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")
82 static void gst_sub_parse_base_init (GstSubParseClass * klass);
83 static void gst_sub_parse_class_init (GstSubParseClass * klass);
84 static void gst_sub_parse_init (GstSubParse * subparse);
86 static gboolean gst_sub_parse_src_event (GstPad * pad, GstEvent * event);
87 static gboolean gst_sub_parse_src_query (GstPad * pad, GstQuery * query);
88 static gboolean gst_sub_parse_sink_event (GstPad * pad, GstEvent * event);
90 static GstStateChangeReturn gst_sub_parse_change_state (GstElement * element,
91 GstStateChange transition);
93 static GstFlowReturn gst_sub_parse_chain (GstPad * sinkpad, GstBuffer * buf);
95 static GstElementClass *parent_class = NULL;
98 gst_sub_parse_get_type (void)
100 static GType sub_parse_type = 0;
102 if (!sub_parse_type) {
103 static const GTypeInfo sub_parse_info = {
104 sizeof (GstSubParseClass),
105 (GBaseInitFunc) gst_sub_parse_base_init,
107 (GClassInitFunc) gst_sub_parse_class_init,
110 sizeof (GstSubParse),
112 (GInstanceInitFunc) gst_sub_parse_init,
115 sub_parse_type = g_type_register_static (GST_TYPE_ELEMENT,
116 "GstSubParse", &sub_parse_info, 0);
119 return sub_parse_type;
123 gst_sub_parse_base_init (GstSubParseClass * klass)
125 GstElementClass *element_class = GST_ELEMENT_CLASS (klass);
127 gst_element_class_add_pad_template (element_class,
128 gst_static_pad_template_get (&sink_templ));
129 gst_element_class_add_pad_template (element_class,
130 gst_static_pad_template_get (&src_templ));
131 gst_element_class_set_details_simple (element_class,
132 "Subtitle parser", "Codec/Parser/Subtitle",
133 "Parses subtitle (.sub) files into text streams",
134 "Gustavo J. A. M. Carneiro <gjc@inescporto.pt>, "
135 "GStreamer maintainers <gstreamer-devel@lists.sourceforge.net>");
139 gst_sub_parse_dispose (GObject * object)
141 GstSubParse *subparse = GST_SUBPARSE (object);
143 GST_DEBUG_OBJECT (subparse, "cleaning up subtitle parser");
145 switch (subparse->parser_type) {
146 case GST_SUB_PARSE_FORMAT_QTTEXT:
147 qttext_context_deinit (&subparse->state);
149 #ifndef GST_DISABLE_XML
150 case GST_SUB_PARSE_FORMAT_SAMI:
151 sami_context_deinit (&subparse->state);
158 if (subparse->encoding) {
159 g_free (subparse->encoding);
160 subparse->encoding = NULL;
163 if (subparse->detected_encoding) {
164 g_free (subparse->detected_encoding);
165 subparse->detected_encoding = NULL;
168 if (subparse->adapter) {
169 g_object_unref (subparse->adapter);
170 subparse->adapter = NULL;
173 if (subparse->textbuf) {
174 g_string_free (subparse->textbuf, TRUE);
175 subparse->textbuf = NULL;
178 GST_CALL_PARENT (G_OBJECT_CLASS, dispose, (object));
182 gst_sub_parse_class_init (GstSubParseClass * klass)
184 GObjectClass *object_class = G_OBJECT_CLASS (klass);
185 GstElementClass *element_class = GST_ELEMENT_CLASS (klass);
187 parent_class = g_type_class_peek_parent (klass);
189 object_class->dispose = gst_sub_parse_dispose;
190 object_class->set_property = gst_sub_parse_set_property;
191 object_class->get_property = gst_sub_parse_get_property;
193 element_class->change_state = gst_sub_parse_change_state;
195 g_object_class_install_property (object_class, PROP_ENCODING,
196 g_param_spec_string ("subtitle-encoding", "subtitle charset encoding",
197 "Encoding to assume if input subtitles are not in UTF-8 or any other "
198 "Unicode encoding. If not set, the GST_SUBTITLE_ENCODING environment "
199 "variable will be checked for an encoding to use. If that is not set "
200 "either, ISO-8859-15 will be assumed.", DEFAULT_ENCODING,
201 G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
203 g_object_class_install_property (object_class, PROP_VIDEOFPS,
204 gst_param_spec_fraction ("video-fps", "Video framerate",
205 "Framerate of the video stream. This is needed by some subtitle "
206 "formats to synchronize subtitles and video properly. If not set "
207 "and the subtitle format requires it subtitles may be out of sync.",
208 0, 1, G_MAXINT, 1, 24000, 1001,
209 G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
213 gst_sub_parse_init (GstSubParse * subparse)
215 subparse->sinkpad = gst_pad_new_from_static_template (&sink_templ, "sink");
216 gst_pad_set_chain_function (subparse->sinkpad,
217 GST_DEBUG_FUNCPTR (gst_sub_parse_chain));
218 gst_pad_set_event_function (subparse->sinkpad,
219 GST_DEBUG_FUNCPTR (gst_sub_parse_sink_event));
220 gst_element_add_pad (GST_ELEMENT (subparse), subparse->sinkpad);
222 subparse->srcpad = gst_pad_new_from_static_template (&src_templ, "src");
223 gst_pad_set_event_function (subparse->srcpad,
224 GST_DEBUG_FUNCPTR (gst_sub_parse_src_event));
225 gst_pad_set_query_function (subparse->srcpad,
226 GST_DEBUG_FUNCPTR (gst_sub_parse_src_query));
227 gst_element_add_pad (GST_ELEMENT (subparse), subparse->srcpad);
229 subparse->textbuf = g_string_new (NULL);
230 subparse->parser_type = GST_SUB_PARSE_FORMAT_UNKNOWN;
231 subparse->flushing = FALSE;
232 gst_segment_init (&subparse->segment, GST_FORMAT_TIME);
233 subparse->need_segment = TRUE;
234 subparse->encoding = g_strdup (DEFAULT_ENCODING);
235 subparse->detected_encoding = NULL;
236 subparse->adapter = gst_adapter_new ();
238 subparse->fps_n = 24000;
239 subparse->fps_d = 1001;
243 * Source pad functions.
247 gst_sub_parse_src_query (GstPad * pad, GstQuery * query)
249 GstSubParse *self = GST_SUBPARSE (gst_pad_get_parent (pad));
250 gboolean ret = FALSE;
252 GST_DEBUG ("Handling %s query", GST_QUERY_TYPE_NAME (query));
254 switch (GST_QUERY_TYPE (query)) {
255 case GST_QUERY_POSITION:{
258 gst_query_parse_position (query, &fmt, NULL);
259 if (fmt != GST_FORMAT_TIME) {
260 ret = gst_pad_peer_query (self->sinkpad, query);
263 gst_query_set_position (query, GST_FORMAT_TIME,
264 self->segment.last_stop);
267 case GST_QUERY_SEEKING:
270 gboolean seekable = FALSE;
274 gst_query_parse_seeking (query, &fmt, NULL, NULL, NULL);
275 if (fmt == GST_FORMAT_TIME) {
276 GstQuery *peerquery = gst_query_new_seeking (GST_FORMAT_BYTES);
278 seekable = gst_pad_peer_query (self->sinkpad, peerquery);
280 gst_query_parse_seeking (peerquery, NULL, &seekable, NULL, NULL);
281 gst_query_unref (peerquery);
284 gst_query_set_seeking (query, fmt, seekable, seekable ? 0 : -1, -1);
289 ret = gst_pad_peer_query (self->sinkpad, query);
293 gst_object_unref (self);
299 gst_sub_parse_src_event (GstPad * pad, GstEvent * event)
301 GstSubParse *self = GST_SUBPARSE (gst_pad_get_parent (pad));
302 gboolean ret = FALSE;
304 GST_DEBUG ("Handling %s event", GST_EVENT_TYPE_NAME (event));
306 switch (GST_EVENT_TYPE (event)) {
310 GstSeekType start_type, stop_type;
315 gst_event_parse_seek (event, &rate, &format, &self->segment_flags,
316 &start_type, &start, &stop_type, &stop);
318 if (format != GST_FORMAT_TIME) {
319 GST_WARNING_OBJECT (self, "we only support seeking in TIME format");
320 gst_event_unref (event);
324 /* Convert that seek to a seeking in bytes at position 0,
325 FIXME: could use an index */
326 ret = gst_pad_push_event (self->sinkpad,
327 gst_event_new_seek (rate, GST_FORMAT_BYTES, self->segment_flags,
328 GST_SEEK_TYPE_SET, 0, GST_SEEK_TYPE_NONE, 0));
331 /* Apply the seek to our segment */
332 gst_segment_set_seek (&self->segment, rate, format, self->segment_flags,
333 start_type, start, stop_type, stop, &update);
335 GST_DEBUG_OBJECT (self, "segment after seek: %" GST_SEGMENT_FORMAT,
338 self->next_offset = 0;
340 self->need_segment = TRUE;
342 GST_WARNING_OBJECT (self, "seek to 0 bytes failed");
345 gst_event_unref (event);
349 ret = gst_pad_event_default (pad, event);
354 gst_object_unref (self);
360 gst_sub_parse_set_property (GObject * object, guint prop_id,
361 const GValue * value, GParamSpec * pspec)
363 GstSubParse *subparse = GST_SUBPARSE (object);
365 GST_OBJECT_LOCK (subparse);
368 g_free (subparse->encoding);
369 subparse->encoding = g_value_dup_string (value);
370 GST_LOG_OBJECT (object, "subtitle encoding set to %s",
371 GST_STR_NULL (subparse->encoding));
375 subparse->fps_n = gst_value_get_fraction_numerator (value);
376 subparse->fps_d = gst_value_get_fraction_denominator (value);
377 GST_DEBUG_OBJECT (object, "video framerate set to %d/%d", subparse->fps_n,
380 if (!subparse->state.have_internal_fps) {
381 subparse->state.fps_n = subparse->fps_n;
382 subparse->state.fps_d = subparse->fps_d;
387 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
390 GST_OBJECT_UNLOCK (subparse);
394 gst_sub_parse_get_property (GObject * object, guint prop_id,
395 GValue * value, GParamSpec * pspec)
397 GstSubParse *subparse = GST_SUBPARSE (object);
399 GST_OBJECT_LOCK (subparse);
402 g_value_set_string (value, subparse->encoding);
405 gst_value_set_fraction (value, subparse->fps_n, subparse->fps_d);
408 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
411 GST_OBJECT_UNLOCK (subparse);
415 gst_sub_parse_get_format_description (GstSubParseFormat format)
418 case GST_SUB_PARSE_FORMAT_MDVDSUB:
420 case GST_SUB_PARSE_FORMAT_SUBRIP:
422 case GST_SUB_PARSE_FORMAT_MPSUB:
424 case GST_SUB_PARSE_FORMAT_SAMI:
426 case GST_SUB_PARSE_FORMAT_TMPLAYER:
428 case GST_SUB_PARSE_FORMAT_MPL2:
430 case GST_SUB_PARSE_FORMAT_SUBVIEWER:
432 case GST_SUB_PARSE_FORMAT_DKS:
434 case GST_SUB_PARSE_FORMAT_QTTEXT:
437 case GST_SUB_PARSE_FORMAT_UNKNOWN:
444 gst_convert_to_utf8 (const gchar * str, gsize len, const gchar * encoding,
445 gsize * consumed, GError ** err)
450 /* The char cast is necessary in glib < 2.24 */
452 g_convert_with_fallback (str, len, "UTF-8", encoding, (char *) "*",
453 consumed, NULL, err);
457 /* + 3 to skip UTF-8 BOM if it was added */
459 if (len >= 3 && (guint8) ret[0] == 0xEF && (guint8) ret[1] == 0xBB
460 && (guint8) ret[2] == 0xBF)
461 g_memmove (ret, ret + 3, len + 1 - 3);
467 detect_encoding (const gchar * str, gsize len)
469 if (len >= 3 && (guint8) str[0] == 0xEF && (guint8) str[1] == 0xBB
470 && (guint8) str[2] == 0xBF)
471 return g_strdup ("UTF-8");
473 if (len >= 2 && (guint8) str[0] == 0xFE && (guint8) str[1] == 0xFF)
474 return g_strdup ("UTF-16BE");
476 if (len >= 2 && (guint8) str[0] == 0xFF && (guint8) str[1] == 0xFE)
477 return g_strdup ("UTF-16LE");
479 if (len >= 4 && (guint8) str[0] == 0x00 && (guint8) str[1] == 0x00
480 && (guint8) str[2] == 0xFE && (guint8) str[3] == 0xFF)
481 return g_strdup ("UTF-32BE");
483 if (len >= 4 && (guint8) str[0] == 0xFF && (guint8) str[1] == 0xFE
484 && (guint8) str[2] == 0x00 && (guint8) str[3] == 0x00)
485 return g_strdup ("UTF-32LE");
491 convert_encoding (GstSubParse * self, const gchar * str, gsize len,
494 const gchar *encoding;
500 /* First try any detected encoding */
501 if (self->detected_encoding) {
503 gst_convert_to_utf8 (str, len, self->detected_encoding, consumed, &err);
508 GST_WARNING_OBJECT (self, "could not convert string from '%s' to UTF-8: %s",
509 self->detected_encoding, err->message);
510 g_free (self->detected_encoding);
511 self->detected_encoding = NULL;
515 /* Otherwise check if it's UTF8 */
516 if (self->valid_utf8) {
517 if (g_utf8_validate (str, len, NULL)) {
518 GST_LOG_OBJECT (self, "valid UTF-8, no conversion needed");
520 return g_strndup (str, len);
522 GST_INFO_OBJECT (self, "invalid UTF-8!");
523 self->valid_utf8 = FALSE;
526 /* Else try fallback */
527 encoding = self->encoding;
528 if (encoding == NULL || *encoding == '\0') {
529 encoding = g_getenv ("GST_SUBTITLE_ENCODING");
531 if (encoding == NULL || *encoding == '\0') {
532 /* if local encoding is UTF-8 and no encoding specified
533 * via the environment variable, assume ISO-8859-15 */
534 if (g_get_charset (&encoding)) {
535 encoding = "ISO-8859-15";
539 ret = gst_convert_to_utf8 (str, len, encoding, consumed, &err);
542 GST_WARNING_OBJECT (self, "could not convert string from '%s' to UTF-8: %s",
543 encoding, err->message);
546 /* invalid input encoding, fall back to ISO-8859-15 (always succeeds) */
547 ret = gst_convert_to_utf8 (str, len, "ISO-8859-15", consumed, NULL);
550 GST_LOG_OBJECT (self,
551 "successfully converted %" G_GSIZE_FORMAT " characters from %s to UTF-8"
552 "%s", len, encoding, (err) ? " , using ISO-8859-15 as fallback" : "");
558 get_next_line (GstSubParse * self)
561 const char *line_end;
563 gboolean have_r = FALSE;
565 line_end = strchr (self->textbuf->str, '\n');
568 /* end-of-line not found; return for more data */
572 /* get rid of '\r' */
573 if (line_end != self->textbuf->str && *(line_end - 1) == '\r') {
578 line_len = line_end - self->textbuf->str;
579 line = g_strndup (self->textbuf->str, line_len);
580 self->textbuf = g_string_erase (self->textbuf, 0,
581 line_len + (have_r ? 2 : 1));
586 parse_mdvdsub (ParserState * state, const gchar * line)
588 const gchar *line_split;
590 guint start_frame, end_frame;
591 gint64 clip_start = 0, clip_stop = 0;
592 gboolean in_seg = FALSE;
596 /* style variables */
602 if (sscanf (line, "{%u}{%u}", &start_frame, &end_frame) != 2) {
603 g_warning ("Parse of the following line, assumed to be in microdvd .sub"
604 " format, failed:\n%s", line);
608 /* skip the {%u}{%u} part */
609 line = strchr (line, '}') + 1;
610 line = strchr (line, '}') + 1;
612 /* see if there's a first line with a framerate */
613 if (start_frame == 1 && end_frame == 1) {
614 gchar *rest, *end = NULL;
616 rest = g_strdup (line);
617 g_strdelimit (rest, ",", '.');
618 fps = g_ascii_strtod (rest, &end);
620 gst_util_double_to_fraction (fps, &state->fps_n, &state->fps_d);
621 GST_INFO ("framerate from file: %d/%d ('%s')", state->fps_n,
629 gst_util_uint64_scale (start_frame, GST_SECOND * state->fps_d,
632 gst_util_uint64_scale (end_frame - start_frame, GST_SECOND * state->fps_d,
635 /* Check our segment start/stop */
636 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
637 state->start_time, state->start_time + state->duration, &clip_start,
640 /* No need to parse that text if it's out of segment */
642 state->start_time = clip_start;
643 state->duration = clip_stop - clip_start;
648 markup = g_string_new (NULL);
653 /* parse style markup */
654 if (strncmp (line, "{y:i}", 5) == 0) {
656 line = strchr (line, '}') + 1;
658 if (strncmp (line, "{y:b}", 5) == 0) {
660 line = strchr (line, '}') + 1;
662 if (sscanf (line, "{s:%u}", &fontsize) == 1) {
663 line = strchr (line, '}') + 1;
665 /* forward slashes at beginning/end signify italics too */
666 if (g_str_has_prefix (line, "/")) {
670 if ((line_split = strchr (line, '|')))
671 line_chunk = g_markup_escape_text (line, line_split - line);
673 line_chunk = g_markup_escape_text (line, strlen (line));
675 /* Remove italics markers at end of line/stanza (CHECKME: are end slashes
676 * always at the end of a line or can they span multiple lines?) */
677 if (g_str_has_suffix (line_chunk, "/")) {
678 line_chunk[strlen (line_chunk) - 1] = '\0';
681 markup = g_string_append (markup, "<span");
683 g_string_append (markup, " style=\"italic\"");
685 g_string_append (markup, " weight=\"bold\"");
687 g_string_append_printf (markup, " size=\"%u\"", fontsize * 1000);
688 g_string_append_printf (markup, ">%s</span>", line_chunk);
691 g_string_append (markup, "\n");
692 line = line_split + 1;
698 g_string_free (markup, FALSE);
699 GST_DEBUG ("parse_mdvdsub returning (%f+%f): %s",
700 state->start_time / (double) GST_SECOND,
701 state->duration / (double) GST_SECOND, ret);
706 strip_trailing_newlines (gchar * txt)
712 while (len > 1 && txt[len - 1] == '\n') {
719 /* we want to escape text in general, but retain basic markup like
720 * <i></i>, <u></u>, and <b></b>. The easiest and safest way is to
721 * just unescape a white list of allowed markups again after
722 * escaping everything (the text between these simple markers isn't
723 * necessarily escaped, so it seems best to do it like this) */
725 subrip_unescape_formatting (gchar * txt)
729 for (pos = txt; pos != NULL && *pos != '\0'; ++pos) {
730 if (g_ascii_strncasecmp (pos, "<u>", 9) == 0 ||
731 g_ascii_strncasecmp (pos, "<i>", 9) == 0 ||
732 g_ascii_strncasecmp (pos, "<b>", 9) == 0) {
734 pos[1] = g_ascii_tolower (pos[4]);
736 /* move NUL terminator as well */
737 g_memmove (pos + 3, pos + 9, strlen (pos + 9) + 1);
742 for (pos = txt; pos != NULL && *pos != '\0'; ++pos) {
743 if (g_ascii_strncasecmp (pos, "</u>", 10) == 0 ||
744 g_ascii_strncasecmp (pos, "</i>", 10) == 0 ||
745 g_ascii_strncasecmp (pos, "</b>", 10) == 0) {
748 pos[2] = g_ascii_tolower (pos[5]);
750 /* move NUL terminator as well */
751 g_memmove (pos + 4, pos + 10, strlen (pos + 10) + 1);
759 subrip_remove_unhandled_tag (gchar * start, gchar * stop)
763 tag = start + strlen ("<");
767 if (g_ascii_tolower (*tag) < 'a' || g_ascii_tolower (*tag) > 'z')
772 GST_LOG ("removing unhandled tag '%s'", start);
774 g_memmove (start, stop, strlen (stop) + 1);
778 /* remove tags we haven't explicitly allowed earlier on, like font tags
781 subrip_remove_unhandled_tags (gchar * txt)
785 for (pos = txt; pos != NULL && *pos != '\0'; ++pos) {
786 if (strncmp (pos, "<", 4) == 0 && (gt = strstr (pos + 4, ">"))) {
787 if (subrip_remove_unhandled_tag (pos, gt + strlen (">")))
793 /* we only allow <i>, <u> and <b>, so let's take a simple approach. This code
794 * assumes the input has been escaped and subrip_unescape_formatting() has then
795 * been run over the input! This function adds missing closing markup tags and
796 * removes broken closing tags for tags that have never been opened. */
798 subrip_fix_up_markup (gchar ** p_txt)
800 gchar *cur, *next_tag;
802 guint num_open_tags = 0;
804 g_assert (*p_txt != NULL);
807 while (*cur != '\0') {
808 next_tag = strchr (cur, '<');
809 if (next_tag == NULL)
815 if (num_open_tags == 0 || open_tags[num_open_tags - 1] != *next_tag) {
816 GST_LOG ("broken input, closing tag '%c' is not open", *next_tag);
817 g_memmove (next_tag - 2, next_tag + 2, strlen (next_tag + 2) + 1);
820 /* it's all good, closing tag which is open */
828 if (num_open_tags == G_N_ELEMENTS (open_tags))
829 return; /* something dodgy is going on, stop parsing */
830 open_tags[num_open_tags] = *next_tag;
834 GST_ERROR ("unexpected tag '%c' (%s)", *next_tag, next_tag);
835 g_assert_not_reached ();
841 if (num_open_tags > 0) {
844 s = g_string_new (*p_txt);
845 while (num_open_tags > 0) {
846 GST_LOG ("adding missing closing tag '%c'", open_tags[num_open_tags - 1]);
847 g_string_append_c (s, '<');
848 g_string_append_c (s, '/');
849 g_string_append_c (s, open_tags[num_open_tags - 1]);
850 g_string_append_c (s, '>');
854 *p_txt = g_string_free (s, FALSE);
859 parse_subrip_time (const gchar * ts_string, GstClockTime * t)
861 gchar s[128] = { '\0', };
863 guint hour, min, sec, msec, len;
865 while (*ts_string == ' ')
868 g_strlcpy (s, ts_string, sizeof (s));
869 if ((end = strstr (s, "-->")))
873 /* ms may be in these formats:
874 * hh:mm:ss,500 = 500ms
877 * hh:mm:ss, 50 = 50ms
879 * and the same with . instead of ,.
880 * sscanf() doesn't differentiate between ' 5' and '5' so munge
881 * the white spaces within the timestamp to '0' (I'm sure there's a
882 * way to make sscanf() do this for us, but how?)
884 g_strdelimit (s, " ", '0');
885 g_strdelimit (s, ".", ',');
887 /* make sure we have exactly three digits after he comma */
889 g_assert (p != NULL);
896 g_strlcat (&p[len], "0", 2);
900 GST_LOG ("parsing timestamp '%s'", s);
901 if (sscanf (s, "%u:%u:%u,%u", &hour, &min, &sec, &msec) != 4) {
902 GST_WARNING ("failed to parse subrip timestamp string '%s'", s);
906 *t = ((hour * 3600) + (min * 60) + sec) * GST_SECOND + msec * GST_MSECOND;
911 parse_subrip (ParserState * state, const gchar * line)
916 switch (state->state) {
918 /* looking for a single integer */
919 if (sscanf (line, "%u", &subnum) == 1)
924 GstClockTime ts_start, ts_end;
927 /* looking for start_time --> end_time */
928 if ((end_time = strstr (line, " --> ")) &&
929 parse_subrip_time (line, &ts_start) &&
930 parse_subrip_time (end_time + strlen (" --> "), &ts_end) &&
931 state->start_time <= ts_end) {
933 state->start_time = ts_start;
934 state->duration = ts_end - ts_start;
936 GST_DEBUG ("error parsing subrip time line '%s'", line);
943 /* No need to parse that text if it's out of segment */
944 gint64 clip_start = 0, clip_stop = 0;
945 gboolean in_seg = FALSE;
947 /* Check our segment start/stop */
948 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
949 state->start_time, state->start_time + state->duration,
950 &clip_start, &clip_stop);
953 state->start_time = clip_start;
954 state->duration = clip_stop - clip_start;
960 /* looking for subtitle text; empty line ends this subtitle entry */
962 g_string_append_c (state->buf, '\n');
963 g_string_append (state->buf, line);
964 if (strlen (line) == 0) {
965 ret = g_markup_escape_text (state->buf->str, state->buf->len);
966 g_string_truncate (state->buf, 0);
968 subrip_unescape_formatting (ret);
969 subrip_remove_unhandled_tags (ret);
970 strip_trailing_newlines (ret);
971 subrip_fix_up_markup (&ret);
976 g_return_val_if_reached (NULL);
981 unescape_newlines_br (gchar * read)
985 /* Replace all occurences of '[br]' with a newline as version 2
986 * of the subviewer format uses this for newlines */
988 if (read[0] == '\0' || read[1] == '\0' || read[2] == '\0' || read[3] == '\0')
992 if (strncmp (read, "[br]", 4) == 0) {
1006 parse_subviewer (ParserState * state, const gchar * line)
1008 guint h1, m1, s1, ms1;
1009 guint h2, m2, s2, ms2;
1012 /* TODO: Maybe also parse the fields in the header, especially DELAY.
1013 * For examples see the unit test or
1014 * http://www.doom9.org/index.html?/sub.htm */
1016 switch (state->state) {
1018 /* looking for start_time,end_time */
1019 if (sscanf (line, "%u:%u:%u.%u,%u:%u:%u.%u",
1020 &h1, &m1, &s1, &ms1, &h2, &m2, &s2, &ms2) == 8) {
1023 (((guint64) h1) * 3600 + m1 * 60 + s1) * GST_SECOND +
1026 (((guint64) h2) * 3600 + m2 * 60 + s2) * GST_SECOND +
1027 ms2 * GST_MSECOND - state->start_time;
1032 /* No need to parse that text if it's out of segment */
1033 gint64 clip_start = 0, clip_stop = 0;
1034 gboolean in_seg = FALSE;
1036 /* Check our segment start/stop */
1037 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1038 state->start_time, state->start_time + state->duration,
1039 &clip_start, &clip_stop);
1042 state->start_time = clip_start;
1043 state->duration = clip_stop - clip_start;
1049 /* looking for subtitle text; empty line ends this subtitle entry */
1050 if (state->buf->len)
1051 g_string_append_c (state->buf, '\n');
1052 g_string_append (state->buf, line);
1053 if (strlen (line) == 0) {
1054 ret = g_strdup (state->buf->str);
1055 unescape_newlines_br (ret);
1056 strip_trailing_newlines (ret);
1057 g_string_truncate (state->buf, 0);
1063 g_assert_not_reached ();
1069 parse_mpsub (ParserState * state, const gchar * line)
1074 switch (state->state) {
1076 /* looking for two floats (offset, duration) */
1077 if (sscanf (line, "%f %f", &t1, &t2) == 2) {
1079 state->start_time += state->duration + GST_SECOND * t1;
1080 state->duration = GST_SECOND * t2;
1084 { /* No need to parse that text if it's out of segment */
1085 gint64 clip_start = 0, clip_stop = 0;
1086 gboolean in_seg = FALSE;
1088 /* Check our segment start/stop */
1089 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1090 state->start_time, state->start_time + state->duration,
1091 &clip_start, &clip_stop);
1094 state->start_time = clip_start;
1095 state->duration = clip_stop - clip_start;
1101 /* looking for subtitle text; empty line ends this
1103 if (state->buf->len)
1104 g_string_append_c (state->buf, '\n');
1105 g_string_append (state->buf, line);
1106 if (strlen (line) == 0) {
1107 ret = g_strdup (state->buf->str);
1108 g_string_truncate (state->buf, 0);
1114 g_assert_not_reached ();
1119 static const gchar *
1120 dks_skip_timestamp (const gchar * line)
1122 while (*line && *line != ']')
1130 parse_dks (ParserState * state, const gchar * line)
1134 switch (state->state) {
1136 /* Looking for the start time and text */
1137 if (sscanf (line, "[%u:%u:%u]", &h, &m, &s) == 3) {
1139 state->start_time = (((guint64) h) * 3600 + m * 60 + s) * GST_SECOND;
1140 text = dks_skip_timestamp (line);
1143 g_string_append (state->buf, text);
1149 gint64 clip_start = 0, clip_stop = 0;
1153 /* Looking for the end time */
1154 if (sscanf (line, "[%u:%u:%u]", &h, &m, &s) == 3) {
1156 state->duration = (((guint64) h) * 3600 + m * 60 + s) * GST_SECOND -
1159 GST_WARNING ("Failed to parse subtitle end time");
1163 /* Check if this subtitle is out of the current segment */
1164 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1165 state->start_time, state->start_time + state->duration,
1166 &clip_start, &clip_stop);
1172 state->start_time = clip_start;
1173 state->duration = clip_stop - clip_start;
1175 ret = g_strdup (state->buf->str);
1176 g_string_truncate (state->buf, 0);
1177 unescape_newlines_br (ret);
1181 g_assert_not_reached ();
1187 parser_state_init (ParserState * state)
1189 GST_DEBUG ("initialising parser");
1192 g_string_truncate (state->buf, 0);
1194 state->buf = g_string_new (NULL);
1197 state->start_time = 0;
1198 state->duration = 0;
1199 state->max_duration = 0; /* no limit */
1201 state->segment = NULL;
1205 parser_state_dispose (GstSubParse * self, ParserState * state)
1208 g_string_free (state->buf, TRUE);
1211 if (state->user_data) {
1212 switch (self->parser_type) {
1213 #ifndef GST_DISABLE_XML
1214 case GST_SUB_PARSE_FORMAT_SAMI:
1215 sami_context_reset (state);
1224 /* regex type enum */
1227 GST_SUB_PARSE_REGEX_UNKNOWN = 0,
1228 GST_SUB_PARSE_REGEX_MDVDSUB = 1,
1229 GST_SUB_PARSE_REGEX_SUBRIP = 2,
1230 GST_SUB_PARSE_REGEX_DKS = 3,
1234 gst_sub_parse_data_format_autodetect_regex_once (GstSubParseRegex regtype)
1236 gpointer result = NULL;
1237 GError *gerr = NULL;
1239 case GST_SUB_PARSE_REGEX_MDVDSUB:
1241 (gpointer) g_regex_new ("^\\{[0-9]+\\}\\{[0-9]+\\}", 0, 0, &gerr);
1242 if (result == NULL) {
1243 g_warning ("Compilation of mdvd regex failed: %s", gerr->message);
1244 g_error_free (gerr);
1247 case GST_SUB_PARSE_REGEX_SUBRIP:
1248 result = (gpointer) g_regex_new ("^([ 0-9]){0,3}[0-9]\\s*(\x0d)?\x0a"
1249 "[ 0-9][0-9]:[ 0-9][0-9]:[ 0-9][0-9][,.][ 0-9]{0,2}[0-9]"
1250 " +--> +([ 0-9])?[0-9]:[ 0-9][0-9]:[ 0-9][0-9][,.][ 0-9]{0,2}[0-9]",
1252 if (result == NULL) {
1253 g_warning ("Compilation of subrip regex failed: %s", gerr->message);
1254 g_error_free (gerr);
1257 case GST_SUB_PARSE_REGEX_DKS:
1258 result = (gpointer) g_regex_new ("^\\[[0-9]+:[0-9]+:[0-9]+\\].*",
1260 if (result == NULL) {
1261 g_warning ("Compilation of dks regex failed: %s", gerr->message);
1262 g_error_free (gerr);
1266 GST_WARNING ("Trying to allocate regex of unknown type %u", regtype);
1272 * FIXME: maybe we should pass along a second argument, the preceding
1273 * text buffer, because that is how this originally worked, even though
1274 * I don't really see the use of that.
1277 static GstSubParseFormat
1278 gst_sub_parse_data_format_autodetect (gchar * match_str)
1282 static GOnce mdvd_rx_once = G_ONCE_INIT;
1283 static GOnce subrip_rx_once = G_ONCE_INIT;
1284 static GOnce dks_rx_once = G_ONCE_INIT;
1290 g_once (&mdvd_rx_once,
1291 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1292 (gpointer) GST_SUB_PARSE_REGEX_MDVDSUB);
1293 g_once (&subrip_rx_once,
1294 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1295 (gpointer) GST_SUB_PARSE_REGEX_SUBRIP);
1296 g_once (&dks_rx_once,
1297 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1298 (gpointer) GST_SUB_PARSE_REGEX_DKS);
1300 mdvd_grx = (GRegex *) mdvd_rx_once.retval;
1301 subrip_grx = (GRegex *) subrip_rx_once.retval;
1302 dks_grx = (GRegex *) dks_rx_once.retval;
1304 if (g_regex_match (mdvd_grx, match_str, 0, NULL) == TRUE) {
1305 GST_LOG ("MicroDVD (frame based) format detected");
1306 return GST_SUB_PARSE_FORMAT_MDVDSUB;
1308 if (g_regex_match (subrip_grx, match_str, 0, NULL) == TRUE) {
1309 GST_LOG ("SubRip (time based) format detected");
1310 return GST_SUB_PARSE_FORMAT_SUBRIP;
1312 if (g_regex_match (dks_grx, match_str, 0, NULL) == TRUE) {
1313 GST_LOG ("DKS (time based) format detected");
1314 return GST_SUB_PARSE_FORMAT_DKS;
1317 if (!strncmp (match_str, "FORMAT=TIME", 11)) {
1318 GST_LOG ("MPSub (time based) format detected");
1319 return GST_SUB_PARSE_FORMAT_MPSUB;
1321 #ifndef GST_DISABLE_XML
1322 if (strstr (match_str, "<SAMI>") != NULL ||
1323 strstr (match_str, "<sami>") != NULL) {
1324 GST_LOG ("SAMI (time based) format detected");
1325 return GST_SUB_PARSE_FORMAT_SAMI;
1328 /* we're boldly assuming the first subtitle appears within the first hour */
1329 if (sscanf (match_str, "0:%02u:%02u:", &n1, &n2) == 2 ||
1330 sscanf (match_str, "0:%02u:%02u=", &n1, &n2) == 2 ||
1331 sscanf (match_str, "00:%02u:%02u:", &n1, &n2) == 2 ||
1332 sscanf (match_str, "00:%02u:%02u=", &n1, &n2) == 2 ||
1333 sscanf (match_str, "00:%02u:%02u,%u=", &n1, &n2, &n3) == 3) {
1334 GST_LOG ("TMPlayer (time based) format detected");
1335 return GST_SUB_PARSE_FORMAT_TMPLAYER;
1337 if (sscanf (match_str, "[%u][%u]", &n1, &n2) == 2) {
1338 GST_LOG ("MPL2 (time based) format detected");
1339 return GST_SUB_PARSE_FORMAT_MPL2;
1341 if (strstr (match_str, "[INFORMATION]") != NULL) {
1342 GST_LOG ("SubViewer (time based) format detected");
1343 return GST_SUB_PARSE_FORMAT_SUBVIEWER;
1345 if (strstr (match_str, "{QTtext}") != NULL) {
1346 GST_LOG ("QTtext (time based) format detected");
1347 return GST_SUB_PARSE_FORMAT_QTTEXT;
1350 GST_DEBUG ("no subtitle format detected");
1351 return GST_SUB_PARSE_FORMAT_UNKNOWN;
1355 gst_sub_parse_format_autodetect (GstSubParse * self)
1358 GstSubParseFormat format;
1360 if (strlen (self->textbuf->str) < 30) {
1361 GST_DEBUG ("File too small to be a subtitles file");
1365 data = g_strndup (self->textbuf->str, 35);
1366 format = gst_sub_parse_data_format_autodetect (data);
1369 self->parser_type = format;
1370 self->subtitle_codec = gst_sub_parse_get_format_description (format);
1371 parser_state_init (&self->state);
1374 case GST_SUB_PARSE_FORMAT_MDVDSUB:
1375 self->parse_line = parse_mdvdsub;
1376 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1377 case GST_SUB_PARSE_FORMAT_SUBRIP:
1378 self->parse_line = parse_subrip;
1379 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1380 case GST_SUB_PARSE_FORMAT_MPSUB:
1381 self->parse_line = parse_mpsub;
1382 return gst_caps_new_simple ("text/plain", NULL);
1383 #ifndef GST_DISABLE_XML
1384 case GST_SUB_PARSE_FORMAT_SAMI:
1385 self->parse_line = parse_sami;
1386 sami_context_init (&self->state);
1387 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1389 case GST_SUB_PARSE_FORMAT_TMPLAYER:
1390 self->parse_line = parse_tmplayer;
1391 self->state.max_duration = 5 * GST_SECOND;
1392 return gst_caps_new_simple ("text/plain", NULL);
1393 case GST_SUB_PARSE_FORMAT_MPL2:
1394 self->parse_line = parse_mpl2;
1395 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1396 case GST_SUB_PARSE_FORMAT_DKS:
1397 self->parse_line = parse_dks;
1398 return gst_caps_new_simple ("text/plain", NULL);
1399 case GST_SUB_PARSE_FORMAT_SUBVIEWER:
1400 self->parse_line = parse_subviewer;
1401 return gst_caps_new_simple ("text/plain", NULL);
1402 case GST_SUB_PARSE_FORMAT_QTTEXT:
1403 self->parse_line = parse_qttext;
1404 qttext_context_init (&self->state);
1405 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1406 case GST_SUB_PARSE_FORMAT_UNKNOWN:
1408 GST_DEBUG ("no subtitle format detected");
1409 GST_ELEMENT_ERROR (self, STREAM, WRONG_TYPE,
1410 ("The input is not a valid/supported subtitle file"), (NULL));
1416 feed_textbuf (GstSubParse * self, GstBuffer * buf)
1420 gchar *input = NULL;
1422 discont = GST_BUFFER_IS_DISCONT (buf);
1424 if (GST_BUFFER_OFFSET_IS_VALID (buf) &&
1425 GST_BUFFER_OFFSET (buf) != self->offset) {
1426 self->offset = GST_BUFFER_OFFSET (buf);
1431 GST_INFO ("discontinuity");
1432 /* flush the parser state */
1433 parser_state_init (&self->state);
1434 g_string_truncate (self->textbuf, 0);
1435 gst_adapter_clear (self->adapter);
1436 #ifndef GST_DISABLE_XML
1437 if (self->parser_type == GST_SUB_PARSE_FORMAT_SAMI)
1438 sami_context_reset (&self->state);
1440 /* we could set a flag to make sure that the next buffer we push out also
1441 * has the DISCONT flag set, but there's no point really given that it's
1442 * subtitles which are discontinuous by nature. */
1445 self->offset = GST_BUFFER_OFFSET (buf) + GST_BUFFER_SIZE (buf);
1446 self->next_offset = self->offset;
1448 gst_adapter_push (self->adapter, buf);
1451 convert_encoding (self, (const gchar *) gst_adapter_peek (self->adapter,
1452 gst_adapter_available (self->adapter)),
1453 (gsize) gst_adapter_available (self->adapter), &consumed);
1455 if (input && consumed > 0) {
1456 self->textbuf = g_string_append (self->textbuf, input);
1457 gst_adapter_flush (self->adapter, consumed);
1463 static GstFlowReturn
1464 handle_buffer (GstSubParse * self, GstBuffer * buf)
1466 GstFlowReturn ret = GST_FLOW_OK;
1467 GstCaps *caps = NULL;
1468 gchar *line, *subtitle;
1470 if (self->first_buffer) {
1471 self->detected_encoding =
1472 detect_encoding ((gchar *) GST_BUFFER_DATA (buf),
1473 GST_BUFFER_SIZE (buf));
1474 self->first_buffer = FALSE;
1475 self->state.fps_n = self->fps_n;
1476 self->state.fps_d = self->fps_d;
1479 feed_textbuf (self, buf);
1481 /* make sure we know the format */
1482 if (G_UNLIKELY (self->parser_type == GST_SUB_PARSE_FORMAT_UNKNOWN)) {
1483 if (!(caps = gst_sub_parse_format_autodetect (self))) {
1484 return GST_FLOW_UNEXPECTED;
1486 if (!gst_pad_set_caps (self->srcpad, caps)) {
1487 gst_caps_unref (caps);
1488 return GST_FLOW_UNEXPECTED;
1490 gst_caps_unref (caps);
1493 if (self->subtitle_codec != NULL) {
1496 tags = gst_tag_list_new ();
1497 gst_tag_list_add (tags, GST_TAG_MERGE_APPEND, GST_TAG_SUBTITLE_CODEC,
1498 self->subtitle_codec, NULL);
1499 gst_element_found_tags_for_pad (GST_ELEMENT (self), self->srcpad, tags);
1503 while (!self->flushing && (line = get_next_line (self))) {
1506 /* Set segment on our parser state machine */
1507 self->state.segment = &self->segment;
1508 /* Now parse the line, out of segment lines will just return NULL */
1509 GST_LOG_OBJECT (self, "Parsing line '%s'", line + offset);
1510 subtitle = self->parse_line (&self->state, line + offset);
1514 guint subtitle_len = strlen (subtitle);
1516 /* +1 for terminating NUL character */
1517 ret = gst_pad_alloc_buffer_and_set_caps (self->srcpad,
1518 GST_BUFFER_OFFSET_NONE, subtitle_len + 1,
1519 GST_PAD_CAPS (self->srcpad), &buf);
1521 if (ret == GST_FLOW_OK) {
1522 /* copy terminating NUL character as well */
1523 memcpy (GST_BUFFER_DATA (buf), subtitle, subtitle_len + 1);
1524 GST_BUFFER_SIZE (buf) = subtitle_len;
1525 GST_BUFFER_TIMESTAMP (buf) = self->state.start_time;
1526 GST_BUFFER_DURATION (buf) = self->state.duration;
1528 /* in some cases (e.g. tmplayer) we can only determine the duration
1529 * of a text chunk from the timestamp of the next text chunk; in those
1530 * cases, we probably want to limit the duration to something
1531 * reasonable, so we don't end up showing some text for e.g. 40 seconds
1532 * just because nothing else is being said during that time */
1533 if (self->state.max_duration > 0 && GST_BUFFER_DURATION_IS_VALID (buf)) {
1534 if (GST_BUFFER_DURATION (buf) > self->state.max_duration)
1535 GST_BUFFER_DURATION (buf) = self->state.max_duration;
1538 gst_segment_set_last_stop (&self->segment, GST_FORMAT_TIME,
1539 self->state.start_time);
1541 GST_DEBUG_OBJECT (self, "Sending text '%s', %" GST_TIME_FORMAT " + %"
1542 GST_TIME_FORMAT, subtitle, GST_TIME_ARGS (self->state.start_time),
1543 GST_TIME_ARGS (self->state.duration));
1545 ret = gst_pad_push (self->srcpad, buf);
1548 /* move this forward (the tmplayer parser needs this) */
1549 if (self->state.duration != GST_CLOCK_TIME_NONE)
1550 self->state.start_time += self->state.duration;
1555 if (ret != GST_FLOW_OK) {
1556 GST_DEBUG_OBJECT (self, "flow: %s", gst_flow_get_name (ret));
1565 static GstFlowReturn
1566 gst_sub_parse_chain (GstPad * sinkpad, GstBuffer * buf)
1571 self = GST_SUBPARSE (GST_PAD_PARENT (sinkpad));
1573 /* Push newsegment if needed */
1574 if (self->need_segment) {
1575 GST_LOG_OBJECT (self, "pushing newsegment event with %" GST_SEGMENT_FORMAT,
1578 gst_pad_push_event (self->srcpad, gst_event_new_new_segment (FALSE,
1579 self->segment.rate, self->segment.format,
1580 self->segment.last_stop, self->segment.stop, self->segment.time));
1581 self->need_segment = FALSE;
1584 ret = handle_buffer (self, buf);
1590 gst_sub_parse_sink_event (GstPad * pad, GstEvent * event)
1592 GstSubParse *self = GST_SUBPARSE (gst_pad_get_parent (pad));
1593 gboolean ret = FALSE;
1595 GST_DEBUG ("Handling %s event", GST_EVENT_TYPE_NAME (event));
1597 switch (GST_EVENT_TYPE (event)) {
1598 case GST_EVENT_EOS:{
1599 /* Make sure the last subrip chunk is pushed out even
1600 * if the file does not have an empty line at the end */
1601 if (self->parser_type == GST_SUB_PARSE_FORMAT_SUBRIP ||
1602 self->parser_type == GST_SUB_PARSE_FORMAT_TMPLAYER ||
1603 self->parser_type == GST_SUB_PARSE_FORMAT_MPL2 ||
1604 self->parser_type == GST_SUB_PARSE_FORMAT_QTTEXT) {
1605 GstBuffer *buf = gst_buffer_new_and_alloc (2 + 1);
1607 GST_DEBUG ("EOS. Pushing remaining text (if any)");
1608 GST_BUFFER_DATA (buf)[0] = '\n';
1609 GST_BUFFER_DATA (buf)[1] = '\n';
1610 GST_BUFFER_DATA (buf)[2] = '\0'; /* play it safe */
1611 GST_BUFFER_SIZE (buf) = 2;
1612 GST_BUFFER_OFFSET (buf) = self->offset;
1613 gst_sub_parse_chain (pad, buf);
1615 ret = gst_pad_event_default (pad, event);
1618 case GST_EVENT_NEWSEGMENT:
1622 gint64 start, stop, time;
1625 gst_event_parse_new_segment (event, &update, &rate, &format, &start,
1628 GST_DEBUG_OBJECT (self, "newsegment (%s)", gst_format_get_name (format));
1630 if (format == GST_FORMAT_TIME) {
1631 gst_segment_set_newsegment (&self->segment, update, rate, format,
1634 /* if not time format, we'll either start with a 0 timestamp anyway or
1635 * it's following a seek in which case we'll have saved the requested
1636 * seek segment and don't want to overwrite it (remember that on a seek
1637 * we always just seek back to the start in BYTES format and just throw
1638 * away all text that's before the requested position; if the subtitles
1639 * come from an upstream demuxer, it won't be able to handle our BYTES
1640 * seek request and instead send us a newsegment from the seek request
1641 * it received via its video pads instead, so all is fine then too) */
1645 gst_event_unref (event);
1648 case GST_EVENT_FLUSH_START:
1650 self->flushing = TRUE;
1652 ret = gst_pad_event_default (pad, event);
1655 case GST_EVENT_FLUSH_STOP:
1657 self->flushing = FALSE;
1659 ret = gst_pad_event_default (pad, event);
1663 ret = gst_pad_event_default (pad, event);
1667 gst_object_unref (self);
1673 static GstStateChangeReturn
1674 gst_sub_parse_change_state (GstElement * element, GstStateChange transition)
1676 GstStateChangeReturn ret = GST_STATE_CHANGE_SUCCESS;
1677 GstSubParse *self = GST_SUBPARSE (element);
1679 switch (transition) {
1680 case GST_STATE_CHANGE_READY_TO_PAUSED:
1681 /* format detection will init the parser state */
1683 self->next_offset = 0;
1684 self->parser_type = GST_SUB_PARSE_FORMAT_UNKNOWN;
1685 self->valid_utf8 = TRUE;
1686 self->first_buffer = TRUE;
1687 g_free (self->detected_encoding);
1688 self->detected_encoding = NULL;
1689 g_string_truncate (self->textbuf, 0);
1690 gst_adapter_clear (self->adapter);
1696 ret = parent_class->change_state (element, transition);
1697 if (ret == GST_STATE_CHANGE_FAILURE)
1700 switch (transition) {
1701 case GST_STATE_CHANGE_PAUSED_TO_READY:
1702 parser_state_dispose (self, &self->state);
1703 self->parser_type = GST_SUB_PARSE_FORMAT_UNKNOWN;
1716 /* FIXME 0.11: these caps are ugly, use app/x-subtitle + type field or so;
1717 * also, give different subtitle formats really different types */
1718 static GstStaticCaps mpl2_caps =
1719 GST_STATIC_CAPS ("application/x-subtitle-mpl2");
1720 #define SUB_CAPS (gst_static_caps_get (&sub_caps))
1722 static GstStaticCaps tmp_caps =
1723 GST_STATIC_CAPS ("application/x-subtitle-tmplayer");
1724 #define TMP_CAPS (gst_static_caps_get (&tmp_caps))
1726 static GstStaticCaps sub_caps = GST_STATIC_CAPS ("application/x-subtitle");
1727 #define MPL2_CAPS (gst_static_caps_get (&mpl2_caps))
1729 #ifndef GST_DISABLE_XML
1730 static GstStaticCaps smi_caps = GST_STATIC_CAPS ("application/x-subtitle-sami");
1731 #define SAMI_CAPS (gst_static_caps_get (&smi_caps))
1734 static GstStaticCaps dks_caps = GST_STATIC_CAPS ("application/x-subtitle-dks");
1735 #define DKS_CAPS (gst_static_caps_get (&dks_caps))
1737 static GstStaticCaps qttext_caps =
1738 GST_STATIC_CAPS ("application/x-subtitle-qttext");
1739 #define QTTEXT_CAPS (gst_static_caps_get (&qttext_caps))
1742 gst_subparse_type_find (GstTypeFind * tf, gpointer private)
1744 GstSubParseFormat format;
1748 gchar *encoding = NULL;
1751 if (!(data = gst_type_find_peek (tf, 0, 129)))
1754 /* make sure string passed to _autodetect() is NUL-terminated */
1755 str = g_malloc0 (129);
1756 memcpy (str, data, 128);
1758 if ((encoding = detect_encoding (str, 128)) != NULL) {
1759 gchar *converted_str;
1763 converted_str = gst_convert_to_utf8 (str, 128, encoding, &tmp, &err);
1764 if (converted_str == NULL) {
1765 GST_DEBUG ("Encoding '%s' detected but conversion failed: %s", encoding,
1771 str = converted_str;
1776 /* Check if at least the first 120 chars are valid UTF8,
1777 * otherwise convert as always */
1778 if (!g_utf8_validate (str, 128, &end) && (end - str) < 120) {
1779 gchar *converted_str;
1784 enc = g_getenv ("GST_SUBTITLE_ENCODING");
1785 if (enc == NULL || *enc == '\0') {
1786 /* if local encoding is UTF-8 and no encoding specified
1787 * via the environment variable, assume ISO-8859-15 */
1788 if (g_get_charset (&enc)) {
1789 enc = "ISO-8859-15";
1792 converted_str = gst_convert_to_utf8 (str, 128, enc, &tmp, &err);
1793 if (converted_str == NULL) {
1794 GST_DEBUG ("Charset conversion failed: %s", err->message);
1800 str = converted_str;
1804 format = gst_sub_parse_data_format_autodetect (str);
1808 case GST_SUB_PARSE_FORMAT_MDVDSUB:
1809 GST_DEBUG ("MicroDVD format detected");
1812 case GST_SUB_PARSE_FORMAT_SUBRIP:
1813 GST_DEBUG ("SubRip format detected");
1816 case GST_SUB_PARSE_FORMAT_MPSUB:
1817 GST_DEBUG ("MPSub format detected");
1820 #ifndef GST_DISABLE_XML
1821 case GST_SUB_PARSE_FORMAT_SAMI:
1822 GST_DEBUG ("SAMI (time-based) format detected");
1826 case GST_SUB_PARSE_FORMAT_TMPLAYER:
1827 GST_DEBUG ("TMPlayer (time based) format detected");
1830 /* FIXME: our MPL2 typefinding is not really good enough to warrant
1831 * returning a high probability (however, since we registered our
1832 * typefinder here with a rank of MARGINAL we should pretty much only
1833 * be called if most other typefinders have already run */
1834 case GST_SUB_PARSE_FORMAT_MPL2:
1835 GST_DEBUG ("MPL2 (time based) format detected");
1838 case GST_SUB_PARSE_FORMAT_SUBVIEWER:
1839 GST_DEBUG ("SubViewer format detected");
1842 case GST_SUB_PARSE_FORMAT_DKS:
1843 GST_DEBUG ("DKS format detected");
1846 case GST_SUB_PARSE_FORMAT_QTTEXT:
1847 GST_DEBUG ("QTtext format detected");
1851 case GST_SUB_PARSE_FORMAT_UNKNOWN:
1852 GST_DEBUG ("no subtitle format detected");
1856 /* if we're here, it's ok */
1857 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, caps);
1861 plugin_init (GstPlugin * plugin)
1863 static const gchar *sub_exts[] =
1864 { "srt", "sub", "mpsub", "mdvd", "smi", "txt", "dks", NULL };
1866 GST_DEBUG_CATEGORY_INIT (sub_parse_debug, "subparse", 0, ".sub parser");
1868 if (!gst_type_find_register (plugin, "subparse_typefind", GST_RANK_MARGINAL,
1869 gst_subparse_type_find, (gchar **) sub_exts, SUB_CAPS, NULL, NULL))
1872 if (!gst_element_register (plugin, "subparse",
1873 GST_RANK_PRIMARY, GST_TYPE_SUBPARSE) ||
1874 !gst_element_register (plugin, "ssaparse",
1875 GST_RANK_PRIMARY, GST_TYPE_SSA_PARSE)) {
1882 GST_PLUGIN_DEFINE (GST_VERSION_MAJOR,
1886 plugin_init, VERSION, "LGPL", GST_PACKAGE_NAME, GST_PACKAGE_ORIGIN)