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"
38 GST_DEBUG_CATEGORY (sub_parse_debug);
40 #define DEFAULT_ENCODING NULL
50 gst_sub_parse_set_property (GObject * object, guint prop_id,
51 const GValue * value, GParamSpec * pspec);
53 gst_sub_parse_get_property (GObject * object, guint prop_id,
54 GValue * value, GParamSpec * pspec);
57 static const GstElementDetails sub_parse_details =
58 GST_ELEMENT_DETAILS ("Subtitle parser",
59 "Codec/Parser/Subtitle",
60 "Parses subtitle (.sub) files into text streams",
61 "Gustavo J. A. M. Carneiro <gjc@inescporto.pt>\n"
62 "GStreamer maintainers <gstreamer-devel@lists.sourceforge.net>");
64 #ifndef GST_DISABLE_XML
65 static GstStaticPadTemplate sink_templ = GST_STATIC_PAD_TEMPLATE ("sink",
68 GST_STATIC_CAPS ("application/x-subtitle; application/x-subtitle-sami; "
69 "application/x-subtitle-tmplayer; application/x-subtitle-mpl2; "
70 "application/x-subtitle-dks")
73 static GstStaticPadTemplate sink_templ = GST_STATIC_PAD_TEMPLATE ("sink",
76 GST_STATIC_CAPS ("application/x-subtitle; application/x-subtitle-dks")
80 static GstStaticPadTemplate src_templ = GST_STATIC_PAD_TEMPLATE ("src",
83 GST_STATIC_CAPS ("text/plain; text/x-pango-markup")
86 static void gst_sub_parse_base_init (GstSubParseClass * klass);
87 static void gst_sub_parse_class_init (GstSubParseClass * klass);
88 static void gst_sub_parse_init (GstSubParse * subparse);
90 static gboolean gst_sub_parse_src_event (GstPad * pad, GstEvent * event);
91 static gboolean gst_sub_parse_src_query (GstPad * pad, GstQuery * query);
92 static gboolean gst_sub_parse_sink_event (GstPad * pad, GstEvent * event);
94 static GstStateChangeReturn gst_sub_parse_change_state (GstElement * element,
95 GstStateChange transition);
97 static GstFlowReturn gst_sub_parse_chain (GstPad * sinkpad, GstBuffer * buf);
99 static GstElementClass *parent_class = NULL;
102 gst_sub_parse_get_type (void)
104 static GType sub_parse_type = 0;
106 if (!sub_parse_type) {
107 static const GTypeInfo sub_parse_info = {
108 sizeof (GstSubParseClass),
109 (GBaseInitFunc) gst_sub_parse_base_init,
111 (GClassInitFunc) gst_sub_parse_class_init,
114 sizeof (GstSubParse),
116 (GInstanceInitFunc) gst_sub_parse_init,
119 sub_parse_type = g_type_register_static (GST_TYPE_ELEMENT,
120 "GstSubParse", &sub_parse_info, 0);
123 return sub_parse_type;
127 gst_sub_parse_base_init (GstSubParseClass * klass)
129 GstElementClass *element_class = GST_ELEMENT_CLASS (klass);
131 gst_element_class_add_pad_template (element_class,
132 gst_static_pad_template_get (&sink_templ));
133 gst_element_class_add_pad_template (element_class,
134 gst_static_pad_template_get (&src_templ));
135 gst_element_class_set_details (element_class, &sub_parse_details);
139 gst_sub_parse_dispose (GObject * object)
141 GstSubParse *subparse = GST_SUBPARSE (object);
143 GST_DEBUG_OBJECT (subparse, "cleaning up subtitle parser");
145 if (subparse->encoding) {
146 g_free (subparse->encoding);
147 subparse->encoding = NULL;
150 if (subparse->detected_encoding) {
151 g_free (subparse->detected_encoding);
152 subparse->detected_encoding = NULL;
155 if (subparse->adapter) {
156 g_object_unref (subparse->adapter);
157 subparse->adapter = NULL;
160 if (subparse->textbuf) {
161 g_string_free (subparse->textbuf, TRUE);
162 subparse->textbuf = NULL;
164 #ifndef GST_DISABLE_XML
165 sami_context_deinit (&subparse->state);
168 GST_CALL_PARENT (G_OBJECT_CLASS, dispose, (object));
172 gst_sub_parse_class_init (GstSubParseClass * klass)
174 GObjectClass *object_class = G_OBJECT_CLASS (klass);
175 GstElementClass *element_class = GST_ELEMENT_CLASS (klass);
177 parent_class = g_type_class_peek_parent (klass);
179 object_class->dispose = gst_sub_parse_dispose;
180 object_class->set_property = gst_sub_parse_set_property;
181 object_class->get_property = gst_sub_parse_get_property;
183 element_class->change_state = gst_sub_parse_change_state;
185 g_object_class_install_property (object_class, PROP_ENCODING,
186 g_param_spec_string ("subtitle-encoding", "subtitle charset encoding",
187 "Encoding to assume if input subtitles are not in UTF-8 or any other "
188 "Unicode encoding. If not set, the GST_SUBTITLE_ENCODING environment "
189 "variable will be checked for an encoding to use. If that is not set "
190 "either, ISO-8859-15 will be assumed.", DEFAULT_ENCODING,
191 G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
193 g_object_class_install_property (object_class, PROP_VIDEOFPS,
194 gst_param_spec_fraction ("video-fps", "Video framerate",
195 "Framerate of the video stream. This is needed by some subtitle "
196 "formats to synchronize subtitles and video properly. If not set "
197 "and the subtitle format requires it subtitles may be out of sync.",
198 0, 1, G_MAXINT, 1, 24000, 1001, G_PARAM_READWRITE));
202 gst_sub_parse_init (GstSubParse * subparse)
204 subparse->sinkpad = gst_pad_new_from_static_template (&sink_templ, "sink");
205 gst_pad_set_chain_function (subparse->sinkpad,
206 GST_DEBUG_FUNCPTR (gst_sub_parse_chain));
207 gst_pad_set_event_function (subparse->sinkpad,
208 GST_DEBUG_FUNCPTR (gst_sub_parse_sink_event));
209 gst_element_add_pad (GST_ELEMENT (subparse), subparse->sinkpad);
211 subparse->srcpad = gst_pad_new_from_static_template (&src_templ, "src");
212 gst_pad_set_event_function (subparse->srcpad,
213 GST_DEBUG_FUNCPTR (gst_sub_parse_src_event));
214 gst_pad_set_query_function (subparse->srcpad,
215 GST_DEBUG_FUNCPTR (gst_sub_parse_src_query));
216 gst_element_add_pad (GST_ELEMENT (subparse), subparse->srcpad);
218 subparse->textbuf = g_string_new (NULL);
219 subparse->parser_type = GST_SUB_PARSE_FORMAT_UNKNOWN;
220 subparse->flushing = FALSE;
221 gst_segment_init (&subparse->segment, GST_FORMAT_TIME);
222 subparse->need_segment = TRUE;
223 subparse->encoding = g_strdup (DEFAULT_ENCODING);
224 subparse->detected_encoding = NULL;
225 subparse->adapter = gst_adapter_new ();
227 subparse->fps_n = 24000;
228 subparse->fps_d = 1001;
232 * Source pad functions.
236 gst_sub_parse_src_query (GstPad * pad, GstQuery * query)
238 GstSubParse *self = GST_SUBPARSE (gst_pad_get_parent (pad));
239 gboolean ret = FALSE;
241 GST_DEBUG ("Handling %s query", GST_QUERY_TYPE_NAME (query));
243 switch (GST_QUERY_TYPE (query)) {
244 case GST_QUERY_POSITION:{
247 gst_query_parse_position (query, &fmt, NULL);
248 if (fmt != GST_FORMAT_TIME) {
249 ret = gst_pad_peer_query (self->sinkpad, query);
252 gst_query_set_position (query, GST_FORMAT_TIME,
253 self->segment.last_stop);
256 case GST_QUERY_SEEKING:
259 gboolean seekable = FALSE;
263 gst_query_parse_seeking (query, &fmt, NULL, NULL, NULL);
264 if (fmt == GST_FORMAT_TIME) {
265 GstQuery *peerquery = gst_query_new_seeking (GST_FORMAT_BYTES);
267 seekable = gst_pad_peer_query (self->sinkpad, peerquery);
269 gst_query_parse_seeking (peerquery, NULL, &seekable, NULL, NULL);
270 gst_query_unref (peerquery);
273 gst_query_set_seeking (query, fmt, seekable, seekable ? 0 : -1, -1);
278 ret = gst_pad_peer_query (self->sinkpad, query);
282 gst_object_unref (self);
288 gst_sub_parse_src_event (GstPad * pad, GstEvent * event)
290 GstSubParse *self = GST_SUBPARSE (gst_pad_get_parent (pad));
291 gboolean ret = FALSE;
293 GST_DEBUG ("Handling %s event", GST_EVENT_TYPE_NAME (event));
295 switch (GST_EVENT_TYPE (event)) {
299 GstSeekType start_type, stop_type;
304 gst_event_parse_seek (event, &rate, &format, &self->segment_flags,
305 &start_type, &start, &stop_type, &stop);
307 if (format != GST_FORMAT_TIME) {
308 GST_WARNING_OBJECT (self, "we only support seeking in TIME format");
309 gst_event_unref (event);
313 /* Convert that seek to a seeking in bytes at position 0,
314 FIXME: could use an index */
315 ret = gst_pad_push_event (self->sinkpad,
316 gst_event_new_seek (rate, GST_FORMAT_BYTES, self->segment_flags,
317 GST_SEEK_TYPE_SET, 0, GST_SEEK_TYPE_NONE, 0));
320 /* Apply the seek to our segment */
321 gst_segment_set_seek (&self->segment, rate, format, self->segment_flags,
322 start_type, start, stop_type, stop, &update);
324 GST_DEBUG_OBJECT (self, "segment after seek: %" GST_SEGMENT_FORMAT,
327 self->next_offset = 0;
329 self->need_segment = TRUE;
331 GST_WARNING_OBJECT (self, "seek to 0 bytes failed");
334 gst_event_unref (event);
338 ret = gst_pad_event_default (pad, event);
343 gst_object_unref (self);
349 gst_sub_parse_set_property (GObject * object, guint prop_id,
350 const GValue * value, GParamSpec * pspec)
352 GstSubParse *subparse = GST_SUBPARSE (object);
354 GST_OBJECT_LOCK (subparse);
357 g_free (subparse->encoding);
358 subparse->encoding = g_value_dup_string (value);
359 GST_LOG_OBJECT (object, "subtitle encoding set to %s",
360 GST_STR_NULL (subparse->encoding));
364 subparse->fps_n = gst_value_get_fraction_numerator (value);
365 subparse->fps_d = gst_value_get_fraction_denominator (value);
366 GST_DEBUG_OBJECT (object, "video framerate set to %d/%d", subparse->fps_n,
369 if (!subparse->state.have_internal_fps) {
370 subparse->state.fps_n = subparse->fps_n;
371 subparse->state.fps_d = subparse->fps_d;
376 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
379 GST_OBJECT_UNLOCK (subparse);
383 gst_sub_parse_get_property (GObject * object, guint prop_id,
384 GValue * value, GParamSpec * pspec)
386 GstSubParse *subparse = GST_SUBPARSE (object);
388 GST_OBJECT_LOCK (subparse);
391 g_value_set_string (value, subparse->encoding);
394 gst_value_set_fraction (value, subparse->fps_n, subparse->fps_d);
397 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
400 GST_OBJECT_UNLOCK (subparse);
404 gst_sub_parse_get_format_description (GstSubParseFormat format)
407 case GST_SUB_PARSE_FORMAT_MDVDSUB:
409 case GST_SUB_PARSE_FORMAT_SUBRIP:
411 case GST_SUB_PARSE_FORMAT_MPSUB:
413 case GST_SUB_PARSE_FORMAT_SAMI:
415 case GST_SUB_PARSE_FORMAT_TMPLAYER:
417 case GST_SUB_PARSE_FORMAT_MPL2:
419 case GST_SUB_PARSE_FORMAT_SUBVIEWER:
421 case GST_SUB_PARSE_FORMAT_DKS:
424 case GST_SUB_PARSE_FORMAT_UNKNOWN:
431 gst_convert_to_utf8 (const gchar * str, gsize len, const gchar * encoding,
432 gsize * consumed, GError ** err)
438 g_convert_with_fallback (str, len, "UTF-8", encoding, "*", consumed, NULL,
443 /* + 3 to skip UTF-8 BOM if it was added */
445 if (len >= 3 && (guint8) ret[0] == 0xEF && (guint8) ret[1] == 0xBB
446 && (guint8) ret[2] == 0xBF)
447 g_memmove (ret, ret + 3, len + 1 - 3);
453 detect_encoding (const gchar * str, gsize len)
455 if (len >= 3 && (guint8) str[0] == 0xEF && (guint8) str[1] == 0xBB
456 && (guint8) str[2] == 0xBF)
457 return g_strdup ("UTF-8");
459 if (len >= 2 && (guint8) str[0] == 0xFE && (guint8) str[1] == 0xFF)
460 return g_strdup ("UTF-16BE");
462 if (len >= 2 && (guint8) str[0] == 0xFF && (guint8) str[1] == 0xFE)
463 return g_strdup ("UTF-16LE");
465 if (len >= 4 && (guint8) str[0] == 0x00 && (guint8) str[1] == 0x00
466 && (guint8) str[2] == 0xFE && (guint8) str[3] == 0xFF)
467 return g_strdup ("UTF-32BE");
469 if (len >= 4 && (guint8) str[0] == 0xFF && (guint8) str[1] == 0xFE
470 && (guint8) str[2] == 0x00 && (guint8) str[3] == 0x00)
471 return g_strdup ("UTF-32LE");
477 convert_encoding (GstSubParse * self, const gchar * str, gsize len,
480 const gchar *encoding;
486 /* First try any detected encoding */
487 if (self->detected_encoding) {
489 gst_convert_to_utf8 (str, len, self->detected_encoding, consumed, &err);
494 GST_WARNING_OBJECT (self, "could not convert string from '%s' to UTF-8: %s",
495 self->detected_encoding, err->message);
496 g_free (self->detected_encoding);
497 self->detected_encoding = NULL;
501 /* Otherwise check if it's UTF8 */
502 if (self->valid_utf8) {
503 if (g_utf8_validate (str, len, NULL)) {
504 GST_LOG_OBJECT (self, "valid UTF-8, no conversion needed");
506 return g_strndup (str, len);
508 GST_INFO_OBJECT (self, "invalid UTF-8!");
509 self->valid_utf8 = FALSE;
512 /* Else try fallback */
513 encoding = self->encoding;
514 if (encoding == NULL || *encoding == '\0') {
515 encoding = g_getenv ("GST_SUBTITLE_ENCODING");
517 if (encoding == NULL || *encoding == '\0') {
518 /* if local encoding is UTF-8 and no encoding specified
519 * via the environment variable, assume ISO-8859-15 */
520 if (g_get_charset (&encoding)) {
521 encoding = "ISO-8859-15";
525 ret = gst_convert_to_utf8 (str, len, encoding, consumed, &err);
528 GST_WARNING_OBJECT (self, "could not convert string from '%s' to UTF-8: %s",
529 encoding, err->message);
532 /* invalid input encoding, fall back to ISO-8859-15 (always succeeds) */
533 ret = gst_convert_to_utf8 (str, len, "ISO-8859-15", consumed, NULL);
536 GST_LOG_OBJECT (self,
537 "successfully converted %" G_GSIZE_FORMAT " characters from %s to UTF-8"
538 "%s", len, encoding, (err) ? " , using ISO-8859-15 as fallback" : "");
544 get_next_line (GstSubParse * self)
547 const char *line_end;
549 gboolean have_r = FALSE;
551 line_end = strchr (self->textbuf->str, '\n');
554 /* end-of-line not found; return for more data */
558 /* get rid of '\r' */
559 if (line_end != self->textbuf->str && *(line_end - 1) == '\r') {
564 line_len = line_end - self->textbuf->str;
565 line = g_strndup (self->textbuf->str, line_len);
566 self->textbuf = g_string_erase (self->textbuf, 0,
567 line_len + (have_r ? 2 : 1));
572 parse_mdvdsub (ParserState * state, const gchar * line)
574 const gchar *line_split;
576 guint start_frame, end_frame;
577 gint64 clip_start = 0, clip_stop = 0;
578 gboolean in_seg = FALSE;
582 /* style variables */
588 if (sscanf (line, "{%u}{%u}", &start_frame, &end_frame) != 2) {
589 g_warning ("Parse of the following line, assumed to be in microdvd .sub"
590 " format, failed:\n%s", line);
594 /* skip the {%u}{%u} part */
595 line = strchr (line, '}') + 1;
596 line = strchr (line, '}') + 1;
598 /* see if there's a first line with a framerate */
599 if (start_frame == 1 && end_frame == 1) {
600 gchar *rest, *end = NULL;
602 rest = g_strdup (line);
603 g_strdelimit (rest, ",", '.');
604 fps = g_ascii_strtod (rest, &end);
609 /* Use double->fraction conversion from gstvalue.c */
610 g_value_init (&d, G_TYPE_DOUBLE);
611 g_value_init (&f, GST_TYPE_FRACTION);
612 g_value_set_double (&d, fps);
613 if (g_value_transform (&d, &f)) {
614 state->have_internal_fps = TRUE;
615 state->fps_n = gst_value_get_fraction_numerator (&f);
616 state->fps_d = gst_value_get_fraction_denominator (&f);
617 GST_INFO ("framerate from file: %d/%d ('%s')", state->fps_n,
628 gst_util_uint64_scale (start_frame, GST_SECOND * state->fps_d,
631 gst_util_uint64_scale (end_frame - start_frame, GST_SECOND * state->fps_d,
634 /* Check our segment start/stop */
635 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
636 state->start_time, state->start_time + state->duration, &clip_start,
639 /* No need to parse that text if it's out of segment */
641 state->start_time = clip_start;
642 state->duration = clip_stop - clip_start;
647 markup = g_string_new (NULL);
652 /* parse style markup */
653 if (strncmp (line, "{y:i}", 5) == 0) {
655 line = strchr (line, '}') + 1;
657 if (strncmp (line, "{y:b}", 5) == 0) {
659 line = strchr (line, '}') + 1;
661 if (sscanf (line, "{s:%u}", &fontsize) == 1) {
662 line = strchr (line, '}') + 1;
664 /* forward slashes at beginning/end signify italics too */
665 if (g_str_has_prefix (line, "/")) {
669 if ((line_split = strchr (line, '|')))
670 line_chunk = g_markup_escape_text (line, line_split - line);
672 line_chunk = g_markup_escape_text (line, strlen (line));
674 /* Remove italics markers at end of line/stanza (CHECKME: are end slashes
675 * always at the end of a line or can they span multiple lines?) */
676 if (g_str_has_suffix (line_chunk, "/")) {
677 line_chunk[strlen (line_chunk) - 1] = '\0';
680 markup = g_string_append (markup, "<span");
682 g_string_append (markup, " style=\"italic\"");
684 g_string_append (markup, " weight=\"bold\"");
686 g_string_append_printf (markup, " size=\"%u\"", fontsize * 1000);
687 g_string_append_printf (markup, ">%s</span>", line_chunk);
690 g_string_append (markup, "\n");
691 line = line_split + 1;
697 g_string_free (markup, FALSE);
698 GST_DEBUG ("parse_mdvdsub returning (%f+%f): %s",
699 state->start_time / (double) GST_SECOND,
700 state->duration / (double) GST_SECOND, ret);
705 strip_trailing_newlines (gchar * txt)
711 while (len > 1 && txt[len - 1] == '\n') {
718 /* we want to escape text in general, but retain basic markup like
719 * <i></i>, <u></u>, and <b></b>. The easiest and safest way is to
720 * just unescape a white list of allowed markups again after
721 * escaping everything (the text between these simple markers isn't
722 * necessarily escaped, so it seems best to do it like this) */
724 subrip_unescape_formatting (gchar * txt)
728 for (pos = txt; pos != NULL && *pos != '\0'; ++pos) {
729 if (g_ascii_strncasecmp (pos, "<u>", 9) == 0 ||
730 g_ascii_strncasecmp (pos, "<i>", 9) == 0 ||
731 g_ascii_strncasecmp (pos, "<b>", 9) == 0) {
733 pos[1] = g_ascii_tolower (pos[4]);
735 /* move NUL terminator as well */
736 g_memmove (pos + 3, pos + 9, strlen (pos + 9) + 1);
741 for (pos = txt; pos != NULL && *pos != '\0'; ++pos) {
742 if (g_ascii_strncasecmp (pos, "</u>", 10) == 0 ||
743 g_ascii_strncasecmp (pos, "</i>", 10) == 0 ||
744 g_ascii_strncasecmp (pos, "</b>", 10) == 0) {
747 pos[2] = g_ascii_tolower (pos[5]);
749 /* move NUL terminator as well */
750 g_memmove (pos + 4, pos + 10, strlen (pos + 10) + 1);
758 subrip_remove_unhandled_tag (gchar * start, gchar * stop)
762 tag = start + strlen ("<");
766 if (g_ascii_tolower (*tag) < 'a' || g_ascii_tolower (*tag) > 'z')
771 GST_LOG ("removing unhandled tag '%s'", start);
773 g_memmove (start, stop, strlen (stop) + 1);
777 /* remove tags we haven't explicitly allowed earlier on, like font tags
780 subrip_remove_unhandled_tags (gchar * txt)
784 for (pos = txt; pos != NULL && *pos != '\0'; ++pos) {
785 if (strncmp (pos, "<", 4) == 0 && (gt = strstr (pos + 4, ">"))) {
786 if (subrip_remove_unhandled_tag (pos, gt + strlen (">")))
792 /* we only allow <i>, <u> and <b>, so let's take a simple approach. This code
793 * assumes the input has been escaped and subrip_unescape_formatting() has then
794 * been run over the input! This function adds missing closing markup tags and
795 * removes broken closing tags for tags that have never been opened. */
797 subrip_fix_up_markup (gchar ** p_txt)
799 gchar *cur, *next_tag;
801 guint num_open_tags = 0;
803 g_assert (*p_txt != NULL);
806 while (*cur != '\0') {
807 next_tag = strchr (cur, '<');
808 if (next_tag == NULL)
814 if (num_open_tags == 0 || open_tags[num_open_tags - 1] != *next_tag) {
815 GST_LOG ("broken input, closing tag '%c' is not open", *next_tag);
816 g_memmove (next_tag - 2, next_tag + 2, strlen (next_tag + 2) + 1);
819 /* it's all good, closing tag which is open */
827 if (num_open_tags == G_N_ELEMENTS (open_tags))
828 return; /* something dodgy is going on, stop parsing */
829 open_tags[num_open_tags] = *next_tag;
833 GST_ERROR ("unexpected tag '%c' (%s)", *next_tag, next_tag);
834 g_assert_not_reached ();
840 if (num_open_tags > 0) {
843 s = g_string_new (*p_txt);
844 while (num_open_tags > 0) {
845 GST_LOG ("adding missing closing tag '%c'", open_tags[num_open_tags - 1]);
846 g_string_append_c (s, '<');
847 g_string_append_c (s, '/');
848 g_string_append_c (s, open_tags[num_open_tags - 1]);
849 g_string_append_c (s, '>');
853 *p_txt = g_string_free (s, FALSE);
858 parse_subrip_time (const gchar * ts_string, GstClockTime * t)
860 gchar s[128] = { '\0', };
862 guint hour, min, sec, msec, len;
864 while (*ts_string == ' ')
867 g_strlcpy (s, ts_string, sizeof (s));
868 if ((end = strstr (s, "-->")))
872 /* ms may be in these formats:
873 * hh:mm:ss,500 = 500ms
876 * hh:mm:ss, 50 = 50ms
878 * and the same with . instead of ,.
879 * sscanf() doesn't differentiate between ' 5' and '5' so munge
880 * the white spaces within the timestamp to '0' (I'm sure there's a
881 * way to make sscanf() do this for us, but how?)
883 g_strdelimit (s, " ", '0');
884 g_strdelimit (s, ".", ',');
886 /* make sure we have exactly three digits after he comma */
888 g_assert (p != NULL);
895 g_strlcat (&p[len], "0", 2);
899 GST_LOG ("parsing timestamp '%s'", s);
900 if (sscanf (s, "%u:%u:%u,%u", &hour, &min, &sec, &msec) != 4) {
901 GST_WARNING ("failed to parse subrip timestamp string '%s'", s);
905 *t = ((hour * 3600) + (min * 60) + sec) * GST_SECOND + msec * GST_MSECOND;
910 parse_subrip (ParserState * state, const gchar * line)
915 switch (state->state) {
917 /* looking for a single integer */
918 if (sscanf (line, "%u", &subnum) == 1)
923 GstClockTime ts_start, ts_end;
926 /* looking for start_time --> end_time */
927 if ((end_time = strstr (line, " --> ")) &&
928 parse_subrip_time (line, &ts_start) &&
929 parse_subrip_time (end_time + strlen (" --> "), &ts_end) &&
930 state->start_time <= ts_end) {
932 state->start_time = ts_start;
933 state->duration = ts_end - ts_start;
935 GST_DEBUG ("error parsing subrip time line '%s'", line);
942 /* No need to parse that text if it's out of segment */
943 gint64 clip_start = 0, clip_stop = 0;
944 gboolean in_seg = FALSE;
946 /* Check our segment start/stop */
947 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
948 state->start_time, state->start_time + state->duration,
949 &clip_start, &clip_stop);
952 state->start_time = clip_start;
953 state->duration = clip_stop - clip_start;
959 /* looking for subtitle text; empty line ends this subtitle entry */
961 g_string_append_c (state->buf, '\n');
962 g_string_append (state->buf, line);
963 if (strlen (line) == 0) {
964 ret = g_markup_escape_text (state->buf->str, state->buf->len);
965 g_string_truncate (state->buf, 0);
967 subrip_unescape_formatting (ret);
968 subrip_remove_unhandled_tags (ret);
969 strip_trailing_newlines (ret);
970 subrip_fix_up_markup (&ret);
975 g_return_val_if_reached (NULL);
980 unescape_newlines_br (gchar * read)
984 /* Replace all occurences of '[br]' with a newline as version 2
985 * of the subviewer format uses this for newlines */
987 if (read[0] == '\0' || read[1] == '\0' || read[2] == '\0' || read[3] == '\0')
991 if (strncmp (read, "[br]", 4) == 0) {
1005 parse_subviewer (ParserState * state, const gchar * line)
1007 guint h1, m1, s1, ms1;
1008 guint h2, m2, s2, ms2;
1011 /* TODO: Maybe also parse the fields in the header, especially DELAY.
1012 * For examples see the unit test or
1013 * http://www.doom9.org/index.html?/sub.htm */
1015 switch (state->state) {
1017 /* looking for start_time,end_time */
1018 if (sscanf (line, "%u:%u:%u.%u,%u:%u:%u.%u",
1019 &h1, &m1, &s1, &ms1, &h2, &m2, &s2, &ms2) == 8) {
1022 (((guint64) h1) * 3600 + m1 * 60 + s1) * GST_SECOND +
1025 (((guint64) h2) * 3600 + m2 * 60 + s2) * GST_SECOND +
1026 ms2 * GST_MSECOND - state->start_time;
1031 /* No need to parse that text if it's out of segment */
1032 gint64 clip_start = 0, clip_stop = 0;
1033 gboolean in_seg = FALSE;
1035 /* Check our segment start/stop */
1036 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1037 state->start_time, state->start_time + state->duration,
1038 &clip_start, &clip_stop);
1041 state->start_time = clip_start;
1042 state->duration = clip_stop - clip_start;
1048 /* looking for subtitle text; empty line ends this subtitle entry */
1049 if (state->buf->len)
1050 g_string_append_c (state->buf, '\n');
1051 g_string_append (state->buf, line);
1052 if (strlen (line) == 0) {
1053 ret = g_strdup (state->buf->str);
1054 unescape_newlines_br (ret);
1055 strip_trailing_newlines (ret);
1056 g_string_truncate (state->buf, 0);
1062 g_assert_not_reached ();
1068 parse_mpsub (ParserState * state, const gchar * line)
1073 switch (state->state) {
1075 /* looking for two floats (offset, duration) */
1076 if (sscanf (line, "%f %f", &t1, &t2) == 2) {
1078 state->start_time += state->duration + GST_SECOND * t1;
1079 state->duration = GST_SECOND * t2;
1083 { /* No need to parse that text if it's out of segment */
1084 gint64 clip_start = 0, clip_stop = 0;
1085 gboolean in_seg = FALSE;
1087 /* Check our segment start/stop */
1088 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1089 state->start_time, state->start_time + state->duration,
1090 &clip_start, &clip_stop);
1093 state->start_time = clip_start;
1094 state->duration = clip_stop - clip_start;
1100 /* looking for subtitle text; empty line ends this
1102 if (state->buf->len)
1103 g_string_append_c (state->buf, '\n');
1104 g_string_append (state->buf, line);
1105 if (strlen (line) == 0) {
1106 ret = g_strdup (state->buf->str);
1107 g_string_truncate (state->buf, 0);
1113 g_assert_not_reached ();
1118 static const gchar *
1119 dks_skip_timestamp (const gchar * line)
1121 while (*line && *line != ']')
1129 parse_dks (ParserState * state, const gchar * line)
1133 switch (state->state) {
1135 /* Looking for the start time and text */
1136 if (sscanf (line, "[%u:%u:%u]", &h, &m, &s) == 3) {
1138 state->start_time = (((guint64) h) * 3600 + m * 60 + s) * GST_SECOND;
1139 text = dks_skip_timestamp (line);
1142 g_string_append (state->buf, text);
1148 gint64 clip_start = 0, clip_stop = 0;
1152 /* Looking for the end time */
1153 if (sscanf (line, "[%u:%u:%u]", &h, &m, &s) == 3) {
1155 state->duration = (((guint64) h) * 3600 + m * 60 + s) * GST_SECOND -
1158 GST_WARNING ("Failed to parse subtitle end time");
1162 /* Check if this subtitle is out of the current segment */
1163 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1164 state->start_time, state->start_time + state->duration,
1165 &clip_start, &clip_stop);
1171 state->start_time = clip_start;
1172 state->duration = clip_stop - clip_start;
1174 ret = g_strdup (state->buf->str);
1175 g_string_truncate (state->buf, 0);
1176 unescape_newlines_br (ret);
1180 g_assert_not_reached ();
1186 parser_state_init (ParserState * state)
1188 GST_DEBUG ("initialising parser");
1191 g_string_truncate (state->buf, 0);
1193 state->buf = g_string_new (NULL);
1196 state->start_time = 0;
1197 state->duration = 0;
1198 state->max_duration = 0; /* no limit */
1200 state->segment = NULL;
1204 parser_state_dispose (ParserState * state)
1207 g_string_free (state->buf, TRUE);
1210 #ifndef GST_DISABLE_XML
1211 if (state->user_data) {
1212 sami_context_reset (state);
1217 /* regex type enum */
1220 GST_SUB_PARSE_REGEX_UNKNOWN = 0,
1221 GST_SUB_PARSE_REGEX_MDVDSUB = 1,
1222 GST_SUB_PARSE_REGEX_SUBRIP = 2,
1223 GST_SUB_PARSE_REGEX_DKS = 3,
1227 gst_sub_parse_data_format_autodetect_regex_once (GstSubParseRegex regtype)
1229 gpointer result = NULL;
1230 GError *gerr = NULL;
1232 case GST_SUB_PARSE_REGEX_MDVDSUB:
1234 (gpointer) g_regex_new ("^\\{[0-9]+\\}\\{[0-9]+\\}", 0, 0, &gerr);
1235 if (result == NULL) {
1236 g_warning ("Compilation of mdvd regex failed: %s", gerr->message);
1237 g_error_free (gerr);
1240 case GST_SUB_PARSE_REGEX_SUBRIP:
1241 result = (gpointer) g_regex_new ("^([ 0-9]){0,3}[0-9]\\s*(\x0d)?\x0a"
1242 "[ 0-9][0-9]:[ 0-9][0-9]:[ 0-9][0-9][,.][ 0-9]{0,2}[0-9]"
1243 " +--> +([ 0-9])?[0-9]:[ 0-9][0-9]:[ 0-9][0-9][,.][ 0-9]{0,2}[0-9]",
1245 if (result == NULL) {
1246 g_warning ("Compilation of subrip regex failed: %s", gerr->message);
1247 g_error_free (gerr);
1250 case GST_SUB_PARSE_REGEX_DKS:
1251 result = (gpointer) g_regex_new ("^\[[0-9]+:[0-9]+:[0-9]+].*",
1253 if (result == NULL) {
1254 g_warning ("Compilation of dks regex failed: %s", gerr->message);
1255 g_error_free (gerr);
1259 GST_WARNING ("Trying to allocate regex of unknown type %u", regtype);
1265 * FIXME: maybe we should pass along a second argument, the preceding
1266 * text buffer, because that is how this originally worked, even though
1267 * I don't really see the use of that.
1270 static GstSubParseFormat
1271 gst_sub_parse_data_format_autodetect (gchar * match_str)
1275 static GOnce mdvd_rx_once = G_ONCE_INIT;
1276 static GOnce subrip_rx_once = G_ONCE_INIT;
1277 static GOnce dks_rx_once = G_ONCE_INIT;
1283 g_once (&mdvd_rx_once,
1284 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1285 (gpointer) GST_SUB_PARSE_REGEX_MDVDSUB);
1286 g_once (&subrip_rx_once,
1287 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1288 (gpointer) GST_SUB_PARSE_REGEX_SUBRIP);
1289 g_once (&dks_rx_once,
1290 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1291 (gpointer) GST_SUB_PARSE_REGEX_DKS);
1293 mdvd_grx = (GRegex *) mdvd_rx_once.retval;
1294 subrip_grx = (GRegex *) subrip_rx_once.retval;
1295 dks_grx = (GRegex *) dks_rx_once.retval;
1297 if (g_regex_match (mdvd_grx, match_str, 0, NULL) == TRUE) {
1298 GST_LOG ("MicroDVD (frame based) format detected");
1299 return GST_SUB_PARSE_FORMAT_MDVDSUB;
1301 if (g_regex_match (subrip_grx, match_str, 0, NULL) == TRUE) {
1302 GST_LOG ("SubRip (time based) format detected");
1303 return GST_SUB_PARSE_FORMAT_SUBRIP;
1305 if (g_regex_match (dks_grx, match_str, 0, NULL) == TRUE) {
1306 GST_LOG ("DKS (time based) format detected");
1307 return GST_SUB_PARSE_FORMAT_DKS;
1310 if (!strncmp (match_str, "FORMAT=TIME", 11)) {
1311 GST_LOG ("MPSub (time based) format detected");
1312 return GST_SUB_PARSE_FORMAT_MPSUB;
1314 #ifndef GST_DISABLE_XML
1315 if (strstr (match_str, "<SAMI>") != NULL ||
1316 strstr (match_str, "<sami>") != NULL) {
1317 GST_LOG ("SAMI (time based) format detected");
1318 return GST_SUB_PARSE_FORMAT_SAMI;
1321 /* we're boldly assuming the first subtitle appears within the first hour */
1322 if (sscanf (match_str, "0:%02u:%02u:", &n1, &n2) == 2 ||
1323 sscanf (match_str, "0:%02u:%02u=", &n1, &n2) == 2 ||
1324 sscanf (match_str, "00:%02u:%02u:", &n1, &n2) == 2 ||
1325 sscanf (match_str, "00:%02u:%02u=", &n1, &n2) == 2 ||
1326 sscanf (match_str, "00:%02u:%02u,%u=", &n1, &n2, &n3) == 3) {
1327 GST_LOG ("TMPlayer (time based) format detected");
1328 return GST_SUB_PARSE_FORMAT_TMPLAYER;
1330 if (sscanf (match_str, "[%u][%u]", &n1, &n2) == 2) {
1331 GST_LOG ("MPL2 (time based) format detected");
1332 return GST_SUB_PARSE_FORMAT_MPL2;
1334 if (strstr (match_str, "[INFORMATION]") != NULL) {
1335 GST_LOG ("SubViewer (time based) format detected");
1336 return GST_SUB_PARSE_FORMAT_SUBVIEWER;
1339 GST_DEBUG ("no subtitle format detected");
1340 return GST_SUB_PARSE_FORMAT_UNKNOWN;
1344 gst_sub_parse_format_autodetect (GstSubParse * self)
1347 GstSubParseFormat format;
1349 if (strlen (self->textbuf->str) < 30) {
1350 GST_DEBUG ("File too small to be a subtitles file");
1354 data = g_strndup (self->textbuf->str, 35);
1355 format = gst_sub_parse_data_format_autodetect (data);
1358 self->parser_type = format;
1359 self->subtitle_codec = gst_sub_parse_get_format_description (format);
1360 parser_state_init (&self->state);
1363 case GST_SUB_PARSE_FORMAT_MDVDSUB:
1364 self->parse_line = parse_mdvdsub;
1365 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1366 case GST_SUB_PARSE_FORMAT_SUBRIP:
1367 self->parse_line = parse_subrip;
1368 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1369 case GST_SUB_PARSE_FORMAT_MPSUB:
1370 self->parse_line = parse_mpsub;
1371 return gst_caps_new_simple ("text/plain", NULL);
1372 #ifndef GST_DISABLE_XML
1373 case GST_SUB_PARSE_FORMAT_SAMI:
1374 self->parse_line = parse_sami;
1375 sami_context_init (&self->state);
1376 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1378 case GST_SUB_PARSE_FORMAT_TMPLAYER:
1379 self->parse_line = parse_tmplayer;
1380 self->state.max_duration = 5 * GST_SECOND;
1381 return gst_caps_new_simple ("text/plain", NULL);
1382 case GST_SUB_PARSE_FORMAT_MPL2:
1383 self->parse_line = parse_mpl2;
1384 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1385 case GST_SUB_PARSE_FORMAT_DKS:
1386 self->parse_line = parse_dks;
1387 return gst_caps_new_simple ("text/plain", NULL);
1388 case GST_SUB_PARSE_FORMAT_SUBVIEWER:
1389 self->parse_line = parse_subviewer;
1390 return gst_caps_new_simple ("text/plain", NULL);
1391 case GST_SUB_PARSE_FORMAT_UNKNOWN:
1393 GST_DEBUG ("no subtitle format detected");
1394 GST_ELEMENT_ERROR (self, STREAM, WRONG_TYPE,
1395 ("The input is not a valid/supported subtitle file"), (NULL));
1401 feed_textbuf (GstSubParse * self, GstBuffer * buf)
1405 gchar *input = NULL;
1407 discont = GST_BUFFER_IS_DISCONT (buf);
1409 if (GST_BUFFER_OFFSET_IS_VALID (buf) &&
1410 GST_BUFFER_OFFSET (buf) != self->offset) {
1411 self->offset = GST_BUFFER_OFFSET (buf);
1416 GST_INFO ("discontinuity");
1417 /* flush the parser state */
1418 parser_state_init (&self->state);
1419 g_string_truncate (self->textbuf, 0);
1420 gst_adapter_clear (self->adapter);
1421 #ifndef GST_DISABLE_XML
1422 sami_context_reset (&self->state);
1424 /* we could set a flag to make sure that the next buffer we push out also
1425 * has the DISCONT flag set, but there's no point really given that it's
1426 * subtitles which are discontinuous by nature. */
1429 self->offset = GST_BUFFER_OFFSET (buf) + GST_BUFFER_SIZE (buf);
1430 self->next_offset = self->offset;
1432 gst_adapter_push (self->adapter, buf);
1435 convert_encoding (self, (const gchar *) gst_adapter_peek (self->adapter,
1436 gst_adapter_available (self->adapter)),
1437 (gsize) gst_adapter_available (self->adapter), &consumed);
1439 if (input && consumed > 0) {
1440 self->textbuf = g_string_append (self->textbuf, input);
1441 gst_adapter_flush (self->adapter, consumed);
1447 static GstFlowReturn
1448 handle_buffer (GstSubParse * self, GstBuffer * buf)
1450 GstFlowReturn ret = GST_FLOW_OK;
1451 GstCaps *caps = NULL;
1452 gchar *line, *subtitle;
1454 if (self->first_buffer) {
1455 self->detected_encoding =
1456 detect_encoding ((gchar *) GST_BUFFER_DATA (buf),
1457 GST_BUFFER_SIZE (buf));
1458 self->first_buffer = FALSE;
1459 self->state.fps_n = self->fps_n;
1460 self->state.fps_d = self->fps_d;
1463 feed_textbuf (self, buf);
1465 /* make sure we know the format */
1466 if (G_UNLIKELY (self->parser_type == GST_SUB_PARSE_FORMAT_UNKNOWN)) {
1467 if (!(caps = gst_sub_parse_format_autodetect (self))) {
1468 return GST_FLOW_UNEXPECTED;
1470 if (!gst_pad_set_caps (self->srcpad, caps)) {
1471 gst_caps_unref (caps);
1472 return GST_FLOW_UNEXPECTED;
1474 gst_caps_unref (caps);
1477 if (self->subtitle_codec != NULL) {
1480 tags = gst_tag_list_new ();
1481 gst_tag_list_add (tags, GST_TAG_MERGE_APPEND, GST_TAG_SUBTITLE_CODEC,
1482 self->subtitle_codec, NULL);
1483 gst_element_found_tags_for_pad (GST_ELEMENT (self), self->srcpad, tags);
1487 while (!self->flushing && (line = get_next_line (self))) {
1490 /* Set segment on our parser state machine */
1491 self->state.segment = &self->segment;
1492 /* Now parse the line, out of segment lines will just return NULL */
1493 GST_LOG_OBJECT (self, "Parsing line '%s'", line + offset);
1494 subtitle = self->parse_line (&self->state, line + offset);
1498 guint subtitle_len = strlen (subtitle);
1500 /* +1 for terminating NUL character */
1501 ret = gst_pad_alloc_buffer_and_set_caps (self->srcpad,
1502 GST_BUFFER_OFFSET_NONE, subtitle_len + 1,
1503 GST_PAD_CAPS (self->srcpad), &buf);
1505 if (ret == GST_FLOW_OK) {
1506 /* copy terminating NUL character as well */
1507 memcpy (GST_BUFFER_DATA (buf), subtitle, subtitle_len + 1);
1508 GST_BUFFER_SIZE (buf) = subtitle_len;
1509 GST_BUFFER_TIMESTAMP (buf) = self->state.start_time;
1510 GST_BUFFER_DURATION (buf) = self->state.duration;
1512 /* in some cases (e.g. tmplayer) we can only determine the duration
1513 * of a text chunk from the timestamp of the next text chunk; in those
1514 * cases, we probably want to limit the duration to something
1515 * reasonable, so we don't end up showing some text for e.g. 40 seconds
1516 * just because nothing else is being said during that time */
1517 if (self->state.max_duration > 0 && GST_BUFFER_DURATION_IS_VALID (buf)) {
1518 if (GST_BUFFER_DURATION (buf) > self->state.max_duration)
1519 GST_BUFFER_DURATION (buf) = self->state.max_duration;
1522 gst_segment_set_last_stop (&self->segment, GST_FORMAT_TIME,
1523 self->state.start_time);
1525 GST_DEBUG_OBJECT (self, "Sending text '%s', %" GST_TIME_FORMAT " + %"
1526 GST_TIME_FORMAT, subtitle, GST_TIME_ARGS (self->state.start_time),
1527 GST_TIME_ARGS (self->state.duration));
1529 ret = gst_pad_push (self->srcpad, buf);
1532 /* move this forward (the tmplayer parser needs this) */
1533 if (self->state.duration != GST_CLOCK_TIME_NONE)
1534 self->state.start_time += self->state.duration;
1539 if (ret != GST_FLOW_OK) {
1540 GST_DEBUG_OBJECT (self, "flow: %s", gst_flow_get_name (ret));
1549 static GstFlowReturn
1550 gst_sub_parse_chain (GstPad * sinkpad, GstBuffer * buf)
1555 self = GST_SUBPARSE (GST_PAD_PARENT (sinkpad));
1557 /* Push newsegment if needed */
1558 if (self->need_segment) {
1559 GST_LOG_OBJECT (self, "pushing newsegment event with %" GST_SEGMENT_FORMAT,
1562 gst_pad_push_event (self->srcpad, gst_event_new_new_segment (FALSE,
1563 self->segment.rate, self->segment.format,
1564 self->segment.last_stop, self->segment.stop, self->segment.time));
1565 self->need_segment = FALSE;
1568 ret = handle_buffer (self, buf);
1574 gst_sub_parse_sink_event (GstPad * pad, GstEvent * event)
1576 GstSubParse *self = GST_SUBPARSE (gst_pad_get_parent (pad));
1577 gboolean ret = FALSE;
1579 GST_DEBUG ("Handling %s event", GST_EVENT_TYPE_NAME (event));
1581 switch (GST_EVENT_TYPE (event)) {
1582 case GST_EVENT_EOS:{
1583 /* Make sure the last subrip chunk is pushed out even
1584 * if the file does not have an empty line at the end */
1585 if (self->parser_type == GST_SUB_PARSE_FORMAT_SUBRIP ||
1586 self->parser_type == GST_SUB_PARSE_FORMAT_TMPLAYER ||
1587 self->parser_type == GST_SUB_PARSE_FORMAT_MPL2) {
1588 GstBuffer *buf = gst_buffer_new_and_alloc (2 + 1);
1590 GST_DEBUG ("EOS. Pushing remaining text (if any)");
1591 GST_BUFFER_DATA (buf)[0] = '\n';
1592 GST_BUFFER_DATA (buf)[1] = '\n';
1593 GST_BUFFER_DATA (buf)[2] = '\0'; /* play it safe */
1594 GST_BUFFER_SIZE (buf) = 2;
1595 GST_BUFFER_OFFSET (buf) = self->offset;
1596 gst_sub_parse_chain (pad, buf);
1598 ret = gst_pad_event_default (pad, event);
1601 case GST_EVENT_NEWSEGMENT:
1605 gint64 start, stop, time;
1608 gst_event_parse_new_segment (event, &update, &rate, &format, &start,
1611 GST_DEBUG_OBJECT (self, "newsegment (%s)", gst_format_get_name (format));
1613 if (format == GST_FORMAT_TIME) {
1614 gst_segment_set_newsegment (&self->segment, update, rate, format,
1617 /* if not time format, we'll either start with a 0 timestamp anyway or
1618 * it's following a seek in which case we'll have saved the requested
1619 * seek segment and don't want to overwrite it (remember that on a seek
1620 * we always just seek back to the start in BYTES format and just throw
1621 * away all text that's before the requested position; if the subtitles
1622 * come from an upstream demuxer, it won't be able to handle our BYTES
1623 * seek request and instead send us a newsegment from the seek request
1624 * it received via its video pads instead, so all is fine then too) */
1628 gst_event_unref (event);
1631 case GST_EVENT_FLUSH_START:
1633 self->flushing = TRUE;
1635 ret = gst_pad_event_default (pad, event);
1638 case GST_EVENT_FLUSH_STOP:
1640 self->flushing = FALSE;
1642 ret = gst_pad_event_default (pad, event);
1646 ret = gst_pad_event_default (pad, event);
1650 gst_object_unref (self);
1656 static GstStateChangeReturn
1657 gst_sub_parse_change_state (GstElement * element, GstStateChange transition)
1659 GstStateChangeReturn ret = GST_STATE_CHANGE_SUCCESS;
1660 GstSubParse *self = GST_SUBPARSE (element);
1662 switch (transition) {
1663 case GST_STATE_CHANGE_READY_TO_PAUSED:
1664 /* format detection will init the parser state */
1666 self->next_offset = 0;
1667 self->parser_type = GST_SUB_PARSE_FORMAT_UNKNOWN;
1668 self->valid_utf8 = TRUE;
1669 self->first_buffer = TRUE;
1670 g_free (self->detected_encoding);
1671 self->detected_encoding = NULL;
1672 g_string_truncate (self->textbuf, 0);
1673 gst_adapter_clear (self->adapter);
1679 ret = parent_class->change_state (element, transition);
1680 if (ret == GST_STATE_CHANGE_FAILURE)
1683 switch (transition) {
1684 case GST_STATE_CHANGE_PAUSED_TO_READY:
1685 parser_state_dispose (&self->state);
1686 self->parser_type = GST_SUB_PARSE_FORMAT_UNKNOWN;
1699 /* FIXME 0.11: these caps are ugly, use app/x-subtitle + type field or so;
1700 * also, give different subtitle formats really different types */
1701 static GstStaticCaps mpl2_caps =
1702 GST_STATIC_CAPS ("application/x-subtitle-mpl2");
1703 #define SUB_CAPS (gst_static_caps_get (&sub_caps))
1705 static GstStaticCaps tmp_caps =
1706 GST_STATIC_CAPS ("application/x-subtitle-tmplayer");
1707 #define TMP_CAPS (gst_static_caps_get (&tmp_caps))
1709 static GstStaticCaps sub_caps = GST_STATIC_CAPS ("application/x-subtitle");
1710 #define MPL2_CAPS (gst_static_caps_get (&mpl2_caps))
1712 #ifndef GST_DISABLE_XML
1713 static GstStaticCaps smi_caps = GST_STATIC_CAPS ("application/x-subtitle-sami");
1714 #define SAMI_CAPS (gst_static_caps_get (&smi_caps))
1717 static GstStaticCaps dks_caps = GST_STATIC_CAPS ("application/x-subtitle-dks");
1718 #define DKS_CAPS (gst_static_caps_get (&dks_caps))
1721 gst_subparse_type_find (GstTypeFind * tf, gpointer private)
1723 GstSubParseFormat format;
1727 gchar *encoding = NULL;
1730 if (!(data = gst_type_find_peek (tf, 0, 129)))
1733 /* make sure string passed to _autodetect() is NUL-terminated */
1734 str = g_malloc0 (129);
1735 memcpy (str, data, 128);
1737 if ((encoding = detect_encoding (str, 128)) != NULL) {
1738 gchar *converted_str;
1742 converted_str = gst_convert_to_utf8 (str, 128, encoding, &tmp, &err);
1743 if (converted_str == NULL) {
1744 GST_DEBUG ("Encoding '%s' detected but conversion failed: %s", encoding,
1750 str = converted_str;
1755 /* Check if at least the first 120 chars are valid UTF8,
1756 * otherwise convert as always */
1757 if (!g_utf8_validate (str, 128, &end) && (end - str) < 120) {
1758 gchar *converted_str;
1763 enc = g_getenv ("GST_SUBTITLE_ENCODING");
1764 if (enc == NULL || *enc == '\0') {
1765 /* if local encoding is UTF-8 and no encoding specified
1766 * via the environment variable, assume ISO-8859-15 */
1767 if (g_get_charset (&enc)) {
1768 enc = "ISO-8859-15";
1771 converted_str = gst_convert_to_utf8 (str, 128, enc, &tmp, &err);
1772 if (converted_str == NULL) {
1773 GST_DEBUG ("Charset conversion failed: %s", err->message);
1779 str = converted_str;
1783 format = gst_sub_parse_data_format_autodetect (str);
1787 case GST_SUB_PARSE_FORMAT_MDVDSUB:
1788 GST_DEBUG ("MicroDVD format detected");
1791 case GST_SUB_PARSE_FORMAT_SUBRIP:
1792 GST_DEBUG ("SubRip format detected");
1795 case GST_SUB_PARSE_FORMAT_MPSUB:
1796 GST_DEBUG ("MPSub format detected");
1799 #ifndef GST_DISABLE_XML
1800 case GST_SUB_PARSE_FORMAT_SAMI:
1801 GST_DEBUG ("SAMI (time-based) format detected");
1805 case GST_SUB_PARSE_FORMAT_TMPLAYER:
1806 GST_DEBUG ("TMPlayer (time based) format detected");
1809 /* FIXME: our MPL2 typefinding is not really good enough to warrant
1810 * returning a high probability (however, since we registered our
1811 * typefinder here with a rank of MARGINAL we should pretty much only
1812 * be called if most other typefinders have already run */
1813 case GST_SUB_PARSE_FORMAT_MPL2:
1814 GST_DEBUG ("MPL2 (time based) format detected");
1817 case GST_SUB_PARSE_FORMAT_SUBVIEWER:
1818 GST_DEBUG ("SubViewer format detected");
1821 case GST_SUB_PARSE_FORMAT_DKS:
1822 GST_DEBUG ("DKS format detected");
1826 case GST_SUB_PARSE_FORMAT_UNKNOWN:
1827 GST_DEBUG ("no subtitle format detected");
1831 /* if we're here, it's ok */
1832 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, caps);
1836 plugin_init (GstPlugin * plugin)
1838 static gchar *sub_exts[] = { "srt", "sub", "mpsub", "mdvd", "smi", "txt",
1842 GST_DEBUG_CATEGORY_INIT (sub_parse_debug, "subparse", 0, ".sub parser");
1844 if (!gst_type_find_register (plugin, "subparse_typefind", GST_RANK_MARGINAL,
1845 gst_subparse_type_find, sub_exts, SUB_CAPS, NULL, NULL))
1848 if (!gst_element_register (plugin, "subparse",
1849 GST_RANK_PRIMARY, GST_TYPE_SUBPARSE) ||
1850 !gst_element_register (plugin, "ssaparse",
1851 GST_RANK_PRIMARY, GST_TYPE_SSA_PARSE)) {
1858 GST_PLUGIN_DEFINE (GST_VERSION_MAJOR,
1862 plugin_init, VERSION, "LGPL", GST_PACKAGE_NAME, GST_PACKAGE_ORIGIN)