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