1 /* GStreamer
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>
5 * Copyright (C) 2016 Philippe Normand <pnormand@igalia.com>
6 * Copyright (C) 2016 Jan Schmidt <jan@centricular.com>
7 *
8 * This library is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU Library General Public
10 * License as published by the Free Software Foundation; either
11 * version 2 of the License, or (at your option) any later version.
12 *
13 * This library is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 * Library General Public License for more details.
17 *
18 * You should have received a copy of the GNU Library General Public
19 * License along with this library; if not, write to the
20 * Free Software Foundation, Inc., 51 Franklin St, Fifth Floor,
21 * Boston, MA 02110-1301, USA.
22 */
23
24 #ifdef HAVE_CONFIG_H
25 #include "config.h"
26 #endif
27
28 #include <stdio.h>
29 #include <stdlib.h>
30 #include <string.h>
31 #include <sys/types.h>
32 #include <glib.h>
33
34 #include "gstsubparse.h"
35 #include "gstssaparse.h"
36 #include "samiparse.h"
37 #include "tmplayerparse.h"
38 #include "mpl2parse.h"
39 #include "qttextparse.h"
40
41 GST_DEBUG_CATEGORY (sub_parse_debug);
42
43 #define DEFAULT_ENCODING NULL
44 #define ATTRIBUTE_REGEX "\\s?[a-zA-Z0-9\\. \t\\(\\)]*"
45 static const gchar *allowed_srt_tags[] = { "i", "b", "u", NULL };
46 static const gchar *allowed_vtt_tags[] =
47 { "i", "b", "c", "u", "v", "ruby", "rt", NULL };
48
49 enum
50 {
51 PROP_0,
52 PROP_ENCODING,
53 PROP_VIDEOFPS
54 };
55
56 static void
57 gst_sub_parse_set_property (GObject * object, guint prop_id,
58 const GValue * value, GParamSpec * pspec);
59 static void
60 gst_sub_parse_get_property (GObject * object, guint prop_id,
61 GValue * value, GParamSpec * pspec);
62
63
64 static GstStaticPadTemplate sink_templ = GST_STATIC_PAD_TEMPLATE ("sink",
65 GST_PAD_SINK,
66 GST_PAD_ALWAYS,
67 GST_STATIC_CAPS ("application/x-subtitle; application/x-subtitle-sami; "
68 "application/x-subtitle-tmplayer; application/x-subtitle-mpl2; "
69 "application/x-subtitle-dks; application/x-subtitle-qttext;"
70 "application/x-subtitle-lrc; application/x-subtitle-vtt")
71 );
72
73 static GstStaticPadTemplate src_templ = GST_STATIC_PAD_TEMPLATE ("src",
74 GST_PAD_SRC,
75 GST_PAD_ALWAYS,
76 GST_STATIC_CAPS ("text/x-raw, format= { pango-markup, utf8 }")
77 );
78
79
80 static gboolean gst_sub_parse_src_event (GstPad * pad, GstObject * parent,
81 GstEvent * event);
82 static gboolean gst_sub_parse_src_query (GstPad * pad, GstObject * parent,
83 GstQuery * query);
84 static gboolean gst_sub_parse_sink_event (GstPad * pad, GstObject * parent,
85 GstEvent * event);
86
87 static GstStateChangeReturn gst_sub_parse_change_state (GstElement * element,
88 GstStateChange transition);
89
90 static GstFlowReturn gst_sub_parse_chain (GstPad * sinkpad, GstObject * parent,
91 GstBuffer * buf);
92
93 #define gst_sub_parse_parent_class parent_class
94 G_DEFINE_TYPE (GstSubParse, gst_sub_parse, GST_TYPE_ELEMENT);
95
96 static void
gst_sub_parse_dispose(GObject * object)97 gst_sub_parse_dispose (GObject * object)
98 {
99 GstSubParse *subparse = GST_SUBPARSE (object);
100
101 GST_DEBUG_OBJECT (subparse, "cleaning up subtitle parser");
102
103 if (subparse->encoding) {
104 g_free (subparse->encoding);
105 subparse->encoding = NULL;
106 }
107
108 if (subparse->detected_encoding) {
109 g_free (subparse->detected_encoding);
110 subparse->detected_encoding = NULL;
111 }
112
113 if (subparse->adapter) {
114 g_object_unref (subparse->adapter);
115 subparse->adapter = NULL;
116 }
117
118 if (subparse->textbuf) {
119 g_string_free (subparse->textbuf, TRUE);
120 subparse->textbuf = NULL;
121 }
122
123 GST_CALL_PARENT (G_OBJECT_CLASS, dispose, (object));
124 }
125
126 static void
gst_sub_parse_class_init(GstSubParseClass * klass)127 gst_sub_parse_class_init (GstSubParseClass * klass)
128 {
129 GObjectClass *object_class = G_OBJECT_CLASS (klass);
130 GstElementClass *element_class = GST_ELEMENT_CLASS (klass);
131
132 object_class->dispose = gst_sub_parse_dispose;
133 object_class->set_property = gst_sub_parse_set_property;
134 object_class->get_property = gst_sub_parse_get_property;
135
136 gst_element_class_add_static_pad_template (element_class, &sink_templ);
137 gst_element_class_add_static_pad_template (element_class, &src_templ);
138 gst_element_class_set_static_metadata (element_class,
139 "Subtitle parser", "Codec/Parser/Subtitle",
140 "Parses subtitle (.sub) files into text streams",
141 "Gustavo J. A. M. Carneiro <gjc@inescporto.pt>, "
142 "GStreamer maintainers <gstreamer-devel@lists.freedesktop.org>");
143
144 element_class->change_state = gst_sub_parse_change_state;
145
146 g_object_class_install_property (object_class, PROP_ENCODING,
147 g_param_spec_string ("subtitle-encoding", "subtitle charset encoding",
148 "Encoding to assume if input subtitles are not in UTF-8 or any other "
149 "Unicode encoding. If not set, the GST_SUBTITLE_ENCODING environment "
150 "variable will be checked for an encoding to use. If that is not set "
151 "either, ISO-8859-15 will be assumed.", DEFAULT_ENCODING,
152 G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
153
154 g_object_class_install_property (object_class, PROP_VIDEOFPS,
155 gst_param_spec_fraction ("video-fps", "Video framerate",
156 "Framerate of the video stream. This is needed by some subtitle "
157 "formats to synchronize subtitles and video properly. If not set "
158 "and the subtitle format requires it subtitles may be out of sync.",
159 0, 1, G_MAXINT, 1, 24000, 1001,
160 G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS));
161 }
162
163 static void
gst_sub_parse_init(GstSubParse * subparse)164 gst_sub_parse_init (GstSubParse * subparse)
165 {
166 subparse->sinkpad = gst_pad_new_from_static_template (&sink_templ, "sink");
167 gst_pad_set_chain_function (subparse->sinkpad,
168 GST_DEBUG_FUNCPTR (gst_sub_parse_chain));
169 gst_pad_set_event_function (subparse->sinkpad,
170 GST_DEBUG_FUNCPTR (gst_sub_parse_sink_event));
171 gst_element_add_pad (GST_ELEMENT (subparse), subparse->sinkpad);
172
173 subparse->srcpad = gst_pad_new_from_static_template (&src_templ, "src");
174 gst_pad_set_event_function (subparse->srcpad,
175 GST_DEBUG_FUNCPTR (gst_sub_parse_src_event));
176 gst_pad_set_query_function (subparse->srcpad,
177 GST_DEBUG_FUNCPTR (gst_sub_parse_src_query));
178 gst_element_add_pad (GST_ELEMENT (subparse), subparse->srcpad);
179
180 subparse->textbuf = g_string_new (NULL);
181 subparse->parser_type = GST_SUB_PARSE_FORMAT_UNKNOWN;
182 subparse->flushing = FALSE;
183 gst_segment_init (&subparse->segment, GST_FORMAT_TIME);
184 subparse->need_segment = TRUE;
185 subparse->encoding = g_strdup (DEFAULT_ENCODING);
186 subparse->detected_encoding = NULL;
187 subparse->adapter = gst_adapter_new ();
188
189 subparse->fps_n = 24000;
190 subparse->fps_d = 1001;
191 }
192
193 /*
194 * Source pad functions.
195 */
196
197 static gboolean
gst_sub_parse_src_query(GstPad * pad,GstObject * parent,GstQuery * query)198 gst_sub_parse_src_query (GstPad * pad, GstObject * parent, GstQuery * query)
199 {
200 GstSubParse *self = GST_SUBPARSE (parent);
201 gboolean ret = FALSE;
202
203 GST_DEBUG ("Handling %s query", GST_QUERY_TYPE_NAME (query));
204
205 switch (GST_QUERY_TYPE (query)) {
206 case GST_QUERY_POSITION:{
207 GstFormat fmt;
208
209 gst_query_parse_position (query, &fmt, NULL);
210 if (fmt != GST_FORMAT_TIME) {
211 ret = gst_pad_peer_query (self->sinkpad, query);
212 } else {
213 ret = TRUE;
214 gst_query_set_position (query, GST_FORMAT_TIME, self->segment.position);
215 }
216 break;
217 }
218 case GST_QUERY_SEEKING:
219 {
220 GstFormat fmt;
221 gboolean seekable = FALSE;
222
223 ret = TRUE;
224
225 gst_query_parse_seeking (query, &fmt, NULL, NULL, NULL);
226 if (fmt == GST_FORMAT_TIME) {
227 GstQuery *peerquery = gst_query_new_seeking (GST_FORMAT_BYTES);
228
229 seekable = gst_pad_peer_query (self->sinkpad, peerquery);
230 if (seekable)
231 gst_query_parse_seeking (peerquery, NULL, &seekable, NULL, NULL);
232 gst_query_unref (peerquery);
233 }
234
235 gst_query_set_seeking (query, fmt, seekable, seekable ? 0 : -1, -1);
236 break;
237 }
238 default:
239 ret = gst_pad_query_default (pad, parent, query);
240 break;
241 }
242
243 return ret;
244 }
245
246 static gboolean
gst_sub_parse_src_event(GstPad * pad,GstObject * parent,GstEvent * event)247 gst_sub_parse_src_event (GstPad * pad, GstObject * parent, GstEvent * event)
248 {
249 GstSubParse *self = GST_SUBPARSE (parent);
250 gboolean ret = FALSE;
251
252 GST_DEBUG ("Handling %s event", GST_EVENT_TYPE_NAME (event));
253
254 switch (GST_EVENT_TYPE (event)) {
255 case GST_EVENT_SEEK:
256 {
257 GstFormat format;
258 GstSeekFlags flags;
259 GstSeekType start_type, stop_type;
260 gint64 start, stop;
261 gdouble rate;
262 gboolean update;
263
264 gst_event_parse_seek (event, &rate, &format, &flags,
265 &start_type, &start, &stop_type, &stop);
266
267 if (format != GST_FORMAT_TIME) {
268 GST_WARNING_OBJECT (self, "we only support seeking in TIME format");
269 gst_event_unref (event);
270 goto beach;
271 }
272
273 /* Convert that seek to a seeking in bytes at position 0,
274 FIXME: could use an index */
275 ret = gst_pad_push_event (self->sinkpad,
276 gst_event_new_seek (rate, GST_FORMAT_BYTES, flags,
277 GST_SEEK_TYPE_SET, 0, GST_SEEK_TYPE_NONE, 0));
278
279 if (ret) {
280 /* Apply the seek to our segment */
281 gst_segment_do_seek (&self->segment, rate, format, flags,
282 start_type, start, stop_type, stop, &update);
283
284 GST_DEBUG_OBJECT (self, "segment after seek: %" GST_SEGMENT_FORMAT,
285 &self->segment);
286
287 /* will mark need_segment when receiving segment from upstream,
288 * after FLUSH and all that has happened,
289 * rather than racing with chain */
290 } else {
291 GST_WARNING_OBJECT (self, "seek to 0 bytes failed");
292 }
293
294 gst_event_unref (event);
295 break;
296 }
297 default:
298 ret = gst_pad_event_default (pad, parent, event);
299 break;
300 }
301
302 beach:
303 return ret;
304 }
305
306 static void
gst_sub_parse_set_property(GObject * object,guint prop_id,const GValue * value,GParamSpec * pspec)307 gst_sub_parse_set_property (GObject * object, guint prop_id,
308 const GValue * value, GParamSpec * pspec)
309 {
310 GstSubParse *subparse = GST_SUBPARSE (object);
311
312 GST_OBJECT_LOCK (subparse);
313 switch (prop_id) {
314 case PROP_ENCODING:
315 g_free (subparse->encoding);
316 subparse->encoding = g_value_dup_string (value);
317 GST_LOG_OBJECT (object, "subtitle encoding set to %s",
318 GST_STR_NULL (subparse->encoding));
319 break;
320 case PROP_VIDEOFPS:
321 {
322 subparse->fps_n = gst_value_get_fraction_numerator (value);
323 subparse->fps_d = gst_value_get_fraction_denominator (value);
324 GST_DEBUG_OBJECT (object, "video framerate set to %d/%d", subparse->fps_n,
325 subparse->fps_d);
326
327 if (!subparse->state.have_internal_fps) {
328 subparse->state.fps_n = subparse->fps_n;
329 subparse->state.fps_d = subparse->fps_d;
330 }
331 break;
332 }
333 default:
334 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
335 break;
336 }
337 GST_OBJECT_UNLOCK (subparse);
338 }
339
340 static void
gst_sub_parse_get_property(GObject * object,guint prop_id,GValue * value,GParamSpec * pspec)341 gst_sub_parse_get_property (GObject * object, guint prop_id,
342 GValue * value, GParamSpec * pspec)
343 {
344 GstSubParse *subparse = GST_SUBPARSE (object);
345
346 GST_OBJECT_LOCK (subparse);
347 switch (prop_id) {
348 case PROP_ENCODING:
349 g_value_set_string (value, subparse->encoding);
350 break;
351 case PROP_VIDEOFPS:
352 gst_value_set_fraction (value, subparse->fps_n, subparse->fps_d);
353 break;
354 default:
355 G_OBJECT_WARN_INVALID_PROPERTY_ID (object, prop_id, pspec);
356 break;
357 }
358 GST_OBJECT_UNLOCK (subparse);
359 }
360
361 static const gchar *
gst_sub_parse_get_format_description(GstSubParseFormat format)362 gst_sub_parse_get_format_description (GstSubParseFormat format)
363 {
364 switch (format) {
365 case GST_SUB_PARSE_FORMAT_MDVDSUB:
366 return "MicroDVD";
367 case GST_SUB_PARSE_FORMAT_SUBRIP:
368 return "SubRip";
369 case GST_SUB_PARSE_FORMAT_MPSUB:
370 return "MPSub";
371 case GST_SUB_PARSE_FORMAT_SAMI:
372 return "SAMI";
373 case GST_SUB_PARSE_FORMAT_TMPLAYER:
374 return "TMPlayer";
375 case GST_SUB_PARSE_FORMAT_MPL2:
376 return "MPL2";
377 case GST_SUB_PARSE_FORMAT_SUBVIEWER:
378 return "SubViewer";
379 case GST_SUB_PARSE_FORMAT_DKS:
380 return "DKS";
381 case GST_SUB_PARSE_FORMAT_VTT:
382 return "WebVTT";
383 case GST_SUB_PARSE_FORMAT_QTTEXT:
384 return "QTtext";
385 case GST_SUB_PARSE_FORMAT_LRC:
386 return "LRC";
387 default:
388 case GST_SUB_PARSE_FORMAT_UNKNOWN:
389 break;
390 }
391 return NULL;
392 }
393
394 static gchar *
gst_convert_to_utf8(const gchar * str,gsize len,const gchar * encoding,gsize * consumed,GError ** err)395 gst_convert_to_utf8 (const gchar * str, gsize len, const gchar * encoding,
396 gsize * consumed, GError ** err)
397 {
398 gchar *ret = NULL;
399
400 *consumed = 0;
401 /* The char cast is necessary in glib < 2.24 */
402 ret =
403 g_convert_with_fallback (str, len, "UTF-8", encoding, (char *) "*",
404 consumed, NULL, err);
405 if (ret == NULL)
406 return ret;
407
408 /* + 3 to skip UTF-8 BOM if it was added */
409 len = strlen (ret);
410 if (len >= 3 && (guint8) ret[0] == 0xEF && (guint8) ret[1] == 0xBB
411 && (guint8) ret[2] == 0xBF)
412 memmove (ret, ret + 3, len + 1 - 3);
413
414 return ret;
415 }
416
417 static gchar *
detect_encoding(const gchar * str,gsize len)418 detect_encoding (const gchar * str, gsize len)
419 {
420 if (len >= 3 && (guint8) str[0] == 0xEF && (guint8) str[1] == 0xBB
421 && (guint8) str[2] == 0xBF)
422 return g_strdup ("UTF-8");
423
424 if (len >= 2 && (guint8) str[0] == 0xFE && (guint8) str[1] == 0xFF)
425 return g_strdup ("UTF-16BE");
426
427 if (len >= 2 && (guint8) str[0] == 0xFF && (guint8) str[1] == 0xFE)
428 return g_strdup ("UTF-16LE");
429
430 if (len >= 4 && (guint8) str[0] == 0x00 && (guint8) str[1] == 0x00
431 && (guint8) str[2] == 0xFE && (guint8) str[3] == 0xFF)
432 return g_strdup ("UTF-32BE");
433
434 if (len >= 4 && (guint8) str[0] == 0xFF && (guint8) str[1] == 0xFE
435 && (guint8) str[2] == 0x00 && (guint8) str[3] == 0x00)
436 return g_strdup ("UTF-32LE");
437
438 return NULL;
439 }
440
441 static gchar *
convert_encoding(GstSubParse * self,const gchar * str,gsize len,gsize * consumed)442 convert_encoding (GstSubParse * self, const gchar * str, gsize len,
443 gsize * consumed)
444 {
445 const gchar *encoding;
446 GError *err = NULL;
447 gchar *ret = NULL;
448
449 *consumed = 0;
450
451 /* First try any detected encoding */
452 if (self->detected_encoding) {
453 ret =
454 gst_convert_to_utf8 (str, len, self->detected_encoding, consumed, &err);
455
456 if (!err)
457 return ret;
458
459 GST_WARNING_OBJECT (self, "could not convert string from '%s' to UTF-8: %s",
460 self->detected_encoding, err->message);
461 g_free (self->detected_encoding);
462 self->detected_encoding = NULL;
463 g_clear_error (&err);
464 }
465
466 /* Otherwise check if it's UTF8 */
467 if (self->valid_utf8) {
468 if (g_utf8_validate (str, len, NULL)) {
469 GST_LOG_OBJECT (self, "valid UTF-8, no conversion needed");
470 *consumed = len;
471 return g_strndup (str, len);
472 }
473 GST_INFO_OBJECT (self, "invalid UTF-8!");
474 self->valid_utf8 = FALSE;
475 }
476
477 /* Else try fallback */
478 encoding = self->encoding;
479 if (encoding == NULL || *encoding == '\0') {
480 encoding = g_getenv ("GST_SUBTITLE_ENCODING");
481 }
482 if (encoding == NULL || *encoding == '\0') {
483 /* if local encoding is UTF-8 and no encoding specified
484 * via the environment variable, assume ISO-8859-15 */
485 if (g_get_charset (&encoding)) {
486 encoding = "ISO-8859-15";
487 }
488 }
489
490 ret = gst_convert_to_utf8 (str, len, encoding, consumed, &err);
491
492 if (err) {
493 GST_WARNING_OBJECT (self, "could not convert string from '%s' to UTF-8: %s",
494 encoding, err->message);
495 g_clear_error (&err);
496
497 /* invalid input encoding, fall back to ISO-8859-15 (always succeeds) */
498 ret = gst_convert_to_utf8 (str, len, "ISO-8859-15", consumed, NULL);
499 }
500
501 GST_LOG_OBJECT (self,
502 "successfully converted %" G_GSIZE_FORMAT " characters from %s to UTF-8"
503 "%s", len, encoding, (err) ? " , using ISO-8859-15 as fallback" : "");
504
505 return ret;
506 }
507
508 static gchar *
get_next_line(GstSubParse * self)509 get_next_line (GstSubParse * self)
510 {
511 char *line = NULL;
512 const char *line_end;
513 int line_len;
514 gboolean have_r = FALSE;
515
516 line_end = strchr (self->textbuf->str, '\n');
517
518 if (!line_end) {
519 /* end-of-line not found; return for more data */
520 return NULL;
521 }
522
523 /* get rid of '\r' */
524 if (line_end != self->textbuf->str && *(line_end - 1) == '\r') {
525 line_end--;
526 have_r = TRUE;
527 }
528
529 line_len = line_end - self->textbuf->str;
530 line = g_strndup (self->textbuf->str, line_len);
531 self->textbuf = g_string_erase (self->textbuf, 0,
532 line_len + (have_r ? 2 : 1));
533 return line;
534 }
535
536 static gchar *
parse_mdvdsub(ParserState * state,const gchar * line)537 parse_mdvdsub (ParserState * state, const gchar * line)
538 {
539 const gchar *line_split;
540 gchar *line_chunk;
541 guint start_frame, end_frame;
542 guint64 clip_start = 0, clip_stop = 0;
543 gboolean in_seg = FALSE;
544 GString *markup;
545 gchar *ret;
546
547 /* style variables */
548 gboolean italic;
549 gboolean bold;
550 guint fontsize;
551 gdouble fps = 0.0;
552
553 if (sscanf (line, "{%u}{%u}", &start_frame, &end_frame) != 2) {
554 g_warning ("Parse of the following line, assumed to be in microdvd .sub"
555 " format, failed:\n%s", line);
556 return NULL;
557 }
558
559 /* skip the {%u}{%u} part */
560 line = strchr (line, '}') + 1;
561 line = strchr (line, '}') + 1;
562
563 /* see if there's a first line with a framerate */
564 if (start_frame == 1 && end_frame == 1) {
565 gchar *rest, *end = NULL;
566
567 rest = g_strdup (line);
568 g_strdelimit (rest, ",", '.');
569 fps = g_ascii_strtod (rest, &end);
570 if (end != rest) {
571 gst_util_double_to_fraction (fps, &state->fps_n, &state->fps_d);
572 GST_INFO ("framerate from file: %d/%d ('%s')", state->fps_n,
573 state->fps_d, rest);
574 }
575 g_free (rest);
576 return NULL;
577 }
578
579 state->start_time =
580 gst_util_uint64_scale (start_frame, GST_SECOND * state->fps_d,
581 state->fps_n);
582 state->duration =
583 gst_util_uint64_scale (end_frame - start_frame, GST_SECOND * state->fps_d,
584 state->fps_n);
585
586 /* Check our segment start/stop */
587 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
588 state->start_time, state->start_time + state->duration, &clip_start,
589 &clip_stop);
590
591 /* No need to parse that text if it's out of segment */
592 if (in_seg) {
593 state->start_time = clip_start;
594 state->duration = clip_stop - clip_start;
595 } else {
596 return NULL;
597 }
598
599 markup = g_string_new (NULL);
600 while (1) {
601 italic = FALSE;
602 bold = FALSE;
603 fontsize = 0;
604 /* parse style markup */
605 if (strncmp (line, "{y:i}", 5) == 0) {
606 italic = TRUE;
607 line = strchr (line, '}') + 1;
608 }
609 if (strncmp (line, "{y:b}", 5) == 0) {
610 bold = TRUE;
611 line = strchr (line, '}') + 1;
612 }
613 if (sscanf (line, "{s:%u}", &fontsize) == 1) {
614 line = strchr (line, '}') + 1;
615 }
616 /* forward slashes at beginning/end signify italics too */
617 if (g_str_has_prefix (line, "/")) {
618 italic = TRUE;
619 ++line;
620 }
621 if ((line_split = strchr (line, '|')))
622 line_chunk = g_markup_escape_text (line, line_split - line);
623 else
624 line_chunk = g_markup_escape_text (line, strlen (line));
625
626 /* Remove italics markers at end of line/stanza (CHECKME: are end slashes
627 * always at the end of a line or can they span multiple lines?) */
628 if (g_str_has_suffix (line_chunk, "/")) {
629 line_chunk[strlen (line_chunk) - 1] = '\0';
630 }
631
632 markup = g_string_append (markup, "<span");
633 if (italic)
634 g_string_append (markup, " style=\"italic\"");
635 if (bold)
636 g_string_append (markup, " weight=\"bold\"");
637 if (fontsize)
638 g_string_append_printf (markup, " size=\"%u\"", fontsize * 1000);
639 g_string_append_printf (markup, ">%s</span>", line_chunk);
640 g_free (line_chunk);
641 if (line_split) {
642 g_string_append (markup, "\n");
643 line = line_split + 1;
644 } else {
645 break;
646 }
647 }
648 ret = markup->str;
649 g_string_free (markup, FALSE);
650 GST_DEBUG ("parse_mdvdsub returning (%f+%f): %s",
651 state->start_time / (double) GST_SECOND,
652 state->duration / (double) GST_SECOND, ret);
653 return ret;
654 }
655
656 static void
strip_trailing_newlines(gchar * txt)657 strip_trailing_newlines (gchar * txt)
658 {
659 if (txt) {
660 guint len;
661
662 len = strlen (txt);
663 while (len > 1 && txt[len - 1] == '\n') {
664 txt[len - 1] = '\0';
665 --len;
666 }
667 }
668 }
669
670 /* we want to escape text in general, but retain basic markup like
671 * <i></i>, <u></u>, and <b></b>. The easiest and safest way is to
672 * just unescape a white list of allowed markups again after
673 * escaping everything (the text between these simple markers isn't
674 * necessarily escaped, so it seems best to do it like this) */
675 static void
subrip_unescape_formatting(gchar * txt,gconstpointer allowed_tags_ptr,gboolean allows_tag_attributes)676 subrip_unescape_formatting (gchar * txt, gconstpointer allowed_tags_ptr,
677 gboolean allows_tag_attributes)
678 {
679 gchar *res;
680 GRegex *tag_regex;
681 gchar *allowed_tags_pattern, *search_pattern;
682 const gchar *replace_pattern;
683
684 /* No processing needed if no escaped tag marker found in the string. */
685 if (strstr (txt, "<") == NULL)
686 return;
687
688 /* Build a list of alternates for our regexp.
689 * FIXME: Could be built once and stored */
690 allowed_tags_pattern = g_strjoinv ("|", (gchar **) allowed_tags_ptr);
691 /* Look for starting/ending escaped tags with optional attributes. */
692 search_pattern = g_strdup_printf ("<(/)?\\ *(%s)(%s)>",
693 allowed_tags_pattern, ATTRIBUTE_REGEX);
694 /* And unescape appropriately */
695 if (allows_tag_attributes) {
696 replace_pattern = "<\\1\\2\\3>";
697 } else {
698 replace_pattern = "<\\1\\2>";
699 }
700
701 tag_regex = g_regex_new (search_pattern, 0, 0, NULL);
702 res = g_regex_replace (tag_regex, txt, strlen (txt), 0,
703 replace_pattern, 0, NULL);
704
705 /* res will always be shorter than the input or identical, so this
706 * copy is OK */
707 strcpy (txt, res);
708
709 g_free (res);
710 g_free (search_pattern);
711 g_free (allowed_tags_pattern);
712
713 g_regex_unref (tag_regex);
714 }
715
716
717 static gboolean
subrip_remove_unhandled_tag(gchar * start,gchar * stop)718 subrip_remove_unhandled_tag (gchar * start, gchar * stop)
719 {
720 gchar *tag, saved;
721
722 tag = start + strlen ("<");
723 if (*tag == '/')
724 ++tag;
725
726 if (g_ascii_tolower (*tag) < 'a' || g_ascii_tolower (*tag) > 'z')
727 return FALSE;
728
729 saved = *stop;
730 *stop = '\0';
731 GST_LOG ("removing unhandled tag '%s'", start);
732 *stop = saved;
733 memmove (start, stop, strlen (stop) + 1);
734 return TRUE;
735 }
736
737 /* remove tags we haven't explicitly allowed earlier on, like font tags
738 * for example */
739 static void
subrip_remove_unhandled_tags(gchar * txt)740 subrip_remove_unhandled_tags (gchar * txt)
741 {
742 gchar *pos, *gt;
743
744 for (pos = txt; pos != NULL && *pos != '\0'; ++pos) {
745 if (strncmp (pos, "<", 4) == 0 && (gt = strstr (pos + 4, ">"))) {
746 if (subrip_remove_unhandled_tag (pos, gt + strlen (">")))
747 --pos;
748 }
749 }
750 }
751
752 /* we only allow a fixed set of tags like <i>, <u> and <b>, so let's
753 * take a simple approach. This code assumes the input has been
754 * escaped and subrip_unescape_formatting() has then been run over the
755 * input! This function adds missing closing markup tags and removes
756 * broken closing tags for tags that have never been opened. */
757 static void
subrip_fix_up_markup(gchar ** p_txt,gconstpointer allowed_tags_ptr)758 subrip_fix_up_markup (gchar ** p_txt, gconstpointer allowed_tags_ptr)
759 {
760 gchar *cur, *next_tag;
761 GPtrArray *open_tags = NULL;
762 guint num_open_tags = 0;
763 const gchar *iter_tag;
764 guint offset = 0;
765 guint index;
766 gchar *cur_tag;
767 gchar *end_tag;
768 GRegex *tag_regex;
769 GMatchInfo *match_info;
770 gchar **allowed_tags = (gchar **) allowed_tags_ptr;
771
772 g_assert (*p_txt != NULL);
773
774 open_tags = g_ptr_array_new_with_free_func (g_free);
775 cur = *p_txt;
776 while (*cur != '\0') {
777 next_tag = strchr (cur, '<');
778 if (next_tag == NULL)
779 break;
780 offset = 0;
781 index = 0;
782 while (index < g_strv_length (allowed_tags)) {
783 iter_tag = allowed_tags[index];
784 /* Look for a white listed tag */
785 cur_tag = g_strconcat ("<", iter_tag, ATTRIBUTE_REGEX, ">", NULL);
786 tag_regex = g_regex_new (cur_tag, 0, 0, NULL);
787 (void) g_regex_match (tag_regex, next_tag, 0, &match_info);
788
789 if (g_match_info_matches (match_info)) {
790 gint start_pos, end_pos;
791 gchar *word = g_match_info_fetch (match_info, 0);
792 g_match_info_fetch_pos (match_info, 0, &start_pos, &end_pos);
793 if (start_pos == 0) {
794 offset = strlen (word);
795 }
796 g_free (word);
797 }
798 g_match_info_free (match_info);
799 g_regex_unref (tag_regex);
800 g_free (cur_tag);
801 index++;
802 if (offset) {
803 /* OK we found a tag, let's keep track of it */
804 g_ptr_array_add (open_tags, g_ascii_strdown (iter_tag, -1));
805 ++num_open_tags;
806 break;
807 }
808 }
809
810 if (offset) {
811 next_tag += offset;
812 cur = next_tag;
813 continue;
814 }
815
816 if (*next_tag == '<' && *(next_tag + 1) == '/') {
817 end_tag = strchr (cur, '>');
818 if (end_tag) {
819 const gchar *last = NULL;
820 if (num_open_tags > 0)
821 last = g_ptr_array_index (open_tags, num_open_tags - 1);
822 if (num_open_tags == 0
823 || g_ascii_strncasecmp (end_tag - 1, last, strlen (last))) {
824 GST_LOG ("broken input, closing tag '%s' is not open", end_tag - 1);
825 memmove (next_tag, end_tag + 1, strlen (end_tag) + 1);
826 next_tag -= strlen (end_tag);
827 } else {
828 --num_open_tags;
829 g_ptr_array_remove_index (open_tags, num_open_tags);
830 }
831 }
832 }
833 ++next_tag;
834 cur = next_tag;
835 }
836
837 if (num_open_tags > 0) {
838 GString *s;
839
840 s = g_string_new (*p_txt);
841 while (num_open_tags > 0) {
842 GST_LOG ("adding missing closing tag '%s'",
843 (char *) g_ptr_array_index (open_tags, num_open_tags - 1));
844 g_string_append_c (s, '<');
845 g_string_append_c (s, '/');
846 g_string_append (s, g_ptr_array_index (open_tags, num_open_tags - 1));
847 g_string_append_c (s, '>');
848 --num_open_tags;
849 }
850 g_free (*p_txt);
851 *p_txt = g_string_free (s, FALSE);
852 }
853 g_ptr_array_free (open_tags, TRUE);
854 }
855
856 static gboolean
parse_subrip_time(const gchar * ts_string,GstClockTime * t)857 parse_subrip_time (const gchar * ts_string, GstClockTime * t)
858 {
859 gchar s[128] = { '\0', };
860 gchar *end, *p;
861 guint hour, min, sec, msec, len;
862
863 while (*ts_string == ' ')
864 ++ts_string;
865
866 g_strlcpy (s, ts_string, sizeof (s));
867 if ((end = strstr (s, "-->")))
868 *end = '\0';
869 g_strchomp (s);
870
871 /* ms may be in these formats:
872 * hh:mm:ss,500 = 500ms
873 * hh:mm:ss, 5 = 5ms
874 * hh:mm:ss, 5 = 50ms
875 * hh:mm:ss, 50 = 50ms
876 * hh:mm:ss,5 = 500ms
877 * and the same with . instead of ,.
878 * sscanf() doesn't differentiate between ' 5' and '5' so munge
879 * the white spaces within the timestamp to '0' (I'm sure there's a
880 * way to make sscanf() do this for us, but how?)
881 */
882 g_strdelimit (s, " ", '0');
883 g_strdelimit (s, ".", ',');
884
885 /* make sure we have exactly three digits after he comma */
886 p = strchr (s, ',');
887 if (p == NULL) {
888 /* If there isn't a ',' the timestamp is broken */
889 /* https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/issues/532#note_100179 */
890 GST_WARNING ("failed to parse subrip timestamp string '%s'", s);
891 return FALSE;
892 }
893
894 ++p;
895 len = strlen (p);
896 if (len > 3) {
897 p[3] = '\0';
898 } else
899 while (len < 3) {
900 g_strlcat (&p[len], "0", 2);
901 ++len;
902 }
903
904 GST_LOG ("parsing timestamp '%s'", s);
905 if (sscanf (s, "%u:%u:%u,%u", &hour, &min, &sec, &msec) != 4) {
906 GST_WARNING ("failed to parse subrip timestamp string '%s'", s);
907 return FALSE;
908 }
909
910 *t = ((hour * 3600) + (min * 60) + sec) * GST_SECOND + msec * GST_MSECOND;
911 return TRUE;
912 }
913
914 /* cue settings are part of the WebVTT specification. They are
915 * declared after the time interval in the first line of the
916 * cue. Example: 00:00:01,000 --> 00:00:02,000 D:vertical-lr A:start
917 * See also http://www.whatwg.org/specs/web-apps/current-work/webvtt.html
918 */
919 static void
parse_webvtt_cue_settings(ParserState * state,const gchar * settings)920 parse_webvtt_cue_settings (ParserState * state, const gchar * settings)
921 {
922 gchar **splitted_settings = g_strsplit_set (settings, " \t", -1);
923 gint i = 0;
924 gint16 text_position, text_size;
925 gint16 line_position;
926 gboolean vertical_found = FALSE;
927 gboolean alignment_found = FALSE;
928
929 while (i < g_strv_length (splitted_settings)) {
930 gboolean valid_tag = FALSE;
931 switch (splitted_settings[i][0]) {
932 case 'T':
933 if (sscanf (splitted_settings[i], "T:%" G_GINT16_FORMAT "%%",
934 &text_position) > 0) {
935 state->text_position = (guint8) text_position;
936 valid_tag = TRUE;
937 }
938 break;
939 case 'D':
940 if (strlen (splitted_settings[i]) > 2) {
941 vertical_found = TRUE;
942 g_free (state->vertical);
943 state->vertical = g_strdup (splitted_settings[i] + 2);
944 valid_tag = TRUE;
945 }
946 break;
947 case 'L':
948 if (g_str_has_suffix (splitted_settings[i], "%")) {
949 if (sscanf (splitted_settings[i], "L:%" G_GINT16_FORMAT "%%",
950 &line_position) > 0) {
951 state->line_position = line_position;
952 valid_tag = TRUE;
953 }
954 } else {
955 if (sscanf (splitted_settings[i], "L:%" G_GINT16_FORMAT,
956 &line_position) > 0) {
957 state->line_number = line_position;
958 valid_tag = TRUE;
959 }
960 }
961 break;
962 case 'S':
963 if (sscanf (splitted_settings[i], "S:%" G_GINT16_FORMAT "%%",
964 &text_size) > 0) {
965 state->text_size = (guint8) text_size;
966 valid_tag = TRUE;
967 }
968 break;
969 case 'A':
970 if (strlen (splitted_settings[i]) > 2) {
971 g_free (state->alignment);
972 state->alignment = g_strdup (splitted_settings[i] + 2);
973 alignment_found = TRUE;
974 valid_tag = TRUE;
975 }
976 break;
977 default:
978 break;
979 }
980 if (!valid_tag) {
981 GST_LOG ("Invalid or unrecognised setting found: %s",
982 splitted_settings[i]);
983 }
984 i++;
985 }
986 g_strfreev (splitted_settings);
987 if (!vertical_found) {
988 g_free (state->vertical);
989 state->vertical = g_strdup ("");
990 }
991 if (!alignment_found) {
992 g_free (state->alignment);
993 state->alignment = g_strdup ("");
994 }
995 }
996
997 static gchar *
parse_subrip(ParserState * state,const gchar * line)998 parse_subrip (ParserState * state, const gchar * line)
999 {
1000 gchar *ret;
1001
1002 switch (state->state) {
1003 case 0:{
1004 char *endptr;
1005 guint64 id;
1006
1007 /* looking for a single integer as a Cue ID, but we
1008 * don't actually use it */
1009 errno = 0;
1010 id = g_ascii_strtoull (line, &endptr, 10);
1011 if (id == G_MAXUINT64 && errno == ERANGE)
1012 state->state = 1;
1013 else if (id == 0 && errno == EINVAL)
1014 state->state = 1;
1015 else if (endptr != line && *endptr == '\0')
1016 state->state = 1;
1017 return NULL;
1018 }
1019 case 1:
1020 {
1021 GstClockTime ts_start, ts_end;
1022 gchar *end_time;
1023
1024 /* looking for start_time --> end_time */
1025 if ((end_time = strstr (line, " --> ")) &&
1026 parse_subrip_time (line, &ts_start) &&
1027 parse_subrip_time (end_time + strlen (" --> "), &ts_end) &&
1028 state->start_time <= ts_end) {
1029 state->state = 2;
1030 state->start_time = ts_start;
1031 state->duration = ts_end - ts_start;
1032 } else {
1033 GST_DEBUG ("error parsing subrip time line '%s'", line);
1034 state->state = 0;
1035 }
1036 return NULL;
1037 }
1038 case 2:
1039 {
1040 /* No need to parse that text if it's out of segment */
1041 guint64 clip_start = 0, clip_stop = 0;
1042 gboolean in_seg = FALSE;
1043
1044 /* Check our segment start/stop */
1045 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1046 state->start_time, state->start_time + state->duration,
1047 &clip_start, &clip_stop);
1048
1049 if (in_seg) {
1050 state->start_time = clip_start;
1051 state->duration = clip_stop - clip_start;
1052 } else {
1053 state->state = 0;
1054 return NULL;
1055 }
1056 }
1057 /* looking for subtitle text; empty line ends this subtitle entry */
1058 if (state->buf->len)
1059 g_string_append_c (state->buf, '\n');
1060 g_string_append (state->buf, line);
1061 if (strlen (line) == 0) {
1062 ret = g_markup_escape_text (state->buf->str, state->buf->len);
1063 g_string_truncate (state->buf, 0);
1064 state->state = 0;
1065 subrip_unescape_formatting (ret, state->allowed_tags,
1066 state->allows_tag_attributes);
1067 subrip_remove_unhandled_tags (ret);
1068 strip_trailing_newlines (ret);
1069 subrip_fix_up_markup (&ret, state->allowed_tags);
1070 return ret;
1071 }
1072 return NULL;
1073 default:
1074 g_return_val_if_reached (NULL);
1075 }
1076 }
1077
1078 static gchar *
parse_lrc(ParserState * state,const gchar * line)1079 parse_lrc (ParserState * state, const gchar * line)
1080 {
1081 gint m, s, c;
1082 const gchar *start;
1083 gint milli;
1084
1085 if (line[0] != '[')
1086 return NULL;
1087
1088 if (sscanf (line, "[%u:%02u.%03u]", &m, &s, &c) != 3 &&
1089 sscanf (line, "[%u:%02u.%02u]", &m, &s, &c) != 3)
1090 return NULL;
1091
1092 start = strchr (line, ']');
1093 if (start - line == 9)
1094 milli = 10;
1095 else
1096 milli = 1;
1097
1098 state->start_time = gst_util_uint64_scale (m, 60 * GST_SECOND, 1)
1099 + gst_util_uint64_scale (s, GST_SECOND, 1)
1100 + gst_util_uint64_scale (c, milli * GST_MSECOND, 1);
1101 state->duration = GST_CLOCK_TIME_NONE;
1102
1103 return g_strdup (start + 1);
1104 }
1105
1106 /* WebVTT is a new subtitle format for the upcoming HTML5 video track
1107 * element. This format is similar to Subrip, the biggest differences
1108 * are that there can be cue settings detailing how to display the cue
1109 * text and more markup tags are allowed.
1110 * See also http://www.whatwg.org/specs/web-apps/current-work/webvtt.html
1111 */
1112 static gchar *
parse_webvtt(ParserState * state,const gchar * line)1113 parse_webvtt (ParserState * state, const gchar * line)
1114 {
1115 /* Cue IDs are optional in WebVTT, but not in subrip,
1116 * so when in state 0 (cue ID), also check if we're
1117 * already at the start --> end time marker */
1118 if (state->state == 0 || state->state == 1) {
1119 GstClockTime ts_start, ts_end;
1120 gchar *end_time;
1121 gchar *cue_settings = NULL;
1122
1123 /* looking for start_time --> end_time */
1124 if ((end_time = strstr (line, " --> ")) &&
1125 parse_subrip_time (line, &ts_start) &&
1126 parse_subrip_time (end_time + strlen (" --> "), &ts_end) &&
1127 state->start_time <= ts_end) {
1128 state->state = 2;
1129 state->start_time = ts_start;
1130 state->duration = ts_end - ts_start;
1131 cue_settings = strstr (end_time + strlen (" --> "), " ");
1132 } else {
1133 GST_DEBUG ("error parsing subrip time line '%s'", line);
1134 state->state = 0;
1135 }
1136
1137 state->text_position = 0;
1138 state->text_size = 0;
1139 state->line_position = 0;
1140 state->line_number = 0;
1141
1142 if (cue_settings)
1143 parse_webvtt_cue_settings (state, cue_settings + 1);
1144 else {
1145 g_free (state->vertical);
1146 state->vertical = g_strdup ("");
1147 g_free (state->alignment);
1148 state->alignment = g_strdup ("");
1149 }
1150
1151 return NULL;
1152 } else
1153 return parse_subrip (state, line);
1154 }
1155
1156 static void
unescape_newlines_br(gchar * read)1157 unescape_newlines_br (gchar * read)
1158 {
1159 gchar *write = read;
1160
1161 /* Replace all occurences of '[br]' with a newline as version 2
1162 * of the subviewer format uses this for newlines */
1163
1164 if (read[0] == '\0' || read[1] == '\0' || read[2] == '\0' || read[3] == '\0')
1165 return;
1166
1167 do {
1168 if (strncmp (read, "[br]", 4) == 0) {
1169 *write = '\n';
1170 read += 4;
1171 } else {
1172 *write = *read;
1173 read++;
1174 }
1175 write++;
1176 } while (*read);
1177
1178 *write = '\0';
1179 }
1180
1181 static gchar *
parse_subviewer(ParserState * state,const gchar * line)1182 parse_subviewer (ParserState * state, const gchar * line)
1183 {
1184 guint h1, m1, s1, ms1;
1185 guint h2, m2, s2, ms2;
1186 gchar *ret;
1187
1188 /* TODO: Maybe also parse the fields in the header, especially DELAY.
1189 * For examples see the unit test or
1190 * http://www.doom9.org/index.html?/sub.htm */
1191
1192 switch (state->state) {
1193 case 0:
1194 /* looking for start_time,end_time */
1195 if (sscanf (line, "%u:%u:%u.%u,%u:%u:%u.%u",
1196 &h1, &m1, &s1, &ms1, &h2, &m2, &s2, &ms2) == 8) {
1197 state->state = 1;
1198 state->start_time =
1199 (((guint64) h1) * 3600 + m1 * 60 + s1) * GST_SECOND +
1200 ms1 * GST_MSECOND;
1201 state->duration =
1202 (((guint64) h2) * 3600 + m2 * 60 + s2) * GST_SECOND +
1203 ms2 * GST_MSECOND - state->start_time;
1204 }
1205 return NULL;
1206 case 1:
1207 {
1208 /* No need to parse that text if it's out of segment */
1209 guint64 clip_start = 0, clip_stop = 0;
1210 gboolean in_seg = FALSE;
1211
1212 /* Check our segment start/stop */
1213 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1214 state->start_time, state->start_time + state->duration,
1215 &clip_start, &clip_stop);
1216
1217 if (in_seg) {
1218 state->start_time = clip_start;
1219 state->duration = clip_stop - clip_start;
1220 } else {
1221 state->state = 0;
1222 return NULL;
1223 }
1224 }
1225 /* looking for subtitle text; empty line ends this subtitle entry */
1226 if (state->buf->len)
1227 g_string_append_c (state->buf, '\n');
1228 g_string_append (state->buf, line);
1229 if (strlen (line) == 0) {
1230 ret = g_strdup (state->buf->str);
1231 unescape_newlines_br (ret);
1232 strip_trailing_newlines (ret);
1233 g_string_truncate (state->buf, 0);
1234 state->state = 0;
1235 return ret;
1236 }
1237 return NULL;
1238 default:
1239 g_assert_not_reached ();
1240 return NULL;
1241 }
1242 }
1243
1244 static gchar *
parse_mpsub(ParserState * state,const gchar * line)1245 parse_mpsub (ParserState * state, const gchar * line)
1246 {
1247 gchar *ret;
1248 float t1, t2;
1249
1250 switch (state->state) {
1251 case 0:
1252 /* looking for two floats (offset, duration) */
1253 if (sscanf (line, "%f %f", &t1, &t2) == 2) {
1254 state->state = 1;
1255 state->start_time += state->duration + GST_SECOND * t1;
1256 state->duration = GST_SECOND * t2;
1257 }
1258 return NULL;
1259 case 1:
1260 { /* No need to parse that text if it's out of segment */
1261 guint64 clip_start = 0, clip_stop = 0;
1262 gboolean in_seg = FALSE;
1263
1264 /* Check our segment start/stop */
1265 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1266 state->start_time, state->start_time + state->duration,
1267 &clip_start, &clip_stop);
1268
1269 if (in_seg) {
1270 state->start_time = clip_start;
1271 state->duration = clip_stop - clip_start;
1272 } else {
1273 state->state = 0;
1274 return NULL;
1275 }
1276 }
1277 /* looking for subtitle text; empty line ends this
1278 * subtitle entry */
1279 if (state->buf->len)
1280 g_string_append_c (state->buf, '\n');
1281 g_string_append (state->buf, line);
1282 if (strlen (line) == 0) {
1283 ret = g_strdup (state->buf->str);
1284 g_string_truncate (state->buf, 0);
1285 state->state = 0;
1286 return ret;
1287 }
1288 return NULL;
1289 default:
1290 g_assert_not_reached ();
1291 return NULL;
1292 }
1293 }
1294
1295 static const gchar *
dks_skip_timestamp(const gchar * line)1296 dks_skip_timestamp (const gchar * line)
1297 {
1298 while (*line && *line != ']')
1299 line++;
1300 if (*line == ']')
1301 line++;
1302 return line;
1303 }
1304
1305 static gchar *
parse_dks(ParserState * state,const gchar * line)1306 parse_dks (ParserState * state, const gchar * line)
1307 {
1308 guint h, m, s;
1309
1310 switch (state->state) {
1311 case 0:
1312 /* Looking for the start time and text */
1313 if (sscanf (line, "[%u:%u:%u]", &h, &m, &s) == 3) {
1314 const gchar *text;
1315 state->start_time = (((guint64) h) * 3600 + m * 60 + s) * GST_SECOND;
1316 text = dks_skip_timestamp (line);
1317 if (*text) {
1318 state->state = 1;
1319 g_string_append (state->buf, text);
1320 }
1321 }
1322 return NULL;
1323 case 1:
1324 {
1325 guint64 clip_start = 0, clip_stop = 0;
1326 gboolean in_seg;
1327 gchar *ret;
1328
1329 /* Looking for the end time */
1330 if (sscanf (line, "[%u:%u:%u]", &h, &m, &s) == 3) {
1331 state->state = 0;
1332 state->duration = (((guint64) h) * 3600 + m * 60 + s) * GST_SECOND -
1333 state->start_time;
1334 } else {
1335 GST_WARNING ("Failed to parse subtitle end time");
1336 return NULL;
1337 }
1338
1339 /* Check if this subtitle is out of the current segment */
1340 in_seg = gst_segment_clip (state->segment, GST_FORMAT_TIME,
1341 state->start_time, state->start_time + state->duration,
1342 &clip_start, &clip_stop);
1343
1344 if (!in_seg) {
1345 return NULL;
1346 }
1347
1348 state->start_time = clip_start;
1349 state->duration = clip_stop - clip_start;
1350
1351 ret = g_strdup (state->buf->str);
1352 g_string_truncate (state->buf, 0);
1353 unescape_newlines_br (ret);
1354 return ret;
1355 }
1356 default:
1357 g_assert_not_reached ();
1358 return NULL;
1359 }
1360 }
1361
1362 static void
parser_state_init(ParserState * state)1363 parser_state_init (ParserState * state)
1364 {
1365 GST_DEBUG ("initialising parser");
1366
1367 if (state->buf) {
1368 g_string_truncate (state->buf, 0);
1369 } else {
1370 state->buf = g_string_new (NULL);
1371 }
1372
1373 state->start_time = 0;
1374 state->duration = 0;
1375 state->max_duration = 0; /* no limit */
1376 state->state = 0;
1377 state->segment = NULL;
1378 }
1379
1380 static void
parser_state_dispose(GstSubParse * self,ParserState * state)1381 parser_state_dispose (GstSubParse * self, ParserState * state)
1382 {
1383 if (state->buf) {
1384 g_string_free (state->buf, TRUE);
1385 state->buf = NULL;
1386 }
1387
1388 g_free (state->vertical);
1389 state->vertical = NULL;
1390 g_free (state->alignment);
1391 state->alignment = NULL;
1392
1393 if (state->user_data) {
1394 switch (self->parser_type) {
1395 case GST_SUB_PARSE_FORMAT_QTTEXT:
1396 qttext_context_deinit (state);
1397 break;
1398 case GST_SUB_PARSE_FORMAT_SAMI:
1399 sami_context_deinit (state);
1400 break;
1401 default:
1402 break;
1403 }
1404 }
1405 state->allowed_tags = NULL;
1406 }
1407
1408 /* regex type enum */
1409 typedef enum
1410 {
1411 GST_SUB_PARSE_REGEX_UNKNOWN = 0,
1412 GST_SUB_PARSE_REGEX_MDVDSUB = 1,
1413 GST_SUB_PARSE_REGEX_SUBRIP = 2,
1414 GST_SUB_PARSE_REGEX_DKS = 3,
1415 GST_SUB_PARSE_REGEX_VTT = 4,
1416 } GstSubParseRegex;
1417
1418 static gpointer
gst_sub_parse_data_format_autodetect_regex_once(GstSubParseRegex regtype)1419 gst_sub_parse_data_format_autodetect_regex_once (GstSubParseRegex regtype)
1420 {
1421 gpointer result = NULL;
1422 GError *gerr = NULL;
1423 switch (regtype) {
1424 case GST_SUB_PARSE_REGEX_MDVDSUB:
1425 result =
1426 (gpointer) g_regex_new ("^\\{[0-9]+\\}\\{[0-9]+\\}",
1427 G_REGEX_RAW | G_REGEX_OPTIMIZE, 0, &gerr);
1428 if (result == NULL) {
1429 g_warning ("Compilation of mdvd regex failed: %s", gerr->message);
1430 g_clear_error (&gerr);
1431 }
1432 break;
1433 case GST_SUB_PARSE_REGEX_SUBRIP:
1434 result = (gpointer)
1435 g_regex_new ("^[\\s\\n]*[\\n]? {0,3}[ 0-9]{1,4}\\s*(\x0d)?\x0a"
1436 " ?[0-9]{1,2}: ?[0-9]{1,2}: ?[0-9]{1,2}[,.] {0,2}[0-9]{1,3}"
1437 " +--> +[0-9]{1,2}: ?[0-9]{1,2}: ?[0-9]{1,2}[,.] {0,2}[0-9]{1,2}",
1438 G_REGEX_RAW | G_REGEX_OPTIMIZE, 0, &gerr);
1439 if (result == NULL) {
1440 g_warning ("Compilation of subrip regex failed: %s", gerr->message);
1441 g_clear_error (&gerr);
1442 }
1443 break;
1444 case GST_SUB_PARSE_REGEX_DKS:
1445 result = (gpointer) g_regex_new ("^\\[[0-9]+:[0-9]+:[0-9]+\\].*",
1446 G_REGEX_RAW | G_REGEX_OPTIMIZE, 0, &gerr);
1447 if (result == NULL) {
1448 g_warning ("Compilation of dks regex failed: %s", gerr->message);
1449 g_clear_error (&gerr);
1450 }
1451 break;
1452 case GST_SUB_PARSE_REGEX_VTT:
1453 result = (gpointer)
1454 g_regex_new ("^(\\xef\\xbb\\xbf)?WEBVTT[\\xa\\xd\\x20\\x9]", 0, 0,
1455 &gerr);
1456 if (result == NULL) {
1457 g_warning ("Compilation of vtt regex failed: %s", gerr->message);
1458 g_error_free (gerr);
1459 }
1460 break;
1461
1462 default:
1463 GST_WARNING ("Trying to allocate regex of unknown type %u", regtype);
1464 }
1465 return result;
1466 }
1467
1468 /*
1469 * FIXME: maybe we should pass along a second argument, the preceding
1470 * text buffer, because that is how this originally worked, even though
1471 * I don't really see the use of that.
1472 */
1473
1474 static GstSubParseFormat
gst_sub_parse_data_format_autodetect(gchar * match_str)1475 gst_sub_parse_data_format_autodetect (gchar * match_str)
1476 {
1477 guint n1, n2, n3;
1478
1479 static GOnce mdvd_rx_once = G_ONCE_INIT;
1480 static GOnce subrip_rx_once = G_ONCE_INIT;
1481 static GOnce dks_rx_once = G_ONCE_INIT;
1482 static GOnce vtt_rx_once = G_ONCE_INIT;
1483
1484 GRegex *mdvd_grx;
1485 GRegex *subrip_grx;
1486 GRegex *dks_grx;
1487 GRegex *vtt_grx;
1488
1489 g_once (&mdvd_rx_once,
1490 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1491 (gpointer) GST_SUB_PARSE_REGEX_MDVDSUB);
1492 g_once (&subrip_rx_once,
1493 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1494 (gpointer) GST_SUB_PARSE_REGEX_SUBRIP);
1495 g_once (&dks_rx_once,
1496 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1497 (gpointer) GST_SUB_PARSE_REGEX_DKS);
1498 g_once (&vtt_rx_once,
1499 (GThreadFunc) gst_sub_parse_data_format_autodetect_regex_once,
1500 (gpointer) GST_SUB_PARSE_REGEX_VTT);
1501
1502 mdvd_grx = (GRegex *) mdvd_rx_once.retval;
1503 subrip_grx = (GRegex *) subrip_rx_once.retval;
1504 dks_grx = (GRegex *) dks_rx_once.retval;
1505 vtt_grx = (GRegex *) vtt_rx_once.retval;
1506
1507 if (g_regex_match (mdvd_grx, match_str, 0, NULL)) {
1508 GST_LOG ("MicroDVD (frame based) format detected");
1509 return GST_SUB_PARSE_FORMAT_MDVDSUB;
1510 }
1511 if (g_regex_match (subrip_grx, match_str, 0, NULL)) {
1512 GST_LOG ("SubRip (time based) format detected");
1513 return GST_SUB_PARSE_FORMAT_SUBRIP;
1514 }
1515 if (g_regex_match (dks_grx, match_str, 0, NULL)) {
1516 GST_LOG ("DKS (time based) format detected");
1517 return GST_SUB_PARSE_FORMAT_DKS;
1518 }
1519 if (g_regex_match (vtt_grx, match_str, 0, NULL) == TRUE) {
1520 GST_LOG ("WebVTT (time based) format detected");
1521 return GST_SUB_PARSE_FORMAT_VTT;
1522 }
1523
1524 if (!strncmp (match_str, "FORMAT=TIME", 11)) {
1525 GST_LOG ("MPSub (time based) format detected");
1526 return GST_SUB_PARSE_FORMAT_MPSUB;
1527 }
1528 if (strstr (match_str, "<SAMI>") != NULL ||
1529 strstr (match_str, "<sami>") != NULL) {
1530 GST_LOG ("SAMI (time based) format detected");
1531 return GST_SUB_PARSE_FORMAT_SAMI;
1532 }
1533 /* we're boldly assuming the first subtitle appears within the first hour */
1534 if (sscanf (match_str, "0:%02u:%02u:", &n1, &n2) == 2 ||
1535 sscanf (match_str, "0:%02u:%02u=", &n1, &n2) == 2 ||
1536 sscanf (match_str, "00:%02u:%02u:", &n1, &n2) == 2 ||
1537 sscanf (match_str, "00:%02u:%02u=", &n1, &n2) == 2 ||
1538 sscanf (match_str, "00:%02u:%02u,%u=", &n1, &n2, &n3) == 3) {
1539 GST_LOG ("TMPlayer (time based) format detected");
1540 return GST_SUB_PARSE_FORMAT_TMPLAYER;
1541 }
1542 if (sscanf (match_str, "[%u][%u]", &n1, &n2) == 2) {
1543 GST_LOG ("MPL2 (time based) format detected");
1544 return GST_SUB_PARSE_FORMAT_MPL2;
1545 }
1546 if (strstr (match_str, "[INFORMATION]") != NULL) {
1547 GST_LOG ("SubViewer (time based) format detected");
1548 return GST_SUB_PARSE_FORMAT_SUBVIEWER;
1549 }
1550 if (strstr (match_str, "{QTtext}") != NULL) {
1551 GST_LOG ("QTtext (time based) format detected");
1552 return GST_SUB_PARSE_FORMAT_QTTEXT;
1553 }
1554 /* We assume the LRC file starts immediately */
1555 if (match_str[0] == '[') {
1556 gboolean all_lines_good = TRUE;
1557 gchar **split;
1558 gchar **ptr;
1559
1560 ptr = split = g_strsplit (match_str, "\n", -1);
1561 while (*ptr && *(ptr + 1)) {
1562 gchar *str = *ptr;
1563 gint len = strlen (str);
1564
1565 if (sscanf (str, "[%u:%02u.%02u]", &n1, &n2, &n3) == 3 ||
1566 sscanf (str, "[%u:%02u.%03u]", &n1, &n2, &n3) == 3) {
1567 all_lines_good = TRUE;
1568 } else if (str[len - 1] == ']' && strchr (str, ':') != NULL) {
1569 all_lines_good = TRUE;
1570 } else {
1571 all_lines_good = FALSE;
1572 break;
1573 }
1574
1575 ptr++;
1576 }
1577 g_strfreev (split);
1578
1579 if (all_lines_good)
1580 return GST_SUB_PARSE_FORMAT_LRC;
1581 }
1582
1583 GST_DEBUG ("no subtitle format detected");
1584 return GST_SUB_PARSE_FORMAT_UNKNOWN;
1585 }
1586
1587 static GstCaps *
gst_sub_parse_format_autodetect(GstSubParse * self)1588 gst_sub_parse_format_autodetect (GstSubParse * self)
1589 {
1590 gchar *data;
1591 GstSubParseFormat format;
1592
1593 if (strlen (self->textbuf->str) < 30) {
1594 GST_DEBUG ("File too small to be a subtitles file");
1595 return NULL;
1596 }
1597
1598 data = g_strndup (self->textbuf->str, 35);
1599 format = gst_sub_parse_data_format_autodetect (data);
1600 g_free (data);
1601
1602 self->parser_type = format;
1603 self->subtitle_codec = gst_sub_parse_get_format_description (format);
1604 parser_state_init (&self->state);
1605 self->state.allowed_tags = NULL;
1606
1607 switch (format) {
1608 case GST_SUB_PARSE_FORMAT_MDVDSUB:
1609 self->parse_line = parse_mdvdsub;
1610 return gst_caps_new_simple ("text/x-raw",
1611 "format", G_TYPE_STRING, "pango-markup", NULL);
1612 case GST_SUB_PARSE_FORMAT_SUBRIP:
1613 self->state.allowed_tags = (gpointer) allowed_srt_tags;
1614 self->state.allows_tag_attributes = FALSE;
1615 self->parse_line = parse_subrip;
1616 return gst_caps_new_simple ("text/x-raw",
1617 "format", G_TYPE_STRING, "pango-markup", NULL);
1618 case GST_SUB_PARSE_FORMAT_MPSUB:
1619 self->parse_line = parse_mpsub;
1620 return gst_caps_new_simple ("text/x-raw",
1621 "format", G_TYPE_STRING, "utf8", NULL);
1622 case GST_SUB_PARSE_FORMAT_SAMI:
1623 self->parse_line = parse_sami;
1624 sami_context_init (&self->state);
1625 return gst_caps_new_simple ("text/x-raw",
1626 "format", G_TYPE_STRING, "pango-markup", NULL);
1627 case GST_SUB_PARSE_FORMAT_TMPLAYER:
1628 self->parse_line = parse_tmplayer;
1629 self->state.max_duration = 5 * GST_SECOND;
1630 return gst_caps_new_simple ("text/x-raw",
1631 "format", G_TYPE_STRING, "utf8", NULL);
1632 case GST_SUB_PARSE_FORMAT_MPL2:
1633 self->parse_line = parse_mpl2;
1634 return gst_caps_new_simple ("text/x-raw",
1635 "format", G_TYPE_STRING, "pango-markup", NULL);
1636 case GST_SUB_PARSE_FORMAT_DKS:
1637 self->parse_line = parse_dks;
1638 return gst_caps_new_simple ("text/x-raw",
1639 "format", G_TYPE_STRING, "utf8", NULL);
1640 case GST_SUB_PARSE_FORMAT_VTT:
1641 self->state.allowed_tags = (gpointer) allowed_vtt_tags;
1642 self->state.allows_tag_attributes = TRUE;
1643 self->parse_line = parse_webvtt;
1644 return gst_caps_new_simple ("text/x-raw",
1645 "format", G_TYPE_STRING, "pango-markup", NULL);
1646 case GST_SUB_PARSE_FORMAT_SUBVIEWER:
1647 self->parse_line = parse_subviewer;
1648 return gst_caps_new_simple ("text/x-raw",
1649 "format", G_TYPE_STRING, "utf8", NULL);
1650 case GST_SUB_PARSE_FORMAT_QTTEXT:
1651 self->parse_line = parse_qttext;
1652 qttext_context_init (&self->state);
1653 return gst_caps_new_simple ("text/x-raw",
1654 "format", G_TYPE_STRING, "pango-markup", NULL);
1655 case GST_SUB_PARSE_FORMAT_LRC:
1656 self->parse_line = parse_lrc;
1657 return gst_caps_new_simple ("text/x-raw",
1658 "format", G_TYPE_STRING, "utf8", NULL);
1659 case GST_SUB_PARSE_FORMAT_UNKNOWN:
1660 default:
1661 GST_DEBUG ("no subtitle format detected");
1662 GST_ELEMENT_ERROR (self, STREAM, WRONG_TYPE,
1663 ("The input is not a valid/supported subtitle file"), (NULL));
1664 return NULL;
1665 }
1666 }
1667
1668 static void
feed_textbuf(GstSubParse * self,GstBuffer * buf)1669 feed_textbuf (GstSubParse * self, GstBuffer * buf)
1670 {
1671 gboolean discont;
1672 gsize consumed;
1673 gchar *input = NULL;
1674 const guint8 *data;
1675 gsize avail;
1676
1677 discont = GST_BUFFER_IS_DISCONT (buf);
1678
1679 if (GST_BUFFER_OFFSET_IS_VALID (buf) &&
1680 GST_BUFFER_OFFSET (buf) != self->offset) {
1681 self->offset = GST_BUFFER_OFFSET (buf);
1682 discont = TRUE;
1683 }
1684
1685 if (discont) {
1686 GST_INFO ("discontinuity");
1687 /* flush the parser state */
1688 parser_state_init (&self->state);
1689 g_string_truncate (self->textbuf, 0);
1690 gst_adapter_clear (self->adapter);
1691 if (self->parser_type == GST_SUB_PARSE_FORMAT_SAMI)
1692 sami_context_reset (&self->state);
1693 /* we could set a flag to make sure that the next buffer we push out also
1694 * has the DISCONT flag set, but there's no point really given that it's
1695 * subtitles which are discontinuous by nature. */
1696 }
1697
1698 self->offset += gst_buffer_get_size (buf);
1699
1700 gst_adapter_push (self->adapter, buf);
1701
1702 avail = gst_adapter_available (self->adapter);
1703 data = gst_adapter_map (self->adapter, avail);
1704 input = convert_encoding (self, (const gchar *) data, avail, &consumed);
1705
1706 if (input && consumed > 0) {
1707 self->textbuf = g_string_append (self->textbuf, input);
1708 gst_adapter_unmap (self->adapter);
1709 gst_adapter_flush (self->adapter, consumed);
1710 } else {
1711 gst_adapter_unmap (self->adapter);
1712 }
1713
1714 g_free (input);
1715 }
1716
1717 static GstFlowReturn
handle_buffer(GstSubParse * self,GstBuffer * buf)1718 handle_buffer (GstSubParse * self, GstBuffer * buf)
1719 {
1720 GstFlowReturn ret = GST_FLOW_OK;
1721 GstCaps *caps = NULL;
1722 gchar *line, *subtitle;
1723 gboolean need_tags = FALSE;
1724
1725 if (self->first_buffer) {
1726 GstMapInfo map;
1727
1728 gst_buffer_map (buf, &map, GST_MAP_READ);
1729 self->detected_encoding = detect_encoding ((gchar *) map.data, map.size);
1730 gst_buffer_unmap (buf, &map);
1731 self->first_buffer = FALSE;
1732 self->state.fps_n = self->fps_n;
1733 self->state.fps_d = self->fps_d;
1734 }
1735
1736 feed_textbuf (self, buf);
1737
1738 /* make sure we know the format */
1739 if (G_UNLIKELY (self->parser_type == GST_SUB_PARSE_FORMAT_UNKNOWN)) {
1740 if (!(caps = gst_sub_parse_format_autodetect (self))) {
1741 return GST_FLOW_EOS;
1742 }
1743 if (!gst_pad_set_caps (self->srcpad, caps)) {
1744 gst_caps_unref (caps);
1745 return GST_FLOW_EOS;
1746 }
1747 gst_caps_unref (caps);
1748 need_tags = TRUE;
1749 }
1750
1751 /* Push newsegment if needed */
1752 if (self->need_segment) {
1753 GST_LOG_OBJECT (self, "pushing newsegment event with %" GST_SEGMENT_FORMAT,
1754 &self->segment);
1755
1756 gst_pad_push_event (self->srcpad, gst_event_new_segment (&self->segment));
1757 self->need_segment = FALSE;
1758 }
1759
1760 if (need_tags) {
1761 /* push tags */
1762 if (self->subtitle_codec != NULL) {
1763 GstTagList *tags;
1764
1765 tags = gst_tag_list_new (GST_TAG_SUBTITLE_CODEC, self->subtitle_codec,
1766 NULL);
1767 gst_pad_push_event (self->srcpad, gst_event_new_tag (tags));
1768 }
1769 }
1770
1771 while (!self->flushing && (line = get_next_line (self))) {
1772 guint offset = 0;
1773
1774 /* Set segment on our parser state machine */
1775 self->state.segment = &self->segment;
1776 /* Now parse the line, out of segment lines will just return NULL */
1777 GST_LOG_OBJECT (self, "State %d. Parsing line '%s'", self->state.state,
1778 line + offset);
1779 subtitle = self->parse_line (&self->state, line + offset);
1780 g_free (line);
1781
1782 if (subtitle) {
1783 guint subtitle_len = strlen (subtitle);
1784
1785 /* +1 for terminating NUL character */
1786 buf = gst_buffer_new_and_alloc (subtitle_len + 1);
1787
1788 /* copy terminating NUL character as well */
1789 gst_buffer_fill (buf, 0, subtitle, subtitle_len + 1);
1790 gst_buffer_set_size (buf, subtitle_len);
1791
1792 GST_BUFFER_TIMESTAMP (buf) = self->state.start_time;
1793 GST_BUFFER_DURATION (buf) = self->state.duration;
1794
1795 /* in some cases (e.g. tmplayer) we can only determine the duration
1796 * of a text chunk from the timestamp of the next text chunk; in those
1797 * cases, we probably want to limit the duration to something
1798 * reasonable, so we don't end up showing some text for e.g. 40 seconds
1799 * just because nothing else is being said during that time */
1800 if (self->state.max_duration > 0 && GST_BUFFER_DURATION_IS_VALID (buf)) {
1801 if (GST_BUFFER_DURATION (buf) > self->state.max_duration)
1802 GST_BUFFER_DURATION (buf) = self->state.max_duration;
1803 }
1804
1805 self->segment.position = self->state.start_time;
1806
1807 GST_DEBUG_OBJECT (self, "Sending text '%s', %" GST_TIME_FORMAT " + %"
1808 GST_TIME_FORMAT, subtitle, GST_TIME_ARGS (self->state.start_time),
1809 GST_TIME_ARGS (self->state.duration));
1810
1811 g_free (self->state.vertical);
1812 self->state.vertical = NULL;
1813 g_free (self->state.alignment);
1814 self->state.alignment = NULL;
1815
1816 ret = gst_pad_push (self->srcpad, buf);
1817
1818 /* move this forward (the tmplayer parser needs this) */
1819 if (self->state.duration != GST_CLOCK_TIME_NONE)
1820 self->state.start_time += self->state.duration;
1821
1822 g_free (subtitle);
1823 subtitle = NULL;
1824
1825 if (ret != GST_FLOW_OK) {
1826 GST_DEBUG_OBJECT (self, "flow: %s", gst_flow_get_name (ret));
1827 break;
1828 }
1829 }
1830 }
1831
1832 return ret;
1833 }
1834
1835 static GstFlowReturn
gst_sub_parse_chain(GstPad * sinkpad,GstObject * parent,GstBuffer * buf)1836 gst_sub_parse_chain (GstPad * sinkpad, GstObject * parent, GstBuffer * buf)
1837 {
1838 GstFlowReturn ret;
1839 GstSubParse *self;
1840
1841 self = GST_SUBPARSE (parent);
1842
1843 ret = handle_buffer (self, buf);
1844
1845 return ret;
1846 }
1847
1848 static gboolean
gst_sub_parse_sink_event(GstPad * pad,GstObject * parent,GstEvent * event)1849 gst_sub_parse_sink_event (GstPad * pad, GstObject * parent, GstEvent * event)
1850 {
1851 GstSubParse *self = GST_SUBPARSE (parent);
1852 gboolean ret = FALSE;
1853
1854 GST_LOG_OBJECT (self, "%s event", GST_EVENT_TYPE_NAME (event));
1855
1856 switch (GST_EVENT_TYPE (event)) {
1857 case GST_EVENT_STREAM_GROUP_DONE:
1858 case GST_EVENT_EOS:{
1859 /* Make sure the last subrip chunk is pushed out even
1860 * if the file does not have an empty line at the end */
1861 if (self->parser_type == GST_SUB_PARSE_FORMAT_SUBRIP ||
1862 self->parser_type == GST_SUB_PARSE_FORMAT_TMPLAYER ||
1863 self->parser_type == GST_SUB_PARSE_FORMAT_MPL2 ||
1864 self->parser_type == GST_SUB_PARSE_FORMAT_QTTEXT ||
1865 self->parser_type == GST_SUB_PARSE_FORMAT_VTT) {
1866 gchar term_chars[] = { '\n', '\n', '\0' };
1867 GstBuffer *buf = gst_buffer_new_and_alloc (2 + 1);
1868
1869 GST_DEBUG_OBJECT (self, "%s: force pushing of any remaining text",
1870 GST_EVENT_TYPE_NAME (event));
1871
1872 gst_buffer_fill (buf, 0, term_chars, 3);
1873 gst_buffer_set_size (buf, 2);
1874
1875 GST_BUFFER_OFFSET (buf) = self->offset;
1876 gst_sub_parse_chain (pad, parent, buf);
1877 }
1878 ret = gst_pad_event_default (pad, parent, event);
1879 break;
1880 }
1881 case GST_EVENT_SEGMENT:
1882 {
1883 const GstSegment *s;
1884 gst_event_parse_segment (event, &s);
1885 if (s->format == GST_FORMAT_TIME)
1886 gst_event_copy_segment (event, &self->segment);
1887 GST_DEBUG_OBJECT (self, "newsegment (%s)",
1888 gst_format_get_name (self->segment.format));
1889
1890 /* if not time format, we'll either start with a 0 timestamp anyway or
1891 * it's following a seek in which case we'll have saved the requested
1892 * seek segment and don't want to overwrite it (remember that on a seek
1893 * we always just seek back to the start in BYTES format and just throw
1894 * away all text that's before the requested position; if the subtitles
1895 * come from an upstream demuxer, it won't be able to handle our BYTES
1896 * seek request and instead send us a newsegment from the seek request
1897 * it received via its video pads instead, so all is fine then too) */
1898 ret = TRUE;
1899 gst_event_unref (event);
1900 /* in either case, let's not simply discard this event;
1901 * trigger sending of the saved requested seek segment
1902 * or the one taken here from upstream */
1903 self->need_segment = TRUE;
1904 break;
1905 }
1906 case GST_EVENT_FLUSH_START:
1907 {
1908 self->flushing = TRUE;
1909
1910 ret = gst_pad_event_default (pad, parent, event);
1911 break;
1912 }
1913 case GST_EVENT_FLUSH_STOP:
1914 {
1915 self->flushing = FALSE;
1916
1917 ret = gst_pad_event_default (pad, parent, event);
1918 break;
1919 }
1920 default:
1921 ret = gst_pad_event_default (pad, parent, event);
1922 break;
1923 }
1924
1925 return ret;
1926 }
1927
1928
1929 static GstStateChangeReturn
gst_sub_parse_change_state(GstElement * element,GstStateChange transition)1930 gst_sub_parse_change_state (GstElement * element, GstStateChange transition)
1931 {
1932 GstStateChangeReturn ret = GST_STATE_CHANGE_SUCCESS;
1933 GstSubParse *self = GST_SUBPARSE (element);
1934
1935 switch (transition) {
1936 case GST_STATE_CHANGE_READY_TO_PAUSED:
1937 /* format detection will init the parser state */
1938 self->offset = 0;
1939 self->parser_type = GST_SUB_PARSE_FORMAT_UNKNOWN;
1940 self->valid_utf8 = TRUE;
1941 self->first_buffer = TRUE;
1942 g_free (self->detected_encoding);
1943 self->detected_encoding = NULL;
1944 g_string_truncate (self->textbuf, 0);
1945 gst_adapter_clear (self->adapter);
1946 break;
1947 default:
1948 break;
1949 }
1950
1951 ret = GST_ELEMENT_CLASS (parent_class)->change_state (element, transition);
1952 if (ret == GST_STATE_CHANGE_FAILURE)
1953 return ret;
1954
1955 switch (transition) {
1956 case GST_STATE_CHANGE_PAUSED_TO_READY:
1957 parser_state_dispose (self, &self->state);
1958 self->parser_type = GST_SUB_PARSE_FORMAT_UNKNOWN;
1959 break;
1960 default:
1961 break;
1962 }
1963
1964 return ret;
1965 }
1966
1967 /*
1968 * Typefind support.
1969 */
1970
1971 /* FIXME 0.11: these caps are ugly, use app/x-subtitle + type field or so;
1972 * also, give different subtitle formats really different types */
1973 static GstStaticCaps mpl2_caps =
1974 GST_STATIC_CAPS ("application/x-subtitle-mpl2");
1975 #define SUB_CAPS (gst_static_caps_get (&sub_caps))
1976
1977 static GstStaticCaps tmp_caps =
1978 GST_STATIC_CAPS ("application/x-subtitle-tmplayer");
1979 #define TMP_CAPS (gst_static_caps_get (&tmp_caps))
1980
1981 static GstStaticCaps sub_caps = GST_STATIC_CAPS ("application/x-subtitle");
1982 #define MPL2_CAPS (gst_static_caps_get (&mpl2_caps))
1983
1984 static GstStaticCaps smi_caps = GST_STATIC_CAPS ("application/x-subtitle-sami");
1985 #define SAMI_CAPS (gst_static_caps_get (&smi_caps))
1986
1987 static GstStaticCaps dks_caps = GST_STATIC_CAPS ("application/x-subtitle-dks");
1988 #define DKS_CAPS (gst_static_caps_get (&dks_caps))
1989
1990 static GstStaticCaps vtt_caps = GST_STATIC_CAPS ("application/x-subtitle-vtt");
1991 #define VTT_CAPS (gst_static_caps_get (&vtt_caps))
1992
1993 static GstStaticCaps qttext_caps =
1994 GST_STATIC_CAPS ("application/x-subtitle-qttext");
1995 #define QTTEXT_CAPS (gst_static_caps_get (&qttext_caps))
1996
1997 static GstStaticCaps lrc_caps = GST_STATIC_CAPS ("application/x-subtitle-lrc");
1998 #define LRC_CAPS (gst_static_caps_get (&lrc_caps))
1999
2000 static void
gst_subparse_type_find(GstTypeFind * tf,gpointer private)2001 gst_subparse_type_find (GstTypeFind * tf, gpointer private)
2002 {
2003 GstSubParseFormat format;
2004 const guint8 *data;
2005 GstCaps *caps;
2006 gchar *str;
2007 gchar *encoding = NULL;
2008 const gchar *end;
2009
2010 if (!(data = gst_type_find_peek (tf, 0, 129)))
2011 return;
2012
2013 /* make sure string passed to _autodetect() is NUL-terminated */
2014 str = g_malloc0 (129);
2015 memcpy (str, data, 128);
2016
2017 if ((encoding = detect_encoding (str, 128)) != NULL) {
2018 gchar *converted_str;
2019 GError *err = NULL;
2020 gsize tmp;
2021
2022 converted_str = gst_convert_to_utf8 (str, 128, encoding, &tmp, &err);
2023 if (converted_str == NULL) {
2024 GST_DEBUG ("Encoding '%s' detected but conversion failed: %s", encoding,
2025 err->message);
2026 g_clear_error (&err);
2027 } else {
2028 g_free (str);
2029 str = converted_str;
2030 }
2031 g_free (encoding);
2032 }
2033
2034 /* Check if at least the first 120 chars are valid UTF8,
2035 * otherwise convert as always */
2036 if (!g_utf8_validate (str, 128, &end) && (end - str) < 120) {
2037 gchar *converted_str;
2038 gsize tmp;
2039 const gchar *enc;
2040
2041 enc = g_getenv ("GST_SUBTITLE_ENCODING");
2042 if (enc == NULL || *enc == '\0') {
2043 /* if local encoding is UTF-8 and no encoding specified
2044 * via the environment variable, assume ISO-8859-15 */
2045 if (g_get_charset (&enc)) {
2046 enc = "ISO-8859-15";
2047 }
2048 }
2049 converted_str = gst_convert_to_utf8 (str, 128, enc, &tmp, NULL);
2050 if (converted_str != NULL) {
2051 g_free (str);
2052 str = converted_str;
2053 }
2054 }
2055
2056 format = gst_sub_parse_data_format_autodetect (str);
2057 g_free (str);
2058
2059 switch (format) {
2060 case GST_SUB_PARSE_FORMAT_MDVDSUB:
2061 GST_DEBUG ("MicroDVD format detected");
2062 caps = SUB_CAPS;
2063 break;
2064 case GST_SUB_PARSE_FORMAT_SUBRIP:
2065 GST_DEBUG ("SubRip format detected");
2066 caps = SUB_CAPS;
2067 break;
2068 case GST_SUB_PARSE_FORMAT_MPSUB:
2069 GST_DEBUG ("MPSub format detected");
2070 caps = SUB_CAPS;
2071 break;
2072 case GST_SUB_PARSE_FORMAT_SAMI:
2073 GST_DEBUG ("SAMI (time-based) format detected");
2074 caps = SAMI_CAPS;
2075 break;
2076 case GST_SUB_PARSE_FORMAT_TMPLAYER:
2077 GST_DEBUG ("TMPlayer (time based) format detected");
2078 caps = TMP_CAPS;
2079 break;
2080 /* FIXME: our MPL2 typefinding is not really good enough to warrant
2081 * returning a high probability (however, since we registered our
2082 * typefinder here with a rank of MARGINAL we should pretty much only
2083 * be called if most other typefinders have already run */
2084 case GST_SUB_PARSE_FORMAT_MPL2:
2085 GST_DEBUG ("MPL2 (time based) format detected");
2086 caps = MPL2_CAPS;
2087 break;
2088 case GST_SUB_PARSE_FORMAT_SUBVIEWER:
2089 GST_DEBUG ("SubViewer format detected");
2090 caps = SUB_CAPS;
2091 break;
2092 case GST_SUB_PARSE_FORMAT_DKS:
2093 GST_DEBUG ("DKS format detected");
2094 caps = DKS_CAPS;
2095 break;
2096 case GST_SUB_PARSE_FORMAT_QTTEXT:
2097 GST_DEBUG ("QTtext format detected");
2098 caps = QTTEXT_CAPS;
2099 break;
2100 case GST_SUB_PARSE_FORMAT_LRC:
2101 GST_DEBUG ("LRC format detected");
2102 caps = LRC_CAPS;
2103 break;
2104 case GST_SUB_PARSE_FORMAT_VTT:
2105 GST_DEBUG ("WebVTT format detected");
2106 caps = VTT_CAPS;
2107 break;
2108 default:
2109 case GST_SUB_PARSE_FORMAT_UNKNOWN:
2110 GST_DEBUG ("no subtitle format detected");
2111 return;
2112 }
2113
2114 /* if we're here, it's ok */
2115 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, caps);
2116 }
2117
2118 static gboolean
plugin_init(GstPlugin * plugin)2119 plugin_init (GstPlugin * plugin)
2120 {
2121 GST_DEBUG_CATEGORY_INIT (sub_parse_debug, "subparse", 0, ".sub parser");
2122
2123 if (!gst_type_find_register (plugin, "subparse_typefind", GST_RANK_MARGINAL,
2124 gst_subparse_type_find, "srt,sub,mpsub,mdvd,smi,txt,dks,vtt",
2125 SUB_CAPS, NULL, NULL))
2126 return FALSE;
2127
2128 if (!gst_element_register (plugin, "subparse",
2129 GST_RANK_PRIMARY, GST_TYPE_SUBPARSE) ||
2130 !gst_element_register (plugin, "ssaparse",
2131 GST_RANK_PRIMARY, GST_TYPE_SSA_PARSE)) {
2132 return FALSE;
2133 }
2134
2135 return TRUE;
2136 }
2137
2138 GST_PLUGIN_DEFINE (GST_VERSION_MAJOR,
2139 GST_VERSION_MINOR,
2140 subparse,
2141 "Subtitle parsing",
2142 plugin_init, VERSION, "LGPL", GST_PACKAGE_NAME, GST_PACKAGE_ORIGIN)
2143