1 /* GStreamer
2 * Copyright (C) 2003 Benjamin Otte <in7y118@public.uni-hamburg.de>
3 * Copyright (C) 2005-2009 Tim-Philipp Müller <tim centricular net>
4 * Copyright (C) 2009 Sebastian Dröge <sebastian.droege@collabora.co.uk>
5 *
6 * gsttypefindfunctions.c: collection of various typefind functions
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 <glib.h>
29 #include <glib/gprintf.h>
30
31 /* don't want to add gio xdgmime typefinder if gio was disabled via configure */
32 #ifdef HAVE_GIO
33 #include <gio/gio.h>
34 #define USE_GIO
35 #endif
36
37 #include <gst/gst.h>
38
39 #include <stdio.h>
40 #include <string.h>
41 #include <ctype.h>
42
43 #include <gst/pbutils/pbutils.h>
44 #include <gst/base/gstbytereader.h>
45
46 GST_DEBUG_CATEGORY_STATIC (type_find_debug);
47 #define GST_CAT_DEFAULT type_find_debug
48
49 /* DataScanCtx: helper for typefind functions that scan through data
50 * step-by-step, to avoid doing a peek at each and every offset */
51
52 #define DATA_SCAN_CTX_CHUNK_SIZE 4096
53
54 typedef struct
55 {
56 guint64 offset;
57 const guint8 *data;
58 guint size;
59 } DataScanCtx;
60
61 static inline void
data_scan_ctx_advance(GstTypeFind * tf,DataScanCtx * c,guint bytes_to_skip)62 data_scan_ctx_advance (GstTypeFind * tf, DataScanCtx * c, guint bytes_to_skip)
63 {
64 c->offset += bytes_to_skip;
65 if (G_LIKELY (c->size > bytes_to_skip)) {
66 c->size -= bytes_to_skip;
67 c->data += bytes_to_skip;
68 } else {
69 c->data += c->size;
70 c->size = 0;
71 }
72 }
73
74 static inline gboolean
data_scan_ctx_ensure_data(GstTypeFind * tf,DataScanCtx * c,guint min_len)75 data_scan_ctx_ensure_data (GstTypeFind * tf, DataScanCtx * c, guint min_len)
76 {
77 const guint8 *data;
78 guint64 len;
79 guint chunk_len = MAX (DATA_SCAN_CTX_CHUNK_SIZE, min_len);
80
81 if (G_LIKELY (c->size >= min_len))
82 return TRUE;
83
84 data = gst_type_find_peek (tf, c->offset, chunk_len);
85 if (G_LIKELY (data != NULL)) {
86 c->data = data;
87 c->size = chunk_len;
88 return TRUE;
89 }
90
91 /* if there's less than our chunk size, try to get as much as we can, but
92 * always at least min_len bytes (we might be typefinding the first buffer
93 * of the stream and not have as much data available as we'd like) */
94 len = gst_type_find_get_length (tf);
95 if (len > 0) {
96 len = CLAMP (len - c->offset, min_len, chunk_len);
97 } else {
98 len = min_len;
99 }
100
101 data = gst_type_find_peek (tf, c->offset, len);
102 if (data != NULL) {
103 c->data = data;
104 c->size = len;
105 return TRUE;
106 }
107
108 return FALSE;
109 }
110
111 static inline gboolean
data_scan_ctx_memcmp(GstTypeFind * tf,DataScanCtx * c,guint offset,const gchar * data,guint len)112 data_scan_ctx_memcmp (GstTypeFind * tf, DataScanCtx * c, guint offset,
113 const gchar * data, guint len)
114 {
115 if (G_UNLIKELY (offset + len >= G_MAXUINT32))
116 return FALSE;
117
118 if (!data_scan_ctx_ensure_data (tf, c, offset + len))
119 return FALSE;
120
121 return (memcmp (c->data + offset, data, len) == 0);
122 }
123
124 /*** text/plain ***/
125 static gboolean xml_check_first_element (GstTypeFind * tf,
126 const gchar * element, guint elen, gboolean strict);
127 static gboolean sdp_check_header (GstTypeFind * tf);
128
129 static GstStaticCaps utf8_caps = GST_STATIC_CAPS ("text/plain");
130
131 #define UTF8_CAPS gst_static_caps_get(&utf8_caps)
132
133 static gboolean
utf8_type_find_have_valid_utf8_at_offset(GstTypeFind * tf,guint64 offset,GstTypeFindProbability * prob)134 utf8_type_find_have_valid_utf8_at_offset (GstTypeFind * tf, guint64 offset,
135 GstTypeFindProbability * prob)
136 {
137 const guint8 *data;
138
139 /* randomly decided values */
140 guint min_size = 16; /* minimum size */
141 guint size = 32 * 1024; /* starting size */
142 guint probability = 95; /* starting probability */
143 guint step = 10; /* how much we reduce probability in each
144 * iteration */
145
146 while (probability > step && size > min_size) {
147 data = gst_type_find_peek (tf, offset, size);
148 if (data) {
149 gchar *end;
150 gchar *start = (gchar *) data;
151
152 if (g_utf8_validate (start, size, (const gchar **) &end) || (end - start + 4 > size)) { /* allow last char to be cut off */
153 *prob = probability;
154 return TRUE;
155 }
156 *prob = 0;
157 return FALSE;
158 }
159 size /= 2;
160 probability -= step;
161 }
162 *prob = 0;
163 return FALSE;
164 }
165
166 static void
utf8_type_find(GstTypeFind * tf,gpointer unused)167 utf8_type_find (GstTypeFind * tf, gpointer unused)
168 {
169 GstTypeFindProbability start_prob, mid_prob;
170 guint64 length;
171
172 /* leave xml to the xml typefinders */
173 if (xml_check_first_element (tf, "", 0, TRUE))
174 return;
175
176 /* leave sdp to the sdp typefinders */
177 if (sdp_check_header (tf))
178 return;
179
180 /* check beginning of stream */
181 if (!utf8_type_find_have_valid_utf8_at_offset (tf, 0, &start_prob))
182 return;
183
184 GST_LOG ("start is plain text with probability of %u", start_prob);
185
186 /* POSSIBLE is the highest probability we ever return if we can't
187 * probe into the middle of the file and don't know its length */
188
189 length = gst_type_find_get_length (tf);
190 if (length == 0 || length == (guint64) - 1) {
191 gst_type_find_suggest (tf, MIN (start_prob, GST_TYPE_FIND_POSSIBLE),
192 UTF8_CAPS);
193 return;
194 }
195
196 if (length < 64 * 1024) {
197 gst_type_find_suggest (tf, start_prob, UTF8_CAPS);
198 return;
199 }
200
201 /* check middle of stream */
202 if (!utf8_type_find_have_valid_utf8_at_offset (tf, length / 2, &mid_prob))
203 return;
204
205 GST_LOG ("middle is plain text with probability of %u", mid_prob);
206 gst_type_find_suggest (tf, (start_prob + mid_prob) / 2, UTF8_CAPS);
207 }
208
209 /*** text/utf-16 and text/utf-32} ***/
210 /* While UTF-8 is unicode too, using text/plain for UTF-16 and UTF-32
211 is going to break stuff. */
212
213 typedef struct
214 {
215 size_t bomlen;
216 const char *const bom;
217 gboolean (*checker) (const guint8 *, gint, gint);
218 int boost;
219 int endianness;
220 } GstUnicodeTester;
221
222 static gboolean
check_utf16(const guint8 * data,gint len,gint endianness)223 check_utf16 (const guint8 * data, gint len, gint endianness)
224 {
225 GstByteReader br;
226 guint16 high, low;
227
228 low = high = 0;
229
230 if (len & 1)
231 return FALSE;
232
233 gst_byte_reader_init (&br, data, len);
234 while (len >= 2) {
235 /* test first for a single 16 bit value in the BMP */
236 if (endianness == G_BIG_ENDIAN)
237 high = gst_byte_reader_get_uint16_be_unchecked (&br);
238 else
239 high = gst_byte_reader_get_uint16_le_unchecked (&br);
240 if (high >= 0xD800 && high <= 0xDBFF) {
241 /* start of a surrogate pair */
242 if (len < 4)
243 return FALSE;
244 len -= 2;
245 if (endianness == G_BIG_ENDIAN)
246 low = gst_byte_reader_get_uint16_be_unchecked (&br);
247 else
248 low = gst_byte_reader_get_uint16_le_unchecked (&br);
249 if (low >= 0xDC00 && low <= 0xDFFF) {
250 /* second half of the surrogate pair */
251 } else
252 return FALSE;
253 } else {
254 if (high >= 0xDC00 && high <= 0xDFFF)
255 return FALSE;
256 }
257 len -= 2;
258 }
259 return TRUE;
260 }
261
262 static gboolean
check_utf32(const guint8 * data,gint len,gint endianness)263 check_utf32 (const guint8 * data, gint len, gint endianness)
264 {
265 if (len & 3)
266 return FALSE;
267 while (len > 3) {
268 guint32 v;
269 if (endianness == G_BIG_ENDIAN)
270 v = GST_READ_UINT32_BE (data);
271 else
272 v = GST_READ_UINT32_LE (data);
273 if (v >= 0x10FFFF)
274 return FALSE;
275 data += 4;
276 len -= 4;
277 }
278 return TRUE;
279 }
280
281 static void
unicode_type_find(GstTypeFind * tf,const GstUnicodeTester * tester,guint n_tester,const char * media_type,gboolean require_bom)282 unicode_type_find (GstTypeFind * tf, const GstUnicodeTester * tester,
283 guint n_tester, const char *media_type, gboolean require_bom)
284 {
285 gsize n;
286 gsize len = 4;
287 const guint8 *data = gst_type_find_peek (tf, 0, len);
288 int prob = -1;
289 const gint max_scan_size = 256 * 1024;
290 int endianness = 0;
291
292 if (!data) {
293 len = 2;
294 data = gst_type_find_peek (tf, 0, len);
295 if (!data)
296 return;
297 }
298
299 /* find a large enough size that works */
300 while (len < max_scan_size) {
301 size_t newlen = len << 1;
302 const guint8 *newdata = gst_type_find_peek (tf, 0, newlen);
303 if (!newdata)
304 break;
305 len = newlen;
306 data = newdata;
307 }
308
309 for (n = 0; n < n_tester; ++n) {
310 int bom_boost = 0, tmpprob;
311 if (len >= tester[n].bomlen) {
312 if (!memcmp (data, tester[n].bom, tester[n].bomlen))
313 bom_boost = tester[n].boost;
314 }
315 if (require_bom && bom_boost == 0)
316 continue;
317 if (!(*tester[n].checker) (data, len, tester[n].endianness))
318 continue;
319 tmpprob = GST_TYPE_FIND_POSSIBLE - 20 + bom_boost;
320 if (tmpprob > prob) {
321 prob = tmpprob;
322 endianness = tester[n].endianness;
323 }
324 }
325
326 if (prob > 0) {
327 GST_DEBUG ("This is valid %s %s", media_type,
328 endianness == G_BIG_ENDIAN ? "be" : "le");
329 gst_type_find_suggest_simple (tf, prob, media_type,
330 "endianness", G_TYPE_INT, endianness, NULL);
331 }
332 }
333
334 static GstStaticCaps utf16_caps = GST_STATIC_CAPS ("text/utf-16");
335
336 #define UTF16_CAPS gst_static_caps_get(&utf16_caps)
337
338 static void
utf16_type_find(GstTypeFind * tf,gpointer unused)339 utf16_type_find (GstTypeFind * tf, gpointer unused)
340 {
341 static const GstUnicodeTester utf16tester[2] = {
342 {2, "\xff\xfe", check_utf16, 10, G_LITTLE_ENDIAN},
343 {2, "\xfe\xff", check_utf16, 20, G_BIG_ENDIAN},
344 };
345 unicode_type_find (tf, utf16tester, G_N_ELEMENTS (utf16tester),
346 "text/utf-16", TRUE);
347 }
348
349 static GstStaticCaps utf32_caps = GST_STATIC_CAPS ("text/utf-32");
350
351 #define UTF32_CAPS gst_static_caps_get(&utf32_caps)
352
353 static void
utf32_type_find(GstTypeFind * tf,gpointer unused)354 utf32_type_find (GstTypeFind * tf, gpointer unused)
355 {
356 static const GstUnicodeTester utf32tester[2] = {
357 {4, "\xff\xfe\x00\x00", check_utf32, 10, G_LITTLE_ENDIAN},
358 {4, "\x00\x00\xfe\xff", check_utf32, 20, G_BIG_ENDIAN}
359 };
360 unicode_type_find (tf, utf32tester, G_N_ELEMENTS (utf32tester),
361 "text/utf-32", TRUE);
362 }
363
364 /*** text/uri-list ***/
365
366 static GstStaticCaps uri_caps = GST_STATIC_CAPS ("text/uri-list");
367
368 #define URI_CAPS (gst_static_caps_get(&uri_caps))
369 #define BUFFER_SIZE 16 /* If the string is < 16 bytes we're screwed */
370 #define INC_BUFFER { \
371 pos++; \
372 if (pos == BUFFER_SIZE) { \
373 pos = 0; \
374 offset += BUFFER_SIZE; \
375 data = gst_type_find_peek (tf, offset, BUFFER_SIZE); \
376 if (data == NULL) return; \
377 } else { \
378 data++; \
379 } \
380 }
381 static void
uri_type_find(GstTypeFind * tf,gpointer unused)382 uri_type_find (GstTypeFind * tf, gpointer unused)
383 {
384 const guint8 *data = gst_type_find_peek (tf, 0, BUFFER_SIZE);
385 guint pos = 0;
386 guint offset = 0;
387
388 if (data) {
389 /* Search for # comment lines */
390 while (*data == '#') {
391 /* Goto end of line */
392 while (*data != '\n') {
393 INC_BUFFER;
394 }
395
396 INC_BUFFER;
397 }
398
399 if (!g_ascii_isalpha (*data)) {
400 /* Had a non alpha char - can't be uri-list */
401 return;
402 }
403
404 INC_BUFFER;
405
406 while (g_ascii_isalnum (*data)) {
407 INC_BUFFER;
408 }
409
410 if (*data != ':') {
411 /* First non alpha char is not a : */
412 return;
413 }
414
415 /* Get the next 2 bytes as well */
416 data = gst_type_find_peek (tf, offset + pos, 3);
417 if (data == NULL)
418 return;
419
420 if (data[1] != '/' && data[2] != '/') {
421 return;
422 }
423
424 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY, URI_CAPS);
425 }
426 }
427
428 /*** application/itc ***/
429 static GstStaticCaps itc_caps = GST_STATIC_CAPS ("application/itc");
430 #define ITC_CAPS (gst_static_caps_get(&itc_caps))
431
432 static void
itc_type_find(GstTypeFind * tf,gpointer unused)433 itc_type_find (GstTypeFind * tf, gpointer unused)
434 {
435 DataScanCtx c = { 0, NULL, 0 };
436 guint8 magic[8] = { 0x00, 0x00, 0x01, 0x1C, 0x69, 0x74, 0x63, 0x68 };
437 guint8 preamble[4] = { 0x00, 0x00, 0x00, 0x02 };
438 guint8 artwork_marker[8] = { 0x00, 0x00, 0x00, 0x00, 0x61, 0x72, 0x74, 0x77 };
439 guint8 item_marker[4] = { 0x69, 0x74, 0x65, 0x6D };
440 GstTypeFindProbability itc_prob = GST_TYPE_FIND_NONE;
441 int i;
442
443 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 8)))
444 return;
445
446 if (memcmp (c.data, magic, 8))
447 return;
448
449 /* At least we found the right magic */
450 itc_prob = GST_TYPE_FIND_MINIMUM;
451 data_scan_ctx_advance (tf, &c, 8);
452
453 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 12)))
454 goto done;
455
456 /* Check preamble 3 consecutive times */
457 for (i = 0; i < 3; i++) {
458 if (memcmp (c.data, preamble, 4))
459 goto done;
460 data_scan_ctx_advance (tf, &c, 4);
461 }
462
463 itc_prob = GST_TYPE_FIND_POSSIBLE;
464
465 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 8)))
466 goto done;
467
468 if (memcmp (c.data, artwork_marker, 8))
469 goto done;
470
471 itc_prob = GST_TYPE_FIND_LIKELY;
472 data_scan_ctx_advance (tf, &c, 8);
473
474 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 256)))
475 goto done;
476
477 /* ...and 256 0x00 padding bytes on what looks like the header's end */
478 for (i = 0; i < 256; i++) {
479 if (c.data[i])
480 goto done;
481 }
482
483 itc_prob = GST_TYPE_FIND_NEARLY_CERTAIN;
484 data_scan_ctx_advance (tf, &c, 256);
485
486 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 8)))
487 goto done;
488
489 if (memcmp (c.data + 4, item_marker, 4))
490 goto done;
491
492 itc_prob = GST_TYPE_FIND_MAXIMUM;
493
494 done:
495 gst_type_find_suggest (tf, itc_prob, ITC_CAPS);
496 }
497
498 /*** application/x-hls ***/
499
500 static GstStaticCaps hls_caps = GST_STATIC_CAPS ("application/x-hls");
501 #define HLS_CAPS (gst_static_caps_get(&hls_caps))
502
503 /* See http://tools.ietf.org/html/draft-pantos-http-live-streaming-05 */
504 static void
hls_type_find(GstTypeFind * tf,gpointer unused)505 hls_type_find (GstTypeFind * tf, gpointer unused)
506 {
507 DataScanCtx c = { 0, NULL, 0 };
508
509 /* Minimum useful size is #EXTM3U\n + 1 tag + ':' = 30 bytes */
510 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 30)))
511 return;
512
513 if (memcmp (c.data, "#EXTM3U", 7))
514 return;
515
516 data_scan_ctx_advance (tf, &c, 7);
517
518 /* Check only the first 4KB */
519 while (c.offset < 4096) {
520 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 21)))
521 return;
522
523 /* Search for # comment lines */
524 if (c.data[0] == '#' && (memcmp (c.data, "#EXT-X-TARGETDURATION", 21) == 0
525 || memcmp (c.data, "#EXT-X-STREAM-INF", 17) == 0
526 || memcmp (c.data, "#EXT-X-MEDIA", 12) == 0)) {
527 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, HLS_CAPS);
528 return;
529 }
530
531 data_scan_ctx_advance (tf, &c, 1);
532 }
533 }
534
535
536 /*** application/xml **********************************************************/
537
538 #define XML_BUFFER_SIZE 16
539 #define XML_INC_BUFFER { \
540 pos++; \
541 if (pos == XML_BUFFER_SIZE) { \
542 pos = 0; \
543 offset += XML_BUFFER_SIZE; \
544 data = gst_type_find_peek (tf, offset, XML_BUFFER_SIZE); \
545 if (data == NULL) return FALSE; \
546 } else { \
547 data++; \
548 } \
549 }
550
551 #define XML_INC_BUFFER_DATA { \
552 pos++; \
553 if (pos >= length) { \
554 return FALSE; \
555 } else { \
556 data++; \
557 } \
558 }
559
560 static gboolean
xml_check_first_element_from_data(const guint8 * data,guint length,const gchar * element,guint elen,gboolean strict)561 xml_check_first_element_from_data (const guint8 * data, guint length,
562 const gchar * element, guint elen, gboolean strict)
563 {
564 gboolean got_xmldec;
565 guint pos = 0;
566
567 g_return_val_if_fail (data != NULL, FALSE);
568
569 if (length <= 5)
570 return FALSE;
571
572 /* look for the XMLDec
573 * see XML spec 2.8, Prolog and Document Type Declaration
574 * http://www.w3.org/TR/2004/REC-xml-20040204/#sec-prolog-dtd */
575 got_xmldec = (memcmp (data, "<?xml", 5) == 0);
576
577 if (strict && !got_xmldec)
578 return FALSE;
579
580 /* skip XMLDec in any case if we've got one */
581 if (got_xmldec) {
582 pos += 5;
583 data += 5;
584 }
585
586 /* look for the first element, it has to be the requested element. Bail
587 * out if it is not within the first 4kB. */
588 while (pos < MIN (4096, length)) {
589 while (*data != '<' && pos < MIN (4096, length)) {
590 XML_INC_BUFFER_DATA;
591 }
592
593 XML_INC_BUFFER_DATA;
594 if (!g_ascii_isalpha (*data)) {
595 /* if not alphabetic, it's a PI or an element / attribute declaration
596 * like <?xxx or <!xxx */
597 XML_INC_BUFFER_DATA;
598 continue;
599 }
600
601 /* the first normal element, check if it's the one asked for */
602 if (pos + elen + 1 >= length)
603 return FALSE;
604 return (element && strncmp ((const char *) data, element, elen) == 0);
605 }
606
607 return FALSE;
608 }
609
610 static gboolean
xml_check_first_element(GstTypeFind * tf,const gchar * element,guint elen,gboolean strict)611 xml_check_first_element (GstTypeFind * tf, const gchar * element, guint elen,
612 gboolean strict)
613 {
614 gboolean got_xmldec;
615 const guint8 *data;
616 guint offset = 0;
617 guint pos = 0;
618
619 data = gst_type_find_peek (tf, 0, XML_BUFFER_SIZE);
620 if (!data)
621 return FALSE;
622
623 /* look for the XMLDec
624 * see XML spec 2.8, Prolog and Document Type Declaration
625 * http://www.w3.org/TR/2004/REC-xml-20040204/#sec-prolog-dtd */
626 got_xmldec = (memcmp (data, "<?xml", 5) == 0);
627
628 if (strict && !got_xmldec)
629 return FALSE;
630
631 /* skip XMLDec in any case if we've got one */
632 if (got_xmldec) {
633 pos += 5;
634 data += 5;
635 }
636
637 /* look for the first element, it has to be the requested element. Bail
638 * out if it is not within the first 4kB. */
639 while (data && (offset + pos) < 4096) {
640 while (*data != '<' && (offset + pos) < 4096) {
641 XML_INC_BUFFER;
642 }
643
644 XML_INC_BUFFER;
645 if (!g_ascii_isalpha (*data)) {
646 /* if not alphabetic, it's a PI or an element / attribute declaration
647 * like <?xxx or <!xxx */
648 XML_INC_BUFFER;
649 continue;
650 }
651
652 /* the first normal element, check if it's the one asked for */
653 data = gst_type_find_peek (tf, offset + pos, elen + 1);
654 return (data && element && strncmp ((char *) data, element, elen) == 0);
655 }
656
657 return FALSE;
658 }
659
660 static GstStaticCaps generic_xml_caps = GST_STATIC_CAPS ("application/xml");
661
662 #define GENERIC_XML_CAPS (gst_static_caps_get(&generic_xml_caps))
663 static void
xml_type_find(GstTypeFind * tf,gpointer unused)664 xml_type_find (GstTypeFind * tf, gpointer unused)
665 {
666 if (xml_check_first_element (tf, "", 0, TRUE)) {
667 gst_type_find_suggest (tf, GST_TYPE_FIND_MINIMUM, GENERIC_XML_CAPS);
668 }
669 }
670
671 /*** application/dash+xml ****************************************************/
672
673 static GstStaticCaps dash_caps = GST_STATIC_CAPS ("application/dash+xml");
674
675 #define DASH_CAPS gst_static_caps_get (&dash_caps)
676
677 static void
dash_mpd_type_find(GstTypeFind * tf,gpointer unused)678 dash_mpd_type_find (GstTypeFind * tf, gpointer unused)
679 {
680 if (xml_check_first_element (tf, "MPD", 3, FALSE) ||
681 xml_check_first_element (tf, "mpd", 3, FALSE)) {
682 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, DASH_CAPS);
683 }
684 }
685
686 /*** application/xges ****************************************************/
687
688 static GstStaticCaps xges_caps = GST_STATIC_CAPS ("application/xges");
689
690 #define XGES_CAPS gst_static_caps_get (&xges_caps)
691
692 static void
xges_type_find(GstTypeFind * tf,gpointer unused)693 xges_type_find (GstTypeFind * tf, gpointer unused)
694 {
695 if (xml_check_first_element (tf, "ges", 3, FALSE)) {
696 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, XGES_CAPS);
697 }
698 }
699
700
701 /*** application/sdp *********************************************************/
702
703 static GstStaticCaps sdp_caps = GST_STATIC_CAPS ("application/sdp");
704
705 #define SDP_CAPS (gst_static_caps_get(&sdp_caps))
706 static gboolean
sdp_check_header(GstTypeFind * tf)707 sdp_check_header (GstTypeFind * tf)
708 {
709 const guint8 *data;
710
711 data = gst_type_find_peek (tf, 0, 5);
712 if (!data)
713 return FALSE;
714
715 /* sdp must start with v=0[\r]\n */
716 if (memcmp (data, "v=0", 3))
717 return FALSE;
718
719 if (data[3] == '\r' && data[4] == '\n')
720 return TRUE;
721 if (data[3] == '\n')
722 return TRUE;
723
724 return FALSE;
725 }
726
727 static void
sdp_type_find(GstTypeFind * tf,gpointer unused)728 sdp_type_find (GstTypeFind * tf, gpointer unused)
729 {
730 if (sdp_check_header (tf))
731 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, SDP_CAPS);
732 }
733
734 /*** application/smil *********************************************************/
735
736 static GstStaticCaps smil_caps = GST_STATIC_CAPS ("application/smil");
737
738 #define SMIL_CAPS (gst_static_caps_get(&smil_caps))
739 static void
smil_type_find(GstTypeFind * tf,gpointer unused)740 smil_type_find (GstTypeFind * tf, gpointer unused)
741 {
742 if (xml_check_first_element (tf, "smil", 4, FALSE)) {
743 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, SMIL_CAPS);
744 }
745 }
746
747 /*** application/ttml+xml *****************************************************/
748
749 static GstStaticCaps ttml_xml_caps = GST_STATIC_CAPS ("application/ttml+xml");
750
751 #define TTML_XML_CAPS (gst_static_caps_get(&ttml_xml_caps))
752 static void
ttml_xml_type_find(GstTypeFind * tf,gpointer unused)753 ttml_xml_type_find (GstTypeFind * tf, gpointer unused)
754 {
755 if (xml_check_first_element (tf, "tt", 2, FALSE)) {
756 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, TTML_XML_CAPS);
757 }
758 }
759
760 /*** text/html ***/
761
762 static GstStaticCaps html_caps = GST_STATIC_CAPS ("text/html");
763
764 #define HTML_CAPS gst_static_caps_get (&html_caps)
765
766 static void
html_type_find(GstTypeFind * tf,gpointer unused)767 html_type_find (GstTypeFind * tf, gpointer unused)
768 {
769 const gchar *d, *data;
770
771 data = (const gchar *) gst_type_find_peek (tf, 0, 16);
772 if (!data)
773 return;
774
775 if (!g_ascii_strncasecmp (data, "<!DOCTYPE HTML", 14)) {
776 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, HTML_CAPS);
777 } else if (xml_check_first_element (tf, "html", 4, FALSE)) {
778 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, HTML_CAPS);
779 } else if ((d = memchr (data, '<', 16))) {
780 data = (const gchar *) gst_type_find_peek (tf, d - data, 6);
781 if (data && g_ascii_strncasecmp (data, "<html>", 6) == 0) {
782 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, HTML_CAPS);
783 }
784 }
785 }
786
787 /*** audio/midi ***/
788
789 static GstStaticCaps mid_caps = GST_STATIC_CAPS ("audio/midi");
790
791 #define MID_CAPS gst_static_caps_get(&mid_caps)
792 static void
mid_type_find(GstTypeFind * tf,gpointer unused)793 mid_type_find (GstTypeFind * tf, gpointer unused)
794 {
795 const guint8 *data = gst_type_find_peek (tf, 0, 4);
796
797 /* http://jedi.ks.uiuc.edu/~johns/links/music/midifile.html */
798 if (data && data[0] == 'M' && data[1] == 'T' && data[2] == 'h'
799 && data[3] == 'd')
800 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MID_CAPS);
801 }
802
803 /*** audio/mobile-xmf ***/
804
805 static GstStaticCaps mxmf_caps = GST_STATIC_CAPS ("audio/mobile-xmf");
806
807 #define MXMF_CAPS gst_static_caps_get(&mxmf_caps)
808 static void
mxmf_type_find(GstTypeFind * tf,gpointer unused)809 mxmf_type_find (GstTypeFind * tf, gpointer unused)
810 {
811 const guint8 *data = NULL;
812
813 /* Search FileId "XMF_" 4 bytes */
814 data = gst_type_find_peek (tf, 0, 4);
815 if (data && data[0] == 'X' && data[1] == 'M' && data[2] == 'F'
816 && data[3] == '_') {
817 /* Search Format version "2.00" 4 bytes */
818 data = gst_type_find_peek (tf, 4, 4);
819 if (data && data[0] == '2' && data[1] == '.' && data[2] == '0'
820 && data[3] == '0') {
821 /* Search TypeId 2 1 byte */
822 data = gst_type_find_peek (tf, 11, 1);
823 if (data && data[0] == 2) {
824 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MXMF_CAPS);
825 }
826 }
827 }
828 }
829
830
831 /*** video/x-fli ***/
832
833 static GstStaticCaps flx_caps = GST_STATIC_CAPS ("video/x-fli");
834
835 #define FLX_CAPS gst_static_caps_get(&flx_caps)
836 static void
flx_type_find(GstTypeFind * tf,gpointer unused)837 flx_type_find (GstTypeFind * tf, gpointer unused)
838 {
839 const guint8 *data = gst_type_find_peek (tf, 0, 134);
840
841 if (data) {
842 /* check magic and the frame type of the first frame */
843 if ((data[4] == 0x11 || data[4] == 0x12 ||
844 data[4] == 0x30 || data[4] == 0x44) &&
845 data[5] == 0xaf &&
846 ((data[132] == 0x00 || data[132] == 0xfa) && data[133] == 0xf1)) {
847 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, FLX_CAPS);
848 }
849 return;
850 }
851 data = gst_type_find_peek (tf, 0, 6);
852 if (data) {
853 /* check magic only */
854 if ((data[4] == 0x11 || data[4] == 0x12 ||
855 data[4] == 0x30 || data[4] == 0x44) && data[5] == 0xaf) {
856 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY, FLX_CAPS);
857 }
858 return;
859 }
860 }
861
862 /*** application/x-id3 ***/
863
864 static GstStaticCaps id3_caps = GST_STATIC_CAPS ("application/x-id3");
865
866 #define ID3_CAPS gst_static_caps_get(&id3_caps)
867 static void
id3v2_type_find(GstTypeFind * tf,gpointer unused)868 id3v2_type_find (GstTypeFind * tf, gpointer unused)
869 {
870 /* ohos.opt.compat.0001: The demux of gstplayer does not accurately parse audio resources in the MP3 format.
871 * As a result, the duration value cannot be obtained in the preparation phase.
872 * Use the demux and typefind of ffmpeg to process audio resources in MP3 format.
873 */
874 #ifdef OHOS_OPT_COMPAT
875 return;
876 #endif
877 const guint8 *data = gst_type_find_peek (tf, 0, 10);
878
879 if (data && memcmp (data, "ID3", 3) == 0 &&
880 data[3] != 0xFF && data[4] != 0xFF &&
881 (data[6] & 0x80) == 0 && (data[7] & 0x80) == 0 &&
882 (data[8] & 0x80) == 0 && (data[9] & 0x80) == 0) {
883 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, ID3_CAPS);
884 }
885 }
886
887 static void
id3v1_type_find(GstTypeFind * tf,gpointer unused)888 id3v1_type_find (GstTypeFind * tf, gpointer unused)
889 {
890 /* ohos.opt.compat.0001: The demux of gstplayer does not accurately parse audio resources in the MP3 format.
891 * As a result, the duration value cannot be obtained in the preparation phase.
892 * Use the demux and typefind of ffmpeg to process audio resources in MP3 format.
893 */
894 #ifdef OHOS_OPT_COMPAT
895 return;
896 #endif
897 const guint8 *data = gst_type_find_peek (tf, -128, 3);
898
899 if (data && memcmp (data, "TAG", 3) == 0) {
900 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, ID3_CAPS);
901 }
902 }
903
904 /*** application/x-ape ***/
905
906 static GstStaticCaps apetag_caps = GST_STATIC_CAPS ("application/x-apetag");
907
908 #define APETAG_CAPS gst_static_caps_get(&apetag_caps)
909 static void
apetag_type_find(GstTypeFind * tf,gpointer unused)910 apetag_type_find (GstTypeFind * tf, gpointer unused)
911 {
912 const guint8 *data;
913
914 /* APEv1/2 at start of file */
915 data = gst_type_find_peek (tf, 0, 8);
916 if (data && !memcmp (data, "APETAGEX", 8)) {
917 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, APETAG_CAPS);
918 return;
919 }
920
921 /* APEv1/2 at end of file */
922 data = gst_type_find_peek (tf, -32, 8);
923 if (data && !memcmp (data, "APETAGEX", 8)) {
924 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, APETAG_CAPS);
925 return;
926 }
927 }
928
929 /*** audio/x-ttafile ***/
930
931 static GstStaticCaps tta_caps = GST_STATIC_CAPS ("audio/x-ttafile");
932
933 #define TTA_CAPS gst_static_caps_get(&tta_caps)
934 static void
tta_type_find(GstTypeFind * tf,gpointer unused)935 tta_type_find (GstTypeFind * tf, gpointer unused)
936 {
937 const guint8 *data = gst_type_find_peek (tf, 0, 3);
938
939 if (data) {
940 if (memcmp (data, "TTA", 3) == 0) {
941 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, TTA_CAPS);
942 return;
943 }
944 }
945 }
946
947 /*** audio/x-flac ***/
948 static GstStaticCaps flac_caps = GST_STATIC_CAPS ("audio/x-flac");
949
950 #define FLAC_CAPS (gst_static_caps_get(&flac_caps))
951
952 static void
flac_type_find(GstTypeFind * tf,gpointer unused)953 flac_type_find (GstTypeFind * tf, gpointer unused)
954 {
955 DataScanCtx c = { 0, NULL, 0 };
956
957 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 4)))
958 return;
959
960 /* standard flac (also old/broken flac-in-ogg with an initial 4-byte marker
961 * packet and without the usual packet framing) */
962 if (memcmp (c.data, "fLaC", 4) == 0) {
963 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, FLAC_CAPS);
964 return;
965 }
966
967 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 6)))
968 return;
969
970 /* flac-in-ogg, see http://flac.sourceforge.net/ogg_mapping.html */
971 if (memcmp (c.data, "\177FLAC\001", 6) == 0) {
972 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, FLAC_CAPS);
973 return;
974 }
975
976 /* disabled because it happily typefinds /dev/urandom as audio/x-flac, and
977 * because I yet have to see header-less flac in the wild */
978 #if 0
979 /* flac without headers (subset format) */
980 /* 64K should be enough */
981 while (c.offset < (64 * 1024)) {
982 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 4)))
983 break;
984
985 /* look for frame header,
986 * http://flac.sourceforge.net/format.html#frame_header
987 */
988 if (c.data[0] == 0xff && (c.data[1] >> 2) == 0x3e) {
989 /* bit 15 in the header must be 0 */
990 if (((c.data[1] >> 1) & 0x01) == 0x01)
991 goto advance;
992
993 /* blocksize must be != 0x00 */
994 if ((c.data[2] >> 4) == 0x00)
995 goto advance;
996
997 /* samplerate must be != 0x0f */
998 if ((c.data[2] & 0x0f) == 0x0f)
999 goto advance;
1000 /* also 0 is invalid, as it means get the info from the header and we
1001 * don't have headers if we are here */
1002 if ((c.data[2] & 0x0f) == 0x00)
1003 goto advance;
1004
1005 /* channel assignment must be < 11 */
1006 if ((c.data[3] >> 4) >= 11)
1007 goto advance;
1008
1009 /* sample size must be != 0x07 and != 0x05 */
1010 if (((c.data[3] >> 1) & 0x07) == 0x07)
1011 goto advance;
1012 if (((c.data[3] >> 1) & 0x07) == 0x05)
1013 goto advance;
1014 /* also 0 is invalid, as it means get the info from the header and we
1015 * don't have headers if we are here */
1016 if (((c.data[3] >> 1) & 0x07) == 0x00)
1017 goto advance;
1018
1019 /* next bit must be 0 */
1020 if ((c.data[3] & 0x01) == 0x01)
1021 goto advance;
1022
1023 /* FIXME: shouldn't we include the crc check ? */
1024
1025 GST_DEBUG ("Found flac without headers at %d", (gint) c.offset);
1026 gst_type_find_suggest (tf, GST_TYPE_FIND_POSSIBLE, FLAC_CAPS);
1027 return;
1028 }
1029 advance:
1030 data_scan_ctx_advance (tf, &c, 1);
1031 }
1032 #endif
1033 }
1034
1035 /* TODO: we could probably make a generic function for this.. */
1036 static gint
aac_type_find_scan_loas_frames_ep(GstTypeFind * tf,DataScanCtx * scan_ctx,gint max_frames)1037 aac_type_find_scan_loas_frames_ep (GstTypeFind * tf, DataScanCtx * scan_ctx,
1038 gint max_frames)
1039 {
1040 DataScanCtx c = *scan_ctx;
1041 guint16 snc;
1042 guint len;
1043 gint count = 0;
1044
1045 do {
1046 if (!data_scan_ctx_ensure_data (tf, &c, 5))
1047 break;
1048
1049 /* EPAudioSyncStream */
1050 len = ((c.data[2] & 0x0f) << 9) | (c.data[3] << 1) |
1051 ((c.data[4] & 0x80) >> 7);
1052
1053 if (len == 0 || !data_scan_ctx_ensure_data (tf, &c, len + 2)) {
1054 GST_DEBUG ("Wrong sync or next frame not within reach, len=%u", len);
1055 break;
1056 }
1057
1058 /* check length of frame */
1059 snc = GST_READ_UINT16_BE (c.data + len);
1060 if (snc != 0x4de1) {
1061 GST_DEBUG ("No sync found at 0x%" G_GINT64_MODIFIER "x", c.offset + len);
1062 break;
1063 }
1064
1065 ++count;
1066
1067 GST_DEBUG ("Found LOAS syncword #%d at offset 0x%" G_GINT64_MODIFIER "x, "
1068 "framelen %u", count, c.offset, len);
1069
1070 data_scan_ctx_advance (tf, &c, len);
1071 } while (count < max_frames && (c.offset - scan_ctx->offset) < 64 * 1024);
1072
1073 GST_DEBUG ("found %d consecutive frames", count);
1074 return count;
1075 }
1076
1077 static gint
aac_type_find_scan_loas_frames(GstTypeFind * tf,DataScanCtx * scan_ctx,gint max_frames)1078 aac_type_find_scan_loas_frames (GstTypeFind * tf, DataScanCtx * scan_ctx,
1079 gint max_frames)
1080 {
1081 DataScanCtx c = *scan_ctx;
1082 guint16 snc;
1083 guint len;
1084 gint count = 0;
1085
1086 do {
1087 if (!data_scan_ctx_ensure_data (tf, &c, 3))
1088 break;
1089
1090 /* AudioSyncStream */
1091 len = ((c.data[1] & 0x1f) << 8) | c.data[2];
1092 /* add size of sync stream header */
1093 len += 3;
1094
1095 if (len == 0 || !data_scan_ctx_ensure_data (tf, &c, len + 2)) {
1096 GST_DEBUG ("Wrong sync or next frame not within reach, len=%u", len);
1097 break;
1098 }
1099
1100 /* check length of frame */
1101 snc = GST_READ_UINT16_BE (c.data + len);
1102 if ((snc & 0xffe0) != 0x56e0) {
1103 GST_DEBUG ("No sync found at 0x%" G_GINT64_MODIFIER "x", c.offset + len);
1104 break;
1105 }
1106
1107 ++count;
1108
1109 GST_DEBUG ("Found LOAS syncword #%d at offset 0x%" G_GINT64_MODIFIER "x, "
1110 "framelen %u", count, c.offset, len);
1111
1112 data_scan_ctx_advance (tf, &c, len);
1113 } while (count < max_frames && (c.offset - scan_ctx->offset) < 64 * 1024);
1114
1115 GST_DEBUG ("found %d consecutive frames", count);
1116 return count;
1117 }
1118
1119 /*** audio/mpeg version 2, 4 ***/
1120
1121 static GstStaticCaps aac_caps = GST_STATIC_CAPS ("audio/mpeg, "
1122 "mpegversion = (int) { 2, 4 }, framed = (bool) false");
1123 #define AAC_CAPS (gst_static_caps_get(&aac_caps))
1124 #define AAC_AMOUNT (4096)
1125 static void
aac_type_find(GstTypeFind * tf,gpointer unused)1126 aac_type_find (GstTypeFind * tf, gpointer unused)
1127 {
1128 /* ohos.opt.compat.0002: the demux of gstplayer does not accurately parse audio resources in the aac format.
1129 * As a result, the duration value cannot be obtained in the preparation phase.
1130 * Use the demux and typefind of ffmpeg to process audio resources in aac format.
1131 */
1132 #ifdef OHOS_OPT_COMPAT
1133 return;
1134 #endif
1135 DataScanCtx c = { 0, NULL, 0 };
1136 GstTypeFindProbability best_probability = GST_TYPE_FIND_NONE;
1137 GstCaps *best_caps = NULL;
1138 gint best_count = 0;
1139
1140 while (c.offset < AAC_AMOUNT) {
1141 guint snc, len, offset, i;
1142
1143 /* detect adts header or adif header.
1144 * The ADIF header is 4 bytes, that should be OK. The ADTS header, on
1145 * the other hand, is 14 bits only, so we require one valid frame with
1146 * again a valid syncpoint on the next one (28 bits) for certainty. We
1147 * require 4 kB, which is quite a lot, since frames are generally 200-400
1148 * bytes.
1149 * LOAS has 2 possible syncwords, which are 11 bits and 16 bits long.
1150 * The following stream syntax depends on which one is found.
1151 */
1152 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 6)))
1153 break;
1154
1155 snc = GST_READ_UINT16_BE (c.data);
1156 if (G_UNLIKELY ((snc & 0xfff6) == 0xfff0)) {
1157 /* ADTS header - find frame length */
1158 GST_DEBUG ("Found one ADTS syncpoint at offset 0x%" G_GINT64_MODIFIER
1159 "x, tracing next...", c.offset);
1160 len = ((c.data[3] & 0x03) << 11) |
1161 (c.data[4] << 3) | ((c.data[5] & 0xe0) >> 5);
1162
1163 if (len == 0 || !data_scan_ctx_ensure_data (tf, &c, len + 6)) {
1164 GST_DEBUG ("Wrong sync or next frame not within reach, len=%u", len);
1165 goto next;
1166 }
1167
1168 offset = len;
1169 /* check if there's a second ADTS frame */
1170 snc = GST_READ_UINT16_BE (c.data + offset);
1171 if ((snc & 0xfff6) == 0xfff0) {
1172 GstCaps *caps;
1173 guint mpegversion, sample_freq_idx, channel_config, profile_idx, rate;
1174 guint8 audio_config[2];
1175
1176 mpegversion = (c.data[1] & 0x08) ? 2 : 4;
1177 profile_idx = c.data[2] >> 6;
1178 sample_freq_idx = ((c.data[2] & 0x3c) >> 2);
1179 channel_config = ((c.data[2] & 0x01) << 2) + (c.data[3] >> 6);
1180
1181 GST_DEBUG ("Found second ADTS-%d syncpoint at offset 0x%"
1182 G_GINT64_MODIFIER "x, framelen %u", mpegversion, c.offset, len);
1183
1184 /* 0xd and 0xe are reserved. 0xf means the sample frequency is directly
1185 * specified in the header, but that's not allowed for ADTS */
1186 if (sample_freq_idx > 0xc) {
1187 GST_DEBUG ("Unexpected sample frequency index %d or wrong sync",
1188 sample_freq_idx);
1189 goto next;
1190 }
1191
1192 rate = gst_codec_utils_aac_get_sample_rate_from_index (sample_freq_idx);
1193 GST_LOG ("ADTS: profile=%u, rate=%u", profile_idx, rate);
1194
1195 /* The ADTS frame header is slightly different from the
1196 * AudioSpecificConfig defined for the MPEG-4 container, so we just
1197 * construct enough of it for getting the level here. */
1198 /* ADTS counts profiles from 0 instead of 1 to save bits */
1199 audio_config[0] = (profile_idx + 1) << 3;
1200 audio_config[0] |= (sample_freq_idx >> 1) & 0x7;
1201 audio_config[1] = (sample_freq_idx & 0x1) << 7;
1202 audio_config[1] |= (channel_config & 0xf) << 3;
1203
1204 caps = gst_caps_new_simple ("audio/mpeg",
1205 "framed", G_TYPE_BOOLEAN, FALSE,
1206 "mpegversion", G_TYPE_INT, mpegversion,
1207 "stream-format", G_TYPE_STRING, "adts", NULL);
1208
1209 gst_codec_utils_aac_caps_set_level_and_profile (caps, audio_config, 2);
1210
1211 /* add rate and number of channels if we can */
1212 if (channel_config != 0 && channel_config <= 7) {
1213 const guint channels_map[] = { 0, 1, 2, 3, 4, 5, 6, 8 };
1214
1215 gst_caps_set_simple (caps, "channels", G_TYPE_INT,
1216 channels_map[channel_config], "rate", G_TYPE_INT, rate, NULL);
1217 }
1218
1219 /* length of the second ADTS frame */
1220 len = ((c.data[offset + 3] & 0x03) << 11) |
1221 (c.data[offset + 4] << 3) | ((c.data[offset + 5] & 0xe0) >> 5);
1222
1223 if (len == 0 || !data_scan_ctx_ensure_data (tf, &c, offset + len + 6)) {
1224 GST_DEBUG ("Wrong sync or next frame not within reach, len=%u", len);
1225 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY, caps);
1226 } else {
1227 offset += len;
1228 /* find more aac sync to select correctly */
1229 /* check if there's a third/fourth/fifth/sixth ADTS frame, if there is a sixth frame, set probability to maximum:100% */
1230 for (i = 3; i <= 6; i++) {
1231 len = ((c.data[offset + 3] & 0x03) << 11) |
1232 (c.data[offset + 4] << 3) | ((c.data[offset + 5] & 0xe0) >> 5);
1233 if (len == 0
1234 || !data_scan_ctx_ensure_data (tf, &c, offset + len + 6)) {
1235 GST_DEBUG ("Wrong sync or next frame not within reach, len=%u",
1236 len);
1237 break;
1238 }
1239 snc = GST_READ_UINT16_BE (c.data + offset);
1240 if ((snc & 0xfff6) == 0xfff0) {
1241 GST_DEBUG ("Find %und Sync..probability is %u ", i,
1242 GST_TYPE_FIND_LIKELY + 5 * (i - 2));
1243 offset += len;
1244 } else {
1245 break;
1246 }
1247 }
1248 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY + 5 * (i - 3), caps);
1249
1250 }
1251 gst_caps_unref (caps);
1252 break;
1253 }
1254
1255 GST_DEBUG ("No next frame found... (should have been at 0x%x)", len);
1256 } else if (G_UNLIKELY ((snc & 0xffe0) == 0x56e0 || snc == 0x4de1)) {
1257 gint count;
1258
1259 /* LOAS frame */
1260 GST_INFO ("Possible LOAS syncword at offset 0x%" G_GINT64_MODIFIER
1261 "x, scanning for more frames...", c.offset);
1262
1263 if (snc == 0x4de1)
1264 count = aac_type_find_scan_loas_frames_ep (tf, &c, 20);
1265 else
1266 count = aac_type_find_scan_loas_frames (tf, &c, 20);
1267
1268 if (count >= 3 && count > best_count) {
1269 gst_caps_replace (&best_caps, NULL);
1270 best_caps = gst_caps_new_simple ("audio/mpeg",
1271 "framed", G_TYPE_BOOLEAN, FALSE,
1272 "mpegversion", G_TYPE_INT, 4,
1273 "stream-format", G_TYPE_STRING, "loas", NULL);
1274 best_count = count;
1275 best_probability = GST_TYPE_FIND_POSSIBLE - 10 + count * 3;
1276 if (best_probability >= GST_TYPE_FIND_LIKELY)
1277 break;
1278 }
1279 } else if (!memcmp (c.data, "ADIF", 4)) {
1280 /* ADIF header */
1281 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_LIKELY, "audio/mpeg",
1282 "framed", G_TYPE_BOOLEAN, FALSE, "mpegversion", G_TYPE_INT, 4,
1283 "stream-format", G_TYPE_STRING, "adif", NULL);
1284 break;
1285 }
1286
1287 next:
1288
1289 data_scan_ctx_advance (tf, &c, 1);
1290 }
1291
1292 if (best_probability > GST_TYPE_FIND_NONE) {
1293 gst_type_find_suggest (tf, best_probability, best_caps);
1294 gst_caps_unref (best_caps);
1295 }
1296 }
1297
1298 /*** audio/mpeg version 1 ***/
1299
1300 /*
1301 * The chance that random data is identified as a valid mp3 header is 63 / 2^18
1302 * (0.024%) per try. This makes the function for calculating false positives
1303 * 1 - (1 - ((63 / 2 ^18) ^ GST_MP3_TYPEFIND_MIN_HEADERS)) ^ buffersize)
1304 * This has the following probabilities of false positives:
1305 * datasize MIN_HEADERS
1306 * (bytes) 1 2 3 4
1307 * 4096 62.6% 0.02% 0% 0%
1308 * 16384 98% 0.09% 0% 0%
1309 * 1 MiB 100% 5.88% 0% 0%
1310 * 1 GiB 100% 100% 1.44% 0%
1311 * 1 TiB 100% 100% 100% 0.35%
1312 * This means that the current choice (3 headers by most of the time 4096 byte
1313 * buffers is pretty safe for now.
1314 *
1315 * The max. size of each frame is 1440 bytes, which means that for N frames to
1316 * be detected, we need 1440 * GST_MP3_TYPEFIND_MIN_HEADERS + 3 bytes of data.
1317 * Assuming we step into the stream right after the frame header, this
1318 * means we need 1440 * (GST_MP3_TYPEFIND_MIN_HEADERS + 1) - 1 + 3 bytes
1319 * of data (5762) to always detect any mp3.
1320 */
1321
1322 static const guint mp3types_bitrates[2][3][16] =
1323 { {{0, 32, 64, 96, 128, 160, 192, 224, 256, 288, 320, 352, 384, 416, 448,},
1324 {0, 32, 48, 56, 64, 80, 96, 112, 128, 160, 192, 224, 256, 320, 384,},
1325 {0, 32, 40, 48, 56, 64, 80, 96, 112, 128, 160, 192, 224, 256, 320,}},
1326 {{0, 32, 48, 56, 64, 80, 96, 112, 128, 144, 160, 176, 192, 224, 256,},
1327 {0, 8, 16, 24, 32, 40, 48, 56, 64, 80, 96, 112, 128, 144, 160,},
1328 {0, 8, 16, 24, 32, 40, 48, 56, 64, 80, 96, 112, 128, 144, 160,}},
1329 };
1330
1331 static const guint mp3types_freqs[3][3] = { {11025, 12000, 8000},
1332 {22050, 24000, 16000},
1333 {44100, 48000, 32000}
1334 };
1335
1336 static inline guint
mp3_type_frame_length_from_header(guint32 header,guint * put_layer,guint * put_channels,guint * put_bitrate,guint * put_samplerate,gboolean * may_be_free_format,gint possible_free_framelen)1337 mp3_type_frame_length_from_header (guint32 header, guint * put_layer,
1338 guint * put_channels, guint * put_bitrate, guint * put_samplerate,
1339 gboolean * may_be_free_format, gint possible_free_framelen)
1340 {
1341 guint bitrate, layer, length, mode, samplerate, version, channels;
1342
1343 if ((header & 0xffe00000) != 0xffe00000)
1344 return 0;
1345
1346 /* we don't need extension, copyright, original or
1347 * emphasis for the frame length */
1348 header >>= 6;
1349
1350 /* mode */
1351 mode = header & 0x3;
1352 header >>= 3;
1353
1354 /* padding */
1355 length = header & 0x1;
1356 header >>= 1;
1357
1358 /* sampling frequency */
1359 samplerate = header & 0x3;
1360 if (samplerate == 3)
1361 return 0;
1362 header >>= 2;
1363
1364 /* bitrate index */
1365 bitrate = header & 0xF;
1366 if (bitrate == 0 && possible_free_framelen == -1) {
1367 GST_LOG ("Possibly a free format mp3 - signaling");
1368 *may_be_free_format = TRUE;
1369 }
1370 if (bitrate == 15 || (bitrate == 0 && possible_free_framelen == -1))
1371 return 0;
1372
1373 /* ignore error correction, too */
1374 header >>= 5;
1375
1376 /* layer */
1377 layer = 4 - (header & 0x3);
1378 if (layer == 4)
1379 return 0;
1380 header >>= 2;
1381
1382 /* version 0=MPEG2.5; 2=MPEG2; 3=MPEG1 */
1383 version = header & 0x3;
1384 if (version == 1)
1385 return 0;
1386
1387 /* lookup */
1388 channels = (mode == 3) ? 1 : 2;
1389 samplerate = mp3types_freqs[version > 0 ? version - 1 : 0][samplerate];
1390 if (bitrate == 0) {
1391 /* possible freeform mp3 */
1392 if (layer == 1) {
1393 length *= 4;
1394 length += possible_free_framelen;
1395 bitrate = length * samplerate / 48000;
1396 } else {
1397 length += possible_free_framelen;
1398 bitrate = length * samplerate /
1399 ((layer == 3 && version != 3) ? 72000 : 144000);
1400 }
1401 /* freeform mp3 should have a higher-than-usually-allowed bitrate */
1402 GST_LOG ("calculated bitrate: %u, max usually: %u", bitrate,
1403 mp3types_bitrates[version == 3 ? 0 : 1][layer - 1][14]);
1404 if (bitrate < mp3types_bitrates[version == 3 ? 0 : 1][layer - 1][14])
1405 return 0;
1406 } else {
1407 /* calculating */
1408 bitrate = mp3types_bitrates[version == 3 ? 0 : 1][layer - 1][bitrate];
1409 if (layer == 1) {
1410 length = ((12000 * bitrate / samplerate) + length) * 4;
1411 } else {
1412 length += ((layer == 3
1413 && version != 3) ? 72000 : 144000) * bitrate / samplerate;
1414 }
1415 }
1416
1417 GST_LOG ("mp3typefind: calculated mp3 frame length of %u bytes", length);
1418 GST_LOG
1419 ("mp3typefind: samplerate = %u - bitrate = %u - layer = %u - version = %u"
1420 " - channels = %u", samplerate, bitrate, layer, version, channels);
1421
1422 if (put_layer)
1423 *put_layer = layer;
1424 if (put_channels)
1425 *put_channels = channels;
1426 if (put_bitrate)
1427 *put_bitrate = bitrate;
1428 if (put_samplerate)
1429 *put_samplerate = samplerate;
1430
1431 return length;
1432 }
1433
1434
1435 static GstStaticCaps mp3_caps = GST_STATIC_CAPS ("audio/mpeg, "
1436 "mpegversion = (int) 1, layer = (int) [ 1, 3 ]");
1437 #define MP3_CAPS (gst_static_caps_get(&mp3_caps))
1438 /*
1439 * random values for typefinding
1440 * if no more data is available, we will return a probability of
1441 * (found_headers/TRY_HEADERS) * (MAXIMUM * (TRY_SYNC - bytes_skipped)
1442 * / TRY_SYNC)
1443 * if found_headers >= MIN_HEADERS
1444 */
1445 #define GST_MP3_TYPEFIND_MIN_HEADERS (2)
1446 #define GST_MP3_TYPEFIND_TRY_HEADERS (5)
1447 #define GST_MP3_TYPEFIND_TRY_SYNC (GST_TYPE_FIND_MAXIMUM * 100) /* 10kB */
1448 #define GST_MP3_TYPEFIND_SYNC_SIZE (2048)
1449 #define GST_MP3_WRONG_HEADER (10)
1450
1451 static void
mp3_type_find_at_offset(GstTypeFind * tf,guint64 start_off,guint * found_layer,GstTypeFindProbability * found_prob)1452 mp3_type_find_at_offset (GstTypeFind * tf, guint64 start_off,
1453 guint * found_layer, GstTypeFindProbability * found_prob)
1454 {
1455 const guint8 *data = NULL;
1456 const guint8 *data_end = NULL;
1457 guint size;
1458 guint64 skipped;
1459 gint last_free_offset = -1;
1460 gint last_free_framelen = -1;
1461 gboolean headerstart = TRUE;
1462
1463 *found_layer = 0;
1464 *found_prob = 0;
1465
1466 size = 0;
1467 skipped = 0;
1468 while (skipped < GST_MP3_TYPEFIND_TRY_SYNC) {
1469 if (size <= 0) {
1470 size = GST_MP3_TYPEFIND_SYNC_SIZE * 2;
1471 do {
1472 size /= 2;
1473 data = gst_type_find_peek (tf, skipped + start_off, size);
1474 } while (size > 10 && !data);
1475 if (!data)
1476 break;
1477 data_end = data + size;
1478 }
1479 if (*data == 0xFF) {
1480 const guint8 *head_data = NULL;
1481 guint layer = 0, bitrate, samplerate, channels;
1482 guint found = 0; /* number of valid headers found */
1483 guint64 offset = skipped;
1484 gboolean changed = FALSE;
1485 guint prev_layer = 0;
1486 guint prev_channels = 0, prev_samplerate = 0;
1487
1488 while (found < GST_MP3_TYPEFIND_TRY_HEADERS) {
1489 guint32 head;
1490 guint length;
1491 gboolean free = FALSE;
1492
1493 if ((gint64) (offset - skipped + 4) >= 0 &&
1494 data + offset - skipped + 4 < data_end) {
1495 head_data = data + offset - skipped;
1496 } else {
1497 head_data = gst_type_find_peek (tf, offset + start_off, 4);
1498 }
1499 if (!head_data)
1500 break;
1501 head = GST_READ_UINT32_BE (head_data);
1502 if (!(length = mp3_type_frame_length_from_header (head, &layer,
1503 &channels, &bitrate, &samplerate, &free,
1504 last_free_framelen))) {
1505 if (free) {
1506 if (last_free_offset == -1)
1507 last_free_offset = offset;
1508 else {
1509 last_free_framelen = offset - last_free_offset;
1510 offset = last_free_offset;
1511 continue;
1512 }
1513 } else {
1514 last_free_framelen = -1;
1515 }
1516
1517 /* Mark the fact that we didn't find a valid header at the beginning */
1518 if (found == 0)
1519 headerstart = FALSE;
1520
1521 GST_LOG ("%d. header at offset %" G_GUINT64_FORMAT
1522 " (0x%" G_GINT64_MODIFIER "x) was not an mp3 header "
1523 "(possibly-free: %s)", found + 1, start_off + offset,
1524 start_off + offset, free ? "yes" : "no");
1525 break;
1526 }
1527 if ((prev_layer && prev_layer != layer) ||
1528 /* (prev_bitrate && prev_bitrate != bitrate) || <-- VBR */
1529 (prev_samplerate && prev_samplerate != samplerate) ||
1530 (prev_channels && prev_channels != channels)) {
1531 /* this means an invalid property, or a change, which might mean
1532 * that this is not a mp3 but just a random bytestream. It could
1533 * be a freaking funky encoded mp3 though. We'll just not count
1534 * this header*/
1535 if (prev_layer)
1536 changed = TRUE;
1537 } else {
1538 found++;
1539 GST_LOG ("found %d. header at offset %" G_GUINT64_FORMAT " (0x%"
1540 G_GINT64_MODIFIER "X)", found, start_off + offset,
1541 start_off + offset);
1542 }
1543 prev_layer = layer;
1544 prev_channels = channels;
1545 prev_samplerate = samplerate;
1546
1547 offset += length;
1548 }
1549 g_assert (found <= GST_MP3_TYPEFIND_TRY_HEADERS);
1550 if (found != 0 && head_data == NULL &&
1551 gst_type_find_peek (tf, offset + start_off - 1, 1) == NULL)
1552 /* Incomplete last frame - don't count it. */
1553 found--;
1554 if (found == GST_MP3_TYPEFIND_TRY_HEADERS ||
1555 (found >= GST_MP3_TYPEFIND_MIN_HEADERS && head_data == NULL)) {
1556 /* we can make a valid guess */
1557 guint probability = found * GST_TYPE_FIND_MAXIMUM *
1558 (GST_MP3_TYPEFIND_TRY_SYNC - skipped) /
1559 GST_MP3_TYPEFIND_TRY_HEADERS / GST_MP3_TYPEFIND_TRY_SYNC;
1560
1561 if (!headerstart
1562 && probability > (GST_TYPE_FIND_MINIMUM + GST_MP3_WRONG_HEADER))
1563 probability -= GST_MP3_WRONG_HEADER;
1564 if (probability < GST_TYPE_FIND_MINIMUM)
1565 probability = GST_TYPE_FIND_MINIMUM;
1566 if (start_off > 0)
1567 probability /= 2;
1568 if (!changed)
1569 probability = (probability + GST_TYPE_FIND_MAXIMUM) / 2;
1570
1571 GST_INFO
1572 ("audio/mpeg calculated %u = %u * %u / %u * (%u - %"
1573 G_GUINT64_FORMAT ") / %u", probability, GST_TYPE_FIND_MAXIMUM,
1574 found, GST_MP3_TYPEFIND_TRY_HEADERS, GST_MP3_TYPEFIND_TRY_SYNC,
1575 (guint64) skipped, GST_MP3_TYPEFIND_TRY_SYNC);
1576 /* make sure we're not id3 tagged */
1577 head_data = gst_type_find_peek (tf, -128, 3);
1578 if (head_data && (memcmp (head_data, "TAG", 3) == 0)) {
1579 probability = 0;
1580 }
1581 g_assert (probability <= GST_TYPE_FIND_MAXIMUM);
1582
1583 *found_prob = probability;
1584 if (probability > 0)
1585 *found_layer = layer;
1586 return;
1587 }
1588 }
1589 data++;
1590 skipped++;
1591 size--;
1592 }
1593 }
1594
1595 static void
mp3_type_find(GstTypeFind * tf,gpointer unused)1596 mp3_type_find (GstTypeFind * tf, gpointer unused)
1597 {
1598 /* ohos.opt.compat.0001: The demux of gstplayer does not accurately parse audio resources in the MP3 format.
1599 * As a result, the duration value cannot be obtained in the preparation phase.
1600 * Use the demux and typefind of ffmpeg to process audio resources in MP3 format.
1601 */
1602 #ifdef OHOS_OPT_COMPAT
1603 return;
1604 #endif
1605 GstTypeFindProbability prob, mid_prob;
1606 const guint8 *data;
1607 guint layer, mid_layer;
1608 guint64 length;
1609
1610 mp3_type_find_at_offset (tf, 0, &layer, &prob);
1611 length = gst_type_find_get_length (tf);
1612
1613 if (length == 0 || length == (guint64) - 1) {
1614 if (prob != 0)
1615 goto suggest;
1616 return;
1617 }
1618
1619 /* if we're pretty certain already, skip the additional check */
1620 if (prob >= GST_TYPE_FIND_LIKELY)
1621 goto suggest;
1622
1623 mp3_type_find_at_offset (tf, length / 2, &mid_layer, &mid_prob);
1624
1625 if (mid_prob > 0) {
1626 if (prob == 0) {
1627 GST_LOG ("detected audio/mpeg only in the middle (p=%u)", mid_prob);
1628 layer = mid_layer;
1629 prob = mid_prob;
1630 goto suggest;
1631 }
1632
1633 if (layer != mid_layer) {
1634 GST_WARNING ("audio/mpeg layer discrepancy: %u vs. %u", layer, mid_layer);
1635 return; /* FIXME: or should we just go with the one in the middle? */
1636 }
1637
1638 /* detected mpeg audio both in middle of the file and at the start */
1639 prob = (prob + mid_prob) / 2;
1640 goto suggest;
1641 }
1642
1643 /* a valid header right at the start makes it more likely
1644 * that this is actually plain mpeg-1 audio */
1645 if (prob > 0) {
1646 data = gst_type_find_peek (tf, 0, 4); /* use min. frame size? */
1647 if (data && mp3_type_frame_length_from_header (GST_READ_UINT32_BE (data),
1648 &layer, NULL, NULL, NULL, NULL, 0) != 0) {
1649 prob = MIN (prob + 10, GST_TYPE_FIND_MAXIMUM);
1650 }
1651 }
1652
1653 if (prob > 0)
1654 goto suggest;
1655
1656 return;
1657
1658 suggest:
1659 {
1660 g_return_if_fail (layer >= 1 && layer <= 3);
1661
1662 gst_type_find_suggest_simple (tf, prob, "audio/mpeg",
1663 "mpegversion", G_TYPE_INT, 1, "layer", G_TYPE_INT, layer,
1664 "parsed", G_TYPE_BOOLEAN, FALSE, NULL);
1665 }
1666 }
1667
1668 /*** audio/x-musepack ***/
1669
1670 static GstStaticCaps musepack_caps =
1671 GST_STATIC_CAPS ("audio/x-musepack, streamversion= (int) { 7, 8 }");
1672
1673 #define MUSEPACK_CAPS (gst_static_caps_get(&musepack_caps))
1674 static void
musepack_type_find(GstTypeFind * tf,gpointer unused)1675 musepack_type_find (GstTypeFind * tf, gpointer unused)
1676 {
1677 const guint8 *data = gst_type_find_peek (tf, 0, 4);
1678 GstTypeFindProbability prop = GST_TYPE_FIND_MINIMUM;
1679 gint streamversion = -1;
1680
1681 if (data && memcmp (data, "MP+", 3) == 0) {
1682 streamversion = 7;
1683 if ((data[3] & 0x7f) == 7) {
1684 prop = GST_TYPE_FIND_MAXIMUM;
1685 } else {
1686 prop = GST_TYPE_FIND_LIKELY + 10;
1687 }
1688 } else if (data && memcmp (data, "MPCK", 4) == 0) {
1689 streamversion = 8;
1690 prop = GST_TYPE_FIND_MAXIMUM;
1691 }
1692
1693 if (streamversion != -1) {
1694 gst_type_find_suggest_simple (tf, prop, "audio/x-musepack",
1695 "streamversion", G_TYPE_INT, streamversion, NULL);
1696 }
1697 }
1698
1699 /*** audio/x-ac3 ***/
1700 /* FIXME 0.11: should be audio/ac3, but isn't for backwards compatibility */
1701 static GstStaticCaps ac3_caps = GST_STATIC_CAPS ("audio/x-ac3");
1702
1703 #define AC3_CAPS (gst_static_caps_get(&ac3_caps))
1704
1705 static GstStaticCaps eac3_caps = GST_STATIC_CAPS ("audio/x-eac3");
1706
1707 #define EAC3_CAPS (gst_static_caps_get(&eac3_caps))
1708
1709 struct ac3_frmsize
1710 {
1711 unsigned short bit_rate;
1712 unsigned short frm_size[3];
1713 };
1714
1715 static const struct ac3_frmsize ac3_frmsizecod_tbl[] = {
1716 {32, {64, 69, 96}},
1717 {32, {64, 70, 96}},
1718 {40, {80, 87, 120}},
1719 {40, {80, 88, 120}},
1720 {48, {96, 104, 144}},
1721 {48, {96, 105, 144}},
1722 {56, {112, 121, 168}},
1723 {56, {112, 122, 168}},
1724 {64, {128, 139, 192}},
1725 {64, {128, 140, 192}},
1726 {80, {160, 174, 240}},
1727 {80, {160, 175, 240}},
1728 {96, {192, 208, 288}},
1729 {96, {192, 209, 288}},
1730 {112, {224, 243, 336}},
1731 {112, {224, 244, 336}},
1732 {128, {256, 278, 384}},
1733 {128, {256, 279, 384}},
1734 {160, {320, 348, 480}},
1735 {160, {320, 349, 480}},
1736 {192, {384, 417, 576}},
1737 {192, {384, 418, 576}},
1738 {224, {448, 487, 672}},
1739 {224, {448, 488, 672}},
1740 {256, {512, 557, 768}},
1741 {256, {512, 558, 768}},
1742 {320, {640, 696, 960}},
1743 {320, {640, 697, 960}},
1744 {384, {768, 835, 1152}},
1745 {384, {768, 836, 1152}},
1746 {448, {896, 975, 1344}},
1747 {448, {896, 976, 1344}},
1748 {512, {1024, 1114, 1536}},
1749 {512, {1024, 1115, 1536}},
1750 {576, {1152, 1253, 1728}},
1751 {576, {1152, 1254, 1728}},
1752 {640, {1280, 1393, 1920}},
1753 {640, {1280, 1394, 1920}}
1754 };
1755
1756 static void
ac3_type_find(GstTypeFind * tf,gpointer unused)1757 ac3_type_find (GstTypeFind * tf, gpointer unused)
1758 {
1759 DataScanCtx c = { 0, NULL, 0 };
1760
1761 /* Search for an ac3 frame; not necessarily right at the start, but give it
1762 * a lower probability if not found right at the start. Check that the
1763 * frame is followed by a second frame at the expected offset.
1764 * We could also check the two ac3 CRCs, but we don't do that right now */
1765 while (c.offset < 1024) {
1766 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 6)))
1767 break;
1768
1769 if (c.data[0] == 0x0b && c.data[1] == 0x77) {
1770 guint bsid = c.data[5] >> 3;
1771
1772 if (bsid <= 8) {
1773 /* ac3 */
1774 guint fscod = c.data[4] >> 6;
1775 guint frmsizecod = c.data[4] & 0x3f;
1776
1777 if (fscod < 3 && frmsizecod < 38) {
1778 DataScanCtx c_next = c;
1779 guint frame_size;
1780
1781 frame_size = ac3_frmsizecod_tbl[frmsizecod].frm_size[fscod];
1782 GST_LOG ("possible AC3 frame sync at offset %"
1783 G_GUINT64_FORMAT ", size=%u", c.offset, frame_size);
1784 if (data_scan_ctx_ensure_data (tf, &c_next, (frame_size * 2) + 5)) {
1785 data_scan_ctx_advance (tf, &c_next, frame_size * 2);
1786
1787 if (c_next.data[0] == 0x0b && c_next.data[1] == 0x77) {
1788 fscod = c_next.data[4] >> 6;
1789 frmsizecod = c_next.data[4] & 0x3f;
1790
1791 if (fscod < 3 && frmsizecod < 38) {
1792 GstTypeFindProbability prob;
1793
1794 GST_LOG ("found second AC3 frame (size=%u), looks good",
1795 ac3_frmsizecod_tbl[frmsizecod].frm_size[fscod]);
1796 if (c.offset == 0)
1797 prob = GST_TYPE_FIND_MAXIMUM;
1798 else
1799 prob = GST_TYPE_FIND_NEARLY_CERTAIN;
1800
1801 gst_type_find_suggest (tf, prob, AC3_CAPS);
1802 return;
1803 }
1804 } else {
1805 GST_LOG ("no second AC3 frame found, false sync");
1806 }
1807 }
1808 }
1809 } else if (bsid <= 16 && bsid > 10) {
1810 /* eac3 */
1811 DataScanCtx c_next = c;
1812 guint frame_size;
1813
1814 frame_size = (((c.data[2] & 0x07) << 8) + c.data[3]) + 1;
1815 GST_LOG ("possible E-AC3 frame sync at offset %"
1816 G_GUINT64_FORMAT ", size=%u", c.offset, frame_size);
1817 if (data_scan_ctx_ensure_data (tf, &c_next, (frame_size * 2) + 5)) {
1818 data_scan_ctx_advance (tf, &c_next, frame_size * 2);
1819
1820 if (c_next.data[0] == 0x0b && c_next.data[1] == 0x77) {
1821 GstTypeFindProbability prob;
1822
1823 GST_LOG ("found second E-AC3 frame, looks good");
1824 if (c.offset == 0)
1825 prob = GST_TYPE_FIND_MAXIMUM;
1826 else
1827 prob = GST_TYPE_FIND_NEARLY_CERTAIN;
1828
1829 gst_type_find_suggest (tf, prob, EAC3_CAPS);
1830 return;
1831 } else {
1832 GST_LOG ("no second E-AC3 frame found, false sync");
1833 }
1834 }
1835 } else {
1836 GST_LOG ("invalid AC3 BSID: %u", bsid);
1837 }
1838 }
1839 data_scan_ctx_advance (tf, &c, 1);
1840 }
1841 }
1842
1843 /*** audio/x-dts ***/
1844 static GstStaticCaps dts_caps = GST_STATIC_CAPS ("audio/x-dts");
1845 #define DTS_CAPS (gst_static_caps_get (&dts_caps))
1846 #define DTS_MIN_FRAMESIZE 96
1847 #define DTS_MAX_FRAMESIZE 18725 /* 16384*16/14 */
1848
1849 static gboolean
dts_parse_frame_header(DataScanCtx * c,guint * frame_size,guint * sample_rate,guint * channels,guint * depth,guint * endianness)1850 dts_parse_frame_header (DataScanCtx * c, guint * frame_size,
1851 guint * sample_rate, guint * channels, guint * depth, guint * endianness)
1852 {
1853 static const int sample_rates[16] = { 0, 8000, 16000, 32000, 0, 0, 11025,
1854 22050, 44100, 0, 0, 12000, 24000, 48000, 96000, 192000
1855 };
1856 static const guint8 channels_table[16] = { 1, 2, 2, 2, 2, 3, 3, 4, 4, 5,
1857 6, 6, 6, 7, 8, 8
1858 };
1859 guint16 hdr[8];
1860 guint32 marker;
1861 guint num_blocks, chans, lfe, i;
1862
1863 marker = GST_READ_UINT32_BE (c->data);
1864
1865 /* raw big endian or 14-bit big endian */
1866 if (marker == 0x7FFE8001 || marker == 0x1FFFE800) {
1867 *endianness = G_BIG_ENDIAN;
1868 for (i = 0; i < G_N_ELEMENTS (hdr); ++i)
1869 hdr[i] = GST_READ_UINT16_BE (c->data + (i * sizeof (guint16)));
1870 } else
1871 /* raw little endian or 14-bit little endian */
1872 if (marker == 0xFE7F0180 || marker == 0xFF1F00E8) {
1873 *endianness = G_LITTLE_ENDIAN;
1874 for (i = 0; i < G_N_ELEMENTS (hdr); ++i)
1875 hdr[i] = GST_READ_UINT16_LE (c->data + (i * sizeof (guint16)));
1876 } else {
1877 return FALSE;
1878 }
1879
1880 GST_LOG ("dts sync marker 0x%08x at offset %u", marker, (guint) c->offset);
1881
1882 /* 14-bit mode */
1883 if (marker == 0x1FFFE800 || marker == 0xFF1F00E8) {
1884 if ((hdr[2] & 0xFFF0) != 0x07F0)
1885 return FALSE;
1886 /* discard top 2 bits (2 void), shift in 2 */
1887 hdr[0] = (hdr[0] << 2) | ((hdr[1] >> 12) & 0x0003);
1888 /* discard top 4 bits (2 void, 2 shifted into hdr[0]), shift in 4 etc. */
1889 hdr[1] = (hdr[1] << 4) | ((hdr[2] >> 10) & 0x000F);
1890 hdr[2] = (hdr[2] << 6) | ((hdr[3] >> 8) & 0x003F);
1891 hdr[3] = (hdr[3] << 8) | ((hdr[4] >> 6) & 0x00FF);
1892 hdr[4] = (hdr[4] << 10) | ((hdr[5] >> 4) & 0x03FF);
1893 hdr[5] = (hdr[5] << 12) | ((hdr[6] >> 2) & 0x0FFF);
1894 hdr[6] = (hdr[6] << 14) | ((hdr[7] >> 0) & 0x3FFF);
1895 g_assert (hdr[0] == 0x7FFE && hdr[1] == 0x8001);
1896 *depth = 14;
1897 } else {
1898 *depth = 16;
1899 }
1900
1901 GST_LOG ("frame header: %04x%04x%04x%04x", hdr[2], hdr[3], hdr[4], hdr[5]);
1902
1903 num_blocks = (hdr[2] >> 2) & 0x7F;
1904 *frame_size = (((hdr[2] & 0x03) << 12) | (hdr[3] >> 4)) + 1;
1905 chans = ((hdr[3] & 0x0F) << 2) | (hdr[4] >> 14);
1906 *sample_rate = sample_rates[(hdr[4] >> 10) & 0x0F];
1907 lfe = (hdr[5] >> 9) & 0x03;
1908
1909 if (num_blocks < 5 || *frame_size < 96 || *sample_rate == 0)
1910 return FALSE;
1911
1912 if (marker == 0x1FFFE800 || marker == 0xFF1F00E8)
1913 *frame_size = (*frame_size * 16) / 14; /* FIXME: round up? */
1914
1915 if (chans < G_N_ELEMENTS (channels_table))
1916 *channels = channels_table[chans] + ((lfe) ? 1 : 0);
1917 else
1918 *channels = 0;
1919
1920 return TRUE;
1921 }
1922
1923 static void
dts_type_find(GstTypeFind * tf,gpointer unused)1924 dts_type_find (GstTypeFind * tf, gpointer unused)
1925 {
1926 DataScanCtx c = { 0, NULL, 0 };
1927
1928 /* Search for an dts frame; not necessarily right at the start, but give it
1929 * a lower probability if not found right at the start. Check that the
1930 * frame is followed by a second frame at the expected offset. */
1931 while (c.offset <= DTS_MAX_FRAMESIZE) {
1932 guint frame_size = 0, rate = 0, chans = 0, depth = 0, endianness = 0;
1933
1934 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, DTS_MIN_FRAMESIZE)))
1935 return;
1936
1937 if (G_UNLIKELY (dts_parse_frame_header (&c, &frame_size, &rate, &chans,
1938 &depth, &endianness))) {
1939 GstTypeFindProbability prob;
1940 DataScanCtx next_c;
1941
1942 prob = (c.offset == 0) ? GST_TYPE_FIND_LIKELY : GST_TYPE_FIND_POSSIBLE;
1943
1944 /* check for second frame sync */
1945 next_c = c;
1946 data_scan_ctx_advance (tf, &next_c, frame_size);
1947 if (data_scan_ctx_ensure_data (tf, &next_c, 4)) {
1948 GST_LOG ("frame size: %u 0x%04x", frame_size, frame_size);
1949 GST_MEMDUMP ("second frame sync", next_c.data, 4);
1950 if (GST_READ_UINT32_BE (c.data) == GST_READ_UINT32_BE (next_c.data))
1951 prob = GST_TYPE_FIND_MAXIMUM;
1952 }
1953
1954 if (chans > 0) {
1955 gst_type_find_suggest_simple (tf, prob, "audio/x-dts",
1956 "rate", G_TYPE_INT, rate, "channels", G_TYPE_INT, chans,
1957 "depth", G_TYPE_INT, depth, "endianness", G_TYPE_INT, endianness,
1958 "framed", G_TYPE_BOOLEAN, FALSE, NULL);
1959 } else {
1960 gst_type_find_suggest_simple (tf, prob, "audio/x-dts",
1961 "rate", G_TYPE_INT, rate, "depth", G_TYPE_INT, depth,
1962 "endianness", G_TYPE_INT, endianness,
1963 "framed", G_TYPE_BOOLEAN, FALSE, NULL);
1964 }
1965
1966 return;
1967 }
1968
1969 data_scan_ctx_advance (tf, &c, 1);
1970 }
1971 }
1972
1973 /*** gsm ***/
1974
1975 /* can only be detected by using the extension, in which case we use the default
1976 * GSM properties */
1977 static GstStaticCaps gsm_caps =
1978 GST_STATIC_CAPS ("audio/x-gsm, rate=8000, channels=1");
1979
1980 #define GSM_CAPS (gst_static_caps_get(&gsm_caps))
1981
1982 /*** wavpack ***/
1983
1984 static GstStaticCaps wavpack_caps =
1985 GST_STATIC_CAPS ("audio/x-wavpack, framed = (boolean) false");
1986
1987 #define WAVPACK_CAPS (gst_static_caps_get(&wavpack_caps))
1988
1989 static GstStaticCaps wavpack_correction_caps =
1990 GST_STATIC_CAPS ("audio/x-wavpack-correction, framed = (boolean) false");
1991
1992 #define WAVPACK_CORRECTION_CAPS (gst_static_caps_get(&wavpack_correction_caps))
1993
1994 static void
wavpack_type_find(GstTypeFind * tf,gpointer unused)1995 wavpack_type_find (GstTypeFind * tf, gpointer unused)
1996 {
1997 GstTypeFindProbability base_prob = GST_TYPE_FIND_POSSIBLE;
1998 guint64 offset;
1999 guint32 blocksize;
2000 const guint8 *data;
2001 guint count_wv, count_wvc;
2002
2003 data = gst_type_find_peek (tf, 0, 32);
2004 if (!data)
2005 return;
2006
2007 if (data[0] != 'w' || data[1] != 'v' || data[2] != 'p' || data[3] != 'k')
2008 return;
2009
2010 /* Note: wavpack blocks can be fairly large (easily 60-110k), possibly
2011 * larger than the max. limits imposed by certain typefinding elements
2012 * like id3demux or apedemux, so typefinding is most likely only going to
2013 * work in pull-mode */
2014 blocksize = GST_READ_UINT32_LE (data + 4);
2015 GST_LOG ("wavpack header, blocksize=0x%04x", blocksize);
2016 /* If bigger than maximum allowed blocksize, refuse */
2017 if (blocksize > 131072)
2018 return;
2019 count_wv = 0;
2020 count_wvc = 0;
2021 offset = 32;
2022 while (offset < 8 + blocksize) {
2023 guint32 sublen;
2024
2025 /* get chunk header */
2026 GST_LOG ("peeking at chunk at offset 0x%04x", (guint) offset);
2027 data = gst_type_find_peek (tf, offset, 4);
2028 if (data == NULL)
2029 break;
2030 sublen = ((guint32) data[1]) << 1;
2031 if (data[0] & 0x80) {
2032 sublen |= (((guint32) data[2]) << 9) | (((guint32) data[3]) << 17);
2033 sublen += 1 + 3; /* id + length */
2034 } else {
2035 sublen += 1 + 1; /* id + length */
2036 }
2037 if (offset + sublen > 8 + blocksize) {
2038 GST_LOG ("chunk length too big (%u > %" G_GUINT64_FORMAT ")", sublen,
2039 blocksize - offset);
2040 break;
2041 }
2042 if ((data[0] & 0x20) == 0) {
2043 switch (data[0] & 0x0f) {
2044 case 0xa: /* ID_WV_BITSTREAM */
2045 case 0xc: /* ID_WVX_BITSTREAM */
2046 ++count_wv;
2047 break;
2048 case 0xb: /* ID_WVC_BITSTREAM */
2049 ++count_wvc;
2050 break;
2051 default:
2052 break;
2053 }
2054 if (count_wv >= 5 || count_wvc >= 5)
2055 break;
2056 }
2057 offset += sublen;
2058 }
2059
2060 /* check for second block header */
2061 data = gst_type_find_peek (tf, 8 + blocksize, 4);
2062 if (data != NULL && memcmp (data, "wvpk", 4) == 0) {
2063 GST_DEBUG ("found second block sync");
2064 base_prob = GST_TYPE_FIND_LIKELY;
2065 }
2066
2067 GST_DEBUG ("wvc=%d, wv=%d", count_wvc, count_wv);
2068
2069 if (count_wvc > 0 && count_wvc > count_wv) {
2070 gst_type_find_suggest (tf,
2071 MIN (base_prob + 5 * count_wvc, GST_TYPE_FIND_NEARLY_CERTAIN),
2072 WAVPACK_CORRECTION_CAPS);
2073 } else if (count_wv > 0) {
2074 gst_type_find_suggest (tf,
2075 MIN (base_prob + 5 * count_wv, GST_TYPE_FIND_NEARLY_CERTAIN),
2076 WAVPACK_CAPS);
2077 }
2078 }
2079
2080 /*** application/postscrip ***/
2081 static GstStaticCaps postscript_caps =
2082 GST_STATIC_CAPS ("application/postscript");
2083
2084 #define POSTSCRIPT_CAPS (gst_static_caps_get(&postscript_caps))
2085
2086 static void
postscript_type_find(GstTypeFind * tf,gpointer unused)2087 postscript_type_find (GstTypeFind * tf, gpointer unused)
2088 {
2089 const guint8 *data = gst_type_find_peek (tf, 0, 3);
2090 if (!data)
2091 return;
2092
2093 if (data[0] == 0x04)
2094 data++;
2095 if (data[0] == '%' && data[1] == '!')
2096 gst_type_find_suggest (tf, GST_TYPE_FIND_POSSIBLE, POSTSCRIPT_CAPS);
2097
2098 }
2099
2100 /*** image/svg+xml ***/
2101 static GstStaticCaps svg_caps = GST_STATIC_CAPS ("image/svg+xml");
2102
2103 #define SVG_CAPS (gst_static_caps_get(&svg_caps))
2104
2105 static void
svg_type_find(GstTypeFind * tf,gpointer unused)2106 svg_type_find (GstTypeFind * tf, gpointer unused)
2107 {
2108 static const gchar svg_doctype[] = "!DOCTYPE svg";
2109 static const gchar svg_tag[] = "<svg";
2110 DataScanCtx c = { 0, NULL, 0 };
2111
2112 while (c.offset <= 1024) {
2113 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 12)))
2114 break;
2115
2116 if (memcmp (svg_doctype, c.data, 12) == 0) {
2117 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, SVG_CAPS);
2118 return;
2119 } else if (memcmp (svg_tag, c.data, 4) == 0) {
2120 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY, SVG_CAPS);
2121 return;
2122 }
2123 data_scan_ctx_advance (tf, &c, 1);
2124 }
2125 }
2126
2127 /*** multipart/x-mixed-replace mimestream ***/
2128
2129 static GstStaticCaps multipart_caps =
2130 GST_STATIC_CAPS ("multipart/x-mixed-replace");
2131 #define MULTIPART_CAPS gst_static_caps_get(&multipart_caps)
2132
2133 /* multipart/x-mixed replace is:
2134 * <maybe some whitespace>--<some ascii chars>[\r]\n
2135 * <more ascii chars>[\r]\nContent-type:<more ascii>[\r]\n */
2136 static void
multipart_type_find(GstTypeFind * tf,gpointer unused)2137 multipart_type_find (GstTypeFind * tf, gpointer unused)
2138 {
2139 const guint8 *data;
2140 const guint8 *x;
2141
2142 #define MULTIPART_MAX_BOUNDARY_OFFSET 16
2143 data = gst_type_find_peek (tf, 0, MULTIPART_MAX_BOUNDARY_OFFSET);
2144 if (!data)
2145 return;
2146
2147 for (x = data;
2148 x - data < MULTIPART_MAX_BOUNDARY_OFFSET - 2 && g_ascii_isspace (*x);
2149 x++);
2150 if (x[0] != '-' || x[1] != '-')
2151 return;
2152
2153 /* Could be okay, peek what should be enough for a complete header */
2154 #define MULTIPART_MAX_HEADER_SIZE 256
2155 data = gst_type_find_peek (tf, 0, MULTIPART_MAX_HEADER_SIZE);
2156 if (!data)
2157 return;
2158
2159 for (x = data; x - data < MULTIPART_MAX_HEADER_SIZE - 14; x++) {
2160 if (!isascii (*x)) {
2161 return;
2162 }
2163 if (*x == '\n' &&
2164 !g_ascii_strncasecmp ("content-type:", (gchar *) x + 1, 13)) {
2165 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MULTIPART_CAPS);
2166 return;
2167 }
2168 }
2169 }
2170
2171 /*** video/mpeg systemstream ***/
2172 static GstStaticCaps mpeg_sys_caps = GST_STATIC_CAPS ("video/mpeg, "
2173 "systemstream = (boolean) true, mpegversion = (int) [ 1, 2 ]");
2174
2175 #define MPEG_SYS_CAPS gst_static_caps_get(&mpeg_sys_caps)
2176 #define IS_MPEG_HEADER(data) (G_UNLIKELY((((guint8 *)(data))[0] == 0x00) && \
2177 (((guint8 *)(data))[1] == 0x00) && \
2178 (((guint8 *)(data))[2] == 0x01)))
2179
2180 #define IS_MPEG_PACK_CODE(b) ((b) == 0xBA)
2181 #define IS_MPEG_SYS_CODE(b) ((b) == 0xBB)
2182 #define IS_MPEG_PACK_HEADER(data) (IS_MPEG_HEADER (data) && \
2183 IS_MPEG_PACK_CODE (((guint8 *)(data))[3]))
2184
2185 #define IS_MPEG_PES_CODE(b) (((b) & 0xF0) == 0xE0 || ((b) & 0xF0) == 0xC0 || \
2186 (b) >= 0xBD)
2187 #define IS_MPEG_PES_HEADER(data) (IS_MPEG_HEADER (data) && \
2188 IS_MPEG_PES_CODE (((guint8 *)(data))[3]))
2189
2190 #define MPEG2_MAX_PROBE_LENGTH (128 * 1024) /* 128kB should be 64 packs of the
2191 * most common 2kB pack size. */
2192
2193 #define MPEG2_MIN_SYS_HEADERS 2
2194 #define MPEG2_MAX_SYS_HEADERS 5
2195
2196 static gboolean
mpeg_sys_is_valid_pack(GstTypeFind * tf,const guint8 * data,guint len,guint * pack_size)2197 mpeg_sys_is_valid_pack (GstTypeFind * tf, const guint8 * data, guint len,
2198 guint * pack_size)
2199 {
2200 /* Check the pack header @ offset for validity, assuming that the 4 byte header
2201 * itself has already been checked. */
2202 guint8 stuff_len;
2203
2204 if (len < 12)
2205 return FALSE;
2206
2207 /* Check marker bits */
2208 if ((data[4] & 0xC4) == 0x44) {
2209 /* MPEG-2 PACK */
2210 if (len < 14)
2211 return FALSE;
2212
2213 if ((data[6] & 0x04) != 0x04 ||
2214 (data[8] & 0x04) != 0x04 ||
2215 (data[9] & 0x01) != 0x01 || (data[12] & 0x03) != 0x03)
2216 return FALSE;
2217
2218 stuff_len = data[13] & 0x07;
2219
2220 /* Check the following header bytes, if we can */
2221 if ((14 + stuff_len + 4) <= len) {
2222 if (!IS_MPEG_HEADER (data + 14 + stuff_len))
2223 return FALSE;
2224 }
2225 if (pack_size)
2226 *pack_size = 14 + stuff_len;
2227 return TRUE;
2228 } else if ((data[4] & 0xF1) == 0x21) {
2229 /* MPEG-1 PACK */
2230 if ((data[6] & 0x01) != 0x01 ||
2231 (data[8] & 0x01) != 0x01 ||
2232 (data[9] & 0x80) != 0x80 || (data[11] & 0x01) != 0x01)
2233 return FALSE;
2234
2235 /* Check the following header bytes, if we can */
2236 if ((12 + 4) <= len) {
2237 if (!IS_MPEG_HEADER (data + 12))
2238 return FALSE;
2239 }
2240 if (pack_size)
2241 *pack_size = 12;
2242 return TRUE;
2243 }
2244
2245 return FALSE;
2246 }
2247
2248 static gboolean
mpeg_sys_is_valid_pes(GstTypeFind * tf,const guint8 * data,guint len,guint * pack_size)2249 mpeg_sys_is_valid_pes (GstTypeFind * tf, const guint8 * data, guint len,
2250 guint * pack_size)
2251 {
2252 guint pes_packet_len;
2253
2254 /* Check the PES header at the given position, assuming the header code itself
2255 * was already checked */
2256 if (len < 6)
2257 return FALSE;
2258
2259 /* For MPEG Program streams, unbounded PES is not allowed, so we must have a
2260 * valid length present */
2261 pes_packet_len = GST_READ_UINT16_BE (data + 4);
2262 if (pes_packet_len == 0)
2263 return FALSE;
2264
2265 /* Check the following header, if we can */
2266 if (6 + pes_packet_len + 4 <= len) {
2267 if (!IS_MPEG_HEADER (data + 6 + pes_packet_len))
2268 return FALSE;
2269 }
2270
2271 if (pack_size)
2272 *pack_size = 6 + pes_packet_len;
2273 return TRUE;
2274 }
2275
2276 static gboolean
mpeg_sys_is_valid_sys(GstTypeFind * tf,const guint8 * data,guint len,guint * pack_size)2277 mpeg_sys_is_valid_sys (GstTypeFind * tf, const guint8 * data, guint len,
2278 guint * pack_size)
2279 {
2280 guint sys_hdr_len;
2281
2282 /* Check the System header at the given position, assuming the header code itself
2283 * was already checked */
2284 if (len < 6)
2285 return FALSE;
2286 sys_hdr_len = GST_READ_UINT16_BE (data + 4);
2287 if (sys_hdr_len < 6)
2288 return FALSE;
2289
2290 /* Check the following header, if we can */
2291 if (6 + sys_hdr_len + 4 <= len) {
2292 if (!IS_MPEG_HEADER (data + 6 + sys_hdr_len))
2293 return FALSE;
2294 }
2295
2296 if (pack_size)
2297 *pack_size = 6 + sys_hdr_len;
2298
2299 return TRUE;
2300 }
2301
2302 /* calculation of possibility to identify random data as mpeg systemstream:
2303 * bits that must match in header detection: 32 (or more)
2304 * chance that random data is identifed: 1/2^32
2305 * chance that MPEG2_MIN_PACK_HEADERS headers are identified:
2306 * 1/2^(32*MPEG2_MIN_PACK_HEADERS)
2307 * chance that this happens in MPEG2_MAX_PROBE_LENGTH bytes:
2308 * 1-(1+1/2^(32*MPEG2_MIN_PACK_HEADERS)^MPEG2_MAX_PROBE_LENGTH)
2309 * for current values:
2310 * 1-(1+1/2^(32*4)^101024)
2311 * = <some_number>
2312 * Since we also check marker bits and pes packet lengths, this probability is a
2313 * very coarse upper bound.
2314 */
2315 static void
mpeg_sys_type_find(GstTypeFind * tf,gpointer unused)2316 mpeg_sys_type_find (GstTypeFind * tf, gpointer unused)
2317 {
2318 const guint8 *data, *data0, *first_sync, *end;
2319 gint mpegversion = 0;
2320 guint pack_headers = 0;
2321 guint pes_headers = 0;
2322 guint pack_size;
2323 guint since_last_sync = 0;
2324 guint32 sync_word = 0xffffffff;
2325 guint potential_headers = 0;
2326
2327 G_STMT_START {
2328 gint len;
2329
2330 len = MPEG2_MAX_PROBE_LENGTH;
2331
2332 while (len >= 16) {
2333 data = gst_type_find_peek (tf, 0, 5 + len);
2334 if (data != NULL)
2335 break;
2336 len = len / 2;
2337 }
2338
2339 if (!data)
2340 return;
2341
2342 end = data + len;
2343 }
2344 G_STMT_END;
2345
2346 data0 = data;
2347 first_sync = NULL;
2348
2349 while (data < end) {
2350 sync_word <<= 8;
2351 if (sync_word == 0x00000100) {
2352 /* Found potential sync word */
2353 if (first_sync == NULL)
2354 first_sync = data - 3;
2355
2356 if (since_last_sync > 4) {
2357 /* If more than 4 bytes since the last sync word, reset our counters,
2358 * as we're only interested in counting contiguous packets */
2359 pes_headers = pack_headers = 0;
2360 }
2361 pack_size = 0;
2362
2363 potential_headers++;
2364 if (IS_MPEG_PACK_CODE (data[0])) {
2365 if ((data[1] & 0xC0) == 0x40) {
2366 /* MPEG-2 */
2367 mpegversion = 2;
2368 } else if ((data[1] & 0xF0) == 0x20) {
2369 mpegversion = 1;
2370 }
2371 if (mpegversion != 0 &&
2372 mpeg_sys_is_valid_pack (tf, data - 3, end - data + 3, &pack_size)) {
2373 pack_headers++;
2374 }
2375 } else if (IS_MPEG_PES_CODE (data[0])) {
2376 /* PES stream */
2377 if (mpeg_sys_is_valid_pes (tf, data - 3, end - data + 3, &pack_size)) {
2378 pes_headers++;
2379 if (mpegversion == 0)
2380 mpegversion = 2;
2381 }
2382 } else if (IS_MPEG_SYS_CODE (data[0])) {
2383 if (mpeg_sys_is_valid_sys (tf, data - 3, end - data + 3, &pack_size)) {
2384 pack_headers++;
2385 }
2386 }
2387
2388 /* If we found a packet with a known size, skip the bytes in it and loop
2389 * around to check the next packet. */
2390 if (pack_size != 0) {
2391 data += pack_size - 3;
2392 sync_word = 0xffffffff;
2393 since_last_sync = 0;
2394 continue;
2395 }
2396 }
2397
2398 sync_word |= data[0];
2399 since_last_sync++;
2400 data++;
2401
2402 /* If we have found MAX headers, and *some* were pes headers (pack headers
2403 * are optional in an mpeg system stream) then return our high-probability
2404 * result */
2405 if (pes_headers > 0 && (pack_headers + pes_headers) > MPEG2_MAX_SYS_HEADERS)
2406 goto suggest;
2407 }
2408
2409 /* If we at least saw MIN headers, and *some* were pes headers (pack headers
2410 * are optional in an mpeg system stream) then return a lower-probability
2411 * result */
2412 if (pes_headers > 0 && (pack_headers + pes_headers) > MPEG2_MIN_SYS_HEADERS)
2413 goto suggest;
2414
2415 return;
2416 suggest:
2417 {
2418 guint prob;
2419
2420 prob = GST_TYPE_FIND_POSSIBLE + (10 * (pack_headers + pes_headers));
2421 prob = MIN (prob, GST_TYPE_FIND_MAXIMUM);
2422
2423 /* With the above test, we get into problems when we try to typefind
2424 a MPEG stream from a small amount of data, which can happen when
2425 we get data pushed from a HTTP source. We thus make a second test
2426 to give higher probability if all the potential headers were either
2427 pack or pes headers (ie, no potential header was unrecognized). */
2428 if (potential_headers == pack_headers + pes_headers) {
2429 GST_LOG ("Only %u headers, but all were recognized", potential_headers);
2430 prob += 10;
2431 prob = MIN (prob, GST_TYPE_FIND_MAXIMUM);
2432 }
2433
2434 /* lower probability if the first packet wasn't right at the start */
2435 if (data0 != first_sync && prob >= 10)
2436 prob -= 10;
2437
2438 GST_LOG ("Suggesting MPEG %d system stream, %d packs, %d pes, prob %u%%",
2439 mpegversion, pack_headers, pes_headers, prob);
2440
2441 gst_type_find_suggest_simple (tf, prob, "video/mpeg",
2442 "systemstream", G_TYPE_BOOLEAN, TRUE,
2443 "mpegversion", G_TYPE_INT, mpegversion, NULL);
2444 }
2445 };
2446
2447 /*** video/mpegts Transport Stream ***/
2448 static GstStaticCaps mpegts_caps = GST_STATIC_CAPS ("video/mpegts, "
2449 "systemstream = (boolean) true, packetsize = (int) [ 188, 208 ]");
2450 #define MPEGTS_CAPS gst_static_caps_get(&mpegts_caps)
2451
2452 #define GST_MPEGTS_TYPEFIND_MIN_HEADERS 4
2453 #define GST_MPEGTS_TYPEFIND_MAX_HEADERS 10
2454 #define GST_MPEGTS_MAX_PACKET_SIZE 208
2455 #define GST_MPEGTS_TYPEFIND_SYNC_SIZE \
2456 (GST_MPEGTS_TYPEFIND_MIN_HEADERS * GST_MPEGTS_MAX_PACKET_SIZE)
2457 #define GST_MPEGTS_TYPEFIND_MAX_SYNC \
2458 (GST_MPEGTS_TYPEFIND_MAX_HEADERS * GST_MPEGTS_MAX_PACKET_SIZE)
2459 #define GST_MPEGTS_TYPEFIND_SCAN_LENGTH \
2460 (GST_MPEGTS_TYPEFIND_MAX_SYNC * 4)
2461
2462 #define MPEGTS_HDR_SIZE 4
2463 /* Check for sync byte, error_indicator == 0 and packet has payload.
2464 * Adaptation control field (data[3] & 0x30) may be zero for TS packets with
2465 * null PIDs. Still, these streams are valid TS streams (for null packets,
2466 * AFC is supposed to be 0x1, but the spec also says decoders should just
2467 * discard any packets with AFC = 0x00) */
2468 #define IS_MPEGTS_HEADER(data) (data[0] == 0x47 && \
2469 (data[1] & 0x80) == 0x00 && \
2470 ((data[3] & 0x30) != 0x00 || \
2471 ((data[3] & 0x30) == 0x00 && (data[1] & 0x1f) == 0x1f && (data[2] & 0xff) == 0xff)))
2472
2473 /* Helper function to search ahead at intervals of packet_size for mpegts
2474 * headers */
2475 static gint
mpeg_ts_probe_headers(GstTypeFind * tf,guint64 offset,gint packet_size)2476 mpeg_ts_probe_headers (GstTypeFind * tf, guint64 offset, gint packet_size)
2477 {
2478 /* We always enter this function having found at least one header already */
2479 gint found = 1;
2480 const guint8 *data = NULL;
2481
2482 GST_LOG ("looking for mpeg-ts packets of size %u", packet_size);
2483 while (found < GST_MPEGTS_TYPEFIND_MAX_HEADERS) {
2484 offset += packet_size;
2485
2486 data = gst_type_find_peek (tf, offset, MPEGTS_HDR_SIZE);
2487 if (data == NULL || !IS_MPEGTS_HEADER (data))
2488 return found;
2489
2490 found++;
2491 GST_LOG ("mpeg-ts sync #%2d at offset %" G_GUINT64_FORMAT, found, offset);
2492 }
2493
2494 return found;
2495 }
2496
2497 /* Try and detect at least 4 packets in at most 10 packets worth of
2498 * data. Need to try several possible packet sizes */
2499 static void
mpeg_ts_type_find(GstTypeFind * tf,gpointer unused)2500 mpeg_ts_type_find (GstTypeFind * tf, gpointer unused)
2501 {
2502 /* TS packet sizes to test: normal, DVHS packet size and
2503 * FEC with 16 or 20 byte codes packet size. */
2504 const gint pack_sizes[] = { 188, 192, 204, 208 };
2505 const guint8 *data = NULL;
2506 guint size = 0;
2507 guint64 skipped = 0;
2508
2509 while (skipped < GST_MPEGTS_TYPEFIND_SCAN_LENGTH) {
2510 if (size < MPEGTS_HDR_SIZE) {
2511 data = gst_type_find_peek (tf, skipped, GST_MPEGTS_TYPEFIND_SYNC_SIZE);
2512 if (!data)
2513 break;
2514 size = GST_MPEGTS_TYPEFIND_SYNC_SIZE;
2515 }
2516
2517 /* Have at least MPEGTS_HDR_SIZE bytes at this point */
2518 if (IS_MPEGTS_HEADER (data)) {
2519 gsize p;
2520
2521 GST_LOG ("possible mpeg-ts sync at offset %" G_GUINT64_FORMAT, skipped);
2522
2523 for (p = 0; p < G_N_ELEMENTS (pack_sizes); p++) {
2524 gint found;
2525
2526 /* Probe ahead at size pack_sizes[p] */
2527 found = mpeg_ts_probe_headers (tf, skipped, pack_sizes[p]);
2528 if (found >= GST_MPEGTS_TYPEFIND_MIN_HEADERS) {
2529 gint probability;
2530
2531 /* found at least 4 headers. 10 headers = MAXIMUM probability.
2532 * Arbitrarily, I assigned 10% probability for each header we
2533 * found, 40% -> 100% */
2534 probability = MIN (10 * found, GST_TYPE_FIND_MAXIMUM);
2535
2536 gst_type_find_suggest_simple (tf, probability, "video/mpegts",
2537 "systemstream", G_TYPE_BOOLEAN, TRUE,
2538 "packetsize", G_TYPE_INT, pack_sizes[p], NULL);
2539 return;
2540 }
2541 }
2542 }
2543 data++;
2544 skipped++;
2545 size--;
2546 }
2547 }
2548
2549 #define GST_MPEGVID_TYPEFIND_TRY_PICTURES 6
2550 #define GST_MPEGVID_TYPEFIND_TRY_SYNC (100 * 1024) /* 100 kB */
2551
2552 /* Scan ahead a maximum of max_extra_offset bytes until the next IS_MPEG_HEADER
2553 * offset. After the call, offset will be after the 0x000001, i.e. at the 4th
2554 * byte of the MPEG header. Returns TRUE if a header was found, FALSE if not.
2555 */
2556 static gboolean
mpeg_find_next_header(GstTypeFind * tf,DataScanCtx * c,guint64 max_extra_offset)2557 mpeg_find_next_header (GstTypeFind * tf, DataScanCtx * c,
2558 guint64 max_extra_offset)
2559 {
2560 guint64 extra_offset;
2561
2562 for (extra_offset = 0; extra_offset <= max_extra_offset; ++extra_offset) {
2563 if (!data_scan_ctx_ensure_data (tf, c, 4))
2564 return FALSE;
2565 if (IS_MPEG_HEADER (c->data)) {
2566 data_scan_ctx_advance (tf, c, 3);
2567 return TRUE;
2568 }
2569 data_scan_ctx_advance (tf, c, 1);
2570 }
2571 return FALSE;
2572 }
2573
2574 /*** video/mpeg MPEG-4 elementary video stream ***/
2575
2576 static GstStaticCaps mpeg4_video_caps = GST_STATIC_CAPS ("video/mpeg, "
2577 "systemstream=(boolean)false, mpegversion=4, parsed=(boolean)false");
2578 #define MPEG4_VIDEO_CAPS gst_static_caps_get(&mpeg4_video_caps)
2579
2580 /*
2581 * This typefind is based on the elementary video header defined in
2582 * http://xhelmboyx.tripod.com/formats/mpeg-layout.txt
2583 * In addition, it allows the visual object sequence header to be
2584 * absent, and even the VOS header to be absent. In the latter case,
2585 * a number of VOPs have to be present.
2586 */
2587 static void
mpeg4_video_type_find(GstTypeFind * tf,gpointer unused)2588 mpeg4_video_type_find (GstTypeFind * tf, gpointer unused)
2589 {
2590 DataScanCtx c = { 0, NULL, 0 };
2591 gboolean seen_vios_at_0 = FALSE;
2592 gboolean seen_vios = FALSE;
2593 gboolean seen_vos = FALSE;
2594 gboolean seen_vol = FALSE;
2595 guint num_vop_headers = 0;
2596 guint8 sc;
2597
2598 while (c.offset < GST_MPEGVID_TYPEFIND_TRY_SYNC) {
2599 if (num_vop_headers >= GST_MPEGVID_TYPEFIND_TRY_PICTURES)
2600 break;
2601
2602 if (!mpeg_find_next_header (tf, &c,
2603 GST_MPEGVID_TYPEFIND_TRY_SYNC - c.offset))
2604 break;
2605
2606 sc = c.data[0];
2607
2608 /* visual_object_sequence_start_code */
2609 if (sc == 0xB0) {
2610 if (seen_vios)
2611 break; /* Terminate at second vios */
2612 if (c.offset == 0)
2613 seen_vios_at_0 = TRUE;
2614 seen_vios = TRUE;
2615 data_scan_ctx_advance (tf, &c, 2);
2616 if (!mpeg_find_next_header (tf, &c, 0))
2617 break;
2618
2619 sc = c.data[0];
2620
2621 /* Optional metadata */
2622 if (sc == 0xB2)
2623 if (!mpeg_find_next_header (tf, &c, 24))
2624 break;
2625 }
2626
2627 /* visual_object_start_code (consider it optional) */
2628 if (sc == 0xB5) {
2629 data_scan_ctx_advance (tf, &c, 2);
2630 /* may contain ID marker and YUV clamping */
2631 if (!mpeg_find_next_header (tf, &c, 7))
2632 break;
2633
2634 sc = c.data[0];
2635 }
2636
2637 /* video_object_start_code */
2638 if (sc <= 0x1F) {
2639 if (seen_vos)
2640 break; /* Terminate at second vos */
2641 seen_vos = TRUE;
2642 data_scan_ctx_advance (tf, &c, 2);
2643 continue;
2644 }
2645
2646 /* video_object_layer_start_code */
2647 if (sc >= 0x20 && sc <= 0x2F) {
2648 seen_vol = TRUE;
2649 data_scan_ctx_advance (tf, &c, 5);
2650 continue;
2651 }
2652
2653 /* video_object_plane_start_code */
2654 if (sc == 0xB6) {
2655 num_vop_headers++;
2656 data_scan_ctx_advance (tf, &c, 2);
2657 continue;
2658 }
2659
2660 /* Unknown start code. */
2661 }
2662
2663 if (num_vop_headers > 0 || seen_vol) {
2664 GstTypeFindProbability probability = 0;
2665
2666 GST_LOG ("Found %d pictures, vios: %d, vos:%d, vol:%d", num_vop_headers,
2667 seen_vios, seen_vos, seen_vol);
2668
2669 if (num_vop_headers >= GST_MPEGVID_TYPEFIND_TRY_PICTURES && seen_vios_at_0
2670 && seen_vos && seen_vol)
2671 probability = GST_TYPE_FIND_MAXIMUM - 1;
2672 else if (num_vop_headers >= GST_MPEGVID_TYPEFIND_TRY_PICTURES && seen_vios
2673 && seen_vos && seen_vol)
2674 probability = GST_TYPE_FIND_NEARLY_CERTAIN - 1;
2675 else if (seen_vios_at_0 && seen_vos && seen_vol)
2676 probability = GST_TYPE_FIND_NEARLY_CERTAIN - 6;
2677 else if (num_vop_headers >= GST_MPEGVID_TYPEFIND_TRY_PICTURES && seen_vos
2678 && seen_vol)
2679 probability = GST_TYPE_FIND_NEARLY_CERTAIN - 6;
2680 else if (num_vop_headers >= GST_MPEGVID_TYPEFIND_TRY_PICTURES && seen_vol)
2681 probability = GST_TYPE_FIND_NEARLY_CERTAIN - 9;
2682 else if (num_vop_headers >= GST_MPEGVID_TYPEFIND_TRY_PICTURES)
2683 probability = GST_TYPE_FIND_LIKELY - 1;
2684 else if (num_vop_headers > 2 && seen_vios && seen_vos && seen_vol)
2685 probability = GST_TYPE_FIND_LIKELY - 9;
2686 else if (seen_vios && seen_vos && seen_vol)
2687 probability = GST_TYPE_FIND_LIKELY - 20;
2688 else if (num_vop_headers > 0 && seen_vos && seen_vol)
2689 probability = GST_TYPE_FIND_POSSIBLE;
2690 else if (num_vop_headers > 0)
2691 probability = GST_TYPE_FIND_POSSIBLE - 10;
2692 else if (seen_vos && seen_vol)
2693 probability = GST_TYPE_FIND_POSSIBLE - 20;
2694
2695 gst_type_find_suggest (tf, probability, MPEG4_VIDEO_CAPS);
2696 }
2697 }
2698
2699 /*** video/x-h263 H263 video stream ***/
2700 static GstStaticCaps h263_video_caps =
2701 GST_STATIC_CAPS ("video/x-h263, variant=(string)itu");
2702
2703 #define H263_VIDEO_CAPS gst_static_caps_get(&h263_video_caps)
2704
2705 #define H263_MAX_PROBE_LENGTH (128 * 1024)
2706
2707 static void
h263_video_type_find(GstTypeFind * tf,gpointer unused)2708 h263_video_type_find (GstTypeFind * tf, gpointer unused)
2709 {
2710 DataScanCtx c = { 0, NULL, 0 };
2711 guint64 data = 0xffff; /* prevents false positive for first 2 bytes */
2712 guint64 psc = 0;
2713 guint8 ptype = 0;
2714 guint format;
2715 guint good = 0;
2716 guint bad = 0;
2717 guint pc_type, pb_mode;
2718
2719 while (c.offset < H263_MAX_PROBE_LENGTH) {
2720 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 4)))
2721 break;
2722
2723 /* Find the picture start code */
2724 data = (data << 8) + c.data[0];
2725 psc = data & G_GUINT64_CONSTANT (0xfffffc0000);
2726 if (psc == 0x800000) {
2727 /* Found PSC */
2728 /* PTYPE */
2729 ptype = (data & 0x3fc) >> 2;
2730 /* Source Format */
2731 format = ptype & 0x07;
2732
2733 /* Now that we have a Valid PSC, check if we also have a valid PTYPE and
2734 the Source Format, which should range between 1 and 5 */
2735 if (((ptype >> 6) == 0x2) && (format > 0 && format < 6)) {
2736 pc_type = data & 0x02;
2737 pb_mode = c.data[1] & 0x20 >> 4;
2738 if (!pc_type && pb_mode)
2739 bad++;
2740 else
2741 good++;
2742 } else
2743 bad++;
2744
2745 /* FIXME: maybe bail out early if we get mostly bad syncs ? */
2746 }
2747
2748 data_scan_ctx_advance (tf, &c, 1);
2749 }
2750
2751 GST_LOG ("good: %d, bad: %d", good, bad);
2752
2753 if (good > 2 * bad)
2754 gst_type_find_suggest (tf, GST_TYPE_FIND_POSSIBLE, H263_VIDEO_CAPS);
2755
2756 return;
2757 }
2758
2759 /*** video/x-h264 H264 elementary video stream ***/
2760
2761 static GstStaticCaps h264_video_caps =
2762 GST_STATIC_CAPS ("video/x-h264,stream-format=byte-stream");
2763
2764 #define H264_VIDEO_CAPS gst_static_caps_get(&h264_video_caps)
2765
2766 #define H264_MAX_PROBE_LENGTH (128 * 1024) /* 128kB for HD should be enough. */
2767
2768 static void
h264_video_type_find(GstTypeFind * tf,gpointer unused)2769 h264_video_type_find (GstTypeFind * tf, gpointer unused)
2770 {
2771 DataScanCtx c = { 0, NULL, 0 };
2772
2773 /* Stream consists of: a series of sync codes (00 00 00 01) followed
2774 * by NALs
2775 */
2776 gboolean seen_idr = FALSE;
2777 gboolean seen_sps = FALSE;
2778 gboolean seen_pps = FALSE;
2779 gboolean seen_ssps = FALSE;
2780 int nut, ref;
2781 int good = 0;
2782 int bad = 0;
2783
2784 while (c.offset < H264_MAX_PROBE_LENGTH) {
2785 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 4)))
2786 break;
2787
2788 if (IS_MPEG_HEADER (c.data)) {
2789 nut = c.data[3] & 0x9f; /* forbiden_zero_bit | nal_unit_type */
2790 ref = c.data[3] & 0x60; /* nal_ref_idc */
2791
2792 /* if forbidden bit is different to 0 won't be h264 */
2793 if (nut > 0x1f) {
2794 bad++;
2795 break;
2796 }
2797
2798 /* collect statistics about the NAL types */
2799 if ((nut >= 1 && nut <= 13) || nut == 19) {
2800 if ((nut == 5 && ref == 0) ||
2801 ((nut == 6 || (nut >= 9 && nut <= 12)) && ref != 0)) {
2802 bad++;
2803 } else {
2804 if (nut == 7)
2805 seen_sps = TRUE;
2806 else if (nut == 8)
2807 seen_pps = TRUE;
2808 else if (nut == 5)
2809 seen_idr = TRUE;
2810
2811 good++;
2812 }
2813 } else if (nut >= 14 && nut <= 33) {
2814 if (nut == 15) {
2815 seen_ssps = TRUE;
2816 good++;
2817 } else if (nut == 14 || nut == 20) {
2818 /* Sometimes we see NAL 14 or 20 without SSPS
2819 * if dropped into the middle of a stream -
2820 * just ignore those (don't add to bad count) */
2821 if (seen_ssps)
2822 good++;
2823 } else {
2824 /* reserved */
2825 /* Theoretically these are good, since if they exist in the
2826 stream it merely means that a newer backwards-compatible
2827 h.264 stream. But we should be identifying that separately. */
2828 bad++;
2829 }
2830 } else {
2831 /* unspecified, application specific */
2832 /* don't consider these bad */
2833 }
2834
2835 GST_LOG ("good:%d, bad:%d, pps:%d, sps:%d, idr:%d ssps:%d", good, bad,
2836 seen_pps, seen_sps, seen_idr, seen_ssps);
2837
2838 if (seen_sps && seen_pps && seen_idr && good >= 10 && bad < 4) {
2839 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY, H264_VIDEO_CAPS);
2840 return;
2841 }
2842
2843 data_scan_ctx_advance (tf, &c, 4);
2844 }
2845 data_scan_ctx_advance (tf, &c, 1);
2846 }
2847
2848 GST_LOG ("good:%d, bad:%d, pps:%d, sps:%d, idr:%d ssps=%d", good, bad,
2849 seen_pps, seen_sps, seen_idr, seen_ssps);
2850
2851 if (good >= 2 && bad == 0) {
2852 gst_type_find_suggest (tf, GST_TYPE_FIND_POSSIBLE, H264_VIDEO_CAPS);
2853 }
2854 }
2855
2856 /*** video/x-h265 H265 elementary video stream ***/
2857
2858 static GstStaticCaps h265_video_caps =
2859 GST_STATIC_CAPS ("video/x-h265,stream-format=byte-stream");
2860
2861 #define H265_VIDEO_CAPS gst_static_caps_get(&h265_video_caps)
2862
2863 #define H265_MAX_PROBE_LENGTH (128 * 1024) /* 128kB for HD should be enough. */
2864
2865 static void
h265_video_type_find(GstTypeFind * tf,gpointer unused)2866 h265_video_type_find (GstTypeFind * tf, gpointer unused)
2867 {
2868 DataScanCtx c = { 0, NULL, 0 };
2869
2870 /* Stream consists of: a series of sync codes (00 00 00 01) followed
2871 * by NALs
2872 */
2873 gboolean seen_irap = FALSE;
2874 gboolean seen_vps = FALSE;
2875 gboolean seen_sps = FALSE;
2876 gboolean seen_pps = FALSE;
2877 int nut;
2878 int good = 0;
2879 int bad = 0;
2880
2881 while (c.offset < H265_MAX_PROBE_LENGTH) {
2882 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 5)))
2883 break;
2884
2885 if (IS_MPEG_HEADER (c.data)) {
2886 /* forbiden_zero_bit | nal_unit_type */
2887 nut = c.data[3] & 0xfe;
2888
2889 /* if forbidden bit is different to 0 won't be h265 */
2890 if (nut > 0x7e) {
2891 bad++;
2892 break;
2893 }
2894 nut = nut >> 1;
2895
2896 /* if nuh_layer_id is not zero or nuh_temporal_id_plus1 is zero then
2897 * it won't be h265 */
2898 if ((c.data[3] & 0x01) || (c.data[4] & 0xf8) || !(c.data[4] & 0x07)) {
2899 bad++;
2900 break;
2901 }
2902
2903 /* collect statistics about the NAL types */
2904 if ((nut >= 0 && nut <= 9) || (nut >= 16 && nut <= 21) || (nut >= 32
2905 && nut <= 40)) {
2906 if (nut == 32)
2907 seen_vps = TRUE;
2908 else if (nut == 33)
2909 seen_sps = TRUE;
2910 else if (nut == 34)
2911 seen_pps = TRUE;
2912 else if (nut >= 16 && nut <= 21) {
2913 /* BLA, IDR and CRA pictures are belongs to be IRAP picture */
2914 /* we are not counting the reserved IRAP pictures (22 and 23) to good */
2915 seen_irap = TRUE;
2916 }
2917
2918 good++;
2919 } else if ((nut >= 10 && nut <= 15) || (nut >= 22 && nut <= 31)
2920 || (nut >= 41 && nut <= 47)) {
2921 /* reserved values are counting as bad */
2922 bad++;
2923 } else {
2924 /* unspecified (48..63), application specific */
2925 /* don't consider these as bad */
2926 }
2927
2928 GST_LOG ("good:%d, bad:%d, pps:%d, sps:%d, vps:%d, irap:%d", good, bad,
2929 seen_pps, seen_sps, seen_vps, seen_irap);
2930
2931 if (seen_sps && seen_pps && seen_irap && good >= 10 && bad < 4) {
2932 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY, H265_VIDEO_CAPS);
2933 return;
2934 }
2935
2936 data_scan_ctx_advance (tf, &c, 5);
2937 }
2938 data_scan_ctx_advance (tf, &c, 1);
2939 }
2940
2941 GST_LOG ("good:%d, bad:%d, pps:%d, sps:%d, vps:%d, irap:%d", good, bad,
2942 seen_pps, seen_sps, seen_vps, seen_irap);
2943
2944 if (good >= 2 && bad == 0) {
2945 gst_type_find_suggest (tf, GST_TYPE_FIND_POSSIBLE, H265_VIDEO_CAPS);
2946 }
2947 }
2948
2949 /*** video/mpeg video stream ***/
2950
2951 static GstStaticCaps mpeg_video_caps = GST_STATIC_CAPS ("video/mpeg, "
2952 "systemstream = (boolean) false");
2953 #define MPEG_VIDEO_CAPS gst_static_caps_get(&mpeg_video_caps)
2954
2955 /*
2956 * Idea is the same as MPEG system stream typefinding: We check each
2957 * byte of the stream to see if - from that point on - the stream
2958 * matches a predefined set of marker bits as defined in the MPEG
2959 * video specs.
2960 *
2961 * I'm sure someone will do a chance calculation here too.
2962 */
2963
2964 static void
mpeg_video_stream_type_find(GstTypeFind * tf,gpointer unused)2965 mpeg_video_stream_type_find (GstTypeFind * tf, gpointer unused)
2966 {
2967 DataScanCtx c = { 0, NULL, 0 };
2968 gboolean seen_seq_at_0 = FALSE;
2969 gboolean seen_seq = FALSE;
2970 gboolean seen_gop = FALSE;
2971 guint64 last_pic_offset = 0;
2972 gint num_pic_headers = 0;
2973 gint found = 0;
2974
2975 while (c.offset < GST_MPEGVID_TYPEFIND_TRY_SYNC) {
2976 if (found >= GST_MPEGVID_TYPEFIND_TRY_PICTURES)
2977 break;
2978
2979 if (!data_scan_ctx_ensure_data (tf, &c, 5))
2980 break;
2981
2982 if (!IS_MPEG_HEADER (c.data))
2983 goto next;
2984
2985 /* a pack header indicates that this isn't an elementary stream */
2986 if (c.data[3] == 0xBA && mpeg_sys_is_valid_pack (tf, c.data, c.size, NULL))
2987 return;
2988
2989 /* do we have a sequence header? */
2990 if (c.data[3] == 0xB3) {
2991 seen_seq_at_0 = seen_seq_at_0 || (c.offset == 0);
2992 seen_seq = TRUE;
2993 data_scan_ctx_advance (tf, &c, 4 + 8);
2994 continue;
2995 }
2996
2997 /* or a GOP header */
2998 if (c.data[3] == 0xB8) {
2999 seen_gop = TRUE;
3000 data_scan_ctx_advance (tf, &c, 8);
3001 continue;
3002 }
3003
3004 /* but what we'd really like to see is a picture header */
3005 if (c.data[3] == 0x00) {
3006 ++num_pic_headers;
3007 last_pic_offset = c.offset;
3008 data_scan_ctx_advance (tf, &c, 8);
3009 continue;
3010 }
3011
3012 /* ... each followed by a slice header with slice_vertical_pos=1 that's
3013 * not too far away from the previously seen picture header. */
3014 if (c.data[3] == 0x01 && num_pic_headers > found &&
3015 (c.offset - last_pic_offset) >= 4 &&
3016 (c.offset - last_pic_offset) <= 64) {
3017 data_scan_ctx_advance (tf, &c, 4);
3018 found += 1;
3019 continue;
3020 }
3021
3022 next:
3023
3024 data_scan_ctx_advance (tf, &c, 1);
3025 }
3026
3027 if (found > 0 || seen_seq) {
3028 GstTypeFindProbability probability = 0;
3029
3030 GST_LOG ("Found %d pictures, seq:%d, gop:%d", found, seen_seq, seen_gop);
3031
3032 if (found >= GST_MPEGVID_TYPEFIND_TRY_PICTURES && seen_seq && seen_gop)
3033 probability = GST_TYPE_FIND_NEARLY_CERTAIN - 1;
3034 else if (found >= GST_MPEGVID_TYPEFIND_TRY_PICTURES && seen_seq)
3035 probability = GST_TYPE_FIND_NEARLY_CERTAIN - 9;
3036 else if (found >= GST_MPEGVID_TYPEFIND_TRY_PICTURES)
3037 probability = GST_TYPE_FIND_LIKELY;
3038 else if (seen_seq_at_0 && seen_gop && found > 2)
3039 probability = GST_TYPE_FIND_LIKELY - 10;
3040 else if (seen_seq && seen_gop && found > 2)
3041 probability = GST_TYPE_FIND_LIKELY - 20;
3042 else if (seen_seq_at_0 && found > 0)
3043 probability = GST_TYPE_FIND_POSSIBLE;
3044 else if (seen_seq && found > 0)
3045 probability = GST_TYPE_FIND_POSSIBLE - 5;
3046 else if (found > 0)
3047 probability = GST_TYPE_FIND_POSSIBLE - 10;
3048 else if (seen_seq)
3049 probability = GST_TYPE_FIND_POSSIBLE - 20;
3050
3051 gst_type_find_suggest_simple (tf, probability, "video/mpeg",
3052 "systemstream", G_TYPE_BOOLEAN, FALSE,
3053 "mpegversion", G_TYPE_INT, 1, "parsed", G_TYPE_BOOLEAN, FALSE, NULL);
3054 }
3055 }
3056
3057 /*** audio/x-aiff ***/
3058
3059 static GstStaticCaps aiff_caps = GST_STATIC_CAPS ("audio/x-aiff");
3060
3061 #define AIFF_CAPS gst_static_caps_get(&aiff_caps)
3062 static void
aiff_type_find(GstTypeFind * tf,gpointer unused)3063 aiff_type_find (GstTypeFind * tf, gpointer unused)
3064 {
3065 const guint8 *data = gst_type_find_peek (tf, 0, 16);
3066
3067 if (data && memcmp (data, "FORM", 4) == 0) {
3068 data += 8;
3069 if (memcmp (data, "AIFF", 4) == 0 || memcmp (data, "AIFC", 4) == 0)
3070 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, AIFF_CAPS);
3071 }
3072 }
3073
3074 /*** audio/x-svx ***/
3075
3076 static GstStaticCaps svx_caps = GST_STATIC_CAPS ("audio/x-svx");
3077
3078 #define SVX_CAPS gst_static_caps_get(&svx_caps)
3079 static void
svx_type_find(GstTypeFind * tf,gpointer unused)3080 svx_type_find (GstTypeFind * tf, gpointer unused)
3081 {
3082 const guint8 *data = gst_type_find_peek (tf, 0, 16);
3083
3084 if (data && memcmp (data, "FORM", 4) == 0) {
3085 data += 8;
3086 if (memcmp (data, "8SVX", 4) == 0 || memcmp (data, "16SV", 4) == 0)
3087 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, SVX_CAPS);
3088 }
3089 }
3090
3091 /*** audio/x-shorten ***/
3092
3093 static GstStaticCaps shn_caps = GST_STATIC_CAPS ("audio/x-shorten");
3094
3095 #define SHN_CAPS gst_static_caps_get(&shn_caps)
3096 static void
shn_type_find(GstTypeFind * tf,gpointer unused)3097 shn_type_find (GstTypeFind * tf, gpointer unused)
3098 {
3099 const guint8 *data = gst_type_find_peek (tf, 0, 4);
3100
3101 if (data && memcmp (data, "ajkg", 4) == 0) {
3102 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, SHN_CAPS);
3103 }
3104 data = gst_type_find_peek (tf, -8, 8);
3105 if (data && memcmp (data, "SHNAMPSK", 8) == 0) {
3106 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, SHN_CAPS);
3107 }
3108 }
3109
3110 /*** application/x-ape ***/
3111
3112 static GstStaticCaps ape_caps = GST_STATIC_CAPS ("application/x-ape");
3113
3114 #define APE_CAPS gst_static_caps_get(&ape_caps)
3115 static void
ape_type_find(GstTypeFind * tf,gpointer unused)3116 ape_type_find (GstTypeFind * tf, gpointer unused)
3117 {
3118 const guint8 *data = gst_type_find_peek (tf, 0, 4);
3119
3120 if (data && memcmp (data, "MAC ", 4) == 0) {
3121 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY + 10, APE_CAPS);
3122 }
3123 }
3124
3125 /*** ISO FORMATS ***/
3126
3127 /*** audio/x-m4a ***/
3128
3129 static GstStaticCaps m4a_caps = GST_STATIC_CAPS ("audio/x-m4a");
3130
3131 #define M4A_CAPS (gst_static_caps_get(&m4a_caps))
3132 static void
m4a_type_find(GstTypeFind * tf,gpointer unused)3133 m4a_type_find (GstTypeFind * tf, gpointer unused)
3134 {
3135 const guint8 *data = gst_type_find_peek (tf, 4, 8);
3136
3137 if (data && (memcmp (data, "ftypM4A ", 8) == 0)) {
3138 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, M4A_CAPS);
3139 }
3140 }
3141
3142 /*** application/x-3gp ***/
3143
3144 /* The Q is there because variables can't start with a number. */
3145 static GstStaticCaps q3gp_caps = GST_STATIC_CAPS ("application/x-3gp");
3146 #define Q3GP_CAPS (gst_static_caps_get(&q3gp_caps))
3147
3148 static const gchar *
q3gp_type_find_get_profile(const guint8 * data)3149 q3gp_type_find_get_profile (const guint8 * data)
3150 {
3151 switch (GST_MAKE_FOURCC (data[0], data[1], data[2], 0)) {
3152 case GST_MAKE_FOURCC ('3', 'g', 'g', 0):
3153 return "general";
3154 case GST_MAKE_FOURCC ('3', 'g', 'p', 0):
3155 return "basic";
3156 case GST_MAKE_FOURCC ('3', 'g', 's', 0):
3157 return "streaming-server";
3158 case GST_MAKE_FOURCC ('3', 'g', 'r', 0):
3159 return "progressive-download";
3160 default:
3161 break;
3162 }
3163 return NULL;
3164 }
3165
3166 static void
q3gp_type_find(GstTypeFind * tf,gpointer unused)3167 q3gp_type_find (GstTypeFind * tf, gpointer unused)
3168 {
3169 const gchar *profile;
3170 guint32 ftyp_size = 0;
3171 guint32 offset = 0;
3172 const guint8 *data = NULL;
3173
3174 if ((data = gst_type_find_peek (tf, 0, 12)) == NULL) {
3175 return;
3176 }
3177
3178 data += 4;
3179 if (memcmp (data, "ftyp", 4) != 0) {
3180 return;
3181 }
3182
3183 /* check major brand */
3184 data += 4;
3185 if ((profile = q3gp_type_find_get_profile (data))) {
3186 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_MAXIMUM,
3187 "application/x-3gp", "profile", G_TYPE_STRING, profile, NULL);
3188 return;
3189 }
3190
3191 /* check compatible brands */
3192 if ((data = gst_type_find_peek (tf, 0, 4)) != NULL) {
3193 ftyp_size = GST_READ_UINT32_BE (data);
3194 }
3195 if ((data = gst_type_find_peek (tf, 0, ftyp_size)) != NULL) {
3196 for (offset = 16; offset + 4 < ftyp_size; offset += 4) {
3197 if ((profile = q3gp_type_find_get_profile (data + offset))) {
3198 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_MAXIMUM,
3199 "application/x-3gp", "profile", G_TYPE_STRING, profile, NULL);
3200 return;
3201 }
3202 }
3203 }
3204
3205 return;
3206
3207 }
3208
3209 /*** video/mj2 and image/jp2 ***/
3210 static GstStaticCaps mj2_caps = GST_STATIC_CAPS ("video/mj2");
3211
3212 #define MJ2_CAPS gst_static_caps_get(&mj2_caps)
3213
3214 static GstStaticCaps jp2_caps = GST_STATIC_CAPS ("image/jp2");
3215
3216 #define JP2_CAPS gst_static_caps_get(&jp2_caps)
3217
3218 static void
jp2_type_find(GstTypeFind * tf,gpointer unused)3219 jp2_type_find (GstTypeFind * tf, gpointer unused)
3220 {
3221 const guint8 *data;
3222
3223 data = gst_type_find_peek (tf, 0, 24);
3224 if (!data)
3225 return;
3226
3227 /* jp2 signature */
3228 if (memcmp (data, "\000\000\000\014jP \015\012\207\012", 12) != 0)
3229 return;
3230
3231 /* check ftyp box */
3232 data += 12;
3233 if (memcmp (data + 4, "ftyp", 4) == 0) {
3234 if (memcmp (data + 8, "jp2 ", 4) == 0)
3235 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, JP2_CAPS);
3236 else if (memcmp (data + 8, "mjp2", 4) == 0)
3237 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MJ2_CAPS);
3238 }
3239 }
3240
3241
3242 static GstStaticCaps jpc_caps = GST_STATIC_CAPS ("image/x-jpc");
3243
3244 #define JPC_CAPS gst_static_caps_get(&jpc_caps)
3245
3246 static void
jpc_type_find(GstTypeFind * tf,gpointer unused)3247 jpc_type_find (GstTypeFind * tf, gpointer unused)
3248 {
3249 gboolean found_cod = FALSE;
3250 gboolean found_qcd = FALSE;
3251 gboolean found_sot = FALSE;
3252 const guint8 *data;
3253 gint offset = 0;
3254 const guint8 soc_siz[] = { 0xff, 0x4f, 0xff, 0x51 };
3255
3256 #define GST_TYPE_FIND_JPC_MARKER_SOT 0xFF90
3257 #define GST_TYPE_FIND_JPC_MARKER_COD 0xFF52
3258 #define GST_TYPE_FIND_JPC_MARKER_QCD 0xFF5C
3259 #define GST_TYPE_FIND_JPC_MARKER_COC 0xFF53
3260 #define GST_TYPE_FIND_JPC_MARKER_RGN 0xFF5E
3261 #define GST_TYPE_FIND_JPC_MARKER_QCC 0xFF5D
3262 #define GST_TYPE_FIND_JPC_MARKER_POC 0xFF5F
3263 #define GST_TYPE_FIND_JPC_MARKER_PLM 0xFF57
3264 #define GST_TYPE_FIND_JPC_MARKER_PPM 0xFF60
3265 #define GST_TYPE_FIND_JPC_MARKER_TLM 0xFF55
3266 #define GST_TYPE_FIND_JPC_MARKER_CRG 0xFF63
3267 #define GST_TYPE_FIND_JPC_MARKER_COM 0xFF64
3268 #define GST_TYPE_FIND_JPC_MARKER_CBD 0xFF78
3269 #define GST_TYPE_FIND_JPC_MARKER_MCC 0xFF75
3270 #define GST_TYPE_FIND_JPC_MARKER_MCT 0xFF74
3271 #define GST_TYPE_FIND_JPC_MARKER_MCO 0xFF77
3272
3273
3274 /* SOC marker + SIZ marker */
3275 if ((data = gst_type_find_peek (tf, 0, 4)) != NULL) {
3276 if (memcmp (data, soc_siz, 4) != 0)
3277 return;
3278 offset += 4;
3279 } else {
3280 return;
3281 }
3282
3283 while (!found_sot) {
3284
3285 /* skip actual marker data */
3286 if ((data = gst_type_find_peek (tf, offset, 2)) != NULL) {
3287 offset += GST_READ_UINT16_BE (data);
3288 } else {
3289 return;
3290 }
3291
3292 /* read marker */
3293 if ((data = gst_type_find_peek (tf, offset, 2)) != NULL) {
3294 guint16 marker = GST_READ_UINT16_BE (data);
3295 switch (marker) {
3296 case GST_TYPE_FIND_JPC_MARKER_SOT:
3297 found_sot = TRUE;
3298 break;
3299 case GST_TYPE_FIND_JPC_MARKER_COD:
3300 found_cod = TRUE;
3301 break;
3302 case GST_TYPE_FIND_JPC_MARKER_QCD:
3303 found_qcd = TRUE;
3304 break;
3305 /* optional header markers */
3306 case GST_TYPE_FIND_JPC_MARKER_COC:
3307 case GST_TYPE_FIND_JPC_MARKER_RGN:
3308 case GST_TYPE_FIND_JPC_MARKER_QCC:
3309 case GST_TYPE_FIND_JPC_MARKER_POC:
3310 case GST_TYPE_FIND_JPC_MARKER_PLM:
3311 case GST_TYPE_FIND_JPC_MARKER_PPM:
3312 case GST_TYPE_FIND_JPC_MARKER_TLM:
3313 case GST_TYPE_FIND_JPC_MARKER_CRG:
3314 case GST_TYPE_FIND_JPC_MARKER_COM:
3315 case GST_TYPE_FIND_JPC_MARKER_CBD:
3316 case GST_TYPE_FIND_JPC_MARKER_MCC:
3317 case GST_TYPE_FIND_JPC_MARKER_MCT:
3318 case GST_TYPE_FIND_JPC_MARKER_MCO:
3319 break;
3320 /* unrecognized marker */
3321 default:
3322 return;
3323 }
3324 offset += 2;
3325 } else {
3326 return;
3327 }
3328 }
3329
3330 if (found_cod && found_qcd && found_sot)
3331 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, JPC_CAPS);
3332 }
3333
3334
3335 /*** video/quicktime ***/
3336
3337 static GstStaticCaps qt_caps = GST_STATIC_CAPS ("video/quicktime");
3338
3339 #define QT_CAPS gst_static_caps_get(&qt_caps)
3340 #define STRNCMP(x,y,z) (strncmp ((char*)(x), (char*)(y), z))
3341
3342 /* FIXME 0.11: go through http://www.ftyps.com/ */
3343 static void
qt_type_find(GstTypeFind * tf,gpointer unused)3344 qt_type_find (GstTypeFind * tf, gpointer unused)
3345 {
3346 const guint8 *data;
3347 guint tip = 0;
3348 guint64 offset = 0;
3349 guint64 size;
3350 const gchar *variant = NULL;
3351
3352 while ((data = gst_type_find_peek (tf, offset, 12)) != NULL) {
3353 guint64 new_offset;
3354
3355 if (STRNCMP (&data[4], "ftypqt ", 8) == 0) {
3356 tip = GST_TYPE_FIND_MAXIMUM;
3357 break;
3358 }
3359
3360 if (STRNCMP (&data[4], "ftypisom", 8) == 0 ||
3361 STRNCMP (&data[4], "ftypavc1", 8) == 0 ||
3362 STRNCMP (&data[4], "ftypmp42", 8) == 0) {
3363 tip = GST_TYPE_FIND_MAXIMUM;
3364 variant = "iso";
3365 break;
3366 }
3367
3368 if (STRNCMP (&data[4], "ftypisml", 8) == 0 ||
3369 STRNCMP (&data[4], "ftypavc3", 8) == 0) {
3370 tip = GST_TYPE_FIND_MAXIMUM;
3371 variant = "iso-fragmented";
3372 break;
3373 }
3374
3375 if (STRNCMP (&data[4], "ftypccff", 8) == 0) {
3376 tip = GST_TYPE_FIND_MAXIMUM;
3377 variant = "ccff";
3378 break;
3379 }
3380
3381 /* box/atom types that are in common with ISO base media file format */
3382 if (STRNCMP (&data[4], "moov", 4) == 0 ||
3383 STRNCMP (&data[4], "mdat", 4) == 0 ||
3384 STRNCMP (&data[4], "ftyp", 4) == 0 ||
3385 STRNCMP (&data[4], "free", 4) == 0 ||
3386 STRNCMP (&data[4], "uuid", 4) == 0 ||
3387 STRNCMP (&data[4], "moof", 4) == 0 ||
3388 STRNCMP (&data[4], "skip", 4) == 0) {
3389 if (tip == 0) {
3390 tip = GST_TYPE_FIND_LIKELY;
3391 } else {
3392 tip = GST_TYPE_FIND_NEARLY_CERTAIN;
3393 }
3394 }
3395 /* other box/atom types, apparently quicktime specific */
3396 else if (STRNCMP (&data[4], "pnot", 4) == 0 ||
3397 STRNCMP (&data[4], "PICT", 4) == 0 ||
3398 STRNCMP (&data[4], "wide", 4) == 0 ||
3399 STRNCMP (&data[4], "prfl", 4) == 0) {
3400 tip = GST_TYPE_FIND_MAXIMUM;
3401 break;
3402 } else {
3403 tip = 0;
3404 break;
3405 }
3406
3407 size = GST_READ_UINT32_BE (data);
3408 if (size + offset >= G_MAXINT64)
3409 break;
3410 /* check compatible brands rather than ever expaning major brands above */
3411 if ((STRNCMP (&data[4], "ftyp", 4) == 0) && (size >= 16)) {
3412 data = gst_type_find_peek (tf, offset, size);
3413 if (data == NULL)
3414 goto done;
3415 new_offset = 12;
3416 while (new_offset + 4 <= size) {
3417 if (STRNCMP (&data[new_offset], "isom", 4) == 0 ||
3418 STRNCMP (&data[new_offset], "dash", 4) == 0 ||
3419 STRNCMP (&data[new_offset], "avc1", 4) == 0 ||
3420 STRNCMP (&data[new_offset], "avc3", 4) == 0 ||
3421 STRNCMP (&data[new_offset], "mp41", 4) == 0 ||
3422 STRNCMP (&data[new_offset], "mp42", 4) == 0) {
3423 tip = GST_TYPE_FIND_MAXIMUM;
3424 variant = "iso";
3425 goto done;
3426 }
3427 new_offset += 4;
3428 }
3429 }
3430 if (size == 1) {
3431 const guint8 *sizedata;
3432
3433 sizedata = gst_type_find_peek (tf, offset + 8, 8);
3434 if (sizedata == NULL)
3435 break;
3436
3437 size = GST_READ_UINT64_BE (sizedata);
3438 } else {
3439 if (size < 8)
3440 break;
3441 }
3442 new_offset = offset + size;
3443 if (new_offset <= offset)
3444 break;
3445 if (new_offset + 16 >= G_MAXINT64)
3446 break;
3447 offset = new_offset;
3448 }
3449
3450 done:
3451 if (tip > 0) {
3452 if (variant) {
3453 GstCaps *caps = gst_caps_copy (QT_CAPS);
3454
3455 gst_caps_set_simple (caps, "variant", G_TYPE_STRING, variant, NULL);
3456 gst_type_find_suggest (tf, tip, caps);
3457 gst_caps_unref (caps);
3458 } else {
3459 gst_type_find_suggest (tf, tip, QT_CAPS);
3460 }
3461 }
3462 };
3463
3464
3465 /*** image/x-quicktime ***/
3466
3467 static GstStaticCaps qtif_caps = GST_STATIC_CAPS ("image/x-quicktime");
3468
3469 #define QTIF_CAPS gst_static_caps_get(&qtif_caps)
3470
3471 /* how many atoms we check before we give up */
3472 #define QTIF_MAXROUNDS 25
3473
3474 static void
qtif_type_find(GstTypeFind * tf,gpointer unused)3475 qtif_type_find (GstTypeFind * tf, gpointer unused)
3476 {
3477 const guint8 *data;
3478 gboolean found_idsc = FALSE;
3479 gboolean found_idat = FALSE;
3480 guint64 offset = 0;
3481 guint rounds = 0;
3482
3483 while ((data = gst_type_find_peek (tf, offset, 8)) != NULL) {
3484 guint64 size;
3485
3486 size = GST_READ_UINT32_BE (data);
3487 if (size == 1) {
3488 const guint8 *sizedata;
3489
3490 sizedata = gst_type_find_peek (tf, offset + 8, 8);
3491 if (sizedata == NULL)
3492 break;
3493
3494 size = GST_READ_UINT64_BE (sizedata);
3495 }
3496 if (size < 8)
3497 break;
3498
3499 if (STRNCMP (data + 4, "idsc", 4) == 0)
3500 found_idsc = TRUE;
3501 if (STRNCMP (data + 4, "idat", 4) == 0)
3502 found_idat = TRUE;
3503
3504 if (found_idsc && found_idat) {
3505 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, QTIF_CAPS);
3506 return;
3507 }
3508
3509 offset += size;
3510 if (offset + 8 >= G_MAXINT64)
3511 break;
3512 if (++rounds > QTIF_MAXROUNDS)
3513 break;
3514 }
3515
3516 if (found_idsc || found_idat) {
3517 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY, QTIF_CAPS);
3518 return;
3519 }
3520 };
3521
3522 /*** audio/x-mod ***/
3523
3524 static GstStaticCaps mod_caps = GST_STATIC_CAPS ("audio/x-mod");
3525
3526 #define MOD_CAPS gst_static_caps_get(&mod_caps)
3527 /* FIXME: M15 CheckType to do */
3528 static void
mod_type_find(GstTypeFind * tf,gpointer unused)3529 mod_type_find (GstTypeFind * tf, gpointer unused)
3530 {
3531 const guint8 *data;
3532 GstTypeFindProbability probability;
3533 const char *mod_type = NULL;
3534
3535 /* MOD */
3536 if ((data = gst_type_find_peek (tf, 1080, 4)) != NULL) {
3537 /* Protracker and variants */
3538 if ((memcmp (data, "M.K.", 4) == 0) ||
3539 (memcmp (data, "M!K!", 4) == 0) ||
3540 (memcmp (data, "M&K!", 4) == 0) || (memcmp (data, "N.T.", 4) == 0) ||
3541 /* Star Tracker */
3542 (memcmp (data, "FLT", 3) == 0 && isdigit (data[3])) ||
3543 (memcmp (data, "EXO", 3) == 0 && isdigit (data[3])) ||
3544 /* Oktalyzer (Amiga) */
3545 (memcmp (data, "OKTA", 4) == 0) || (memcmp (data, "OCTA", 4) == 0) ||
3546 /* Oktalyser (Atari) */
3547 (memcmp (data, "CD81", 4) == 0) ||
3548 /* Taketracker */
3549 (memcmp (data, "TDZ", 3) == 0 && isdigit (data[3])) ||
3550 /* Fasttracker */
3551 (memcmp (data + 1, "CHN", 3) == 0 && isdigit (data[0])) ||
3552 /* Fasttracker or Taketracker */
3553 (memcmp (data + 2, "CH", 2) == 0 && isdigit (data[0])
3554 && isdigit (data[1])) || (memcmp (data + 2, "CN", 2) == 0
3555 && isdigit (data[0]) && isdigit (data[1]))) {
3556 mod_type = "mod";
3557 probability = GST_TYPE_FIND_MAXIMUM;
3558 goto suggest_audio_mod_caps;
3559 }
3560 }
3561 /* J2B (Jazz Jackrabbit 2) */
3562 if ((data = gst_type_find_peek (tf, 0, 8)) != NULL) {
3563 if ((memcmp (data, "MUSE\xDE\xAD", 4) == 0) &&
3564 ((memcmp (data + 6, "\xBE\xEF", 2) == 0) ||
3565 (memcmp (data + 6, "\xBA\xBE", 2) == 0))) {
3566 mod_type = "j2b";
3567 probability = GST_TYPE_FIND_MAXIMUM;
3568 goto suggest_audio_mod_caps;
3569 }
3570 }
3571 /* AMS (Velvet Studio) */
3572 if ((data = gst_type_find_peek (tf, 0, 7)) != NULL) {
3573 if (memcmp (data, "AMShdr\x1A", 7) == 0) {
3574 mod_type = "velvet-ams";
3575 probability = GST_TYPE_FIND_MAXIMUM;
3576 goto suggest_audio_mod_caps;
3577 }
3578 }
3579 /* AMS (Extreme Tracker) */
3580 if ((data = gst_type_find_peek (tf, 0, 9)) != NULL) {
3581 if ((memcmp (data, "Extreme", 7) == 0) && (data[8] == 1)) {
3582 mod_type = "extreme-ams";
3583 probability = GST_TYPE_FIND_LIKELY;
3584 goto suggest_audio_mod_caps;
3585 }
3586 }
3587 /* ULT (Ultratracker) */
3588 if ((data = gst_type_find_peek (tf, 0, 14)) != NULL) {
3589 if (memcmp (data, "MAS_UTrack_V00", 14) == 0) {
3590 mod_type = "ult";
3591 probability = GST_TYPE_FIND_MAXIMUM;
3592 goto suggest_audio_mod_caps;
3593 }
3594 }
3595 /* DIGI (DigiBooster) */
3596 if ((data = gst_type_find_peek (tf, 0, 20)) != NULL) {
3597 if (memcmp (data, "DIGI Booster module\0", 20) == 0) {
3598 mod_type = "digi";
3599 probability = GST_TYPE_FIND_MAXIMUM;
3600 goto suggest_audio_mod_caps;
3601 }
3602 }
3603 /* PTM (PolyTracker) */
3604 if ((data = gst_type_find_peek (tf, 0x2C, 4)) != NULL) {
3605 if (memcmp (data, "PTMF", 4) == 0) {
3606 mod_type = "ptm";
3607 probability = GST_TYPE_FIND_LIKELY;
3608 goto suggest_audio_mod_caps;
3609 }
3610 }
3611 /* XM */
3612 if ((data = gst_type_find_peek (tf, 0, 38)) != NULL) {
3613 if ((memcmp (data, "Extended Module: ", 17) == 0) && (data[37] == 0x1A)) {
3614 mod_type = "xm";
3615 probability = GST_TYPE_FIND_MAXIMUM;
3616 goto suggest_audio_mod_caps;
3617 }
3618 }
3619 /* OKT */
3620 if (data || (data = gst_type_find_peek (tf, 0, 8)) != NULL) {
3621 if (memcmp (data, "OKTASONG", 8) == 0) {
3622 mod_type = "okt";
3623 probability = GST_TYPE_FIND_MAXIMUM;
3624 goto suggest_audio_mod_caps;
3625 }
3626 }
3627 /* Various formats with a 4-byte magic ID at the beginning of the file */
3628 if (data || (data = gst_type_find_peek (tf, 0, 4)) != NULL) {
3629 /* PSM (Protracker Studio PSM) */
3630 if (memcmp (data, "PSM", 3) == 0) {
3631 unsigned char fbyte = data[3];
3632 if ((fbyte == ' ') || (fbyte == 254)) {
3633 mod_type = "psm";
3634 probability = GST_TYPE_FIND_MAXIMUM;
3635 goto suggest_audio_mod_caps;
3636 }
3637 }
3638 /* 669 */
3639 if ((memcmp (data, "if", 2) == 0) || (memcmp (data, "JN", 2) == 0)) {
3640 mod_type = "669";
3641 probability = GST_TYPE_FIND_LIKELY;
3642 goto suggest_audio_mod_caps;
3643 }
3644 /* AMF */
3645 if ((memcmp (data, "AMF", 3) == 0) && (data[3] > 10) && (data[3] < 14)) {
3646 mod_type = "dsmi-amf";
3647 probability = GST_TYPE_FIND_MAXIMUM;
3648 goto suggest_audio_mod_caps;
3649 }
3650 /* IT */
3651 if (memcmp (data, "IMPM", 4) == 0) {
3652 mod_type = "it";
3653 probability = GST_TYPE_FIND_MAXIMUM;
3654 goto suggest_audio_mod_caps;
3655 }
3656 /* DBM (DigiBooster Pro) */
3657 if (memcmp (data, "DBM0", 4) == 0) {
3658 mod_type = "dbm";
3659 probability = GST_TYPE_FIND_MAXIMUM;
3660 goto suggest_audio_mod_caps;
3661 }
3662 /* MDL (DigiTrakker) */
3663 if (memcmp (data, "DMDL", 4) == 0) {
3664 mod_type = "mdl";
3665 probability = GST_TYPE_FIND_MAXIMUM;
3666 goto suggest_audio_mod_caps;
3667 }
3668 /* MT2 (MadTracker 2.0) */
3669 if (memcmp (data, "MT20", 4) == 0) {
3670 mod_type = "mt2";
3671 probability = GST_TYPE_FIND_MAXIMUM;
3672 goto suggest_audio_mod_caps;
3673 }
3674 /* DMF (X-Tracker) */
3675 if (memcmp (data, "DDMF", 4) == 0) {
3676 mod_type = "dmf";
3677 probability = GST_TYPE_FIND_MAXIMUM;
3678 goto suggest_audio_mod_caps;
3679 }
3680 /* MED */
3681 if ((memcmp (data, "MMD0", 4) == 0) || (memcmp (data, "MMD1", 4) == 0)) {
3682 mod_type = "med";
3683 probability = GST_TYPE_FIND_MAXIMUM;
3684 goto suggest_audio_mod_caps;
3685 }
3686 /* MTM */
3687 if (memcmp (data, "MTM", 3) == 0) {
3688 mod_type = "mtm";
3689 probability = GST_TYPE_FIND_MAXIMUM;
3690 goto suggest_audio_mod_caps;
3691 }
3692 /* DSM */
3693 if (memcmp (data, "RIFF", 4) == 0) {
3694 const guint8 *data2 = gst_type_find_peek (tf, 8, 4);
3695
3696 if (data2) {
3697 if (memcmp (data2, "DSMF", 4) == 0) {
3698 mod_type = "dsm";
3699 probability = GST_TYPE_FIND_MAXIMUM;
3700 goto suggest_audio_mod_caps;
3701 }
3702 }
3703 }
3704 /* FAR (Farandole) */
3705 if (memcmp (data, "FAR\xFE", 4) == 0) {
3706 mod_type = "far";
3707 probability = GST_TYPE_FIND_MAXIMUM;
3708 goto suggest_audio_mod_caps;
3709 }
3710 /* FAM */
3711 if (memcmp (data, "FAM\xFE", 4) == 0) {
3712 const guint8 *data2 = gst_type_find_peek (tf, 44, 3);
3713
3714 if (data2) {
3715 if (memcmp (data2, "compare", 3) == 0) {
3716 mod_type = "fam";
3717 probability = GST_TYPE_FIND_MAXIMUM;
3718 goto suggest_audio_mod_caps;
3719 }
3720 /* otherwise do not suggest anything */
3721 } else {
3722 mod_type = "fam";
3723 probability = GST_TYPE_FIND_LIKELY;
3724 goto suggest_audio_mod_caps;
3725 }
3726 }
3727 /* GDM */
3728 if (memcmp (data, "GDM\xFE", 4) == 0) {
3729 const guint8 *data2 = gst_type_find_peek (tf, 71, 4);
3730
3731 if (data2) {
3732 if (memcmp (data2, "GMFS", 4) == 0) {
3733 mod_type = "gdm";
3734 probability = GST_TYPE_FIND_MAXIMUM;
3735 goto suggest_audio_mod_caps;
3736 }
3737 /* otherwise do not suggest anything */
3738 } else {
3739 mod_type = "gdm";
3740 probability = GST_TYPE_FIND_LIKELY;
3741 goto suggest_audio_mod_caps;
3742 }
3743 }
3744 /* UMX */
3745 if (memcmp (data, "\xC1\x83\x2A\x9E", 4) == 0) {
3746 mod_type = "umx";
3747 probability = GST_TYPE_FIND_POSSIBLE;
3748 goto suggest_audio_mod_caps;
3749 }
3750 }
3751 /* FAR (Farandole) (secondary detection) */
3752 if ((data = gst_type_find_peek (tf, 44, 3)) != NULL) {
3753 if (memcmp (data, "\x0D\x0A\x1A", 3) == 0) {
3754 mod_type = "far";
3755 probability = GST_TYPE_FIND_POSSIBLE;
3756 goto suggest_audio_mod_caps;
3757 }
3758 }
3759 /* IMF */
3760 if ((data = gst_type_find_peek (tf, 60, 4)) != NULL) {
3761 if (memcmp (data, "IM10", 4) == 0) {
3762 mod_type = "imf";
3763 probability = GST_TYPE_FIND_MAXIMUM;
3764 goto suggest_audio_mod_caps;
3765 }
3766 }
3767 /* S3M */
3768 if ((data = gst_type_find_peek (tf, 44, 4)) != NULL) {
3769 if (memcmp (data, "SCRM", 4) == 0) {
3770 mod_type = "s3m";
3771 probability = GST_TYPE_FIND_MAXIMUM;
3772 goto suggest_audio_mod_caps;
3773 }
3774 }
3775 /* STM */
3776 if ((data = gst_type_find_peek (tf, 20, 8)) != NULL) {
3777 if (g_ascii_strncasecmp ((gchar *) data, "!Scream!", 8) == 0 ||
3778 g_ascii_strncasecmp ((gchar *) data, "BMOD2STM", 8) == 0) {
3779 const guint8 *id, *stmtype;
3780
3781 if ((id = gst_type_find_peek (tf, 28, 1)) == NULL)
3782 return;
3783 if ((stmtype = gst_type_find_peek (tf, 29, 1)) == NULL)
3784 return;
3785 if (*id == 0x1A && *stmtype == 2) {
3786 mod_type = "stm";
3787 probability = GST_TYPE_FIND_MAXIMUM;
3788 goto suggest_audio_mod_caps;
3789 }
3790 }
3791 }
3792 /* AMF */
3793 if ((data = gst_type_find_peek (tf, 0, 19)) != NULL) {
3794 if (memcmp (data, "ASYLUM Music Format", 19) == 0) {
3795 mod_type = "asylum-amf";
3796 probability = GST_TYPE_FIND_MAXIMUM;
3797 goto suggest_audio_mod_caps;
3798 }
3799 }
3800
3801 suggest_audio_mod_caps:
3802 if (mod_type != NULL) {
3803 GstCaps *caps = gst_caps_new_simple ("audio/x-mod",
3804 "type", G_TYPE_STRING, mod_type, NULL);
3805
3806 gst_type_find_suggest (tf, probability, caps);
3807 gst_caps_unref (caps);
3808 }
3809 }
3810
3811 /*** application/x-shockwave-flash ***/
3812
3813 static GstStaticCaps swf_caps =
3814 GST_STATIC_CAPS ("application/x-shockwave-flash");
3815 #define SWF_CAPS (gst_static_caps_get(&swf_caps))
3816 static void
swf_type_find(GstTypeFind * tf,gpointer unused)3817 swf_type_find (GstTypeFind * tf, gpointer unused)
3818 {
3819 const guint8 *data = gst_type_find_peek (tf, 0, 4);
3820
3821 if (data && (data[0] == 'F' || data[0] == 'C') &&
3822 data[1] == 'W' && data[2] == 'S') {
3823 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, SWF_CAPS);
3824 }
3825 }
3826
3827 /*** application/vnd.ms-sstr+xml ***/
3828
3829 static void
mss_manifest_load_utf16(gunichar2 * utf16_ne,const guint8 * utf16_data,gsize data_size,guint data_endianness)3830 mss_manifest_load_utf16 (gunichar2 * utf16_ne, const guint8 * utf16_data,
3831 gsize data_size, guint data_endianness)
3832 {
3833 memcpy (utf16_ne, utf16_data, data_size);
3834 if (data_endianness != G_BYTE_ORDER) {
3835 guint i;
3836
3837 for (i = 0; i < data_size / 2; ++i)
3838 utf16_ne[i] = GUINT16_SWAP_LE_BE (utf16_ne[i]);
3839 }
3840 }
3841
3842 static GstStaticCaps mss_manifest_caps =
3843 GST_STATIC_CAPS ("application/vnd.ms-sstr+xml");
3844 #define MSS_MANIFEST_CAPS (gst_static_caps_get(&mss_manifest_caps))
3845 static void
mss_manifest_type_find(GstTypeFind * tf,gpointer unused)3846 mss_manifest_type_find (GstTypeFind * tf, gpointer unused)
3847 {
3848 gunichar2 utf16_ne[512];
3849 const guint8 *data;
3850 guint data_endianness = 0;
3851 glong n_read = 0, size = 0;
3852 guint length;
3853 gchar *utf8;
3854 gboolean utf8_bom_detected = FALSE;
3855
3856 if (xml_check_first_element (tf, "SmoothStreamingMedia", 20, TRUE)) {
3857 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MSS_MANIFEST_CAPS);
3858 return;
3859 }
3860
3861 length = gst_type_find_get_length (tf);
3862
3863 /* try detecting the charset */
3864 data = gst_type_find_peek (tf, 0, 3);
3865
3866 if (data == NULL)
3867 return;
3868
3869 /* look for a possible BOM */
3870 if (data[0] == 0xEF && data[1] == 0xBB && data[2] == 0xBF)
3871 utf8_bom_detected = TRUE;
3872 else if (data[0] == 0xFF && data[1] == 0xFE)
3873 data_endianness = G_LITTLE_ENDIAN;
3874 else if (data[0] == 0xFE && data[1] == 0xFF)
3875 data_endianness = G_BIG_ENDIAN;
3876 else
3877 return;
3878
3879 /* try a default that should be enough */
3880 if (length == 0)
3881 length = 512;
3882 else if (length < 64)
3883 return;
3884 else /* the first few bytes should be enough */
3885 length = MIN (1024, length);
3886
3887 data = gst_type_find_peek (tf, 0, length);
3888
3889 if (data == NULL)
3890 return;
3891
3892 /* skip the BOM */
3893 data += 2;
3894 length -= 2;
3895
3896 if (utf8_bom_detected) {
3897 /* skip last byte of the BOM */
3898 data++;
3899 length--;
3900
3901 if (xml_check_first_element_from_data (data, length,
3902 "SmoothStreamingMedia", 20, TRUE))
3903 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MSS_MANIFEST_CAPS);
3904 } else {
3905 length = GST_ROUND_DOWN_2 (length);
3906
3907 /* convert to native endian UTF-16 */
3908 mss_manifest_load_utf16 (utf16_ne, data, length, data_endianness);
3909
3910 /* and now convert to UTF-8 */
3911 utf8 = g_utf16_to_utf8 (utf16_ne, length / 2, &n_read, &size, NULL);
3912 if (utf8 != NULL && n_read > 0) {
3913 if (xml_check_first_element_from_data ((const guint8 *) utf8, size,
3914 "SmoothStreamingMedia", 20, TRUE))
3915 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MSS_MANIFEST_CAPS);
3916 }
3917 g_free (utf8);
3918 }
3919 }
3920
3921 /*** image/jpeg ***/
3922
3923 #define JPEG_MARKER_IS_START_OF_FRAME(x) \
3924 ((x)>=0xc0 && (x) <= 0xcf && (x)!=0xc4 && (x)!=0xc8 && (x)!=0xcc)
3925
3926 static GstStaticCaps jpeg_caps = GST_STATIC_CAPS ("image/jpeg");
3927
3928 #define JPEG_CAPS (gst_static_caps_get(&jpeg_caps))
3929 static void
jpeg_type_find(GstTypeFind * tf,gpointer unused)3930 jpeg_type_find (GstTypeFind * tf, gpointer unused)
3931 {
3932 GstTypeFindProbability prob = GST_TYPE_FIND_POSSIBLE;
3933 DataScanCtx c = { 0, NULL, 0 };
3934 GstCaps *caps;
3935 guint num_markers;
3936
3937 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 2)))
3938 return;
3939
3940 if (c.data[0] != 0xff || c.data[1] != 0xd8)
3941 return;
3942
3943 num_markers = 1;
3944 data_scan_ctx_advance (tf, &c, 2);
3945
3946 caps = gst_caps_copy (JPEG_CAPS);
3947
3948 while (data_scan_ctx_ensure_data (tf, &c, 4) && c.offset < (200 * 1024)) {
3949 guint16 len;
3950 guint8 marker;
3951
3952 if (c.data[0] != 0xff)
3953 break;
3954
3955 marker = c.data[1];
3956 if (G_UNLIKELY (marker == 0xff)) {
3957 data_scan_ctx_advance (tf, &c, 1);
3958 continue;
3959 }
3960
3961 data_scan_ctx_advance (tf, &c, 2);
3962
3963 /* we assume all markers we'll see before SOF have a payload length; if
3964 * that's not the case we'll just detect a false sync and bail out, but
3965 * still report POSSIBLE probability */
3966 len = GST_READ_UINT16_BE (c.data);
3967
3968 GST_LOG ("possible JPEG marker 0x%02x (@0x%04x), segment length %u",
3969 marker, (guint) c.offset, len);
3970
3971 if (!data_scan_ctx_ensure_data (tf, &c, len))
3972 break;
3973
3974 if (marker == 0xc4 || /* DEFINE_HUFFMAN_TABLES */
3975 marker == 0xcc || /* DEFINE_ARITHMETIC_CONDITIONING */
3976 marker == 0xdb || /* DEFINE_QUANTIZATION_TABLES */
3977 marker == 0xdd || /* DEFINE_RESTART_INTERVAL */
3978 marker == 0xfe) { /* COMMENT */
3979 data_scan_ctx_advance (tf, &c, len);
3980 ++num_markers;
3981 } else if (marker == 0xe0 && len >= (2 + 4) && /* APP0 */
3982 data_scan_ctx_memcmp (tf, &c, 2, "JFIF", 4)) {
3983 GST_LOG ("found JFIF tag");
3984 prob = GST_TYPE_FIND_MAXIMUM;
3985 data_scan_ctx_advance (tf, &c, len);
3986 ++num_markers;
3987 /* we continue until we find a start of frame marker */
3988 } else if (marker == 0xe1 && len >= (2 + 4) && /* APP1 */
3989 data_scan_ctx_memcmp (tf, &c, 2, "Exif", 4)) {
3990 GST_LOG ("found Exif tag");
3991 prob = GST_TYPE_FIND_MAXIMUM;
3992 data_scan_ctx_advance (tf, &c, len);
3993 ++num_markers;
3994 /* we continue until we find a start of frame marker */
3995 } else if (marker >= 0xe0 && marker <= 0xef) { /* APPn */
3996 data_scan_ctx_advance (tf, &c, len);
3997 ++num_markers;
3998 } else if (JPEG_MARKER_IS_START_OF_FRAME (marker) && len >= (2 + 8)) {
3999 int h, w;
4000
4001 h = GST_READ_UINT16_BE (c.data + 2 + 1);
4002 w = GST_READ_UINT16_BE (c.data + 2 + 1 + 2);
4003 if (h == 0 || w == 0) {
4004 GST_WARNING ("bad width %u and/or height %u in SOF header", w, h);
4005 break;
4006 }
4007
4008 GST_LOG ("SOF at offset %" G_GUINT64_FORMAT ", num_markers=%d, "
4009 "WxH=%dx%d", c.offset - 2, num_markers, w, h);
4010
4011 if (num_markers >= 5 || prob == GST_TYPE_FIND_MAXIMUM)
4012 prob = GST_TYPE_FIND_MAXIMUM;
4013 else
4014 prob = GST_TYPE_FIND_LIKELY;
4015
4016 gst_caps_set_simple (caps, "width", G_TYPE_INT, w,
4017 "height", G_TYPE_INT, h, "sof-marker", G_TYPE_INT, marker & 0xf,
4018 NULL);
4019
4020 break;
4021 } else {
4022 GST_WARNING ("bad length or unexpected JPEG marker 0xff 0x%02x", marker);
4023 break;
4024 }
4025 }
4026
4027 gst_type_find_suggest (tf, prob, caps);
4028 gst_caps_unref (caps);
4029 }
4030
4031 /*** image/bmp ***/
4032
4033 static GstStaticCaps bmp_caps = GST_STATIC_CAPS ("image/bmp");
4034
4035 #define BMP_CAPS (gst_static_caps_get(&bmp_caps))
4036 static void
bmp_type_find(GstTypeFind * tf,gpointer unused)4037 bmp_type_find (GstTypeFind * tf, gpointer unused)
4038 {
4039 DataScanCtx c = { 0, NULL, 0 };
4040 guint32 struct_size, w, h, planes, bpp;
4041
4042 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 54)))
4043 return;
4044
4045 if (c.data[0] != 'B' || c.data[1] != 'M')
4046 return;
4047
4048 /* skip marker + size */
4049 data_scan_ctx_advance (tf, &c, 2 + 4);
4050
4051 /* reserved, must be 0 */
4052 if (c.data[0] != 0 || c.data[1] != 0 || c.data[2] != 0 || c.data[3] != 0)
4053 return;
4054
4055 data_scan_ctx_advance (tf, &c, 2 + 2);
4056
4057 /* offset to start of image data in bytes (check for sanity) */
4058 GST_LOG ("offset=%u", GST_READ_UINT32_LE (c.data));
4059 if (GST_READ_UINT32_LE (c.data) > (10 * 1024 * 1024))
4060 return;
4061
4062 struct_size = GST_READ_UINT32_LE (c.data + 4);
4063 GST_LOG ("struct_size=%u", struct_size);
4064
4065 data_scan_ctx_advance (tf, &c, 4 + 4);
4066
4067 if (struct_size == 0x0C) {
4068 w = GST_READ_UINT16_LE (c.data);
4069 h = GST_READ_UINT16_LE (c.data + 2);
4070 planes = GST_READ_UINT16_LE (c.data + 2 + 2);
4071 bpp = GST_READ_UINT16_LE (c.data + 2 + 2 + 2);
4072 } else if (struct_size == 40 || struct_size == 64 || struct_size == 108
4073 || struct_size == 124 || struct_size == 0xF0) {
4074 w = GST_READ_UINT32_LE (c.data);
4075 h = GST_READ_UINT32_LE (c.data + 4);
4076 planes = GST_READ_UINT16_LE (c.data + 4 + 4);
4077 bpp = GST_READ_UINT16_LE (c.data + 4 + 4 + 2);
4078 } else {
4079 return;
4080 }
4081
4082 /* image sizes sanity check */
4083 GST_LOG ("w=%u, h=%u, planes=%u, bpp=%u", w, h, planes, bpp);
4084 if (w == 0 || w > 0xfffff || h == 0 || h > 0xfffff || planes != 1 ||
4085 (bpp != 1 && bpp != 4 && bpp != 8 && bpp != 16 && bpp != 24 && bpp != 32))
4086 return;
4087
4088 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_MAXIMUM, "image/bmp",
4089 "width", G_TYPE_INT, w, "height", G_TYPE_INT, h, "bpp", G_TYPE_INT, bpp,
4090 NULL);
4091 }
4092
4093 /*** image/tiff ***/
4094 static GstStaticCaps tiff_caps = GST_STATIC_CAPS ("image/tiff, "
4095 "endianness = (int) { BIG_ENDIAN, LITTLE_ENDIAN }");
4096 #define TIFF_CAPS (gst_static_caps_get(&tiff_caps))
4097 static GstStaticCaps tiff_be_caps = GST_STATIC_CAPS ("image/tiff, "
4098 "endianness = (int) BIG_ENDIAN");
4099 #define TIFF_BE_CAPS (gst_static_caps_get(&tiff_be_caps))
4100 static GstStaticCaps tiff_le_caps = GST_STATIC_CAPS ("image/tiff, "
4101 "endianness = (int) LITTLE_ENDIAN");
4102 #define TIFF_LE_CAPS (gst_static_caps_get(&tiff_le_caps))
4103 static void
tiff_type_find(GstTypeFind * tf,gpointer ununsed)4104 tiff_type_find (GstTypeFind * tf, gpointer ununsed)
4105 {
4106 const guint8 *data = gst_type_find_peek (tf, 0, 8);
4107 guint8 le_header[4] = { 0x49, 0x49, 0x2A, 0x00 };
4108 guint8 be_header[4] = { 0x4D, 0x4D, 0x00, 0x2A };
4109
4110 if (data) {
4111 if (memcmp (data, le_header, 4) == 0) {
4112 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, TIFF_LE_CAPS);
4113 } else if (memcmp (data, be_header, 4) == 0) {
4114 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, TIFF_BE_CAPS);
4115 }
4116 }
4117 }
4118
4119 /*** image/x-exr ***/
4120 static GstStaticCaps exr_caps = GST_STATIC_CAPS ("image/x-exr");
4121 #define EXR_CAPS (gst_static_caps_get(&exr_caps))
4122 static void
exr_type_find(GstTypeFind * tf,gpointer ununsed)4123 exr_type_find (GstTypeFind * tf, gpointer ununsed)
4124 {
4125 const guint8 *data = gst_type_find_peek (tf, 0, 8);
4126
4127 if (data) {
4128 guint32 flags;
4129
4130 if (GST_READ_UINT32_LE (data) != 0x01312f76)
4131 return;
4132
4133 flags = GST_READ_UINT32_LE (data + 4);
4134 if ((flags & 0xff) != 1 && (flags & 0xff) != 2)
4135 return;
4136
4137 /* If bit 9 is set, bit 11 and 12 must be 0 */
4138 if ((flags & 0x200) && (flags & 0x1800))
4139 return;
4140
4141 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, EXR_CAPS);
4142 }
4143 }
4144
4145
4146 /*** PNM ***/
4147
4148 static GstStaticCaps pnm_caps = GST_STATIC_CAPS ("image/x-portable-bitmap; "
4149 "image/x-portable-graymap; image/x-portable-pixmap; "
4150 "image/x-portable-anymap");
4151
4152 #define PNM_CAPS (gst_static_caps_get(&pnm_caps))
4153
4154 #define IS_PNM_WHITESPACE(c) \
4155 ((c) == ' ' || (c) == '\r' || (c) == '\n' || (c) == 't')
4156
4157 static void
pnm_type_find(GstTypeFind * tf,gpointer ununsed)4158 pnm_type_find (GstTypeFind * tf, gpointer ununsed)
4159 {
4160 const gchar *media_type = NULL;
4161 DataScanCtx c = { 0, NULL, 0 };
4162 guint h = 0, w = 0;
4163
4164 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 16)))
4165 return;
4166
4167 /* see http://en.wikipedia.org/wiki/Netpbm_format */
4168 if (c.data[0] != 'P' || c.data[1] < '1' || c.data[1] > '7' ||
4169 !IS_PNM_WHITESPACE (c.data[2]) ||
4170 (c.data[3] != '#' && c.data[3] < '0' && c.data[3] > '9'))
4171 return;
4172
4173 switch (c.data[1]) {
4174 case '1':
4175 media_type = "image/x-portable-bitmap"; /* ASCII */
4176 break;
4177 case '2':
4178 media_type = "image/x-portable-graymap"; /* ASCII */
4179 break;
4180 case '3':
4181 media_type = "image/x-portable-pixmap"; /* ASCII */
4182 break;
4183 case '4':
4184 media_type = "image/x-portable-bitmap"; /* Raw */
4185 break;
4186 case '5':
4187 media_type = "image/x-portable-graymap"; /* Raw */
4188 break;
4189 case '6':
4190 media_type = "image/x-portable-pixmap"; /* Raw */
4191 break;
4192 case '7':
4193 media_type = "image/x-portable-anymap";
4194 break;
4195 default:
4196 g_return_if_reached ();
4197 }
4198
4199 /* try to extract width and height as well */
4200 if (c.data[1] != '7') {
4201 gchar s[64] = { 0, }
4202 , sep1, sep2;
4203
4204 /* need to skip any comment lines first */
4205 data_scan_ctx_advance (tf, &c, 3);
4206
4207 if (!data_scan_ctx_ensure_data (tf, &c, 1))
4208 return;
4209
4210 while (c.data[0] == '#') { /* we know there's still data left */
4211 data_scan_ctx_advance (tf, &c, 1);
4212 if (!data_scan_ctx_ensure_data (tf, &c, 1))
4213 return;
4214
4215 while (c.data[0] != '\n' && c.data[0] != '\r') {
4216 data_scan_ctx_advance (tf, &c, 1);
4217 if (!data_scan_ctx_ensure_data (tf, &c, 1))
4218 return;
4219 }
4220 data_scan_ctx_advance (tf, &c, 1);
4221 GST_LOG ("skipped comment line in PNM header");
4222 if (!data_scan_ctx_ensure_data (tf, &c, 1))
4223 return;
4224 }
4225
4226 if (!data_scan_ctx_ensure_data (tf, &c, 32) &&
4227 !data_scan_ctx_ensure_data (tf, &c, 4)) {
4228 return;
4229 }
4230
4231 /* need to NUL-terminate data for sscanf */
4232 memcpy (s, c.data, MIN (sizeof (s) - 1, c.size));
4233 if (sscanf (s, "%u%c%u%c", &w, &sep1, &h, &sep2) == 4 &&
4234 IS_PNM_WHITESPACE (sep1) && IS_PNM_WHITESPACE (sep2) &&
4235 w > 0 && w < G_MAXINT && h > 0 && h < G_MAXINT) {
4236 GST_LOG ("extracted PNM width and height: %dx%d", w, h);
4237 } else {
4238 w = 0;
4239 h = 0;
4240 }
4241 } else {
4242 /* FIXME: extract width + height for anymaps too */
4243 }
4244
4245 if (w > 0 && h > 0) {
4246 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_MAXIMUM, media_type,
4247 "width", G_TYPE_INT, w, "height", G_TYPE_INT, h, NULL);
4248 } else {
4249 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_LIKELY, media_type, NULL);
4250 }
4251 }
4252
4253 static GstStaticCaps sds_caps = GST_STATIC_CAPS ("audio/x-sds");
4254
4255 #define SDS_CAPS (gst_static_caps_get(&sds_caps))
4256 static void
sds_type_find(GstTypeFind * tf,gpointer ununsed)4257 sds_type_find (GstTypeFind * tf, gpointer ununsed)
4258 {
4259 const guint8 *data = gst_type_find_peek (tf, 0, 4);
4260 guint8 mask[4] = { 0xFF, 0xFF, 0x80, 0xFF };
4261 guint8 match[4] = { 0xF0, 0x7E, 0, 0x01 };
4262 gint x;
4263
4264 if (data) {
4265 for (x = 0; x < 4; x++) {
4266 if ((data[x] & mask[x]) != match[x]) {
4267 return;
4268 }
4269 }
4270 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, SDS_CAPS);
4271 }
4272 }
4273
4274 static GstStaticCaps ircam_caps = GST_STATIC_CAPS ("audio/x-ircam");
4275
4276 #define IRCAM_CAPS (gst_static_caps_get(&ircam_caps))
4277 static void
ircam_type_find(GstTypeFind * tf,gpointer ununsed)4278 ircam_type_find (GstTypeFind * tf, gpointer ununsed)
4279 {
4280 const guint8 *data = gst_type_find_peek (tf, 0, 4);
4281 guint8 mask[4] = { 0xFF, 0xFF, 0xF8, 0xFF };
4282 guint8 match[4] = { 0x64, 0xA3, 0x00, 0x00 };
4283 gint x;
4284 gboolean matched = TRUE;
4285
4286 if (!data) {
4287 return;
4288 }
4289 for (x = 0; x < 4; x++) {
4290 if ((data[x] & mask[x]) != match[x]) {
4291 matched = FALSE;
4292 }
4293 }
4294 if (matched) {
4295 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, IRCAM_CAPS);
4296 return;
4297 }
4298 /* now try the reverse version */
4299 matched = TRUE;
4300 for (x = 0; x < 4; x++) {
4301 if ((data[x] & mask[3 - x]) != match[3 - x]) {
4302 matched = FALSE;
4303 }
4304 }
4305 }
4306
4307 /*** Matroska/WebM ***/
4308
4309 #define EBML_HEADER 0x1A45DFA3
4310 #define EBML_VERSION 0x4286
4311 #define EBML_DOCTYPE 0x4282
4312 #define EBML_DOCTYPE_VERSION 0x4287
4313 #define MATROSKA_SEGMENT 0x18538067
4314 #define MATROSKA_CLUSTER 0x1F43B675
4315 #define MATROSKA_TRACKS 0x1654AE6B
4316 #define MATROSKA_TRACK_ENTRY 0xAE
4317 #define MATROSKA_TRACK_TYPE 0x83
4318 #define MATROSKA_STEREO_MODE 0x53B8
4319
4320 #define EBML_MAX_LEN (2 * 1024 * 1024)
4321
4322 typedef enum
4323 {
4324 EBML_DOCTYPE_UNKNOWN = 0,
4325 EBML_DOCTYPE_MATROSKA,
4326 EBML_DOCTYPE_WEBM
4327 } GstEbmlDocType;
4328
4329 typedef struct
4330 {
4331 GstEbmlDocType doctype;
4332 guint audio;
4333 guint video;
4334 guint other;
4335 guint video_stereo;
4336 guint chunks;
4337 guint tracks_ok; /* if we've seen and fully parsed the TRACKS element */
4338 } GstMatroskaInfo;
4339
4340 static inline guint
ebml_read_chunk_header(GstTypeFind * tf,DataScanCtx * c,guint max_size,guint32 * id,guint64 * size)4341 ebml_read_chunk_header (GstTypeFind * tf, DataScanCtx * c, guint max_size,
4342 guint32 * id, guint64 * size)
4343 {
4344 guint64 mask;
4345 guint msbit_set, i, len, id_len;
4346
4347 if (c->size < 12 || max_size < 1)
4348 return 0;
4349
4350 /* element ID */
4351 *id = c->data[0];
4352 if ((c->data[0] & 0x80) == 0x80) {
4353 id_len = 1;
4354 } else if ((c->data[0] & 0xC0) == 0x40) {
4355 id_len = 2;
4356 } else if ((c->data[0] & 0xE0) == 0x20) {
4357 id_len = 3;
4358 } else if ((c->data[0] & 0xF0) == 0x10) {
4359 id_len = 4;
4360 } else {
4361 return 0;
4362 }
4363
4364 if (max_size < id_len)
4365 return 0;
4366
4367 for (i = 1; i < id_len; ++i) {
4368 *id = (*id << 8) | c->data[i];
4369 }
4370
4371 data_scan_ctx_advance (tf, c, id_len);
4372 max_size -= id_len;
4373
4374 /* size */
4375 if (max_size < 1 || c->data[0] == 0)
4376 return 0;
4377
4378 msbit_set = g_bit_nth_msf (c->data[0], 8);
4379 mask = ((1 << msbit_set) - 1);
4380 *size = c->data[0] & mask;
4381 len = 7 - msbit_set;
4382
4383 if (max_size < 1 + len)
4384 return 0;
4385 for (i = 0; i < len; ++i) {
4386 mask = (mask << 8) | 0xff;
4387 *size = (*size << 8) | c->data[1 + i];
4388 }
4389
4390 data_scan_ctx_advance (tf, c, 1 + len);
4391
4392 /* undefined/unknown size? (all bits 1) */
4393 if (*size == mask) {
4394 /* allow unknown size for SEGMENT chunk, bail out otherwise */
4395 if (*id == MATROSKA_SEGMENT)
4396 *size = G_MAXUINT64;
4397 else
4398 return 0;
4399 }
4400
4401 return id_len + (1 + len);
4402 }
4403
4404 static gboolean
ebml_parse_chunk(GstTypeFind * tf,DataScanCtx * ctx,guint32 chunk_id,guint chunk_size,GstMatroskaInfo * info,guint depth)4405 ebml_parse_chunk (GstTypeFind * tf, DataScanCtx * ctx, guint32 chunk_id,
4406 guint chunk_size, GstMatroskaInfo * info, guint depth)
4407 { /* FIXME: make sure input size is clipped to 32 bit */
4408 static const gchar SPACES[] = " ";
4409 DataScanCtx c = *ctx;
4410 guint64 element_size = 0;
4411 guint32 id, hdr_len;
4412
4413 if (depth >= 8) /* keep SPACES large enough for depth */
4414 return FALSE;
4415
4416 while (chunk_size > 0) {
4417 if (c.offset > EBML_MAX_LEN || !data_scan_ctx_ensure_data (tf, &c, 64))
4418 return FALSE;
4419
4420 hdr_len = ebml_read_chunk_header (tf, &c, chunk_size, &id, &element_size);
4421 if (hdr_len == 0)
4422 return FALSE;
4423
4424 g_assert (hdr_len <= chunk_size);
4425 chunk_size -= hdr_len;
4426
4427 if (element_size > chunk_size)
4428 return FALSE;
4429
4430 GST_DEBUG ("%s %08x, size %" G_GUINT64_FORMAT " / %" G_GUINT64_FORMAT,
4431 SPACES + sizeof (SPACES) - 1 - (2 * depth), id, element_size,
4432 hdr_len + element_size);
4433
4434 if (element_size >= G_MAXUINT32) {
4435 GST_DEBUG ("Chunk too big for typefinding");
4436 return FALSE;
4437 }
4438
4439 if (!data_scan_ctx_ensure_data (tf, &c, element_size)) {
4440 GST_DEBUG ("not enough data");
4441 return FALSE;
4442 }
4443
4444 switch (id) {
4445 case EBML_DOCTYPE:
4446 if (element_size >= 8 && memcmp (c.data, "matroska", 8) == 0)
4447 info->doctype = EBML_DOCTYPE_MATROSKA;
4448 else if (element_size >= 4 && memcmp (c.data, "webm", 4) == 0)
4449 info->doctype = EBML_DOCTYPE_WEBM;
4450 break;
4451 case MATROSKA_SEGMENT:
4452 GST_LOG ("parsing segment");
4453 ebml_parse_chunk (tf, &c, id, element_size, info, depth + 1);
4454 GST_LOG ("parsed segment, done");
4455 return FALSE;
4456 case MATROSKA_TRACKS:
4457 GST_LOG ("parsing tracks");
4458 info->tracks_ok =
4459 ebml_parse_chunk (tf, &c, id, element_size, info, depth + 1);
4460 GST_LOG ("parsed tracks: %s, done (after %" G_GUINT64_FORMAT " bytes)",
4461 info->tracks_ok ? "ok" : "FAIL", c.offset + element_size);
4462 return FALSE;
4463 case MATROSKA_TRACK_ENTRY:
4464 GST_LOG ("parsing track entry");
4465 if (!ebml_parse_chunk (tf, &c, id, element_size, info, depth + 1))
4466 return FALSE;
4467 break;
4468 case MATROSKA_TRACK_TYPE:{
4469 guint type = 0, i;
4470
4471 /* is supposed to always be 1-byte, but not everyone's following that */
4472 for (i = 0; i < element_size; ++i)
4473 type = (type << 8) | c.data[i];
4474
4475 GST_DEBUG ("%s track type %u",
4476 SPACES + sizeof (SPACES) - 1 - (2 * depth), type);
4477
4478 if (type == 1)
4479 ++info->video;
4480 else if (c.data[0] == 2)
4481 ++info->audio;
4482 else
4483 ++info->other;
4484 break;
4485 }
4486 case MATROSKA_STEREO_MODE:
4487 ++info->video_stereo;
4488 break;
4489 case MATROSKA_CLUSTER:
4490 GST_WARNING ("cluster, bailing out (should've found tracks by now)");
4491 return FALSE;
4492 default:
4493 break;
4494 }
4495 data_scan_ctx_advance (tf, &c, element_size);
4496 chunk_size -= element_size;
4497 ++info->chunks;
4498 }
4499
4500 return TRUE;
4501 }
4502
4503 static GstStaticCaps matroska_caps = GST_STATIC_CAPS ("video/x-matroska");
4504
4505 #define MATROSKA_CAPS (gst_static_caps_get(&matroska_caps))
4506 static void
matroska_type_find(GstTypeFind * tf,gpointer ununsed)4507 matroska_type_find (GstTypeFind * tf, gpointer ununsed)
4508 {
4509 GstTypeFindProbability prob;
4510 GstMatroskaInfo info = { 0, };
4511 const gchar *type_name;
4512 DataScanCtx c = { 0, NULL, 0 };
4513 gboolean is_audio;
4514 guint64 size;
4515 guint32 id, hdr_len;
4516
4517 if (!data_scan_ctx_ensure_data (tf, &c, 64))
4518 return;
4519
4520 if (GST_READ_UINT32_BE (c.data) != EBML_HEADER)
4521 return;
4522
4523 while (c.offset < EBML_MAX_LEN && data_scan_ctx_ensure_data (tf, &c, 64)) {
4524 hdr_len = ebml_read_chunk_header (tf, &c, c.size, &id, &size);
4525 if (hdr_len == 0)
4526 return;
4527
4528 GST_INFO ("=== top-level chunk %08x, size %" G_GUINT64_FORMAT
4529 " / %" G_GUINT64_FORMAT, id, size, size + hdr_len);
4530
4531 if (!ebml_parse_chunk (tf, &c, id, size, &info, 0))
4532 break;
4533 data_scan_ctx_advance (tf, &c, size);
4534 GST_INFO ("=== done with chunk %08x", id);
4535 if (id == MATROSKA_SEGMENT)
4536 break;
4537 }
4538
4539 GST_INFO ("audio=%u video=%u other=%u chunks=%u doctype=%d all_tracks=%d",
4540 info.audio, info.video, info.other, info.chunks, info.doctype,
4541 info.tracks_ok);
4542
4543 /* perhaps we should bail out if tracks_ok is FALSE and wait for more data?
4544 * (we would need new API to signal this properly and prevent other
4545 * typefinders from taking over the decision then) */
4546 is_audio = (info.audio > 0 && info.video == 0 && info.other == 0);
4547
4548 if (info.doctype == EBML_DOCTYPE_WEBM) {
4549 type_name = (is_audio) ? "audio/webm" : "video/webm";
4550 } else if (info.video > 0 && info.video_stereo) {
4551 type_name = "video/x-matroska-3d";
4552 } else {
4553 type_name = (is_audio) ? "audio/x-matroska" : "video/x-matroska";
4554 }
4555
4556 if (info.doctype == EBML_DOCTYPE_UNKNOWN)
4557 prob = GST_TYPE_FIND_LIKELY;
4558 else
4559 prob = GST_TYPE_FIND_MAXIMUM;
4560
4561 gst_type_find_suggest_simple (tf, prob, type_name, NULL);
4562 }
4563
4564 /*** application/mxf ***/
4565 static GstStaticCaps mxf_caps = GST_STATIC_CAPS ("application/mxf");
4566
4567 #define MXF_MAX_PROBE_LENGTH (1024 * 64)
4568 #define MXF_CAPS (gst_static_caps_get(&mxf_caps))
4569
4570 /*
4571 * MXF files start with a header partition pack key of 16 bytes which is defined
4572 * at SMPTE-377M 6.1. Before this there can be up to 64K of run-in which _must_
4573 * not contain the partition pack key.
4574 */
4575 static void
mxf_type_find(GstTypeFind * tf,gpointer ununsed)4576 mxf_type_find (GstTypeFind * tf, gpointer ununsed)
4577 {
4578 static const guint8 partition_pack_key[] =
4579 { 0x06, 0x0e, 0x2b, 0x34, 0x02, 0x05, 0x01, 0x01, 0x0d, 0x01, 0x02, 0x01,
4580 0x01
4581 };
4582 DataScanCtx c = { 0, NULL, 0 };
4583
4584 while (c.offset <= MXF_MAX_PROBE_LENGTH) {
4585 guint i;
4586 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 1024)))
4587 break;
4588
4589 /* look over in chunks of 1kbytes to avoid too much overhead */
4590
4591 for (i = 0; i < 1024 - 16; i++) {
4592 /* Check first byte before calling more expensive memcmp function */
4593 if (G_UNLIKELY (c.data[i] == 0x06
4594 && memcmp (c.data + i, partition_pack_key, 13) == 0)) {
4595 /* Header partition pack? */
4596 if (c.data[i + 13] != 0x02)
4597 goto advance;
4598
4599 /* Partition status */
4600 if (c.data[i + 14] >= 0x05)
4601 goto advance;
4602
4603 /* Reserved, must be 0x00 */
4604 if (c.data[i + 15] != 0x00)
4605 goto advance;
4606
4607 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MXF_CAPS);
4608 return;
4609 }
4610 }
4611
4612 advance:
4613 data_scan_ctx_advance (tf, &c, 1024 - 16);
4614 }
4615 }
4616
4617 /*** video/x-dv ***/
4618
4619 static GstStaticCaps dv_caps = GST_STATIC_CAPS ("video/x-dv, "
4620 "systemstream = (boolean) true");
4621 #define DV_CAPS (gst_static_caps_get(&dv_caps))
4622 static void
dv_type_find(GstTypeFind * tf,gpointer private)4623 dv_type_find (GstTypeFind * tf, gpointer private)
4624 {
4625 const guint8 *data;
4626
4627 data = gst_type_find_peek (tf, 0, 5);
4628
4629 /* check for DIF and DV flag */
4630 if (data && (data[0] == 0x1f) && (data[1] == 0x07) && (data[2] == 0x00)) {
4631 const gchar *format;
4632
4633 if (data[3] & 0x80) {
4634 format = "PAL";
4635 } else {
4636 format = "NTSC";
4637 }
4638
4639 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_MAXIMUM, "video/x-dv",
4640 "systemstream", G_TYPE_BOOLEAN, TRUE,
4641 "format", G_TYPE_STRING, format, NULL);
4642 }
4643 }
4644
4645
4646 /*** Ogg variants ***/
4647 static GstStaticCaps ogg_caps =
4648 GST_STATIC_CAPS ("application/ogg;video/ogg;audio/ogg;application/kate");
4649
4650 #define OGG_CAPS (gst_static_caps_get(&ogg_caps))
4651
4652 typedef enum
4653 {
4654 OGG_AUDIO = 0,
4655 OGG_VIDEO,
4656 OGG_KATE,
4657 OGG_OTHER,
4658 OGG_SKELETON,
4659 OGG_ANNODEX,
4660 OGG_NUM
4661 } GstOggStreamType;
4662
4663 static void
ogganx_type_find(GstTypeFind * tf,gpointer private)4664 ogganx_type_find (GstTypeFind * tf, gpointer private)
4665 {
4666 const gchar *media_type;
4667 DataScanCtx c = { 0, NULL, 0 };
4668 guint ogg_syncs = 0;
4669 guint hdr_count[OGG_NUM] = { 0, };
4670 static const struct
4671 {
4672 const gchar marker[10];
4673 guint8 marker_size;
4674 GstOggStreamType stream_type;
4675 } markers[] = {
4676 {
4677 "\001vorbis", 7, OGG_AUDIO}, {
4678 "\200theora", 7, OGG_VIDEO}, {
4679 "fLaC", 4, OGG_AUDIO}, {
4680 "\177FLAC", 5, OGG_AUDIO}, {
4681 "Speex", 5, OGG_AUDIO}, {
4682 "CMML\0\0\0\0", 8, OGG_OTHER}, {
4683 "PCM ", 8, OGG_AUDIO}, {
4684 "Annodex", 7, OGG_ANNODEX}, {
4685 "fishead", 7, OGG_SKELETON}, {
4686 "AnxData", 7, OGG_ANNODEX}, {
4687 "CELT ", 8, OGG_AUDIO}, {
4688 "\200kate\0\0\0", 8, OGG_KATE}, {
4689 "BBCD\0", 5, OGG_VIDEO}, {
4690 "OVP80\1\1", 7, OGG_VIDEO}, {
4691 "OpusHead", 8, OGG_AUDIO}, {
4692 "\001audio\0\0\0", 9, OGG_AUDIO}, {
4693 "\001video\0\0\0", 9, OGG_VIDEO}, {
4694 "\001text\0\0\0", 9, OGG_OTHER}
4695 };
4696
4697 while (c.offset < 4096 && data_scan_ctx_ensure_data (tf, &c, 64)) {
4698 guint size, i;
4699
4700 if (memcmp (c.data, "OggS", 5) != 0)
4701 break;
4702
4703 ++ogg_syncs;
4704
4705 /* check if BOS */
4706 if (c.data[5] != 0x02)
4707 break;
4708
4709 /* headers should only have one segment */
4710 if (c.data[26] != 1)
4711 break;
4712
4713 size = c.data[27];
4714 if (size < 8)
4715 break;
4716
4717 data_scan_ctx_advance (tf, &c, 28);
4718
4719 if (!data_scan_ctx_ensure_data (tf, &c, MAX (size, 8)))
4720 break;
4721
4722 for (i = 0; i < G_N_ELEMENTS (markers); ++i) {
4723 if (memcmp (c.data, markers[i].marker, markers[i].marker_size) == 0) {
4724 ++hdr_count[markers[i].stream_type];
4725 break;
4726 }
4727 }
4728
4729 if (i == G_N_ELEMENTS (markers)) {
4730 GST_MEMDUMP ("unknown Ogg stream marker", c.data, size);
4731 ++hdr_count[OGG_OTHER];
4732 }
4733
4734 data_scan_ctx_advance (tf, &c, size);
4735 }
4736
4737 if (ogg_syncs == 0)
4738 return;
4739
4740 /* We don't bother with annodex types. FIXME: what about XSPF? */
4741 if (hdr_count[OGG_VIDEO] > 0) {
4742 media_type = "video/ogg";
4743 } else if (hdr_count[OGG_AUDIO] > 0) {
4744 media_type = "audio/ogg";
4745 } else if (hdr_count[OGG_KATE] > 0 && hdr_count[OGG_OTHER] == 0) {
4746 media_type = "application/kate";
4747 } else {
4748 media_type = "application/ogg";
4749 }
4750
4751 GST_INFO ("found %s (audio:%u, video:%u, annodex:%u, skeleton:%u, other:%u)",
4752 media_type, hdr_count[OGG_AUDIO], hdr_count[OGG_VIDEO],
4753 hdr_count[OGG_ANNODEX], hdr_count[OGG_SKELETON], hdr_count[OGG_OTHER]);
4754
4755 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_MAXIMUM, media_type, NULL);
4756 }
4757
4758 /*** audio/x-vorbis ***/
4759 static GstStaticCaps vorbis_caps = GST_STATIC_CAPS ("audio/x-vorbis");
4760
4761 #define VORBIS_CAPS (gst_static_caps_get(&vorbis_caps))
4762 static void
vorbis_type_find(GstTypeFind * tf,gpointer private)4763 vorbis_type_find (GstTypeFind * tf, gpointer private)
4764 {
4765 const guint8 *data = gst_type_find_peek (tf, 0, 30);
4766
4767 if (data) {
4768 guint blocksize_0;
4769 guint blocksize_1;
4770
4771 /* 1 byte packet type (identification=0x01)
4772 6 byte string "vorbis"
4773 4 byte vorbis version */
4774 if (memcmp (data, "\001vorbis\000\000\000\000", 11) != 0)
4775 return;
4776 data += 11;
4777 /* 1 byte channels must be != 0 */
4778 if (data[0] == 0)
4779 return;
4780 data++;
4781 /* 4 byte samplerate must be != 0 */
4782 if (GST_READ_UINT32_LE (data) == 0)
4783 return;
4784 data += 16;
4785 /* blocksize checks */
4786 blocksize_0 = data[0] & 0x0F;
4787 blocksize_1 = (data[0] & 0xF0) >> 4;
4788 if (blocksize_0 > blocksize_1)
4789 return;
4790 if (blocksize_0 < 6 || blocksize_0 > 13)
4791 return;
4792 if (blocksize_1 < 6 || blocksize_1 > 13)
4793 return;
4794 data++;
4795 /* framing bit */
4796 if ((data[0] & 0x01) != 1)
4797 return;
4798 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, VORBIS_CAPS);
4799 }
4800 }
4801
4802 /*** video/x-theora ***/
4803
4804 static GstStaticCaps theora_caps = GST_STATIC_CAPS ("video/x-theora");
4805
4806 #define THEORA_CAPS (gst_static_caps_get(&theora_caps))
4807 static void
theora_type_find(GstTypeFind * tf,gpointer private)4808 theora_type_find (GstTypeFind * tf, gpointer private)
4809 {
4810 const guint8 *data = gst_type_find_peek (tf, 0, 7); //42);
4811
4812 if (data) {
4813 if (data[0] != 0x80)
4814 return;
4815 if (memcmp (&data[1], "theora", 6) != 0)
4816 return;
4817 /* FIXME: make this more reliable when specs are out */
4818
4819 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, THEORA_CAPS);
4820 }
4821 }
4822
4823 /*** kate ***/
4824 static void
kate_type_find(GstTypeFind * tf,gpointer private)4825 kate_type_find (GstTypeFind * tf, gpointer private)
4826 {
4827 const guint8 *data = gst_type_find_peek (tf, 0, 64);
4828 gchar category[16] = { 0, };
4829
4830 if (G_UNLIKELY (data == NULL))
4831 return;
4832
4833 /* see: http://wiki.xiph.org/index.php/OggKate#Format_specification */
4834 if (G_LIKELY (memcmp (data, "\200kate\0\0\0", 8) != 0))
4835 return;
4836
4837 /* make sure we always have a NUL-terminated string */
4838 memcpy (category, data + 48, 15);
4839 GST_LOG ("kate category: %s", category);
4840 /* canonical categories for subtitles: subtitles, spu-subtitles, SUB, K-SPU */
4841 if (strcmp (category, "subtitles") == 0 || strcmp (category, "SUB") == 0 ||
4842 strcmp (category, "spu-subtitles") == 0 ||
4843 strcmp (category, "K-SPU") == 0) {
4844 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_MAXIMUM,
4845 "subtitle/x-kate", NULL);
4846 } else {
4847 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_MAXIMUM,
4848 "application/x-kate", NULL);
4849 }
4850 }
4851
4852 /*** WEBVTTT subtitles ***/
4853 static GstStaticCaps webvtt_caps =
4854 GST_STATIC_CAPS ("application/x-subtitle-vtt, parsed=(boolean)false");
4855 #define WEBVTT_CAPS (gst_static_caps_get(&webvtt_caps))
4856
4857 static void
webvtt_type_find(GstTypeFind * tf,gpointer private)4858 webvtt_type_find (GstTypeFind * tf, gpointer private)
4859 {
4860 const guint8 *data;
4861
4862 data = gst_type_find_peek (tf, 0, 10);
4863
4864 if (data == NULL)
4865 return;
4866
4867 /* there might be a UTF-8 BOM at the beginning */
4868 if (memcmp (data, "WEBVTT", 6) != 0 && memcmp (data + 3, "WEBVTT", 6) != 0) {
4869 return;
4870 }
4871
4872 if (data[0] != 'W') {
4873 if (data[0] != 0xef || data[1] != 0xbb || data[2] != 0xbf)
4874 return; /* Not a UTF-8 BOM */
4875 data += 3;
4876 }
4877
4878 /* After the WEBVTT magic must be one of these chars:
4879 * 0x20 (space), 0x9 (tab), 0xa (LF) or 0xd (CR) */
4880 if (data[6] != 0x20 && data[6] != 0x9 && data[6] != 0xa && data[6] != 0xd) {
4881 return;
4882 }
4883
4884 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, WEBVTT_CAPS);
4885 }
4886
4887 /*** application/x-ogm-video or audio***/
4888
4889 static GstStaticCaps ogmvideo_caps =
4890 GST_STATIC_CAPS ("application/x-ogm-video");
4891 #define OGMVIDEO_CAPS (gst_static_caps_get(&ogmvideo_caps))
4892 static void
ogmvideo_type_find(GstTypeFind * tf,gpointer private)4893 ogmvideo_type_find (GstTypeFind * tf, gpointer private)
4894 {
4895 const guint8 *data = gst_type_find_peek (tf, 0, 9);
4896
4897 if (data) {
4898 if (memcmp (data, "\001video\000\000\000", 9) != 0)
4899 return;
4900 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, OGMVIDEO_CAPS);
4901 }
4902 }
4903
4904 static GstStaticCaps ogmaudio_caps =
4905 GST_STATIC_CAPS ("application/x-ogm-audio");
4906 #define OGMAUDIO_CAPS (gst_static_caps_get(&ogmaudio_caps))
4907 static void
ogmaudio_type_find(GstTypeFind * tf,gpointer private)4908 ogmaudio_type_find (GstTypeFind * tf, gpointer private)
4909 {
4910 const guint8 *data = gst_type_find_peek (tf, 0, 9);
4911
4912 if (data) {
4913 if (memcmp (data, "\001audio\000\000\000", 9) != 0)
4914 return;
4915 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, OGMAUDIO_CAPS);
4916 }
4917 }
4918
4919 static GstStaticCaps ogmtext_caps = GST_STATIC_CAPS ("application/x-ogm-text");
4920
4921 #define OGMTEXT_CAPS (gst_static_caps_get(&ogmtext_caps))
4922 static void
ogmtext_type_find(GstTypeFind * tf,gpointer private)4923 ogmtext_type_find (GstTypeFind * tf, gpointer private)
4924 {
4925 const guint8 *data = gst_type_find_peek (tf, 0, 9);
4926
4927 if (data) {
4928 if (memcmp (data, "\001text\000\000\000\000", 9) != 0)
4929 return;
4930 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, OGMTEXT_CAPS);
4931 }
4932 }
4933
4934 /*** audio/x-speex ***/
4935
4936 static GstStaticCaps speex_caps = GST_STATIC_CAPS ("audio/x-speex");
4937
4938 #define SPEEX_CAPS (gst_static_caps_get(&speex_caps))
4939 static void
speex_type_find(GstTypeFind * tf,gpointer private)4940 speex_type_find (GstTypeFind * tf, gpointer private)
4941 {
4942 const guint8 *data = gst_type_find_peek (tf, 0, 80);
4943
4944 if (data) {
4945 /* 8 byte string "Speex "
4946 24 byte speex version string + int */
4947 if (memcmp (data, "Speex ", 8) != 0)
4948 return;
4949 data += 32;
4950
4951 /* 4 byte header size >= 80 */
4952 if (GST_READ_UINT32_LE (data) < 80)
4953 return;
4954 data += 4;
4955
4956 /* 4 byte sample rate <= 48000 */
4957 if (GST_READ_UINT32_LE (data) > 48000)
4958 return;
4959 data += 4;
4960
4961 /* currently there are only 3 speex modes. */
4962 if (GST_READ_UINT32_LE (data) > 3)
4963 return;
4964 data += 12;
4965
4966 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, SPEEX_CAPS);
4967 }
4968 }
4969
4970 /*** audio/x-celt ***/
4971
4972 static GstStaticCaps celt_caps = GST_STATIC_CAPS ("audio/x-celt");
4973
4974 #define CELT_CAPS (gst_static_caps_get(&celt_caps))
4975 static void
celt_type_find(GstTypeFind * tf,gpointer private)4976 celt_type_find (GstTypeFind * tf, gpointer private)
4977 {
4978 const guint8 *data = gst_type_find_peek (tf, 0, 8);
4979
4980 if (data) {
4981 /* 8 byte string "CELT " */
4982 if (memcmp (data, "CELT ", 8) != 0)
4983 return;
4984
4985 /* TODO: Check other values of the CELT header */
4986 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, CELT_CAPS);
4987 }
4988 }
4989
4990 /*** application/x-ogg-skeleton ***/
4991 static GstStaticCaps ogg_skeleton_caps =
4992 GST_STATIC_CAPS ("application/x-ogg-skeleton, parsed=(boolean)FALSE");
4993 #define OGG_SKELETON_CAPS (gst_static_caps_get(&ogg_skeleton_caps))
4994 static void
oggskel_type_find(GstTypeFind * tf,gpointer private)4995 oggskel_type_find (GstTypeFind * tf, gpointer private)
4996 {
4997 const guint8 *data = gst_type_find_peek (tf, 0, 12);
4998
4999 if (data) {
5000 /* 8 byte string "fishead\0" for the ogg skeleton stream */
5001 if (memcmp (data, "fishead\0", 8) != 0)
5002 return;
5003 data += 8;
5004
5005 /* Require that the header contains version 3.0 */
5006 if (GST_READ_UINT16_LE (data) != 3)
5007 return;
5008 data += 2;
5009 if (GST_READ_UINT16_LE (data) != 0)
5010 return;
5011
5012 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, OGG_SKELETON_CAPS);
5013 }
5014 }
5015
5016 static GstStaticCaps cmml_caps = GST_STATIC_CAPS ("text/x-cmml");
5017
5018 #define CMML_CAPS (gst_static_caps_get(&cmml_caps))
5019 static void
cmml_type_find(GstTypeFind * tf,gpointer private)5020 cmml_type_find (GstTypeFind * tf, gpointer private)
5021 {
5022 /* Header is 12 bytes minimum (though we don't check the minor version */
5023 const guint8 *data = gst_type_find_peek (tf, 0, 12);
5024
5025 if (data) {
5026
5027 /* 8 byte string "CMML\0\0\0\0" for the magic number */
5028 if (memcmp (data, "CMML\0\0\0\0", 8) != 0)
5029 return;
5030 data += 8;
5031
5032 /* Require that the header contains at least version 2.0 */
5033 if (GST_READ_UINT16_LE (data) < 2)
5034 return;
5035
5036 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, CMML_CAPS);
5037 }
5038 }
5039
5040 /*** application/x-tar ***/
5041
5042 static GstStaticCaps tar_caps = GST_STATIC_CAPS ("application/x-tar");
5043
5044 #define TAR_CAPS (gst_static_caps_get(&tar_caps))
5045 #define OLDGNU_MAGIC "ustar " /* 7 chars and a NUL */
5046 #define NEWGNU_MAGIC "ustar" /* 5 chars and a NUL */
5047 static void
tar_type_find(GstTypeFind * tf,gpointer unused)5048 tar_type_find (GstTypeFind * tf, gpointer unused)
5049 {
5050 const guint8 *data = gst_type_find_peek (tf, 257, 8);
5051
5052 /* of course we are not certain, but we don't want other typefind funcs
5053 * to detect formats of files within the tar archive, e.g. mp3s */
5054 if (data) {
5055 if (memcmp (data, OLDGNU_MAGIC, 8) == 0) { /* sic */
5056 gst_type_find_suggest (tf, GST_TYPE_FIND_NEARLY_CERTAIN, TAR_CAPS);
5057 } else if (memcmp (data, NEWGNU_MAGIC, 6) == 0 && /* sic */
5058 g_ascii_isdigit (data[6]) && g_ascii_isdigit (data[7])) {
5059 gst_type_find_suggest (tf, GST_TYPE_FIND_NEARLY_CERTAIN, TAR_CAPS);
5060 }
5061 }
5062 }
5063
5064 /*** application/x-ar ***/
5065
5066 static GstStaticCaps ar_caps = GST_STATIC_CAPS ("application/x-ar");
5067
5068 #define AR_CAPS (gst_static_caps_get(&ar_caps))
5069 static void
ar_type_find(GstTypeFind * tf,gpointer unused)5070 ar_type_find (GstTypeFind * tf, gpointer unused)
5071 {
5072 const guint8 *data = gst_type_find_peek (tf, 0, 24);
5073
5074 if (data && memcmp (data, "!<arch>", 7) == 0) {
5075 gint i;
5076
5077 for (i = 7; i < 24; ++i) {
5078 if (!g_ascii_isprint (data[i]) && data[i] != '\n') {
5079 gst_type_find_suggest (tf, GST_TYPE_FIND_POSSIBLE, AR_CAPS);
5080 }
5081 }
5082
5083 gst_type_find_suggest (tf, GST_TYPE_FIND_NEARLY_CERTAIN, AR_CAPS);
5084 }
5085 }
5086
5087 /*** audio/x-au ***/
5088
5089 /* NOTE: we cannot replace this function with TYPE_FIND_REGISTER_START_WITH,
5090 * as it is only possible to register one typefind factory per 'name'
5091 * (which is in this case the caps), and the first one would be replaced by
5092 * the second one. */
5093 static GstStaticCaps au_caps = GST_STATIC_CAPS ("audio/x-au");
5094
5095 #define AU_CAPS (gst_static_caps_get(&au_caps))
5096 static void
au_type_find(GstTypeFind * tf,gpointer unused)5097 au_type_find (GstTypeFind * tf, gpointer unused)
5098 {
5099 const guint8 *data = gst_type_find_peek (tf, 0, 4);
5100
5101 if (data) {
5102 if (memcmp (data, ".snd", 4) == 0 || memcmp (data, "dns.", 4) == 0) {
5103 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, AU_CAPS);
5104 }
5105 }
5106 }
5107
5108
5109 /*** video/x-nuv ***/
5110
5111 /* NOTE: we cannot replace this function with TYPE_FIND_REGISTER_START_WITH,
5112 * as it is only possible to register one typefind factory per 'name'
5113 * (which is in this case the caps), and the first one would be replaced by
5114 * the second one. */
5115 static GstStaticCaps nuv_caps = GST_STATIC_CAPS ("video/x-nuv");
5116
5117 #define NUV_CAPS (gst_static_caps_get(&nuv_caps))
5118 static void
nuv_type_find(GstTypeFind * tf,gpointer unused)5119 nuv_type_find (GstTypeFind * tf, gpointer unused)
5120 {
5121 const guint8 *data = gst_type_find_peek (tf, 0, 11);
5122
5123 if (data) {
5124 if (memcmp (data, "MythTVVideo", 11) == 0
5125 || memcmp (data, "NuppelVideo", 11) == 0) {
5126 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, NUV_CAPS);
5127 }
5128 }
5129 }
5130
5131 /*** audio/x-paris ***/
5132 /* NOTE: do not replace this function with two TYPE_FIND_REGISTER_START_WITH */
5133 static GstStaticCaps paris_caps = GST_STATIC_CAPS ("audio/x-paris");
5134
5135 #define PARIS_CAPS (gst_static_caps_get(&paris_caps))
5136 static void
paris_type_find(GstTypeFind * tf,gpointer unused)5137 paris_type_find (GstTypeFind * tf, gpointer unused)
5138 {
5139 const guint8 *data = gst_type_find_peek (tf, 0, 4);
5140
5141 if (data) {
5142 if (memcmp (data, " paf", 4) == 0 || memcmp (data, "fap ", 4) == 0) {
5143 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, PARIS_CAPS);
5144 }
5145 }
5146 }
5147
5148 /*** audio/x-sbc ***/
5149 static GstStaticCaps sbc_caps = GST_STATIC_CAPS ("audio/x-sbc");
5150 #define SBC_CAPS (gst_static_caps_get(&sbc_caps))
5151
5152 static gsize
sbc_check_header(const guint8 * data,gsize len,guint * rate,guint * channels)5153 sbc_check_header (const guint8 * data, gsize len, guint * rate,
5154 guint * channels)
5155 {
5156 static const guint16 sbc_rates[4] = { 16000, 32000, 44100, 48000 };
5157 static const guint8 sbc_blocks[4] = { 4, 8, 12, 16 };
5158 guint n_blocks, ch_mode, n_subbands, bitpool;
5159
5160 if (data[0] != 0x9C || len < 4)
5161 return 0;
5162
5163 n_blocks = sbc_blocks[(data[1] >> 4) & 0x03];
5164 ch_mode = (data[1] >> 2) & 0x03;
5165 n_subbands = (data[1] & 0x01) ? 8 : 4;
5166 bitpool = data[2];
5167 if (bitpool < 2)
5168 return 0;
5169
5170 *rate = sbc_rates[(data[1] >> 6) & 0x03];
5171 *channels = (ch_mode == 0) ? 1 : 2;
5172
5173 if (ch_mode == 0)
5174 return 4 + (n_subbands * 1) / 2 + (n_blocks * 1 * bitpool) / 8;
5175 else if (ch_mode == 1)
5176 return 4 + (n_subbands * 2) / 2 + (n_blocks * 2 * bitpool) / 8;
5177 else if (ch_mode == 2)
5178 return 4 + (n_subbands * 2) / 2 + (n_blocks * bitpool) / 8;
5179 else if (ch_mode == 3)
5180 return 4 + (n_subbands * 2) / 2 + (n_subbands + n_blocks * bitpool) / 8;
5181
5182 return 0;
5183 }
5184
5185 static void
sbc_type_find(GstTypeFind * tf,gpointer unused)5186 sbc_type_find (GstTypeFind * tf, gpointer unused)
5187 {
5188 const guint8 *data;
5189 gsize frame_len;
5190 guint i, rate, channels, offset = 0;
5191
5192 for (i = 0; i < 10; ++i) {
5193 data = gst_type_find_peek (tf, offset, 8);
5194 if (data == NULL)
5195 return;
5196
5197 frame_len = sbc_check_header (data, 8, &rate, &channels);
5198 if (frame_len == 0)
5199 return;
5200
5201 offset += frame_len;
5202 }
5203 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_POSSIBLE, "audio/x-sbc",
5204 "rate", G_TYPE_INT, rate, "channels", G_TYPE_INT, channels,
5205 "parsed", G_TYPE_BOOLEAN, FALSE, NULL);
5206 }
5207
5208 /*** audio/iLBC-sh ***/
5209 /* NOTE: do not replace this function with two TYPE_FIND_REGISTER_START_WITH */
5210 static GstStaticCaps ilbc_caps = GST_STATIC_CAPS ("audio/iLBC-sh");
5211
5212 #define ILBC_CAPS (gst_static_caps_get(&ilbc_caps))
5213 static void
ilbc_type_find(GstTypeFind * tf,gpointer unused)5214 ilbc_type_find (GstTypeFind * tf, gpointer unused)
5215 {
5216 const guint8 *data = gst_type_find_peek (tf, 0, 8);
5217
5218 if (data) {
5219 if (memcmp (data, "#!iLBC30", 8) == 0 || memcmp (data, "#!iLBC20", 8) == 0) {
5220 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY, ILBC_CAPS);
5221 }
5222 }
5223 }
5224
5225 /*** application/x-ms-dos-executable ***/
5226
5227 static GstStaticCaps msdos_caps =
5228 GST_STATIC_CAPS ("application/x-ms-dos-executable");
5229 #define MSDOS_CAPS (gst_static_caps_get(&msdos_caps))
5230 /* see http://www.madchat.org/vxdevl/papers/winsys/pefile/pefile.htm */
5231 static void
msdos_type_find(GstTypeFind * tf,gpointer unused)5232 msdos_type_find (GstTypeFind * tf, gpointer unused)
5233 {
5234 const guint8 *data = gst_type_find_peek (tf, 0, 64);
5235
5236 if (data && data[0] == 'M' && data[1] == 'Z' &&
5237 GST_READ_UINT16_LE (data + 8) == 4) {
5238 guint32 pe_offset = GST_READ_UINT32_LE (data + 60);
5239
5240 data = gst_type_find_peek (tf, pe_offset, 2);
5241 if (data && data[0] == 'P' && data[1] == 'E') {
5242 gst_type_find_suggest (tf, GST_TYPE_FIND_NEARLY_CERTAIN, MSDOS_CAPS);
5243 }
5244 }
5245 }
5246
5247 /*** application/x-mmsh ***/
5248
5249 static GstStaticCaps mmsh_caps = GST_STATIC_CAPS ("application/x-mmsh");
5250
5251 #define MMSH_CAPS gst_static_caps_get(&mmsh_caps)
5252
5253 /* This is to recognise mssh-over-http */
5254 static void
mmsh_type_find(GstTypeFind * tf,gpointer unused)5255 mmsh_type_find (GstTypeFind * tf, gpointer unused)
5256 {
5257 static const guint8 asf_marker[16] = { 0x30, 0x26, 0xb2, 0x75, 0x8e, 0x66,
5258 0xcf, 0x11, 0xa6, 0xd9, 0x00, 0xaa, 0x00, 0x62, 0xce, 0x6c
5259 };
5260
5261 const guint8 *data;
5262
5263 data = gst_type_find_peek (tf, 0, 2 + 2 + 4 + 2 + 2 + 16);
5264 if (data && data[0] == 0x24 && data[1] == 0x48 &&
5265 GST_READ_UINT16_LE (data + 2) > 2 + 2 + 4 + 2 + 2 + 16 &&
5266 memcmp (data + 2 + 2 + 4 + 2 + 2, asf_marker, 16) == 0) {
5267 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY, MMSH_CAPS);
5268 }
5269 }
5270
5271 /*** video/x-dirac ***/
5272
5273 /* NOTE: we cannot replace this function with TYPE_FIND_REGISTER_START_WITH,
5274 * as it is only possible to register one typefind factory per 'name'
5275 * (which is in this case the caps), and the first one would be replaced by
5276 * the second one. */
5277 static GstStaticCaps dirac_caps = GST_STATIC_CAPS ("video/x-dirac");
5278
5279 #define DIRAC_CAPS (gst_static_caps_get(&dirac_caps))
5280 static void
dirac_type_find(GstTypeFind * tf,gpointer unused)5281 dirac_type_find (GstTypeFind * tf, gpointer unused)
5282 {
5283 const guint8 *data = gst_type_find_peek (tf, 0, 8);
5284
5285 if (data) {
5286 if (memcmp (data, "BBCD", 4) == 0 || memcmp (data, "KW-DIRAC", 8) == 0) {
5287 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, DIRAC_CAPS);
5288 }
5289 }
5290 }
5291
5292 /*** audio/x-tap-tap ***/
5293
5294 /* NOTE: we cannot replace this function with TYPE_FIND_REGISTER_START_WITH,
5295 * as it is only possible to register one typefind factory per 'name'
5296 * (which is in this case the caps), and the first one would be replaced by
5297 * the second one. */
5298 static GstStaticCaps tap_caps = GST_STATIC_CAPS ("audio/x-tap-tap");
5299
5300 #define TAP_CAPS (gst_static_caps_get(&tap_caps))
5301 static void
tap_type_find(GstTypeFind * tf,gpointer unused)5302 tap_type_find (GstTypeFind * tf, gpointer unused)
5303 {
5304 const guint8 *data = gst_type_find_peek (tf, 0, 16);
5305
5306 if (data) {
5307 if (memcmp (data, "C64-TAPE-RAW", 12) == 0
5308 || memcmp (data, "C16-TAPE-RAW", 12) == 0) {
5309 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, TAP_CAPS);
5310 }
5311 }
5312 }
5313
5314 /*** video/vivo ***/
5315
5316 static GstStaticCaps vivo_caps = GST_STATIC_CAPS ("video/vivo");
5317
5318 #define VIVO_CAPS gst_static_caps_get(&vivo_caps)
5319
5320 static void
vivo_type_find(GstTypeFind * tf,gpointer unused)5321 vivo_type_find (GstTypeFind * tf, gpointer unused)
5322 {
5323 static const guint8 vivo_marker[] = { 'V', 'e', 'r', 's', 'i', 'o', 'n',
5324 ':', 'V', 'i', 'v', 'o', '/'
5325 };
5326 const guint8 *data;
5327 guint hdr_len, pos;
5328
5329 data = gst_type_find_peek (tf, 0, 1024);
5330 if (data == NULL || data[0] != 0x00)
5331 return;
5332
5333 if ((data[1] & 0x80)) {
5334 if ((data[2] & 0x80))
5335 return;
5336 hdr_len = ((guint) (data[1] & 0x7f)) << 7;
5337 hdr_len += data[2];
5338 if (hdr_len > 2048)
5339 return;
5340 pos = 3;
5341 } else {
5342 hdr_len = data[1];
5343 pos = 2;
5344 }
5345
5346 /* 1008 = 1022 - strlen ("Version:Vivo/") - 1 */
5347 while (pos < 1008 && data[pos] == '\r' && data[pos + 1] == '\n')
5348 pos += 2;
5349
5350 if (memcmp (data + pos, vivo_marker, sizeof (vivo_marker)) == 0) {
5351 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, VIVO_CAPS);
5352 }
5353 }
5354
5355 /*** XDG MIME typefinder (to avoid false positives mostly) ***/
5356
5357 #ifdef USE_GIO
5358 static void
xdgmime_typefind(GstTypeFind * find,gpointer user_data)5359 xdgmime_typefind (GstTypeFind * find, gpointer user_data)
5360 {
5361 gchar *mimetype;
5362 gsize length = 16384;
5363 guint64 tf_length;
5364 const guint8 *data;
5365 gchar *tmp;
5366
5367 if ((tf_length = gst_type_find_get_length (find)) > 0)
5368 length = MIN (length, tf_length);
5369
5370 if ((data = gst_type_find_peek (find, 0, length)) == NULL)
5371 return;
5372
5373 tmp = g_content_type_guess (NULL, data, length, NULL);
5374 if (tmp == NULL || g_content_type_is_unknown (tmp)) {
5375 g_free (tmp);
5376 return;
5377 }
5378
5379 mimetype = g_content_type_get_mime_type (tmp);
5380 g_free (tmp);
5381
5382 if (mimetype == NULL)
5383 return;
5384
5385 GST_DEBUG ("Got mimetype '%s'", mimetype);
5386
5387 /* Ignore audio/video types:
5388 * - our own typefinders in -base are likely to be better at this
5389 * (and if they're not, we really want to fix them, that's why we don't
5390 * report xdg-detected audio/video types at all, not even with a low
5391 * probability)
5392 * - we want to detect GStreamer media types and not MIME types
5393 * - the purpose of this xdg mime finder is mainly to prevent false
5394 * positives of non-media formats, not to typefind audio/video formats */
5395 if (g_str_has_prefix (mimetype, "audio/") ||
5396 g_str_has_prefix (mimetype, "video/")) {
5397 GST_LOG ("Ignoring audio/video mime type");
5398 g_free (mimetype);
5399 return;
5400 }
5401
5402 /* Again, we mainly want the xdg typefinding to prevent false-positives on
5403 * non-media formats, so suggest the type with a probability that trumps
5404 * uncertain results of our typefinders, but not more than that. */
5405 GST_LOG ("Suggesting '%s' with probability POSSIBLE", mimetype);
5406 gst_type_find_suggest_simple (find, GST_TYPE_FIND_POSSIBLE, mimetype, NULL);
5407 g_free (mimetype);
5408 }
5409 #endif /* USE_GIO */
5410
5411 /*** Windows icon typefinder (to avoid false positives mostly) ***/
5412
5413 static void
windows_icon_typefind(GstTypeFind * find,gpointer user_data)5414 windows_icon_typefind (GstTypeFind * find, gpointer user_data)
5415 {
5416 const guint8 *data;
5417 gint64 datalen;
5418 guint16 type, nimages;
5419 gint32 size, offset;
5420
5421 datalen = gst_type_find_get_length (find);
5422 if (datalen < 22)
5423 return;
5424 if ((data = gst_type_find_peek (find, 0, 6)) == NULL)
5425 return;
5426
5427 /* header - simple and not enough to rely on it alone */
5428 if (GST_READ_UINT16_LE (data) != 0)
5429 return;
5430 type = GST_READ_UINT16_LE (data + 2);
5431 if (type != 1 && type != 2)
5432 return;
5433 nimages = GST_READ_UINT16_LE (data + 4);
5434 if (nimages == 0) /* we can assume we can't have an empty image file ? */
5435 return;
5436
5437 /* first image */
5438 if (data[6 + 3] != 0)
5439 return;
5440 if (type == 1) {
5441 guint16 planes = GST_READ_UINT16_LE (data + 6 + 4);
5442 if (planes > 1)
5443 return;
5444 }
5445 size = GST_READ_UINT32_LE (data + 6 + 8);
5446 offset = GST_READ_UINT32_LE (data + 6 + 12);
5447 if (offset < 0 || size <= 0 || size >= datalen || offset >= datalen
5448 || size + offset > datalen)
5449 return;
5450
5451 gst_type_find_suggest_simple (find, GST_TYPE_FIND_NEARLY_CERTAIN,
5452 "image/x-icon", NULL);
5453 }
5454
5455 /*** WAP WBMP typefinder ***/
5456
5457 static void
wbmp_typefind(GstTypeFind * find,gpointer user_data)5458 wbmp_typefind (GstTypeFind * find, gpointer user_data)
5459 {
5460 const guint8 *data;
5461 gint64 datalen;
5462 guint w, h, size;
5463
5464 /* http://en.wikipedia.org/wiki/Wireless_Application_Protocol_Bitmap_Format */
5465 datalen = gst_type_find_get_length (find);
5466 if (datalen == 0)
5467 return;
5468
5469 data = gst_type_find_peek (find, 0, 5);
5470 if (data == NULL)
5471 return;
5472
5473 /* want 0x00 0x00 at start */
5474 if (*data++ != 0 || *data++ != 0)
5475 return;
5476
5477 /* min header size */
5478 size = 4;
5479
5480 /* let's assume max width/height is 65536 */
5481 w = *data++;
5482 if ((w & 0x80)) {
5483 w = (w << 8) | *data++;
5484 if ((w & 0x80))
5485 return;
5486 ++size;
5487 data = gst_type_find_peek (find, 4, 2);
5488 if (data == NULL)
5489 return;
5490 }
5491 h = *data++;
5492 if ((h & 0x80)) {
5493 h = (h << 8) | *data++;
5494 if ((h & 0x80))
5495 return;
5496 ++size;
5497 }
5498
5499 if (w == 0 || h == 0)
5500 return;
5501
5502 /* now add bitmap size */
5503 size += h * (GST_ROUND_UP_8 (w) / 8);
5504
5505 if (datalen == size) {
5506 gst_type_find_suggest_simple (find, GST_TYPE_FIND_POSSIBLE - 10,
5507 "image/vnd.wap.wbmp", NULL);
5508 }
5509 }
5510
5511 /*** DEGAS Atari images (also to avoid false positives, see #625129) ***/
5512 static void
degas_type_find(GstTypeFind * tf,gpointer private)5513 degas_type_find (GstTypeFind * tf, gpointer private)
5514 {
5515 /* No magic, but it should have a fixed size and a few invalid values */
5516 /* http://www.fileformat.info/format/atari/spec/6ecf9f6eb5be494284a47feb8a214687/view.htm */
5517 gint64 len;
5518 const guint8 *data;
5519 guint16 resolution;
5520 int n;
5521
5522 len = gst_type_find_get_length (tf);
5523 if (len < 34) /* smallest header of the lot */
5524 return;
5525 data = gst_type_find_peek (tf, 0, 4);
5526 if (G_UNLIKELY (data == NULL))
5527 return;
5528 resolution = GST_READ_UINT16_BE (data);
5529 if (len == 32034) {
5530 /* could be DEGAS */
5531 if (resolution <= 2)
5532 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_POSSIBLE + 5,
5533 "image/x-degas", NULL);
5534 } else if (len == 32066) {
5535 /* could be DEGAS Elite */
5536 if (resolution <= 2) {
5537 data = gst_type_find_peek (tf, len - 16, 8);
5538 if (G_UNLIKELY (data == NULL))
5539 return;
5540 for (n = 0; n < 4; n++) {
5541 if (GST_READ_UINT16_BE (data + n * 2) > 2)
5542 return;
5543 }
5544 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_POSSIBLE + 5,
5545 "image/x-degas", NULL);
5546 }
5547 } else if (len >= 66 && len < 32066) {
5548 /* could be compressed DEGAS Elite, but it's compressed and so we can't rely on size,
5549 it does have 4 16 bytes values near the end that are 0-2 though. */
5550 if ((resolution & 0x8000) && (resolution & 0x7fff) <= 2) {
5551 data = gst_type_find_peek (tf, len - 16, 8);
5552 if (G_UNLIKELY (data == NULL))
5553 return;
5554 for (n = 0; n < 4; n++) {
5555 if (GST_READ_UINT16_BE (data + n * 2) > 2)
5556 return;
5557 }
5558 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_POSSIBLE + 5,
5559 "image/x-degas", NULL);
5560 }
5561 }
5562 }
5563
5564 /*** y4m ***/
5565
5566 static void
y4m_typefind(GstTypeFind * tf,gpointer private)5567 y4m_typefind (GstTypeFind * tf, gpointer private)
5568 {
5569 const guint8 *data;
5570
5571 data = gst_type_find_peek (tf, 0, 10);
5572 if (data != NULL && memcmp (data, "YUV4MPEG2 ", 10) == 0) {
5573 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_LIKELY,
5574 "application/x-yuv4mpeg", "y4mversion", G_TYPE_INT, 2, NULL);
5575 }
5576 }
5577
5578 /*** DVD ISO images (looks like H.264, see #674069) ***/
5579 static void
dvdiso_type_find(GstTypeFind * tf,gpointer private)5580 dvdiso_type_find (GstTypeFind * tf, gpointer private)
5581 {
5582 /* 0x8000 bytes of zeros, then "\001CD001" */
5583 gint64 len;
5584 const guint8 *data;
5585
5586 len = gst_type_find_get_length (tf);
5587 if (len < 0x8006)
5588 return;
5589 data = gst_type_find_peek (tf, 0, 0x8006);
5590 if (G_UNLIKELY (data == NULL))
5591 return;
5592 for (len = 0; len < 0x8000; len++)
5593 if (data[len])
5594 return;
5595 /* Can the '1' be anything else ? My three samples all have '1'. */
5596 if (memcmp (data + 0x8000, "\001CD001", 6))
5597 return;
5598
5599 /* May need more inspection, we may be able to demux some of them */
5600 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_LIKELY,
5601 "application/octet-stream", NULL);
5602 }
5603
5604 /* SSA/ASS subtitles
5605 *
5606 * http://en.wikipedia.org/wiki/SubStation_Alpha
5607 * http://matroska.org/technical/specs/subtitles/ssa.html
5608 */
5609 static void
ssa_type_find(GstTypeFind * tf,gpointer private)5610 ssa_type_find (GstTypeFind * tf, gpointer private)
5611 {
5612 const gchar *start, *end, *ver_str, *media_type = NULL;
5613 const guint8 *data;
5614 gchar *str, *script_type, *p = NULL;
5615 gint64 len;
5616
5617 data = gst_type_find_peek (tf, 0, 32);
5618
5619 if (data == NULL)
5620 return;
5621
5622 /* FIXME: detect utf-16/32 BOM and convert before typefinding the rest */
5623
5624 /* there might be a UTF-8 BOM at the beginning */
5625 if (memcmp (data, "[Script Info]", 13) != 0 &&
5626 memcmp (data + 3, "[Script Info]", 13) != 0) {
5627 return;
5628 }
5629
5630 /* now check if we have SSA or ASS */
5631 len = gst_type_find_get_length (tf);
5632 if (len > 8192)
5633 len = 8192;
5634
5635 data = gst_type_find_peek (tf, 0, len);
5636 if (data == NULL)
5637 return;
5638
5639 /* skip BOM */
5640 start = (gchar *) memchr (data, '[', 5);
5641 g_assert (start);
5642 len -= (start - (gchar *) data);
5643
5644 /* ignore anything non-UTF8 for now, in future we might at least allow
5645 * other UTF variants that are clearly prefixed with the appropriate BOM */
5646 if (!g_utf8_validate (start, len, &end) && (len - (end - start)) > 6) {
5647 GST_FIXME ("non-UTF8 SSA/ASS file");
5648 return;
5649 }
5650
5651 /* something at start, but not a UTF-8 BOM? */
5652 if (data[0] != '[' && (data[0] != 0xEF || data[1] != 0xBB || data[2] != 0xBF))
5653 return;
5654
5655 /* ignore any partial UTF-8 characters at the end */
5656 len = end - start;
5657
5658 /* create a NUL-terminated string so it's easier to process it safely */
5659 str = g_strndup (start, len - 1);
5660 script_type = strstr (str, "ScriptType:");
5661 if (script_type != NULL) {
5662 gdouble version;
5663
5664 ver_str = script_type + 11;
5665 while (*ver_str == ' ' || *ver_str == 'v' || *ver_str == 'V')
5666 ++ver_str;
5667 version = g_ascii_strtod (ver_str, &p);
5668 if (version == 4.0 && p != NULL && *p == '+')
5669 media_type = "application/x-ass";
5670 else if (version >= 1.0 && version <= 4.0)
5671 media_type = "application/x-ssa";
5672 }
5673
5674 if (media_type == NULL) {
5675 if (strstr (str, "[v4+ Styles]") || strstr (str, "[V4+ Styles]"))
5676 media_type = "application/x-ass";
5677 else if (strstr (str, "[v4 Styles]") || strstr (str, "[V4 Styles]"))
5678 media_type = "application/x-ssa";
5679 }
5680
5681 if (media_type != NULL) {
5682 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_MAXIMUM,
5683 media_type, "parsed", G_TYPE_BOOLEAN, FALSE, NULL);
5684 } else {
5685 GST_WARNING ("could not detect SSA/ASS variant");
5686 }
5687
5688 g_free (str);
5689 }
5690
5691 /*** application/x-mcc ***/
5692 static GstStaticCaps mcc_caps = GST_STATIC_CAPS ("application/x-mcc");
5693
5694 #define MCC_CAPS gst_static_caps_get(&mcc_caps)
5695
5696 static void
mcc_type_find(GstTypeFind * tf,gpointer private)5697 mcc_type_find (GstTypeFind * tf, gpointer private)
5698 {
5699 const guint8 *data;
5700
5701 data = gst_type_find_peek (tf, 0, 31);
5702
5703 if (data == NULL)
5704 return;
5705
5706 /* MCC files always start with this followed by the version */
5707 if (memcmp (data, "File Format=MacCaption_MCC V", 28) != 0 ||
5708 !g_ascii_isdigit (data[28]) || data[29] != '.' ||
5709 !g_ascii_isdigit (data[30])) {
5710 return;
5711 }
5712
5713 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_MAXIMUM,
5714 "application/x-mcc", "version", G_TYPE_INT, data[28] - '0', NULL);
5715 }
5716
5717 /*** video/x-pva ***/
5718
5719 static GstStaticCaps pva_caps = GST_STATIC_CAPS ("video/x-pva");
5720
5721 #define PVA_CAPS gst_static_caps_get(&pva_caps)
5722
5723 static void
pva_type_find(GstTypeFind * tf,gpointer private)5724 pva_type_find (GstTypeFind * tf, gpointer private)
5725 {
5726 const guint8 *data;
5727
5728 data = gst_type_find_peek (tf, 0, 5);
5729
5730 if (data == NULL)
5731 return;
5732
5733 if (data[0] == 'A' && data[1] == 'V' && data[2] < 3 && data[4] == 0x55)
5734 gst_type_find_suggest (tf, GST_TYPE_FIND_NEARLY_CERTAIN, PVA_CAPS);
5735 }
5736
5737 /*** audio/audible ***/
5738
5739 /* derived from pyaudibletags
5740 * http://code.google.com/p/pyaudibletags/source/browse/trunk/pyaudibletags.py
5741 */
5742 static GstStaticCaps aa_caps = GST_STATIC_CAPS ("audio/x-audible");
5743
5744 #define AA_CAPS gst_static_caps_get(&aa_caps)
5745
5746 static void
aa_type_find(GstTypeFind * tf,gpointer private)5747 aa_type_find (GstTypeFind * tf, gpointer private)
5748 {
5749 const guint8 *data;
5750
5751 data = gst_type_find_peek (tf, 0, 12);
5752 if (data == NULL)
5753 return;
5754
5755 if (GST_READ_UINT32_BE (data + 4) == 0x57907536) {
5756 guint64 media_len;
5757
5758 media_len = gst_type_find_get_length (tf);
5759 if (media_len > 0 && GST_READ_UINT32_BE (data) == media_len)
5760 gst_type_find_suggest (tf, GST_TYPE_FIND_NEARLY_CERTAIN, AA_CAPS);
5761 else
5762 gst_type_find_suggest (tf, GST_TYPE_FIND_POSSIBLE, AA_CAPS);
5763 }
5764 }
5765
5766 /*** generic typefind for streams that have some data at a specific position***/
5767 typedef struct
5768 {
5769 const guint8 *data;
5770 guint size;
5771 guint probability;
5772 GstCaps *caps;
5773 }
5774 GstTypeFindData;
5775
5776 static void
start_with_type_find(GstTypeFind * tf,gpointer private)5777 start_with_type_find (GstTypeFind * tf, gpointer private)
5778 {
5779 GstTypeFindData *start_with = (GstTypeFindData *) private;
5780 const guint8 *data;
5781
5782 GST_LOG ("trying to find mime type %s with the first %u bytes of data",
5783 gst_structure_get_name (gst_caps_get_structure (start_with->caps, 0)),
5784 start_with->size);
5785 data = gst_type_find_peek (tf, 0, start_with->size);
5786 if (data && memcmp (data, start_with->data, start_with->size) == 0) {
5787 gst_type_find_suggest (tf, start_with->probability, start_with->caps);
5788 }
5789 }
5790
5791 static void
sw_data_destroy(GstTypeFindData * sw_data)5792 sw_data_destroy (GstTypeFindData * sw_data)
5793 {
5794 if (G_LIKELY (sw_data->caps != NULL))
5795 gst_caps_unref (sw_data->caps);
5796 g_slice_free (GstTypeFindData, sw_data);
5797 }
5798
5799 #define TYPE_FIND_REGISTER_START_WITH(plugin,name,rank,ext,_data,_size,_probability)\
5800 G_BEGIN_DECLS{ \
5801 GstTypeFindData *sw_data = g_slice_new (GstTypeFindData); \
5802 sw_data->data = (const guint8 *)_data; \
5803 sw_data->size = _size; \
5804 sw_data->probability = _probability; \
5805 sw_data->caps = gst_caps_new_empty_simple (name); \
5806 if (!gst_type_find_register (plugin, name, rank, start_with_type_find,\
5807 ext, sw_data->caps, sw_data, \
5808 (GDestroyNotify) (sw_data_destroy))) { \
5809 sw_data_destroy (sw_data); \
5810 } \
5811 }G_END_DECLS
5812
5813 /*** same for riff types ***/
5814
5815 static void
riff_type_find(GstTypeFind * tf,gpointer private)5816 riff_type_find (GstTypeFind * tf, gpointer private)
5817 {
5818 GstTypeFindData *riff_data = (GstTypeFindData *) private;
5819 const guint8 *data = gst_type_find_peek (tf, 0, 12);
5820
5821 if (data && (memcmp (data, "RIFF", 4) == 0 || memcmp (data, "AVF0", 4) == 0)) {
5822 data += 8;
5823 if (memcmp (data, riff_data->data, 4) == 0)
5824 gst_type_find_suggest (tf, riff_data->probability, riff_data->caps);
5825 }
5826 }
5827
5828 #define TYPE_FIND_REGISTER_RIFF(plugin,name,rank,ext,_data) \
5829 G_BEGIN_DECLS{ \
5830 GstTypeFindData *sw_data = g_slice_new (GstTypeFindData); \
5831 sw_data->data = (gpointer)_data; \
5832 sw_data->size = 4; \
5833 sw_data->probability = GST_TYPE_FIND_MAXIMUM; \
5834 sw_data->caps = gst_caps_new_empty_simple (name); \
5835 if (!gst_type_find_register (plugin, name, rank, riff_type_find, \
5836 ext, sw_data->caps, sw_data, \
5837 (GDestroyNotify) (sw_data_destroy))) { \
5838 sw_data_destroy (sw_data); \
5839 } \
5840 }G_END_DECLS
5841
5842
5843 /*** plugin initialization ***/
5844
5845 #define TYPE_FIND_REGISTER(plugin,name,rank,func,ext,caps,priv,notify) \
5846 G_BEGIN_DECLS{\
5847 if (!gst_type_find_register (plugin, name, rank, func, ext, caps, priv, notify))\
5848 return FALSE; \
5849 }G_END_DECLS
5850
5851
5852 static gboolean
plugin_init(GstPlugin * plugin)5853 plugin_init (GstPlugin * plugin)
5854 {
5855 /* can't initialize this via a struct as caps can't be statically initialized */
5856
5857 GST_DEBUG_CATEGORY_INIT (type_find_debug, "typefindfunctions",
5858 GST_DEBUG_FG_GREEN | GST_DEBUG_BG_RED, "generic type find functions");
5859
5860 /* note: asx/wax/wmx are XML files, asf doesn't handle them */
5861 /* must use strings, macros don't accept initializers */
5862 TYPE_FIND_REGISTER_START_WITH (plugin, "video/x-ms-asf", GST_RANK_SECONDARY,
5863 "asf,wm,wma,wmv",
5864 "\060\046\262\165\216\146\317\021\246\331\000\252\000\142\316\154", 16,
5865 GST_TYPE_FIND_MAXIMUM);
5866 TYPE_FIND_REGISTER (plugin, "audio/x-musepack", GST_RANK_PRIMARY,
5867 musepack_type_find, "mpc,mpp,mp+", MUSEPACK_CAPS, NULL, NULL);
5868 TYPE_FIND_REGISTER (plugin, "audio/x-au", GST_RANK_MARGINAL,
5869 au_type_find, "au,snd", AU_CAPS, NULL, NULL);
5870 TYPE_FIND_REGISTER_RIFF (plugin, "video/x-msvideo", GST_RANK_PRIMARY,
5871 "avi", "AVI ");
5872 TYPE_FIND_REGISTER_RIFF (plugin, "audio/qcelp", GST_RANK_PRIMARY,
5873 "qcp", "QLCM");
5874 TYPE_FIND_REGISTER_RIFF (plugin, "video/x-cdxa", GST_RANK_PRIMARY,
5875 "dat", "CDXA");
5876 TYPE_FIND_REGISTER_START_WITH (plugin, "video/x-vcd", GST_RANK_PRIMARY,
5877 "dat", "\000\377\377\377\377\377\377\377\377\377\377\000", 12,
5878 GST_TYPE_FIND_MAXIMUM);
5879 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-imelody", GST_RANK_PRIMARY,
5880 "imy,ime,imelody", "BEGIN:IMELODY", 13, GST_TYPE_FIND_MAXIMUM);
5881 TYPE_FIND_REGISTER_START_WITH (plugin, "application/x-scc", GST_RANK_PRIMARY,
5882 "scc", "Scenarist_SCC V1.0", 18, GST_TYPE_FIND_MAXIMUM);
5883 TYPE_FIND_REGISTER (plugin, "application/x-mcc", GST_RANK_PRIMARY,
5884 mcc_type_find, "mcc", MCC_CAPS, NULL, NULL);
5885
5886 #if 0
5887 TYPE_FIND_REGISTER_START_WITH (plugin, "video/x-smoke", GST_RANK_PRIMARY,
5888 NULL, "\x80smoke\x00\x01\x00", 6, GST_TYPE_FIND_MAXIMUM);
5889 #endif
5890 TYPE_FIND_REGISTER (plugin, "audio/midi", GST_RANK_PRIMARY, mid_type_find,
5891 "mid,midi", MID_CAPS, NULL, NULL);
5892 TYPE_FIND_REGISTER_RIFF (plugin, "audio/riff-midi", GST_RANK_PRIMARY,
5893 "mid,midi", "RMID");
5894 TYPE_FIND_REGISTER (plugin, "audio/mobile-xmf", GST_RANK_PRIMARY,
5895 mxmf_type_find, "mxmf", MXMF_CAPS, NULL, NULL);
5896 TYPE_FIND_REGISTER (plugin, "video/x-fli", GST_RANK_MARGINAL, flx_type_find,
5897 "flc,fli", FLX_CAPS, NULL, NULL);
5898 TYPE_FIND_REGISTER (plugin, "application/x-id3v2", GST_RANK_PRIMARY + 103,
5899 id3v2_type_find, "mp3,mp2,mp1,mpga,ogg,flac,tta", ID3_CAPS, NULL, NULL);
5900 TYPE_FIND_REGISTER (plugin, "application/x-id3v1", GST_RANK_PRIMARY + 101,
5901 id3v1_type_find, "mp3,mp2,mp1,mpga,ogg,flac,tta", ID3_CAPS, NULL, NULL);
5902 TYPE_FIND_REGISTER (plugin, "application/x-apetag", GST_RANK_PRIMARY + 102,
5903 apetag_type_find, "mp3,ape,mpc,wv", APETAG_CAPS, NULL, NULL);
5904 TYPE_FIND_REGISTER (plugin, "audio/x-ttafile", GST_RANK_PRIMARY,
5905 tta_type_find, "tta", TTA_CAPS, NULL, NULL);
5906 TYPE_FIND_REGISTER (plugin, "audio/x-mod", GST_RANK_SECONDARY, mod_type_find,
5907 "669,amf,ams,dbm,digi,dmf,dsm,gdm,far,imf,it,j2b,mdl,med,mod,mt2,mtm,"
5908 "okt,psm,ptm,sam,s3m,stm,stx,ult,umx,xm", MOD_CAPS, NULL, NULL);
5909 TYPE_FIND_REGISTER (plugin, "audio/mpeg", GST_RANK_PRIMARY, mp3_type_find,
5910 "mp3,mp2,mp1,mpga", MP3_CAPS, NULL, NULL);
5911 TYPE_FIND_REGISTER (plugin, "audio/x-ac3", GST_RANK_PRIMARY, ac3_type_find,
5912 "ac3,eac3", AC3_CAPS, NULL, NULL);
5913 TYPE_FIND_REGISTER (plugin, "audio/x-dts", GST_RANK_SECONDARY, dts_type_find,
5914 "dts", DTS_CAPS, NULL, NULL);
5915 TYPE_FIND_REGISTER (plugin, "audio/x-gsm", GST_RANK_PRIMARY, NULL, "gsm",
5916 GSM_CAPS, NULL, NULL);
5917 TYPE_FIND_REGISTER (plugin, "video/mpeg-sys", GST_RANK_PRIMARY,
5918 mpeg_sys_type_find, "mpe,mpeg,mpg", MPEG_SYS_CAPS, NULL, NULL);
5919 TYPE_FIND_REGISTER (plugin, "video/mpegts", GST_RANK_PRIMARY,
5920 mpeg_ts_type_find, "ts,mts", MPEGTS_CAPS, NULL, NULL);
5921 TYPE_FIND_REGISTER (plugin, "application/ogg", GST_RANK_PRIMARY,
5922 ogganx_type_find, "ogg,oga,ogv,ogm,ogx,spx,anx,axa,axv", OGG_CAPS,
5923 NULL, NULL);
5924 TYPE_FIND_REGISTER (plugin, "video/mpeg-elementary", GST_RANK_MARGINAL,
5925 mpeg_video_stream_type_find, "mpv,mpeg,mpg", MPEG_VIDEO_CAPS, NULL, NULL);
5926 TYPE_FIND_REGISTER (plugin, "video/mpeg4", GST_RANK_PRIMARY,
5927 mpeg4_video_type_find, "m4v", MPEG_VIDEO_CAPS, NULL, NULL);
5928 TYPE_FIND_REGISTER (plugin, "video/x-h263", GST_RANK_SECONDARY,
5929 h263_video_type_find, "h263,263", H263_VIDEO_CAPS, NULL, NULL);
5930 TYPE_FIND_REGISTER (plugin, "video/x-h264", GST_RANK_PRIMARY,
5931 h264_video_type_find, "h264,x264,264", H264_VIDEO_CAPS, NULL, NULL);
5932 TYPE_FIND_REGISTER (plugin, "video/x-h265", GST_RANK_PRIMARY,
5933 h265_video_type_find, "h265,x265,265", H265_VIDEO_CAPS, NULL, NULL);
5934 TYPE_FIND_REGISTER (plugin, "video/x-nuv", GST_RANK_SECONDARY, nuv_type_find,
5935 "nuv", NUV_CAPS, NULL, NULL);
5936
5937 /* ISO formats */
5938 TYPE_FIND_REGISTER (plugin, "audio/x-m4a", GST_RANK_PRIMARY, m4a_type_find,
5939 "m4a", M4A_CAPS, NULL, NULL);
5940 TYPE_FIND_REGISTER (plugin, "application/x-3gp", GST_RANK_PRIMARY,
5941 q3gp_type_find, "3gp", Q3GP_CAPS, NULL, NULL);
5942 TYPE_FIND_REGISTER (plugin, "video/quicktime", GST_RANK_PRIMARY,
5943 qt_type_find, "mov,mp4", QT_CAPS, NULL, NULL);
5944 TYPE_FIND_REGISTER (plugin, "image/x-quicktime", GST_RANK_SECONDARY,
5945 qtif_type_find, "qif,qtif,qti", QTIF_CAPS, NULL, NULL);
5946 TYPE_FIND_REGISTER (plugin, "image/jp2", GST_RANK_PRIMARY,
5947 jp2_type_find, "jp2", JP2_CAPS, NULL, NULL);
5948 TYPE_FIND_REGISTER (plugin, "image/x-jpc", GST_RANK_PRIMARY,
5949 jpc_type_find, "jpc,j2k", JPC_CAPS, NULL, NULL);
5950 TYPE_FIND_REGISTER (plugin, "video/mj2", GST_RANK_PRIMARY,
5951 jp2_type_find, "mj2", MJ2_CAPS, NULL, NULL);
5952
5953 TYPE_FIND_REGISTER (plugin, "text/html", GST_RANK_SECONDARY, html_type_find,
5954 "htm,html", HTML_CAPS, NULL, NULL);
5955 TYPE_FIND_REGISTER_START_WITH (plugin, "application/vnd.rn-realmedia",
5956 GST_RANK_SECONDARY, "ra,ram,rm,rmvb", ".RMF", 4, GST_TYPE_FIND_MAXIMUM);
5957 TYPE_FIND_REGISTER_START_WITH (plugin, "application/x-pn-realaudio",
5958 GST_RANK_SECONDARY, "ra,ram,rm,rmvb", ".ra\375", 4,
5959 GST_TYPE_FIND_MAXIMUM);
5960 TYPE_FIND_REGISTER (plugin, "application/x-shockwave-flash",
5961 GST_RANK_SECONDARY, swf_type_find, "swf,swfl", SWF_CAPS, NULL, NULL);
5962 TYPE_FIND_REGISTER (plugin, "application/xges",
5963 GST_RANK_PRIMARY, xges_type_find, "xges", XGES_CAPS, NULL, NULL);
5964 TYPE_FIND_REGISTER (plugin, "application/dash+xml",
5965 GST_RANK_PRIMARY, dash_mpd_type_find, "mpd,MPD", DASH_CAPS, NULL, NULL);
5966 TYPE_FIND_REGISTER (plugin, "application/vnd.ms-sstr+xml",
5967 GST_RANK_PRIMARY, mss_manifest_type_find, NULL, MSS_MANIFEST_CAPS, NULL,
5968 NULL);
5969 TYPE_FIND_REGISTER_START_WITH (plugin, "video/x-flv", GST_RANK_SECONDARY,
5970 "flv", "FLV", 3, GST_TYPE_FIND_MAXIMUM);
5971 TYPE_FIND_REGISTER (plugin, "text/plain", GST_RANK_MARGINAL, utf8_type_find,
5972 "txt", UTF8_CAPS, NULL, NULL);
5973 TYPE_FIND_REGISTER (plugin, "text/utf-16", GST_RANK_MARGINAL, utf16_type_find,
5974 "txt", UTF16_CAPS, NULL, NULL);
5975 TYPE_FIND_REGISTER (plugin, "text/utf-32", GST_RANK_MARGINAL, utf32_type_find,
5976 "txt", UTF32_CAPS, NULL, NULL);
5977 TYPE_FIND_REGISTER (plugin, "text/uri-list", GST_RANK_MARGINAL, uri_type_find,
5978 "ram", URI_CAPS, NULL, NULL);
5979 TYPE_FIND_REGISTER (plugin, "application/itc", GST_RANK_SECONDARY,
5980 itc_type_find, "itc", ITC_CAPS, NULL, NULL);
5981 TYPE_FIND_REGISTER (plugin, "application/x-hls", GST_RANK_MARGINAL,
5982 hls_type_find, "m3u8", HLS_CAPS, NULL, NULL);
5983 TYPE_FIND_REGISTER (plugin, "application/sdp", GST_RANK_SECONDARY,
5984 sdp_type_find, "sdp", SDP_CAPS, NULL, NULL);
5985 TYPE_FIND_REGISTER (plugin, "application/smil", GST_RANK_SECONDARY,
5986 smil_type_find, "smil", SMIL_CAPS, NULL, NULL);
5987 TYPE_FIND_REGISTER (plugin, "application/ttml+xml", GST_RANK_SECONDARY,
5988 ttml_xml_type_find, "ttml+xml", TTML_XML_CAPS, NULL, NULL);
5989 TYPE_FIND_REGISTER (plugin, "application/xml", GST_RANK_MARGINAL,
5990 xml_type_find, "xml", GENERIC_XML_CAPS, NULL, NULL);
5991 TYPE_FIND_REGISTER_RIFF (plugin, "audio/x-wav", GST_RANK_PRIMARY, "wav",
5992 "WAVE");
5993 TYPE_FIND_REGISTER (plugin, "audio/x-aiff", GST_RANK_SECONDARY,
5994 aiff_type_find, "aiff,aif,aifc", AIFF_CAPS, NULL, NULL);
5995 TYPE_FIND_REGISTER (plugin, "audio/x-svx", GST_RANK_SECONDARY, svx_type_find,
5996 "iff,svx", SVX_CAPS, NULL, NULL);
5997 TYPE_FIND_REGISTER (plugin, "audio/x-paris", GST_RANK_SECONDARY,
5998 paris_type_find, "paf", PARIS_CAPS, NULL, NULL);
5999 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-nist", GST_RANK_SECONDARY,
6000 "nist", "NIST", 4, GST_TYPE_FIND_MAXIMUM);
6001 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-voc", GST_RANK_SECONDARY,
6002 "voc", "Creative", 8, GST_TYPE_FIND_MAXIMUM);
6003 TYPE_FIND_REGISTER (plugin, "audio/x-sds", GST_RANK_SECONDARY, sds_type_find,
6004 "sds", SDS_CAPS, NULL, NULL);
6005 TYPE_FIND_REGISTER (plugin, "audio/x-ircam", GST_RANK_SECONDARY,
6006 ircam_type_find, "sf", IRCAM_CAPS, NULL, NULL);
6007 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-w64", GST_RANK_SECONDARY,
6008 "w64", "riff", 4, GST_TYPE_FIND_MAXIMUM);
6009 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-rf64", GST_RANK_PRIMARY,
6010 "rf64", "RF64", 4, GST_TYPE_FIND_MAXIMUM);
6011 TYPE_FIND_REGISTER (plugin, "audio/x-shorten", GST_RANK_SECONDARY,
6012 shn_type_find, "shn", SHN_CAPS, NULL, NULL);
6013 TYPE_FIND_REGISTER (plugin, "application/x-ape", GST_RANK_SECONDARY,
6014 ape_type_find, "ape", APE_CAPS, NULL, NULL);
6015 TYPE_FIND_REGISTER (plugin, "image/jpeg", GST_RANK_PRIMARY + 15,
6016 jpeg_type_find, "jpg,jpe,jpeg", JPEG_CAPS, NULL, NULL);
6017 TYPE_FIND_REGISTER_START_WITH (plugin, "image/gif", GST_RANK_PRIMARY, "gif",
6018 "GIF8", 4, GST_TYPE_FIND_MAXIMUM);
6019 TYPE_FIND_REGISTER_START_WITH (plugin, "image/png", GST_RANK_PRIMARY + 14,
6020 "png", "\211PNG\015\012\032\012", 8, GST_TYPE_FIND_MAXIMUM);
6021 TYPE_FIND_REGISTER (plugin, "image/bmp", GST_RANK_PRIMARY, bmp_type_find,
6022 "bmp", BMP_CAPS, NULL, NULL);
6023 TYPE_FIND_REGISTER (plugin, "image/tiff", GST_RANK_PRIMARY, tiff_type_find,
6024 "tif,tiff", TIFF_CAPS, NULL, NULL);
6025 TYPE_FIND_REGISTER_RIFF (plugin, "image/webp", GST_RANK_PRIMARY,
6026 "webp", "WEBP");
6027 TYPE_FIND_REGISTER (plugin, "image/x-exr", GST_RANK_PRIMARY, exr_type_find,
6028 "exr", EXR_CAPS, NULL, NULL);
6029 TYPE_FIND_REGISTER (plugin, "image/x-portable-pixmap", GST_RANK_SECONDARY,
6030 pnm_type_find, "pnm,ppm,pgm,pbm", PNM_CAPS, NULL, NULL);
6031 TYPE_FIND_REGISTER (plugin, "video/x-matroska", GST_RANK_PRIMARY,
6032 matroska_type_find, "mkv,mka,mk3d,webm", MATROSKA_CAPS, NULL, NULL);
6033 TYPE_FIND_REGISTER (plugin, "application/mxf", GST_RANK_PRIMARY,
6034 mxf_type_find, "mxf", MXF_CAPS, NULL, NULL);
6035 TYPE_FIND_REGISTER_START_WITH (plugin, "video/x-mve", GST_RANK_SECONDARY,
6036 "mve", "Interplay MVE File\032\000\032\000\000\001\063\021", 26,
6037 GST_TYPE_FIND_MAXIMUM);
6038 TYPE_FIND_REGISTER (plugin, "video/x-dv", GST_RANK_SECONDARY, dv_type_find,
6039 "dv,dif", DV_CAPS, NULL, NULL);
6040 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-amr-nb-sh", GST_RANK_PRIMARY,
6041 "amr", "#!AMR", 5, GST_TYPE_FIND_LIKELY);
6042 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-amr-wb-sh", GST_RANK_PRIMARY,
6043 "amr", "#!AMR-WB", 7, GST_TYPE_FIND_MAXIMUM);
6044 TYPE_FIND_REGISTER (plugin, "audio/iLBC-sh", GST_RANK_PRIMARY, ilbc_type_find,
6045 "ilbc", ILBC_CAPS, NULL, NULL);
6046 TYPE_FIND_REGISTER (plugin, "audio/x-sbc", GST_RANK_MARGINAL, sbc_type_find,
6047 "sbc", SBC_CAPS, NULL, NULL);
6048 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-sid", GST_RANK_MARGINAL,
6049 "sid", "PSID", 4, GST_TYPE_FIND_MAXIMUM);
6050 TYPE_FIND_REGISTER_START_WITH (plugin, "image/x-xcf", GST_RANK_SECONDARY,
6051 "xcf", "gimp xcf", 8, GST_TYPE_FIND_MAXIMUM);
6052 TYPE_FIND_REGISTER_START_WITH (plugin, "video/x-mng", GST_RANK_SECONDARY,
6053 "mng", "\212MNG\015\012\032\012", 8, GST_TYPE_FIND_MAXIMUM);
6054 TYPE_FIND_REGISTER_START_WITH (plugin, "image/x-jng", GST_RANK_SECONDARY,
6055 "jng", "\213JNG\015\012\032\012", 8, GST_TYPE_FIND_MAXIMUM);
6056 TYPE_FIND_REGISTER_START_WITH (plugin, "image/x-xpixmap", GST_RANK_SECONDARY,
6057 "xpm", "/* XPM */", 9, GST_TYPE_FIND_MAXIMUM);
6058 TYPE_FIND_REGISTER_START_WITH (plugin, "image/x-sun-raster",
6059 GST_RANK_SECONDARY, "ras", "\131\246\152\225", 4, GST_TYPE_FIND_MAXIMUM);
6060 TYPE_FIND_REGISTER_START_WITH (plugin, "application/x-bzip",
6061 GST_RANK_SECONDARY, "bz2", "BZh", 3, GST_TYPE_FIND_LIKELY);
6062 TYPE_FIND_REGISTER_START_WITH (plugin, "application/x-gzip",
6063 GST_RANK_SECONDARY, "gz", "\037\213", 2, GST_TYPE_FIND_LIKELY);
6064 TYPE_FIND_REGISTER_START_WITH (plugin, "application/zip", GST_RANK_SECONDARY,
6065 "zip", "PK\003\004", 4, GST_TYPE_FIND_LIKELY);
6066 TYPE_FIND_REGISTER_START_WITH (plugin, "application/x-compress",
6067 GST_RANK_SECONDARY, "Z", "\037\235", 2, GST_TYPE_FIND_LIKELY);
6068 TYPE_FIND_REGISTER (plugin, "subtitle/x-kate", GST_RANK_MARGINAL,
6069 kate_type_find, NULL, NULL, NULL, NULL);
6070 TYPE_FIND_REGISTER (plugin, "application/x-subtitle-vtt", GST_RANK_MARGINAL,
6071 webvtt_type_find, "vtt", WEBVTT_CAPS, NULL, NULL);
6072 TYPE_FIND_REGISTER (plugin, "audio/x-flac", GST_RANK_PRIMARY, flac_type_find,
6073 "flac", FLAC_CAPS, NULL, NULL);
6074 TYPE_FIND_REGISTER (plugin, "audio/x-vorbis", GST_RANK_PRIMARY,
6075 vorbis_type_find, NULL, VORBIS_CAPS, NULL, NULL);
6076 TYPE_FIND_REGISTER (plugin, "video/x-theora", GST_RANK_PRIMARY,
6077 theora_type_find, NULL, THEORA_CAPS, NULL, NULL);
6078 TYPE_FIND_REGISTER (plugin, "application/x-ogm-video", GST_RANK_PRIMARY,
6079 ogmvideo_type_find, NULL, OGMVIDEO_CAPS, NULL, NULL);
6080 TYPE_FIND_REGISTER (plugin, "application/x-ogm-audio", GST_RANK_PRIMARY,
6081 ogmaudio_type_find, NULL, OGMAUDIO_CAPS, NULL, NULL);
6082 TYPE_FIND_REGISTER (plugin, "application/x-ogm-text", GST_RANK_PRIMARY,
6083 ogmtext_type_find, NULL, OGMTEXT_CAPS, NULL, NULL);
6084 TYPE_FIND_REGISTER (plugin, "audio/x-speex", GST_RANK_PRIMARY,
6085 speex_type_find, NULL, SPEEX_CAPS, NULL, NULL);
6086 TYPE_FIND_REGISTER (plugin, "audio/x-celt", GST_RANK_PRIMARY, celt_type_find,
6087 NULL, CELT_CAPS, NULL, NULL);
6088 TYPE_FIND_REGISTER (plugin, "application/x-ogg-skeleton", GST_RANK_PRIMARY,
6089 oggskel_type_find, NULL, OGG_SKELETON_CAPS, NULL, NULL);
6090 TYPE_FIND_REGISTER (plugin, "text/x-cmml", GST_RANK_PRIMARY, cmml_type_find,
6091 NULL, CMML_CAPS, NULL, NULL);
6092 TYPE_FIND_REGISTER_START_WITH (plugin, "application/x-executable",
6093 GST_RANK_MARGINAL, NULL, "\177ELF", 4, GST_TYPE_FIND_MAXIMUM);
6094 TYPE_FIND_REGISTER (plugin, "audio/aac", GST_RANK_SECONDARY, aac_type_find,
6095 "aac,adts,adif,loas", AAC_CAPS, NULL, NULL);
6096 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-spc", GST_RANK_SECONDARY,
6097 "spc", "SNES-SPC700 Sound File Data", 27, GST_TYPE_FIND_MAXIMUM);
6098 TYPE_FIND_REGISTER (plugin, "audio/x-wavpack", GST_RANK_SECONDARY,
6099 wavpack_type_find, "wv,wvp", WAVPACK_CAPS, NULL, NULL);
6100 TYPE_FIND_REGISTER (plugin, "audio/x-wavpack-correction", GST_RANK_SECONDARY,
6101 wavpack_type_find, "wvc", WAVPACK_CORRECTION_CAPS, NULL, NULL);
6102 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-caf", GST_RANK_SECONDARY,
6103 "caf", "caff\000\001", 6, GST_TYPE_FIND_MAXIMUM);
6104 TYPE_FIND_REGISTER (plugin, "application/postscript", GST_RANK_SECONDARY,
6105 postscript_type_find, "ps", POSTSCRIPT_CAPS, NULL, NULL);
6106 TYPE_FIND_REGISTER (plugin, "image/svg+xml", GST_RANK_SECONDARY,
6107 svg_type_find, "svg", SVG_CAPS, NULL, NULL);
6108 TYPE_FIND_REGISTER_START_WITH (plugin, "application/x-rar",
6109 GST_RANK_SECONDARY, "rar", "Rar!", 4, GST_TYPE_FIND_LIKELY);
6110 TYPE_FIND_REGISTER (plugin, "application/x-tar", GST_RANK_SECONDARY,
6111 tar_type_find, "tar", TAR_CAPS, NULL, NULL);
6112 TYPE_FIND_REGISTER (plugin, "application/x-ar", GST_RANK_SECONDARY,
6113 ar_type_find, "a", AR_CAPS, NULL, NULL);
6114 TYPE_FIND_REGISTER (plugin, "application/x-ms-dos-executable",
6115 GST_RANK_SECONDARY, msdos_type_find, "dll,exe,ocx,sys,scr,msstyles,cpl",
6116 MSDOS_CAPS, NULL, NULL);
6117 TYPE_FIND_REGISTER (plugin, "video/x-dirac", GST_RANK_PRIMARY,
6118 dirac_type_find, NULL, DIRAC_CAPS, NULL, NULL);
6119 TYPE_FIND_REGISTER (plugin, "multipart/x-mixed-replace", GST_RANK_SECONDARY,
6120 multipart_type_find, NULL, MULTIPART_CAPS, NULL, NULL);
6121 TYPE_FIND_REGISTER (plugin, "application/x-mmsh", GST_RANK_SECONDARY,
6122 mmsh_type_find, NULL, MMSH_CAPS, NULL, NULL);
6123 TYPE_FIND_REGISTER (plugin, "video/vivo", GST_RANK_SECONDARY, vivo_type_find,
6124 "viv", VIVO_CAPS, NULL, NULL);
6125 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-nsf", GST_RANK_SECONDARY,
6126 "nsf", "NESM\x1a", 5, GST_TYPE_FIND_MAXIMUM);
6127 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-gym", GST_RANK_SECONDARY,
6128 "gym", "GYMX", 4, GST_TYPE_FIND_MAXIMUM);
6129 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-ay", GST_RANK_SECONDARY, "ay",
6130 "ZXAYEMUL", 8, GST_TYPE_FIND_MAXIMUM);
6131 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-gbs", GST_RANK_SECONDARY,
6132 "gbs", "GBS\x01", 4, GST_TYPE_FIND_MAXIMUM);
6133 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-vgm", GST_RANK_SECONDARY,
6134 "vgm", "Vgm\x20", 4, GST_TYPE_FIND_MAXIMUM);
6135 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-sap", GST_RANK_SECONDARY,
6136 "sap", "SAP\x0d\x0a" "AUTHOR\x20", 12, GST_TYPE_FIND_MAXIMUM);
6137 TYPE_FIND_REGISTER_START_WITH (plugin, "video/x-ivf", GST_RANK_SECONDARY,
6138 "ivf", "DKIF", 4, GST_TYPE_FIND_NEARLY_CERTAIN);
6139 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-kss", GST_RANK_SECONDARY,
6140 "kss", "KSSX\0", 5, GST_TYPE_FIND_MAXIMUM);
6141 TYPE_FIND_REGISTER_START_WITH (plugin, "application/pdf", GST_RANK_SECONDARY,
6142 "pdf", "%PDF-", 5, GST_TYPE_FIND_LIKELY);
6143 TYPE_FIND_REGISTER_START_WITH (plugin, "application/msword",
6144 GST_RANK_SECONDARY, "doc", "\320\317\021\340\241\261\032\341", 8,
6145 GST_TYPE_FIND_LIKELY);
6146 /* Mac OS X .DS_Store files tend to be taken for video/mpeg */
6147 TYPE_FIND_REGISTER_START_WITH (plugin, "application/octet-stream",
6148 GST_RANK_SECONDARY, "DS_Store", "\000\000\000\001Bud1", 8,
6149 GST_TYPE_FIND_LIKELY);
6150 TYPE_FIND_REGISTER_START_WITH (plugin, "image/vnd.adobe.photoshop",
6151 GST_RANK_SECONDARY, "psd", "8BPS\000\001\000\000\000\000", 10,
6152 GST_TYPE_FIND_LIKELY);
6153 TYPE_FIND_REGISTER (plugin, "image/vnd.wap.wbmp", GST_RANK_MARGINAL,
6154 wbmp_typefind, NULL, NULL, NULL, NULL);
6155 TYPE_FIND_REGISTER (plugin, "application/x-yuv4mpeg", GST_RANK_SECONDARY,
6156 y4m_typefind, NULL, NULL, NULL, NULL);
6157 TYPE_FIND_REGISTER (plugin, "image/x-icon", GST_RANK_MARGINAL,
6158 windows_icon_typefind, NULL, NULL, NULL, NULL);
6159
6160 #ifdef USE_GIO
6161 TYPE_FIND_REGISTER (plugin, "xdgmime-base", GST_RANK_MARGINAL,
6162 xdgmime_typefind, NULL, NULL, NULL, NULL);
6163 #endif
6164
6165 TYPE_FIND_REGISTER (plugin, "image/x-degas", GST_RANK_MARGINAL,
6166 degas_type_find, NULL, NULL, NULL, NULL);
6167 TYPE_FIND_REGISTER (plugin, "application/octet-stream", GST_RANK_MARGINAL,
6168 dvdiso_type_find, NULL, NULL, NULL, NULL);
6169
6170 TYPE_FIND_REGISTER (plugin, "application/x-ssa", GST_RANK_SECONDARY,
6171 ssa_type_find, "ssa,ass", NULL, NULL, NULL);
6172
6173 TYPE_FIND_REGISTER (plugin, "video/x-pva", GST_RANK_SECONDARY,
6174 pva_type_find, "pva", PVA_CAPS, NULL, NULL);
6175
6176 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-xi", GST_RANK_SECONDARY,
6177 "xi", "Extended Instrument: ", 21, GST_TYPE_FIND_MAXIMUM);
6178
6179 TYPE_FIND_REGISTER (plugin, "audio/audible", GST_RANK_MARGINAL,
6180 aa_type_find, "aa,aax", AA_CAPS, NULL, NULL);
6181
6182 TYPE_FIND_REGISTER (plugin, "audio/x-tap-tap", GST_RANK_PRIMARY,
6183 tap_type_find, "tap", TAP_CAPS, NULL, NULL);
6184 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-tap-dmp",
6185 GST_RANK_SECONDARY, "dmp", "DC2N-TAP-RAW", 12, GST_TYPE_FIND_LIKELY);
6186
6187 return TRUE;
6188 }
6189
6190 GST_PLUGIN_DEFINE (GST_VERSION_MAJOR,
6191 GST_VERSION_MINOR,
6192 typefindfunctions,
6193 "default typefind functions",
6194 plugin_init, VERSION, GST_LICENSE, GST_PACKAGE_NAME, GST_PACKAGE_ORIGIN)
6195