ass.c

来自「君正早期ucos系统(只有早期的才不没有打包成库),MPLAYER,文件系统,图」· C语言 代码 · 共 1,093 行 · 第 1/2 页

	bytes[0] = (value & 0xff0000) >> 16;

	for (i = 0; i < cnt; ++i)
		*dst++ = bytes[i];
	return dst;
}

static int decode_font(ass_track_t* track)
{
	unsigned char* p;
	unsigned char* q;
	int i;
	int size; // original size
	int dsize; // decoded size
	unsigned char* buf = 0;

	mp_msg(MSGT_ASS, MSGL_V, "font: %d bytes encoded data \n", track->parser_priv->fontdata_used);
	size = track->parser_priv->fontdata_used;
	if (size % 4 == 1) {
		mp_msg(MSGT_ASS, MSGL_ERR, MSGTR_LIBASS_BadEncodedDataSize);
		goto error_decode_font;
	}
	buf = malloc(size / 4 * 3 + 2);
	q = buf;
	for (i = 0, p = (unsigned char*)track->parser_priv->fontdata; i < size / 4; i++, p+=4) {
		q = decode_chars(p[0], p[1], p[2], p[3], q, 3);
	}
	if (size % 4 == 2) {
		q = decode_chars(p[0], p[1], 0, 0, q, 1);
	} else if (size % 4 == 3) {
		q = decode_chars(p[0], p[1], p[2], 0, q, 2);
	}
	dsize = q - buf;
	assert(dsize <= size / 4 * 3 + 2);
	
	if (track->library->extract_fonts) {
		ass_add_font(track->library, track->parser_priv->fontname, (char*)buf, dsize);
		buf = 0;
	}

error_decode_font:
	if (buf) free(buf);
	free(track->parser_priv->fontname);
	free(track->parser_priv->fontdata);
	track->parser_priv->fontname = 0;
	track->parser_priv->fontdata = 0;
	track->parser_priv->fontdata_size = 0;
	track->parser_priv->fontdata_used = 0;
	return 0;
}

static int process_fonts_line(ass_track_t* track, char *str)
{
	int len;

	if (!strncmp(str, "fontname:", 9)) {
		char* p = str + 9;
		skip_spaces(&p);
		if (track->parser_priv->fontname) {
			decode_font(track);
		}
		track->parser_priv->fontname = strdup(p);
		mp_msg(MSGT_ASS, MSGL_V, "fontname: %s\n", track->parser_priv->fontname);
		return 0;
	}
	
	if (!track->parser_priv->fontname) {
		mp_msg(MSGT_ASS, MSGL_V, "Not understood: %s  \n", str);
		return 0;
	}

	len = strlen(str);
	if (len > 80) {
		mp_msg(MSGT_ASS, MSGL_WARN, MSGTR_LIBASS_FontLineTooLong, len, str);
		return 0;
	}
	if (track->parser_priv->fontdata_used + len > track->parser_priv->fontdata_size) {
		track->parser_priv->fontdata_size += 100 * 1024;
		track->parser_priv->fontdata = realloc(track->parser_priv->fontdata, track->parser_priv->fontdata_size);
	}
	memcpy(track->parser_priv->fontdata + track->parser_priv->fontdata_used, str, len);
	track->parser_priv->fontdata_used += len;
	
	return 0;
}

/**
 * \brief Parse a header line
 * \param track track
 * \param str string to parse, zero-terminated
*/ 
static int process_line(ass_track_t* track, char *str)
{
	if (strstr(str, "[Script Info]")) { // FIXME: strstr to skip possible BOM at the beginning of the script
		track->parser_priv->state = PST_INFO;
	} else if (!strncmp(str, "[V4 Styles]", 11)) {
		track->parser_priv->state = PST_STYLES;
		track->track_type = TRACK_TYPE_SSA;
	} else if (!strncmp(str, "[V4+ Styles]", 12)) {
		track->parser_priv->state = PST_STYLES;
		track->track_type = TRACK_TYPE_ASS;
	} else if (!strncmp(str, "[Events]", 8)) {
		track->parser_priv->state = PST_EVENTS;
	} else if (!strncmp(str, "[Fonts]", 7)) {
		track->parser_priv->state = PST_FONTS;
	} else {
		switch (track->parser_priv->state) {
		case PST_INFO:
			process_info_line(track, str);
			break;
		case PST_STYLES:
			process_styles_line(track, str);
			break;
		case PST_EVENTS:
			process_events_line(track, str);
			break;
		case PST_FONTS:
			process_fonts_line(track, str);
			break;
		default:
			break;
		}
	}

	// there is no explicit end-of-font marker in ssa/ass
	if ((track->parser_priv->state != PST_FONTS) && (track->parser_priv->fontname))
		decode_font(track);

	return 0;
}

static int process_text(ass_track_t* track, char* str)
{
	char* p = str;
	while(1) {
		char* q;
		for (;((*p=='\r')||(*p=='\n'));++p) {}
		for (q=p; ((*q!='\0')&&(*q!='\r')&&(*q!='\n')); ++q) {};
		if (q==p)
			break;
		if (*q != '\0')
			*(q++) = '\0';
		process_line(track, p);
		if (*q == '\0')
			break;
		p = q;
	}
	return 0;
}

/**
 * \brief Process CodecPrivate section of subtitle stream
 * \param track track
 * \param data string to parse
 * \param size length of data
 CodecPrivate section contains [Stream Info] and [V4+ Styles] ([V4 Styles] for SSA) sections
*/ 
void ass_process_codec_private(ass_track_t* track, char *data, int size)
{
	char* str = malloc(size + 1);

	memcpy(str, data, size);
	str[size] = '\0';

	process_text(track, str);
	free(str);

	if (!track->event_format) {
		// probably an mkv produced by ancient mkvtoolnix
		// such files don't have [Events] and Format: headers
		track->parser_priv->state = PST_EVENTS;
		if (track->track_type == TRACK_TYPE_SSA)
			track->event_format = strdup("Format: Marked, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text");
		else
			track->event_format = strdup("Format: Layer, Start, End, Style, Actor, MarginL, MarginR, MarginV, Effect, Text");
	}

	process_force_style(track);
}

static int check_duplicate_event(ass_track_t* track, int ReadOrder)
{
	int i;
	for (i = 0; i<track->n_events - 1; ++i) // ignoring last event, it is the one we are comparing with
		if (track->events[i].ReadOrder == ReadOrder)
			return 1;
	return 0;
}

/**
 * \brief Process a chunk of subtitle stream data. In matroska, this containes exactly 1 event (or a commentary)
 * \param track track
 * \param data string to parse
 * \param size length of data
 * \param timecode starting time of the event (milliseconds)
 * \param duration duration of the event (milliseconds)
*/ 
void ass_process_chunk(ass_track_t* track, char *data, int size, long long timecode, long long duration)
{
	char* str;
	int eid;
	char* p;
	char* token;
	ass_event_t* event;

	if (!track->event_format) {
		mp_msg(MSGT_ASS, MSGL_WARN, MSGTR_LIBASS_EventFormatHeaderMissing);
		return;
	}
	
	str = malloc(size + 1);
	memcpy(str, data, size);
	str[size] = '\0';
	mp_msg(MSGT_ASS, MSGL_V, "event at %" PRId64 ", +%" PRId64 ": %s  \n", (int64_t)timecode, (int64_t)duration, str);

	eid = ass_alloc_event(track);
	event = track->events + eid;

	p = str;
	
	do { 
		NEXT(p, token);
		event->ReadOrder = atoi(token);
		if (check_duplicate_event(track, event->ReadOrder))
			break;

		NEXT(p, token);
		event->Layer = atoi(token);

		process_event_tail(track, event, p, 3);

		event->Start = timecode;
		event->Duration = duration;
		
		free(str);
		return;
//		dump_events(tid);
	} while (0);
	// some error
	ass_free_event(track, eid);
	track->n_events--;
	free(str);
}

#ifdef USE_ICONV
/** \brief recode buffer to utf-8
 * constraint: codepage != 0
 * \param data pointer to text buffer
 * \param size buffer size
 * \return a pointer to recoded buffer, caller is responsible for freeing it
**/
static char* sub_recode(char* data, size_t size, char* codepage)
{
	static iconv_t icdsc = (iconv_t)(-1);
	char* tocp = "UTF-8";
	char* outbuf;
	assert(codepage);

	{
		char* cp_tmp = codepage ? strdup(codepage) : 0;
#ifdef HAVE_ENCA
		char enca_lang[3], enca_fallback[100];
		if (sscanf(codepage, "enca:%2s:%99s", enca_lang, enca_fallback) == 2
				|| sscanf(codepage, "ENCA:%2s:%99s", enca_lang, enca_fallback) == 2) {
			cp_tmp = guess_buffer_cp((unsigned char*)data, size, enca_lang, enca_fallback);
		}
#endif
		if ((icdsc = iconv_open (tocp, cp_tmp)) != (iconv_t)(-1)){
			mp_msg(MSGT_ASS,MSGL_V,"LIBSUB: opened iconv descriptor.\n");
		} else
			mp_msg(MSGT_ASS,MSGL_ERR,MSGTR_LIBASS_ErrorOpeningIconvDescriptor);
#ifdef HAVE_ENCA
		if (cp_tmp) free(cp_tmp);
#endif
	}

	{
		size_t osize = size;
		size_t ileft = size;
		size_t oleft = size - 1;
		char* ip;
		char* op;
		size_t rc;
		
		outbuf = malloc(size);
		ip = data;
		op = outbuf;
		
		while (ileft) {
			rc = iconv(icdsc, &ip, &ileft, &op, &oleft);
			if (rc == (size_t)(-1)) {
				if (errno == E2BIG) {
					int offset = op - outbuf;
					outbuf = (char*)realloc(outbuf, osize + size);
					op = outbuf + offset;
					osize += size;
					oleft += size;
				} else {
					mp_msg(MSGT_ASS, MSGL_WARN, MSGTR_LIBASS_ErrorRecodingFile);
					return NULL;
				}
			}
		}
		outbuf[osize - oleft - 1] = 0;
	}

	if (icdsc != (iconv_t)(-1)) {
		(void)iconv_close(icdsc);
		icdsc = (iconv_t)(-1);
		mp_msg(MSGT_ASS,MSGL_V,"LIBSUB: closed iconv descriptor.\n");
	}
	
	return outbuf;
}
#endif // ICONV

/**
 * \brief read file contents into newly allocated buffer
 * \param fname file name
 * \param bufsize out: file size
 * \return pointer to file contents. Caller is responsible for its deallocation.
 */
static char* read_file(char* fname, size_t *bufsize)
{
	int res;
	long sz;
	long bytes_read;
	char* buf;

	FILE* fp = fopen(fname, "rb");
	if (!fp) {
		mp_msg(MSGT_ASS, MSGL_WARN, MSGTR_LIBASS_FopenFailed, fname);
		return 0;
	}
	res = fseek(fp, 0, SEEK_END);
	if (res == -1) {
		mp_msg(MSGT_ASS, MSGL_WARN, MSGTR_LIBASS_FseekFailed, fname);
		fclose(fp);
		return 0;
	}
	
	sz = ftell(fp);
	rewind(fp);

	if (sz > 10*1024*1024) {
		mp_msg(MSGT_ASS, MSGL_INFO, MSGTR_LIBASS_RefusingToLoadSubtitlesLargerThan10M, fname);
		fclose(fp);
		return 0;
	}
	
	mp_msg(MSGT_ASS, MSGL_V, "file size: %ld\n", sz);
	
	buf = malloc(sz + 1);
	assert(buf);
	bytes_read = 0;
	do {
		res = fread(buf + bytes_read, 1, sz - bytes_read, fp);
		if (res <= 0) {
			mp_msg(MSGT_ASS, MSGL_INFO, MSGTR_LIBASS_ReadFailed, errno, strerror(errno));
			fclose(fp);
			free(buf);
			return 0;
		}
		bytes_read += res;
	} while (sz - bytes_read > 0);
	buf[sz] = '\0';
	fclose(fp);
	
	if (bufsize)
		*bufsize = sz;
	return buf;
}

/*
 * \param buf pointer to subtitle text in utf-8
 */
static ass_track_t* parse_memory(ass_library_t* library, char* buf)
{
	ass_track_t* track;
	int i;
	
	track = ass_new_track(library);
	
	// process header
	process_text(track, buf);

	// external SSA/ASS subs does not have ReadOrder field
	for (i = 0; i < track->n_events; ++i)
		track->events[i].ReadOrder = i;

	// there is no explicit end-of-font marker in ssa/ass
	if (track->parser_priv->fontname)
		decode_font(track);

	if (track->track_type == TRACK_TYPE_UNKNOWN) {
		ass_free_track(track);
		return 0;
	}

	process_force_style(track);

	return track;
}

/**
 * \brief Read subtitles from memory.
 * \param library libass library object
 * \param buf pointer to subtitles text
 * \param bufsize size of buffer
 * \param codepage recode buffer contents from given codepage
 * \return newly allocated track
*/ 
ass_track_t* ass_read_memory(ass_library_t* library, char* buf, size_t bufsize, char* codepage)
{
	ass_track_t* track;
	int need_free = 0;
	
	if (!buf)
		return 0;
	
#ifdef USE_ICONV
	if (codepage)
		buf = sub_recode(buf, bufsize, codepage);
	if (!buf)
		return 0;
	else
		need_free = 1;
#endif
	track = parse_memory(library, buf);
	if (need_free)
		free(buf);
	if (!track)
		return 0;

	mp_msg(MSGT_ASS, MSGL_INFO, MSGTR_LIBASS_AddedSubtitleFileMemory, track->n_styles, track->n_events);
	return track;
}

char* read_file_recode(char* fname, char* codepage, int* size)
{
	char* buf;
	size_t bufsize;
	
	buf = read_file(fname, &bufsize);
	if (!buf)
		return 0;
#ifdef USE_ICONV
	if (codepage) {
		 char* tmpbuf = sub_recode(buf, bufsize, codepage);
		 free(buf);
		 buf = tmpbuf;
	}
	if (!buf)
		return 0;
#endif
	*size = bufsize;
	return buf;
}

/**
 * \brief Read subtitles from file.
 * \param library libass library object
 * \param fname file name
 * \param codepage recode buffer contents from given codepage
 * \return newly allocated track
*/ 
ass_track_t* ass_read_file(ass_library_t* library, char* fname, char* codepage)
{
	char* buf;
	ass_track_t* track;
	size_t bufsize;

	buf = read_file_recode(fname, codepage, &bufsize);
	if (!buf)
		return 0;
	track = parse_memory(library, buf);
	free(buf);
	if (!track)
		return 0;
	
	track->name = strdup(fname);

	mp_msg(MSGT_ASS, MSGL_INFO, MSGTR_LIBASS_AddedSubtitleFileFname, fname, track->n_styles, track->n_events);
	
//	dump_events(forced_tid);
	return track;
}

/**
 * \brief read styles from file into already initialized track
 */
int ass_read_styles(ass_track_t* track, char* fname, char* codepage)
{
	char* buf;
	parser_state_t old_state;
	size_t sz;

	buf = read_file(fname, &sz);
	if (!buf)
		return 1;
#ifdef USE_ICONV
	if (codepage) {
		char* tmpbuf;
		tmpbuf = sub_recode(buf, sz, codepage);
		free(buf);
		buf = tmpbuf;
	}
	if (!buf)
		return 0;
#endif

	old_state = track->parser_priv->state;
	track->parser_priv->state = PST_STYLES;
	process_text(track, buf);
	track->parser_priv->state = old_state;

	return 0;
}

long long ass_step_sub(ass_track_t* track, long long now, int movement) {
	int i;

	if (movement == 0) return 0;
	if (track->n_events == 0) return 0;
	
	if (movement < 0)
		for (i = 0; (i < track->n_events) && ((long long)(track->events[i].Start + track->events[i].Duration) <= now); ++i) {}
	else
		for (i = track->n_events - 1; (i >= 0) && ((long long)(track->events[i].Start) > now); --i) {}
	
	// -1 and n_events are ok
	assert(i >= -1); assert(i <= track->n_events);
	i += movement;
	if (i < 0) i = 0;
	if (i >= track->n_events) i = track->n_events - 1;
	return ((long long)track->events[i].Start) - now;
}

ass_track_t* ass_new_track(ass_library_t* library) {
	ass_track_t* track = calloc(1, sizeof(ass_track_t));
	track->library = library;
	track->parser_priv = calloc(1, sizeof(parser_priv_t));
	return track;
}