text_import.c

一个用于智能手机的多媒体库适合S60 WinCE的跨平台开发库

			break;
		case 1:
			if (sscanf(szLine, "%d:%d:%d,%d --> %d:%d:%d,%d", &sh, &sm, &ss, &sms, &eh, &em, &es, &ems) != 8) {
				e = gf_import_message(import, GF_CORRUPTED_DATA, "Error scanning SRT frame %d timing", curLine);
				goto exit;
			}
			start = (3600*sh + 60*sm + ss)*1000 + sms;
			if (start<end) {
				gf_import_message(import, GF_OK, "WARNING: corrupted SRT frame %d - starts (at %d ms) before end of previous one (%d ms) - adjusting time stamps", curLine, start, end);
				start = end;
			}

			end = (3600*eh + 60*em + es)*1000 + ems;
			/*make stream start at 0 by inserting a fake AU*/
			if (first_samp && (start>0)) {
				s = gf_isom_text_to_sample(samp);
				s->DTS = 0;
				gf_isom_add_sample(import->dest, track, 1, s);
				gf_isom_sample_del(&s);
				nb_samp++;
			}
			rec.style_flags = 0;
			state = 2;			
			break;

		default:
			/*reset only when text is present*/
			first_samp = 0;

			/*go to line*/
			if (txt_line) {
				gf_isom_text_add_text(samp, "\n", 1);
				char_len += 1;
			}

			ptr = (char *) szLine;
			len = gf_utf8_mbstowcs(uniLine, 5000, (const char **) &ptr);
			if (len == (u32) -1) {
				e = gf_import_message(import, GF_CORRUPTED_DATA, "Invalid UTF data (line %d)", curLine);
				goto exit;
			}
			char_line = 0;
			i=j=0;
			rem_styles = 0;
			while (i<len) {
				/*start of new style*/
				if ( (uniLine[i]=='<') && (uniLine[i+2]=='>')) {
					/*store prev style*/
					if (set_end_char) {
						assert(set_start_char);
						gf_isom_text_add_style(samp, &rec);
						set_end_char = set_start_char = 0;
						rec.style_flags &= ~rem_styles;
						rem_styles = 0;
					}
					if (set_start_char && (rec.startCharOffset != j)) {
						rec.endCharOffset = char_len + j;
						if (rec.style_flags) gf_isom_text_add_style(samp, &rec);
					}
					switch (uniLine[i+1]) {
					case 'b': case 'B': 
						rec.style_flags |= GF_TXT_STYLE_BOLD; 
						set_start_char = 1;
						rec.startCharOffset = char_len + j;
						break;
					case 'i': case 'I': 
						rec.style_flags |= GF_TXT_STYLE_ITALIC; 
						set_start_char = 1;
						rec.startCharOffset = char_len + j;
						break;
					case 'u': case 'U': 
						rec.style_flags |= GF_TXT_STYLE_UNDERLINED; 
						set_start_char = 1;
						rec.startCharOffset = char_len + j;
						break;
					}
					i+=3;
					continue;
				}

				/*end of prev style*/
				if ( (uniLine[i]=='<') && (uniLine[i+1]=='/') && (uniLine[i+3]=='>')) {
					switch (uniLine[i+2]) {
					case 'b': case 'B': 
						rem_styles |= GF_TXT_STYLE_BOLD; 
						set_end_char = 1;
						rec.endCharOffset = char_len + j;
						break;
					case 'i': case 'I': 
						rem_styles |= GF_TXT_STYLE_ITALIC; 
						set_end_char = 1;
						rec.endCharOffset = char_len + j;
						break;
					case 'u': case 'U': 
						rem_styles |= GF_TXT_STYLE_UNDERLINED; 
						set_end_char = 1;
						rec.endCharOffset = char_len + j;
						break;
					}
					i+=4;
					continue;
				}
				/*store style*/
				if (set_end_char) {
					gf_isom_text_add_style(samp, &rec);
					set_end_char = 0;
					set_start_char = 1;
					rec.startCharOffset = char_len + j;
					rec.style_flags &= ~rem_styles;
					rem_styles = 0;
				}

				uniText[j] = uniLine[i];
				j++;
				i++;
			}
			/*store last style*/
			if (set_end_char) {
				gf_isom_text_add_style(samp, &rec);
				set_end_char = 0;
				set_start_char = 1;
				rec.startCharOffset = char_len + j;
				rec.style_flags &= ~rem_styles;
				rem_styles = 0;
			}

			char_line = j;
			uniText[j] = 0;

			sptr = (u16 *) uniText;
			len = gf_utf8_wcstombs(szText, 5000, (const u16 **) &sptr);

			gf_isom_text_add_text(samp, szText, len);
			char_len += char_line;
			txt_line ++;
			break;
		}
		if (duration && (start >= duration)) break;
	}

	/*final flush*/	
	if (end) {
		gf_isom_text_reset(samp);
		s = gf_isom_text_to_sample(samp);
		s->DTS = (u64) (scale*(s64)end);
		s->IsRAP = 1;
		gf_isom_add_sample(import->dest, track, 1, s);
		gf_isom_sample_del(&s);
		nb_samp++;
	}
	gf_isom_delete_text_sample(samp);
	gf_isom_set_last_sample_duration(import->dest, track, 0);
	gf_set_progress("Importing SRT", nb_samp, nb_samp);

exit:
	if (e) gf_isom_remove_track(import->dest, track);
	fclose(srt_in);
	return e;
}


static GF_Err gf_text_import_sub(GF_MediaImporter *import)
{
	FILE *sub_in;
	u32 track, ID, timescale, i, j, desc_idx, start, end, prev_end, nb_samp, duration, file_size, len, line;
	GF_TextConfig*cfg;
	GF_Err e;
	Double FPS;
	GF_TextSample * samp;
	Bool first_samp;
	s32 unicode_type;
	char szLine[2048], szTime[20], szText[2048];
	GF_ISOSample *s;

	sub_in = fopen(import->in_name, "rt");
	fseek(sub_in, 0, SEEK_END);
	file_size = ftell(sub_in);
	fseek(sub_in, 0, SEEK_SET);
	
	unicode_type = gf_text_get_utf_type(sub_in);
	if (unicode_type<0) {
		fclose(sub_in);
		return gf_import_message(import, GF_NOT_SUPPORTED, "Unsupported SUB UTF encoding");
	}

	FPS = 25.0;
	if (import->video_fps) FPS = import->video_fps;

	cfg = NULL;
	if (import->esd) {
		if (!import->esd->slConfig) {
			import->esd->slConfig = (GF_SLConfig *) gf_odf_desc_new(GF_ODF_SLC_TAG);
			import->esd->slConfig->predefined = 2;
			import->esd->slConfig->timestampResolution = 1000;
		}
		timescale = import->esd->slConfig->timestampResolution;
		if (!timescale) timescale = 1000;

		/*explicit text config*/
		if (import->esd->decoderConfig && import->esd->decoderConfig->decoderSpecificInfo->tag == GF_ODF_TEXT_CFG_TAG) {
			cfg = (GF_TextConfig *) import->esd->decoderConfig->decoderSpecificInfo;
			import->esd->decoderConfig->decoderSpecificInfo = NULL;
		}
		ID = import->esd->ESID;
	} else {
		timescale = 1000;
		ID = 0;
	}
	
	if (cfg && cfg->timescale) timescale = cfg->timescale;
	track = gf_isom_new_track(import->dest, ID, GF_ISOM_MEDIA_TEXT, timescale);
	if (!track) {
		fclose(sub_in);
		return gf_import_message(import, gf_isom_last_error(import->dest), "Error creating text track");
	}
	gf_isom_set_track_enabled(import->dest, track, 1);
	if (import->esd && !import->esd->ESID) import->esd->ESID = gf_isom_get_track_id(import->dest, track);

	gf_text_import_set_language(import, track);

	file_size = 0;
	/*setup track*/
	if (cfg) {
		u32 count;
		char *firstFont = NULL;
		/*set track info*/
		gf_isom_set_track_layout_info(import->dest, track, cfg->text_width<<16, cfg->text_height<<16, 0, 0, cfg->layer);

		/*and set sample descriptions*/
		count = gf_list_count(cfg->sample_descriptions);
		for (i=0; i<count; i++) {
			GF_TextSampleDescriptor *sd= (GF_TextSampleDescriptor *)gf_list_get(cfg->sample_descriptions, i);
			if (!sd->font_count) {
				sd->fonts = (GF_FontRecord*)malloc(sizeof(GF_FontRecord));
				sd->font_count = 1;
				sd->fonts[0].fontID = 1;
				sd->fonts[0].fontName = strdup("Serif");
			}
			if (!sd->default_style.fontID) sd->default_style.fontID = sd->fonts[0].fontID;
			if (!sd->default_style.font_size) sd->default_style.font_size = 16;
			if (!sd->default_style.text_color) sd->default_style.text_color = 0xFF000000;
			file_size = sd->default_style.font_size;
			gf_isom_new_text_description(import->dest, track, sd, NULL, NULL, &desc_idx);
			if (!firstFont) firstFont = sd->fonts[0].fontName;
		}
		gf_import_message(import, GF_OK, "Timed Text (SUB @ %02.2f) import - text track %d x %d, font %s (size %d)", FPS, cfg->text_width, cfg->text_height, firstFont, file_size);

		gf_odf_desc_del((GF_Descriptor *)cfg);
	} else {
		u32 w, h;
		GF_TextSampleDescriptor *sd;
		gf_text_get_video_size(import->dest, &w, &h);

		/*have to work with default - use max size (if only one video, this means the text region is the
		entire display, and with bottom alignment things should be fine...*/
		gf_isom_set_track_layout_info(import->dest, track, w<<16, h<<16, 0, 0, 0);
		sd = (GF_TextSampleDescriptor*)gf_odf_desc_new(GF_ODF_TX3G_TAG);
		sd->fonts = (GF_FontRecord*)malloc(sizeof(GF_FontRecord));
		sd->font_count = 1;
		sd->fonts[0].fontID = 1;
		sd->fonts[0].fontName = strdup("Serif");
		sd->back_color = 0x00000000;	/*transparent*/
		sd->default_style.fontID = 1;
		sd->default_style.font_size = TTXT_DEFAULT_FONT_SIZE;
		sd->default_style.text_color = 0xFFFFFFFF;	/*white*/
		sd->default_style.style_flags = 0;
		sd->horiz_justif = 1; /*center of scene*/
		sd->vert_justif = (s8) -1;	/*bottom of scene*/

		if (import->flags & GF_IMPORT_SKIP_TXT_BOX) {
			sd->default_pos.top = sd->default_pos.left = sd->default_pos.right = sd->default_pos.bottom = 0;
		} else {
			if ((sd->default_pos.bottom==sd->default_pos.top) || (sd->default_pos.right==sd->default_pos.left)) {
				sd->default_pos.top = sd->default_pos.left = 0;
				sd->default_pos.right = w;
				sd->default_pos.bottom = h;
			}
		}

		gf_isom_new_text_description(import->dest, track, sd, NULL, NULL, &desc_idx);
		gf_import_message(import, GF_OK, "Timed Text (SUB @ %02.2f) import - text track %d x %d, font %s (size %d)", FPS, w, h, sd->fonts[0].fontName, TTXT_DEFAULT_FONT_SIZE);
		gf_odf_desc_del((GF_Descriptor *)sd);
	}

	duration = (u32) (((Double) import->duration)*timescale/1000.0);

	e = GF_OK;
	nb_samp = 0;
	samp = gf_isom_new_text_sample();

	FPS = ((Double) timescale ) / FPS;
	start = end = prev_end = 0;

	line = 0;
	first_samp = 1;
	while (1) {
		char *sOK = gf_text_get_utf8_line(szLine, 2048, sub_in, unicode_type);
		if (!sOK) break;

		REM_TRAIL_MARKS(szLine, "\r\n\t ")

		line++;
		len = strlen(szLine); 
		if (!len) continue;

		i=0;
		if (szLine[i] != '{') {
			e = gf_import_message(import, GF_NON_COMPLIANT_BITSTREAM, "Bad SUB file (line %d): expecting \"{\" got \"%c\"", line, szLine[i]);
			goto exit;
		}
		while (szLine[i+1] && szLine[i+1]!='}') { szTime[i] = szLine[i+1]; i++; }
		szTime[i] = 0;
		start = atoi(szTime);
		if (start<end) {
			gf_import_message(import, GF_OK, "WARNING: corrupted SUB frame (line %d) - starts (at %d ms) before end of previous one (%d ms) - adjusting time stamps", line, start, end);
			start = end;
		}
		j=i+2;
		i=0;
		if (szLine[i+j] != '{') {
			e = gf_import_message(import, GF_NON_COMPLIANT_BITSTREAM, "Bad SUB file - expecting \"{\" got \"%c\"", szLine[i]);
			goto exit;
		}
		while (szLine[i+1+j] && szLine[i+1+j]!='}') { szTime[i] = szLine[i+1+j]; i++; }
		szTime[i] = 0;
		end = atoi(szTime);
		j+=i+2;

		if (start>end) {
			gf_import_message(import, GF_OK, "WARNING: corrupted SUB frame (line %d) - ends (at %d ms) before start of current frame (%d ms) - skipping", line, end, start);
			continue;
		}

		gf_isom_text_reset(samp);

		if (start && first_samp) {
			s = gf_isom_text_to_sample(samp);
			s->DTS = 0;
			s->IsRAP = 1;
			gf_isom_add_sample(import->dest, track, 1, s);
			gf_isom_sample_del(&s);
			first_samp = 0;
			nb_samp++;
		}

		for (i=j; i<len; i++) {
			if (szLine[i]=='|') {
				szText[i-j] = '\n';
			} else {
				szText[i-j] = szLine[i];
			}
		}
		szText[i-j] = 0;
		gf_isom_text_add_text(samp, szText, strlen(szText) );

		if (prev_end) {
			GF_TextSample * empty_samp = gf_isom_new_text_sample();
			s = gf_isom_text_to_sample(empty_samp);
			s->DTS = (u64) (FPS*(s64)prev_end);
			gf_isom_add_sample(import->dest, track, 1, s);
			gf_isom_sample_del(&s);
			nb_samp++;
			gf_isom_delete_text_sample(empty_samp);
		}

		s = gf_isom_text_to_sample(samp);
		s->DTS = (u64) (FPS*(s64)start);
		gf_isom_add_sample(import->dest, track, 1, s);
		gf_isom_sample_del(&s);
		nb_samp++;
		gf_isom_text_reset(samp);
		prev_end = end;
		gf_set_progress("Importing SUB", ftell(sub_in), file_size);
		if (duration && (end >= duration)) break;
	}
	/*final flush*/
	if (end) {
		gf_isom_text_reset(samp);
		s = gf_isom_text_to_sample(samp);
		s->DTS = (u64)(FPS*(s64)end);
		gf_isom_add_sample(import->dest, track, 1, s);
		gf_isom_sample_del(&s);
		nb_samp++;
	}
	gf_isom_delete_text_sample(samp);
	
	gf_isom_set_last_sample_duration(import->dest, track, 0);
	gf_set_progress("Importing SUB", nb_samp, nb_samp);

exit:
	if (e) gf_isom_remove_track(import->dest, track);
	fclose(sub_in);
	return e;
}


#define CHECK_STR(__str)	\
	if (!__str) { \
		e = gf_import_message(import, GF_BAD_PARAM, "Invalid XML formatting (line %d)", parser.line);	\
		goto exit;	\
	}	\


u32 ttxt_get_color(GF_MediaImporter *import, char *val)
{
	u32 r, g, b, a, res;
	r = g = b = a = 0;
	if (sscanf(val, "%x %x %x %x", &r, &g, &b, &a) != 4) {
		gf_import_message(import, GF_OK, "Warning: color badly formatted");
	}
	res = (a&0xFF); res<<=8;
	res |= (r&0xFF); res<<=8;
	res |= (g&0xFF); res<<=8;
	res |= (b&0xFF);
	return res;
}

void ttxt_parse_text_box(GF_MediaImporter *import, GF_XMLNode *n, GF_BoxRecord *box)
{
	u32 i=0;
	GF_XMLAttribute *att;
	memset(box, 0, sizeof(GF_BoxRecord));
	while ( (att=(GF_XMLAttribute *)gf_list_enum(n->attributes, &i))) {
		if (!stricmp(att->name, "top")) box->top = atoi(att->value);
		else if (!stricmp(att->name, "bottom")) box->bottom = atoi(att->value);
		else if (!stricmp(att->name, "left")) box->left = atoi(att->value);