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., 59 Temple Place - Suite 330,
19 * Boston, MA 02111-1307, 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/plain; text/x-pango-markup")
83 static gboolean gst_sub_parse_src_event (GstPad * pad, GstEvent * event);
84 static gboolean gst_sub_parse_src_query (GstPad * pad, GstQuery ** query);
85 static gboolean gst_sub_parse_sink_event (GstPad * pad, GstEvent * event);
87 static GstStateChangeReturn gst_sub_parse_change_state (GstElement * element,
88 GstStateChange transition);
90 static GstFlowReturn gst_sub_parse_chain (GstPad * sinkpad, GstBuffer * buf);
92 #define gst_sub_parse_parent_class parent_class
93 G_DEFINE_TYPE (GstSubParse, gst_sub_parse, GST_TYPE_ELEMENT);
96 gst_sub_parse_dispose (GObject * object)
98 GstSubParse *subparse = GST_SUBPARSE (object);
100 GST_DEBUG_OBJECT (subparse, "cleaning up subtitle parser");
102 switch (subparse->parser_type) {
103 case GST_SUB_PARSE_FORMAT_QTTEXT:
104 qttext_context_deinit (&subparse->state);
106 #ifndef GST_DISABLE_XML
107 case GST_SUB_PARSE_FORMAT_SAMI:
108 sami_context_deinit (&subparse->state);
115 if (subparse->encoding) {
116 g_free (subparse->encoding);
117 subparse->encoding = NULL;
120 if (subparse->detected_encoding) {
121 g_free (subparse->detected_encoding);
122 subparse->detected_encoding = NULL;
125 if (subparse->adapter) {
126 g_object_unref (subparse->adapter);
127 subparse->adapter = NULL;
130 if (subparse->textbuf) {
131 g_string_free (subparse->textbuf, TRUE);
132 subparse->textbuf = NULL;
135 GST_CALL_PARENT (G_OBJECT_CLASS, dispose, (object));
139 gst_sub_parse_class_init (GstSubParseClass * klass)
141 GObjectClass *object_class = G_OBJECT_CLASS (klass);
142 GstElementClass *element_class = GST_ELEMENT_CLASS (klass);
144 object_class->dispose = gst_sub_parse_dispose;
145 object_class->set_property = gst_sub_parse_set_property;
146 object_class->get_property = gst_sub_parse_get_property;
148 gst_element_class_add_pad_template (element_class,
149 gst_static_pad_template_get (&sink_templ));
150 gst_element_class_add_pad_template (element_class,
151 gst_static_pad_template_get (&src_templ));
152 gst_element_class_set_details_simple (element_class,
153 "Subtitle parser", "Codec/Parser/Subtitle",
154 "Parses subtitle (.sub) files into text streams",
155 "Gustavo J. A. M. Carneiro <gjc@inescporto.pt>, "
156 "GStreamer maintainers <gstreamer-devel@lists.sourceforge.net>");
158 element_class->change_state = gst_sub_parse_change_state;
160 g_object_class_install_property (object_class, PROP_ENCODING,
161 g_param_spec_string ("subtitle-encoding", "subtitle charset encoding",
162 "Encoding to assume if input subtitles are not in UTF-8 or any other "
163 "Unicode encoding. If not set, the GST_SUBTITLE_ENCODING environment "
164 "variable will be checked for an encoding to use. If that is not set "
165 "either, ISO-8859-15 will be assumed.", DEFAULT_ENCODING,
166 G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
168 g_object_class_install_property (object_class, PROP_VIDEOFPS,
169 gst_param_spec_fraction ("video-fps", "Video framerate",
170 "Framerate of the video stream. This is needed by some subtitle "
171 "formats to synchronize subtitles and video properly. If not set "
172 "and the subtitle format requires it subtitles may be out of sync.",
173 0, 1, G_MAXINT, 1, 24000, 1001,
174 G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
178 gst_sub_parse_init (GstSubParse * subparse)
180 subparse->sinkpad = gst_pad_new_from_static_template (&sink_templ, "sink");
181 gst_pad_set_chain_function (subparse->sinkpad,
182 GST_DEBUG_FUNCPTR (gst_sub_parse_chain));
183 gst_pad_set_event_function (subparse->sinkpad,
184 GST_DEBUG_FUNCPTR (gst_sub_parse_sink_event));
185 gst_element_add_pad (GST_ELEMENT (subparse), subparse->sinkpad);
187 subparse->srcpad = gst_pad_new_from_static_template (&src_templ, "src");
188 gst_pad_set_event_function (subparse->srcpad,
189 GST_DEBUG_FUNCPTR (gst_sub_parse_src_event));
190 gst_pad_set_query_function (subparse->srcpad,
191 GST_DEBUG_FUNCPTR (gst_sub_parse_src_query));
192 gst_element_add_pad (GST_ELEMENT (subparse), subparse->srcpad);
194 subparse->textbuf = g_string_new (NULL);
195 subparse->parser_type = GST_SUB_PARSE_FORMAT_UNKNOWN;
196 subparse->flushing = FALSE;
197 gst_segment_init (&subparse->segment, GST_FORMAT_TIME);
198 subparse->need_segment = TRUE;
199 subparse->encoding = g_strdup (DEFAULT_ENCODING);
200 subparse->detected_encoding = NULL;
201 subparse->adapter = gst_adapter_new ();
203 subparse->fps_n = 24000;
204 subparse->fps_d = 1001;
208 * Source pad functions.
212 gst_sub_parse_src_query (GstPad * pad, GstQuery ** query)
214 GstSubParse *self = GST_SUBPARSE (gst_pad_get_parent (pad));
215 gboolean ret = FALSE;
217 GST_DEBUG ("Handling %s query", GST_QUERY_TYPE_NAME (*query));
219 switch (GST_QUERY_TYPE (*query)) {
220 case GST_QUERY_POSITION:{
223 gst_query_parse_position (*query, &fmt, NULL);
224 if (fmt != GST_FORMAT_TIME) {
225 ret = gst_pad_peer_query (self->sinkpad, query);
228 gst_query_set_position (*query, GST_FORMAT_TIME,
229 self->segment.position);
232 case GST_QUERY_SEEKING:
235 gboolean seekable = FALSE;
239 gst_query_parse_seeking (*query, &fmt, NULL, NULL, NULL);
240 if (fmt == GST_FORMAT_TIME) {
241 GstQuery *peerquery = gst_query_new_seeking (GST_FORMAT_BYTES);
243 seekable = gst_pad_peer_query (self->sinkpad, &peerquery);
245 gst_query_parse_seeking (peerquery, NULL, &seekable, NULL, NULL);
246 gst_query_unref (peerquery);
249 gst_query_set_seeking (*query, fmt, seekable, seekable ? 0 : -1, -1);
254 ret = gst_pad_peer_query (self->sinkpad, query);
258 gst_object_unref (self);
264 gst_sub_parse_src_event (GstPad * pad, GstEvent * event)
266 GstSubParse *self = GST_SUBPARSE (gst_pad_get_parent (pad));
267 gboolean ret = FALSE;
269 GST_DEBUG ("Handling %s event", GST_EVENT_TYPE_NAME (event));
271 switch (GST_EVENT_TYPE (event)) {
276 GstSeekType start_type, stop_type;
281 gst_event_parse_seek (event, &rate, &format, &flags,
282 &start_type, &start, &stop_type, &stop);
284 if (format != GST_FORMAT_TIME) {
285 GST_WARNING_OBJECT (self, "we only support seeking in TIME format");
286 gst_event_unref (event);
290 /* Convert that seek to a seeking in bytes at position 0,
291 FIXME: could use an index */
292 ret = gst_pad_push_event (self->sinkpad,
293 gst_event_new_seek (rate, GST_FORMAT_BYTES, flags,
294 GST_SEEK_TYPE_SET, 0, GST_SEEK_TYPE_NONE, 0));
297 /* Apply the seek to our segment */
298 gst_segment_do_seek (&self->segment, rate, format, flags,
299 start_type, start, stop_type, stop, &update);
301 GST_DEBUG_OBJECT (self, "segment after seek: %" GST_SEGMENT_FORMAT,
304 self->next_offset = 0;
306 self->need_segment = TRUE;
308 GST_WARNING_OBJECT (self, "seek to 0 bytes failed");
311 gst_event_unref (event);
315 ret = gst_pad_event_default (pad, event);
320 gst_object_unref (self);
326 gst_sub_parse_set_property (GObject * object, guint prop_id,
327 const GValue * value, GParamSpec * pspec)
329 GstSubParse *subparse = GST_SUBPARSE (object);
331 GST_OBJECT_LOCK (subparse);
334 g_free (subparse->encoding);
335 subparse->encoding = g_value_dup_string (value);
336 GST_LOG_OBJECT (object, "subtitle encoding set to %s",
337 GST_STR_NULL (subparse->encoding));
341 subparse->fps_n = gst_value_get_fraction_numerator (value);
342 subparse->fps_d = gst_value_get_fraction_denominator (value);
343 GST_DEBUG_OBJECT (object, "video framerate set to %d/%d", subparse->fps_n,
346 if (!subparse->state.have_internal_fps) {
347 subparse->state.fps_n = subparse->fps_n;
348 subparse->state.fps_d = subparse->fps_d;
353 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
356 GST_OBJECT_UNLOCK (subparse);
360 gst_sub_parse_get_property (GObject * object, guint prop_id,
361 GValue * value, GParamSpec * pspec)
363 GstSubParse *subparse = GST_SUBPARSE (object);
365 GST_OBJECT_LOCK (subparse);
368 g_value_set_string (value, subparse->encoding);
371 gst_value_set_fraction (value, subparse->fps_n, subparse->fps_d);
374 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
377 GST_OBJECT_UNLOCK (subparse);
381 gst_sub_parse_get_format_description (GstSubParseFormat format)
384 case GST_SUB_PARSE_FORMAT_MDVDSUB:
386 case GST_SUB_PARSE_FORMAT_SUBRIP:
388 case GST_SUB_PARSE_FORMAT_MPSUB:
390 case GST_SUB_PARSE_FORMAT_SAMI:
392 case GST_SUB_PARSE_FORMAT_TMPLAYER:
394 case GST_SUB_PARSE_FORMAT_MPL2:
396 case GST_SUB_PARSE_FORMAT_SUBVIEWER:
398 case GST_SUB_PARSE_FORMAT_DKS:
400 case GST_SUB_PARSE_FORMAT_QTTEXT:
403 case GST_SUB_PARSE_FORMAT_UNKNOWN:
410 gst_convert_to_utf8 (const gchar * str, gsize len, const gchar * encoding,
411 gsize * consumed, GError ** err)
416 /* The char cast is necessary in glib < 2.24 */
418 g_convert_with_fallback (str, len, "UTF-8", encoding, (char *) "*",
419 consumed, NULL, err);
423 /* + 3 to skip UTF-8 BOM if it was added */
425 if (len >= 3 && (guint8) ret[0] == 0xEF && (guint8) ret[1] == 0xBB
426 && (guint8) ret[2] == 0xBF)
427 g_memmove (ret, ret + 3, len + 1 - 3);
433 detect_encoding (const gchar * str, gsize len)
435 if (len >= 3 && (guint8) str[0] == 0xEF && (guint8) str[1] == 0xBB
436 && (guint8) str[2] == 0xBF)
437 return g_strdup ("UTF-8");
439 if (len >= 2 && (guint8) str[0] == 0xFE && (guint8) str[1] == 0xFF)
440 return g_strdup ("UTF-16BE");
442 if (len >= 2 && (guint8) str[0] == 0xFF && (guint8) str[1] == 0xFE)
443 return g_strdup ("UTF-16LE");
445 if (len >= 4 && (guint8) str[0] == 0x00 && (guint8) str[1] == 0x00
446 && (guint8) str[2] == 0xFE && (guint8) str[3] == 0xFF)
447 return g_strdup ("UTF-32BE");
449 if (len >= 4 && (guint8) str[0] == 0xFF && (guint8) str[1] == 0xFE
450 && (guint8) str[2] == 0x00 && (guint8) str[3] == 0x00)
451 return g_strdup ("UTF-32LE");
457 convert_encoding (GstSubParse * self, const gchar * str, gsize len,
460 const gchar *encoding;
466 /* First try any detected encoding */
467 if (self->detected_encoding) {
469 gst_convert_to_utf8 (str, len, self->detected_encoding, consumed, &err);
474 GST_WARNING_OBJECT (self, "could not convert string from '%s' to UTF-8: %s",
475 self->detected_encoding, err->message);
476 g_free (self->detected_encoding);
477 self->detected_encoding = NULL;
481 /* Otherwise check if it's UTF8 */
482 if (self->valid_utf8) {
483 if (g_utf8_validate (str, len, NULL)) {
484 GST_LOG_OBJECT (self, "valid UTF-8, no conversion needed");
486 return g_strndup (str, len);
488 GST_INFO_OBJECT (self, "invalid UTF-8!");
489 self->valid_utf8 = FALSE;
492 /* Else try fallback */
493 encoding = self->encoding;
494 if (encoding == NULL || *encoding == '\0') {
495 encoding = g_getenv ("GST_SUBTITLE_ENCODING");
497 if (encoding == NULL || *encoding == '\0') {
498 /* if local encoding is UTF-8 and no encoding specified
499 * via the environment variable, assume ISO-8859-15 */
500 if (g_get_charset (&encoding)) {
501 encoding = "ISO-8859-15";
505 ret = gst_convert_to_utf8 (str, len, encoding, consumed, &err);
508 GST_WARNING_OBJECT (self, "could not convert string from '%s' to UTF-8: %s",
509 encoding, err->message);
512 /* invalid input encoding, fall back to ISO-8859-15 (always succeeds) */
513 ret = gst_convert_to_utf8 (str, len, "ISO-8859-15", consumed, NULL);
516 GST_LOG_OBJECT (self,
517 "successfully converted %" G_GSIZE_FORMAT " characters from %s to UTF-8"
518 "%s", len, encoding, (err) ? " , using ISO-8859-15 as fallback" : "");
524 get_next_line (GstSubParse * self)
527 const char *line_end;
529 gboolean have_r = FALSE;
531 line_end = strchr (self->textbuf->str, '\n');
534 /* end-of-line not found; return for more data */
538 /* get rid of '\r' */
539 if (line_end != self->textbuf->str && *(line_end - 1) == '\r') {
544 line_len = line_end - self->textbuf->str;
545 line = g_strndup (self->textbuf->str, line_len);
546 self->textbuf = g_string_erase (self->textbuf, 0,
547 line_len + (have_r ? 2 : 1));
552 parse_mdvdsub (ParserState * state, const gchar * line)
554 const gchar *line_split;
556 guint start_frame, end_frame;
557 guint64 clip_start = 0, clip_stop = 0;
558 gboolean in_seg = FALSE;
562 /* style variables */
568 if (sscanf (line, "{%u}{%u}", &start_frame, &end_frame) != 2) {
569 g_warning ("Parse of the following line, assumed to be in microdvd .sub"
570 " format, failed:\n%s", line);
574 /* skip the {%u}{%u} part */
575 line = strchr (line, '}') + 1;
576 line = strchr (line, '}') + 1;
578 /* see if there's a first line with a framerate */
579 if (start_frame == 1 && end_frame == 1) {
580 gchar *rest, *end = NULL;
582 rest = g_strdup (line);
583 g_strdelimit (rest, ",", '.');
584 fps = g_ascii_strtod (rest, &end);
586 gst_util_double_to_fraction (fps, &state->fps_n, &state->fps_d);
587 GST_INFO ("framerate from file: %d/%d ('%s')", state->fps_n,
595 gst_util_uint64_scale (start_frame, GST_SECOND * state->fps_d,
598 gst_util_uint64_scale (end_frame - start_frame, GST_SECOND * state->fps_d,
601 /* Check our segment start/stop */
602 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
603 state->start_time, state->start_time + state->duration, &clip_start,
606 /* No need to parse that text if it's out of segment */
608 state->start_time = clip_start;
609 state->duration = clip_stop - clip_start;
614 markup = g_string_new (NULL);
619 /* parse style markup */
620 if (strncmp (line, "{y:i}", 5) == 0) {
622 line = strchr (line, '}') + 1;
624 if (strncmp (line, "{y:b}", 5) == 0) {
626 line = strchr (line, '}') + 1;
628 if (sscanf (line, "{s:%u}", &fontsize) == 1) {
629 line = strchr (line, '}') + 1;
631 /* forward slashes at beginning/end signify italics too */
632 if (g_str_has_prefix (line, "/")) {
636 if ((line_split = strchr (line, '|')))
637 line_chunk = g_markup_escape_text (line, line_split - line);
639 line_chunk = g_markup_escape_text (line, strlen (line));
641 /* Remove italics markers at end of line/stanza (CHECKME: are end slashes
642 * always at the end of a line or can they span multiple lines?) */
643 if (g_str_has_suffix (line_chunk, "/")) {
644 line_chunk[strlen (line_chunk) - 1] = '\0';
647 markup = g_string_append (markup, "<span");
649 g_string_append (markup, " style=\"italic\"");
651 g_string_append (markup, " weight=\"bold\"");
653 g_string_append_printf (markup, " size=\"%u\"", fontsize * 1000);
654 g_string_append_printf (markup, ">%s</span>", line_chunk);
657 g_string_append (markup, "\n");
658 line = line_split + 1;
664 g_string_free (markup, FALSE);
665 GST_DEBUG ("parse_mdvdsub returning (%f+%f): %s",
666 state->start_time / (double) GST_SECOND,
667 state->duration / (double) GST_SECOND, ret);
672 strip_trailing_newlines (gchar * txt)
678 while (len > 1 && txt[len - 1] == '\n') {
685 /* we want to escape text in general, but retain basic markup like
686 * <i></i>, <u></u>, and <b></b>. The easiest and safest way is to
687 * just unescape a white list of allowed markups again after
688 * escaping everything (the text between these simple markers isn't
689 * necessarily escaped, so it seems best to do it like this) */
691 subrip_unescape_formatting (gchar * txt)
695 for (pos = txt; pos != NULL && *pos != '\0'; ++pos) {
696 if (g_ascii_strncasecmp (pos, "<u>", 9) == 0 ||
697 g_ascii_strncasecmp (pos, "<i>", 9) == 0 ||
698 g_ascii_strncasecmp (pos, "<b>", 9) == 0) {
700 pos[1] = g_ascii_tolower (pos[4]);
702 /* move NUL terminator as well */
703 g_memmove (pos + 3, pos + 9, strlen (pos + 9) + 1);
708 for (pos = txt; pos != NULL && *pos != '\0'; ++pos) {
709 if (g_ascii_strncasecmp (pos, "</u>", 10) == 0 ||
710 g_ascii_strncasecmp (pos, "</i>", 10) == 0 ||
711 g_ascii_strncasecmp (pos, "</b>", 10) == 0) {
714 pos[2] = g_ascii_tolower (pos[5]);
716 /* move NUL terminator as well */
717 g_memmove (pos + 4, pos + 10, strlen (pos + 10) + 1);
725 subrip_remove_unhandled_tag (gchar * start, gchar * stop)
729 tag = start + strlen ("<");
733 if (g_ascii_tolower (*tag) < 'a' || g_ascii_tolower (*tag) > 'z')
738 GST_LOG ("removing unhandled tag '%s'", start);
740 g_memmove (start, stop, strlen (stop) + 1);
744 /* remove tags we haven't explicitly allowed earlier on, like font tags
747 subrip_remove_unhandled_tags (gchar * txt)
751 for (pos = txt; pos != NULL && *pos != '\0'; ++pos) {
752 if (strncmp (pos, "<", 4) == 0 && (gt = strstr (pos + 4, ">"))) {
753 if (subrip_remove_unhandled_tag (pos, gt + strlen (">")))
759 /* we only allow <i>, <u> and <b>, so let's take a simple approach. This code
760 * assumes the input has been escaped and subrip_unescape_formatting() has then
761 * been run over the input! This function adds missing closing markup tags and
762 * removes broken closing tags for tags that have never been opened. */
764 subrip_fix_up_markup (gchar ** p_txt)
766 gchar *cur, *next_tag;
768 guint num_open_tags = 0;
770 g_assert (*p_txt != NULL);
773 while (*cur != '\0') {
774 next_tag = strchr (cur, '<');
775 if (next_tag == NULL)
781 if (num_open_tags == 0 || open_tags[num_open_tags - 1] != *next_tag) {
782 GST_LOG ("broken input, closing tag '%c' is not open", *next_tag);
783 g_memmove (next_tag - 2, next_tag + 2, strlen (next_tag + 2) + 1);
786 /* it's all good, closing tag which is open */
794 if (num_open_tags == G_N_ELEMENTS (open_tags))
795 return; /* something dodgy is going on, stop parsing */
796 open_tags[num_open_tags] = *next_tag;
800 GST_ERROR ("unexpected tag '%c' (%s)", *next_tag, next_tag);
801 g_assert_not_reached ();
807 if (num_open_tags > 0) {
810 s = g_string_new (*p_txt);
811 while (num_open_tags > 0) {
812 GST_LOG ("adding missing closing tag '%c'", open_tags[num_open_tags - 1]);
813 g_string_append_c (s, '<');
814 g_string_append_c (s, '/');
815 g_string_append_c (s, open_tags[num_open_tags - 1]);
816 g_string_append_c (s, '>');
820 *p_txt = g_string_free (s, FALSE);
825 parse_subrip_time (const gchar * ts_string, GstClockTime * t)
827 gchar s[128] = { '\0', };
829 guint hour, min, sec, msec, len;
831 while (*ts_string == ' ')
834 g_strlcpy (s, ts_string, sizeof (s));
835 if ((end = strstr (s, "-->")))
839 /* ms may be in these formats:
840 * hh:mm:ss,500 = 500ms
843 * hh:mm:ss, 50 = 50ms
845 * and the same with . instead of ,.
846 * sscanf() doesn't differentiate between ' 5' and '5' so munge
847 * the white spaces within the timestamp to '0' (I'm sure there's a
848 * way to make sscanf() do this for us, but how?)
850 g_strdelimit (s, " ", '0');
851 g_strdelimit (s, ".", ',');
853 /* make sure we have exactly three digits after he comma */
855 g_assert (p != NULL);
862 g_strlcat (&p[len], "0", 2);
866 GST_LOG ("parsing timestamp '%s'", s);
867 if (sscanf (s, "%u:%u:%u,%u", &hour, &min, &sec, &msec) != 4) {
868 GST_WARNING ("failed to parse subrip timestamp string '%s'", s);
872 *t = ((hour * 3600) + (min * 60) + sec) * GST_SECOND + msec * GST_MSECOND;
877 parse_subrip (ParserState * state, const gchar * line)
882 switch (state->state) {
884 /* looking for a single integer */
885 if (sscanf (line, "%u", &subnum) == 1)
890 GstClockTime ts_start, ts_end;
893 /* looking for start_time --> end_time */
894 if ((end_time = strstr (line, " --> ")) &&
895 parse_subrip_time (line, &ts_start) &&
896 parse_subrip_time (end_time + strlen (" --> "), &ts_end) &&
897 state->start_time <= ts_end) {
899 state->start_time = ts_start;
900 state->duration = ts_end - ts_start;
902 GST_DEBUG ("error parsing subrip time line '%s'", line);
909 /* No need to parse that text if it's out of segment */
910 guint64 clip_start = 0, clip_stop = 0;
911 gboolean in_seg = FALSE;
913 /* Check our segment start/stop */
914 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
915 state->start_time, state->start_time + state->duration,
916 &clip_start, &clip_stop);
919 state->start_time = clip_start;
920 state->duration = clip_stop - clip_start;
926 /* looking for subtitle text; empty line ends this subtitle entry */
928 g_string_append_c (state->buf, '\n');
929 g_string_append (state->buf, line);
930 if (strlen (line) == 0) {
931 ret = g_markup_escape_text (state->buf->str, state->buf->len);
932 g_string_truncate (state->buf, 0);
934 subrip_unescape_formatting (ret);
935 subrip_remove_unhandled_tags (ret);
936 strip_trailing_newlines (ret);
937 subrip_fix_up_markup (&ret);
942 g_return_val_if_reached (NULL);
947 unescape_newlines_br (gchar * read)
951 /* Replace all occurences of '[br]' with a newline as version 2
952 * of the subviewer format uses this for newlines */
954 if (read[0] == '\0' || read[1] == '\0' || read[2] == '\0' || read[3] == '\0')
958 if (strncmp (read, "[br]", 4) == 0) {
972 parse_subviewer (ParserState * state, const gchar * line)
974 guint h1, m1, s1, ms1;
975 guint h2, m2, s2, ms2;
978 /* TODO: Maybe also parse the fields in the header, especially DELAY.
979 * For examples see the unit test or
980 * http://www.doom9.org/index.html?/sub.htm */
982 switch (state->state) {
984 /* looking for start_time,end_time */
985 if (sscanf (line, "%u:%u:%u.%u,%u:%u:%u.%u",
986 &h1, &m1, &s1, &ms1, &h2, &m2, &s2, &ms2) == 8) {
989 (((guint64) h1) * 3600 + m1 * 60 + s1) * GST_SECOND +
992 (((guint64) h2) * 3600 + m2 * 60 + s2) * GST_SECOND +
993 ms2 * GST_MSECOND - state->start_time;
998 /* No need to parse that text if it's out of segment */
999 guint64 clip_start = 0, clip_stop = 0;
1000 gboolean in_seg = FALSE;
1002 /* Check our segment start/stop */
1003 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1004 state->start_time, state->start_time + state->duration,
1005 &clip_start, &clip_stop);
1008 state->start_time = clip_start;
1009 state->duration = clip_stop - clip_start;
1015 /* looking for subtitle text; empty line ends this subtitle entry */
1016 if (state->buf->len)
1017 g_string_append_c (state->buf, '\n');
1018 g_string_append (state->buf, line);
1019 if (strlen (line) == 0) {
1020 ret = g_strdup (state->buf->str);
1021 unescape_newlines_br (ret);
1022 strip_trailing_newlines (ret);
1023 g_string_truncate (state->buf, 0);
1029 g_assert_not_reached ();
1035 parse_mpsub (ParserState * state, const gchar * line)
1040 switch (state->state) {
1042 /* looking for two floats (offset, duration) */
1043 if (sscanf (line, "%f %f", &t1, &t2) == 2) {
1045 state->start_time += state->duration + GST_SECOND * t1;
1046 state->duration = GST_SECOND * t2;
1050 { /* No need to parse that text if it's out of segment */
1051 guint64 clip_start = 0, clip_stop = 0;
1052 gboolean in_seg = FALSE;
1054 /* Check our segment start/stop */
1055 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1056 state->start_time, state->start_time + state->duration,
1057 &clip_start, &clip_stop);
1060 state->start_time = clip_start;
1061 state->duration = clip_stop - clip_start;
1067 /* looking for subtitle text; empty line ends this
1069 if (state->buf->len)
1070 g_string_append_c (state->buf, '\n');
1071 g_string_append (state->buf, line);
1072 if (strlen (line) == 0) {
1073 ret = g_strdup (state->buf->str);
1074 g_string_truncate (state->buf, 0);
1080 g_assert_not_reached ();
1085 static const gchar *
1086 dks_skip_timestamp (const gchar * line)
1088 while (*line && *line != ']')
1096 parse_dks (ParserState * state, const gchar * line)
1100 switch (state->state) {
1102 /* Looking for the start time and text */
1103 if (sscanf (line, "[%u:%u:%u]", &h, &m, &s) == 3) {
1105 state->start_time = (((guint64) h) * 3600 + m * 60 + s) * GST_SECOND;
1106 text = dks_skip_timestamp (line);
1109 g_string_append (state->buf, text);
1115 guint64 clip_start = 0, clip_stop = 0;
1119 /* Looking for the end time */
1120 if (sscanf (line, "[%u:%u:%u]", &h, &m, &s) == 3) {
1122 state->duration = (((guint64) h) * 3600 + m * 60 + s) * GST_SECOND -
1125 GST_WARNING ("Failed to parse subtitle end time");
1129 /* Check if this subtitle is out of the current segment */
1130 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1131 state->start_time, state->start_time + state->duration,
1132 &clip_start, &clip_stop);
1138 state->start_time = clip_start;
1139 state->duration = clip_stop - clip_start;
1141 ret = g_strdup (state->buf->str);
1142 g_string_truncate (state->buf, 0);
1143 unescape_newlines_br (ret);
1147 g_assert_not_reached ();
1153 parser_state_init (ParserState * state)
1155 GST_DEBUG ("initialising parser");
1158 g_string_truncate (state->buf, 0);
1160 state->buf = g_string_new (NULL);
1163 state->start_time = 0;
1164 state->duration = 0;
1165 state->max_duration = 0; /* no limit */
1167 state->segment = NULL;
1171 parser_state_dispose (GstSubParse * self, ParserState * state)
1174 g_string_free (state->buf, TRUE);
1177 if (state->user_data) {
1178 switch (self->parser_type) {
1179 #ifndef GST_DISABLE_XML
1180 case GST_SUB_PARSE_FORMAT_SAMI:
1181 sami_context_reset (state);
1190 /* regex type enum */
1193 GST_SUB_PARSE_REGEX_UNKNOWN = 0,
1194 GST_SUB_PARSE_REGEX_MDVDSUB = 1,
1195 GST_SUB_PARSE_REGEX_SUBRIP = 2,
1196 GST_SUB_PARSE_REGEX_DKS = 3,
1200 gst_sub_parse_data_format_autodetect_regex_once (GstSubParseRegex regtype)
1202 gpointer result = NULL;
1203 GError *gerr = NULL;
1205 case GST_SUB_PARSE_REGEX_MDVDSUB:
1207 (gpointer) g_regex_new ("^\\{[0-9]+\\}\\{[0-9]+\\}", 0, 0, &gerr);
1208 if (result == NULL) {
1209 g_warning ("Compilation of mdvd regex failed: %s", gerr->message);
1210 g_error_free (gerr);
1213 case GST_SUB_PARSE_REGEX_SUBRIP:
1214 result = (gpointer) g_regex_new ("^([ 0-9]){0,3}[0-9]\\s*(\x0d)?\x0a"
1215 "[ 0-9][0-9]:[ 0-9][0-9]:[ 0-9][0-9][,.][ 0-9]{0,2}[0-9]"
1216 " +--> +([ 0-9])?[0-9]:[ 0-9][0-9]:[ 0-9][0-9][,.][ 0-9]{0,2}[0-9]",
1218 if (result == NULL) {
1219 g_warning ("Compilation of subrip regex failed: %s", gerr->message);
1220 g_error_free (gerr);
1223 case GST_SUB_PARSE_REGEX_DKS:
1224 result = (gpointer) g_regex_new ("^\\[[0-9]+:[0-9]+:[0-9]+\\].*",
1226 if (result == NULL) {
1227 g_warning ("Compilation of dks regex failed: %s", gerr->message);
1228 g_error_free (gerr);
1232 GST_WARNING ("Trying to allocate regex of unknown type %u", regtype);
1238 * FIXME: maybe we should pass along a second argument, the preceding
1239 * text buffer, because that is how this originally worked, even though
1240 * I don't really see the use of that.
1243 static GstSubParseFormat
1244 gst_sub_parse_data_format_autodetect (gchar * match_str)
1248 static GOnce mdvd_rx_once = G_ONCE_INIT;
1249 static GOnce subrip_rx_once = G_ONCE_INIT;
1250 static GOnce dks_rx_once = G_ONCE_INIT;
1256 g_once (&mdvd_rx_once,
1257 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1258 (gpointer) GST_SUB_PARSE_REGEX_MDVDSUB);
1259 g_once (&subrip_rx_once,
1260 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1261 (gpointer) GST_SUB_PARSE_REGEX_SUBRIP);
1262 g_once (&dks_rx_once,
1263 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1264 (gpointer) GST_SUB_PARSE_REGEX_DKS);
1266 mdvd_grx = (GRegex *) mdvd_rx_once.retval;
1267 subrip_grx = (GRegex *) subrip_rx_once.retval;
1268 dks_grx = (GRegex *) dks_rx_once.retval;
1270 if (g_regex_match (mdvd_grx, match_str, 0, NULL) == TRUE) {
1271 GST_LOG ("MicroDVD (frame based) format detected");
1272 return GST_SUB_PARSE_FORMAT_MDVDSUB;
1274 if (g_regex_match (subrip_grx, match_str, 0, NULL) == TRUE) {
1275 GST_LOG ("SubRip (time based) format detected");
1276 return GST_SUB_PARSE_FORMAT_SUBRIP;
1278 if (g_regex_match (dks_grx, match_str, 0, NULL) == TRUE) {
1279 GST_LOG ("DKS (time based) format detected");
1280 return GST_SUB_PARSE_FORMAT_DKS;
1283 if (!strncmp (match_str, "FORMAT=TIME", 11)) {
1284 GST_LOG ("MPSub (time based) format detected");
1285 return GST_SUB_PARSE_FORMAT_MPSUB;
1287 #ifndef GST_DISABLE_XML
1288 if (strstr (match_str, "<SAMI>") != NULL ||
1289 strstr (match_str, "<sami>") != NULL) {
1290 GST_LOG ("SAMI (time based) format detected");
1291 return GST_SUB_PARSE_FORMAT_SAMI;
1294 /* we're boldly assuming the first subtitle appears within the first hour */
1295 if (sscanf (match_str, "0:%02u:%02u:", &n1, &n2) == 2 ||
1296 sscanf (match_str, "0:%02u:%02u=", &n1, &n2) == 2 ||
1297 sscanf (match_str, "00:%02u:%02u:", &n1, &n2) == 2 ||
1298 sscanf (match_str, "00:%02u:%02u=", &n1, &n2) == 2 ||
1299 sscanf (match_str, "00:%02u:%02u,%u=", &n1, &n2, &n3) == 3) {
1300 GST_LOG ("TMPlayer (time based) format detected");
1301 return GST_SUB_PARSE_FORMAT_TMPLAYER;
1303 if (sscanf (match_str, "[%u][%u]", &n1, &n2) == 2) {
1304 GST_LOG ("MPL2 (time based) format detected");
1305 return GST_SUB_PARSE_FORMAT_MPL2;
1307 if (strstr (match_str, "[INFORMATION]") != NULL) {
1308 GST_LOG ("SubViewer (time based) format detected");
1309 return GST_SUB_PARSE_FORMAT_SUBVIEWER;
1311 if (strstr (match_str, "{QTtext}") != NULL) {
1312 GST_LOG ("QTtext (time based) format detected");
1313 return GST_SUB_PARSE_FORMAT_QTTEXT;
1316 GST_DEBUG ("no subtitle format detected");
1317 return GST_SUB_PARSE_FORMAT_UNKNOWN;
1321 gst_sub_parse_format_autodetect (GstSubParse * self)
1324 GstSubParseFormat format;
1326 if (strlen (self->textbuf->str) < 30) {
1327 GST_DEBUG ("File too small to be a subtitles file");
1331 data = g_strndup (self->textbuf->str, 35);
1332 format = gst_sub_parse_data_format_autodetect (data);
1335 self->parser_type = format;
1336 self->subtitle_codec = gst_sub_parse_get_format_description (format);
1337 parser_state_init (&self->state);
1340 case GST_SUB_PARSE_FORMAT_MDVDSUB:
1341 self->parse_line = parse_mdvdsub;
1342 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1343 case GST_SUB_PARSE_FORMAT_SUBRIP:
1344 self->parse_line = parse_subrip;
1345 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1346 case GST_SUB_PARSE_FORMAT_MPSUB:
1347 self->parse_line = parse_mpsub;
1348 return gst_caps_new_simple ("text/plain", NULL);
1349 #ifndef GST_DISABLE_XML
1350 case GST_SUB_PARSE_FORMAT_SAMI:
1351 self->parse_line = parse_sami;
1352 sami_context_init (&self->state);
1353 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1355 case GST_SUB_PARSE_FORMAT_TMPLAYER:
1356 self->parse_line = parse_tmplayer;
1357 self->state.max_duration = 5 * GST_SECOND;
1358 return gst_caps_new_simple ("text/plain", NULL);
1359 case GST_SUB_PARSE_FORMAT_MPL2:
1360 self->parse_line = parse_mpl2;
1361 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1362 case GST_SUB_PARSE_FORMAT_DKS:
1363 self->parse_line = parse_dks;
1364 return gst_caps_new_simple ("text/plain", NULL);
1365 case GST_SUB_PARSE_FORMAT_SUBVIEWER:
1366 self->parse_line = parse_subviewer;
1367 return gst_caps_new_simple ("text/plain", NULL);
1368 case GST_SUB_PARSE_FORMAT_QTTEXT:
1369 self->parse_line = parse_qttext;
1370 qttext_context_init (&self->state);
1371 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1372 case GST_SUB_PARSE_FORMAT_UNKNOWN:
1374 GST_DEBUG ("no subtitle format detected");
1375 GST_ELEMENT_ERROR (self, STREAM, WRONG_TYPE,
1376 ("The input is not a valid/supported subtitle file"), (NULL));
1382 feed_textbuf (GstSubParse * self, GstBuffer * buf)
1386 gchar *input = NULL;
1390 discont = GST_BUFFER_IS_DISCONT (buf);
1392 if (GST_BUFFER_OFFSET_IS_VALID (buf) &&
1393 GST_BUFFER_OFFSET (buf) != self->offset) {
1394 self->offset = GST_BUFFER_OFFSET (buf);
1399 GST_INFO ("discontinuity");
1400 /* flush the parser state */
1401 parser_state_init (&self->state);
1402 g_string_truncate (self->textbuf, 0);
1403 gst_adapter_clear (self->adapter);
1404 #ifndef GST_DISABLE_XML
1405 if (self->parser_type == GST_SUB_PARSE_FORMAT_SAMI)
1406 sami_context_reset (&self->state);
1408 /* we could set a flag to make sure that the next buffer we push out also
1409 * has the DISCONT flag set, but there's no point really given that it's
1410 * subtitles which are discontinuous by nature. */
1413 self->offset = GST_BUFFER_OFFSET (buf) + gst_buffer_get_size (buf);
1414 self->next_offset = self->offset;
1416 gst_adapter_push (self->adapter, buf);
1418 avail = gst_adapter_available (self->adapter);
1419 data = gst_adapter_map (self->adapter, avail),
1420 input = convert_encoding (self, (const gchar *) data, avail, &consumed);
1422 if (input && consumed > 0) {
1423 self->textbuf = g_string_append (self->textbuf, input);
1424 gst_adapter_unmap (self->adapter, consumed);
1426 gst_adapter_unmap (self->adapter, 0);
1432 static GstFlowReturn
1433 handle_buffer (GstSubParse * self, GstBuffer * buf)
1435 GstFlowReturn ret = GST_FLOW_OK;
1436 GstCaps *caps = NULL;
1437 gchar *line, *subtitle, *data;
1440 if (self->first_buffer) {
1441 data = gst_buffer_map (buf, &size, NULL, GST_MAP_READ);
1442 self->detected_encoding = detect_encoding (data, size);
1443 gst_buffer_unmap (buf, data, size);
1444 self->first_buffer = FALSE;
1445 self->state.fps_n = self->fps_n;
1446 self->state.fps_d = self->fps_d;
1449 feed_textbuf (self, buf);
1451 /* make sure we know the format */
1452 if (G_UNLIKELY (self->parser_type == GST_SUB_PARSE_FORMAT_UNKNOWN)) {
1453 if (!(caps = gst_sub_parse_format_autodetect (self))) {
1454 return GST_FLOW_UNEXPECTED;
1456 if (!gst_pad_set_caps (self->srcpad, caps)) {
1457 gst_caps_unref (caps);
1458 return GST_FLOW_UNEXPECTED;
1460 gst_caps_unref (caps);
1463 if (self->subtitle_codec != NULL) {
1466 tags = gst_tag_list_new ();
1467 gst_tag_list_add (tags, GST_TAG_MERGE_APPEND, GST_TAG_SUBTITLE_CODEC,
1468 self->subtitle_codec, NULL);
1469 gst_element_found_tags_for_pad (GST_ELEMENT (self), self->srcpad, tags);
1473 while (!self->flushing && (line = get_next_line (self))) {
1476 /* Set segment on our parser state machine */
1477 self->state.segment = &self->segment;
1478 /* Now parse the line, out of segment lines will just return NULL */
1479 GST_LOG_OBJECT (self, "Parsing line '%s'", line + offset);
1480 subtitle = self->parse_line (&self->state, line + offset);
1484 guint subtitle_len = strlen (subtitle);
1486 /* +1 for terminating NUL character */
1487 buf = gst_buffer_new_and_alloc (subtitle_len + 1);
1489 /* copy terminating NUL character as well */
1490 gst_buffer_fill (buf, 0, subtitle, subtitle_len + 1);
1491 gst_buffer_set_size (buf, subtitle_len);
1493 GST_BUFFER_TIMESTAMP (buf) = self->state.start_time;
1494 GST_BUFFER_DURATION (buf) = self->state.duration;
1496 /* in some cases (e.g. tmplayer) we can only determine the duration
1497 * of a text chunk from the timestamp of the next text chunk; in those
1498 * cases, we probably want to limit the duration to something
1499 * reasonable, so we don't end up showing some text for e.g. 40 seconds
1500 * just because nothing else is being said during that time */
1501 if (self->state.max_duration > 0 && GST_BUFFER_DURATION_IS_VALID (buf)) {
1502 if (GST_BUFFER_DURATION (buf) > self->state.max_duration)
1503 GST_BUFFER_DURATION (buf) = self->state.max_duration;
1506 self->segment.position = self->state.start_time;
1508 GST_DEBUG_OBJECT (self, "Sending text '%s', %" GST_TIME_FORMAT " + %"
1509 GST_TIME_FORMAT, subtitle, GST_TIME_ARGS (self->state.start_time),
1510 GST_TIME_ARGS (self->state.duration));
1512 ret = gst_pad_push (self->srcpad, buf);
1514 /* move this forward (the tmplayer parser needs this) */
1515 if (self->state.duration != GST_CLOCK_TIME_NONE)
1516 self->state.start_time += self->state.duration;
1521 if (ret != GST_FLOW_OK) {
1522 GST_DEBUG_OBJECT (self, "flow: %s", gst_flow_get_name (ret));
1531 static GstFlowReturn
1532 gst_sub_parse_chain (GstPad * sinkpad, GstBuffer * buf)
1537 self = GST_SUBPARSE (GST_PAD_PARENT (sinkpad));
1539 /* Push newsegment if needed */
1540 if (self->need_segment) {
1541 GST_LOG_OBJECT (self, "pushing newsegment event with %" GST_SEGMENT_FORMAT,
1544 gst_pad_push_event (self->srcpad, gst_event_new_segment (&self->segment));
1545 self->need_segment = FALSE;
1548 ret = handle_buffer (self, buf);
1554 gst_sub_parse_sink_event (GstPad * pad, GstEvent * event)
1556 GstSubParse *self = GST_SUBPARSE (gst_pad_get_parent (pad));
1557 gboolean ret = FALSE;
1559 GST_DEBUG ("Handling %s event", GST_EVENT_TYPE_NAME (event));
1561 switch (GST_EVENT_TYPE (event)) {
1562 case GST_EVENT_EOS:{
1563 /* Make sure the last subrip chunk is pushed out even
1564 * if the file does not have an empty line at the end */
1565 if (self->parser_type == GST_SUB_PARSE_FORMAT_SUBRIP ||
1566 self->parser_type == GST_SUB_PARSE_FORMAT_TMPLAYER ||
1567 self->parser_type == GST_SUB_PARSE_FORMAT_MPL2 ||
1568 self->parser_type == GST_SUB_PARSE_FORMAT_QTTEXT) {
1569 gchar term_chars[] = { '\n', '\n', '\0' };
1570 GstBuffer *buf = gst_buffer_new_and_alloc (2 + 1);
1572 GST_DEBUG ("EOS. Pushing remaining text (if any)");
1573 gst_buffer_fill (buf, 0, term_chars, 3);
1574 gst_buffer_set_size (buf, 2);
1576 GST_BUFFER_OFFSET (buf) = self->offset;
1577 gst_sub_parse_chain (pad, buf);
1579 ret = gst_pad_event_default (pad, event);
1582 case GST_EVENT_SEGMENT:
1584 gst_event_parse_segment (event, &self->segment);
1585 GST_DEBUG_OBJECT (self, "newsegment (%s)",
1586 gst_format_get_name (self->segment.format));
1588 /* if not time format, we'll either start with a 0 timestamp anyway or
1589 * it's following a seek in which case we'll have saved the requested
1590 * seek segment and don't want to overwrite it (remember that on a seek
1591 * we always just seek back to the start in BYTES format and just throw
1592 * away all text that's before the requested position; if the subtitles
1593 * come from an upstream demuxer, it won't be able to handle our BYTES
1594 * seek request and instead send us a newsegment from the seek request
1595 * it received via its video pads instead, so all is fine then too) */
1597 gst_event_unref (event);
1600 case GST_EVENT_FLUSH_START:
1602 self->flushing = TRUE;
1604 ret = gst_pad_event_default (pad, event);
1607 case GST_EVENT_FLUSH_STOP:
1609 self->flushing = FALSE;
1611 ret = gst_pad_event_default (pad, event);
1615 ret = gst_pad_event_default (pad, event);
1619 gst_object_unref (self);
1625 static GstStateChangeReturn
1626 gst_sub_parse_change_state (GstElement * element, GstStateChange transition)
1628 GstStateChangeReturn ret = GST_STATE_CHANGE_SUCCESS;
1629 GstSubParse *self = GST_SUBPARSE (element);
1631 switch (transition) {
1632 case GST_STATE_CHANGE_READY_TO_PAUSED:
1633 /* format detection will init the parser state */
1635 self->next_offset = 0;
1636 self->parser_type = GST_SUB_PARSE_FORMAT_UNKNOWN;
1637 self->valid_utf8 = TRUE;
1638 self->first_buffer = TRUE;
1639 g_free (self->detected_encoding);
1640 self->detected_encoding = NULL;
1641 g_string_truncate (self->textbuf, 0);
1642 gst_adapter_clear (self->adapter);
1648 ret = GST_ELEMENT_CLASS (parent_class)->change_state (element, transition);
1649 if (ret == GST_STATE_CHANGE_FAILURE)
1652 switch (transition) {
1653 case GST_STATE_CHANGE_PAUSED_TO_READY:
1654 parser_state_dispose (self, &self->state);
1655 self->parser_type = GST_SUB_PARSE_FORMAT_UNKNOWN;
1668 /* FIXME 0.11: these caps are ugly, use app/x-subtitle + type field or so;
1669 * also, give different subtitle formats really different types */
1670 static GstStaticCaps mpl2_caps =
1671 GST_STATIC_CAPS ("application/x-subtitle-mpl2");
1672 #define SUB_CAPS (gst_static_caps_get (&sub_caps))
1674 static GstStaticCaps tmp_caps =
1675 GST_STATIC_CAPS ("application/x-subtitle-tmplayer");
1676 #define TMP_CAPS (gst_static_caps_get (&tmp_caps))
1678 static GstStaticCaps sub_caps = GST_STATIC_CAPS ("application/x-subtitle");
1679 #define MPL2_CAPS (gst_static_caps_get (&mpl2_caps))
1681 #ifndef GST_DISABLE_XML
1682 static GstStaticCaps smi_caps = GST_STATIC_CAPS ("application/x-subtitle-sami");
1683 #define SAMI_CAPS (gst_static_caps_get (&smi_caps))
1686 static GstStaticCaps dks_caps = GST_STATIC_CAPS ("application/x-subtitle-dks");
1687 #define DKS_CAPS (gst_static_caps_get (&dks_caps))
1689 static GstStaticCaps qttext_caps =
1690 GST_STATIC_CAPS ("application/x-subtitle-qttext");
1691 #define QTTEXT_CAPS (gst_static_caps_get (&qttext_caps))
1694 gst_subparse_type_find (GstTypeFind * tf, gpointer private)
1696 GstSubParseFormat format;
1700 gchar *encoding = NULL;
1703 if (!(data = gst_type_find_peek (tf, 0, 129)))
1706 /* make sure string passed to _autodetect() is NUL-terminated */
1707 str = g_malloc0 (129);
1708 memcpy (str, data, 128);
1710 if ((encoding = detect_encoding (str, 128)) != NULL) {
1711 gchar *converted_str;
1715 converted_str = gst_convert_to_utf8 (str, 128, encoding, &tmp, &err);
1716 if (converted_str == NULL) {
1717 GST_DEBUG ("Encoding '%s' detected but conversion failed: %s", encoding,
1723 str = converted_str;
1728 /* Check if at least the first 120 chars are valid UTF8,
1729 * otherwise convert as always */
1730 if (!g_utf8_validate (str, 128, &end) && (end - str) < 120) {
1731 gchar *converted_str;
1736 enc = g_getenv ("GST_SUBTITLE_ENCODING");
1737 if (enc == NULL || *enc == '\0') {
1738 /* if local encoding is UTF-8 and no encoding specified
1739 * via the environment variable, assume ISO-8859-15 */
1740 if (g_get_charset (&enc)) {
1741 enc = "ISO-8859-15";
1744 converted_str = gst_convert_to_utf8 (str, 128, enc, &tmp, &err);
1745 if (converted_str == NULL) {
1746 GST_DEBUG ("Charset conversion failed: %s", err->message);
1752 str = converted_str;
1756 format = gst_sub_parse_data_format_autodetect (str);
1760 case GST_SUB_PARSE_FORMAT_MDVDSUB:
1761 GST_DEBUG ("MicroDVD format detected");
1764 case GST_SUB_PARSE_FORMAT_SUBRIP:
1765 GST_DEBUG ("SubRip format detected");
1768 case GST_SUB_PARSE_FORMAT_MPSUB:
1769 GST_DEBUG ("MPSub format detected");
1772 #ifndef GST_DISABLE_XML
1773 case GST_SUB_PARSE_FORMAT_SAMI:
1774 GST_DEBUG ("SAMI (time-based) format detected");
1778 case GST_SUB_PARSE_FORMAT_TMPLAYER:
1779 GST_DEBUG ("TMPlayer (time based) format detected");
1782 /* FIXME: our MPL2 typefinding is not really good enough to warrant
1783 * returning a high probability (however, since we registered our
1784 * typefinder here with a rank of MARGINAL we should pretty much only
1785 * be called if most other typefinders have already run */
1786 case GST_SUB_PARSE_FORMAT_MPL2:
1787 GST_DEBUG ("MPL2 (time based) format detected");
1790 case GST_SUB_PARSE_FORMAT_SUBVIEWER:
1791 GST_DEBUG ("SubViewer format detected");
1794 case GST_SUB_PARSE_FORMAT_DKS:
1795 GST_DEBUG ("DKS format detected");
1798 case GST_SUB_PARSE_FORMAT_QTTEXT:
1799 GST_DEBUG ("QTtext format detected");
1803 case GST_SUB_PARSE_FORMAT_UNKNOWN:
1804 GST_DEBUG ("no subtitle format detected");
1808 /* if we're here, it's ok */
1809 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, caps);
1813 plugin_init (GstPlugin * plugin)
1815 static const gchar *sub_exts[] =
1816 { "srt", "sub", "mpsub", "mdvd", "smi", "txt", "dks", NULL };
1818 GST_DEBUG_CATEGORY_INIT (sub_parse_debug, "subparse", 0, ".sub parser");
1820 if (!gst_type_find_register (plugin, "subparse_typefind", GST_RANK_MARGINAL,
1821 gst_subparse_type_find, (gchar **) sub_exts, SUB_CAPS, NULL, NULL))
1824 if (!gst_element_register (plugin, "subparse",
1825 GST_RANK_PRIMARY, GST_TYPE_SUBPARSE) ||
1826 !gst_element_register (plugin, "ssaparse",
1827 GST_RANK_PRIMARY, GST_TYPE_SSA_PARSE)) {
1834 GST_PLUGIN_DEFINE (GST_VERSION_MAJOR,
1838 plugin_init, VERSION, "LGPL", GST_PACKAGE_NAME, GST_PACKAGE_ORIGIN)