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.last_stop);
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)) {
275 GstSeekType start_type, stop_type;
280 gst_event_parse_seek (event, &rate, &format, &self->segment_flags,
281 &start_type, &start, &stop_type, &stop);
283 if (format != GST_FORMAT_TIME) {
284 GST_WARNING_OBJECT (self, "we only support seeking in TIME format");
285 gst_event_unref (event);
289 /* Convert that seek to a seeking in bytes at position 0,
290 FIXME: could use an index */
291 ret = gst_pad_push_event (self->sinkpad,
292 gst_event_new_seek (rate, GST_FORMAT_BYTES, self->segment_flags,
293 GST_SEEK_TYPE_SET, 0, GST_SEEK_TYPE_NONE, 0));
296 /* Apply the seek to our segment */
297 gst_segment_set_seek (&self->segment, rate, format, self->segment_flags,
298 start_type, start, stop_type, stop, &update);
300 GST_DEBUG_OBJECT (self, "segment after seek: %" GST_SEGMENT_FORMAT,
303 self->next_offset = 0;
305 self->need_segment = TRUE;
307 GST_WARNING_OBJECT (self, "seek to 0 bytes failed");
310 gst_event_unref (event);
314 ret = gst_pad_event_default (pad, event);
319 gst_object_unref (self);
325 gst_sub_parse_set_property (GObject * object, guint prop_id,
326 const GValue * value, GParamSpec * pspec)
328 GstSubParse *subparse = GST_SUBPARSE (object);
330 GST_OBJECT_LOCK (subparse);
333 g_free (subparse->encoding);
334 subparse->encoding = g_value_dup_string (value);
335 GST_LOG_OBJECT (object, "subtitle encoding set to %s",
336 GST_STR_NULL (subparse->encoding));
340 subparse->fps_n = gst_value_get_fraction_numerator (value);
341 subparse->fps_d = gst_value_get_fraction_denominator (value);
342 GST_DEBUG_OBJECT (object, "video framerate set to %d/%d", subparse->fps_n,
345 if (!subparse->state.have_internal_fps) {
346 subparse->state.fps_n = subparse->fps_n;
347 subparse->state.fps_d = subparse->fps_d;
352 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
355 GST_OBJECT_UNLOCK (subparse);
359 gst_sub_parse_get_property (GObject * object, guint prop_id,
360 GValue * value, GParamSpec * pspec)
362 GstSubParse *subparse = GST_SUBPARSE (object);
364 GST_OBJECT_LOCK (subparse);
367 g_value_set_string (value, subparse->encoding);
370 gst_value_set_fraction (value, subparse->fps_n, subparse->fps_d);
373 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
376 GST_OBJECT_UNLOCK (subparse);
380 gst_sub_parse_get_format_description (GstSubParseFormat format)
383 case GST_SUB_PARSE_FORMAT_MDVDSUB:
385 case GST_SUB_PARSE_FORMAT_SUBRIP:
387 case GST_SUB_PARSE_FORMAT_MPSUB:
389 case GST_SUB_PARSE_FORMAT_SAMI:
391 case GST_SUB_PARSE_FORMAT_TMPLAYER:
393 case GST_SUB_PARSE_FORMAT_MPL2:
395 case GST_SUB_PARSE_FORMAT_SUBVIEWER:
397 case GST_SUB_PARSE_FORMAT_DKS:
399 case GST_SUB_PARSE_FORMAT_QTTEXT:
402 case GST_SUB_PARSE_FORMAT_UNKNOWN:
409 gst_convert_to_utf8 (const gchar * str, gsize len, const gchar * encoding,
410 gsize * consumed, GError ** err)
415 /* The char cast is necessary in glib < 2.24 */
417 g_convert_with_fallback (str, len, "UTF-8", encoding, (char *) "*",
418 consumed, NULL, err);
422 /* + 3 to skip UTF-8 BOM if it was added */
424 if (len >= 3 && (guint8) ret[0] == 0xEF && (guint8) ret[1] == 0xBB
425 && (guint8) ret[2] == 0xBF)
426 g_memmove (ret, ret + 3, len + 1 - 3);
432 detect_encoding (const gchar * str, gsize len)
434 if (len >= 3 && (guint8) str[0] == 0xEF && (guint8) str[1] == 0xBB
435 && (guint8) str[2] == 0xBF)
436 return g_strdup ("UTF-8");
438 if (len >= 2 && (guint8) str[0] == 0xFE && (guint8) str[1] == 0xFF)
439 return g_strdup ("UTF-16BE");
441 if (len >= 2 && (guint8) str[0] == 0xFF && (guint8) str[1] == 0xFE)
442 return g_strdup ("UTF-16LE");
444 if (len >= 4 && (guint8) str[0] == 0x00 && (guint8) str[1] == 0x00
445 && (guint8) str[2] == 0xFE && (guint8) str[3] == 0xFF)
446 return g_strdup ("UTF-32BE");
448 if (len >= 4 && (guint8) str[0] == 0xFF && (guint8) str[1] == 0xFE
449 && (guint8) str[2] == 0x00 && (guint8) str[3] == 0x00)
450 return g_strdup ("UTF-32LE");
456 convert_encoding (GstSubParse * self, const gchar * str, gsize len,
459 const gchar *encoding;
465 /* First try any detected encoding */
466 if (self->detected_encoding) {
468 gst_convert_to_utf8 (str, len, self->detected_encoding, consumed, &err);
473 GST_WARNING_OBJECT (self, "could not convert string from '%s' to UTF-8: %s",
474 self->detected_encoding, err->message);
475 g_free (self->detected_encoding);
476 self->detected_encoding = NULL;
480 /* Otherwise check if it's UTF8 */
481 if (self->valid_utf8) {
482 if (g_utf8_validate (str, len, NULL)) {
483 GST_LOG_OBJECT (self, "valid UTF-8, no conversion needed");
485 return g_strndup (str, len);
487 GST_INFO_OBJECT (self, "invalid UTF-8!");
488 self->valid_utf8 = FALSE;
491 /* Else try fallback */
492 encoding = self->encoding;
493 if (encoding == NULL || *encoding == '\0') {
494 encoding = g_getenv ("GST_SUBTITLE_ENCODING");
496 if (encoding == NULL || *encoding == '\0') {
497 /* if local encoding is UTF-8 and no encoding specified
498 * via the environment variable, assume ISO-8859-15 */
499 if (g_get_charset (&encoding)) {
500 encoding = "ISO-8859-15";
504 ret = gst_convert_to_utf8 (str, len, encoding, consumed, &err);
507 GST_WARNING_OBJECT (self, "could not convert string from '%s' to UTF-8: %s",
508 encoding, err->message);
511 /* invalid input encoding, fall back to ISO-8859-15 (always succeeds) */
512 ret = gst_convert_to_utf8 (str, len, "ISO-8859-15", consumed, NULL);
515 GST_LOG_OBJECT (self,
516 "successfully converted %" G_GSIZE_FORMAT " characters from %s to UTF-8"
517 "%s", len, encoding, (err) ? " , using ISO-8859-15 as fallback" : "");
523 get_next_line (GstSubParse * self)
526 const char *line_end;
528 gboolean have_r = FALSE;
530 line_end = strchr (self->textbuf->str, '\n');
533 /* end-of-line not found; return for more data */
537 /* get rid of '\r' */
538 if (line_end != self->textbuf->str && *(line_end - 1) == '\r') {
543 line_len = line_end - self->textbuf->str;
544 line = g_strndup (self->textbuf->str, line_len);
545 self->textbuf = g_string_erase (self->textbuf, 0,
546 line_len + (have_r ? 2 : 1));
551 parse_mdvdsub (ParserState * state, const gchar * line)
553 const gchar *line_split;
555 guint start_frame, end_frame;
556 gint64 clip_start = 0, clip_stop = 0;
557 gboolean in_seg = FALSE;
561 /* style variables */
567 if (sscanf (line, "{%u}{%u}", &start_frame, &end_frame) != 2) {
568 g_warning ("Parse of the following line, assumed to be in microdvd .sub"
569 " format, failed:\n%s", line);
573 /* skip the {%u}{%u} part */
574 line = strchr (line, '}') + 1;
575 line = strchr (line, '}') + 1;
577 /* see if there's a first line with a framerate */
578 if (start_frame == 1 && end_frame == 1) {
579 gchar *rest, *end = NULL;
581 rest = g_strdup (line);
582 g_strdelimit (rest, ",", '.');
583 fps = g_ascii_strtod (rest, &end);
585 gst_util_double_to_fraction (fps, &state->fps_n, &state->fps_d);
586 GST_INFO ("framerate from file: %d/%d ('%s')", state->fps_n,
594 gst_util_uint64_scale (start_frame, GST_SECOND * state->fps_d,
597 gst_util_uint64_scale (end_frame - start_frame, GST_SECOND * state->fps_d,
600 /* Check our segment start/stop */
601 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
602 state->start_time, state->start_time + state->duration, &clip_start,
605 /* No need to parse that text if it's out of segment */
607 state->start_time = clip_start;
608 state->duration = clip_stop - clip_start;
613 markup = g_string_new (NULL);
618 /* parse style markup */
619 if (strncmp (line, "{y:i}", 5) == 0) {
621 line = strchr (line, '}') + 1;
623 if (strncmp (line, "{y:b}", 5) == 0) {
625 line = strchr (line, '}') + 1;
627 if (sscanf (line, "{s:%u}", &fontsize) == 1) {
628 line = strchr (line, '}') + 1;
630 /* forward slashes at beginning/end signify italics too */
631 if (g_str_has_prefix (line, "/")) {
635 if ((line_split = strchr (line, '|')))
636 line_chunk = g_markup_escape_text (line, line_split - line);
638 line_chunk = g_markup_escape_text (line, strlen (line));
640 /* Remove italics markers at end of line/stanza (CHECKME: are end slashes
641 * always at the end of a line or can they span multiple lines?) */
642 if (g_str_has_suffix (line_chunk, "/")) {
643 line_chunk[strlen (line_chunk) - 1] = '\0';
646 markup = g_string_append (markup, "<span");
648 g_string_append (markup, " style=\"italic\"");
650 g_string_append (markup, " weight=\"bold\"");
652 g_string_append_printf (markup, " size=\"%u\"", fontsize * 1000);
653 g_string_append_printf (markup, ">%s</span>", line_chunk);
656 g_string_append (markup, "\n");
657 line = line_split + 1;
663 g_string_free (markup, FALSE);
664 GST_DEBUG ("parse_mdvdsub returning (%f+%f): %s",
665 state->start_time / (double) GST_SECOND,
666 state->duration / (double) GST_SECOND, ret);
671 strip_trailing_newlines (gchar * txt)
677 while (len > 1 && txt[len - 1] == '\n') {
684 /* we want to escape text in general, but retain basic markup like
685 * <i></i>, <u></u>, and <b></b>. The easiest and safest way is to
686 * just unescape a white list of allowed markups again after
687 * escaping everything (the text between these simple markers isn't
688 * necessarily escaped, so it seems best to do it like this) */
690 subrip_unescape_formatting (gchar * txt)
694 for (pos = txt; pos != NULL && *pos != '\0'; ++pos) {
695 if (g_ascii_strncasecmp (pos, "<u>", 9) == 0 ||
696 g_ascii_strncasecmp (pos, "<i>", 9) == 0 ||
697 g_ascii_strncasecmp (pos, "<b>", 9) == 0) {
699 pos[1] = g_ascii_tolower (pos[4]);
701 /* move NUL terminator as well */
702 g_memmove (pos + 3, pos + 9, strlen (pos + 9) + 1);
707 for (pos = txt; pos != NULL && *pos != '\0'; ++pos) {
708 if (g_ascii_strncasecmp (pos, "</u>", 10) == 0 ||
709 g_ascii_strncasecmp (pos, "</i>", 10) == 0 ||
710 g_ascii_strncasecmp (pos, "</b>", 10) == 0) {
713 pos[2] = g_ascii_tolower (pos[5]);
715 /* move NUL terminator as well */
716 g_memmove (pos + 4, pos + 10, strlen (pos + 10) + 1);
724 subrip_remove_unhandled_tag (gchar * start, gchar * stop)
728 tag = start + strlen ("<");
732 if (g_ascii_tolower (*tag) < 'a' || g_ascii_tolower (*tag) > 'z')
737 GST_LOG ("removing unhandled tag '%s'", start);
739 g_memmove (start, stop, strlen (stop) + 1);
743 /* remove tags we haven't explicitly allowed earlier on, like font tags
746 subrip_remove_unhandled_tags (gchar * txt)
750 for (pos = txt; pos != NULL && *pos != '\0'; ++pos) {
751 if (strncmp (pos, "<", 4) == 0 && (gt = strstr (pos + 4, ">"))) {
752 if (subrip_remove_unhandled_tag (pos, gt + strlen (">")))
758 /* we only allow <i>, <u> and <b>, so let's take a simple approach. This code
759 * assumes the input has been escaped and subrip_unescape_formatting() has then
760 * been run over the input! This function adds missing closing markup tags and
761 * removes broken closing tags for tags that have never been opened. */
763 subrip_fix_up_markup (gchar ** p_txt)
765 gchar *cur, *next_tag;
767 guint num_open_tags = 0;
769 g_assert (*p_txt != NULL);
772 while (*cur != '\0') {
773 next_tag = strchr (cur, '<');
774 if (next_tag == NULL)
780 if (num_open_tags == 0 || open_tags[num_open_tags - 1] != *next_tag) {
781 GST_LOG ("broken input, closing tag '%c' is not open", *next_tag);
782 g_memmove (next_tag - 2, next_tag + 2, strlen (next_tag + 2) + 1);
785 /* it's all good, closing tag which is open */
793 if (num_open_tags == G_N_ELEMENTS (open_tags))
794 return; /* something dodgy is going on, stop parsing */
795 open_tags[num_open_tags] = *next_tag;
799 GST_ERROR ("unexpected tag '%c' (%s)", *next_tag, next_tag);
800 g_assert_not_reached ();
806 if (num_open_tags > 0) {
809 s = g_string_new (*p_txt);
810 while (num_open_tags > 0) {
811 GST_LOG ("adding missing closing tag '%c'", open_tags[num_open_tags - 1]);
812 g_string_append_c (s, '<');
813 g_string_append_c (s, '/');
814 g_string_append_c (s, open_tags[num_open_tags - 1]);
815 g_string_append_c (s, '>');
819 *p_txt = g_string_free (s, FALSE);
824 parse_subrip_time (const gchar * ts_string, GstClockTime * t)
826 gchar s[128] = { '\0', };
828 guint hour, min, sec, msec, len;
830 while (*ts_string == ' ')
833 g_strlcpy (s, ts_string, sizeof (s));
834 if ((end = strstr (s, "-->")))
838 /* ms may be in these formats:
839 * hh:mm:ss,500 = 500ms
842 * hh:mm:ss, 50 = 50ms
844 * and the same with . instead of ,.
845 * sscanf() doesn't differentiate between ' 5' and '5' so munge
846 * the white spaces within the timestamp to '0' (I'm sure there's a
847 * way to make sscanf() do this for us, but how?)
849 g_strdelimit (s, " ", '0');
850 g_strdelimit (s, ".", ',');
852 /* make sure we have exactly three digits after he comma */
854 g_assert (p != NULL);
861 g_strlcat (&p[len], "0", 2);
865 GST_LOG ("parsing timestamp '%s'", s);
866 if (sscanf (s, "%u:%u:%u,%u", &hour, &min, &sec, &msec) != 4) {
867 GST_WARNING ("failed to parse subrip timestamp string '%s'", s);
871 *t = ((hour * 3600) + (min * 60) + sec) * GST_SECOND + msec * GST_MSECOND;
876 parse_subrip (ParserState * state, const gchar * line)
881 switch (state->state) {
883 /* looking for a single integer */
884 if (sscanf (line, "%u", &subnum) == 1)
889 GstClockTime ts_start, ts_end;
892 /* looking for start_time --> end_time */
893 if ((end_time = strstr (line, " --> ")) &&
894 parse_subrip_time (line, &ts_start) &&
895 parse_subrip_time (end_time + strlen (" --> "), &ts_end) &&
896 state->start_time <= ts_end) {
898 state->start_time = ts_start;
899 state->duration = ts_end - ts_start;
901 GST_DEBUG ("error parsing subrip time line '%s'", line);
908 /* No need to parse that text if it's out of segment */
909 gint64 clip_start = 0, clip_stop = 0;
910 gboolean in_seg = FALSE;
912 /* Check our segment start/stop */
913 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
914 state->start_time, state->start_time + state->duration,
915 &clip_start, &clip_stop);
918 state->start_time = clip_start;
919 state->duration = clip_stop - clip_start;
925 /* looking for subtitle text; empty line ends this subtitle entry */
927 g_string_append_c (state->buf, '\n');
928 g_string_append (state->buf, line);
929 if (strlen (line) == 0) {
930 ret = g_markup_escape_text (state->buf->str, state->buf->len);
931 g_string_truncate (state->buf, 0);
933 subrip_unescape_formatting (ret);
934 subrip_remove_unhandled_tags (ret);
935 strip_trailing_newlines (ret);
936 subrip_fix_up_markup (&ret);
941 g_return_val_if_reached (NULL);
946 unescape_newlines_br (gchar * read)
950 /* Replace all occurences of '[br]' with a newline as version 2
951 * of the subviewer format uses this for newlines */
953 if (read[0] == '\0' || read[1] == '\0' || read[2] == '\0' || read[3] == '\0')
957 if (strncmp (read, "[br]", 4) == 0) {
971 parse_subviewer (ParserState * state, const gchar * line)
973 guint h1, m1, s1, ms1;
974 guint h2, m2, s2, ms2;
977 /* TODO: Maybe also parse the fields in the header, especially DELAY.
978 * For examples see the unit test or
979 * http://www.doom9.org/index.html?/sub.htm */
981 switch (state->state) {
983 /* looking for start_time,end_time */
984 if (sscanf (line, "%u:%u:%u.%u,%u:%u:%u.%u",
985 &h1, &m1, &s1, &ms1, &h2, &m2, &s2, &ms2) == 8) {
988 (((guint64) h1) * 3600 + m1 * 60 + s1) * GST_SECOND +
991 (((guint64) h2) * 3600 + m2 * 60 + s2) * GST_SECOND +
992 ms2 * GST_MSECOND - state->start_time;
997 /* No need to parse that text if it's out of segment */
998 gint64 clip_start = 0, clip_stop = 0;
999 gboolean in_seg = FALSE;
1001 /* Check our segment start/stop */
1002 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1003 state->start_time, state->start_time + state->duration,
1004 &clip_start, &clip_stop);
1007 state->start_time = clip_start;
1008 state->duration = clip_stop - clip_start;
1014 /* looking for subtitle text; empty line ends this subtitle entry */
1015 if (state->buf->len)
1016 g_string_append_c (state->buf, '\n');
1017 g_string_append (state->buf, line);
1018 if (strlen (line) == 0) {
1019 ret = g_strdup (state->buf->str);
1020 unescape_newlines_br (ret);
1021 strip_trailing_newlines (ret);
1022 g_string_truncate (state->buf, 0);
1028 g_assert_not_reached ();
1034 parse_mpsub (ParserState * state, const gchar * line)
1039 switch (state->state) {
1041 /* looking for two floats (offset, duration) */
1042 if (sscanf (line, "%f %f", &t1, &t2) == 2) {
1044 state->start_time += state->duration + GST_SECOND * t1;
1045 state->duration = GST_SECOND * t2;
1049 { /* No need to parse that text if it's out of segment */
1050 gint64 clip_start = 0, clip_stop = 0;
1051 gboolean in_seg = FALSE;
1053 /* Check our segment start/stop */
1054 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1055 state->start_time, state->start_time + state->duration,
1056 &clip_start, &clip_stop);
1059 state->start_time = clip_start;
1060 state->duration = clip_stop - clip_start;
1066 /* looking for subtitle text; empty line ends this
1068 if (state->buf->len)
1069 g_string_append_c (state->buf, '\n');
1070 g_string_append (state->buf, line);
1071 if (strlen (line) == 0) {
1072 ret = g_strdup (state->buf->str);
1073 g_string_truncate (state->buf, 0);
1079 g_assert_not_reached ();
1084 static const gchar *
1085 dks_skip_timestamp (const gchar * line)
1087 while (*line && *line != ']')
1095 parse_dks (ParserState * state, const gchar * line)
1099 switch (state->state) {
1101 /* Looking for the start time and text */
1102 if (sscanf (line, "[%u:%u:%u]", &h, &m, &s) == 3) {
1104 state->start_time = (((guint64) h) * 3600 + m * 60 + s) * GST_SECOND;
1105 text = dks_skip_timestamp (line);
1108 g_string_append (state->buf, text);
1114 gint64 clip_start = 0, clip_stop = 0;
1118 /* Looking for the end time */
1119 if (sscanf (line, "[%u:%u:%u]", &h, &m, &s) == 3) {
1121 state->duration = (((guint64) h) * 3600 + m * 60 + s) * GST_SECOND -
1124 GST_WARNING ("Failed to parse subtitle end time");
1128 /* Check if this subtitle is out of the current segment */
1129 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1130 state->start_time, state->start_time + state->duration,
1131 &clip_start, &clip_stop);
1137 state->start_time = clip_start;
1138 state->duration = clip_stop - clip_start;
1140 ret = g_strdup (state->buf->str);
1141 g_string_truncate (state->buf, 0);
1142 unescape_newlines_br (ret);
1146 g_assert_not_reached ();
1152 parser_state_init (ParserState * state)
1154 GST_DEBUG ("initialising parser");
1157 g_string_truncate (state->buf, 0);
1159 state->buf = g_string_new (NULL);
1162 state->start_time = 0;
1163 state->duration = 0;
1164 state->max_duration = 0; /* no limit */
1166 state->segment = NULL;
1170 parser_state_dispose (GstSubParse * self, ParserState * state)
1173 g_string_free (state->buf, TRUE);
1176 if (state->user_data) {
1177 switch (self->parser_type) {
1178 #ifndef GST_DISABLE_XML
1179 case GST_SUB_PARSE_FORMAT_SAMI:
1180 sami_context_reset (state);
1189 /* regex type enum */
1192 GST_SUB_PARSE_REGEX_UNKNOWN = 0,
1193 GST_SUB_PARSE_REGEX_MDVDSUB = 1,
1194 GST_SUB_PARSE_REGEX_SUBRIP = 2,
1195 GST_SUB_PARSE_REGEX_DKS = 3,
1199 gst_sub_parse_data_format_autodetect_regex_once (GstSubParseRegex regtype)
1201 gpointer result = NULL;
1202 GError *gerr = NULL;
1204 case GST_SUB_PARSE_REGEX_MDVDSUB:
1206 (gpointer) g_regex_new ("^\\{[0-9]+\\}\\{[0-9]+\\}", 0, 0, &gerr);
1207 if (result == NULL) {
1208 g_warning ("Compilation of mdvd regex failed: %s", gerr->message);
1209 g_error_free (gerr);
1212 case GST_SUB_PARSE_REGEX_SUBRIP:
1213 result = (gpointer) g_regex_new ("^([ 0-9]){0,3}[0-9]\\s*(\x0d)?\x0a"
1214 "[ 0-9][0-9]:[ 0-9][0-9]:[ 0-9][0-9][,.][ 0-9]{0,2}[0-9]"
1215 " +--> +([ 0-9])?[0-9]:[ 0-9][0-9]:[ 0-9][0-9][,.][ 0-9]{0,2}[0-9]",
1217 if (result == NULL) {
1218 g_warning ("Compilation of subrip regex failed: %s", gerr->message);
1219 g_error_free (gerr);
1222 case GST_SUB_PARSE_REGEX_DKS:
1223 result = (gpointer) g_regex_new ("^\\[[0-9]+:[0-9]+:[0-9]+\\].*",
1225 if (result == NULL) {
1226 g_warning ("Compilation of dks regex failed: %s", gerr->message);
1227 g_error_free (gerr);
1231 GST_WARNING ("Trying to allocate regex of unknown type %u", regtype);
1237 * FIXME: maybe we should pass along a second argument, the preceding
1238 * text buffer, because that is how this originally worked, even though
1239 * I don't really see the use of that.
1242 static GstSubParseFormat
1243 gst_sub_parse_data_format_autodetect (gchar * match_str)
1247 static GOnce mdvd_rx_once = G_ONCE_INIT;
1248 static GOnce subrip_rx_once = G_ONCE_INIT;
1249 static GOnce dks_rx_once = G_ONCE_INIT;
1255 g_once (&mdvd_rx_once,
1256 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1257 (gpointer) GST_SUB_PARSE_REGEX_MDVDSUB);
1258 g_once (&subrip_rx_once,
1259 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1260 (gpointer) GST_SUB_PARSE_REGEX_SUBRIP);
1261 g_once (&dks_rx_once,
1262 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1263 (gpointer) GST_SUB_PARSE_REGEX_DKS);
1265 mdvd_grx = (GRegex *) mdvd_rx_once.retval;
1266 subrip_grx = (GRegex *) subrip_rx_once.retval;
1267 dks_grx = (GRegex *) dks_rx_once.retval;
1269 if (g_regex_match (mdvd_grx, match_str, 0, NULL) == TRUE) {
1270 GST_LOG ("MicroDVD (frame based) format detected");
1271 return GST_SUB_PARSE_FORMAT_MDVDSUB;
1273 if (g_regex_match (subrip_grx, match_str, 0, NULL) == TRUE) {
1274 GST_LOG ("SubRip (time based) format detected");
1275 return GST_SUB_PARSE_FORMAT_SUBRIP;
1277 if (g_regex_match (dks_grx, match_str, 0, NULL) == TRUE) {
1278 GST_LOG ("DKS (time based) format detected");
1279 return GST_SUB_PARSE_FORMAT_DKS;
1282 if (!strncmp (match_str, "FORMAT=TIME", 11)) {
1283 GST_LOG ("MPSub (time based) format detected");
1284 return GST_SUB_PARSE_FORMAT_MPSUB;
1286 #ifndef GST_DISABLE_XML
1287 if (strstr (match_str, "<SAMI>") != NULL ||
1288 strstr (match_str, "<sami>") != NULL) {
1289 GST_LOG ("SAMI (time based) format detected");
1290 return GST_SUB_PARSE_FORMAT_SAMI;
1293 /* we're boldly assuming the first subtitle appears within the first hour */
1294 if (sscanf (match_str, "0:%02u:%02u:", &n1, &n2) == 2 ||
1295 sscanf (match_str, "0:%02u:%02u=", &n1, &n2) == 2 ||
1296 sscanf (match_str, "00:%02u:%02u:", &n1, &n2) == 2 ||
1297 sscanf (match_str, "00:%02u:%02u=", &n1, &n2) == 2 ||
1298 sscanf (match_str, "00:%02u:%02u,%u=", &n1, &n2, &n3) == 3) {
1299 GST_LOG ("TMPlayer (time based) format detected");
1300 return GST_SUB_PARSE_FORMAT_TMPLAYER;
1302 if (sscanf (match_str, "[%u][%u]", &n1, &n2) == 2) {
1303 GST_LOG ("MPL2 (time based) format detected");
1304 return GST_SUB_PARSE_FORMAT_MPL2;
1306 if (strstr (match_str, "[INFORMATION]") != NULL) {
1307 GST_LOG ("SubViewer (time based) format detected");
1308 return GST_SUB_PARSE_FORMAT_SUBVIEWER;
1310 if (strstr (match_str, "{QTtext}") != NULL) {
1311 GST_LOG ("QTtext (time based) format detected");
1312 return GST_SUB_PARSE_FORMAT_QTTEXT;
1315 GST_DEBUG ("no subtitle format detected");
1316 return GST_SUB_PARSE_FORMAT_UNKNOWN;
1320 gst_sub_parse_format_autodetect (GstSubParse * self)
1323 GstSubParseFormat format;
1325 if (strlen (self->textbuf->str) < 30) {
1326 GST_DEBUG ("File too small to be a subtitles file");
1330 data = g_strndup (self->textbuf->str, 35);
1331 format = gst_sub_parse_data_format_autodetect (data);
1334 self->parser_type = format;
1335 self->subtitle_codec = gst_sub_parse_get_format_description (format);
1336 parser_state_init (&self->state);
1339 case GST_SUB_PARSE_FORMAT_MDVDSUB:
1340 self->parse_line = parse_mdvdsub;
1341 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1342 case GST_SUB_PARSE_FORMAT_SUBRIP:
1343 self->parse_line = parse_subrip;
1344 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1345 case GST_SUB_PARSE_FORMAT_MPSUB:
1346 self->parse_line = parse_mpsub;
1347 return gst_caps_new_simple ("text/plain", NULL);
1348 #ifndef GST_DISABLE_XML
1349 case GST_SUB_PARSE_FORMAT_SAMI:
1350 self->parse_line = parse_sami;
1351 sami_context_init (&self->state);
1352 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1354 case GST_SUB_PARSE_FORMAT_TMPLAYER:
1355 self->parse_line = parse_tmplayer;
1356 self->state.max_duration = 5 * GST_SECOND;
1357 return gst_caps_new_simple ("text/plain", NULL);
1358 case GST_SUB_PARSE_FORMAT_MPL2:
1359 self->parse_line = parse_mpl2;
1360 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1361 case GST_SUB_PARSE_FORMAT_DKS:
1362 self->parse_line = parse_dks;
1363 return gst_caps_new_simple ("text/plain", NULL);
1364 case GST_SUB_PARSE_FORMAT_SUBVIEWER:
1365 self->parse_line = parse_subviewer;
1366 return gst_caps_new_simple ("text/plain", NULL);
1367 case GST_SUB_PARSE_FORMAT_QTTEXT:
1368 self->parse_line = parse_qttext;
1369 qttext_context_init (&self->state);
1370 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1371 case GST_SUB_PARSE_FORMAT_UNKNOWN:
1373 GST_DEBUG ("no subtitle format detected");
1374 GST_ELEMENT_ERROR (self, STREAM, WRONG_TYPE,
1375 ("The input is not a valid/supported subtitle file"), (NULL));
1381 feed_textbuf (GstSubParse * self, GstBuffer * buf)
1385 gchar *input = NULL;
1389 discont = GST_BUFFER_IS_DISCONT (buf);
1391 if (GST_BUFFER_OFFSET_IS_VALID (buf) &&
1392 GST_BUFFER_OFFSET (buf) != self->offset) {
1393 self->offset = GST_BUFFER_OFFSET (buf);
1398 GST_INFO ("discontinuity");
1399 /* flush the parser state */
1400 parser_state_init (&self->state);
1401 g_string_truncate (self->textbuf, 0);
1402 gst_adapter_clear (self->adapter);
1403 #ifndef GST_DISABLE_XML
1404 if (self->parser_type == GST_SUB_PARSE_FORMAT_SAMI)
1405 sami_context_reset (&self->state);
1407 /* we could set a flag to make sure that the next buffer we push out also
1408 * has the DISCONT flag set, but there's no point really given that it's
1409 * subtitles which are discontinuous by nature. */
1412 self->offset = GST_BUFFER_OFFSET (buf) + gst_buffer_get_size (buf);
1413 self->next_offset = self->offset;
1415 gst_adapter_push (self->adapter, buf);
1417 avail = gst_adapter_available (self->adapter);
1418 data = gst_adapter_map (self->adapter, avail),
1419 input = convert_encoding (self, (const gchar *) data, avail, &consumed);
1421 if (input && consumed > 0) {
1422 self->textbuf = g_string_append (self->textbuf, input);
1423 gst_adapter_unmap (self->adapter, consumed);
1425 gst_adapter_unmap (self->adapter, 0);
1431 static GstFlowReturn
1432 handle_buffer (GstSubParse * self, GstBuffer * buf)
1434 GstFlowReturn ret = GST_FLOW_OK;
1435 GstCaps *caps = NULL;
1436 gchar *line, *subtitle, *data;
1439 if (self->first_buffer) {
1440 data = gst_buffer_map (buf, &size, NULL, GST_MAP_READ);
1441 self->detected_encoding = detect_encoding (data, size);
1442 gst_buffer_unmap (buf, data, size);
1443 self->first_buffer = FALSE;
1444 self->state.fps_n = self->fps_n;
1445 self->state.fps_d = self->fps_d;
1448 feed_textbuf (self, buf);
1450 /* make sure we know the format */
1451 if (G_UNLIKELY (self->parser_type == GST_SUB_PARSE_FORMAT_UNKNOWN)) {
1452 if (!(caps = gst_sub_parse_format_autodetect (self))) {
1453 return GST_FLOW_UNEXPECTED;
1455 if (!gst_pad_set_caps (self->srcpad, caps)) {
1456 gst_caps_unref (caps);
1457 return GST_FLOW_UNEXPECTED;
1459 gst_caps_unref (caps);
1462 if (self->subtitle_codec != NULL) {
1465 tags = gst_tag_list_new ();
1466 gst_tag_list_add (tags, GST_TAG_MERGE_APPEND, GST_TAG_SUBTITLE_CODEC,
1467 self->subtitle_codec, NULL);
1468 gst_element_found_tags_for_pad (GST_ELEMENT (self), self->srcpad, tags);
1472 while (!self->flushing && (line = get_next_line (self))) {
1475 /* Set segment on our parser state machine */
1476 self->state.segment = &self->segment;
1477 /* Now parse the line, out of segment lines will just return NULL */
1478 GST_LOG_OBJECT (self, "Parsing line '%s'", line + offset);
1479 subtitle = self->parse_line (&self->state, line + offset);
1483 guint subtitle_len = strlen (subtitle);
1485 /* +1 for terminating NUL character */
1486 buf = gst_buffer_new_and_alloc (subtitle_len + 1);
1488 /* copy terminating NUL character as well */
1489 gst_buffer_fill (buf, 0, subtitle, subtitle_len + 1);
1490 gst_buffer_set_size (buf, subtitle_len);
1492 GST_BUFFER_TIMESTAMP (buf) = self->state.start_time;
1493 GST_BUFFER_DURATION (buf) = self->state.duration;
1495 /* in some cases (e.g. tmplayer) we can only determine the duration
1496 * of a text chunk from the timestamp of the next text chunk; in those
1497 * cases, we probably want to limit the duration to something
1498 * reasonable, so we don't end up showing some text for e.g. 40 seconds
1499 * just because nothing else is being said during that time */
1500 if (self->state.max_duration > 0 && GST_BUFFER_DURATION_IS_VALID (buf)) {
1501 if (GST_BUFFER_DURATION (buf) > self->state.max_duration)
1502 GST_BUFFER_DURATION (buf) = self->state.max_duration;
1505 gst_segment_set_last_stop (&self->segment, GST_FORMAT_TIME,
1506 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_new_segment (FALSE,
1545 self->segment.rate, self->segment.format,
1546 self->segment.last_stop, self->segment.stop, self->segment.time));
1547 self->need_segment = FALSE;
1550 ret = handle_buffer (self, buf);
1556 gst_sub_parse_sink_event (GstPad * pad, GstEvent * event)
1558 GstSubParse *self = GST_SUBPARSE (gst_pad_get_parent (pad));
1559 gboolean ret = FALSE;
1561 GST_DEBUG ("Handling %s event", GST_EVENT_TYPE_NAME (event));
1563 switch (GST_EVENT_TYPE (event)) {
1564 case GST_EVENT_EOS:{
1565 /* Make sure the last subrip chunk is pushed out even
1566 * if the file does not have an empty line at the end */
1567 if (self->parser_type == GST_SUB_PARSE_FORMAT_SUBRIP ||
1568 self->parser_type == GST_SUB_PARSE_FORMAT_TMPLAYER ||
1569 self->parser_type == GST_SUB_PARSE_FORMAT_MPL2 ||
1570 self->parser_type == GST_SUB_PARSE_FORMAT_QTTEXT) {
1571 gchar term_chars[] = { '\n', '\n', '\0' };
1572 GstBuffer *buf = gst_buffer_new_and_alloc (2 + 1);
1574 GST_DEBUG ("EOS. Pushing remaining text (if any)");
1575 gst_buffer_fill (buf, 0, term_chars, 3);
1576 gst_buffer_set_size (buf, 2);
1578 GST_BUFFER_OFFSET (buf) = self->offset;
1579 gst_sub_parse_chain (pad, buf);
1581 ret = gst_pad_event_default (pad, event);
1584 case GST_EVENT_NEWSEGMENT:
1588 gint64 start, stop, time;
1591 gst_event_parse_new_segment (event, &update, &rate, &format, &start,
1594 GST_DEBUG_OBJECT (self, "newsegment (%s)", gst_format_get_name (format));
1596 if (format == GST_FORMAT_TIME) {
1597 gst_segment_set_newsegment (&self->segment, update, rate, format,
1600 /* if not time format, we'll either start with a 0 timestamp anyway or
1601 * it's following a seek in which case we'll have saved the requested
1602 * seek segment and don't want to overwrite it (remember that on a seek
1603 * we always just seek back to the start in BYTES format and just throw
1604 * away all text that's before the requested position; if the subtitles
1605 * come from an upstream demuxer, it won't be able to handle our BYTES
1606 * seek request and instead send us a newsegment from the seek request
1607 * it received via its video pads instead, so all is fine then too) */
1611 gst_event_unref (event);
1614 case GST_EVENT_FLUSH_START:
1616 self->flushing = TRUE;
1618 ret = gst_pad_event_default (pad, event);
1621 case GST_EVENT_FLUSH_STOP:
1623 self->flushing = FALSE;
1625 ret = gst_pad_event_default (pad, event);
1629 ret = gst_pad_event_default (pad, event);
1633 gst_object_unref (self);
1639 static GstStateChangeReturn
1640 gst_sub_parse_change_state (GstElement * element, GstStateChange transition)
1642 GstStateChangeReturn ret = GST_STATE_CHANGE_SUCCESS;
1643 GstSubParse *self = GST_SUBPARSE (element);
1645 switch (transition) {
1646 case GST_STATE_CHANGE_READY_TO_PAUSED:
1647 /* format detection will init the parser state */
1649 self->next_offset = 0;
1650 self->parser_type = GST_SUB_PARSE_FORMAT_UNKNOWN;
1651 self->valid_utf8 = TRUE;
1652 self->first_buffer = TRUE;
1653 g_free (self->detected_encoding);
1654 self->detected_encoding = NULL;
1655 g_string_truncate (self->textbuf, 0);
1656 gst_adapter_clear (self->adapter);
1662 ret = GST_ELEMENT_CLASS (parent_class)->change_state (element, transition);
1663 if (ret == GST_STATE_CHANGE_FAILURE)
1666 switch (transition) {
1667 case GST_STATE_CHANGE_PAUSED_TO_READY:
1668 parser_state_dispose (self, &self->state);
1669 self->parser_type = GST_SUB_PARSE_FORMAT_UNKNOWN;
1682 /* FIXME 0.11: these caps are ugly, use app/x-subtitle + type field or so;
1683 * also, give different subtitle formats really different types */
1684 static GstStaticCaps mpl2_caps =
1685 GST_STATIC_CAPS ("application/x-subtitle-mpl2");
1686 #define SUB_CAPS (gst_static_caps_get (&sub_caps))
1688 static GstStaticCaps tmp_caps =
1689 GST_STATIC_CAPS ("application/x-subtitle-tmplayer");
1690 #define TMP_CAPS (gst_static_caps_get (&tmp_caps))
1692 static GstStaticCaps sub_caps = GST_STATIC_CAPS ("application/x-subtitle");
1693 #define MPL2_CAPS (gst_static_caps_get (&mpl2_caps))
1695 #ifndef GST_DISABLE_XML
1696 static GstStaticCaps smi_caps = GST_STATIC_CAPS ("application/x-subtitle-sami");
1697 #define SAMI_CAPS (gst_static_caps_get (&smi_caps))
1700 static GstStaticCaps dks_caps = GST_STATIC_CAPS ("application/x-subtitle-dks");
1701 #define DKS_CAPS (gst_static_caps_get (&dks_caps))
1703 static GstStaticCaps qttext_caps =
1704 GST_STATIC_CAPS ("application/x-subtitle-qttext");
1705 #define QTTEXT_CAPS (gst_static_caps_get (&qttext_caps))
1708 gst_subparse_type_find (GstTypeFind * tf, gpointer private)
1710 GstSubParseFormat format;
1714 gchar *encoding = NULL;
1717 if (!(data = gst_type_find_peek (tf, 0, 129)))
1720 /* make sure string passed to _autodetect() is NUL-terminated */
1721 str = g_malloc0 (129);
1722 memcpy (str, data, 128);
1724 if ((encoding = detect_encoding (str, 128)) != NULL) {
1725 gchar *converted_str;
1729 converted_str = gst_convert_to_utf8 (str, 128, encoding, &tmp, &err);
1730 if (converted_str == NULL) {
1731 GST_DEBUG ("Encoding '%s' detected but conversion failed: %s", encoding,
1737 str = converted_str;
1742 /* Check if at least the first 120 chars are valid UTF8,
1743 * otherwise convert as always */
1744 if (!g_utf8_validate (str, 128, &end) && (end - str) < 120) {
1745 gchar *converted_str;
1750 enc = g_getenv ("GST_SUBTITLE_ENCODING");
1751 if (enc == NULL || *enc == '\0') {
1752 /* if local encoding is UTF-8 and no encoding specified
1753 * via the environment variable, assume ISO-8859-15 */
1754 if (g_get_charset (&enc)) {
1755 enc = "ISO-8859-15";
1758 converted_str = gst_convert_to_utf8 (str, 128, enc, &tmp, &err);
1759 if (converted_str == NULL) {
1760 GST_DEBUG ("Charset conversion failed: %s", err->message);
1766 str = converted_str;
1770 format = gst_sub_parse_data_format_autodetect (str);
1774 case GST_SUB_PARSE_FORMAT_MDVDSUB:
1775 GST_DEBUG ("MicroDVD format detected");
1778 case GST_SUB_PARSE_FORMAT_SUBRIP:
1779 GST_DEBUG ("SubRip format detected");
1782 case GST_SUB_PARSE_FORMAT_MPSUB:
1783 GST_DEBUG ("MPSub format detected");
1786 #ifndef GST_DISABLE_XML
1787 case GST_SUB_PARSE_FORMAT_SAMI:
1788 GST_DEBUG ("SAMI (time-based) format detected");
1792 case GST_SUB_PARSE_FORMAT_TMPLAYER:
1793 GST_DEBUG ("TMPlayer (time based) format detected");
1796 /* FIXME: our MPL2 typefinding is not really good enough to warrant
1797 * returning a high probability (however, since we registered our
1798 * typefinder here with a rank of MARGINAL we should pretty much only
1799 * be called if most other typefinders have already run */
1800 case GST_SUB_PARSE_FORMAT_MPL2:
1801 GST_DEBUG ("MPL2 (time based) format detected");
1804 case GST_SUB_PARSE_FORMAT_SUBVIEWER:
1805 GST_DEBUG ("SubViewer format detected");
1808 case GST_SUB_PARSE_FORMAT_DKS:
1809 GST_DEBUG ("DKS format detected");
1812 case GST_SUB_PARSE_FORMAT_QTTEXT:
1813 GST_DEBUG ("QTtext format detected");
1817 case GST_SUB_PARSE_FORMAT_UNKNOWN:
1818 GST_DEBUG ("no subtitle format detected");
1822 /* if we're here, it's ok */
1823 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, caps);
1827 plugin_init (GstPlugin * plugin)
1829 static const gchar *sub_exts[] =
1830 { "srt", "sub", "mpsub", "mdvd", "smi", "txt", "dks", NULL };
1832 GST_DEBUG_CATEGORY_INIT (sub_parse_debug, "subparse", 0, ".sub parser");
1834 if (!gst_type_find_register (plugin, "subparse_typefind", GST_RANK_MARGINAL,
1835 gst_subparse_type_find, (gchar **) sub_exts, SUB_CAPS, NULL, NULL))
1838 if (!gst_element_register (plugin, "subparse",
1839 GST_RANK_PRIMARY, GST_TYPE_SUBPARSE) ||
1840 !gst_element_register (plugin, "ssaparse",
1841 GST_RANK_PRIMARY, GST_TYPE_SSA_PARSE)) {
1848 GST_PLUGIN_DEFINE (GST_VERSION_MAJOR,
1852 plugin_init, VERSION, "LGPL", GST_PACKAGE_NAME, GST_PACKAGE_ORIGIN)