2 * Copyright (C) <1999> Erik Walthinsen <omega@cse.ogi.edu>
3 * Copyright (C) 2004 Ronald S. Bultje <rbultje@ronald.bitfreak.net>
4 * Copyright (C) 2006 Tim-Philipp Müller <tim centricular net>
6 * This library is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Library General Public
8 * License as published by the Free Software Foundation; either
9 * version 2 of the License, or (at your option) any later version.
11 * This library is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Library General Public License for more details.
16 * You should have received a copy of the GNU Library General Public
17 * License along with this library; if not, write to the
18 * Free Software Foundation, Inc., 51 Franklin St, Fifth Floor,
19 * Boston, MA 02110-1301, USA.
29 #include <sys/types.h>
32 #include "gstsubparse.h"
33 #include "gstssaparse.h"
34 #include "samiparse.h"
35 #include "tmplayerparse.h"
36 #include "mpl2parse.h"
37 #include "qttextparse.h"
39 GST_DEBUG_CATEGORY (sub_parse_debug);
41 #define DEFAULT_ENCODING NULL
51 gst_sub_parse_set_property (GObject * object, guint prop_id,
52 const GValue * value, GParamSpec * pspec);
54 gst_sub_parse_get_property (GObject * object, guint prop_id,
55 GValue * value, GParamSpec * pspec);
58 #ifndef GST_DISABLE_XML
59 static GstStaticPadTemplate sink_templ = GST_STATIC_PAD_TEMPLATE ("sink",
62 GST_STATIC_CAPS ("application/x-subtitle; application/x-subtitle-sami; "
63 "application/x-subtitle-tmplayer; application/x-subtitle-mpl2; "
64 "application/x-subtitle-dks; application/x-subtitle-qttext")
67 static GstStaticPadTemplate sink_templ = GST_STATIC_PAD_TEMPLATE ("sink",
70 GST_STATIC_CAPS ("application/x-subtitle; application/x-subtitle-dks; "
71 "application/x-subtitle-tmplayer; application/x-subtitle-mpl2; "
72 "application/x-subtitle-qttext")
76 static GstStaticPadTemplate src_templ = GST_STATIC_PAD_TEMPLATE ("src",
79 GST_STATIC_CAPS ("text/x-raw, format= { pango-markup, utf8 }")
83 static gboolean gst_sub_parse_src_event (GstPad * pad, GstObject * parent,
85 static gboolean gst_sub_parse_src_query (GstPad * pad, GstObject * parent,
87 static gboolean gst_sub_parse_sink_event (GstPad * pad, GstObject * parent,
90 static GstStateChangeReturn gst_sub_parse_change_state (GstElement * element,
91 GstStateChange transition);
93 static GstFlowReturn gst_sub_parse_chain (GstPad * sinkpad, GstObject * parent,
96 #define gst_sub_parse_parent_class parent_class
97 G_DEFINE_TYPE (GstSubParse, gst_sub_parse, GST_TYPE_ELEMENT);
100 gst_sub_parse_dispose (GObject * object)
102 GstSubParse *subparse = GST_SUBPARSE (object);
104 GST_DEBUG_OBJECT (subparse, "cleaning up subtitle parser");
106 switch (subparse->parser_type) {
107 case GST_SUB_PARSE_FORMAT_QTTEXT:
108 qttext_context_deinit (&subparse->state);
110 #ifndef GST_DISABLE_XML
111 case GST_SUB_PARSE_FORMAT_SAMI:
112 sami_context_deinit (&subparse->state);
119 if (subparse->encoding) {
120 g_free (subparse->encoding);
121 subparse->encoding = NULL;
124 if (subparse->detected_encoding) {
125 g_free (subparse->detected_encoding);
126 subparse->detected_encoding = NULL;
129 if (subparse->adapter) {
130 g_object_unref (subparse->adapter);
131 subparse->adapter = NULL;
134 if (subparse->textbuf) {
135 g_string_free (subparse->textbuf, TRUE);
136 subparse->textbuf = NULL;
139 GST_CALL_PARENT (G_OBJECT_CLASS, dispose, (object));
143 gst_sub_parse_class_init (GstSubParseClass * klass)
145 GObjectClass *object_class = G_OBJECT_CLASS (klass);
146 GstElementClass *element_class = GST_ELEMENT_CLASS (klass);
148 object_class->dispose = gst_sub_parse_dispose;
149 object_class->set_property = gst_sub_parse_set_property;
150 object_class->get_property = gst_sub_parse_get_property;
152 gst_element_class_add_pad_template (element_class,
153 gst_static_pad_template_get (&sink_templ));
154 gst_element_class_add_pad_template (element_class,
155 gst_static_pad_template_get (&src_templ));
156 gst_element_class_set_static_metadata (element_class,
157 "Subtitle parser", "Codec/Parser/Subtitle",
158 "Parses subtitle (.sub) files into text streams",
159 "Gustavo J. A. M. Carneiro <gjc@inescporto.pt>, "
160 "GStreamer maintainers <gstreamer-devel@lists.sourceforge.net>");
162 element_class->change_state = gst_sub_parse_change_state;
164 g_object_class_install_property (object_class, PROP_ENCODING,
165 g_param_spec_string ("subtitle-encoding", "subtitle charset encoding",
166 "Encoding to assume if input subtitles are not in UTF-8 or any other "
167 "Unicode encoding. If not set, the GST_SUBTITLE_ENCODING environment "
168 "variable will be checked for an encoding to use. If that is not set "
169 "either, ISO-8859-15 will be assumed.", DEFAULT_ENCODING,
170 G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
172 g_object_class_install_property (object_class, PROP_VIDEOFPS,
173 gst_param_spec_fraction ("video-fps", "Video framerate",
174 "Framerate of the video stream. This is needed by some subtitle "
175 "formats to synchronize subtitles and video properly. If not set "
176 "and the subtitle format requires it subtitles may be out of sync.",
177 0, 1, G_MAXINT, 1, 24000, 1001,
178 G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
182 gst_sub_parse_init (GstSubParse * subparse)
184 subparse->sinkpad = gst_pad_new_from_static_template (&sink_templ, "sink");
185 gst_pad_set_chain_function (subparse->sinkpad,
186 GST_DEBUG_FUNCPTR (gst_sub_parse_chain));
187 gst_pad_set_event_function (subparse->sinkpad,
188 GST_DEBUG_FUNCPTR (gst_sub_parse_sink_event));
189 gst_element_add_pad (GST_ELEMENT (subparse), subparse->sinkpad);
191 subparse->srcpad = gst_pad_new_from_static_template (&src_templ, "src");
192 gst_pad_set_event_function (subparse->srcpad,
193 GST_DEBUG_FUNCPTR (gst_sub_parse_src_event));
194 gst_pad_set_query_function (subparse->srcpad,
195 GST_DEBUG_FUNCPTR (gst_sub_parse_src_query));
196 gst_element_add_pad (GST_ELEMENT (subparse), subparse->srcpad);
198 subparse->textbuf = g_string_new (NULL);
199 subparse->parser_type = GST_SUB_PARSE_FORMAT_UNKNOWN;
200 subparse->flushing = FALSE;
201 gst_segment_init (&subparse->segment, GST_FORMAT_TIME);
202 subparse->need_segment = TRUE;
203 subparse->encoding = g_strdup (DEFAULT_ENCODING);
204 subparse->detected_encoding = NULL;
205 subparse->adapter = gst_adapter_new ();
207 subparse->fps_n = 24000;
208 subparse->fps_d = 1001;
212 * Source pad functions.
216 gst_sub_parse_src_query (GstPad * pad, GstObject * parent, GstQuery * query)
218 GstSubParse *self = GST_SUBPARSE (parent);
219 gboolean ret = FALSE;
221 GST_DEBUG ("Handling %s query", GST_QUERY_TYPE_NAME (query));
223 switch (GST_QUERY_TYPE (query)) {
224 case GST_QUERY_POSITION:{
227 gst_query_parse_position (query, &fmt, NULL);
228 if (fmt != GST_FORMAT_TIME) {
229 ret = gst_pad_peer_query (self->sinkpad, query);
232 gst_query_set_position (query, GST_FORMAT_TIME, self->segment.position);
236 case GST_QUERY_SEEKING:
239 gboolean seekable = FALSE;
243 gst_query_parse_seeking (query, &fmt, NULL, NULL, NULL);
244 if (fmt == GST_FORMAT_TIME) {
245 GstQuery *peerquery = gst_query_new_seeking (GST_FORMAT_BYTES);
247 seekable = gst_pad_peer_query (self->sinkpad, peerquery);
249 gst_query_parse_seeking (peerquery, NULL, &seekable, NULL, NULL);
250 gst_query_unref (peerquery);
253 gst_query_set_seeking (query, fmt, seekable, seekable ? 0 : -1, -1);
257 ret = gst_pad_query_default (pad, parent, query);
265 gst_sub_parse_src_event (GstPad * pad, GstObject * parent, GstEvent * event)
267 GstSubParse *self = GST_SUBPARSE (parent);
268 gboolean ret = FALSE;
270 GST_DEBUG ("Handling %s event", GST_EVENT_TYPE_NAME (event));
272 switch (GST_EVENT_TYPE (event)) {
277 GstSeekType start_type, stop_type;
282 gst_event_parse_seek (event, &rate, &format, &flags,
283 &start_type, &start, &stop_type, &stop);
285 if (format != GST_FORMAT_TIME) {
286 GST_WARNING_OBJECT (self, "we only support seeking in TIME format");
287 gst_event_unref (event);
291 /* Convert that seek to a seeking in bytes at position 0,
292 FIXME: could use an index */
293 ret = gst_pad_push_event (self->sinkpad,
294 gst_event_new_seek (rate, GST_FORMAT_BYTES, flags,
295 GST_SEEK_TYPE_SET, 0, GST_SEEK_TYPE_NONE, 0));
298 /* Apply the seek to our segment */
299 gst_segment_do_seek (&self->segment, rate, format, flags,
300 start_type, start, stop_type, stop, &update);
302 GST_DEBUG_OBJECT (self, "segment after seek: %" GST_SEGMENT_FORMAT,
305 self->need_segment = TRUE;
307 GST_WARNING_OBJECT (self, "seek to 0 bytes failed");
310 gst_event_unref (event);
314 ret = gst_pad_event_default (pad, parent, event);
323 gst_sub_parse_set_property (GObject * object, guint prop_id,
324 const GValue * value, GParamSpec * pspec)
326 GstSubParse *subparse = GST_SUBPARSE (object);
328 GST_OBJECT_LOCK (subparse);
331 g_free (subparse->encoding);
332 subparse->encoding = g_value_dup_string (value);
333 GST_LOG_OBJECT (object, "subtitle encoding set to %s",
334 GST_STR_NULL (subparse->encoding));
338 subparse->fps_n = gst_value_get_fraction_numerator (value);
339 subparse->fps_d = gst_value_get_fraction_denominator (value);
340 GST_DEBUG_OBJECT (object, "video framerate set to %d/%d", subparse->fps_n,
343 if (!subparse->state.have_internal_fps) {
344 subparse->state.fps_n = subparse->fps_n;
345 subparse->state.fps_d = subparse->fps_d;
350 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
353 GST_OBJECT_UNLOCK (subparse);
357 gst_sub_parse_get_property (GObject * object, guint prop_id,
358 GValue * value, GParamSpec * pspec)
360 GstSubParse *subparse = GST_SUBPARSE (object);
362 GST_OBJECT_LOCK (subparse);
365 g_value_set_string (value, subparse->encoding);
368 gst_value_set_fraction (value, subparse->fps_n, subparse->fps_d);
371 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
374 GST_OBJECT_UNLOCK (subparse);
378 gst_sub_parse_get_format_description (GstSubParseFormat format)
381 case GST_SUB_PARSE_FORMAT_MDVDSUB:
383 case GST_SUB_PARSE_FORMAT_SUBRIP:
385 case GST_SUB_PARSE_FORMAT_MPSUB:
387 case GST_SUB_PARSE_FORMAT_SAMI:
389 case GST_SUB_PARSE_FORMAT_TMPLAYER:
391 case GST_SUB_PARSE_FORMAT_MPL2:
393 case GST_SUB_PARSE_FORMAT_SUBVIEWER:
395 case GST_SUB_PARSE_FORMAT_DKS:
397 case GST_SUB_PARSE_FORMAT_QTTEXT:
400 case GST_SUB_PARSE_FORMAT_UNKNOWN:
407 gst_convert_to_utf8 (const gchar * str, gsize len, const gchar * encoding,
408 gsize * consumed, GError ** err)
413 /* The char cast is necessary in glib < 2.24 */
415 g_convert_with_fallback (str, len, "UTF-8", encoding, (char *) "*",
416 consumed, NULL, err);
420 /* + 3 to skip UTF-8 BOM if it was added */
422 if (len >= 3 && (guint8) ret[0] == 0xEF && (guint8) ret[1] == 0xBB
423 && (guint8) ret[2] == 0xBF)
424 g_memmove (ret, ret + 3, len + 1 - 3);
430 detect_encoding (const gchar * str, gsize len)
432 if (len >= 3 && (guint8) str[0] == 0xEF && (guint8) str[1] == 0xBB
433 && (guint8) str[2] == 0xBF)
434 return g_strdup ("UTF-8");
436 if (len >= 2 && (guint8) str[0] == 0xFE && (guint8) str[1] == 0xFF)
437 return g_strdup ("UTF-16BE");
439 if (len >= 2 && (guint8) str[0] == 0xFF && (guint8) str[1] == 0xFE)
440 return g_strdup ("UTF-16LE");
442 if (len >= 4 && (guint8) str[0] == 0x00 && (guint8) str[1] == 0x00
443 && (guint8) str[2] == 0xFE && (guint8) str[3] == 0xFF)
444 return g_strdup ("UTF-32BE");
446 if (len >= 4 && (guint8) str[0] == 0xFF && (guint8) str[1] == 0xFE
447 && (guint8) str[2] == 0x00 && (guint8) str[3] == 0x00)
448 return g_strdup ("UTF-32LE");
454 convert_encoding (GstSubParse * self, const gchar * str, gsize len,
457 const gchar *encoding;
463 /* First try any detected encoding */
464 if (self->detected_encoding) {
466 gst_convert_to_utf8 (str, len, self->detected_encoding, consumed, &err);
471 GST_WARNING_OBJECT (self, "could not convert string from '%s' to UTF-8: %s",
472 self->detected_encoding, err->message);
473 g_free (self->detected_encoding);
474 self->detected_encoding = NULL;
478 /* Otherwise check if it's UTF8 */
479 if (self->valid_utf8) {
480 if (g_utf8_validate (str, len, NULL)) {
481 GST_LOG_OBJECT (self, "valid UTF-8, no conversion needed");
483 return g_strndup (str, len);
485 GST_INFO_OBJECT (self, "invalid UTF-8!");
486 self->valid_utf8 = FALSE;
489 /* Else try fallback */
490 encoding = self->encoding;
491 if (encoding == NULL || *encoding == '\0') {
492 encoding = g_getenv ("GST_SUBTITLE_ENCODING");
494 if (encoding == NULL || *encoding == '\0') {
495 /* if local encoding is UTF-8 and no encoding specified
496 * via the environment variable, assume ISO-8859-15 */
497 if (g_get_charset (&encoding)) {
498 encoding = "ISO-8859-15";
502 ret = gst_convert_to_utf8 (str, len, encoding, consumed, &err);
505 GST_WARNING_OBJECT (self, "could not convert string from '%s' to UTF-8: %s",
506 encoding, err->message);
509 /* invalid input encoding, fall back to ISO-8859-15 (always succeeds) */
510 ret = gst_convert_to_utf8 (str, len, "ISO-8859-15", consumed, NULL);
513 GST_LOG_OBJECT (self,
514 "successfully converted %" G_GSIZE_FORMAT " characters from %s to UTF-8"
515 "%s", len, encoding, (err) ? " , using ISO-8859-15 as fallback" : "");
521 get_next_line (GstSubParse * self)
524 const char *line_end;
526 gboolean have_r = FALSE;
528 line_end = strchr (self->textbuf->str, '\n');
531 /* end-of-line not found; return for more data */
535 /* get rid of '\r' */
536 if (line_end != self->textbuf->str && *(line_end - 1) == '\r') {
541 line_len = line_end - self->textbuf->str;
542 line = g_strndup (self->textbuf->str, line_len);
543 self->textbuf = g_string_erase (self->textbuf, 0,
544 line_len + (have_r ? 2 : 1));
549 parse_mdvdsub (ParserState * state, const gchar * line)
551 const gchar *line_split;
553 guint start_frame, end_frame;
554 guint64 clip_start = 0, clip_stop = 0;
555 gboolean in_seg = FALSE;
559 /* style variables */
565 if (sscanf (line, "{%u}{%u}", &start_frame, &end_frame) != 2) {
566 g_warning ("Parse of the following line, assumed to be in microdvd .sub"
567 " format, failed:\n%s", line);
571 /* skip the {%u}{%u} part */
572 line = strchr (line, '}') + 1;
573 line = strchr (line, '}') + 1;
575 /* see if there's a first line with a framerate */
576 if (start_frame == 1 && end_frame == 1) {
577 gchar *rest, *end = NULL;
579 rest = g_strdup (line);
580 g_strdelimit (rest, ",", '.');
581 fps = g_ascii_strtod (rest, &end);
583 gst_util_double_to_fraction (fps, &state->fps_n, &state->fps_d);
584 GST_INFO ("framerate from file: %d/%d ('%s')", state->fps_n,
592 gst_util_uint64_scale (start_frame, GST_SECOND * state->fps_d,
595 gst_util_uint64_scale (end_frame - start_frame, GST_SECOND * state->fps_d,
598 /* Check our segment start/stop */
599 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
600 state->start_time, state->start_time + state->duration, &clip_start,
603 /* No need to parse that text if it's out of segment */
605 state->start_time = clip_start;
606 state->duration = clip_stop - clip_start;
611 markup = g_string_new (NULL);
616 /* parse style markup */
617 if (strncmp (line, "{y:i}", 5) == 0) {
619 line = strchr (line, '}') + 1;
621 if (strncmp (line, "{y:b}", 5) == 0) {
623 line = strchr (line, '}') + 1;
625 if (sscanf (line, "{s:%u}", &fontsize) == 1) {
626 line = strchr (line, '}') + 1;
628 /* forward slashes at beginning/end signify italics too */
629 if (g_str_has_prefix (line, "/")) {
633 if ((line_split = strchr (line, '|')))
634 line_chunk = g_markup_escape_text (line, line_split - line);
636 line_chunk = g_markup_escape_text (line, strlen (line));
638 /* Remove italics markers at end of line/stanza (CHECKME: are end slashes
639 * always at the end of a line or can they span multiple lines?) */
640 if (g_str_has_suffix (line_chunk, "/")) {
641 line_chunk[strlen (line_chunk) - 1] = '\0';
644 markup = g_string_append (markup, "<span");
646 g_string_append (markup, " style=\"italic\"");
648 g_string_append (markup, " weight=\"bold\"");
650 g_string_append_printf (markup, " size=\"%u\"", fontsize * 1000);
651 g_string_append_printf (markup, ">%s</span>", line_chunk);
654 g_string_append (markup, "\n");
655 line = line_split + 1;
661 g_string_free (markup, FALSE);
662 GST_DEBUG ("parse_mdvdsub returning (%f+%f): %s",
663 state->start_time / (double) GST_SECOND,
664 state->duration / (double) GST_SECOND, ret);
669 strip_trailing_newlines (gchar * txt)
675 while (len > 1 && txt[len - 1] == '\n') {
682 /* we want to escape text in general, but retain basic markup like
683 * <i></i>, <u></u>, and <b></b>. The easiest and safest way is to
684 * just unescape a white list of allowed markups again after
685 * escaping everything (the text between these simple markers isn't
686 * necessarily escaped, so it seems best to do it like this) */
688 subrip_unescape_formatting (gchar * txt)
692 for (pos = txt; pos != NULL && *pos != '\0'; ++pos) {
693 if (g_ascii_strncasecmp (pos, "<u>", 9) == 0 ||
694 g_ascii_strncasecmp (pos, "<i>", 9) == 0 ||
695 g_ascii_strncasecmp (pos, "<b>", 9) == 0) {
697 pos[1] = g_ascii_tolower (pos[4]);
699 /* move NUL terminator as well */
700 g_memmove (pos + 3, pos + 9, strlen (pos + 9) + 1);
705 for (pos = txt; pos != NULL && *pos != '\0'; ++pos) {
706 if (g_ascii_strncasecmp (pos, "</u>", 10) == 0 ||
707 g_ascii_strncasecmp (pos, "</i>", 10) == 0 ||
708 g_ascii_strncasecmp (pos, "</b>", 10) == 0) {
711 pos[2] = g_ascii_tolower (pos[5]);
713 /* move NUL terminator as well */
714 g_memmove (pos + 4, pos + 10, strlen (pos + 10) + 1);
722 subrip_remove_unhandled_tag (gchar * start, gchar * stop)
726 tag = start + strlen ("<");
730 if (g_ascii_tolower (*tag) < 'a' || g_ascii_tolower (*tag) > 'z')
735 GST_LOG ("removing unhandled tag '%s'", start);
737 g_memmove (start, stop, strlen (stop) + 1);
741 /* remove tags we haven't explicitly allowed earlier on, like font tags
744 subrip_remove_unhandled_tags (gchar * txt)
748 for (pos = txt; pos != NULL && *pos != '\0'; ++pos) {
749 if (strncmp (pos, "<", 4) == 0 && (gt = strstr (pos + 4, ">"))) {
750 if (subrip_remove_unhandled_tag (pos, gt + strlen (">")))
756 /* we only allow <i>, <u> and <b>, so let's take a simple approach. This code
757 * assumes the input has been escaped and subrip_unescape_formatting() has then
758 * been run over the input! This function adds missing closing markup tags and
759 * removes broken closing tags for tags that have never been opened. */
761 subrip_fix_up_markup (gchar ** p_txt)
763 gchar *cur, *next_tag;
765 guint num_open_tags = 0;
767 g_assert (*p_txt != NULL);
770 while (*cur != '\0') {
771 next_tag = strchr (cur, '<');
772 if (next_tag == NULL)
778 if (num_open_tags == 0 || open_tags[num_open_tags - 1] != *next_tag) {
779 GST_LOG ("broken input, closing tag '%c' is not open", *next_tag);
780 g_memmove (next_tag - 2, next_tag + 2, strlen (next_tag + 2) + 1);
783 /* it's all good, closing tag which is open */
791 if (num_open_tags == G_N_ELEMENTS (open_tags))
792 return; /* something dodgy is going on, stop parsing */
793 open_tags[num_open_tags] = *next_tag;
797 GST_ERROR ("unexpected tag '%c' (%s)", *next_tag, next_tag);
798 g_assert_not_reached ();
804 if (num_open_tags > 0) {
807 s = g_string_new (*p_txt);
808 while (num_open_tags > 0) {
809 GST_LOG ("adding missing closing tag '%c'", open_tags[num_open_tags - 1]);
810 g_string_append_c (s, '<');
811 g_string_append_c (s, '/');
812 g_string_append_c (s, open_tags[num_open_tags - 1]);
813 g_string_append_c (s, '>');
817 *p_txt = g_string_free (s, FALSE);
822 parse_subrip_time (const gchar * ts_string, GstClockTime * t)
824 gchar s[128] = { '\0', };
826 guint hour, min, sec, msec, len;
828 while (*ts_string == ' ')
831 g_strlcpy (s, ts_string, sizeof (s));
832 if ((end = strstr (s, "-->")))
836 /* ms may be in these formats:
837 * hh:mm:ss,500 = 500ms
840 * hh:mm:ss, 50 = 50ms
842 * and the same with . instead of ,.
843 * sscanf() doesn't differentiate between ' 5' and '5' so munge
844 * the white spaces within the timestamp to '0' (I'm sure there's a
845 * way to make sscanf() do this for us, but how?)
847 g_strdelimit (s, " ", '0');
848 g_strdelimit (s, ".", ',');
850 /* make sure we have exactly three digits after he comma */
852 g_assert (p != NULL);
859 g_strlcat (&p[len], "0", 2);
863 GST_LOG ("parsing timestamp '%s'", s);
864 if (sscanf (s, "%u:%u:%u,%u", &hour, &min, &sec, &msec) != 4) {
865 GST_WARNING ("failed to parse subrip timestamp string '%s'", s);
869 *t = ((hour * 3600) + (min * 60) + sec) * GST_SECOND + msec * GST_MSECOND;
874 parse_subrip (ParserState * state, const gchar * line)
879 switch (state->state) {
881 /* looking for a single integer */
882 if (sscanf (line, "%u", &subnum) == 1)
887 GstClockTime ts_start, ts_end;
890 /* looking for start_time --> end_time */
891 if ((end_time = strstr (line, " --> ")) &&
892 parse_subrip_time (line, &ts_start) &&
893 parse_subrip_time (end_time + strlen (" --> "), &ts_end) &&
894 state->start_time <= ts_end) {
896 state->start_time = ts_start;
897 state->duration = ts_end - ts_start;
899 GST_DEBUG ("error parsing subrip time line '%s'", line);
906 /* No need to parse that text if it's out of segment */
907 guint64 clip_start = 0, clip_stop = 0;
908 gboolean in_seg = FALSE;
910 /* Check our segment start/stop */
911 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
912 state->start_time, state->start_time + state->duration,
913 &clip_start, &clip_stop);
916 state->start_time = clip_start;
917 state->duration = clip_stop - clip_start;
923 /* looking for subtitle text; empty line ends this subtitle entry */
925 g_string_append_c (state->buf, '\n');
926 g_string_append (state->buf, line);
927 if (strlen (line) == 0) {
928 ret = g_markup_escape_text (state->buf->str, state->buf->len);
929 g_string_truncate (state->buf, 0);
931 subrip_unescape_formatting (ret);
932 subrip_remove_unhandled_tags (ret);
933 strip_trailing_newlines (ret);
934 subrip_fix_up_markup (&ret);
939 g_return_val_if_reached (NULL);
944 unescape_newlines_br (gchar * read)
948 /* Replace all occurences of '[br]' with a newline as version 2
949 * of the subviewer format uses this for newlines */
951 if (read[0] == '\0' || read[1] == '\0' || read[2] == '\0' || read[3] == '\0')
955 if (strncmp (read, "[br]", 4) == 0) {
969 parse_subviewer (ParserState * state, const gchar * line)
971 guint h1, m1, s1, ms1;
972 guint h2, m2, s2, ms2;
975 /* TODO: Maybe also parse the fields in the header, especially DELAY.
976 * For examples see the unit test or
977 * http://www.doom9.org/index.html?/sub.htm */
979 switch (state->state) {
981 /* looking for start_time,end_time */
982 if (sscanf (line, "%u:%u:%u.%u,%u:%u:%u.%u",
983 &h1, &m1, &s1, &ms1, &h2, &m2, &s2, &ms2) == 8) {
986 (((guint64) h1) * 3600 + m1 * 60 + s1) * GST_SECOND +
989 (((guint64) h2) * 3600 + m2 * 60 + s2) * GST_SECOND +
990 ms2 * GST_MSECOND - state->start_time;
995 /* No need to parse that text if it's out of segment */
996 guint64 clip_start = 0, clip_stop = 0;
997 gboolean in_seg = FALSE;
999 /* Check our segment start/stop */
1000 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1001 state->start_time, state->start_time + state->duration,
1002 &clip_start, &clip_stop);
1005 state->start_time = clip_start;
1006 state->duration = clip_stop - clip_start;
1012 /* looking for subtitle text; empty line ends this subtitle entry */
1013 if (state->buf->len)
1014 g_string_append_c (state->buf, '\n');
1015 g_string_append (state->buf, line);
1016 if (strlen (line) == 0) {
1017 ret = g_strdup (state->buf->str);
1018 unescape_newlines_br (ret);
1019 strip_trailing_newlines (ret);
1020 g_string_truncate (state->buf, 0);
1026 g_assert_not_reached ();
1032 parse_mpsub (ParserState * state, const gchar * line)
1037 switch (state->state) {
1039 /* looking for two floats (offset, duration) */
1040 if (sscanf (line, "%f %f", &t1, &t2) == 2) {
1042 state->start_time += state->duration + GST_SECOND * t1;
1043 state->duration = GST_SECOND * t2;
1047 { /* No need to parse that text if it's out of segment */
1048 guint64 clip_start = 0, clip_stop = 0;
1049 gboolean in_seg = FALSE;
1051 /* Check our segment start/stop */
1052 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1053 state->start_time, state->start_time + state->duration,
1054 &clip_start, &clip_stop);
1057 state->start_time = clip_start;
1058 state->duration = clip_stop - clip_start;
1064 /* looking for subtitle text; empty line ends this
1066 if (state->buf->len)
1067 g_string_append_c (state->buf, '\n');
1068 g_string_append (state->buf, line);
1069 if (strlen (line) == 0) {
1070 ret = g_strdup (state->buf->str);
1071 g_string_truncate (state->buf, 0);
1077 g_assert_not_reached ();
1082 static const gchar *
1083 dks_skip_timestamp (const gchar * line)
1085 while (*line && *line != ']')
1093 parse_dks (ParserState * state, const gchar * line)
1097 switch (state->state) {
1099 /* Looking for the start time and text */
1100 if (sscanf (line, "[%u:%u:%u]", &h, &m, &s) == 3) {
1102 state->start_time = (((guint64) h) * 3600 + m * 60 + s) * GST_SECOND;
1103 text = dks_skip_timestamp (line);
1106 g_string_append (state->buf, text);
1112 guint64 clip_start = 0, clip_stop = 0;
1116 /* Looking for the end time */
1117 if (sscanf (line, "[%u:%u:%u]", &h, &m, &s) == 3) {
1119 state->duration = (((guint64) h) * 3600 + m * 60 + s) * GST_SECOND -
1122 GST_WARNING ("Failed to parse subtitle end time");
1126 /* Check if this subtitle is out of the current segment */
1127 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1128 state->start_time, state->start_time + state->duration,
1129 &clip_start, &clip_stop);
1135 state->start_time = clip_start;
1136 state->duration = clip_stop - clip_start;
1138 ret = g_strdup (state->buf->str);
1139 g_string_truncate (state->buf, 0);
1140 unescape_newlines_br (ret);
1144 g_assert_not_reached ();
1150 parser_state_init (ParserState * state)
1152 GST_DEBUG ("initialising parser");
1155 g_string_truncate (state->buf, 0);
1157 state->buf = g_string_new (NULL);
1160 state->start_time = 0;
1161 state->duration = 0;
1162 state->max_duration = 0; /* no limit */
1164 state->segment = NULL;
1168 parser_state_dispose (GstSubParse * self, ParserState * state)
1171 g_string_free (state->buf, TRUE);
1174 if (state->user_data) {
1175 switch (self->parser_type) {
1176 #ifndef GST_DISABLE_XML
1177 case GST_SUB_PARSE_FORMAT_SAMI:
1178 sami_context_reset (state);
1187 /* regex type enum */
1190 GST_SUB_PARSE_REGEX_UNKNOWN = 0,
1191 GST_SUB_PARSE_REGEX_MDVDSUB = 1,
1192 GST_SUB_PARSE_REGEX_SUBRIP = 2,
1193 GST_SUB_PARSE_REGEX_DKS = 3,
1197 gst_sub_parse_data_format_autodetect_regex_once (GstSubParseRegex regtype)
1199 gpointer result = NULL;
1200 GError *gerr = NULL;
1202 case GST_SUB_PARSE_REGEX_MDVDSUB:
1204 (gpointer) g_regex_new ("^\\{[0-9]+\\}\\{[0-9]+\\}",
1205 G_REGEX_RAW | G_REGEX_OPTIMIZE, 0, &gerr);
1206 if (result == NULL) {
1207 g_warning ("Compilation of mdvd regex failed: %s", gerr->message);
1208 g_error_free (gerr);
1211 case GST_SUB_PARSE_REGEX_SUBRIP:
1213 g_regex_new ("^[\\s\\n]*[\\n]? {0,3}[ 0-9]{1,4}\\s*(\x0d)?\x0a"
1214 " ?[0-9]{1,2}: ?[0-9]{1,2}: ?[0-9]{1,2}[,.] {0,2}[0-9]{1,3}"
1215 " +--> +[0-9]{1,2}: ?[0-9]{1,2}: ?[0-9]{1,2}[,.] {0,2}[0-9]{1,2}",
1216 G_REGEX_RAW | G_REGEX_OPTIMIZE, 0, &gerr);
1217 if (result == NULL) {
1218 g_warning ("Compilation of subrip regex failed: %s", gerr->message);
1219 g_error_free (gerr);
1222 case GST_SUB_PARSE_REGEX_DKS:
1223 result = (gpointer) g_regex_new ("^\\[[0-9]+:[0-9]+:[0-9]+\\].*",
1224 G_REGEX_RAW | G_REGEX_OPTIMIZE, 0, &gerr);
1225 if (result == NULL) {
1226 g_warning ("Compilation of dks regex failed: %s", gerr->message);
1227 g_error_free (gerr);
1231 GST_WARNING ("Trying to allocate regex of unknown type %u", regtype);
1237 * FIXME: maybe we should pass along a second argument, the preceding
1238 * text buffer, because that is how this originally worked, even though
1239 * I don't really see the use of that.
1242 static GstSubParseFormat
1243 gst_sub_parse_data_format_autodetect (gchar * match_str)
1247 static GOnce mdvd_rx_once = G_ONCE_INIT;
1248 static GOnce subrip_rx_once = G_ONCE_INIT;
1249 static GOnce dks_rx_once = G_ONCE_INIT;
1255 g_once (&mdvd_rx_once,
1256 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1257 (gpointer) GST_SUB_PARSE_REGEX_MDVDSUB);
1258 g_once (&subrip_rx_once,
1259 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1260 (gpointer) GST_SUB_PARSE_REGEX_SUBRIP);
1261 g_once (&dks_rx_once,
1262 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1263 (gpointer) GST_SUB_PARSE_REGEX_DKS);
1265 mdvd_grx = (GRegex *) mdvd_rx_once.retval;
1266 subrip_grx = (GRegex *) subrip_rx_once.retval;
1267 dks_grx = (GRegex *) dks_rx_once.retval;
1269 if (g_regex_match (mdvd_grx, match_str, 0, NULL) == TRUE) {
1270 GST_LOG ("MicroDVD (frame based) format detected");
1271 return GST_SUB_PARSE_FORMAT_MDVDSUB;
1273 if (g_regex_match (subrip_grx, match_str, 0, NULL) == TRUE) {
1274 GST_LOG ("SubRip (time based) format detected");
1275 return GST_SUB_PARSE_FORMAT_SUBRIP;
1277 if (g_regex_match (dks_grx, match_str, 0, NULL) == TRUE) {
1278 GST_LOG ("DKS (time based) format detected");
1279 return GST_SUB_PARSE_FORMAT_DKS;
1282 if (!strncmp (match_str, "FORMAT=TIME", 11)) {
1283 GST_LOG ("MPSub (time based) format detected");
1284 return GST_SUB_PARSE_FORMAT_MPSUB;
1286 #ifndef GST_DISABLE_XML
1287 if (strstr (match_str, "<SAMI>") != NULL ||
1288 strstr (match_str, "<sami>") != NULL) {
1289 GST_LOG ("SAMI (time based) format detected");
1290 return GST_SUB_PARSE_FORMAT_SAMI;
1293 /* we're boldly assuming the first subtitle appears within the first hour */
1294 if (sscanf (match_str, "0:%02u:%02u:", &n1, &n2) == 2 ||
1295 sscanf (match_str, "0:%02u:%02u=", &n1, &n2) == 2 ||
1296 sscanf (match_str, "00:%02u:%02u:", &n1, &n2) == 2 ||
1297 sscanf (match_str, "00:%02u:%02u=", &n1, &n2) == 2 ||
1298 sscanf (match_str, "00:%02u:%02u,%u=", &n1, &n2, &n3) == 3) {
1299 GST_LOG ("TMPlayer (time based) format detected");
1300 return GST_SUB_PARSE_FORMAT_TMPLAYER;
1302 if (sscanf (match_str, "[%u][%u]", &n1, &n2) == 2) {
1303 GST_LOG ("MPL2 (time based) format detected");
1304 return GST_SUB_PARSE_FORMAT_MPL2;
1306 if (strstr (match_str, "[INFORMATION]") != NULL) {
1307 GST_LOG ("SubViewer (time based) format detected");
1308 return GST_SUB_PARSE_FORMAT_SUBVIEWER;
1310 if (strstr (match_str, "{QTtext}") != NULL) {
1311 GST_LOG ("QTtext (time based) format detected");
1312 return GST_SUB_PARSE_FORMAT_QTTEXT;
1315 GST_DEBUG ("no subtitle format detected");
1316 return GST_SUB_PARSE_FORMAT_UNKNOWN;
1320 gst_sub_parse_format_autodetect (GstSubParse * self)
1323 GstSubParseFormat format;
1325 if (strlen (self->textbuf->str) < 30) {
1326 GST_DEBUG ("File too small to be a subtitles file");
1330 data = g_strndup (self->textbuf->str, 35);
1331 format = gst_sub_parse_data_format_autodetect (data);
1334 self->parser_type = format;
1335 self->subtitle_codec = gst_sub_parse_get_format_description (format);
1336 parser_state_init (&self->state);
1339 case GST_SUB_PARSE_FORMAT_MDVDSUB:
1340 self->parse_line = parse_mdvdsub;
1341 return gst_caps_new_simple ("text/x-raw",
1342 "format", G_TYPE_STRING, "pango-markup", NULL);
1343 case GST_SUB_PARSE_FORMAT_SUBRIP:
1344 self->parse_line = parse_subrip;
1345 return gst_caps_new_simple ("text/x-raw",
1346 "format", G_TYPE_STRING, "pango-markup", NULL);
1347 case GST_SUB_PARSE_FORMAT_MPSUB:
1348 self->parse_line = parse_mpsub;
1349 return gst_caps_new_simple ("text/x-raw",
1350 "format", G_TYPE_STRING, "utf8", NULL);
1351 #ifndef GST_DISABLE_XML
1352 case GST_SUB_PARSE_FORMAT_SAMI:
1353 self->parse_line = parse_sami;
1354 sami_context_init (&self->state);
1355 return gst_caps_new_simple ("text/x-raw",
1356 "format", G_TYPE_STRING, "pango-markup", NULL);
1358 case GST_SUB_PARSE_FORMAT_TMPLAYER:
1359 self->parse_line = parse_tmplayer;
1360 self->state.max_duration = 5 * GST_SECOND;
1361 return gst_caps_new_simple ("text/x-raw",
1362 "format", G_TYPE_STRING, "utf8", NULL);
1363 case GST_SUB_PARSE_FORMAT_MPL2:
1364 self->parse_line = parse_mpl2;
1365 return gst_caps_new_simple ("text/x-raw",
1366 "format", G_TYPE_STRING, "pango-markup", NULL);
1367 case GST_SUB_PARSE_FORMAT_DKS:
1368 self->parse_line = parse_dks;
1369 return gst_caps_new_simple ("text/x-raw",
1370 "format", G_TYPE_STRING, "utf8", NULL);
1371 case GST_SUB_PARSE_FORMAT_SUBVIEWER:
1372 self->parse_line = parse_subviewer;
1373 return gst_caps_new_simple ("text/x-raw",
1374 "format", G_TYPE_STRING, "utf8", NULL);
1375 case GST_SUB_PARSE_FORMAT_QTTEXT:
1376 self->parse_line = parse_qttext;
1377 qttext_context_init (&self->state);
1378 return gst_caps_new_simple ("text/x-raw",
1379 "format", G_TYPE_STRING, "pango-markup", NULL);
1380 case GST_SUB_PARSE_FORMAT_UNKNOWN:
1382 GST_DEBUG ("no subtitle format detected");
1383 GST_ELEMENT_ERROR (self, STREAM, WRONG_TYPE,
1384 ("The input is not a valid/supported subtitle file"), (NULL));
1390 feed_textbuf (GstSubParse * self, GstBuffer * buf)
1394 gchar *input = NULL;
1398 discont = GST_BUFFER_IS_DISCONT (buf);
1400 if (GST_BUFFER_OFFSET_IS_VALID (buf) &&
1401 GST_BUFFER_OFFSET (buf) != self->offset) {
1402 self->offset = GST_BUFFER_OFFSET (buf);
1407 GST_INFO ("discontinuity");
1408 /* flush the parser state */
1409 parser_state_init (&self->state);
1410 g_string_truncate (self->textbuf, 0);
1411 gst_adapter_clear (self->adapter);
1412 #ifndef GST_DISABLE_XML
1413 if (self->parser_type == GST_SUB_PARSE_FORMAT_SAMI)
1414 sami_context_reset (&self->state);
1416 /* we could set a flag to make sure that the next buffer we push out also
1417 * has the DISCONT flag set, but there's no point really given that it's
1418 * subtitles which are discontinuous by nature. */
1421 self->offset += gst_buffer_get_size (buf);
1423 gst_adapter_push (self->adapter, buf);
1425 avail = gst_adapter_available (self->adapter);
1426 data = gst_adapter_map (self->adapter, avail);
1427 input = convert_encoding (self, (const gchar *) data, avail, &consumed);
1429 if (input && consumed > 0) {
1430 self->textbuf = g_string_append (self->textbuf, input);
1431 gst_adapter_unmap (self->adapter);
1432 gst_adapter_flush (self->adapter, consumed);
1434 gst_adapter_unmap (self->adapter);
1440 static GstFlowReturn
1441 handle_buffer (GstSubParse * self, GstBuffer * buf)
1443 GstFlowReturn ret = GST_FLOW_OK;
1444 GstCaps *caps = NULL;
1445 gchar *line, *subtitle;
1447 if (self->first_buffer) {
1450 gst_buffer_map (buf, &map, GST_MAP_READ);
1451 self->detected_encoding = detect_encoding ((gchar *) map.data, map.size);
1452 gst_buffer_unmap (buf, &map);
1453 self->first_buffer = FALSE;
1454 self->state.fps_n = self->fps_n;
1455 self->state.fps_d = self->fps_d;
1458 feed_textbuf (self, buf);
1460 /* make sure we know the format */
1461 if (G_UNLIKELY (self->parser_type == GST_SUB_PARSE_FORMAT_UNKNOWN)) {
1462 if (!(caps = gst_sub_parse_format_autodetect (self))) {
1463 return GST_FLOW_EOS;
1465 if (!gst_pad_set_caps (self->srcpad, caps)) {
1466 gst_caps_unref (caps);
1467 return GST_FLOW_EOS;
1469 gst_caps_unref (caps);
1472 if (self->subtitle_codec != NULL) {
1475 tags = gst_tag_list_new (GST_TAG_SUBTITLE_CODEC, self->subtitle_codec,
1477 gst_pad_push_event (self->srcpad, gst_event_new_tag (tags));
1481 while (!self->flushing && (line = get_next_line (self))) {
1484 /* Set segment on our parser state machine */
1485 self->state.segment = &self->segment;
1486 /* Now parse the line, out of segment lines will just return NULL */
1487 GST_LOG_OBJECT (self, "Parsing line '%s'", line + offset);
1488 subtitle = self->parse_line (&self->state, line + offset);
1492 guint subtitle_len = strlen (subtitle);
1494 /* +1 for terminating NUL character */
1495 buf = gst_buffer_new_and_alloc (subtitle_len + 1);
1497 /* copy terminating NUL character as well */
1498 gst_buffer_fill (buf, 0, subtitle, subtitle_len + 1);
1499 gst_buffer_set_size (buf, subtitle_len);
1501 GST_BUFFER_TIMESTAMP (buf) = self->state.start_time;
1502 GST_BUFFER_DURATION (buf) = self->state.duration;
1504 /* in some cases (e.g. tmplayer) we can only determine the duration
1505 * of a text chunk from the timestamp of the next text chunk; in those
1506 * cases, we probably want to limit the duration to something
1507 * reasonable, so we don't end up showing some text for e.g. 40 seconds
1508 * just because nothing else is being said during that time */
1509 if (self->state.max_duration > 0 && GST_BUFFER_DURATION_IS_VALID (buf)) {
1510 if (GST_BUFFER_DURATION (buf) > self->state.max_duration)
1511 GST_BUFFER_DURATION (buf) = self->state.max_duration;
1514 self->segment.position = self->state.start_time;
1516 GST_DEBUG_OBJECT (self, "Sending text '%s', %" GST_TIME_FORMAT " + %"
1517 GST_TIME_FORMAT, subtitle, GST_TIME_ARGS (self->state.start_time),
1518 GST_TIME_ARGS (self->state.duration));
1520 ret = gst_pad_push (self->srcpad, buf);
1522 /* move this forward (the tmplayer parser needs this) */
1523 if (self->state.duration != GST_CLOCK_TIME_NONE)
1524 self->state.start_time += self->state.duration;
1529 if (ret != GST_FLOW_OK) {
1530 GST_DEBUG_OBJECT (self, "flow: %s", gst_flow_get_name (ret));
1539 static GstFlowReturn
1540 gst_sub_parse_chain (GstPad * sinkpad, GstObject * parent, GstBuffer * buf)
1545 self = GST_SUBPARSE (parent);
1547 /* Push newsegment if needed */
1548 if (self->need_segment) {
1549 GST_LOG_OBJECT (self, "pushing newsegment event with %" GST_SEGMENT_FORMAT,
1552 gst_pad_push_event (self->srcpad, gst_event_new_segment (&self->segment));
1553 self->need_segment = FALSE;
1556 ret = handle_buffer (self, buf);
1562 gst_sub_parse_sink_event (GstPad * pad, GstObject * parent, GstEvent * event)
1564 GstSubParse *self = GST_SUBPARSE (parent);
1565 gboolean ret = FALSE;
1567 GST_DEBUG ("Handling %s event", GST_EVENT_TYPE_NAME (event));
1569 switch (GST_EVENT_TYPE (event)) {
1570 case GST_EVENT_EOS:{
1571 /* Make sure the last subrip chunk is pushed out even
1572 * if the file does not have an empty line at the end */
1573 if (self->parser_type == GST_SUB_PARSE_FORMAT_SUBRIP ||
1574 self->parser_type == GST_SUB_PARSE_FORMAT_TMPLAYER ||
1575 self->parser_type == GST_SUB_PARSE_FORMAT_MPL2 ||
1576 self->parser_type == GST_SUB_PARSE_FORMAT_QTTEXT) {
1577 gchar term_chars[] = { '\n', '\n', '\0' };
1578 GstBuffer *buf = gst_buffer_new_and_alloc (2 + 1);
1580 GST_DEBUG ("EOS. Pushing remaining text (if any)");
1581 gst_buffer_fill (buf, 0, term_chars, 3);
1582 gst_buffer_set_size (buf, 2);
1584 GST_BUFFER_OFFSET (buf) = self->offset;
1585 gst_sub_parse_chain (pad, parent, buf);
1587 ret = gst_pad_event_default (pad, parent, event);
1590 case GST_EVENT_SEGMENT:
1592 const GstSegment *s;
1593 gst_event_parse_segment (event, &s);
1594 if (s->format == GST_FORMAT_TIME)
1595 gst_event_copy_segment (event, &self->segment);
1596 GST_DEBUG_OBJECT (self, "newsegment (%s)",
1597 gst_format_get_name (self->segment.format));
1599 /* if not time format, we'll either start with a 0 timestamp anyway or
1600 * it's following a seek in which case we'll have saved the requested
1601 * seek segment and don't want to overwrite it (remember that on a seek
1602 * we always just seek back to the start in BYTES format and just throw
1603 * away all text that's before the requested position; if the subtitles
1604 * come from an upstream demuxer, it won't be able to handle our BYTES
1605 * seek request and instead send us a newsegment from the seek request
1606 * it received via its video pads instead, so all is fine then too) */
1608 gst_event_unref (event);
1611 case GST_EVENT_FLUSH_START:
1613 self->flushing = TRUE;
1615 ret = gst_pad_event_default (pad, parent, event);
1618 case GST_EVENT_FLUSH_STOP:
1620 self->flushing = FALSE;
1622 ret = gst_pad_event_default (pad, parent, event);
1626 ret = gst_pad_event_default (pad, parent, event);
1634 static GstStateChangeReturn
1635 gst_sub_parse_change_state (GstElement * element, GstStateChange transition)
1637 GstStateChangeReturn ret = GST_STATE_CHANGE_SUCCESS;
1638 GstSubParse *self = GST_SUBPARSE (element);
1640 switch (transition) {
1641 case GST_STATE_CHANGE_READY_TO_PAUSED:
1642 /* format detection will init the parser state */
1644 self->parser_type = GST_SUB_PARSE_FORMAT_UNKNOWN;
1645 self->valid_utf8 = TRUE;
1646 self->first_buffer = TRUE;
1647 g_free (self->detected_encoding);
1648 self->detected_encoding = NULL;
1649 g_string_truncate (self->textbuf, 0);
1650 gst_adapter_clear (self->adapter);
1656 ret = GST_ELEMENT_CLASS (parent_class)->change_state (element, transition);
1657 if (ret == GST_STATE_CHANGE_FAILURE)
1660 switch (transition) {
1661 case GST_STATE_CHANGE_PAUSED_TO_READY:
1662 parser_state_dispose (self, &self->state);
1663 self->parser_type = GST_SUB_PARSE_FORMAT_UNKNOWN;
1676 /* FIXME 0.11: these caps are ugly, use app/x-subtitle + type field or so;
1677 * also, give different subtitle formats really different types */
1678 static GstStaticCaps mpl2_caps =
1679 GST_STATIC_CAPS ("application/x-subtitle-mpl2");
1680 #define SUB_CAPS (gst_static_caps_get (&sub_caps))
1682 static GstStaticCaps tmp_caps =
1683 GST_STATIC_CAPS ("application/x-subtitle-tmplayer");
1684 #define TMP_CAPS (gst_static_caps_get (&tmp_caps))
1686 static GstStaticCaps sub_caps = GST_STATIC_CAPS ("application/x-subtitle");
1687 #define MPL2_CAPS (gst_static_caps_get (&mpl2_caps))
1689 #ifndef GST_DISABLE_XML
1690 static GstStaticCaps smi_caps = GST_STATIC_CAPS ("application/x-subtitle-sami");
1691 #define SAMI_CAPS (gst_static_caps_get (&smi_caps))
1694 static GstStaticCaps dks_caps = GST_STATIC_CAPS ("application/x-subtitle-dks");
1695 #define DKS_CAPS (gst_static_caps_get (&dks_caps))
1697 static GstStaticCaps qttext_caps =
1698 GST_STATIC_CAPS ("application/x-subtitle-qttext");
1699 #define QTTEXT_CAPS (gst_static_caps_get (&qttext_caps))
1702 gst_subparse_type_find (GstTypeFind * tf, gpointer private)
1704 GstSubParseFormat format;
1708 gchar *encoding = NULL;
1711 if (!(data = gst_type_find_peek (tf, 0, 129)))
1714 /* make sure string passed to _autodetect() is NUL-terminated */
1715 str = g_malloc0 (129);
1716 memcpy (str, data, 128);
1718 if ((encoding = detect_encoding (str, 128)) != NULL) {
1719 gchar *converted_str;
1723 converted_str = gst_convert_to_utf8 (str, 128, encoding, &tmp, &err);
1724 if (converted_str == NULL) {
1725 GST_DEBUG ("Encoding '%s' detected but conversion failed: %s", encoding,
1730 str = converted_str;
1735 /* Check if at least the first 120 chars are valid UTF8,
1736 * otherwise convert as always */
1737 if (!g_utf8_validate (str, 128, &end) && (end - str) < 120) {
1738 gchar *converted_str;
1742 enc = g_getenv ("GST_SUBTITLE_ENCODING");
1743 if (enc == NULL || *enc == '\0') {
1744 /* if local encoding is UTF-8 and no encoding specified
1745 * via the environment variable, assume ISO-8859-15 */
1746 if (g_get_charset (&enc)) {
1747 enc = "ISO-8859-15";
1750 converted_str = gst_convert_to_utf8 (str, 128, enc, &tmp, NULL);
1751 if (converted_str != NULL) {
1753 str = converted_str;
1757 format = gst_sub_parse_data_format_autodetect (str);
1761 case GST_SUB_PARSE_FORMAT_MDVDSUB:
1762 GST_DEBUG ("MicroDVD format detected");
1765 case GST_SUB_PARSE_FORMAT_SUBRIP:
1766 GST_DEBUG ("SubRip format detected");
1769 case GST_SUB_PARSE_FORMAT_MPSUB:
1770 GST_DEBUG ("MPSub format detected");
1773 #ifndef GST_DISABLE_XML
1774 case GST_SUB_PARSE_FORMAT_SAMI:
1775 GST_DEBUG ("SAMI (time-based) format detected");
1779 case GST_SUB_PARSE_FORMAT_TMPLAYER:
1780 GST_DEBUG ("TMPlayer (time based) format detected");
1783 /* FIXME: our MPL2 typefinding is not really good enough to warrant
1784 * returning a high probability (however, since we registered our
1785 * typefinder here with a rank of MARGINAL we should pretty much only
1786 * be called if most other typefinders have already run */
1787 case GST_SUB_PARSE_FORMAT_MPL2:
1788 GST_DEBUG ("MPL2 (time based) format detected");
1791 case GST_SUB_PARSE_FORMAT_SUBVIEWER:
1792 GST_DEBUG ("SubViewer format detected");
1795 case GST_SUB_PARSE_FORMAT_DKS:
1796 GST_DEBUG ("DKS format detected");
1799 case GST_SUB_PARSE_FORMAT_QTTEXT:
1800 GST_DEBUG ("QTtext format detected");
1804 case GST_SUB_PARSE_FORMAT_UNKNOWN:
1805 GST_DEBUG ("no subtitle format detected");
1809 /* if we're here, it's ok */
1810 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, caps);
1814 plugin_init (GstPlugin * plugin)
1816 GST_DEBUG_CATEGORY_INIT (sub_parse_debug, "subparse", 0, ".sub parser");
1818 if (!gst_type_find_register (plugin, "subparse_typefind", GST_RANK_MARGINAL,
1819 gst_subparse_type_find, "srt,sub,mpsub,mdvd,smi,txt,dks", SUB_CAPS,
1823 if (!gst_element_register (plugin, "subparse",
1824 GST_RANK_PRIMARY, GST_TYPE_SUBPARSE) ||
1825 !gst_element_register (plugin, "ssaparse",
1826 GST_RANK_PRIMARY, GST_TYPE_SSA_PARSE)) {
1833 GST_PLUGIN_DEFINE (GST_VERSION_MAJOR,
1837 plugin_init, VERSION, "LGPL", GST_PACKAGE_NAME, GST_PACKAGE_ORIGIN)