz_text2++;
+ }
+ break;
+ case '\\':
+ if( (*(psz_text + 1 ) ) == 'n' )
+ {
+ *psz_text2 = '\n';
+ psz_text++;
+ psz_text2++;
break;
- case '~':
- if( !i_comment )
- {
- *psz_text2 = ' ';
- psz_text2++;
- }
+ }
+ if( ( toupper(*(psz_text + 1 ) ) == 'C' ) ||
+ ( toupper(*(psz_text + 1 ) ) == 'F' ) )
+ {
+ psz_text++; psz_text++;
break;
- case ' ':
- case '\t':
- if( (*(psz_text + 1 ) ) == ' ' || (*(psz_text + 1 ) ) == '\t' )
- break;
- if( !i_comment )
- {
- *psz_text2 = ' ';
- psz_text2++;
- }
+ }
+ if( (*(psz_text + 1 ) ) == 'B' || (*(psz_text + 1 ) ) == 'b' ||
+ (*(psz_text + 1 ) ) == 'I' || (*(psz_text + 1 ) ) == 'i' ||
+ (*(psz_text + 1 ) ) == 'U' || (*(psz_text + 1 ) ) == 'u' ||
+ (*(psz_text + 1 ) ) == 'D' || (*(psz_text + 1 ) ) == 'N' )
+ {
+ psz_text++;
break;
- case '\\':
- if( (*(psz_text + 1 ) ) == 'n' )
- {
- *psz_text2 = '\n';
- psz_text++;
- psz_text2++;
- break;
- }
- if( ( toupper(*(psz_text + 1 ) ) == 'C' ) ||
- ( toupper(*(psz_text + 1 ) ) == 'F' ) )
- {
- psz_text++; psz_text++;
- break;
- }
- if( (*(psz_text + 1 ) ) == 'B' || (*(psz_text + 1 ) ) == 'b' ||
- (*(psz_text + 1 ) ) == 'I' || (*(psz_text + 1 ) ) == 'i' ||
- (*(psz_text + 1 ) ) == 'U' || (*(psz_text + 1 ) ) == 'u' ||
- (*(psz_text + 1 ) ) == 'D' || (*(psz_text + 1 ) ) == 'N' )
- {
- psz_text++;
- break;
- }
- if( (*(psz_text + 1 ) ) == '~' || (*(psz_text + 1 ) ) == '{' ||
- (*(psz_text + 1 ) ) == '\\' )
- psz_text++;
- else if( *(psz_text + 1 ) == '\r' || *(psz_text + 1 ) == '\n'
- || *(psz_text + 1 ) == '\0' )
- {
- char *s2 = TextGetLine( txt );
- if( !s2 )
- return VLC_EGENERIC;
-
- while ( *s2 == ' ' ) s2++;
-
- /* Here to parse the second line, we should add s2 to
- psz_text and go on the for( ) line 1556 in order to
- parse the next line.
- */
- }
- default:
- if( !i_comment )
- {
- *psz_text2 = *psz_text;
- psz_text2++;
- }
}
- psz_text++;
+ if( (*(psz_text + 1 ) ) == '~' || (*(psz_text + 1 ) ) == '{' ||
+ (*(psz_text + 1 ) ) == '\\' )
+ psz_text++;
+ else if( *(psz_text + 1 ) == '\r' || *(psz_text + 1 ) == '\n' ||
+ *(psz_text + 1 ) == '\0' )
+ {
+ psz_text++;
+ }
+ break;
+ default:
+ if( !i_comment )
+ {
+ *psz_text2 = *psz_text;
+ psz_text2++;
+ }
}
-
- p_subtitle->psz_text = psz_orig2;
- free( psz_orig );
- return VLC_SUCCESS;
+ psz_text++;
}
+
+ p_subtitle->psz_text = psz_orig2;
+ msg_Dbg( p_demux, "%s", p_subtitle->psz_text );
+ free( psz_orig );
+ return VLC_SUCCESS;
}
static int ParsePSB( demux_t *p_demux, subtitle_t *p_subtitle, int i_idx )
return VLC_SUCCESS;
}
+static int64_t ParseRealTime( char *psz, int *h, int *m, int *s, int *f )
+{
+ if( strlen( psz ) == 0 ) return 0;
+ if( sscanf( psz, "%d:%d:%d.%d", h, m, s, f ) == 4 ||
+ sscanf( psz, "%d:%d.%d", m, s, f ) == 3 ||
+ sscanf( psz, "%d.%d", s, f ) == 2 ||
+ sscanf( psz, "%d:%d", m, s ) == 2 ||
+ sscanf( psz, "%d", s ) == 1 )
+ {
+ return (int64_t)((( *h * 60 + *m ) * 60 ) + *s ) * 1000 * 1000
+ + (int64_t)*f * 10 * 1000;
+ }
+ else return VLC_EGENERIC;
+}
+
+static int ParseRealText( demux_t *p_demux, subtitle_t *p_subtitle, int i_idx )
+{
+ VLC_UNUSED( i_idx );
+ demux_sys_t *p_sys = p_demux->p_sys;
+ text_t *txt = &p_sys->txt;
+ char *psz_text;
+ char psz_end[12]= "", psz_begin[12] = "";
+
+ for( ;; )
+ {
+ int h1 = 0, m1 = 0, s1 = 0, f1 = 0;
+ int h2 = 0, m2 = 0, s2 = 0, f2 = 0;
+ const char *s = TextGetLine( txt );
+
+ if( !s )
+ return VLC_EGENERIC;
+
+ psz_text = malloc( strlen( s ) + 1 );
+ if( !psz_text )
+ return VLC_ENOMEM;
+
+ /* Find the good begining. This removes extra spaces at the beginning
+ of the line.*/
+ char *psz_temp = strcasestr( s, "<time");
+ if( psz_temp != NULL )
+ {
+ /* Line has begin and end */
+ if( ( sscanf( psz_temp,
+ "<%*[t|T]ime %*[b|B]egin=\"%[^\"]\" %*[e|E]nd=\"%[^\"]%*[^>]%[^\n\r]",
+ psz_begin, psz_end, psz_text) != 3 ) &&
+ /* Line has begin and no end */
+ ( sscanf( psz_temp,
+ "<%*[t|T]ime %*[b|B]egin=\"%[^\"]\"%*[^>]%[^\n\r]",
+ psz_begin, psz_text ) != 2) )
+ /* Line is not recognized */
+ {
+ free( psz_text );
+ continue;
+ }
+
+ /* Get the times */
+ int64_t i_time = ParseRealTime( psz_begin, &h1, &m1, &s1, &f1 );
+ if( i_time >= 0)
+ {
+ p_subtitle->i_start = i_time;
+ }
+
+ i_time = ParseRealTime( psz_end, &h2, &m2, &s2, &f2 );
+ if( i_time >= 0 )
+ {
+ p_subtitle->i_stop = i_time;
+ }
+ break;
+ }
+ /* Line is not recognized */
+ else continue;
+ free( psz_text );
+ }
+
+ /* Get the following Lines */
+ for( ;; )
+ {
+ const char *s = TextGetLine( txt );
+
+ if( !s )
+ return VLC_EGENERIC;
+
+ int i_len = strlen( s );
+ if( i_len == 0 ) break;
+
+ if( strcasestr( s, "<time" ) ||
+ strcasestr( s, "<clear/") )
+ {
+ TextPreviousLine( txt );
+ break;
+ }
+
+ int i_old = strlen( psz_text );
+
+ psz_text = realloc( psz_text, i_old + i_len + 1 + 1 );
+ if( !psz_text )
+ return VLC_ENOMEM;
+
+ strcat( psz_text, s );
+ strcat( psz_text, "\n" );
+ }
+
+ /* Remove the starting ">" that remained after the sscanf */
+ memmove( &psz_text[0], &psz_text[1], strlen( psz_text ) );
+
+ p_subtitle->psz_text = psz_text;
+
+ return VLC_SUCCESS;
+}
+
+static int ParseDKS( demux_t *p_demux, subtitle_t *p_subtitle, int i_idx )
+{
+ VLC_UNUSED( i_idx );
+
+ demux_sys_t *p_sys = p_demux->p_sys;
+ text_t *txt = &p_sys->txt;
+ char *psz_text;
+
+ for( ;; )
+ {
+ int h1, m1, s1;
+ int h2, m2, s2;
+ char *s = TextGetLine( txt );
+
+ if( !s )
+ return VLC_EGENERIC;
+
+ psz_text = malloc( strlen( s ) + 1 );
+ if( !psz_text )
+ return VLC_ENOMEM;
+
+ if( sscanf( s, "[%d:%d:%d]%[^\r\n]",
+ &h1, &m1, &s1, psz_text ) == 4 )
+ {
+ p_subtitle->i_start = ( (int64_t)h1 * 3600*1000 +
+ (int64_t)m1 * 60*1000 +
+ (int64_t)s1 * 1000 ) * 1000;
+
+ char *s = TextGetLine( txt );
+ if( !s )
+ return VLC_EGENERIC;
+
+ if( sscanf( s, "[%d:%d:%d]", &h2, &m2, &s2 ) == 3 )
+ p_subtitle->i_stop = ( (int64_t)h2 * 3600*1000 +
+ (int64_t)m2 * 60*1000 +
+ (int64_t)s2 * 1000 ) * 1000;
+ break;
+ }
+ free( psz_text );
+ }
+
+ /* replace [br] by \n */
+ char *p;
+ while( ( p = strstr( psz_text, "[br]" ) ) )
+ {
+ *p++ = '\n';
+ memmove( p, &p[3], strlen(&p[3])+1 );
+ }
+
+ p_subtitle->psz_text = psz_text;
+ return VLC_SUCCESS;
+}
+
+static int ParseSubViewer1( demux_t *p_demux, subtitle_t *p_subtitle, int i_idx )
+{
+ VLC_UNUSED( i_idx );
+
+ demux_sys_t *p_sys = p_demux->p_sys;
+ text_t *txt = &p_sys->txt;
+ char *psz_text;
+
+ for( ;; )
+ {
+ int h1, m1, s1;
+ int h2, m2, s2;
+ char *s = TextGetLine( txt );
+
+ if( !s )
+ return VLC_EGENERIC;
+
+ if( sscanf( s, "[%d:%d:%d]", &h1, &m1, &s1 ) == 3 )
+ {
+ p_subtitle->i_start = ( (int64_t)h1 * 3600*1000 +
+ (int64_t)m1 * 60*1000 +
+ (int64_t)s1 * 1000 ) * 1000;
+
+ char *s = TextGetLine( txt );
+ if( !s )
+ return VLC_EGENERIC;
+
+ psz_text = strdup( s );
+ if( !psz_text )
+ return VLC_ENOMEM;
+
+ s = TextGetLine( txt );
+ if( !s )
+ return VLC_EGENERIC;
+
+ if( sscanf( s, "[%d:%d:%d]", &h2, &m2, &s2 ) == 3 )
+ p_subtitle->i_stop = ( (int64_t)h2 * 3600*1000 +
+ (int64_t)m2 * 60*1000 +
+ (int64_t)s2 * 1000 ) * 1000;
+ break;
+ }
+ }
+
+ p_subtitle->psz_text = psz_text;
+
+ return VLC_SUCCESS;
+}
projects / vlc / blobdiff