subsdec.c

VLC Player Source Code

    /* 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_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 );
    if( !p_spu->p_region )
    {
        msg_Err( p_dec, "cannot allocate SPU region" );
        free( psz_subtitle );
        p_dec->pf_spu_buffer_del( p_dec, p_spu );
        return NULL;
    }

    /* Decode and format the subpicture unit */
    if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
    {
        /* Normal text subs, easy markup */
        p_spu->p_region->i_align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
        p_spu->i_x = p_sys->i_align ? 20 : 0;
        p_spu->i_y = 10;

        /* Remove formatting from string */

        p_spu->p_region->psz_text = StripTags( psz_subtitle );
        if( var_CreateGetBool( p_dec, "subsdec-formatted" ) )
        {
            p_spu->p_region->psz_html = CreateHtmlSubtitle( 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 = false;
    }
    else
    {
        /* Decode SSA/USF strings */
        if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ') )
            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 = false;
        p_spu->i_original_picture_width = p_sys->i_original_width;
        p_spu->i_original_picture_height = p_sys->i_original_height;
    }
    free( psz_subtitle );

    return p_spu;
}

char* GotoNextLine( char *psz_text )
{
    char *p_newline = 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;
}

/* 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;

    while( *psz_subtitle )
    {
        if( *psz_subtitle == '<' )
        {
            if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
                *psz_text++ = '\n';

            psz_subtitle += strcspn( psz_subtitle, ">" );
        }
        else if( *psz_subtitle == '&' )
        {
            if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
            {
                *psz_text++ = '<';
                psz_subtitle += strcspn( psz_subtitle, ";" );
            }
            else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
            {
                *psz_text++ = '>';
                psz_subtitle += strcspn( psz_subtitle, ";" );
            }
            else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
            {
                *psz_text++ = '&';
                psz_subtitle += strcspn( psz_subtitle, ";" );
            }
            else if( !strncasecmp( psz_subtitle, "&quot;", 6 ))
            {
                *psz_text++ = '\"';
                psz_subtitle += strcspn( psz_subtitle, ";" );
            }
            else
            {
                /* Assume it is just a normal ampersand */
                *psz_text++ = '&';
            }
        }
        else
        {
            *psz_text++ = *psz_subtitle;
        }

        psz_subtitle++;
    }
    *psz_text = '\0';
    psz_text_start = realloc( psz_text_start, strlen( psz_text_start ) + 1 );

    return psz_text_start;
}

/* 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>
 *  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 )
{
    const int i_len = strlen(psz_text);

    strncpy( *ppsz_html, psz_text, i_max );
    *ppsz_html += __MIN(i_max,i_len);
}

static void HtmlPut( char **ppsz_html, const char *psz_text )
{
    strcpy( *ppsz_html, psz_text );
    *ppsz_html += strlen(psz_text);
}
static void HtmlCopy( char **ppsz_html, char **ppsz_subtitle, const char *psz_text )
{
    HtmlPut( ppsz_html, psz_text );
    *ppsz_subtitle += strlen(psz_text);
}

static char *CreateHtmlSubtitle( char *psz_subtitle )
{
    char   *psz_tag = malloc( ( strlen( psz_subtitle ) / 3 ) + 1 );
    if( !psz_tag ) return NULL;
    size_t  i_buf_size     = strlen( psz_subtitle ) + 100;
    char   *psz_html_start = malloc( i_buf_size );

    psz_tag[ 0 ] = '\0';

    if( psz_html_start == NULL )
    {
        free( psz_tag );
        return NULL;
    }
    
    char *psz_html = psz_html_start;

    strcpy( psz_html, "<text>" );
    psz_html += 6;

    while( *psz_subtitle )
    {
        if( *psz_subtitle == '\n' )
        {
            HtmlPut( &psz_html, "<br/>" );
            psz_subtitle++;
        }
        else if( *psz_subtitle == '<' )
        {
            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, "<font ", 6 ))
            {
                const char *psz_attribs[] = { "face=", "family=", "size=",
                        "color=", "outline-color=", "shadow-color=",
                        "outline-level=", "shadow-level=", "back-color=",
                        "alpha=", NULL };

                HtmlCopy( &psz_html, &psz_subtitle, "<font " );
                strcat( psz_tag, "f" );

                while( *psz_subtitle != '>' )
                {
                    int  k;

                    for( k=0; psz_attribs[ k ]; k++ )
                    {
                        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;
                        }
                    }
                    if( psz_attribs[ k ] == NULL )
                    {
                        /* Jump over unrecognised tag */
                        int i_len = strcspn( psz_subtitle, "\"" ) + 1;

                        i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
                        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;
                int    i_len       = strlen( psz_tag ) - 1;
                char  *psz_lastTag = NULL;

                if( i_len >= 0 )
                {
                    psz_lastTag = psz_tag + i_len;
                    i_len = 0;

                    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 'f':
                        b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
                        i_len   = 7;
                        break;
                    }
                }
                if( ! b_match )
                {
                    /* Not well formed -- kill everything */
                    free( psz_html_start );
                    psz_html_start = NULL;
                    break;
                }
                *psz_lastTag = '\0';
                strncpy( psz_html, psz_subtitle, i_len );
                psz_html += i_len;
                psz_subtitle += i_len;
            }
            else
            {
                psz_subtitle += strcspn( psz_subtitle, ">" );
            }
        }
        else if( *psz_subtitle == '&' )
        {
            if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
            {
                HtmlCopy( &psz_html, &psz_subtitle, "&lt;" );
            }
            else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
            {
                HtmlCopy( &psz_html, &psz_subtitle, "&gt;" );
            }
            else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
            {
                HtmlCopy( &psz_html, &psz_subtitle, "&amp;" );
            }
            else
            {
                HtmlPut( &psz_html, "&amp;" );
                psz_subtitle++;
            }
        }
        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 )
        {
            int i_len = psz_html - psz_html_start;

            i_buf_size += 200;
            psz_html_start = realloc( psz_html_start, i_buf_size );
            psz_html = psz_html_start + i_len;
            *psz_html = '\0';
        }
    }
    strcpy( psz_html, "</text>" );
    psz_html += 7;

    if( psz_tag[ 0 ] != '\0' )
    {
        /* Not well formed -- kill everything */
        free( psz_html_start );
        psz_html_start = NULL;
    }
    else if( psz_html_start )
    {
        /* Shrink the memory requirements */
        psz_html_start = realloc( psz_html_start,  psz_html - psz_html_start + 1 );
    }
    free( psz_tag );

    return psz_html_start;
}