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; "
77 "application/x-subtitle-tmplayer; application/x-subtitle-mpl2")
81 static GstStaticPadTemplate src_templ = GST_STATIC_PAD_TEMPLATE ("src",
84 GST_STATIC_CAPS ("text/plain; text/x-pango-markup")
87 static void gst_sub_parse_base_init (GstSubParseClass * klass);
88 static void gst_sub_parse_class_init (GstSubParseClass * klass);
89 static void gst_sub_parse_init (GstSubParse * subparse);
91 static gboolean gst_sub_parse_src_event (GstPad * pad, GstEvent * event);
92 static gboolean gst_sub_parse_src_query (GstPad * pad, GstQuery * query);
93 static gboolean gst_sub_parse_sink_event (GstPad * pad, GstEvent * event);
95 static GstStateChangeReturn gst_sub_parse_change_state (GstElement * element,
96 GstStateChange transition);
98 static GstFlowReturn gst_sub_parse_chain (GstPad * sinkpad, GstBuffer * buf);
100 static GstElementClass *parent_class = NULL;
103 gst_sub_parse_get_type (void)
105 static GType sub_parse_type = 0;
107 if (!sub_parse_type) {
108 static const GTypeInfo sub_parse_info = {
109 sizeof (GstSubParseClass),
110 (GBaseInitFunc) gst_sub_parse_base_init,
112 (GClassInitFunc) gst_sub_parse_class_init,
115 sizeof (GstSubParse),
117 (GInstanceInitFunc) gst_sub_parse_init,
120 sub_parse_type = g_type_register_static (GST_TYPE_ELEMENT,
121 "GstSubParse", &sub_parse_info, 0);
124 return sub_parse_type;
128 gst_sub_parse_base_init (GstSubParseClass * klass)
130 GstElementClass *element_class = GST_ELEMENT_CLASS (klass);
132 gst_element_class_add_pad_template (element_class,
133 gst_static_pad_template_get (&sink_templ));
134 gst_element_class_add_pad_template (element_class,
135 gst_static_pad_template_get (&src_templ));
136 gst_element_class_set_details (element_class, &sub_parse_details);
140 gst_sub_parse_dispose (GObject * object)
142 GstSubParse *subparse = GST_SUBPARSE (object);
144 GST_DEBUG_OBJECT (subparse, "cleaning up subtitle parser");
146 if (subparse->encoding) {
147 g_free (subparse->encoding);
148 subparse->encoding = NULL;
151 if (subparse->detected_encoding) {
152 g_free (subparse->detected_encoding);
153 subparse->detected_encoding = NULL;
156 if (subparse->adapter) {
157 g_object_unref (subparse->adapter);
158 subparse->adapter = NULL;
161 if (subparse->textbuf) {
162 g_string_free (subparse->textbuf, TRUE);
163 subparse->textbuf = NULL;
165 #ifndef GST_DISABLE_XML
166 sami_context_deinit (&subparse->state);
169 GST_CALL_PARENT (G_OBJECT_CLASS, dispose, (object));
173 gst_sub_parse_class_init (GstSubParseClass * klass)
175 GObjectClass *object_class = G_OBJECT_CLASS (klass);
176 GstElementClass *element_class = GST_ELEMENT_CLASS (klass);
178 parent_class = g_type_class_peek_parent (klass);
180 object_class->dispose = gst_sub_parse_dispose;
181 object_class->set_property = gst_sub_parse_set_property;
182 object_class->get_property = gst_sub_parse_get_property;
184 element_class->change_state = gst_sub_parse_change_state;
186 g_object_class_install_property (object_class, PROP_ENCODING,
187 g_param_spec_string ("subtitle-encoding", "subtitle charset encoding",
188 "Encoding to assume if input subtitles are not in UTF-8 or any other "
189 "Unicode encoding. If not set, the GST_SUBTITLE_ENCODING environment "
190 "variable will be checked for an encoding to use. If that is not set "
191 "either, ISO-8859-15 will be assumed.", DEFAULT_ENCODING,
192 G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
194 g_object_class_install_property (object_class, PROP_VIDEOFPS,
195 gst_param_spec_fraction ("video-fps", "Video framerate",
196 "Framerate of the video stream. This is needed by some subtitle "
197 "formats to synchronize subtitles and video properly. If not set "
198 "and the subtitle format requires it subtitles may be out of sync.",
199 0, 1, G_MAXINT, 1, 24000, 1001, G_PARAM_READWRITE));
203 gst_sub_parse_init (GstSubParse * subparse)
205 subparse->sinkpad = gst_pad_new_from_static_template (&sink_templ, "sink");
206 gst_pad_set_chain_function (subparse->sinkpad,
207 GST_DEBUG_FUNCPTR (gst_sub_parse_chain));
208 gst_pad_set_event_function (subparse->sinkpad,
209 GST_DEBUG_FUNCPTR (gst_sub_parse_sink_event));
210 gst_element_add_pad (GST_ELEMENT (subparse), subparse->sinkpad);
212 subparse->srcpad = gst_pad_new_from_static_template (&src_templ, "src");
213 gst_pad_set_event_function (subparse->srcpad,
214 GST_DEBUG_FUNCPTR (gst_sub_parse_src_event));
215 gst_pad_set_query_function (subparse->srcpad,
216 GST_DEBUG_FUNCPTR (gst_sub_parse_src_query));
217 gst_element_add_pad (GST_ELEMENT (subparse), subparse->srcpad);
219 subparse->textbuf = g_string_new (NULL);
220 subparse->parser_type = GST_SUB_PARSE_FORMAT_UNKNOWN;
221 subparse->flushing = FALSE;
222 gst_segment_init (&subparse->segment, GST_FORMAT_TIME);
223 subparse->need_segment = TRUE;
224 subparse->encoding = g_strdup (DEFAULT_ENCODING);
225 subparse->detected_encoding = NULL;
226 subparse->adapter = gst_adapter_new ();
228 subparse->fps_n = 24000;
229 subparse->fps_d = 1001;
233 * Source pad functions.
237 gst_sub_parse_src_query (GstPad * pad, GstQuery * query)
239 GstSubParse *self = GST_SUBPARSE (gst_pad_get_parent (pad));
240 gboolean ret = FALSE;
242 GST_DEBUG ("Handling %s query", GST_QUERY_TYPE_NAME (query));
244 switch (GST_QUERY_TYPE (query)) {
245 case GST_QUERY_POSITION:{
248 gst_query_parse_position (query, &fmt, NULL);
249 if (fmt != GST_FORMAT_TIME) {
250 ret = gst_pad_peer_query (self->sinkpad, query);
253 gst_query_set_position (query, GST_FORMAT_TIME,
254 self->segment.last_stop);
257 case GST_QUERY_SEEKING:
260 gboolean seekable = FALSE;
264 gst_query_parse_seeking (query, &fmt, NULL, NULL, NULL);
265 if (fmt == GST_FORMAT_TIME) {
266 GstQuery *peerquery = gst_query_new_seeking (GST_FORMAT_BYTES);
268 seekable = gst_pad_peer_query (self->sinkpad, peerquery);
270 gst_query_parse_seeking (peerquery, NULL, &seekable, NULL, NULL);
271 gst_query_unref (peerquery);
274 gst_query_set_seeking (query, fmt, seekable, seekable ? 0 : -1, -1);
279 ret = gst_pad_peer_query (self->sinkpad, query);
283 gst_object_unref (self);
289 gst_sub_parse_src_event (GstPad * pad, GstEvent * event)
291 GstSubParse *self = GST_SUBPARSE (gst_pad_get_parent (pad));
292 gboolean ret = FALSE;
294 GST_DEBUG ("Handling %s event", GST_EVENT_TYPE_NAME (event));
296 switch (GST_EVENT_TYPE (event)) {
300 GstSeekType start_type, stop_type;
305 gst_event_parse_seek (event, &rate, &format, &self->segment_flags,
306 &start_type, &start, &stop_type, &stop);
308 if (format != GST_FORMAT_TIME) {
309 GST_WARNING_OBJECT (self, "we only support seeking in TIME format");
310 gst_event_unref (event);
314 /* Convert that seek to a seeking in bytes at position 0,
315 FIXME: could use an index */
316 ret = gst_pad_push_event (self->sinkpad,
317 gst_event_new_seek (rate, GST_FORMAT_BYTES, self->segment_flags,
318 GST_SEEK_TYPE_SET, 0, GST_SEEK_TYPE_NONE, 0));
321 /* Apply the seek to our segment */
322 gst_segment_set_seek (&self->segment, rate, format, self->segment_flags,
323 start_type, start, stop_type, stop, &update);
325 GST_DEBUG_OBJECT (self, "segment after seek: %" GST_SEGMENT_FORMAT,
328 self->next_offset = 0;
330 self->need_segment = TRUE;
332 GST_WARNING_OBJECT (self, "seek to 0 bytes failed");
335 gst_event_unref (event);
339 ret = gst_pad_event_default (pad, event);
344 gst_object_unref (self);
350 gst_sub_parse_set_property (GObject * object, guint prop_id,
351 const GValue * value, GParamSpec * pspec)
353 GstSubParse *subparse = GST_SUBPARSE (object);
355 GST_OBJECT_LOCK (subparse);
358 g_free (subparse->encoding);
359 subparse->encoding = g_value_dup_string (value);
360 GST_LOG_OBJECT (object, "subtitle encoding set to %s",
361 GST_STR_NULL (subparse->encoding));
365 subparse->fps_n = gst_value_get_fraction_numerator (value);
366 subparse->fps_d = gst_value_get_fraction_denominator (value);
367 GST_DEBUG_OBJECT (object, "video framerate set to %d/%d", subparse->fps_n,
370 if (!subparse->state.have_internal_fps) {
371 subparse->state.fps_n = subparse->fps_n;
372 subparse->state.fps_d = subparse->fps_d;
377 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
380 GST_OBJECT_UNLOCK (subparse);
384 gst_sub_parse_get_property (GObject * object, guint prop_id,
385 GValue * value, GParamSpec * pspec)
387 GstSubParse *subparse = GST_SUBPARSE (object);
389 GST_OBJECT_LOCK (subparse);
392 g_value_set_string (value, subparse->encoding);
395 gst_value_set_fraction (value, subparse->fps_n, subparse->fps_d);
398 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
401 GST_OBJECT_UNLOCK (subparse);
405 gst_sub_parse_get_format_description (GstSubParseFormat format)
408 case GST_SUB_PARSE_FORMAT_MDVDSUB:
410 case GST_SUB_PARSE_FORMAT_SUBRIP:
412 case GST_SUB_PARSE_FORMAT_MPSUB:
414 case GST_SUB_PARSE_FORMAT_SAMI:
416 case GST_SUB_PARSE_FORMAT_TMPLAYER:
418 case GST_SUB_PARSE_FORMAT_MPL2:
420 case GST_SUB_PARSE_FORMAT_SUBVIEWER:
422 case GST_SUB_PARSE_FORMAT_DKS:
425 case GST_SUB_PARSE_FORMAT_UNKNOWN:
432 gst_convert_to_utf8 (const gchar * str, gsize len, const gchar * encoding,
433 gsize * consumed, GError ** err)
439 g_convert_with_fallback (str, len, "UTF-8", encoding, "*", consumed, NULL,
444 /* + 3 to skip UTF-8 BOM if it was added */
446 if (len >= 3 && (guint8) ret[0] == 0xEF && (guint8) ret[1] == 0xBB
447 && (guint8) ret[2] == 0xBF)
448 g_memmove (ret, ret + 3, len + 1 - 3);
454 detect_encoding (const gchar * str, gsize len)
456 if (len >= 3 && (guint8) str[0] == 0xEF && (guint8) str[1] == 0xBB
457 && (guint8) str[2] == 0xBF)
458 return g_strdup ("UTF-8");
460 if (len >= 2 && (guint8) str[0] == 0xFE && (guint8) str[1] == 0xFF)
461 return g_strdup ("UTF-16BE");
463 if (len >= 2 && (guint8) str[0] == 0xFF && (guint8) str[1] == 0xFE)
464 return g_strdup ("UTF-16LE");
466 if (len >= 4 && (guint8) str[0] == 0x00 && (guint8) str[1] == 0x00
467 && (guint8) str[2] == 0xFE && (guint8) str[3] == 0xFF)
468 return g_strdup ("UTF-32BE");
470 if (len >= 4 && (guint8) str[0] == 0xFF && (guint8) str[1] == 0xFE
471 && (guint8) str[2] == 0x00 && (guint8) str[3] == 0x00)
472 return g_strdup ("UTF-32LE");
478 convert_encoding (GstSubParse * self, const gchar * str, gsize len,
481 const gchar *encoding;
487 /* First try any detected encoding */
488 if (self->detected_encoding) {
490 gst_convert_to_utf8 (str, len, self->detected_encoding, consumed, &err);
495 GST_WARNING_OBJECT (self, "could not convert string from '%s' to UTF-8: %s",
496 self->detected_encoding, err->message);
497 g_free (self->detected_encoding);
498 self->detected_encoding = NULL;
502 /* Otherwise check if it's UTF8 */
503 if (self->valid_utf8) {
504 if (g_utf8_validate (str, len, NULL)) {
505 GST_LOG_OBJECT (self, "valid UTF-8, no conversion needed");
507 return g_strndup (str, len);
509 GST_INFO_OBJECT (self, "invalid UTF-8!");
510 self->valid_utf8 = FALSE;
513 /* Else try fallback */
514 encoding = self->encoding;
515 if (encoding == NULL || *encoding == '\0') {
516 encoding = g_getenv ("GST_SUBTITLE_ENCODING");
518 if (encoding == NULL || *encoding == '\0') {
519 /* if local encoding is UTF-8 and no encoding specified
520 * via the environment variable, assume ISO-8859-15 */
521 if (g_get_charset (&encoding)) {
522 encoding = "ISO-8859-15";
526 ret = gst_convert_to_utf8 (str, len, encoding, consumed, &err);
529 GST_WARNING_OBJECT (self, "could not convert string from '%s' to UTF-8: %s",
530 encoding, err->message);
533 /* invalid input encoding, fall back to ISO-8859-15 (always succeeds) */
534 ret = gst_convert_to_utf8 (str, len, "ISO-8859-15", consumed, NULL);
537 GST_LOG_OBJECT (self,
538 "successfully converted %" G_GSIZE_FORMAT " characters from %s to UTF-8"
539 "%s", len, encoding, (err) ? " , using ISO-8859-15 as fallback" : "");
545 get_next_line (GstSubParse * self)
548 const char *line_end;
550 gboolean have_r = FALSE;
552 line_end = strchr (self->textbuf->str, '\n');
555 /* end-of-line not found; return for more data */
559 /* get rid of '\r' */
560 if (line_end != self->textbuf->str && *(line_end - 1) == '\r') {
565 line_len = line_end - self->textbuf->str;
566 line = g_strndup (self->textbuf->str, line_len);
567 self->textbuf = g_string_erase (self->textbuf, 0,
568 line_len + (have_r ? 2 : 1));
573 parse_mdvdsub (ParserState * state, const gchar * line)
575 const gchar *line_split;
577 guint start_frame, end_frame;
578 gint64 clip_start = 0, clip_stop = 0;
579 gboolean in_seg = FALSE;
583 /* style variables */
589 if (sscanf (line, "{%u}{%u}", &start_frame, &end_frame) != 2) {
590 g_warning ("Parse of the following line, assumed to be in microdvd .sub"
591 " format, failed:\n%s", line);
595 /* skip the {%u}{%u} part */
596 line = strchr (line, '}') + 1;
597 line = strchr (line, '}') + 1;
599 /* see if there's a first line with a framerate */
600 if (start_frame == 1 && end_frame == 1) {
601 gchar *rest, *end = NULL;
603 rest = g_strdup (line);
604 g_strdelimit (rest, ",", '.');
605 fps = g_ascii_strtod (rest, &end);
607 gst_util_double_to_fraction (fps, &state->fps_n, &state->fps_d);
608 GST_INFO ("framerate from file: %d/%d ('%s')", state->fps_n,
616 gst_util_uint64_scale (start_frame, GST_SECOND * state->fps_d,
619 gst_util_uint64_scale (end_frame - start_frame, GST_SECOND * state->fps_d,
622 /* Check our segment start/stop */
623 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
624 state->start_time, state->start_time + state->duration, &clip_start,
627 /* No need to parse that text if it's out of segment */
629 state->start_time = clip_start;
630 state->duration = clip_stop - clip_start;
635 markup = g_string_new (NULL);
640 /* parse style markup */
641 if (strncmp (line, "{y:i}", 5) == 0) {
643 line = strchr (line, '}') + 1;
645 if (strncmp (line, "{y:b}", 5) == 0) {
647 line = strchr (line, '}') + 1;
649 if (sscanf (line, "{s:%u}", &fontsize) == 1) {
650 line = strchr (line, '}') + 1;
652 /* forward slashes at beginning/end signify italics too */
653 if (g_str_has_prefix (line, "/")) {
657 if ((line_split = strchr (line, '|')))
658 line_chunk = g_markup_escape_text (line, line_split - line);
660 line_chunk = g_markup_escape_text (line, strlen (line));
662 /* Remove italics markers at end of line/stanza (CHECKME: are end slashes
663 * always at the end of a line or can they span multiple lines?) */
664 if (g_str_has_suffix (line_chunk, "/")) {
665 line_chunk[strlen (line_chunk) - 1] = '\0';
668 markup = g_string_append (markup, "<span");
670 g_string_append (markup, " style=\"italic\"");
672 g_string_append (markup, " weight=\"bold\"");
674 g_string_append_printf (markup, " size=\"%u\"", fontsize * 1000);
675 g_string_append_printf (markup, ">%s</span>", line_chunk);
678 g_string_append (markup, "\n");
679 line = line_split + 1;
685 g_string_free (markup, FALSE);
686 GST_DEBUG ("parse_mdvdsub returning (%f+%f): %s",
687 state->start_time / (double) GST_SECOND,
688 state->duration / (double) GST_SECOND, ret);
693 strip_trailing_newlines (gchar * txt)
699 while (len > 1 && txt[len - 1] == '\n') {
706 /* we want to escape text in general, but retain basic markup like
707 * <i></i>, <u></u>, and <b></b>. The easiest and safest way is to
708 * just unescape a white list of allowed markups again after
709 * escaping everything (the text between these simple markers isn't
710 * necessarily escaped, so it seems best to do it like this) */
712 subrip_unescape_formatting (gchar * txt)
716 for (pos = txt; pos != NULL && *pos != '\0'; ++pos) {
717 if (g_ascii_strncasecmp (pos, "<u>", 9) == 0 ||
718 g_ascii_strncasecmp (pos, "<i>", 9) == 0 ||
719 g_ascii_strncasecmp (pos, "<b>", 9) == 0) {
721 pos[1] = g_ascii_tolower (pos[4]);
723 /* move NUL terminator as well */
724 g_memmove (pos + 3, pos + 9, strlen (pos + 9) + 1);
729 for (pos = txt; pos != NULL && *pos != '\0'; ++pos) {
730 if (g_ascii_strncasecmp (pos, "</u>", 10) == 0 ||
731 g_ascii_strncasecmp (pos, "</i>", 10) == 0 ||
732 g_ascii_strncasecmp (pos, "</b>", 10) == 0) {
735 pos[2] = g_ascii_tolower (pos[5]);
737 /* move NUL terminator as well */
738 g_memmove (pos + 4, pos + 10, strlen (pos + 10) + 1);
746 subrip_remove_unhandled_tag (gchar * start, gchar * stop)
750 tag = start + strlen ("<");
754 if (g_ascii_tolower (*tag) < 'a' || g_ascii_tolower (*tag) > 'z')
759 GST_LOG ("removing unhandled tag '%s'", start);
761 g_memmove (start, stop, strlen (stop) + 1);
765 /* remove tags we haven't explicitly allowed earlier on, like font tags
768 subrip_remove_unhandled_tags (gchar * txt)
772 for (pos = txt; pos != NULL && *pos != '\0'; ++pos) {
773 if (strncmp (pos, "<", 4) == 0 && (gt = strstr (pos + 4, ">"))) {
774 if (subrip_remove_unhandled_tag (pos, gt + strlen (">")))
780 /* we only allow <i>, <u> and <b>, so let's take a simple approach. This code
781 * assumes the input has been escaped and subrip_unescape_formatting() has then
782 * been run over the input! This function adds missing closing markup tags and
783 * removes broken closing tags for tags that have never been opened. */
785 subrip_fix_up_markup (gchar ** p_txt)
787 gchar *cur, *next_tag;
789 guint num_open_tags = 0;
791 g_assert (*p_txt != NULL);
794 while (*cur != '\0') {
795 next_tag = strchr (cur, '<');
796 if (next_tag == NULL)
802 if (num_open_tags == 0 || open_tags[num_open_tags - 1] != *next_tag) {
803 GST_LOG ("broken input, closing tag '%c' is not open", *next_tag);
804 g_memmove (next_tag - 2, next_tag + 2, strlen (next_tag + 2) + 1);
807 /* it's all good, closing tag which is open */
815 if (num_open_tags == G_N_ELEMENTS (open_tags))
816 return; /* something dodgy is going on, stop parsing */
817 open_tags[num_open_tags] = *next_tag;
821 GST_ERROR ("unexpected tag '%c' (%s)", *next_tag, next_tag);
822 g_assert_not_reached ();
828 if (num_open_tags > 0) {
831 s = g_string_new (*p_txt);
832 while (num_open_tags > 0) {
833 GST_LOG ("adding missing closing tag '%c'", open_tags[num_open_tags - 1]);
834 g_string_append_c (s, '<');
835 g_string_append_c (s, '/');
836 g_string_append_c (s, open_tags[num_open_tags - 1]);
837 g_string_append_c (s, '>');
841 *p_txt = g_string_free (s, FALSE);
846 parse_subrip_time (const gchar * ts_string, GstClockTime * t)
848 gchar s[128] = { '\0', };
850 guint hour, min, sec, msec, len;
852 while (*ts_string == ' ')
855 g_strlcpy (s, ts_string, sizeof (s));
856 if ((end = strstr (s, "-->")))
860 /* ms may be in these formats:
861 * hh:mm:ss,500 = 500ms
864 * hh:mm:ss, 50 = 50ms
866 * and the same with . instead of ,.
867 * sscanf() doesn't differentiate between ' 5' and '5' so munge
868 * the white spaces within the timestamp to '0' (I'm sure there's a
869 * way to make sscanf() do this for us, but how?)
871 g_strdelimit (s, " ", '0');
872 g_strdelimit (s, ".", ',');
874 /* make sure we have exactly three digits after he comma */
876 g_assert (p != NULL);
883 g_strlcat (&p[len], "0", 2);
887 GST_LOG ("parsing timestamp '%s'", s);
888 if (sscanf (s, "%u:%u:%u,%u", &hour, &min, &sec, &msec) != 4) {
889 GST_WARNING ("failed to parse subrip timestamp string '%s'", s);
893 *t = ((hour * 3600) + (min * 60) + sec) * GST_SECOND + msec * GST_MSECOND;
898 parse_subrip (ParserState * state, const gchar * line)
903 switch (state->state) {
905 /* looking for a single integer */
906 if (sscanf (line, "%u", &subnum) == 1)
911 GstClockTime ts_start, ts_end;
914 /* looking for start_time --> end_time */
915 if ((end_time = strstr (line, " --> ")) &&
916 parse_subrip_time (line, &ts_start) &&
917 parse_subrip_time (end_time + strlen (" --> "), &ts_end) &&
918 state->start_time <= ts_end) {
920 state->start_time = ts_start;
921 state->duration = ts_end - ts_start;
923 GST_DEBUG ("error parsing subrip time line '%s'", line);
930 /* No need to parse that text if it's out of segment */
931 gint64 clip_start = 0, clip_stop = 0;
932 gboolean in_seg = FALSE;
934 /* Check our segment start/stop */
935 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
936 state->start_time, state->start_time + state->duration,
937 &clip_start, &clip_stop);
940 state->start_time = clip_start;
941 state->duration = clip_stop - clip_start;
947 /* looking for subtitle text; empty line ends this subtitle entry */
949 g_string_append_c (state->buf, '\n');
950 g_string_append (state->buf, line);
951 if (strlen (line) == 0) {
952 ret = g_markup_escape_text (state->buf->str, state->buf->len);
953 g_string_truncate (state->buf, 0);
955 subrip_unescape_formatting (ret);
956 subrip_remove_unhandled_tags (ret);
957 strip_trailing_newlines (ret);
958 subrip_fix_up_markup (&ret);
963 g_return_val_if_reached (NULL);
968 unescape_newlines_br (gchar * read)
972 /* Replace all occurences of '[br]' with a newline as version 2
973 * of the subviewer format uses this for newlines */
975 if (read[0] == '\0' || read[1] == '\0' || read[2] == '\0' || read[3] == '\0')
979 if (strncmp (read, "[br]", 4) == 0) {
993 parse_subviewer (ParserState * state, const gchar * line)
995 guint h1, m1, s1, ms1;
996 guint h2, m2, s2, ms2;
999 /* TODO: Maybe also parse the fields in the header, especially DELAY.
1000 * For examples see the unit test or
1001 * http://www.doom9.org/index.html?/sub.htm */
1003 switch (state->state) {
1005 /* looking for start_time,end_time */
1006 if (sscanf (line, "%u:%u:%u.%u,%u:%u:%u.%u",
1007 &h1, &m1, &s1, &ms1, &h2, &m2, &s2, &ms2) == 8) {
1010 (((guint64) h1) * 3600 + m1 * 60 + s1) * GST_SECOND +
1013 (((guint64) h2) * 3600 + m2 * 60 + s2) * GST_SECOND +
1014 ms2 * GST_MSECOND - state->start_time;
1019 /* No need to parse that text if it's out of segment */
1020 gint64 clip_start = 0, clip_stop = 0;
1021 gboolean in_seg = FALSE;
1023 /* Check our segment start/stop */
1024 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1025 state->start_time, state->start_time + state->duration,
1026 &clip_start, &clip_stop);
1029 state->start_time = clip_start;
1030 state->duration = clip_stop - clip_start;
1036 /* looking for subtitle text; empty line ends this subtitle entry */
1037 if (state->buf->len)
1038 g_string_append_c (state->buf, '\n');
1039 g_string_append (state->buf, line);
1040 if (strlen (line) == 0) {
1041 ret = g_strdup (state->buf->str);
1042 unescape_newlines_br (ret);
1043 strip_trailing_newlines (ret);
1044 g_string_truncate (state->buf, 0);
1050 g_assert_not_reached ();
1056 parse_mpsub (ParserState * state, const gchar * line)
1061 switch (state->state) {
1063 /* looking for two floats (offset, duration) */
1064 if (sscanf (line, "%f %f", &t1, &t2) == 2) {
1066 state->start_time += state->duration + GST_SECOND * t1;
1067 state->duration = GST_SECOND * t2;
1071 { /* No need to parse that text if it's out of segment */
1072 gint64 clip_start = 0, clip_stop = 0;
1073 gboolean in_seg = FALSE;
1075 /* Check our segment start/stop */
1076 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1077 state->start_time, state->start_time + state->duration,
1078 &clip_start, &clip_stop);
1081 state->start_time = clip_start;
1082 state->duration = clip_stop - clip_start;
1088 /* looking for subtitle text; empty line ends this
1090 if (state->buf->len)
1091 g_string_append_c (state->buf, '\n');
1092 g_string_append (state->buf, line);
1093 if (strlen (line) == 0) {
1094 ret = g_strdup (state->buf->str);
1095 g_string_truncate (state->buf, 0);
1101 g_assert_not_reached ();
1106 static const gchar *
1107 dks_skip_timestamp (const gchar * line)
1109 while (*line && *line != ']')
1117 parse_dks (ParserState * state, const gchar * line)
1121 switch (state->state) {
1123 /* Looking for the start time and text */
1124 if (sscanf (line, "[%u:%u:%u]", &h, &m, &s) == 3) {
1126 state->start_time = (((guint64) h) * 3600 + m * 60 + s) * GST_SECOND;
1127 text = dks_skip_timestamp (line);
1130 g_string_append (state->buf, text);
1136 gint64 clip_start = 0, clip_stop = 0;
1140 /* Looking for the end time */
1141 if (sscanf (line, "[%u:%u:%u]", &h, &m, &s) == 3) {
1143 state->duration = (((guint64) h) * 3600 + m * 60 + s) * GST_SECOND -
1146 GST_WARNING ("Failed to parse subtitle end time");
1150 /* Check if this subtitle is out of the current segment */
1151 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1152 state->start_time, state->start_time + state->duration,
1153 &clip_start, &clip_stop);
1159 state->start_time = clip_start;
1160 state->duration = clip_stop - clip_start;
1162 ret = g_strdup (state->buf->str);
1163 g_string_truncate (state->buf, 0);
1164 unescape_newlines_br (ret);
1168 g_assert_not_reached ();
1174 parser_state_init (ParserState * state)
1176 GST_DEBUG ("initialising parser");
1179 g_string_truncate (state->buf, 0);
1181 state->buf = g_string_new (NULL);
1184 state->start_time = 0;
1185 state->duration = 0;
1186 state->max_duration = 0; /* no limit */
1188 state->segment = NULL;
1192 parser_state_dispose (ParserState * state)
1195 g_string_free (state->buf, TRUE);
1198 #ifndef GST_DISABLE_XML
1199 if (state->user_data) {
1200 sami_context_reset (state);
1205 /* regex type enum */
1208 GST_SUB_PARSE_REGEX_UNKNOWN = 0,
1209 GST_SUB_PARSE_REGEX_MDVDSUB = 1,
1210 GST_SUB_PARSE_REGEX_SUBRIP = 2,
1211 GST_SUB_PARSE_REGEX_DKS = 3,
1215 gst_sub_parse_data_format_autodetect_regex_once (GstSubParseRegex regtype)
1217 gpointer result = NULL;
1218 GError *gerr = NULL;
1220 case GST_SUB_PARSE_REGEX_MDVDSUB:
1222 (gpointer) g_regex_new ("^\\{[0-9]+\\}\\{[0-9]+\\}", 0, 0, &gerr);
1223 if (result == NULL) {
1224 g_warning ("Compilation of mdvd regex failed: %s", gerr->message);
1225 g_error_free (gerr);
1228 case GST_SUB_PARSE_REGEX_SUBRIP:
1229 result = (gpointer) g_regex_new ("^([ 0-9]){0,3}[0-9]\\s*(\x0d)?\x0a"
1230 "[ 0-9][0-9]:[ 0-9][0-9]:[ 0-9][0-9][,.][ 0-9]{0,2}[0-9]"
1231 " +--> +([ 0-9])?[0-9]:[ 0-9][0-9]:[ 0-9][0-9][,.][ 0-9]{0,2}[0-9]",
1233 if (result == NULL) {
1234 g_warning ("Compilation of subrip regex failed: %s", gerr->message);
1235 g_error_free (gerr);
1238 case GST_SUB_PARSE_REGEX_DKS:
1239 result = (gpointer) g_regex_new ("^\[[0-9]+:[0-9]+:[0-9]+].*",
1241 if (result == NULL) {
1242 g_warning ("Compilation of dks regex failed: %s", gerr->message);
1243 g_error_free (gerr);
1247 GST_WARNING ("Trying to allocate regex of unknown type %u", regtype);
1253 * FIXME: maybe we should pass along a second argument, the preceding
1254 * text buffer, because that is how this originally worked, even though
1255 * I don't really see the use of that.
1258 static GstSubParseFormat
1259 gst_sub_parse_data_format_autodetect (gchar * match_str)
1263 static GOnce mdvd_rx_once = G_ONCE_INIT;
1264 static GOnce subrip_rx_once = G_ONCE_INIT;
1265 static GOnce dks_rx_once = G_ONCE_INIT;
1271 g_once (&mdvd_rx_once,
1272 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1273 (gpointer) GST_SUB_PARSE_REGEX_MDVDSUB);
1274 g_once (&subrip_rx_once,
1275 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1276 (gpointer) GST_SUB_PARSE_REGEX_SUBRIP);
1277 g_once (&dks_rx_once,
1278 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1279 (gpointer) GST_SUB_PARSE_REGEX_DKS);
1281 mdvd_grx = (GRegex *) mdvd_rx_once.retval;
1282 subrip_grx = (GRegex *) subrip_rx_once.retval;
1283 dks_grx = (GRegex *) dks_rx_once.retval;
1285 if (g_regex_match (mdvd_grx, match_str, 0, NULL) == TRUE) {
1286 GST_LOG ("MicroDVD (frame based) format detected");
1287 return GST_SUB_PARSE_FORMAT_MDVDSUB;
1289 if (g_regex_match (subrip_grx, match_str, 0, NULL) == TRUE) {
1290 GST_LOG ("SubRip (time based) format detected");
1291 return GST_SUB_PARSE_FORMAT_SUBRIP;
1293 if (g_regex_match (dks_grx, match_str, 0, NULL) == TRUE) {
1294 GST_LOG ("DKS (time based) format detected");
1295 return GST_SUB_PARSE_FORMAT_DKS;
1298 if (!strncmp (match_str, "FORMAT=TIME", 11)) {
1299 GST_LOG ("MPSub (time based) format detected");
1300 return GST_SUB_PARSE_FORMAT_MPSUB;
1302 #ifndef GST_DISABLE_XML
1303 if (strstr (match_str, "<SAMI>") != NULL ||
1304 strstr (match_str, "<sami>") != NULL) {
1305 GST_LOG ("SAMI (time based) format detected");
1306 return GST_SUB_PARSE_FORMAT_SAMI;
1309 /* we're boldly assuming the first subtitle appears within the first hour */
1310 if (sscanf (match_str, "0:%02u:%02u:", &n1, &n2) == 2 ||
1311 sscanf (match_str, "0:%02u:%02u=", &n1, &n2) == 2 ||
1312 sscanf (match_str, "00:%02u:%02u:", &n1, &n2) == 2 ||
1313 sscanf (match_str, "00:%02u:%02u=", &n1, &n2) == 2 ||
1314 sscanf (match_str, "00:%02u:%02u,%u=", &n1, &n2, &n3) == 3) {
1315 GST_LOG ("TMPlayer (time based) format detected");
1316 return GST_SUB_PARSE_FORMAT_TMPLAYER;
1318 if (sscanf (match_str, "[%u][%u]", &n1, &n2) == 2) {
1319 GST_LOG ("MPL2 (time based) format detected");
1320 return GST_SUB_PARSE_FORMAT_MPL2;
1322 if (strstr (match_str, "[INFORMATION]") != NULL) {
1323 GST_LOG ("SubViewer (time based) format detected");
1324 return GST_SUB_PARSE_FORMAT_SUBVIEWER;
1327 GST_DEBUG ("no subtitle format detected");
1328 return GST_SUB_PARSE_FORMAT_UNKNOWN;
1332 gst_sub_parse_format_autodetect (GstSubParse * self)
1335 GstSubParseFormat format;
1337 if (strlen (self->textbuf->str) < 30) {
1338 GST_DEBUG ("File too small to be a subtitles file");
1342 data = g_strndup (self->textbuf->str, 35);
1343 format = gst_sub_parse_data_format_autodetect (data);
1346 self->parser_type = format;
1347 self->subtitle_codec = gst_sub_parse_get_format_description (format);
1348 parser_state_init (&self->state);
1351 case GST_SUB_PARSE_FORMAT_MDVDSUB:
1352 self->parse_line = parse_mdvdsub;
1353 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1354 case GST_SUB_PARSE_FORMAT_SUBRIP:
1355 self->parse_line = parse_subrip;
1356 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1357 case GST_SUB_PARSE_FORMAT_MPSUB:
1358 self->parse_line = parse_mpsub;
1359 return gst_caps_new_simple ("text/plain", NULL);
1360 #ifndef GST_DISABLE_XML
1361 case GST_SUB_PARSE_FORMAT_SAMI:
1362 self->parse_line = parse_sami;
1363 sami_context_init (&self->state);
1364 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1366 case GST_SUB_PARSE_FORMAT_TMPLAYER:
1367 self->parse_line = parse_tmplayer;
1368 self->state.max_duration = 5 * GST_SECOND;
1369 return gst_caps_new_simple ("text/plain", NULL);
1370 case GST_SUB_PARSE_FORMAT_MPL2:
1371 self->parse_line = parse_mpl2;
1372 return gst_caps_new_simple ("text/x-pango-markup", NULL);
1373 case GST_SUB_PARSE_FORMAT_DKS:
1374 self->parse_line = parse_dks;
1375 return gst_caps_new_simple ("text/plain", NULL);
1376 case GST_SUB_PARSE_FORMAT_SUBVIEWER:
1377 self->parse_line = parse_subviewer;
1378 return gst_caps_new_simple ("text/plain", NULL);
1379 case GST_SUB_PARSE_FORMAT_UNKNOWN:
1381 GST_DEBUG ("no subtitle format detected");
1382 GST_ELEMENT_ERROR (self, STREAM, WRONG_TYPE,
1383 ("The input is not a valid/supported subtitle file"), (NULL));
1389 feed_textbuf (GstSubParse * self, GstBuffer * buf)
1393 gchar *input = NULL;
1395 discont = GST_BUFFER_IS_DISCONT (buf);
1397 if (GST_BUFFER_OFFSET_IS_VALID (buf) &&
1398 GST_BUFFER_OFFSET (buf) != self->offset) {
1399 self->offset = GST_BUFFER_OFFSET (buf);
1404 GST_INFO ("discontinuity");
1405 /* flush the parser state */
1406 parser_state_init (&self->state);
1407 g_string_truncate (self->textbuf, 0);
1408 gst_adapter_clear (self->adapter);
1409 #ifndef GST_DISABLE_XML
1410 sami_context_reset (&self->state);
1412 /* we could set a flag to make sure that the next buffer we push out also
1413 * has the DISCONT flag set, but there's no point really given that it's
1414 * subtitles which are discontinuous by nature. */
1417 self->offset = GST_BUFFER_OFFSET (buf) + GST_BUFFER_SIZE (buf);
1418 self->next_offset = self->offset;
1420 gst_adapter_push (self->adapter, buf);
1423 convert_encoding (self, (const gchar *) gst_adapter_peek (self->adapter,
1424 gst_adapter_available (self->adapter)),
1425 (gsize) gst_adapter_available (self->adapter), &consumed);
1427 if (input && consumed > 0) {
1428 self->textbuf = g_string_append (self->textbuf, input);
1429 gst_adapter_flush (self->adapter, consumed);
1435 static GstFlowReturn
1436 handle_buffer (GstSubParse * self, GstBuffer * buf)
1438 GstFlowReturn ret = GST_FLOW_OK;
1439 GstCaps *caps = NULL;
1440 gchar *line, *subtitle;
1442 if (self->first_buffer) {
1443 self->detected_encoding =
1444 detect_encoding ((gchar *) GST_BUFFER_DATA (buf),
1445 GST_BUFFER_SIZE (buf));
1446 self->first_buffer = FALSE;
1447 self->state.fps_n = self->fps_n;
1448 self->state.fps_d = self->fps_d;
1451 feed_textbuf (self, buf);
1453 /* make sure we know the format */
1454 if (G_UNLIKELY (self->parser_type == GST_SUB_PARSE_FORMAT_UNKNOWN)) {
1455 if (!(caps = gst_sub_parse_format_autodetect (self))) {
1456 return GST_FLOW_UNEXPECTED;
1458 if (!gst_pad_set_caps (self->srcpad, caps)) {
1459 gst_caps_unref (caps);
1460 return GST_FLOW_UNEXPECTED;
1462 gst_caps_unref (caps);
1465 if (self->subtitle_codec != NULL) {
1468 tags = gst_tag_list_new ();
1469 gst_tag_list_add (tags, GST_TAG_MERGE_APPEND, GST_TAG_SUBTITLE_CODEC,
1470 self->subtitle_codec, NULL);
1471 gst_element_found_tags_for_pad (GST_ELEMENT (self), self->srcpad, tags);
1475 while (!self->flushing && (line = get_next_line (self))) {
1478 /* Set segment on our parser state machine */
1479 self->state.segment = &self->segment;
1480 /* Now parse the line, out of segment lines will just return NULL */
1481 GST_LOG_OBJECT (self, "Parsing line '%s'", line + offset);
1482 subtitle = self->parse_line (&self->state, line + offset);
1486 guint subtitle_len = strlen (subtitle);
1488 /* +1 for terminating NUL character */
1489 ret = gst_pad_alloc_buffer_and_set_caps (self->srcpad,
1490 GST_BUFFER_OFFSET_NONE, subtitle_len + 1,
1491 GST_PAD_CAPS (self->srcpad), &buf);
1493 if (ret == GST_FLOW_OK) {
1494 /* copy terminating NUL character as well */
1495 memcpy (GST_BUFFER_DATA (buf), subtitle, subtitle_len + 1);
1496 GST_BUFFER_SIZE (buf) = subtitle_len;
1497 GST_BUFFER_TIMESTAMP (buf) = self->state.start_time;
1498 GST_BUFFER_DURATION (buf) = self->state.duration;
1500 /* in some cases (e.g. tmplayer) we can only determine the duration
1501 * of a text chunk from the timestamp of the next text chunk; in those
1502 * cases, we probably want to limit the duration to something
1503 * reasonable, so we don't end up showing some text for e.g. 40 seconds
1504 * just because nothing else is being said during that time */
1505 if (self->state.max_duration > 0 && GST_BUFFER_DURATION_IS_VALID (buf)) {
1506 if (GST_BUFFER_DURATION (buf) > self->state.max_duration)
1507 GST_BUFFER_DURATION (buf) = self->state.max_duration;
1510 gst_segment_set_last_stop (&self->segment, GST_FORMAT_TIME,
1511 self->state.start_time);
1513 GST_DEBUG_OBJECT (self, "Sending text '%s', %" GST_TIME_FORMAT " + %"
1514 GST_TIME_FORMAT, subtitle, GST_TIME_ARGS (self->state.start_time),
1515 GST_TIME_ARGS (self->state.duration));
1517 ret = gst_pad_push (self->srcpad, buf);
1520 /* move this forward (the tmplayer parser needs this) */
1521 if (self->state.duration != GST_CLOCK_TIME_NONE)
1522 self->state.start_time += self->state.duration;
1527 if (ret != GST_FLOW_OK) {
1528 GST_DEBUG_OBJECT (self, "flow: %s", gst_flow_get_name (ret));
1537 static GstFlowReturn
1538 gst_sub_parse_chain (GstPad * sinkpad, GstBuffer * buf)
1543 self = GST_SUBPARSE (GST_PAD_PARENT (sinkpad));
1545 /* Push newsegment if needed */
1546 if (self->need_segment) {
1547 GST_LOG_OBJECT (self, "pushing newsegment event with %" GST_SEGMENT_FORMAT,
1550 gst_pad_push_event (self->srcpad, gst_event_new_new_segment (FALSE,
1551 self->segment.rate, self->segment.format,
1552 self->segment.last_stop, self->segment.stop, self->segment.time));
1553 self->need_segment = FALSE;
1556 ret = handle_buffer (self, buf);
1562 gst_sub_parse_sink_event (GstPad * pad, GstEvent * event)
1564 GstSubParse *self = GST_SUBPARSE (gst_pad_get_parent (pad));
1565 gboolean ret = FALSE;
1567 GST_DEBUG ("Handling %s event", GST_EVENT_TYPE_NAME (event));
1569 switch (GST_EVENT_TYPE (event)) {
1570 case GST_EVENT_EOS:{
1571 /* Make sure the last subrip chunk is pushed out even
1572 * if the file does not have an empty line at the end */
1573 if (self->parser_type == GST_SUB_PARSE_FORMAT_SUBRIP ||
1574 self->parser_type == GST_SUB_PARSE_FORMAT_TMPLAYER ||
1575 self->parser_type == GST_SUB_PARSE_FORMAT_MPL2) {
1576 GstBuffer *buf = gst_buffer_new_and_alloc (2 + 1);
1578 GST_DEBUG ("EOS. Pushing remaining text (if any)");
1579 GST_BUFFER_DATA (buf)[0] = '\n';
1580 GST_BUFFER_DATA (buf)[1] = '\n';
1581 GST_BUFFER_DATA (buf)[2] = '\0'; /* play it safe */
1582 GST_BUFFER_SIZE (buf) = 2;
1583 GST_BUFFER_OFFSET (buf) = self->offset;
1584 gst_sub_parse_chain (pad, buf);
1586 ret = gst_pad_event_default (pad, event);
1589 case GST_EVENT_NEWSEGMENT:
1593 gint64 start, stop, time;
1596 gst_event_parse_new_segment (event, &update, &rate, &format, &start,
1599 GST_DEBUG_OBJECT (self, "newsegment (%s)", gst_format_get_name (format));
1601 if (format == GST_FORMAT_TIME) {
1602 gst_segment_set_newsegment (&self->segment, update, rate, format,
1605 /* if not time format, we'll either start with a 0 timestamp anyway or
1606 * it's following a seek in which case we'll have saved the requested
1607 * seek segment and don't want to overwrite it (remember that on a seek
1608 * we always just seek back to the start in BYTES format and just throw
1609 * away all text that's before the requested position; if the subtitles
1610 * come from an upstream demuxer, it won't be able to handle our BYTES
1611 * seek request and instead send us a newsegment from the seek request
1612 * it received via its video pads instead, so all is fine then too) */
1616 gst_event_unref (event);
1619 case GST_EVENT_FLUSH_START:
1621 self->flushing = TRUE;
1623 ret = gst_pad_event_default (pad, event);
1626 case GST_EVENT_FLUSH_STOP:
1628 self->flushing = FALSE;
1630 ret = gst_pad_event_default (pad, event);
1634 ret = gst_pad_event_default (pad, event);
1638 gst_object_unref (self);
1644 static GstStateChangeReturn
1645 gst_sub_parse_change_state (GstElement * element, GstStateChange transition)
1647 GstStateChangeReturn ret = GST_STATE_CHANGE_SUCCESS;
1648 GstSubParse *self = GST_SUBPARSE (element);
1650 switch (transition) {
1651 case GST_STATE_CHANGE_READY_TO_PAUSED:
1652 /* format detection will init the parser state */
1654 self->next_offset = 0;
1655 self->parser_type = GST_SUB_PARSE_FORMAT_UNKNOWN;
1656 self->valid_utf8 = TRUE;
1657 self->first_buffer = TRUE;
1658 g_free (self->detected_encoding);
1659 self->detected_encoding = NULL;
1660 g_string_truncate (self->textbuf, 0);
1661 gst_adapter_clear (self->adapter);
1667 ret = parent_class->change_state (element, transition);
1668 if (ret == GST_STATE_CHANGE_FAILURE)
1671 switch (transition) {
1672 case GST_STATE_CHANGE_PAUSED_TO_READY:
1673 parser_state_dispose (&self->state);
1674 self->parser_type = GST_SUB_PARSE_FORMAT_UNKNOWN;
1687 /* FIXME 0.11: these caps are ugly, use app/x-subtitle + type field or so;
1688 * also, give different subtitle formats really different types */
1689 static GstStaticCaps mpl2_caps =
1690 GST_STATIC_CAPS ("application/x-subtitle-mpl2");
1691 #define SUB_CAPS (gst_static_caps_get (&sub_caps))
1693 static GstStaticCaps tmp_caps =
1694 GST_STATIC_CAPS ("application/x-subtitle-tmplayer");
1695 #define TMP_CAPS (gst_static_caps_get (&tmp_caps))
1697 static GstStaticCaps sub_caps = GST_STATIC_CAPS ("application/x-subtitle");
1698 #define MPL2_CAPS (gst_static_caps_get (&mpl2_caps))
1700 #ifndef GST_DISABLE_XML
1701 static GstStaticCaps smi_caps = GST_STATIC_CAPS ("application/x-subtitle-sami");
1702 #define SAMI_CAPS (gst_static_caps_get (&smi_caps))
1705 static GstStaticCaps dks_caps = GST_STATIC_CAPS ("application/x-subtitle-dks");
1706 #define DKS_CAPS (gst_static_caps_get (&dks_caps))
1709 gst_subparse_type_find (GstTypeFind * tf, gpointer private)
1711 GstSubParseFormat format;
1715 gchar *encoding = NULL;
1718 if (!(data = gst_type_find_peek (tf, 0, 129)))
1721 /* make sure string passed to _autodetect() is NUL-terminated */
1722 str = g_malloc0 (129);
1723 memcpy (str, data, 128);
1725 if ((encoding = detect_encoding (str, 128)) != NULL) {
1726 gchar *converted_str;
1730 converted_str = gst_convert_to_utf8 (str, 128, encoding, &tmp, &err);
1731 if (converted_str == NULL) {
1732 GST_DEBUG ("Encoding '%s' detected but conversion failed: %s", encoding,
1738 str = converted_str;
1743 /* Check if at least the first 120 chars are valid UTF8,
1744 * otherwise convert as always */
1745 if (!g_utf8_validate (str, 128, &end) && (end - str) < 120) {
1746 gchar *converted_str;
1751 enc = g_getenv ("GST_SUBTITLE_ENCODING");
1752 if (enc == NULL || *enc == '\0') {
1753 /* if local encoding is UTF-8 and no encoding specified
1754 * via the environment variable, assume ISO-8859-15 */
1755 if (g_get_charset (&enc)) {
1756 enc = "ISO-8859-15";
1759 converted_str = gst_convert_to_utf8 (str, 128, enc, &tmp, &err);
1760 if (converted_str == NULL) {
1761 GST_DEBUG ("Charset conversion failed: %s", err->message);
1767 str = converted_str;
1771 format = gst_sub_parse_data_format_autodetect (str);
1775 case GST_SUB_PARSE_FORMAT_MDVDSUB:
1776 GST_DEBUG ("MicroDVD format detected");
1779 case GST_SUB_PARSE_FORMAT_SUBRIP:
1780 GST_DEBUG ("SubRip format detected");
1783 case GST_SUB_PARSE_FORMAT_MPSUB:
1784 GST_DEBUG ("MPSub format detected");
1787 #ifndef GST_DISABLE_XML
1788 case GST_SUB_PARSE_FORMAT_SAMI:
1789 GST_DEBUG ("SAMI (time-based) format detected");
1793 case GST_SUB_PARSE_FORMAT_TMPLAYER:
1794 GST_DEBUG ("TMPlayer (time based) format detected");
1797 /* FIXME: our MPL2 typefinding is not really good enough to warrant
1798 * returning a high probability (however, since we registered our
1799 * typefinder here with a rank of MARGINAL we should pretty much only
1800 * be called if most other typefinders have already run */
1801 case GST_SUB_PARSE_FORMAT_MPL2:
1802 GST_DEBUG ("MPL2 (time based) format detected");
1805 case GST_SUB_PARSE_FORMAT_SUBVIEWER:
1806 GST_DEBUG ("SubViewer format detected");
1809 case GST_SUB_PARSE_FORMAT_DKS:
1810 GST_DEBUG ("DKS format detected");
1814 case GST_SUB_PARSE_FORMAT_UNKNOWN:
1815 GST_DEBUG ("no subtitle format detected");
1819 /* if we're here, it's ok */
1820 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, caps);
1824 plugin_init (GstPlugin * plugin)
1826 static gchar *sub_exts[] = { "srt", "sub", "mpsub", "mdvd", "smi", "txt",
1830 GST_DEBUG_CATEGORY_INIT (sub_parse_debug, "subparse", 0, ".sub parser");
1832 if (!gst_type_find_register (plugin, "subparse_typefind", GST_RANK_MARGINAL,
1833 gst_subparse_type_find, sub_exts, SUB_CAPS, NULL, NULL))
1836 if (!gst_element_register (plugin, "subparse",
1837 GST_RANK_PRIMARY, GST_TYPE_SUBPARSE) ||
1838 !gst_element_register (plugin, "ssaparse",
1839 GST_RANK_PRIMARY, GST_TYPE_SSA_PARSE)) {
1846 GST_PLUGIN_DEFINE (GST_VERSION_MAJOR,
1850 plugin_init, VERSION, "LGPL", GST_PACKAGE_NAME, GST_PACKAGE_ORIGIN)