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>
6 * gsttypefindfunctions.c: collection of various typefind functions
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.
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.
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.
29 #include <glib/gprintf.h>
31 /* don't want to add gio xdgmime typefinder if gio was disabled via configure */
43 #include <gst/pbutils/pbutils.h>
44 #include <gst/base/gstbytereader.h>
46 GST_DEBUG_CATEGORY_STATIC (type_find_debug);
47 #define GST_CAT_DEFAULT type_find_debug
49 /* so our code stays ready for 0.11 */
50 #define gst_type_find_peek(tf,off,len) \
51 ((const guint8 *)gst_type_find_peek((tf),(off),(len)))
53 /* DataScanCtx: helper for typefind functions that scan through data
54 * step-by-step, to avoid doing a peek at each and every offset */
56 #define DATA_SCAN_CTX_CHUNK_SIZE 4096
66 data_scan_ctx_advance (GstTypeFind * tf, DataScanCtx * c, guint bytes_to_skip)
68 c->offset += bytes_to_skip;
69 if (G_LIKELY (c->size > bytes_to_skip)) {
70 c->size -= bytes_to_skip;
71 c->data += bytes_to_skip;
78 static inline gboolean
79 data_scan_ctx_ensure_data (GstTypeFind * tf, DataScanCtx * c, gint min_len)
83 guint chunk_len = MAX (DATA_SCAN_CTX_CHUNK_SIZE, min_len);
85 if (G_LIKELY (c->size >= min_len))
88 data = gst_type_find_peek (tf, c->offset, chunk_len);
89 if (G_LIKELY (data != NULL)) {
95 /* if there's less than our chunk size, try to get as much as we can, but
96 * always at least min_len bytes (we might be typefinding the first buffer
97 * of the stream and not have as much data available as we'd like) */
98 len = gst_type_find_get_length (tf);
100 len = CLAMP (len - c->offset, min_len, chunk_len);
105 data = gst_type_find_peek (tf, c->offset, len);
115 static inline gboolean
116 data_scan_ctx_memcmp (GstTypeFind * tf, DataScanCtx * c, guint offset,
117 const gchar * data, guint len)
119 if (!data_scan_ctx_ensure_data (tf, c, offset + len))
122 return (memcmp (c->data + offset, data, len) == 0);
126 static gboolean xml_check_first_element (GstTypeFind * tf,
127 const gchar * element, guint elen, gboolean strict);
128 static gboolean sdp_check_header (GstTypeFind * tf);
130 static GstStaticCaps utf8_caps = GST_STATIC_CAPS ("text/plain");
132 #define UTF8_CAPS gst_static_caps_get(&utf8_caps)
135 utf8_type_find_have_valid_utf8_at_offset (GstTypeFind * tf, guint64 offset,
136 GstTypeFindProbability * prob)
140 /* randomly decided values */
141 guint min_size = 16; /* minimum size */
142 guint size = 32 * 1024; /* starting size */
143 guint probability = 95; /* starting probability */
144 guint step = 10; /* how much we reduce probability in each
147 while (probability > step && size > min_size) {
148 data = gst_type_find_peek (tf, offset, size);
151 gchar *start = (gchar *) data;
153 if (g_utf8_validate (start, size, (const gchar **) &end) || (end - start + 4 > size)) { /* allow last char to be cut off */
168 utf8_type_find (GstTypeFind * tf, gpointer unused)
170 GstTypeFindProbability start_prob, mid_prob;
173 /* leave xml to the xml typefinders */
174 if (xml_check_first_element (tf, "", 0, TRUE))
177 /* leave sdp to the sdp typefinders */
178 if (sdp_check_header (tf))
181 /* check beginning of stream */
182 if (!utf8_type_find_have_valid_utf8_at_offset (tf, 0, &start_prob))
185 GST_LOG ("start is plain text with probability of %u", start_prob);
187 /* POSSIBLE is the highest probability we ever return if we can't
188 * probe into the middle of the file and don't know its length */
190 length = gst_type_find_get_length (tf);
191 if (length == 0 || length == (guint64) - 1) {
192 gst_type_find_suggest (tf, MIN (start_prob, GST_TYPE_FIND_POSSIBLE),
197 if (length < 64 * 1024) {
198 gst_type_find_suggest (tf, start_prob, UTF8_CAPS);
202 /* check middle of stream */
203 if (!utf8_type_find_have_valid_utf8_at_offset (tf, length / 2, &mid_prob))
206 GST_LOG ("middle is plain text with probability of %u", mid_prob);
207 gst_type_find_suggest (tf, (start_prob + mid_prob) / 2, UTF8_CAPS);
210 /*** text/utf-16 and text/utf-32} ***/
211 /* While UTF-8 is unicode too, using text/plain for UTF-16 and UTF-32
212 is going to break stuff. */
217 const char *const bom;
218 gboolean (*checker) (const guint8 *, gint, gint);
224 check_utf16 (const guint8 * data, gint len, gint endianness)
234 gst_byte_reader_init (&br, data, len);
236 /* test first for a single 16 bit value in the BMP */
237 if (endianness == G_BIG_ENDIAN)
238 high = gst_byte_reader_get_uint16_be_unchecked (&br);
240 high = gst_byte_reader_get_uint16_le_unchecked (&br);
241 if (high >= 0xD800 && high <= 0xDBFF) {
242 /* start of a surrogate pair */
246 if (endianness == G_BIG_ENDIAN)
247 low = gst_byte_reader_get_uint16_be_unchecked (&br);
249 low = gst_byte_reader_get_uint16_le_unchecked (&br);
250 if (low >= 0xDC00 && low <= 0xDFFF) {
251 /* second half of the surrogate pair */
255 if (high >= 0xDC00 && high <= 0xDFFF)
264 check_utf32 (const guint8 * data, gint len, gint endianness)
270 if (endianness == G_BIG_ENDIAN)
271 v = (data[0] << 24) | (data[1] << 16) | (data[2] << 8) | data[3];
273 v = (data[3] << 24) | (data[2] << 16) | (data[1] << 8) | data[0];
283 unicode_type_find (GstTypeFind * tf, const GstUnicodeTester * tester,
284 guint n_tester, const char *media_type, gboolean require_bom)
288 const guint8 *data = gst_type_find_peek (tf, 0, len);
290 const gint max_scan_size = 256 * 1024;
295 data = gst_type_find_peek (tf, 0, len);
300 /* find a large enough size that works */
301 while (len < max_scan_size) {
302 size_t newlen = len << 1;
303 const guint8 *newdata = gst_type_find_peek (tf, 0, newlen);
310 for (n = 0; n < n_tester; ++n) {
311 int bom_boost = 0, tmpprob;
312 if (len >= tester[n].bomlen) {
313 if (!memcmp (data, tester[n].bom, tester[n].bomlen))
314 bom_boost = tester[n].boost;
316 if (require_bom && bom_boost == 0)
318 if (!(*tester[n].checker) (data, len, tester[n].endianness))
320 tmpprob = GST_TYPE_FIND_POSSIBLE - 20 + bom_boost;
321 if (tmpprob > prob) {
323 endianness = tester[n].endianness;
328 GST_DEBUG ("This is valid %s %s", media_type,
329 endianness == G_BIG_ENDIAN ? "be" : "le");
330 gst_type_find_suggest_simple (tf, prob, media_type,
331 "endianness", G_TYPE_INT, endianness, NULL);
335 static GstStaticCaps utf16_caps = GST_STATIC_CAPS ("text/utf-16");
337 #define UTF16_CAPS gst_static_caps_get(&utf16_caps)
340 utf16_type_find (GstTypeFind * tf, gpointer unused)
342 static const GstUnicodeTester utf16tester[2] = {
343 {2, "\xff\xfe", check_utf16, 10, G_LITTLE_ENDIAN},
344 {2, "\xfe\xff", check_utf16, 20, G_BIG_ENDIAN},
346 unicode_type_find (tf, utf16tester, G_N_ELEMENTS (utf16tester),
347 "text/utf-16", TRUE);
350 static GstStaticCaps utf32_caps = GST_STATIC_CAPS ("text/utf-32");
352 #define UTF32_CAPS gst_static_caps_get(&utf32_caps)
355 utf32_type_find (GstTypeFind * tf, gpointer unused)
357 static const GstUnicodeTester utf32tester[2] = {
358 {4, "\xff\xfe\x00\x00", check_utf32, 10, G_LITTLE_ENDIAN},
359 {4, "\x00\x00\xfe\xff", check_utf32, 20, G_BIG_ENDIAN}
361 unicode_type_find (tf, utf32tester, G_N_ELEMENTS (utf32tester),
362 "text/utf-32", TRUE);
365 /*** text/uri-list ***/
367 static GstStaticCaps uri_caps = GST_STATIC_CAPS ("text/uri-list");
369 #define URI_CAPS (gst_static_caps_get(&uri_caps))
370 #define BUFFER_SIZE 16 /* If the string is < 16 bytes we're screwed */
371 #define INC_BUFFER { \
373 if (pos == BUFFER_SIZE) { \
375 offset += BUFFER_SIZE; \
376 data = gst_type_find_peek (tf, offset, BUFFER_SIZE); \
377 if (data == NULL) return; \
383 uri_type_find (GstTypeFind * tf, gpointer unused)
385 const guint8 *data = gst_type_find_peek (tf, 0, BUFFER_SIZE);
390 /* Search for # comment lines */
391 while (*data == '#') {
392 /* Goto end of line */
393 while (*data != '\n') {
400 if (!g_ascii_isalpha (*data)) {
401 /* Had a non alpha char - can't be uri-list */
407 while (g_ascii_isalnum (*data)) {
412 /* First non alpha char is not a : */
416 /* Get the next 2 bytes as well */
417 data = gst_type_find_peek (tf, offset + pos, 3);
421 if (data[1] != '/' && data[2] != '/') {
425 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, URI_CAPS);
429 /*** application/x-hls ***/
431 static GstStaticCaps hls_caps = GST_STATIC_CAPS ("application/x-hls");
432 #define HLS_CAPS (gst_static_caps_get(&hls_caps))
434 /* See http://tools.ietf.org/html/draft-pantos-http-live-streaming-05 */
436 hls_type_find (GstTypeFind * tf, gpointer unused)
438 DataScanCtx c = { 0, NULL, 0 };
440 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 7)))
443 if (memcmp (c.data, "#EXTM3U", 7))
446 data_scan_ctx_advance (tf, &c, 7);
448 /* Check only the first 256 bytes */
449 while (c.offset < 256) {
450 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 21)))
453 /* Search for # comment lines */
454 if (c.data[0] == '#' && (memcmp (c.data, "#EXT-X-TARGETDURATION", 21) == 0
455 || memcmp (c.data, "#EXT-X-STREAM-INF", 17) == 0)) {
456 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, HLS_CAPS);
460 data_scan_ctx_advance (tf, &c, 1);
465 /*** application/xml **********************************************************/
467 #define XML_BUFFER_SIZE 16
468 #define XML_INC_BUFFER { \
470 if (pos == XML_BUFFER_SIZE) { \
472 offset += XML_BUFFER_SIZE; \
473 data = gst_type_find_peek (tf, offset, XML_BUFFER_SIZE); \
474 if (data == NULL) return FALSE; \
481 xml_check_first_element (GstTypeFind * tf, const gchar * element, guint elen,
489 data = gst_type_find_peek (tf, 0, XML_BUFFER_SIZE);
493 /* look for the XMLDec
494 * see XML spec 2.8, Prolog and Document Type Declaration
495 * http://www.w3.org/TR/2004/REC-xml-20040204/#sec-prolog-dtd */
496 got_xmldec = (memcmp (data, "<?xml", 5) == 0);
498 if (strict && !got_xmldec)
501 /* skip XMLDec in any case if we've got one */
507 /* look for the first element, it has to be the requested element. Bail
508 * out if it is not within the first 4kB. */
509 while (data && (offset + pos) < 4096) {
510 while (*data != '<' && (offset + pos) < 4096) {
515 if (!g_ascii_isalpha (*data)) {
516 /* if not alphabetic, it's a PI or an element / attribute declaration
517 * like <?xxx or <!xxx */
522 /* the first normal element, check if it's the one asked for */
523 data = gst_type_find_peek (tf, offset + pos, elen + 1);
524 return (data && element && strncmp ((char *) data, element, elen) == 0);
530 static GstStaticCaps generic_xml_caps = GST_STATIC_CAPS ("application/xml");
532 #define GENERIC_XML_CAPS (gst_static_caps_get(&generic_xml_caps))
534 xml_type_find (GstTypeFind * tf, gpointer unused)
536 if (xml_check_first_element (tf, "", 0, TRUE)) {
537 gst_type_find_suggest (tf, GST_TYPE_FIND_MINIMUM, GENERIC_XML_CAPS);
541 /*** application/sdp *********************************************************/
543 static GstStaticCaps sdp_caps = GST_STATIC_CAPS ("application/sdp");
545 #define SDP_CAPS (gst_static_caps_get(&sdp_caps))
547 sdp_check_header (GstTypeFind * tf)
551 data = gst_type_find_peek (tf, 0, 5);
555 /* sdp must start with v=0[\r]\n */
556 if (memcmp (data, "v=0", 3))
559 if (data[3] == '\r' && data[4] == '\n')
568 sdp_type_find (GstTypeFind * tf, gpointer unused)
570 if (sdp_check_header (tf))
571 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, SDP_CAPS);
574 /*** application/smil *********************************************************/
576 static GstStaticCaps smil_caps = GST_STATIC_CAPS ("application/smil");
578 #define SMIL_CAPS (gst_static_caps_get(&smil_caps))
580 smil_type_find (GstTypeFind * tf, gpointer unused)
582 if (xml_check_first_element (tf, "smil", 4, FALSE)) {
583 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, SMIL_CAPS);
589 static GstStaticCaps html_caps = GST_STATIC_CAPS ("text/html");
591 #define HTML_CAPS gst_static_caps_get (&html_caps)
594 html_type_find (GstTypeFind * tf, gpointer unused)
596 const gchar *d, *data;
598 data = (const gchar *) gst_type_find_peek (tf, 0, 16);
602 if (!g_ascii_strncasecmp (data, "<!DOCTYPE HTML", 14)) {
603 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, HTML_CAPS);
604 } else if (xml_check_first_element (tf, "html", 4, FALSE)) {
605 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, HTML_CAPS);
606 } else if ((d = memchr (data, '<', 16))) {
607 data = (const gchar *) gst_type_find_peek (tf, d - data, 6);
608 if (data && g_ascii_strncasecmp (data, "<html>", 6) == 0) {
609 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, HTML_CAPS);
616 static GstStaticCaps mid_caps = GST_STATIC_CAPS ("audio/midi");
618 #define MID_CAPS gst_static_caps_get(&mid_caps)
620 mid_type_find (GstTypeFind * tf, gpointer unused)
622 const guint8 *data = gst_type_find_peek (tf, 0, 4);
624 /* http://jedi.ks.uiuc.edu/~johns/links/music/midifile.html */
625 if (data && data[0] == 'M' && data[1] == 'T' && data[2] == 'h'
627 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MID_CAPS);
630 /*** audio/mobile-xmf ***/
632 static GstStaticCaps mxmf_caps = GST_STATIC_CAPS ("audio/mobile-xmf");
634 #define MXMF_CAPS gst_static_caps_get(&mxmf_caps)
636 mxmf_type_find (GstTypeFind * tf, gpointer unused)
638 const guint8 *data = NULL;
640 /* Search FileId "XMF_" 4 bytes */
641 data = gst_type_find_peek (tf, 0, 4);
642 if (data && data[0] == 'X' && data[1] == 'M' && data[2] == 'F'
644 /* Search Format version "2.00" 4 bytes */
645 data = gst_type_find_peek (tf, 4, 4);
646 if (data && data[0] == '2' && data[1] == '.' && data[2] == '0'
648 /* Search TypeId 2 1 byte */
649 data = gst_type_find_peek (tf, 11, 1);
650 if (data && data[0] == 2) {
651 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MXMF_CAPS);
658 /*** video/x-fli ***/
660 static GstStaticCaps flx_caps = GST_STATIC_CAPS ("video/x-fli");
662 #define FLX_CAPS gst_static_caps_get(&flx_caps)
664 flx_type_find (GstTypeFind * tf, gpointer unused)
666 const guint8 *data = gst_type_find_peek (tf, 0, 134);
669 /* check magic and the frame type of the first frame */
670 if ((data[4] == 0x11 || data[4] == 0x12 ||
671 data[4] == 0x30 || data[4] == 0x44) &&
673 ((data[132] == 0x00 || data[132] == 0xfa) && data[133] == 0xf1)) {
674 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, FLX_CAPS);
678 data = gst_type_find_peek (tf, 0, 6);
680 /* check magic only */
681 if ((data[4] == 0x11 || data[4] == 0x12 ||
682 data[4] == 0x30 || data[4] == 0x44) && data[5] == 0xaf) {
683 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY, FLX_CAPS);
689 /*** application/x-id3 ***/
691 static GstStaticCaps id3_caps = GST_STATIC_CAPS ("application/x-id3");
693 #define ID3_CAPS gst_static_caps_get(&id3_caps)
695 id3v2_type_find (GstTypeFind * tf, gpointer unused)
697 const guint8 *data = gst_type_find_peek (tf, 0, 10);
699 if (data && memcmp (data, "ID3", 3) == 0 &&
700 data[3] != 0xFF && data[4] != 0xFF &&
701 (data[6] & 0x80) == 0 && (data[7] & 0x80) == 0 &&
702 (data[8] & 0x80) == 0 && (data[9] & 0x80) == 0) {
703 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, ID3_CAPS);
708 id3v1_type_find (GstTypeFind * tf, gpointer unused)
710 const guint8 *data = gst_type_find_peek (tf, -128, 3);
712 if (data && memcmp (data, "TAG", 3) == 0) {
713 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, ID3_CAPS);
717 /*** application/x-ape ***/
719 static GstStaticCaps apetag_caps = GST_STATIC_CAPS ("application/x-apetag");
721 #define APETAG_CAPS gst_static_caps_get(&apetag_caps)
723 apetag_type_find (GstTypeFind * tf, gpointer unused)
727 /* APEv1/2 at start of file */
728 data = gst_type_find_peek (tf, 0, 8);
729 if (data && !memcmp (data, "APETAGEX", 8)) {
730 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, APETAG_CAPS);
734 /* APEv1/2 at end of file */
735 data = gst_type_find_peek (tf, -32, 8);
736 if (data && !memcmp (data, "APETAGEX", 8)) {
737 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, APETAG_CAPS);
742 /*** audio/x-ttafile ***/
744 static GstStaticCaps tta_caps = GST_STATIC_CAPS ("audio/x-ttafile");
746 #define TTA_CAPS gst_static_caps_get(&tta_caps)
748 tta_type_find (GstTypeFind * tf, gpointer unused)
750 const guint8 *data = gst_type_find_peek (tf, 0, 3);
753 if (memcmp (data, "TTA", 3) == 0) {
754 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, TTA_CAPS);
760 /*** audio/x-flac ***/
761 static GstStaticCaps flac_caps = GST_STATIC_CAPS ("audio/x-flac");
763 #define FLAC_CAPS (gst_static_caps_get(&flac_caps))
766 flac_type_find (GstTypeFind * tf, gpointer unused)
768 DataScanCtx c = { 0, NULL, 0 };
770 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 4)))
773 /* standard flac (also old/broken flac-in-ogg with an initial 4-byte marker
774 * packet and without the usual packet framing) */
775 if (memcmp (c.data, "fLaC", 4) == 0) {
776 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, FLAC_CAPS);
780 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 6)))
783 /* flac-in-ogg, see http://flac.sourceforge.net/ogg_mapping.html */
784 if (memcmp (c.data, "\177FLAC\001", 6) == 0) {
785 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, FLAC_CAPS);
789 /* disabled because it happily typefinds /dev/urandom as audio/x-flac, and
790 * because I yet have to see header-less flac in the wild */
792 /* flac without headers (subset format) */
793 /* 64K should be enough */
794 while (c.offset < (64 * 1024)) {
795 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 4)))
798 /* look for frame header,
799 * http://flac.sourceforge.net/format.html#frame_header
801 if (c.data[0] == 0xff && (c.data[1] >> 2) == 0x3e) {
802 /* bit 15 in the header must be 0 */
803 if (((c.data[1] >> 1) & 0x01) == 0x01)
806 /* blocksize must be != 0x00 */
807 if ((c.data[2] >> 4) == 0x00)
810 /* samplerate must be != 0x0f */
811 if ((c.data[2] & 0x0f) == 0x0f)
813 /* also 0 is invalid, as it means get the info from the header and we
814 * don't have headers if we are here */
815 if ((c.data[2] & 0x0f) == 0x00)
818 /* channel assignment must be < 11 */
819 if ((c.data[3] >> 4) >= 11)
822 /* sample size must be != 0x07 and != 0x05 */
823 if (((c.data[3] >> 1) & 0x07) == 0x07)
825 if (((c.data[3] >> 1) & 0x07) == 0x05)
827 /* also 0 is invalid, as it means get the info from the header and we
828 * don't have headers if we are here */
829 if (((c.data[3] >> 1) & 0x07) == 0x00)
832 /* next bit must be 0 */
833 if ((c.data[3] & 0x01) == 0x01)
836 /* FIXME: shouldn't we include the crc check ? */
838 GST_DEBUG ("Found flac without headers at %d", (gint) c.offset);
839 gst_type_find_suggest (tf, GST_TYPE_FIND_POSSIBLE, FLAC_CAPS);
843 data_scan_ctx_advance (tf, &c, 1);
848 /* TODO: we could probably make a generic function for this.. */
850 aac_type_find_scan_loas_frames_ep (GstTypeFind * tf, DataScanCtx * scan_ctx,
853 DataScanCtx c = *scan_ctx;
859 if (!data_scan_ctx_ensure_data (tf, &c, 5))
862 /* EPAudioSyncStream */
863 len = ((c.data[2] & 0x0f) << 9) | (c.data[3] << 1) |
864 ((c.data[4] & 0x80) >> 7);
866 if (len == 0 || !data_scan_ctx_ensure_data (tf, &c, len + 2)) {
867 GST_DEBUG ("Wrong sync or next frame not within reach, len=%u", len);
871 /* check length of frame */
872 snc = GST_READ_UINT16_BE (c.data + len);
874 GST_DEBUG ("No sync found at 0x%" G_GINT64_MODIFIER "x", c.offset + len);
880 GST_DEBUG ("Found LOAS syncword #%d at offset 0x%" G_GINT64_MODIFIER "x, "
881 "framelen %u", count, c.offset, len);
883 data_scan_ctx_advance (tf, &c, len);
884 } while (count < max_frames && (c.offset - scan_ctx->offset) < 64 * 1024);
886 GST_DEBUG ("found %d consecutive frames", count);
891 aac_type_find_scan_loas_frames (GstTypeFind * tf, DataScanCtx * scan_ctx,
894 DataScanCtx c = *scan_ctx;
900 if (!data_scan_ctx_ensure_data (tf, &c, 3))
903 /* AudioSyncStream */
904 len = ((c.data[1] & 0x1f) << 8) | c.data[2];
905 /* add size of sync stream header */
908 if (len == 0 || !data_scan_ctx_ensure_data (tf, &c, len)) {
909 GST_DEBUG ("Wrong sync or next frame not within reach, len=%u", len);
913 /* check length of frame */
914 snc = GST_READ_UINT16_BE (c.data + len);
915 if ((snc & 0xffe0) != 0x56e0) {
916 GST_DEBUG ("No sync found at 0x%" G_GINT64_MODIFIER "x", c.offset + len);
922 GST_DEBUG ("Found LOAS syncword #%d at offset 0x%" G_GINT64_MODIFIER "x, "
923 "framelen %u", count, c.offset, len);
925 data_scan_ctx_advance (tf, &c, len);
926 } while (count < max_frames && (c.offset - scan_ctx->offset) < 64 * 1024);
928 GST_DEBUG ("found %d consecutive frames", count);
932 /*** audio/mpeg version 2, 4 ***/
934 static GstStaticCaps aac_caps = GST_STATIC_CAPS ("audio/mpeg, "
935 "mpegversion = (int) { 2, 4 }, framed = (bool) false");
936 #define AAC_CAPS (gst_static_caps_get(&aac_caps))
937 #define AAC_AMOUNT (4096)
939 aac_type_find (GstTypeFind * tf, gpointer unused)
941 DataScanCtx c = { 0, NULL, 0 };
942 GstTypeFindProbability best_probability = GST_TYPE_FIND_NONE;
943 GstCaps *best_caps = NULL;
944 guint best_count = 0;
946 while (c.offset < AAC_AMOUNT) {
949 /* detect adts header or adif header.
950 * The ADIF header is 4 bytes, that should be OK. The ADTS header, on
951 * the other hand, is 14 bits only, so we require one valid frame with
952 * again a valid syncpoint on the next one (28 bits) for certainty. We
953 * require 4 kB, which is quite a lot, since frames are generally 200-400
955 * LOAS has 2 possible syncwords, which are 11 bits and 16 bits long.
956 * The following stream syntax depends on which one is found.
958 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 6)))
961 snc = GST_READ_UINT16_BE (c.data);
962 if (G_UNLIKELY ((snc & 0xfff6) == 0xfff0)) {
963 /* ADTS header - find frame length */
964 GST_DEBUG ("Found one ADTS syncpoint at offset 0x%" G_GINT64_MODIFIER
965 "x, tracing next...", c.offset);
966 len = ((c.data[3] & 0x03) << 11) |
967 (c.data[4] << 3) | ((c.data[5] & 0xe0) >> 5);
969 if (len == 0 || !data_scan_ctx_ensure_data (tf, &c, len + 2)) {
970 GST_DEBUG ("Wrong sync or next frame not within reach, len=%u", len);
974 /* check if there's a second ADTS frame */
975 snc = GST_READ_UINT16_BE (c.data + len);
976 if ((snc & 0xfff6) == 0xfff0) {
978 guint mpegversion, sample_freq_idx, channel_config, profile_idx, rate;
979 guint8 audio_config[2];
981 mpegversion = (c.data[1] & 0x08) ? 2 : 4;
982 profile_idx = c.data[2] >> 6;
983 sample_freq_idx = ((c.data[2] & 0x3c) >> 2);
984 channel_config = ((c.data[2] & 0x01) << 2) + (c.data[3] >> 6);
986 GST_DEBUG ("Found second ADTS-%d syncpoint at offset 0x%"
987 G_GINT64_MODIFIER "x, framelen %u", mpegversion, c.offset, len);
989 /* 0xd and 0xe are reserved. 0xf means the sample frequency is directly
990 * specified in the header, but that's not allowed for ADTS */
991 if (sample_freq_idx > 0xc) {
992 GST_DEBUG ("Unexpected sample frequency index %d or wrong sync",
997 rate = gst_codec_utils_aac_get_sample_rate_from_index (sample_freq_idx);
998 GST_LOG ("ADTS: profile=%u, rate=%u", profile_idx, rate);
1000 /* The ADTS frame header is slightly different from the
1001 * AudioSpecificConfig defined for the MPEG-4 container, so we just
1002 * construct enough of it for getting the level here. */
1003 /* ADTS counts profiles from 0 instead of 1 to save bits */
1004 audio_config[0] = (profile_idx + 1) << 3;
1005 audio_config[0] |= (sample_freq_idx >> 1) & 0x7;
1006 audio_config[1] = (sample_freq_idx & 0x1) << 7;
1007 audio_config[1] |= (channel_config & 0xf) << 3;
1009 caps = gst_caps_new_simple ("audio/mpeg",
1010 "framed", G_TYPE_BOOLEAN, FALSE,
1011 "mpegversion", G_TYPE_INT, mpegversion,
1012 "stream-format", G_TYPE_STRING, "adts", NULL);
1014 gst_codec_utils_aac_caps_set_level_and_profile (caps, audio_config, 2);
1016 /* add rate and number of channels if we can */
1017 if (channel_config != 0 && channel_config <= 7) {
1018 const guint channels_map[] = { 0, 1, 2, 3, 4, 5, 6, 8 };
1020 gst_caps_set_simple (caps, "channels", G_TYPE_INT,
1021 channels_map[channel_config], "rate", G_TYPE_INT, rate, NULL);
1024 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY, caps);
1025 gst_caps_unref (caps);
1029 GST_DEBUG ("No next frame found... (should have been at 0x%x)", len);
1030 } else if (G_UNLIKELY ((snc & 0xffe0) == 0x56e0 || snc == 0x4de1)) {
1034 GST_INFO ("Possible LOAS syncword at offset 0x%" G_GINT64_MODIFIER
1035 "x, scanning for more frames...", c.offset);
1038 count = aac_type_find_scan_loas_frames_ep (tf, &c, 20);
1040 count = aac_type_find_scan_loas_frames (tf, &c, 20);
1042 if (count >= 3 && count > best_count) {
1043 gst_caps_replace (&best_caps, NULL);
1044 best_caps = gst_caps_new_simple ("audio/mpeg",
1045 "framed", G_TYPE_BOOLEAN, FALSE,
1046 "mpegversion", G_TYPE_INT, 4,
1047 "stream-format", G_TYPE_STRING, "loas", NULL);
1049 best_probability = GST_TYPE_FIND_POSSIBLE - 10 + count * 3;
1050 if (best_probability >= GST_TYPE_FIND_LIKELY)
1053 } else if (!memcmp (c.data, "ADIF", 4)) {
1055 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_LIKELY, "audio/mpeg",
1056 "framed", G_TYPE_BOOLEAN, FALSE, "mpegversion", G_TYPE_INT, 4,
1057 "stream-format", G_TYPE_STRING, "adif", NULL);
1063 data_scan_ctx_advance (tf, &c, 1);
1066 if (best_probability > GST_TYPE_FIND_NONE) {
1067 gst_type_find_suggest (tf, best_probability, best_caps);
1068 gst_caps_unref (best_caps);
1072 /*** audio/mpeg version 1 ***/
1075 * The chance that random data is identified as a valid mp3 header is 63 / 2^18
1076 * (0.024%) per try. This makes the function for calculating false positives
1077 * 1 - (1 - ((63 / 2 ^18) ^ GST_MP3_TYPEFIND_MIN_HEADERS)) ^ buffersize)
1078 * This has the following probabilities of false positives:
1079 * datasize MIN_HEADERS
1081 * 4096 62.6% 0.02% 0% 0%
1082 * 16384 98% 0.09% 0% 0%
1083 * 1 MiB 100% 5.88% 0% 0%
1084 * 1 GiB 100% 100% 1.44% 0%
1085 * 1 TiB 100% 100% 100% 0.35%
1086 * This means that the current choice (3 headers by most of the time 4096 byte
1087 * buffers is pretty safe for now.
1089 * The max. size of each frame is 1440 bytes, which means that for N frames to
1090 * be detected, we need 1440 * GST_MP3_TYPEFIND_MIN_HEADERS + 3 bytes of data.
1091 * Assuming we step into the stream right after the frame header, this
1092 * means we need 1440 * (GST_MP3_TYPEFIND_MIN_HEADERS + 1) - 1 + 3 bytes
1093 * of data (5762) to always detect any mp3.
1096 static const guint mp3types_bitrates[2][3][16] =
1097 { {{0, 32, 64, 96, 128, 160, 192, 224, 256, 288, 320, 352, 384, 416, 448,},
1098 {0, 32, 48, 56, 64, 80, 96, 112, 128, 160, 192, 224, 256, 320, 384,},
1099 {0, 32, 40, 48, 56, 64, 80, 96, 112, 128, 160, 192, 224, 256, 320,}},
1100 {{0, 32, 48, 56, 64, 80, 96, 112, 128, 144, 160, 176, 192, 224, 256,},
1101 {0, 8, 16, 24, 32, 40, 48, 56, 64, 80, 96, 112, 128, 144, 160,},
1102 {0, 8, 16, 24, 32, 40, 48, 56, 64, 80, 96, 112, 128, 144, 160,}},
1105 static const guint mp3types_freqs[3][3] = { {11025, 12000, 8000},
1106 {22050, 24000, 16000},
1107 {44100, 48000, 32000}
1111 mp3_type_frame_length_from_header (guint32 header, guint * put_layer,
1112 guint * put_channels, guint * put_bitrate, guint * put_samplerate,
1113 gboolean * may_be_free_format, gint possible_free_framelen)
1115 guint bitrate, layer, length, mode, samplerate, version, channels;
1117 if ((header & 0xffe00000) != 0xffe00000)
1120 /* we don't need extension, copyright, original or
1121 * emphasis for the frame length */
1125 mode = header & 0x3;
1129 length = header & 0x1;
1132 /* sampling frequency */
1133 samplerate = header & 0x3;
1134 if (samplerate == 3)
1139 bitrate = header & 0xF;
1140 if (bitrate == 0 && possible_free_framelen == -1) {
1141 GST_LOG ("Possibly a free format mp3 - signaling");
1142 *may_be_free_format = TRUE;
1144 if (bitrate == 15 || (bitrate == 0 && possible_free_framelen == -1))
1147 /* ignore error correction, too */
1151 layer = 4 - (header & 0x3);
1156 /* version 0=MPEG2.5; 2=MPEG2; 3=MPEG1 */
1157 version = header & 0x3;
1162 channels = (mode == 3) ? 1 : 2;
1163 samplerate = mp3types_freqs[version > 0 ? version - 1 : 0][samplerate];
1167 length += possible_free_framelen;
1168 bitrate = length * samplerate / 48000;
1170 length += possible_free_framelen;
1171 bitrate = length * samplerate /
1172 ((layer == 3 && version != 3) ? 72000 : 144000);
1176 bitrate = mp3types_bitrates[version == 3 ? 0 : 1][layer - 1][bitrate];
1178 length = ((12000 * bitrate / samplerate) + length) * 4;
1180 length += ((layer == 3
1181 && version != 3) ? 72000 : 144000) * bitrate / samplerate;
1185 GST_LOG ("mp3typefind: calculated mp3 frame length of %u bytes", length);
1187 ("mp3typefind: samplerate = %u - bitrate = %u - layer = %u - version = %u"
1188 " - channels = %u", samplerate, bitrate, layer, version, channels);
1193 *put_channels = channels;
1195 *put_bitrate = bitrate;
1197 *put_samplerate = samplerate;
1203 static GstStaticCaps mp3_caps = GST_STATIC_CAPS ("audio/mpeg, "
1204 "mpegversion = (int) 1, layer = (int) [ 1, 3 ]");
1205 #define MP3_CAPS (gst_static_caps_get(&mp3_caps))
1207 * random values for typefinding
1208 * if no more data is available, we will return a probability of
1209 * (found_headers/TRY_HEADERS) * (MAXIMUM * (TRY_SYNC - bytes_skipped)
1211 * if found_headers >= MIN_HEADERS
1213 #define GST_MP3_TYPEFIND_MIN_HEADERS (2)
1214 #define GST_MP3_TYPEFIND_TRY_HEADERS (5)
1215 #define GST_MP3_TYPEFIND_TRY_SYNC (GST_TYPE_FIND_MAXIMUM * 100) /* 10kB */
1216 #define GST_MP3_TYPEFIND_SYNC_SIZE (2048)
1217 #define GST_MP3_WRONG_HEADER (10)
1220 mp3_type_find_at_offset (GstTypeFind * tf, guint64 start_off,
1221 guint * found_layer, GstTypeFindProbability * found_prob)
1223 const guint8 *data = NULL;
1224 const guint8 *data_end = NULL;
1227 gint last_free_offset = -1;
1228 gint last_free_framelen = -1;
1229 gboolean headerstart = TRUE;
1236 while (skipped < GST_MP3_TYPEFIND_TRY_SYNC) {
1238 size = GST_MP3_TYPEFIND_SYNC_SIZE * 2;
1241 data = gst_type_find_peek (tf, skipped + start_off, size);
1242 } while (size > 10 && !data);
1245 data_end = data + size;
1247 if (*data == 0xFF) {
1248 const guint8 *head_data = NULL;
1249 guint layer = 0, bitrate, samplerate, channels;
1250 guint found = 0; /* number of valid headers found */
1251 guint64 offset = skipped;
1252 gboolean changed = FALSE;
1254 while (found < GST_MP3_TYPEFIND_TRY_HEADERS) {
1257 guint prev_layer = 0;
1258 guint prev_channels = 0, prev_samplerate = 0;
1259 gboolean free = FALSE;
1261 if ((gint64) (offset - skipped + 4) >= 0 &&
1262 data + offset - skipped + 4 < data_end) {
1263 head_data = data + offset - skipped;
1265 head_data = gst_type_find_peek (tf, offset + start_off, 4);
1269 head = GST_READ_UINT32_BE (head_data);
1270 if (!(length = mp3_type_frame_length_from_header (head, &layer,
1271 &channels, &bitrate, &samplerate, &free,
1272 last_free_framelen))) {
1274 if (last_free_offset == -1)
1275 last_free_offset = offset;
1277 last_free_framelen = offset - last_free_offset;
1278 offset = last_free_offset;
1282 last_free_framelen = -1;
1285 /* Mark the fact that we didn't find a valid header at the beginning */
1287 headerstart = FALSE;
1289 GST_LOG ("%d. header at offset %" G_GUINT64_FORMAT
1290 " (0x%" G_GINT64_MODIFIER "x) was not an mp3 header "
1291 "(possibly-free: %s)", found + 1, start_off + offset,
1292 start_off + offset, free ? "yes" : "no");
1295 if ((prev_layer && prev_layer != layer) ||
1296 /* (prev_bitrate && prev_bitrate != bitrate) || <-- VBR */
1297 (prev_samplerate && prev_samplerate != samplerate) ||
1298 (prev_channels && prev_channels != channels)) {
1299 /* this means an invalid property, or a change, which might mean
1300 * that this is not a mp3 but just a random bytestream. It could
1301 * be a freaking funky encoded mp3 though. We'll just not count
1306 prev_channels = channels;
1307 prev_samplerate = samplerate;
1310 GST_LOG ("found %d. header at offset %" G_GUINT64_FORMAT " (0x%"
1311 G_GINT64_MODIFIER "X)", found, start_off + offset,
1312 start_off + offset);
1316 g_assert (found <= GST_MP3_TYPEFIND_TRY_HEADERS);
1317 if (head_data == NULL &&
1318 gst_type_find_peek (tf, offset + start_off - 1, 1) == NULL)
1319 /* Incomplete last frame - don't count it. */
1321 if (found == GST_MP3_TYPEFIND_TRY_HEADERS ||
1322 (found >= GST_MP3_TYPEFIND_MIN_HEADERS && head_data == NULL)) {
1323 /* we can make a valid guess */
1324 guint probability = found * GST_TYPE_FIND_MAXIMUM *
1325 (GST_MP3_TYPEFIND_TRY_SYNC - skipped) /
1326 GST_MP3_TYPEFIND_TRY_HEADERS / GST_MP3_TYPEFIND_TRY_SYNC;
1329 && probability > (GST_TYPE_FIND_MINIMUM + GST_MP3_WRONG_HEADER))
1330 probability -= GST_MP3_WRONG_HEADER;
1331 if (probability < GST_TYPE_FIND_MINIMUM)
1332 probability = GST_TYPE_FIND_MINIMUM;
1336 probability = (probability + GST_TYPE_FIND_MAXIMUM) / 2;
1339 ("audio/mpeg calculated %u = %u * %u / %u * (%u - %"
1340 G_GUINT64_FORMAT ") / %u", probability, GST_TYPE_FIND_MAXIMUM,
1341 found, GST_MP3_TYPEFIND_TRY_HEADERS, GST_MP3_TYPEFIND_TRY_SYNC,
1342 (guint64) skipped, GST_MP3_TYPEFIND_TRY_SYNC);
1343 /* make sure we're not id3 tagged */
1344 head_data = gst_type_find_peek (tf, -128, 3);
1345 if (head_data && (memcmp (head_data, "TAG", 3) == 0)) {
1348 g_assert (probability <= GST_TYPE_FIND_MAXIMUM);
1350 *found_prob = probability;
1351 if (probability > 0)
1352 *found_layer = layer;
1363 mp3_type_find (GstTypeFind * tf, gpointer unused)
1365 GstTypeFindProbability prob, mid_prob;
1367 guint layer, mid_layer;
1370 mp3_type_find_at_offset (tf, 0, &layer, &prob);
1371 length = gst_type_find_get_length (tf);
1373 if (length == 0 || length == (guint64) - 1) {
1379 /* if we're pretty certain already, skip the additional check */
1380 if (prob >= GST_TYPE_FIND_LIKELY)
1383 mp3_type_find_at_offset (tf, length / 2, &mid_layer, &mid_prob);
1387 GST_LOG ("detected audio/mpeg only in the middle (p=%u)", mid_prob);
1393 if (layer != mid_layer) {
1394 GST_WARNING ("audio/mpeg layer discrepancy: %u vs. %u", layer, mid_layer);
1395 return; /* FIXME: or should we just go with the one in the middle? */
1398 /* detected mpeg audio both in middle of the file and at the start */
1399 prob = (prob + mid_prob) / 2;
1403 /* let's see if there's a valid header right at the start */
1404 data = gst_type_find_peek (tf, 0, 4); /* use min. frame size? */
1405 if (data && mp3_type_frame_length_from_header (GST_READ_UINT32_BE (data),
1406 &layer, NULL, NULL, NULL, NULL, 0) != 0) {
1408 prob = GST_TYPE_FIND_POSSIBLE - 10;
1410 prob = MAX (GST_TYPE_FIND_POSSIBLE - 10, prob + 10);
1420 g_return_if_fail (layer >= 1 && layer <= 3);
1422 gst_type_find_suggest_simple (tf, prob, "audio/mpeg",
1423 "mpegversion", G_TYPE_INT, 1, "layer", G_TYPE_INT, layer, NULL);
1427 /*** audio/x-musepack ***/
1429 static GstStaticCaps musepack_caps =
1430 GST_STATIC_CAPS ("audio/x-musepack, streamversion= (int) { 7, 8 }");
1432 #define MUSEPACK_CAPS (gst_static_caps_get(&musepack_caps))
1434 musepack_type_find (GstTypeFind * tf, gpointer unused)
1436 const guint8 *data = gst_type_find_peek (tf, 0, 4);
1437 GstTypeFindProbability prop = GST_TYPE_FIND_MINIMUM;
1438 gint streamversion = -1;
1440 if (data && memcmp (data, "MP+", 3) == 0) {
1442 if ((data[3] & 0x7f) == 7) {
1443 prop = GST_TYPE_FIND_MAXIMUM;
1445 prop = GST_TYPE_FIND_LIKELY + 10;
1447 } else if (data && memcmp (data, "MPCK", 4) == 0) {
1449 prop = GST_TYPE_FIND_MAXIMUM;
1452 if (streamversion != -1) {
1453 gst_type_find_suggest_simple (tf, prop, "audio/x-musepack",
1454 "streamversion", G_TYPE_INT, streamversion, NULL);
1458 /*** audio/x-ac3 ***/
1459 /* FIXME 0.11: should be audio/ac3, but isn't for backwards compatibility */
1460 static GstStaticCaps ac3_caps = GST_STATIC_CAPS ("audio/x-ac3");
1462 #define AC3_CAPS (gst_static_caps_get(&ac3_caps))
1464 static GstStaticCaps eac3_caps = GST_STATIC_CAPS ("audio/x-eac3");
1466 #define EAC3_CAPS (gst_static_caps_get(&eac3_caps))
1470 unsigned short bit_rate;
1471 unsigned short frm_size[3];
1474 static const struct ac3_frmsize ac3_frmsizecod_tbl[] = {
1477 {40, {80, 87, 120}},
1478 {40, {80, 88, 120}},
1479 {48, {96, 104, 144}},
1480 {48, {96, 105, 144}},
1481 {56, {112, 121, 168}},
1482 {56, {112, 122, 168}},
1483 {64, {128, 139, 192}},
1484 {64, {128, 140, 192}},
1485 {80, {160, 174, 240}},
1486 {80, {160, 175, 240}},
1487 {96, {192, 208, 288}},
1488 {96, {192, 209, 288}},
1489 {112, {224, 243, 336}},
1490 {112, {224, 244, 336}},
1491 {128, {256, 278, 384}},
1492 {128, {256, 279, 384}},
1493 {160, {320, 348, 480}},
1494 {160, {320, 349, 480}},
1495 {192, {384, 417, 576}},
1496 {192, {384, 418, 576}},
1497 {224, {448, 487, 672}},
1498 {224, {448, 488, 672}},
1499 {256, {512, 557, 768}},
1500 {256, {512, 558, 768}},
1501 {320, {640, 696, 960}},
1502 {320, {640, 697, 960}},
1503 {384, {768, 835, 1152}},
1504 {384, {768, 836, 1152}},
1505 {448, {896, 975, 1344}},
1506 {448, {896, 976, 1344}},
1507 {512, {1024, 1114, 1536}},
1508 {512, {1024, 1115, 1536}},
1509 {576, {1152, 1253, 1728}},
1510 {576, {1152, 1254, 1728}},
1511 {640, {1280, 1393, 1920}},
1512 {640, {1280, 1394, 1920}}
1516 ac3_type_find (GstTypeFind * tf, gpointer unused)
1518 DataScanCtx c = { 0, NULL, 0 };
1520 /* Search for an ac3 frame; not necessarily right at the start, but give it
1521 * a lower probability if not found right at the start. Check that the
1522 * frame is followed by a second frame at the expected offset.
1523 * We could also check the two ac3 CRCs, but we don't do that right now */
1524 while (c.offset < 1024) {
1525 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 5)))
1528 if (c.data[0] == 0x0b && c.data[1] == 0x77) {
1529 guint bsid = c.data[5] >> 3;
1533 guint fscod = c.data[4] >> 6;
1534 guint frmsizecod = c.data[4] & 0x3f;
1536 if (fscod < 3 && frmsizecod < 38) {
1537 DataScanCtx c_next = c;
1540 frame_size = ac3_frmsizecod_tbl[frmsizecod].frm_size[fscod];
1541 GST_LOG ("possible AC3 frame sync at offset %"
1542 G_GUINT64_FORMAT ", size=%u", c.offset, frame_size);
1543 if (data_scan_ctx_ensure_data (tf, &c_next, (frame_size * 2) + 5)) {
1544 data_scan_ctx_advance (tf, &c_next, frame_size * 2);
1546 if (c_next.data[0] == 0x0b && c_next.data[1] == 0x77) {
1547 fscod = c_next.data[4] >> 6;
1548 frmsizecod = c_next.data[4] & 0x3f;
1550 if (fscod < 3 && frmsizecod < 38) {
1551 GstTypeFindProbability prob;
1553 GST_LOG ("found second AC3 frame (size=%u), looks good",
1554 ac3_frmsizecod_tbl[frmsizecod].frm_size[fscod]);
1556 prob = GST_TYPE_FIND_MAXIMUM;
1558 prob = GST_TYPE_FIND_NEARLY_CERTAIN;
1560 gst_type_find_suggest (tf, prob, AC3_CAPS);
1564 GST_LOG ("no second AC3 frame found, false sync");
1568 } else if (bsid <= 16 && bsid > 10) {
1570 DataScanCtx c_next = c;
1573 frame_size = (((c.data[2] & 0x07) << 8) + c.data[3]) + 1;
1574 GST_LOG ("possible E-AC3 frame sync at offset %"
1575 G_GUINT64_FORMAT ", size=%u", c.offset, frame_size);
1576 if (data_scan_ctx_ensure_data (tf, &c_next, (frame_size * 2) + 5)) {
1577 data_scan_ctx_advance (tf, &c_next, frame_size * 2);
1579 if (c_next.data[0] == 0x0b && c_next.data[1] == 0x77) {
1580 GstTypeFindProbability prob;
1582 GST_LOG ("found second E-AC3 frame, looks good");
1584 prob = GST_TYPE_FIND_MAXIMUM;
1586 prob = GST_TYPE_FIND_NEARLY_CERTAIN;
1588 gst_type_find_suggest (tf, prob, EAC3_CAPS);
1591 GST_LOG ("no second E-AC3 frame found, false sync");
1595 GST_LOG ("invalid AC3 BSID: %u", bsid);
1598 data_scan_ctx_advance (tf, &c, 1);
1602 /*** audio/x-dts ***/
1603 static GstStaticCaps dts_caps = GST_STATIC_CAPS ("audio/x-dts");
1604 #define DTS_CAPS (gst_static_caps_get (&dts_caps))
1605 #define DTS_MIN_FRAMESIZE 96
1606 #define DTS_MAX_FRAMESIZE 18725 /* 16384*16/14 */
1609 dts_parse_frame_header (DataScanCtx * c, guint * frame_size,
1610 guint * sample_rate, guint * channels, guint * depth, guint * endianness)
1612 static const int sample_rates[16] = { 0, 8000, 16000, 32000, 0, 0, 11025,
1613 22050, 44100, 0, 0, 12000, 24000, 48000, 96000, 192000
1615 static const guint8 channels_table[16] = { 1, 2, 2, 2, 2, 3, 3, 4, 4, 5,
1620 guint num_blocks, chans, lfe, i;
1622 marker = GST_READ_UINT32_BE (c->data);
1624 /* raw big endian or 14-bit big endian */
1625 if (marker == 0x7FFE8001 || marker == 0x1FFFE800) {
1626 *endianness = G_BIG_ENDIAN;
1627 for (i = 0; i < G_N_ELEMENTS (hdr); ++i)
1628 hdr[i] = GST_READ_UINT16_BE (c->data + (i * sizeof (guint16)));
1630 /* raw little endian or 14-bit little endian */
1631 if (marker == 0xFE7F0180 || marker == 0xFF1F00E8) {
1632 *endianness = G_LITTLE_ENDIAN;
1633 for (i = 0; i < G_N_ELEMENTS (hdr); ++i)
1634 hdr[i] = GST_READ_UINT16_LE (c->data + (i * sizeof (guint16)));
1639 GST_LOG ("dts sync marker 0x%08x at offset %u", marker, (guint) c->offset);
1642 if (marker == 0x1FFFE800 || marker == 0xFF1F00E8) {
1643 if ((hdr[2] & 0xFFF0) != 0x07F0)
1645 /* discard top 2 bits (2 void), shift in 2 */
1646 hdr[0] = (hdr[0] << 2) | ((hdr[1] >> 12) & 0x0003);
1647 /* discard top 4 bits (2 void, 2 shifted into hdr[0]), shift in 4 etc. */
1648 hdr[1] = (hdr[1] << 4) | ((hdr[2] >> 10) & 0x000F);
1649 hdr[2] = (hdr[2] << 6) | ((hdr[3] >> 8) & 0x003F);
1650 hdr[3] = (hdr[3] << 8) | ((hdr[4] >> 6) & 0x00FF);
1651 hdr[4] = (hdr[4] << 10) | ((hdr[5] >> 4) & 0x03FF);
1652 hdr[5] = (hdr[5] << 12) | ((hdr[6] >> 2) & 0x0FFF);
1653 hdr[6] = (hdr[6] << 14) | ((hdr[7] >> 0) & 0x3FFF);
1654 g_assert (hdr[0] == 0x7FFE && hdr[1] == 0x8001);
1660 GST_LOG ("frame header: %04x%04x%04x%04x", hdr[2], hdr[3], hdr[4], hdr[5]);
1662 num_blocks = (hdr[2] >> 2) & 0x7F;
1663 *frame_size = (((hdr[2] & 0x03) << 12) | (hdr[3] >> 4)) + 1;
1664 chans = ((hdr[3] & 0x0F) << 2) | (hdr[4] >> 14);
1665 *sample_rate = sample_rates[(hdr[4] >> 10) & 0x0F];
1666 lfe = (hdr[5] >> 9) & 0x03;
1668 if (num_blocks < 5 || *frame_size < 96 || *sample_rate == 0)
1671 if (marker == 0x1FFFE800 || marker == 0xFF1F00E8)
1672 *frame_size = (*frame_size * 16) / 14; /* FIXME: round up? */
1674 if (chans < G_N_ELEMENTS (channels_table))
1675 *channels = channels_table[chans] + ((lfe) ? 1 : 0);
1683 dts_type_find (GstTypeFind * tf, gpointer unused)
1685 DataScanCtx c = { 0, NULL, 0 };
1687 /* Search for an dts frame; not necessarily right at the start, but give it
1688 * a lower probability if not found right at the start. Check that the
1689 * frame is followed by a second frame at the expected offset. */
1690 while (c.offset <= DTS_MAX_FRAMESIZE) {
1691 guint frame_size = 0, rate = 0, chans = 0, depth = 0, endianness = 0;
1693 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, DTS_MIN_FRAMESIZE)))
1696 if (G_UNLIKELY (dts_parse_frame_header (&c, &frame_size, &rate, &chans,
1697 &depth, &endianness))) {
1698 GstTypeFindProbability prob;
1701 prob = (c.offset == 0) ? GST_TYPE_FIND_LIKELY : GST_TYPE_FIND_POSSIBLE;
1703 /* check for second frame sync */
1705 data_scan_ctx_advance (tf, &next_c, frame_size);
1706 if (data_scan_ctx_ensure_data (tf, &next_c, 4)) {
1707 GST_LOG ("frame size: %u 0x%04x", frame_size, frame_size);
1708 GST_MEMDUMP ("second frame sync", next_c.data, 4);
1709 if (GST_READ_UINT32_BE (c.data) == GST_READ_UINT32_BE (next_c.data))
1710 prob = GST_TYPE_FIND_MAXIMUM;
1714 gst_type_find_suggest_simple (tf, prob, "audio/x-dts",
1715 "rate", G_TYPE_INT, rate, "channels", G_TYPE_INT, chans,
1716 "depth", G_TYPE_INT, depth, "endianness", G_TYPE_INT, endianness,
1717 "framed", G_TYPE_BOOLEAN, FALSE, NULL);
1719 gst_type_find_suggest_simple (tf, prob, "audio/x-dts",
1720 "rate", G_TYPE_INT, rate, "depth", G_TYPE_INT, depth,
1721 "endianness", G_TYPE_INT, endianness,
1722 "framed", G_TYPE_BOOLEAN, FALSE, NULL);
1728 data_scan_ctx_advance (tf, &c, 1);
1734 /* can only be detected by using the extension, in which case we use the default
1736 static GstStaticCaps gsm_caps =
1737 GST_STATIC_CAPS ("audio/x-gsm, rate=8000, channels=1");
1739 #define GSM_CAPS (gst_static_caps_get(&gsm_caps))
1743 static GstStaticCaps wavpack_caps =
1744 GST_STATIC_CAPS ("audio/x-wavpack, framed = (boolean) false");
1746 #define WAVPACK_CAPS (gst_static_caps_get(&wavpack_caps))
1748 static GstStaticCaps wavpack_correction_caps =
1749 GST_STATIC_CAPS ("audio/x-wavpack-correction, framed = (boolean) false");
1751 #define WAVPACK_CORRECTION_CAPS (gst_static_caps_get(&wavpack_correction_caps))
1754 wavpack_type_find (GstTypeFind * tf, gpointer unused)
1756 GstTypeFindProbability base_prob = GST_TYPE_FIND_POSSIBLE;
1760 guint count_wv, count_wvc;
1762 data = gst_type_find_peek (tf, 0, 32);
1766 if (data[0] != 'w' || data[1] != 'v' || data[2] != 'p' || data[3] != 'k')
1769 /* Note: wavpack blocks can be fairly large (easily 60-110k), possibly
1770 * larger than the max. limits imposed by certain typefinding elements
1771 * like id3demux or apedemux, so typefinding is most likely only going to
1772 * work in pull-mode */
1773 blocksize = GST_READ_UINT32_LE (data + 4);
1774 GST_LOG ("wavpack header, blocksize=0x%04x", blocksize);
1778 while (offset < 8 + blocksize) {
1781 /* get chunk header */
1782 GST_LOG ("peeking at chunk at offset 0x%04x", (guint) offset);
1783 data = gst_type_find_peek (tf, offset, 4);
1786 sublen = ((guint32) data[1]) << 1;
1787 if (data[0] & 0x80) {
1788 sublen |= (((guint32) data[2]) << 9) | (((guint32) data[3]) << 17);
1789 sublen += 1 + 3; /* id + length */
1791 sublen += 1 + 1; /* id + length */
1793 if (offset + sublen > 8 + blocksize) {
1794 GST_LOG ("chunk length too big (%u > %" G_GUINT64_FORMAT ")", sublen,
1795 blocksize - offset);
1798 if ((data[0] & 0x20) == 0) {
1799 switch (data[0] & 0x0f) {
1800 case 0xa: /* ID_WV_BITSTREAM */
1801 case 0xc: /* ID_WVX_BITSTREAM */
1804 case 0xb: /* ID_WVC_BITSTREAM */
1810 if (count_wv >= 5 || count_wvc >= 5)
1816 /* check for second block header */
1817 data = gst_type_find_peek (tf, 8 + blocksize, 4);
1818 if (data != NULL && memcmp (data, "wvpk", 4) == 0) {
1819 GST_DEBUG ("found second block sync");
1820 base_prob = GST_TYPE_FIND_LIKELY;
1823 GST_DEBUG ("wvc=%d, wv=%d", count_wvc, count_wv);
1825 if (count_wvc > 0 && count_wvc > count_wv) {
1826 gst_type_find_suggest (tf,
1827 MIN (base_prob + 5 * count_wvc, GST_TYPE_FIND_NEARLY_CERTAIN),
1828 WAVPACK_CORRECTION_CAPS);
1829 } else if (count_wv > 0) {
1830 gst_type_find_suggest (tf,
1831 MIN (base_prob + 5 * count_wv, GST_TYPE_FIND_NEARLY_CERTAIN),
1836 /*** application/postscrip ***/
1837 static GstStaticCaps postscript_caps =
1838 GST_STATIC_CAPS ("application/postscript");
1840 #define POSTSCRIPT_CAPS (gst_static_caps_get(&postscript_caps))
1843 postscript_type_find (GstTypeFind * tf, gpointer unused)
1845 const guint8 *data = gst_type_find_peek (tf, 0, 3);
1849 if (data[0] == 0x04)
1851 if (data[0] == '%' && data[1] == '!')
1852 gst_type_find_suggest (tf, GST_TYPE_FIND_POSSIBLE, POSTSCRIPT_CAPS);
1856 /*** image/svg+xml ***/
1857 static GstStaticCaps svg_caps = GST_STATIC_CAPS ("image/svg+xml");
1859 #define SVG_CAPS (gst_static_caps_get(&svg_caps))
1862 svg_type_find (GstTypeFind * tf, gpointer unused)
1864 static const gchar svg_doctype[] = "!DOCTYPE svg";
1865 static const gchar svg_tag[] = "<svg";
1866 DataScanCtx c = { 0, NULL, 0 };
1868 while (c.offset <= 1024) {
1869 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 12)))
1872 if (memcmp (svg_doctype, c.data, 12) == 0) {
1873 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, SVG_CAPS);
1875 } else if (memcmp (svg_tag, c.data, 4) == 0) {
1876 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY, SVG_CAPS);
1879 data_scan_ctx_advance (tf, &c, 1);
1883 /*** multipart/x-mixed-replace mimestream ***/
1885 static GstStaticCaps multipart_caps =
1886 GST_STATIC_CAPS ("multipart/x-mixed-replace");
1887 #define MULTIPART_CAPS gst_static_caps_get(&multipart_caps)
1889 /* multipart/x-mixed replace is:
1890 * <maybe some whitespace>--<some ascii chars>[\r]\n
1891 * <more ascii chars>[\r]\nContent-type:<more ascii>[\r]\n */
1893 multipart_type_find (GstTypeFind * tf, gpointer unused)
1898 #define MULTIPART_MAX_BOUNDARY_OFFSET 16
1899 data = gst_type_find_peek (tf, 0, MULTIPART_MAX_BOUNDARY_OFFSET);
1904 x - data < MULTIPART_MAX_BOUNDARY_OFFSET - 2 && g_ascii_isspace (*x);
1906 if (x[0] != '-' || x[1] != '-')
1909 /* Could be okay, peek what should be enough for a complete header */
1910 #define MULTIPART_MAX_HEADER_SIZE 256
1911 data = gst_type_find_peek (tf, 0, MULTIPART_MAX_HEADER_SIZE);
1915 for (x = data; x - data < MULTIPART_MAX_HEADER_SIZE - 14; x++) {
1916 if (!isascii (*x)) {
1920 !g_ascii_strncasecmp ("content-type:", (gchar *) x + 1, 13)) {
1921 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MULTIPART_CAPS);
1927 /*** video/mpeg systemstream ***/
1928 static GstStaticCaps mpeg_sys_caps = GST_STATIC_CAPS ("video/mpeg, "
1929 "systemstream = (boolean) true, mpegversion = (int) [ 1, 2 ]");
1931 #define MPEG_SYS_CAPS gst_static_caps_get(&mpeg_sys_caps)
1932 #define IS_MPEG_HEADER(data) (G_UNLIKELY((((guint8 *)(data))[0] == 0x00) && \
1933 (((guint8 *)(data))[1] == 0x00) && \
1934 (((guint8 *)(data))[2] == 0x01)))
1936 #define IS_MPEG_PACK_CODE(b) ((b) == 0xBA)
1937 #define IS_MPEG_SYS_CODE(b) ((b) == 0xBB)
1938 #define IS_MPEG_PACK_HEADER(data) (IS_MPEG_HEADER (data) && \
1939 IS_MPEG_PACK_CODE (((guint8 *)(data))[3]))
1941 #define IS_MPEG_PES_CODE(b) (((b) & 0xF0) == 0xE0 || ((b) & 0xF0) == 0xC0 || \
1943 #define IS_MPEG_PES_HEADER(data) (IS_MPEG_HEADER (data) && \
1944 IS_MPEG_PES_CODE (((guint8 *)(data))[3]))
1946 #define MPEG2_MAX_PROBE_LENGTH (128 * 1024) /* 128kB should be 64 packs of the
1947 * most common 2kB pack size. */
1949 #define MPEG2_MIN_SYS_HEADERS 2
1950 #define MPEG2_MAX_SYS_HEADERS 5
1953 mpeg_sys_is_valid_pack (GstTypeFind * tf, const guint8 * data, guint len,
1956 /* Check the pack header @ offset for validity, assuming that the 4 byte header
1957 * itself has already been checked. */
1963 /* Check marker bits */
1964 if ((data[4] & 0xC4) == 0x44) {
1969 if ((data[6] & 0x04) != 0x04 ||
1970 (data[8] & 0x04) != 0x04 ||
1971 (data[9] & 0x01) != 0x01 || (data[12] & 0x03) != 0x03)
1974 stuff_len = data[13] & 0x07;
1976 /* Check the following header bytes, if we can */
1977 if ((14 + stuff_len + 4) <= len) {
1978 if (!IS_MPEG_HEADER (data + 14 + stuff_len))
1982 *pack_size = 14 + stuff_len;
1984 } else if ((data[4] & 0xF1) == 0x21) {
1986 if ((data[6] & 0x01) != 0x01 ||
1987 (data[8] & 0x01) != 0x01 ||
1988 (data[9] & 0x80) != 0x80 || (data[11] & 0x01) != 0x01)
1991 /* Check the following header bytes, if we can */
1992 if ((12 + 4) <= len) {
1993 if (!IS_MPEG_HEADER (data + 12))
2005 mpeg_sys_is_valid_pes (GstTypeFind * tf, const guint8 * data, guint len,
2008 guint pes_packet_len;
2010 /* Check the PES header at the given position, assuming the header code itself
2011 * was already checked */
2015 /* For MPEG Program streams, unbounded PES is not allowed, so we must have a
2016 * valid length present */
2017 pes_packet_len = GST_READ_UINT16_BE (data + 4);
2018 if (pes_packet_len == 0)
2021 /* Check the following header, if we can */
2022 if (6 + pes_packet_len + 4 <= len) {
2023 if (!IS_MPEG_HEADER (data + 6 + pes_packet_len))
2028 *pack_size = 6 + pes_packet_len;
2033 mpeg_sys_is_valid_sys (GstTypeFind * tf, const guint8 * data, guint len,
2038 /* Check the System header at the given position, assuming the header code itself
2039 * was already checked */
2042 sys_hdr_len = GST_READ_UINT16_BE (data + 4);
2043 if (sys_hdr_len < 6)
2046 /* Check the following header, if we can */
2047 if (6 + sys_hdr_len + 4 <= len) {
2048 if (!IS_MPEG_HEADER (data + 6 + sys_hdr_len))
2053 *pack_size = 6 + sys_hdr_len;
2058 /* calculation of possibility to identify random data as mpeg systemstream:
2059 * bits that must match in header detection: 32 (or more)
2060 * chance that random data is identifed: 1/2^32
2061 * chance that MPEG2_MIN_PACK_HEADERS headers are identified:
2062 * 1/2^(32*MPEG2_MIN_PACK_HEADERS)
2063 * chance that this happens in MPEG2_MAX_PROBE_LENGTH bytes:
2064 * 1-(1+1/2^(32*MPEG2_MIN_PACK_HEADERS)^MPEG2_MAX_PROBE_LENGTH)
2065 * for current values:
2066 * 1-(1+1/2^(32*4)^101024)
2068 * Since we also check marker bits and pes packet lengths, this probability is a
2069 * very coarse upper bound.
2072 mpeg_sys_type_find (GstTypeFind * tf, gpointer unused)
2074 const guint8 *data, *data0, *first_sync, *end;
2075 gint mpegversion = 0;
2076 guint pack_headers = 0;
2077 guint pes_headers = 0;
2079 guint since_last_sync = 0;
2080 guint32 sync_word = 0xffffffff;
2085 len = MPEG2_MAX_PROBE_LENGTH;
2088 data = gst_type_find_peek (tf, 0, 5 + len);
2089 } while (data == NULL && len >= 32);
2101 while (data < end) {
2103 if (sync_word == 0x00000100) {
2104 /* Found potential sync word */
2105 if (first_sync == NULL)
2106 first_sync = data - 3;
2108 if (since_last_sync > 4) {
2109 /* If more than 4 bytes since the last sync word, reset our counters,
2110 * as we're only interested in counting contiguous packets */
2111 pes_headers = pack_headers = 0;
2115 if (IS_MPEG_PACK_CODE (data[0])) {
2116 if ((data[1] & 0xC0) == 0x40) {
2119 } else if ((data[1] & 0xF0) == 0x20) {
2122 if (mpegversion != 0 &&
2123 mpeg_sys_is_valid_pack (tf, data - 3, end - data + 3, &pack_size)) {
2126 } else if (IS_MPEG_PES_CODE (data[0])) {
2128 if (mpeg_sys_is_valid_pes (tf, data - 3, end - data + 3, &pack_size)) {
2130 if (mpegversion == 0)
2133 } else if (IS_MPEG_SYS_CODE (data[0])) {
2134 if (mpeg_sys_is_valid_sys (tf, data - 3, end - data + 3, &pack_size)) {
2139 /* If we found a packet with a known size, skip the bytes in it and loop
2140 * around to check the next packet. */
2141 if (pack_size != 0) {
2142 data += pack_size - 3;
2143 sync_word = 0xffffffff;
2144 since_last_sync = 0;
2149 sync_word |= data[0];
2153 /* If we have found MAX headers, and *some* were pes headers (pack headers
2154 * are optional in an mpeg system stream) then return our high-probability
2156 if (pes_headers > 0 && (pack_headers + pes_headers) > MPEG2_MAX_SYS_HEADERS)
2160 /* If we at least saw MIN headers, and *some* were pes headers (pack headers
2161 * are optional in an mpeg system stream) then return a lower-probability
2163 if (pes_headers > 0 && (pack_headers + pes_headers) > MPEG2_MIN_SYS_HEADERS)
2171 prob = GST_TYPE_FIND_POSSIBLE + (10 * (pack_headers + pes_headers));
2172 prob = MIN (prob, GST_TYPE_FIND_MAXIMUM);
2174 /* lower probability if the first packet wasn't right at the start */
2175 if (data0 != first_sync && prob >= 10)
2178 GST_LOG ("Suggesting MPEG %d system stream, %d packs, %d pes, prob %u%%\n",
2179 mpegversion, pack_headers, pes_headers, prob);
2181 gst_type_find_suggest_simple (tf, prob, "video/mpeg",
2182 "systemstream", G_TYPE_BOOLEAN, TRUE,
2183 "mpegversion", G_TYPE_INT, mpegversion, NULL);
2187 /*** video/mpegts Transport Stream ***/
2188 static GstStaticCaps mpegts_caps = GST_STATIC_CAPS ("video/mpegts, "
2189 "systemstream = (boolean) true, packetsize = (int) [ 188, 208 ]");
2190 #define MPEGTS_CAPS gst_static_caps_get(&mpegts_caps)
2192 #define GST_MPEGTS_TYPEFIND_MIN_HEADERS 4
2193 #define GST_MPEGTS_TYPEFIND_MAX_HEADERS 10
2194 #define GST_MPEGTS_MAX_PACKET_SIZE 208
2195 #define GST_MPEGTS_TYPEFIND_SYNC_SIZE \
2196 (GST_MPEGTS_TYPEFIND_MIN_HEADERS * GST_MPEGTS_MAX_PACKET_SIZE)
2197 #define GST_MPEGTS_TYPEFIND_MAX_SYNC \
2198 (GST_MPEGTS_TYPEFIND_MAX_HEADERS * GST_MPEGTS_MAX_PACKET_SIZE)
2199 #define GST_MPEGTS_TYPEFIND_SCAN_LENGTH \
2200 (GST_MPEGTS_TYPEFIND_MAX_SYNC * 4)
2202 #define MPEGTS_HDR_SIZE 4
2203 /* Check for sync byte, error_indicator == 0 and packet has payload */
2204 #define IS_MPEGTS_HEADER(data) (((data)[0] == 0x47) && \
2205 (((data)[1] & 0x80) == 0x00) && \
2206 (((data)[3] & 0x30) != 0x00))
2208 /* Helper function to search ahead at intervals of packet_size for mpegts
2211 mpeg_ts_probe_headers (GstTypeFind * tf, guint64 offset, gint packet_size)
2213 /* We always enter this function having found at least one header already */
2215 const guint8 *data = NULL;
2217 GST_LOG ("looking for mpeg-ts packets of size %u", packet_size);
2218 while (found < GST_MPEGTS_TYPEFIND_MAX_HEADERS) {
2219 offset += packet_size;
2221 data = gst_type_find_peek (tf, offset, MPEGTS_HDR_SIZE);
2222 if (data == NULL || !IS_MPEGTS_HEADER (data))
2226 GST_LOG ("mpeg-ts sync #%2d at offset %" G_GUINT64_FORMAT, found, offset);
2232 /* Try and detect at least 4 packets in at most 10 packets worth of
2233 * data. Need to try several possible packet sizes */
2235 mpeg_ts_type_find (GstTypeFind * tf, gpointer unused)
2237 /* TS packet sizes to test: normal, DVHS packet size and
2238 * FEC with 16 or 20 byte codes packet size. */
2239 const gint pack_sizes[] = { 188, 192, 204, 208 };
2240 const guint8 *data = NULL;
2242 guint64 skipped = 0;
2244 while (skipped < GST_MPEGTS_TYPEFIND_SCAN_LENGTH) {
2245 if (size < MPEGTS_HDR_SIZE) {
2246 data = gst_type_find_peek (tf, skipped, GST_MPEGTS_TYPEFIND_SYNC_SIZE);
2249 size = GST_MPEGTS_TYPEFIND_SYNC_SIZE;
2252 /* Have at least MPEGTS_HDR_SIZE bytes at this point */
2253 if (IS_MPEGTS_HEADER (data)) {
2256 GST_LOG ("possible mpeg-ts sync at offset %" G_GUINT64_FORMAT, skipped);
2258 for (p = 0; p < G_N_ELEMENTS (pack_sizes); p++) {
2261 /* Probe ahead at size pack_sizes[p] */
2262 found = mpeg_ts_probe_headers (tf, skipped, pack_sizes[p]);
2263 if (found >= GST_MPEGTS_TYPEFIND_MIN_HEADERS) {
2266 /* found at least 4 headers. 10 headers = MAXIMUM probability.
2267 * Arbitrarily, I assigned 10% probability for each header we
2268 * found, 40% -> 100% */
2269 probability = MIN (10 * found, GST_TYPE_FIND_MAXIMUM);
2271 gst_type_find_suggest_simple (tf, probability, "video/mpegts",
2272 "systemstream", G_TYPE_BOOLEAN, TRUE,
2273 "packetsize", G_TYPE_INT, pack_sizes[p], NULL);
2284 #define GST_MPEGVID_TYPEFIND_TRY_PICTURES 6
2285 #define GST_MPEGVID_TYPEFIND_TRY_SYNC (100 * 1024) /* 100 kB */
2287 /* Scan ahead a maximum of max_extra_offset bytes until the next IS_MPEG_HEADER
2288 * offset. After the call, offset will be after the 0x000001, i.e. at the 4th
2289 * byte of the MPEG header. Returns TRUE if a header was found, FALSE if not.
2292 mpeg_find_next_header (GstTypeFind * tf, DataScanCtx * c,
2293 guint64 max_extra_offset)
2295 guint64 extra_offset;
2297 for (extra_offset = 0; extra_offset <= max_extra_offset; ++extra_offset) {
2298 if (!data_scan_ctx_ensure_data (tf, c, 4))
2300 if (IS_MPEG_HEADER (c->data)) {
2301 data_scan_ctx_advance (tf, c, 3);
2304 data_scan_ctx_advance (tf, c, 1);
2309 /*** video/mpeg MPEG-4 elementary video stream ***/
2311 static GstStaticCaps mpeg4_video_caps = GST_STATIC_CAPS ("video/mpeg, "
2312 "systemstream=(boolean)false, mpegversion=4, parsed=(boolean)false");
2313 #define MPEG4_VIDEO_CAPS gst_static_caps_get(&mpeg4_video_caps)
2316 * This typefind is based on the elementary video header defined in
2317 * http://xhelmboyx.tripod.com/formats/mpeg-layout.txt
2318 * In addition, it allows the visual object sequence header to be
2319 * absent, and even the VOS header to be absent. In the latter case,
2320 * a number of VOPs have to be present.
2323 mpeg4_video_type_find (GstTypeFind * tf, gpointer unused)
2325 DataScanCtx c = { 0, NULL, 0 };
2326 gboolean seen_vios_at_0 = FALSE;
2327 gboolean seen_vios = FALSE;
2328 gboolean seen_vos = FALSE;
2329 gboolean seen_vol = FALSE;
2330 guint num_vop_headers = 0;
2333 while (c.offset < GST_MPEGVID_TYPEFIND_TRY_SYNC) {
2334 if (num_vop_headers >= GST_MPEGVID_TYPEFIND_TRY_PICTURES)
2337 if (!mpeg_find_next_header (tf, &c,
2338 GST_MPEGVID_TYPEFIND_TRY_SYNC - c.offset))
2343 /* visual_object_sequence_start_code */
2346 break; /* Terminate at second vios */
2348 seen_vios_at_0 = TRUE;
2350 data_scan_ctx_advance (tf, &c, 2);
2351 if (!mpeg_find_next_header (tf, &c, 0))
2356 /* Optional metadata */
2358 if (!mpeg_find_next_header (tf, &c, 24))
2362 /* visual_object_start_code (consider it optional) */
2364 data_scan_ctx_advance (tf, &c, 2);
2365 /* may contain ID marker and YUV clamping */
2366 if (!mpeg_find_next_header (tf, &c, 7))
2372 /* video_object_start_code */
2375 break; /* Terminate at second vos */
2377 data_scan_ctx_advance (tf, &c, 2);
2381 /* video_object_layer_start_code */
2382 if (sc >= 0x20 && sc <= 0x2F) {
2384 data_scan_ctx_advance (tf, &c, 5);
2388 /* video_object_plane_start_code */
2391 data_scan_ctx_advance (tf, &c, 2);
2395 /* Unknown start code. */
2398 if (num_vop_headers > 0 || seen_vol) {
2399 GstTypeFindProbability probability = 0;
2401 GST_LOG ("Found %d pictures, vios: %d, vos:%d, vol:%d", num_vop_headers,
2402 seen_vios, seen_vos, seen_vol);
2404 if (num_vop_headers >= GST_MPEGVID_TYPEFIND_TRY_PICTURES && seen_vios_at_0
2405 && seen_vos && seen_vol)
2406 probability = GST_TYPE_FIND_MAXIMUM - 1;
2407 else if (num_vop_headers >= GST_MPEGVID_TYPEFIND_TRY_PICTURES && seen_vios
2408 && seen_vos && seen_vol)
2409 probability = GST_TYPE_FIND_NEARLY_CERTAIN - 1;
2410 else if (seen_vios_at_0 && seen_vos && seen_vol)
2411 probability = GST_TYPE_FIND_NEARLY_CERTAIN - 6;
2412 else if (num_vop_headers >= GST_MPEGVID_TYPEFIND_TRY_PICTURES && seen_vos
2414 probability = GST_TYPE_FIND_NEARLY_CERTAIN - 6;
2415 else if (num_vop_headers >= GST_MPEGVID_TYPEFIND_TRY_PICTURES && seen_vol)
2416 probability = GST_TYPE_FIND_NEARLY_CERTAIN - 9;
2417 else if (num_vop_headers >= GST_MPEGVID_TYPEFIND_TRY_PICTURES)
2418 probability = GST_TYPE_FIND_LIKELY - 1;
2419 else if (num_vop_headers > 2 && seen_vios && seen_vos && seen_vol)
2420 probability = GST_TYPE_FIND_LIKELY - 9;
2421 else if (seen_vios && seen_vos && seen_vol)
2422 probability = GST_TYPE_FIND_LIKELY - 20;
2423 else if (num_vop_headers > 0 && seen_vos && seen_vol)
2424 probability = GST_TYPE_FIND_POSSIBLE;
2425 else if (num_vop_headers > 0)
2426 probability = GST_TYPE_FIND_POSSIBLE - 10;
2427 else if (seen_vos && seen_vol)
2428 probability = GST_TYPE_FIND_POSSIBLE - 20;
2430 gst_type_find_suggest (tf, probability, MPEG4_VIDEO_CAPS);
2434 /*** video/x-h263 H263 video stream ***/
2435 static GstStaticCaps h263_video_caps = GST_STATIC_CAPS ("video/x-h263");
2437 #define H263_VIDEO_CAPS gst_static_caps_get(&h263_video_caps)
2439 #define H263_MAX_PROBE_LENGTH (128 * 1024)
2442 h263_video_type_find (GstTypeFind * tf, gpointer unused)
2444 DataScanCtx c = { 0, NULL, 0 };
2452 while (c.offset < H263_MAX_PROBE_LENGTH) {
2453 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 4)))
2456 /* Find the picture start code */
2457 data = (data << 8) + c.data[0];
2458 psc = data & G_GUINT64_CONSTANT (0xfffffc0000);
2459 if (psc == 0x800000) {
2462 tr = (data & 0x3fc) >> 2;
2466 /* Now that we have a Valid PSC, check if we also have a valid PTYPE and
2467 the Source Format, which should range between 1 and 5 */
2468 if (((tr >> 6) == 0x2) && (format > 0 && format < 6))
2473 /* FIXME: maybe bail out early if we get mostly bad syncs ? */
2476 data_scan_ctx_advance (tf, &c, 1);
2479 if (good > 0 && bad == 0)
2480 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY, H263_VIDEO_CAPS);
2481 else if (good > 2 * bad)
2482 gst_type_find_suggest (tf, GST_TYPE_FIND_POSSIBLE, H263_VIDEO_CAPS);
2487 /*** video/x-h264 H264 elementary video stream ***/
2489 static GstStaticCaps h264_video_caps =
2490 GST_STATIC_CAPS ("video/x-h264,stream-format=byte-stream");
2492 #define H264_VIDEO_CAPS gst_static_caps_get(&h264_video_caps)
2494 #define H264_MAX_PROBE_LENGTH (128 * 1024) /* 128kB for HD should be enough. */
2497 h264_video_type_find (GstTypeFind * tf, gpointer unused)
2499 DataScanCtx c = { 0, NULL, 0 };
2501 /* Stream consists of: a series of sync codes (00 00 00 01) followed
2504 gboolean seen_idr = FALSE;
2505 gboolean seen_sps = FALSE;
2506 gboolean seen_pps = FALSE;
2511 while (c.offset < H264_MAX_PROBE_LENGTH) {
2512 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 4)))
2515 if (IS_MPEG_HEADER (c.data)) {
2516 nut = c.data[3] & 0x9f; /* forbiden_zero_bit | nal_unit_type */
2517 ref = c.data[3] & 0x60; /* nal_ref_idc */
2519 /* if forbidden bit is different to 0 won't be h264 */
2525 /* collect statistics about the NAL types */
2526 if ((nut >= 1 && nut <= 13) || nut == 19) {
2527 if ((nut == 5 && ref == 0) ||
2528 ((nut == 6 || (nut >= 9 && nut <= 12)) && ref != 0)) {
2540 } else if (nut >= 14 && nut <= 33) {
2542 /* Theoretically these are good, since if they exist in the
2543 stream it merely means that a newer backwards-compatible
2544 h.264 stream. But we should be identifying that separately. */
2547 /* unspecified, application specific */
2548 /* don't consider these bad */
2551 GST_LOG ("good:%d, bad:%d, pps:%d, sps:%d, idr:%d", good, bad, seen_pps,
2552 seen_sps, seen_idr);
2554 if (seen_sps && seen_pps && seen_idr && good >= 10 && bad < 4) {
2555 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY, H264_VIDEO_CAPS);
2559 data_scan_ctx_advance (tf, &c, 4);
2561 data_scan_ctx_advance (tf, &c, 1);
2564 GST_LOG ("good:%d, bad:%d, pps:%d, sps:%d, idr:%d", good, bad, seen_pps,
2565 seen_sps, seen_idr);
2567 if (good >= 2 && bad == 0) {
2568 gst_type_find_suggest (tf, GST_TYPE_FIND_POSSIBLE, H264_VIDEO_CAPS);
2572 /*** video/mpeg video stream ***/
2574 static GstStaticCaps mpeg_video_caps = GST_STATIC_CAPS ("video/mpeg, "
2575 "systemstream = (boolean) false");
2576 #define MPEG_VIDEO_CAPS gst_static_caps_get(&mpeg_video_caps)
2579 * Idea is the same as MPEG system stream typefinding: We check each
2580 * byte of the stream to see if - from that point on - the stream
2581 * matches a predefined set of marker bits as defined in the MPEG
2584 * I'm sure someone will do a chance calculation here too.
2588 mpeg_video_stream_type_find (GstTypeFind * tf, gpointer unused)
2590 DataScanCtx c = { 0, NULL, 0 };
2591 gboolean seen_seq_at_0 = FALSE;
2592 gboolean seen_seq = FALSE;
2593 gboolean seen_gop = FALSE;
2594 guint64 last_pic_offset = 0;
2595 guint num_pic_headers = 0;
2598 while (c.offset < GST_MPEGVID_TYPEFIND_TRY_SYNC) {
2599 if (found >= GST_MPEGVID_TYPEFIND_TRY_PICTURES)
2602 if (!data_scan_ctx_ensure_data (tf, &c, 5))
2605 if (!IS_MPEG_HEADER (c.data))
2608 /* a pack header indicates that this isn't an elementary stream */
2609 if (c.data[3] == 0xBA && mpeg_sys_is_valid_pack (tf, c.data, c.size, NULL))
2612 /* do we have a sequence header? */
2613 if (c.data[3] == 0xB3) {
2614 seen_seq_at_0 = seen_seq_at_0 || (c.offset == 0);
2616 data_scan_ctx_advance (tf, &c, 4 + 8);
2620 /* or a GOP header */
2621 if (c.data[3] == 0xB8) {
2623 data_scan_ctx_advance (tf, &c, 8);
2627 /* but what we'd really like to see is a picture header */
2628 if (c.data[3] == 0x00) {
2630 last_pic_offset = c.offset;
2631 data_scan_ctx_advance (tf, &c, 8);
2635 /* ... each followed by a slice header with slice_vertical_pos=1 that's
2636 * not too far away from the previously seen picture header. */
2637 if (c.data[3] == 0x01 && num_pic_headers > found &&
2638 (c.offset - last_pic_offset) >= 4 &&
2639 (c.offset - last_pic_offset) <= 64) {
2640 data_scan_ctx_advance (tf, &c, 4);
2647 data_scan_ctx_advance (tf, &c, 1);
2650 if (found > 0 || seen_seq) {
2651 GstTypeFindProbability probability = 0;
2653 GST_LOG ("Found %d pictures, seq:%d, gop:%d", found, seen_seq, seen_gop);
2655 if (found >= GST_MPEGVID_TYPEFIND_TRY_PICTURES && seen_seq && seen_gop)
2656 probability = GST_TYPE_FIND_NEARLY_CERTAIN - 1;
2657 else if (found >= GST_MPEGVID_TYPEFIND_TRY_PICTURES && seen_seq)
2658 probability = GST_TYPE_FIND_NEARLY_CERTAIN - 9;
2659 else if (found >= GST_MPEGVID_TYPEFIND_TRY_PICTURES)
2660 probability = GST_TYPE_FIND_LIKELY;
2661 else if (seen_seq_at_0 && seen_gop && found > 2)
2662 probability = GST_TYPE_FIND_LIKELY - 10;
2663 else if (seen_seq && seen_gop && found > 2)
2664 probability = GST_TYPE_FIND_LIKELY - 20;
2665 else if (seen_seq_at_0 && found > 0)
2666 probability = GST_TYPE_FIND_POSSIBLE;
2667 else if (seen_seq && found > 0)
2668 probability = GST_TYPE_FIND_POSSIBLE - 5;
2670 probability = GST_TYPE_FIND_POSSIBLE - 10;
2672 probability = GST_TYPE_FIND_POSSIBLE - 20;
2674 gst_type_find_suggest_simple (tf, probability, "video/mpeg",
2675 "systemstream", G_TYPE_BOOLEAN, FALSE,
2676 "mpegversion", G_TYPE_INT, 1, NULL);
2680 /*** audio/x-aiff ***/
2682 static GstStaticCaps aiff_caps = GST_STATIC_CAPS ("audio/x-aiff");
2684 #define AIFF_CAPS gst_static_caps_get(&aiff_caps)
2686 aiff_type_find (GstTypeFind * tf, gpointer unused)
2688 const guint8 *data = gst_type_find_peek (tf, 0, 4);
2690 if (data && memcmp (data, "FORM", 4) == 0) {
2692 if (memcmp (data, "AIFF", 4) == 0 || memcmp (data, "AIFC", 4) == 0)
2693 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, AIFF_CAPS);
2697 /*** audio/x-svx ***/
2699 static GstStaticCaps svx_caps = GST_STATIC_CAPS ("audio/x-svx");
2701 #define SVX_CAPS gst_static_caps_get(&svx_caps)
2703 svx_type_find (GstTypeFind * tf, gpointer unused)
2705 const guint8 *data = gst_type_find_peek (tf, 0, 4);
2707 if (data && memcmp (data, "FORM", 4) == 0) {
2709 if (memcmp (data, "8SVX", 4) == 0 || memcmp (data, "16SV", 4) == 0)
2710 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, SVX_CAPS);
2714 /*** audio/x-shorten ***/
2716 static GstStaticCaps shn_caps = GST_STATIC_CAPS ("audio/x-shorten");
2718 #define SHN_CAPS gst_static_caps_get(&shn_caps)
2720 shn_type_find (GstTypeFind * tf, gpointer unused)
2722 const guint8 *data = gst_type_find_peek (tf, 0, 4);
2724 if (data && memcmp (data, "ajkg", 4) == 0) {
2725 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, SHN_CAPS);
2727 data = gst_type_find_peek (tf, -8, 8);
2728 if (data && memcmp (data, "SHNAMPSK", 8) == 0) {
2729 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, SHN_CAPS);
2733 /*** application/x-ape ***/
2735 static GstStaticCaps ape_caps = GST_STATIC_CAPS ("application/x-ape");
2737 #define APE_CAPS gst_static_caps_get(&ape_caps)
2739 ape_type_find (GstTypeFind * tf, gpointer unused)
2741 const guint8 *data = gst_type_find_peek (tf, 0, 4);
2743 if (data && memcmp (data, "MAC ", 4) == 0) {
2744 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY + 10, APE_CAPS);
2748 /*** ISO FORMATS ***/
2750 /*** audio/x-m4a ***/
2752 static GstStaticCaps m4a_caps = GST_STATIC_CAPS ("audio/x-m4a");
2754 #define M4A_CAPS (gst_static_caps_get(&m4a_caps))
2756 m4a_type_find (GstTypeFind * tf, gpointer unused)
2758 const guint8 *data = gst_type_find_peek (tf, 4, 8);
2760 if (data && (memcmp (data, "ftypM4A ", 8) == 0)) {
2761 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, M4A_CAPS);
2765 /*** application/x-3gp ***/
2767 /* The Q is there because variables can't start with a number. */
2768 static GstStaticCaps q3gp_caps = GST_STATIC_CAPS ("application/x-3gp");
2769 #define Q3GP_CAPS (gst_static_caps_get(&q3gp_caps))
2771 static const gchar *
2772 q3gp_type_find_get_profile (const guint8 * data)
2774 switch (GST_MAKE_FOURCC (data[0], data[1], data[2], 0)) {
2775 case GST_MAKE_FOURCC ('3', 'g', 'g', 0):
2777 case GST_MAKE_FOURCC ('3', 'g', 'p', 0):
2779 case GST_MAKE_FOURCC ('3', 'g', 's', 0):
2780 return "streaming-server";
2781 case GST_MAKE_FOURCC ('3', 'g', 'r', 0):
2782 return "progressive-download";
2790 q3gp_type_find (GstTypeFind * tf, gpointer unused)
2792 const gchar *profile;
2793 guint32 ftyp_size = 0;
2795 const guint8 *data = NULL;
2797 if ((data = gst_type_find_peek (tf, 0, 12)) == NULL) {
2802 if (memcmp (data, "ftyp", 4) != 0) {
2806 /* check major brand */
2808 if ((profile = q3gp_type_find_get_profile (data))) {
2809 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_MAXIMUM,
2810 "application/x-3gp", "profile", G_TYPE_STRING, profile, NULL);
2814 /* check compatible brands */
2815 if ((data = gst_type_find_peek (tf, 0, 4)) != NULL) {
2816 ftyp_size = GST_READ_UINT32_BE (data);
2818 for (offset = 16; offset < ftyp_size; offset += 4) {
2819 if ((data = gst_type_find_peek (tf, offset, 3)) == NULL) {
2822 if ((profile = q3gp_type_find_get_profile (data))) {
2823 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_MAXIMUM,
2824 "application/x-3gp", "profile", G_TYPE_STRING, profile, NULL);
2833 /*** video/mj2 and image/jp2 ***/
2834 static GstStaticCaps mj2_caps = GST_STATIC_CAPS ("video/mj2");
2836 #define MJ2_CAPS gst_static_caps_get(&mj2_caps)
2838 static GstStaticCaps jp2_caps = GST_STATIC_CAPS ("image/jp2");
2840 #define JP2_CAPS gst_static_caps_get(&jp2_caps)
2843 jp2_type_find (GstTypeFind * tf, gpointer unused)
2847 data = gst_type_find_peek (tf, 0, 24);
2852 if (memcmp (data, "\000\000\000\014jP \015\012\207\012", 12) != 0)
2855 /* check ftyp box */
2857 if (memcmp (data + 4, "ftyp", 4) == 0) {
2858 if (memcmp (data + 8, "jp2 ", 4) == 0)
2859 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, JP2_CAPS);
2860 else if (memcmp (data + 8, "mjp2", 4) == 0)
2861 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MJ2_CAPS);
2865 /*** video/quicktime ***/
2867 static GstStaticCaps qt_caps = GST_STATIC_CAPS ("video/quicktime");
2869 #define QT_CAPS gst_static_caps_get(&qt_caps)
2870 #define STRNCMP(x,y,z) (strncmp ((char*)(x), (char*)(y), z))
2872 /* FIXME 0.11: go through http://www.ftyps.com/ */
2874 qt_type_find (GstTypeFind * tf, gpointer unused)
2880 const gchar *variant = NULL;
2882 while ((data = gst_type_find_peek (tf, offset, 12)) != NULL) {
2885 if (STRNCMP (&data[4], "ftypqt ", 8) == 0) {
2886 tip = GST_TYPE_FIND_MAXIMUM;
2890 if (STRNCMP (&data[4], "ftypisom", 8) == 0 ||
2891 STRNCMP (&data[4], "ftypavc1", 8) == 0 ||
2892 STRNCMP (&data[4], "ftypmp42", 8) == 0) {
2893 tip = GST_TYPE_FIND_MAXIMUM;
2898 if (STRNCMP (&data[4], "ftypisml", 8) == 0) {
2899 tip = GST_TYPE_FIND_MAXIMUM;
2900 variant = "iso-fragmented";
2904 /* box/atom types that are in common with ISO base media file format */
2905 if (STRNCMP (&data[4], "moov", 4) == 0 ||
2906 STRNCMP (&data[4], "mdat", 4) == 0 ||
2907 STRNCMP (&data[4], "ftyp", 4) == 0 ||
2908 STRNCMP (&data[4], "free", 4) == 0 ||
2909 STRNCMP (&data[4], "uuid", 4) == 0 ||
2910 STRNCMP (&data[4], "skip", 4) == 0) {
2912 tip = GST_TYPE_FIND_LIKELY;
2914 tip = GST_TYPE_FIND_NEARLY_CERTAIN;
2917 /* other box/atom types, apparently quicktime specific */
2918 else if (STRNCMP (&data[4], "pnot", 4) == 0 ||
2919 STRNCMP (&data[4], "PICT", 4) == 0 ||
2920 STRNCMP (&data[4], "wide", 4) == 0 ||
2921 STRNCMP (&data[4], "prfl", 4) == 0) {
2922 tip = GST_TYPE_FIND_MAXIMUM;
2929 size = GST_READ_UINT32_BE (data);
2930 /* check compatible brands rather than ever expaning major brands above */
2931 if ((STRNCMP (&data[4], "ftyp", 4) == 0) && (size >= 16)) {
2932 new_offset = offset + 12;
2933 while (new_offset + 4 <= offset + size) {
2934 data = gst_type_find_peek (tf, new_offset, 4);
2937 if (STRNCMP (&data[4], "isom", 4) == 0 ||
2938 STRNCMP (&data[4], "avc1", 4) == 0 ||
2939 STRNCMP (&data[4], "mp41", 4) == 0 ||
2940 STRNCMP (&data[4], "mp42", 4) == 0) {
2941 tip = GST_TYPE_FIND_MAXIMUM;
2949 const guint8 *sizedata;
2951 sizedata = gst_type_find_peek (tf, offset + 8, 8);
2952 if (sizedata == NULL)
2955 size = GST_READ_UINT64_BE (sizedata);
2960 new_offset = offset + size;
2961 if (new_offset <= offset)
2963 offset = new_offset;
2969 GstCaps *caps = gst_caps_copy (QT_CAPS);
2971 gst_caps_set_simple (caps, "variant", G_TYPE_STRING, variant, NULL);
2972 gst_type_find_suggest (tf, tip, caps);
2973 gst_caps_unref (caps);
2975 gst_type_find_suggest (tf, tip, QT_CAPS);
2981 /*** image/x-quicktime ***/
2983 static GstStaticCaps qtif_caps = GST_STATIC_CAPS ("image/x-quicktime");
2985 #define QTIF_CAPS gst_static_caps_get(&qtif_caps)
2987 /* how many atoms we check before we give up */
2988 #define QTIF_MAXROUNDS 25
2991 qtif_type_find (GstTypeFind * tf, gpointer unused)
2994 gboolean found_idsc = FALSE;
2995 gboolean found_idat = FALSE;
2999 while ((data = gst_type_find_peek (tf, offset, 8)) != NULL) {
3002 size = GST_READ_UINT32_BE (data);
3004 const guint8 *sizedata;
3006 sizedata = gst_type_find_peek (tf, offset + 8, 8);
3007 if (sizedata == NULL)
3010 size = GST_READ_UINT64_BE (sizedata);
3015 if (STRNCMP (data + 4, "idsc", 4) == 0)
3017 if (STRNCMP (data + 4, "idat", 4) == 0)
3020 if (found_idsc && found_idat) {
3021 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, QTIF_CAPS);
3026 if (++rounds > QTIF_MAXROUNDS)
3030 if (found_idsc || found_idat) {
3031 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY, QTIF_CAPS);
3036 /*** audio/x-mod ***/
3038 static GstStaticCaps mod_caps = GST_STATIC_CAPS ("audio/x-mod");
3040 #define MOD_CAPS gst_static_caps_get(&mod_caps)
3041 /* FIXME: M15 CheckType to do */
3043 mod_type_find (GstTypeFind * tf, gpointer unused)
3048 if ((data = gst_type_find_peek (tf, 1080, 4)) != NULL) {
3049 /* Protracker and variants */
3050 if ((memcmp (data, "M.K.", 4) == 0) || (memcmp (data, "M!K!", 4) == 0) ||
3052 (memcmp (data, "FLT", 3) == 0 && isdigit (data[3])) ||
3053 (memcmp (data, "EXO", 3) == 0 && isdigit (data[3])) ||
3054 /* Oktalyzer (Amiga) */
3055 (memcmp (data, "OKTA", 4) == 0) ||
3056 /* Oktalyser (Atari) */
3057 (memcmp (data, "CD81", 4) == 0) ||
3059 (memcmp (data + 1, "CHN", 3) == 0 && isdigit (data[0])) ||
3060 /* Fasttracker or Taketracker */
3061 (memcmp (data + 2, "CH", 2) == 0 && isdigit (data[0])
3062 && isdigit (data[1])) || (memcmp (data + 2, "CN", 2) == 0
3063 && isdigit (data[0]) && isdigit (data[1]))) {
3064 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MOD_CAPS);
3069 if ((data = gst_type_find_peek (tf, 0, 38)) != NULL) {
3070 if (memcmp (data, "Extended Module: ", 17) == 0 && data[37] == 0x1A) {
3071 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MOD_CAPS);
3076 if (data || (data = gst_type_find_peek (tf, 0, 8)) != NULL) {
3077 if (memcmp (data, "OKTASONG", 8) == 0) {
3078 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MOD_CAPS);
3082 if (data || (data = gst_type_find_peek (tf, 0, 4)) != NULL) {
3084 if ((memcmp (data, "if", 2) == 0) || (memcmp (data, "JN", 2) == 0)) {
3085 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY, MOD_CAPS);
3089 if ((memcmp (data, "AMF", 3) == 0 && data[3] > 10 && data[3] < 14) ||
3091 (memcmp (data, "IMPM", 4) == 0) ||
3093 (memcmp (data, "MMD0", 4) == 0) || (memcmp (data, "MMD1", 4) == 0) ||
3095 (memcmp (data, "MTM", 3) == 0)) {
3096 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MOD_CAPS);
3100 if (memcmp (data, "RIFF", 4) == 0) {
3101 const guint8 *data2 = gst_type_find_peek (tf, 8, 4);
3104 if (memcmp (data2, "DSMF", 4) == 0) {
3105 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MOD_CAPS);
3111 if (memcmp (data, "FAM\xFE", 4) == 0) {
3112 const guint8 *data2 = gst_type_find_peek (tf, 44, 3);
3115 if (memcmp (data2, "compare", 3) == 0) {
3116 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MOD_CAPS);
3120 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY, MOD_CAPS);
3125 if (memcmp (data, "GDM\xFE", 4) == 0) {
3126 const guint8 *data2 = gst_type_find_peek (tf, 71, 4);
3129 if (memcmp (data2, "GMFS", 4) == 0) {
3130 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MOD_CAPS);
3134 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY, MOD_CAPS);
3140 if ((data = gst_type_find_peek (tf, 60, 4)) != NULL) {
3141 if (memcmp (data, "IM10", 4) == 0) {
3142 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MOD_CAPS);
3147 if ((data = gst_type_find_peek (tf, 44, 4)) != NULL) {
3148 if (memcmp (data, "SCRM", 4) == 0) {
3149 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MOD_CAPS);
3154 if ((data = gst_type_find_peek (tf, 20, 8)) != NULL) {
3155 if (g_ascii_strncasecmp ((gchar *) data, "!Scream!", 8) == 0 ||
3156 g_ascii_strncasecmp ((gchar *) data, "BMOD2STM", 8) == 0) {
3157 const guint8 *id, *stmtype;
3159 if ((id = gst_type_find_peek (tf, 28, 1)) == NULL)
3161 if ((stmtype = gst_type_find_peek (tf, 29, 1)) == NULL)
3163 if (*id == 0x1A && *stmtype == 2)
3164 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MOD_CAPS);
3169 if ((data = gst_type_find_peek (tf, 0, 19)) != NULL) {
3170 if (memcmp (data, "ASYLUM Music Format", 19) == 0) {
3171 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MOD_CAPS);
3177 /*** application/x-shockwave-flash ***/
3179 static GstStaticCaps swf_caps =
3180 GST_STATIC_CAPS ("application/x-shockwave-flash");
3181 #define SWF_CAPS (gst_static_caps_get(&swf_caps))
3183 swf_type_find (GstTypeFind * tf, gpointer unused)
3185 const guint8 *data = gst_type_find_peek (tf, 0, 4);
3187 if (data && (data[0] == 'F' || data[0] == 'C') &&
3188 data[1] == 'W' && data[2] == 'S') {
3189 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, SWF_CAPS);
3193 /*** image/jpeg ***/
3195 #define JPEG_MARKER_IS_START_OF_FRAME(x) \
3196 ((x)>=0xc0 && (x) <= 0xcf && (x)!=0xc4 && (x)!=0xc8 && (x)!=0xcc)
3198 static GstStaticCaps jpeg_caps = GST_STATIC_CAPS ("image/jpeg");
3200 #define JPEG_CAPS (gst_static_caps_get(&jpeg_caps))
3202 jpeg_type_find (GstTypeFind * tf, gpointer unused)
3204 GstTypeFindProbability prob = GST_TYPE_FIND_POSSIBLE;
3205 DataScanCtx c = { 0, NULL, 0 };
3209 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 2)))
3212 if (c.data[0] != 0xff || c.data[1] != 0xd8)
3216 data_scan_ctx_advance (tf, &c, 2);
3218 caps = gst_caps_copy (JPEG_CAPS);
3220 while (data_scan_ctx_ensure_data (tf, &c, 4) && c.offset < (200 * 1024)) {
3224 if (c.data[0] != 0xff)
3228 if (G_UNLIKELY (marker == 0xff)) {
3229 data_scan_ctx_advance (tf, &c, 1);
3233 data_scan_ctx_advance (tf, &c, 2);
3235 /* we assume all markers we'll see before SOF have a payload length; if
3236 * that's not the case we'll just detect a false sync and bail out, but
3237 * still report POSSIBLE probability */
3238 len = GST_READ_UINT16_BE (c.data);
3240 GST_LOG ("possible JPEG marker 0x%02x (@0x%04x), segment length %u",
3241 marker, (guint) c.offset, len);
3243 if (!data_scan_ctx_ensure_data (tf, &c, len))
3246 if (marker == 0xc4 || /* DEFINE_HUFFMAN_TABLES */
3247 marker == 0xcc || /* DEFINE_ARITHMETIC_CONDITIONING */
3248 marker == 0xdb || /* DEFINE_QUANTIZATION_TABLES */
3249 marker == 0xdd || /* DEFINE_RESTART_INTERVAL */
3250 marker == 0xfe) { /* COMMENT */
3251 data_scan_ctx_advance (tf, &c, len);
3253 } else if (marker == 0xe0 && len >= (2 + 4) && /* APP0 */
3254 data_scan_ctx_memcmp (tf, &c, 2, "JFIF", 4)) {
3255 GST_LOG ("found JFIF tag");
3256 prob = GST_TYPE_FIND_MAXIMUM;
3257 data_scan_ctx_advance (tf, &c, len);
3259 /* we continue until we find a start of frame marker */
3260 } else if (marker == 0xe1 && len >= (2 + 4) && /* APP1 */
3261 data_scan_ctx_memcmp (tf, &c, 2, "Exif", 4)) {
3262 GST_LOG ("found Exif tag");
3263 prob = GST_TYPE_FIND_MAXIMUM;
3264 data_scan_ctx_advance (tf, &c, len);
3266 /* we continue until we find a start of frame marker */
3267 } else if (marker >= 0xe0 && marker <= 0xef) { /* APPn */
3268 data_scan_ctx_advance (tf, &c, len);
3270 } else if (JPEG_MARKER_IS_START_OF_FRAME (marker) && len >= (2 + 8)) {
3273 h = GST_READ_UINT16_BE (c.data + 2 + 1);
3274 w = GST_READ_UINT16_BE (c.data + 2 + 1 + 2);
3275 if (h == 0 || w == 0) {
3276 GST_WARNING ("bad width %u and/or height %u in SOF header", w, h);
3280 GST_LOG ("SOF at offset %" G_GUINT64_FORMAT ", num_markers=%d, "
3281 "WxH=%dx%d", c.offset - 2, num_markers, w, h);
3283 if (num_markers >= 5 || prob == GST_TYPE_FIND_MAXIMUM)
3284 prob = GST_TYPE_FIND_MAXIMUM;
3286 prob = GST_TYPE_FIND_LIKELY;
3288 gst_caps_set_simple (caps, "width", G_TYPE_INT, w,
3289 "height", G_TYPE_INT, h, "sof-marker", G_TYPE_INT, marker & 0xf,
3294 GST_WARNING ("bad length or unexpected JPEG marker 0xff 0x%02x", marker);
3299 gst_type_find_suggest (tf, prob, caps);
3300 gst_caps_unref (caps);
3305 static GstStaticCaps bmp_caps = GST_STATIC_CAPS ("image/bmp");
3307 #define BMP_CAPS (gst_static_caps_get(&bmp_caps))
3309 bmp_type_find (GstTypeFind * tf, gpointer unused)
3311 DataScanCtx c = { 0, NULL, 0 };
3312 guint32 struct_size, w, h, planes, bpp;
3314 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 54)))
3317 if (c.data[0] != 'B' || c.data[1] != 'M')
3320 /* skip marker + size */
3321 data_scan_ctx_advance (tf, &c, 2 + 4);
3323 /* reserved, must be 0 */
3324 if (c.data[0] != 0 || c.data[1] != 0 || c.data[2] != 0 || c.data[3] != 0)
3327 data_scan_ctx_advance (tf, &c, 2 + 2);
3329 /* offset to start of image data in bytes (check for sanity) */
3330 GST_LOG ("offset=%u", GST_READ_UINT32_LE (c.data));
3331 if (GST_READ_UINT32_LE (c.data) > (10 * 1024 * 1024))
3334 struct_size = GST_READ_UINT32_LE (c.data + 4);
3335 GST_LOG ("struct_size=%u", struct_size);
3337 data_scan_ctx_advance (tf, &c, 4 + 4);
3339 if (struct_size == 0x0C) {
3340 w = GST_READ_UINT16_LE (c.data);
3341 h = GST_READ_UINT16_LE (c.data + 2);
3342 planes = GST_READ_UINT16_LE (c.data + 2 + 2);
3343 bpp = GST_READ_UINT16_LE (c.data + 2 + 2 + 2);
3344 } else if (struct_size == 40 || struct_size == 64 || struct_size == 108
3345 || struct_size == 124 || struct_size == 0xF0) {
3346 w = GST_READ_UINT32_LE (c.data);
3347 h = GST_READ_UINT32_LE (c.data + 4);
3348 planes = GST_READ_UINT16_LE (c.data + 4 + 4);
3349 bpp = GST_READ_UINT16_LE (c.data + 4 + 4 + 2);
3354 /* image sizes sanity check */
3355 GST_LOG ("w=%u, h=%u, planes=%u, bpp=%u", w, h, planes, bpp);
3356 if (w == 0 || w > 0xfffff || h == 0 || h > 0xfffff || planes != 1 ||
3357 (bpp != 1 && bpp != 4 && bpp != 8 && bpp != 16 && bpp != 24 && bpp != 32))
3360 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_MAXIMUM, "image/bmp",
3361 "width", G_TYPE_INT, w, "height", G_TYPE_INT, h, "bpp", G_TYPE_INT, bpp,
3365 /*** image/tiff ***/
3366 static GstStaticCaps tiff_caps = GST_STATIC_CAPS ("image/tiff, "
3367 "endianness = (int) { BIG_ENDIAN, LITTLE_ENDIAN }");
3368 #define TIFF_CAPS (gst_static_caps_get(&tiff_caps))
3369 static GstStaticCaps tiff_be_caps = GST_STATIC_CAPS ("image/tiff, "
3370 "endianness = (int) BIG_ENDIAN");
3371 #define TIFF_BE_CAPS (gst_static_caps_get(&tiff_be_caps))
3372 static GstStaticCaps tiff_le_caps = GST_STATIC_CAPS ("image/tiff, "
3373 "endianness = (int) LITTLE_ENDIAN");
3374 #define TIFF_LE_CAPS (gst_static_caps_get(&tiff_le_caps))
3376 tiff_type_find (GstTypeFind * tf, gpointer ununsed)
3378 const guint8 *data = gst_type_find_peek (tf, 0, 8);
3379 guint8 le_header[4] = { 0x49, 0x49, 0x2A, 0x00 };
3380 guint8 be_header[4] = { 0x4D, 0x4D, 0x00, 0x2A };
3383 if (memcmp (data, le_header, 4) == 0) {
3384 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, TIFF_LE_CAPS);
3385 } else if (memcmp (data, be_header, 4) == 0) {
3386 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, TIFF_BE_CAPS);
3393 static GstStaticCaps pnm_caps = GST_STATIC_CAPS ("image/x-portable-bitmap; "
3394 "image/x-portable-graymap; image/x-portable-pixmap; "
3395 "image/x-portable-anymap");
3397 #define PNM_CAPS (gst_static_caps_get(&pnm_caps))
3399 #define IS_PNM_WHITESPACE(c) \
3400 ((c) == ' ' || (c) == '\r' || (c) == '\n' || (c) == 't')
3403 pnm_type_find (GstTypeFind * tf, gpointer ununsed)
3405 const gchar *media_type = NULL;
3406 DataScanCtx c = { 0, NULL, 0 };
3409 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 16)))
3412 /* see http://en.wikipedia.org/wiki/Netpbm_format */
3413 if (c.data[0] != 'P' || c.data[1] < '1' || c.data[1] > '7' ||
3414 !IS_PNM_WHITESPACE (c.data[2]) ||
3415 (c.data[3] != '#' && c.data[3] < '0' && c.data[3] > '9'))
3418 switch (c.data[1]) {
3420 media_type = "image/x-portable-bitmap"; /* ASCII */
3423 media_type = "image/x-portable-graymap"; /* ASCII */
3426 media_type = "image/x-portable-pixmap"; /* ASCII */
3429 media_type = "image/x-portable-bitmap"; /* Raw */
3432 media_type = "image/x-portable-graymap"; /* Raw */
3435 media_type = "image/x-portable-pixmap"; /* Raw */
3438 media_type = "image/x-portable-anymap";
3441 g_return_if_reached ();
3444 /* try to extract width and height as well */
3445 if (c.data[1] != '7') {
3446 gchar s[64] = { 0, }
3449 /* need to skip any comment lines first */
3450 data_scan_ctx_advance (tf, &c, 3);
3451 while (c.data[0] == '#') { /* we know there's still data left */
3452 data_scan_ctx_advance (tf, &c, 1);
3453 while (c.data[0] != '\n' && c.data[0] != '\r') {
3454 if (!data_scan_ctx_ensure_data (tf, &c, 4))
3456 data_scan_ctx_advance (tf, &c, 1);
3458 data_scan_ctx_advance (tf, &c, 1);
3459 GST_LOG ("skipped comment line in PNM header");
3462 if (!data_scan_ctx_ensure_data (tf, &c, 32) &&
3463 !data_scan_ctx_ensure_data (tf, &c, 4)) {
3467 /* need to NUL-terminate data for sscanf */
3468 memcpy (s, c.data, MIN (sizeof (s) - 1, c.size));
3469 if (sscanf (s, "%u%c%u%c", &w, &sep1, &h, &sep2) == 4 &&
3470 IS_PNM_WHITESPACE (sep1) && IS_PNM_WHITESPACE (sep2) &&
3471 w > 0 && w < G_MAXINT && h > 0 && h < G_MAXINT) {
3472 GST_LOG ("extracted PNM width and height: %dx%d", w, h);
3478 /* FIXME: extract width + height for anymaps too */
3481 if (w > 0 && h > 0) {
3482 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_MAXIMUM, media_type,
3483 "width", G_TYPE_INT, w, "height", G_TYPE_INT, h, NULL);
3485 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_LIKELY, media_type, NULL);
3489 static GstStaticCaps sds_caps = GST_STATIC_CAPS ("audio/x-sds");
3491 #define SDS_CAPS (gst_static_caps_get(&sds_caps))
3493 sds_type_find (GstTypeFind * tf, gpointer ununsed)
3495 const guint8 *data = gst_type_find_peek (tf, 0, 4);
3496 guint8 mask[4] = { 0xFF, 0xFF, 0x80, 0xFF };
3497 guint8 match[4] = { 0xF0, 0x7E, 0, 0x01 };
3501 for (x = 0; x < 4; x++) {
3502 if ((data[x] & mask[x]) != match[x]) {
3506 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, SDS_CAPS);
3510 static GstStaticCaps ircam_caps = GST_STATIC_CAPS ("audio/x-ircam");
3512 #define IRCAM_CAPS (gst_static_caps_get(&ircam_caps))
3514 ircam_type_find (GstTypeFind * tf, gpointer ununsed)
3516 const guint8 *data = gst_type_find_peek (tf, 0, 4);
3517 guint8 mask[4] = { 0xFF, 0xFF, 0xF8, 0xFF };
3518 guint8 match[4] = { 0x64, 0xA3, 0x00, 0x00 };
3520 gboolean matched = TRUE;
3525 for (x = 0; x < 4; x++) {
3526 if ((data[x] & mask[x]) != match[x]) {
3531 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, IRCAM_CAPS);
3534 /* now try the reverse version */
3536 for (x = 0; x < 4; x++) {
3537 if ((data[x] & mask[3 - x]) != match[3 - x]) {
3543 /*** Matroska/WebM ***/
3545 #define EBML_HEADER 0x1A45DFA3
3546 #define EBML_VERSION 0x4286
3547 #define EBML_DOCTYPE 0x4282
3548 #define EBML_DOCTYPE_VERSION 0x4287
3549 #define MATROSKA_SEGMENT 0x18538067
3550 #define MATROSKA_CLUSTER 0x1F43B675
3551 #define MATROSKA_TRACKS 0x1654AE6B
3552 #define MATROSKA_TRACK_ENTRY 0xAE
3553 #define MATROSKA_TRACK_TYPE 0x83
3554 #define MATROSKA_STEREO_MODE 0x53B8
3556 #define EBML_MAX_LEN (2 * 1024 * 1024)
3560 EBML_DOCTYPE_UNKNOWN = 0,
3561 EBML_DOCTYPE_MATROSKA,
3567 GstEbmlDocType doctype;
3573 guint tracks_ok; /* if we've seen and fully parsed the TRACKS element */
3577 ebml_read_chunk_header (GstTypeFind * tf, DataScanCtx * c, guint max_size,
3578 guint32 * id, guint64 * size)
3581 guint msbit_set, i, len, id_len;
3583 if (c->size < 12 || max_size < 1)
3588 if ((c->data[0] & 0x80) == 0x80) {
3590 } else if ((c->data[0] & 0xC0) == 0x40) {
3592 } else if ((c->data[0] & 0xE0) == 0x20) {
3594 } else if ((c->data[0] & 0xF0) == 0x10) {
3600 if (max_size < id_len)
3603 for (i = 1; i < id_len; ++i) {
3604 *id = (*id << 8) | c->data[i];
3607 data_scan_ctx_advance (tf, c, id_len);
3611 if (max_size < 1 || c->data[0] == 0)
3614 msbit_set = g_bit_nth_msf (c->data[0], 8);
3615 mask = ((1 << msbit_set) - 1);
3616 *size = c->data[0] & mask;
3617 len = 7 - msbit_set;
3619 if (max_size < 1 + len)
3621 for (i = 0; i < len; ++i) {
3622 mask = (mask << 8) | 0xff;
3623 *size = (*size << 8) | c->data[1 + i];
3626 data_scan_ctx_advance (tf, c, 1 + len);
3628 /* undefined/unknown size? (all bits 1) */
3629 if (*size == mask) {
3630 /* allow unknown size for SEGMENT chunk, bail out otherwise */
3631 if (*id == MATROSKA_SEGMENT)
3632 *size = G_MAXUINT64;
3637 return id_len + (1 + len);
3641 ebml_parse_chunk (GstTypeFind * tf, DataScanCtx * ctx, guint32 chunk_id,
3642 guint chunk_size, GstMatroskaInfo * info, guint depth)
3643 { /* FIXME: make sure input size is clipped to 32 bit */
3644 static const gchar SPACES[] = " ";
3645 DataScanCtx c = *ctx;
3646 guint64 element_size;
3647 guint32 id, hdr_len;
3649 if (depth >= 8) /* keep SPACES large enough for depth */
3652 while (chunk_size > 0) {
3653 if (c.offset > EBML_MAX_LEN || !data_scan_ctx_ensure_data (tf, &c, 64))
3656 hdr_len = ebml_read_chunk_header (tf, &c, chunk_size, &id, &element_size);
3660 g_assert (hdr_len <= chunk_size);
3661 chunk_size -= hdr_len;
3663 if (element_size > chunk_size)
3666 GST_DEBUG ("%s %08x, size %" G_GUINT64_FORMAT " / %" G_GUINT64_FORMAT,
3667 SPACES + sizeof (SPACES) - 1 - (2 * depth), id, element_size,
3668 hdr_len + element_size);
3670 if (!data_scan_ctx_ensure_data (tf, &c, element_size)) {
3671 GST_DEBUG ("not enough data");
3677 if (element_size >= 8 && memcmp (c.data, "matroska", 8) == 0)
3678 info->doctype = EBML_DOCTYPE_MATROSKA;
3679 else if (element_size >= 4 && memcmp (c.data, "webm", 4) == 0)
3680 info->doctype = EBML_DOCTYPE_WEBM;
3682 case MATROSKA_SEGMENT:
3683 GST_LOG ("parsing segment");
3684 ebml_parse_chunk (tf, &c, id, element_size, info, depth + 1);
3685 GST_LOG ("parsed segment, done");
3687 case MATROSKA_TRACKS:
3688 GST_LOG ("parsing tracks");
3690 ebml_parse_chunk (tf, &c, id, element_size, info, depth + 1);
3691 GST_LOG ("parsed tracks: %s, done (after %" G_GUINT64_FORMAT " bytes)",
3692 info->tracks_ok ? "ok" : "FAIL", c.offset + element_size);
3694 case MATROSKA_TRACK_ENTRY:
3695 GST_LOG ("parsing track entry");
3696 if (!ebml_parse_chunk (tf, &c, id, element_size, info, depth + 1))
3699 case MATROSKA_TRACK_TYPE:{
3702 /* is supposed to always be 1-byte, but not everyone's following that */
3703 for (i = 0; i < element_size; ++i)
3704 type = (type << 8) | c.data[i];
3706 GST_DEBUG ("%s track type %u",
3707 SPACES + sizeof (SPACES) - 1 - (2 * depth), type);
3711 else if (c.data[0] == 2)
3717 case MATROSKA_STEREO_MODE:
3718 ++info->video_stereo;
3720 case MATROSKA_CLUSTER:
3721 GST_WARNING ("cluster, bailing out (should've found tracks by now)");
3726 data_scan_ctx_advance (tf, &c, element_size);
3727 chunk_size -= element_size;
3734 static GstStaticCaps matroska_caps = GST_STATIC_CAPS ("video/x-matroska");
3736 #define MATROSKA_CAPS (gst_static_caps_get(&matroska_caps))
3738 matroska_type_find (GstTypeFind * tf, gpointer ununsed)
3740 GstTypeFindProbability prob;
3741 GstMatroskaInfo info = { 0, };
3742 const gchar *type_name;
3743 DataScanCtx c = { 0, NULL, 0 };
3746 guint32 id, hdr_len;
3748 if (!data_scan_ctx_ensure_data (tf, &c, 64))
3751 if (GST_READ_UINT32_BE (c.data) != EBML_HEADER)
3754 while (c.offset < EBML_MAX_LEN && data_scan_ctx_ensure_data (tf, &c, 64)) {
3755 hdr_len = ebml_read_chunk_header (tf, &c, c.size, &id, &size);
3759 GST_INFO ("=== top-level chunk %08x, size %" G_GUINT64_FORMAT
3760 " / %" G_GUINT64_FORMAT, id, size, size + hdr_len);
3762 if (!ebml_parse_chunk (tf, &c, id, size, &info, 0))
3764 data_scan_ctx_advance (tf, &c, size);
3765 GST_INFO ("=== done with chunk %08x", id);
3766 if (id == MATROSKA_SEGMENT)
3770 GST_INFO ("audio=%u video=%u other=%u chunks=%u doctype=%d all_tracks=%d",
3771 info.audio, info.video, info.other, info.chunks, info.doctype,
3774 /* perhaps we should bail out if tracks_ok is FALSE and wait for more data?
3775 * (we would need new API to signal this properly and prevent other
3776 * typefinders from taking over the decision then) */
3777 is_audio = (info.audio > 0 && info.video == 0 && info.other == 0);
3779 if (info.doctype == EBML_DOCTYPE_WEBM) {
3780 type_name = (is_audio) ? "audio/webm" : "video/webm";
3781 } else if (info.video > 0 && info.video_stereo) {
3782 type_name = "video/x-matroska-3d";
3784 type_name = (is_audio) ? "audio/x-matroska" : "video/x-matroska";
3787 if (info.doctype == EBML_DOCTYPE_UNKNOWN)
3788 prob = GST_TYPE_FIND_LIKELY;
3790 prob = GST_TYPE_FIND_MAXIMUM;
3792 gst_type_find_suggest_simple (tf, prob, type_name, NULL);
3795 /*** application/mxf ***/
3796 static GstStaticCaps mxf_caps = GST_STATIC_CAPS ("application/mxf");
3798 #define MXF_MAX_PROBE_LENGTH (1024 * 64)
3799 #define MXF_CAPS (gst_static_caps_get(&mxf_caps))
3802 * MXF files start with a header partition pack key of 16 bytes which is defined
3803 * at SMPTE-377M 6.1. Before this there can be up to 64K of run-in which _must_
3804 * not contain the partition pack key.
3807 mxf_type_find (GstTypeFind * tf, gpointer ununsed)
3809 static const guint8 partition_pack_key[] =
3810 { 0x06, 0x0e, 0x2b, 0x34, 0x02, 0x05, 0x01, 0x01, 0x0d, 0x01, 0x02, 0x01,
3813 DataScanCtx c = { 0, NULL, 0 };
3815 while (c.offset <= MXF_MAX_PROBE_LENGTH) {
3817 if (G_UNLIKELY (!data_scan_ctx_ensure_data (tf, &c, 1024)))
3820 /* look over in chunks of 1kbytes to avoid too much overhead */
3822 for (i = 0; i < 1024 - 16; i++) {
3823 /* Check first byte before calling more expensive memcmp function */
3824 if (G_UNLIKELY (c.data[i] == 0x06
3825 && memcmp (c.data + i, partition_pack_key, 13) == 0)) {
3826 /* Header partition pack? */
3827 if (c.data[i + 13] != 0x02)
3830 /* Partition status */
3831 if (c.data[i + 14] >= 0x05)
3834 /* Reserved, must be 0x00 */
3835 if (c.data[i + 15] != 0x00)
3838 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, MXF_CAPS);
3844 data_scan_ctx_advance (tf, &c, 1024 - 16);
3848 /*** video/x-dv ***/
3850 static GstStaticCaps dv_caps = GST_STATIC_CAPS ("video/x-dv, "
3851 "systemstream = (boolean) true");
3852 #define DV_CAPS (gst_static_caps_get(&dv_caps))
3854 dv_type_find (GstTypeFind * tf, gpointer private)
3858 data = gst_type_find_peek (tf, 0, 5);
3860 /* check for DIF and DV flag */
3861 if (data && (data[0] == 0x1f) && (data[1] == 0x07) && (data[2] == 0x00)) {
3862 const gchar *format;
3864 if (data[3] & 0x80) {
3870 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_MAXIMUM, "video/x-dv",
3871 "systemstream", G_TYPE_BOOLEAN, TRUE,
3872 "format", G_TYPE_STRING, format, NULL);
3877 /*** Ogg variants ***/
3878 static GstStaticCaps ogg_caps =
3879 GST_STATIC_CAPS ("application/ogg;video/ogg;audio/ogg;application/kate");
3881 #define OGG_CAPS (gst_static_caps_get(&ogg_caps))
3895 ogganx_type_find (GstTypeFind * tf, gpointer private)
3897 const gchar *media_type;
3898 DataScanCtx c = { 0, NULL, 0 };
3899 guint ogg_syncs = 0;
3900 guint hdr_count[OGG_NUM] = { 0, };
3903 const gchar marker[10];
3905 GstOggStreamType stream_type;
3908 "\001vorbis", 7, OGG_AUDIO}, {
3909 "\200theora", 7, OGG_VIDEO}, {
3910 "fLaC", 4, OGG_AUDIO}, {
3911 "\177FLAC", 5, OGG_AUDIO}, {
3912 "Speex", 5, OGG_AUDIO}, {
3913 "CMML\0\0\0\0", 8, OGG_OTHER}, {
3914 "PCM ", 8, OGG_AUDIO}, {
3915 "Annodex", 7, OGG_ANNODEX}, {
3916 "fishead", 7, OGG_SKELETON}, {
3917 "AnxData", 7, OGG_ANNODEX}, {
3918 "CELT ", 8, OGG_AUDIO}, {
3919 "\200kate\0\0\0", 8, OGG_KATE}, {
3920 "BBCD\0", 5, OGG_VIDEO}, {
3921 "OVP80\1\1", 7, OGG_VIDEO}, {
3922 "OpusHead", 8, OGG_AUDIO}, {
3923 "\001audio\0\0\0", 9, OGG_AUDIO}, {
3924 "\001video\0\0\0", 9, OGG_VIDEO}, {
3925 "\001text\0\0\0", 9, OGG_OTHER}
3928 while (c.offset < 4096 && data_scan_ctx_ensure_data (tf, &c, 64)) {
3931 if (memcmp (c.data, "OggS", 5) != 0)
3937 if (c.data[5] != 0x02)
3940 /* headers should only have one segment */
3941 if (c.data[26] != 1)
3948 data_scan_ctx_advance (tf, &c, 28);
3950 if (!data_scan_ctx_ensure_data (tf, &c, MAX (size, 8)))
3953 for (i = 0; i < G_N_ELEMENTS (markers); ++i) {
3954 if (memcmp (c.data, markers[i].marker, markers[i].marker_size) == 0) {
3955 ++hdr_count[markers[i].stream_type];
3960 if (i == G_N_ELEMENTS (markers)) {
3961 GST_MEMDUMP ("unknown Ogg stream marker", c.data, size);
3962 ++hdr_count[OGG_OTHER];
3965 data_scan_ctx_advance (tf, &c, size);
3971 /* We don't bother with annodex types. FIXME: what about XSPF? */
3972 if (hdr_count[OGG_VIDEO] > 0) {
3973 media_type = "video/ogg";
3974 } else if (hdr_count[OGG_AUDIO] > 0) {
3975 media_type = "audio/ogg";
3976 } else if (hdr_count[OGG_KATE] > 0 && hdr_count[OGG_OTHER] == 0) {
3977 media_type = "application/kate";
3979 media_type = "application/ogg";
3982 GST_INFO ("found %s (audio:%u, video:%u, annodex:%u, skeleton:%u, other:%u)",
3983 media_type, hdr_count[OGG_AUDIO], hdr_count[OGG_VIDEO],
3984 hdr_count[OGG_ANNODEX], hdr_count[OGG_SKELETON], hdr_count[OGG_OTHER]);
3986 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_MAXIMUM, media_type, NULL);
3989 /*** audio/x-vorbis ***/
3990 static GstStaticCaps vorbis_caps = GST_STATIC_CAPS ("audio/x-vorbis");
3992 #define VORBIS_CAPS (gst_static_caps_get(&vorbis_caps))
3994 vorbis_type_find (GstTypeFind * tf, gpointer private)
3996 const guint8 *data = gst_type_find_peek (tf, 0, 30);
4002 /* 1 byte packet type (identification=0x01)
4003 6 byte string "vorbis"
4004 4 byte vorbis version */
4005 if (memcmp (data, "\001vorbis\000\000\000\000", 11) != 0)
4008 /* 1 byte channels must be != 0 */
4012 /* 4 byte samplerate must be != 0 */
4013 if (GST_READ_UINT32_LE (data) == 0)
4016 /* blocksize checks */
4017 blocksize_0 = data[0] & 0x0F;
4018 blocksize_1 = (data[0] & 0xF0) >> 4;
4019 if (blocksize_0 > blocksize_1)
4021 if (blocksize_0 < 6 || blocksize_0 > 13)
4023 if (blocksize_1 < 6 || blocksize_1 > 13)
4027 if ((data[0] & 0x01) != 1)
4029 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, VORBIS_CAPS);
4033 /*** video/x-theora ***/
4035 static GstStaticCaps theora_caps = GST_STATIC_CAPS ("video/x-theora");
4037 #define THEORA_CAPS (gst_static_caps_get(&theora_caps))
4039 theora_type_find (GstTypeFind * tf, gpointer private)
4041 const guint8 *data = gst_type_find_peek (tf, 0, 7); //42);
4044 if (data[0] != 0x80)
4046 if (memcmp (&data[1], "theora", 6) != 0)
4048 /* FIXME: make this more reliable when specs are out */
4050 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, THEORA_CAPS);
4056 kate_type_find (GstTypeFind * tf, gpointer private)
4058 const guint8 *data = gst_type_find_peek (tf, 0, 64);
4059 gchar category[16] = { 0, };
4061 if (G_UNLIKELY (data == NULL))
4064 /* see: http://wiki.xiph.org/index.php/OggKate#Format_specification */
4065 if (G_LIKELY (memcmp (data, "\200kate\0\0\0", 8) != 0))
4068 /* make sure we always have a NUL-terminated string */
4069 memcpy (category, data + 48, 15);
4070 GST_LOG ("kate category: %s", category);
4071 /* canonical categories for subtitles: subtitles, spu-subtitles, SUB, K-SPU */
4072 if (strcmp (category, "subtitles") == 0 || strcmp (category, "SUB") == 0 ||
4073 strcmp (category, "spu-subtitles") == 0 ||
4074 strcmp (category, "K-SPU") == 0) {
4075 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_MAXIMUM,
4076 "subtitle/x-kate", NULL);
4078 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_MAXIMUM,
4079 "application/x-kate", NULL);
4083 /*** application/x-ogm-video or audio***/
4085 static GstStaticCaps ogmvideo_caps =
4086 GST_STATIC_CAPS ("application/x-ogm-video");
4087 #define OGMVIDEO_CAPS (gst_static_caps_get(&ogmvideo_caps))
4089 ogmvideo_type_find (GstTypeFind * tf, gpointer private)
4091 const guint8 *data = gst_type_find_peek (tf, 0, 9);
4094 if (memcmp (data, "\001video\000\000\000", 9) != 0)
4096 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, OGMVIDEO_CAPS);
4100 static GstStaticCaps ogmaudio_caps =
4101 GST_STATIC_CAPS ("application/x-ogm-audio");
4102 #define OGMAUDIO_CAPS (gst_static_caps_get(&ogmaudio_caps))
4104 ogmaudio_type_find (GstTypeFind * tf, gpointer private)
4106 const guint8 *data = gst_type_find_peek (tf, 0, 9);
4109 if (memcmp (data, "\001audio\000\000\000", 9) != 0)
4111 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, OGMAUDIO_CAPS);
4115 static GstStaticCaps ogmtext_caps = GST_STATIC_CAPS ("application/x-ogm-text");
4117 #define OGMTEXT_CAPS (gst_static_caps_get(&ogmtext_caps))
4119 ogmtext_type_find (GstTypeFind * tf, gpointer private)
4121 const guint8 *data = gst_type_find_peek (tf, 0, 9);
4124 if (memcmp (data, "\001text\000\000\000\000", 9) != 0)
4126 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, OGMTEXT_CAPS);
4130 /*** audio/x-speex ***/
4132 static GstStaticCaps speex_caps = GST_STATIC_CAPS ("audio/x-speex");
4134 #define SPEEX_CAPS (gst_static_caps_get(&speex_caps))
4136 speex_type_find (GstTypeFind * tf, gpointer private)
4138 const guint8 *data = gst_type_find_peek (tf, 0, 80);
4141 /* 8 byte string "Speex "
4142 24 byte speex version string + int */
4143 if (memcmp (data, "Speex ", 8) != 0)
4147 /* 4 byte header size >= 80 */
4148 if (GST_READ_UINT32_LE (data) < 80)
4152 /* 4 byte sample rate <= 48000 */
4153 if (GST_READ_UINT32_LE (data) > 48000)
4157 /* currently there are only 3 speex modes. */
4158 if (GST_READ_UINT32_LE (data) > 3)
4162 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, SPEEX_CAPS);
4166 /*** audio/x-celt ***/
4168 static GstStaticCaps celt_caps = GST_STATIC_CAPS ("audio/x-celt");
4170 #define CELT_CAPS (gst_static_caps_get(&celt_caps))
4172 celt_type_find (GstTypeFind * tf, gpointer private)
4174 const guint8 *data = gst_type_find_peek (tf, 0, 8);
4177 /* 8 byte string "CELT " */
4178 if (memcmp (data, "CELT ", 8) != 0)
4181 /* TODO: Check other values of the CELT header */
4182 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, CELT_CAPS);
4186 /*** application/x-ogg-skeleton ***/
4187 static GstStaticCaps ogg_skeleton_caps =
4188 GST_STATIC_CAPS ("application/x-ogg-skeleton, parsed=(boolean)FALSE");
4189 #define OGG_SKELETON_CAPS (gst_static_caps_get(&ogg_skeleton_caps))
4191 oggskel_type_find (GstTypeFind * tf, gpointer private)
4193 const guint8 *data = gst_type_find_peek (tf, 0, 12);
4196 /* 8 byte string "fishead\0" for the ogg skeleton stream */
4197 if (memcmp (data, "fishead\0", 8) != 0)
4201 /* Require that the header contains version 3.0 */
4202 if (GST_READ_UINT16_LE (data) != 3)
4205 if (GST_READ_UINT16_LE (data) != 0)
4208 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, OGG_SKELETON_CAPS);
4212 static GstStaticCaps cmml_caps = GST_STATIC_CAPS ("text/x-cmml");
4214 #define CMML_CAPS (gst_static_caps_get(&cmml_caps))
4216 cmml_type_find (GstTypeFind * tf, gpointer private)
4218 /* Header is 12 bytes minimum (though we don't check the minor version */
4219 const guint8 *data = gst_type_find_peek (tf, 0, 12);
4223 /* 8 byte string "CMML\0\0\0\0" for the magic number */
4224 if (memcmp (data, "CMML\0\0\0\0", 8) != 0)
4228 /* Require that the header contains at least version 2.0 */
4229 if (GST_READ_UINT16_LE (data) < 2)
4232 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, CMML_CAPS);
4236 /*** application/x-tar ***/
4238 static GstStaticCaps tar_caps = GST_STATIC_CAPS ("application/x-tar");
4240 #define TAR_CAPS (gst_static_caps_get(&tar_caps))
4241 #define OLDGNU_MAGIC "ustar " /* 7 chars and a NUL */
4242 #define NEWGNU_MAGIC "ustar" /* 5 chars and a NUL */
4244 tar_type_find (GstTypeFind * tf, gpointer unused)
4246 const guint8 *data = gst_type_find_peek (tf, 257, 8);
4248 /* of course we are not certain, but we don't want other typefind funcs
4249 * to detect formats of files within the tar archive, e.g. mp3s */
4251 if (memcmp (data, OLDGNU_MAGIC, 8) == 0) { /* sic */
4252 gst_type_find_suggest (tf, GST_TYPE_FIND_NEARLY_CERTAIN, TAR_CAPS);
4253 } else if (memcmp (data, NEWGNU_MAGIC, 6) == 0 && /* sic */
4254 g_ascii_isdigit (data[6]) && g_ascii_isdigit (data[7])) {
4255 gst_type_find_suggest (tf, GST_TYPE_FIND_NEARLY_CERTAIN, TAR_CAPS);
4260 /*** application/x-ar ***/
4262 static GstStaticCaps ar_caps = GST_STATIC_CAPS ("application/x-ar");
4264 #define AR_CAPS (gst_static_caps_get(&ar_caps))
4266 ar_type_find (GstTypeFind * tf, gpointer unused)
4268 const guint8 *data = gst_type_find_peek (tf, 0, 24);
4270 if (data && memcmp (data, "!<arch>", 7) == 0) {
4273 for (i = 7; i < 24; ++i) {
4274 if (!g_ascii_isprint (data[i]) && data[i] != '\n') {
4275 gst_type_find_suggest (tf, GST_TYPE_FIND_POSSIBLE, AR_CAPS);
4279 gst_type_find_suggest (tf, GST_TYPE_FIND_NEARLY_CERTAIN, AR_CAPS);
4283 /*** audio/x-au ***/
4285 /* NOTE: we cannot replace this function with TYPE_FIND_REGISTER_START_WITH,
4286 * as it is only possible to register one typefind factory per 'name'
4287 * (which is in this case the caps), and the first one would be replaced by
4288 * the second one. */
4289 static GstStaticCaps au_caps = GST_STATIC_CAPS ("audio/x-au");
4291 #define AU_CAPS (gst_static_caps_get(&au_caps))
4293 au_type_find (GstTypeFind * tf, gpointer unused)
4295 const guint8 *data = gst_type_find_peek (tf, 0, 4);
4298 if (memcmp (data, ".snd", 4) == 0 || memcmp (data, "dns.", 4) == 0) {
4299 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, AU_CAPS);
4305 /*** video/x-nuv ***/
4307 /* NOTE: we cannot replace this function with TYPE_FIND_REGISTER_START_WITH,
4308 * as it is only possible to register one typefind factory per 'name'
4309 * (which is in this case the caps), and the first one would be replaced by
4310 * the second one. */
4311 static GstStaticCaps nuv_caps = GST_STATIC_CAPS ("video/x-nuv");
4313 #define NUV_CAPS (gst_static_caps_get(&nuv_caps))
4315 nuv_type_find (GstTypeFind * tf, gpointer unused)
4317 const guint8 *data = gst_type_find_peek (tf, 0, 11);
4320 if (memcmp (data, "MythTVVideo", 11) == 0
4321 || memcmp (data, "NuppelVideo", 11) == 0) {
4322 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, NUV_CAPS);
4327 /*** audio/x-paris ***/
4328 /* NOTE: do not replace this function with two TYPE_FIND_REGISTER_START_WITH */
4329 static GstStaticCaps paris_caps = GST_STATIC_CAPS ("audio/x-paris");
4331 #define PARIS_CAPS (gst_static_caps_get(&paris_caps))
4333 paris_type_find (GstTypeFind * tf, gpointer unused)
4335 const guint8 *data = gst_type_find_peek (tf, 0, 4);
4338 if (memcmp (data, " paf", 4) == 0 || memcmp (data, "fap ", 4) == 0) {
4339 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, PARIS_CAPS);
4344 /*** audio/x-sbc ***/
4345 static GstStaticCaps sbc_caps = GST_STATIC_CAPS ("audio/x-sbc");
4346 #define SBC_CAPS (gst_static_caps_get(&sbc_caps))
4349 sbc_check_header (const guint8 * data, gsize len, guint * rate,
4352 static const guint16 sbc_rates[4] = { 16000, 32000, 44100, 48000 };
4353 static const guint8 sbc_blocks[4] = { 4, 8, 12, 16 };
4354 guint n_blocks, ch_mode, n_subbands, bitpool;
4356 if (data[0] != 0x9C || len < 4)
4359 n_blocks = sbc_blocks[(data[1] >> 4) & 0x03];
4360 ch_mode = (data[1] >> 2) & 0x03;
4361 n_subbands = (data[1] & 0x01) ? 8 : 4;
4366 *rate = sbc_rates[(data[1] >> 6) & 0x03];
4367 *channels = (ch_mode == 0) ? 1 : 2;
4370 return 4 + (n_subbands * 1) / 2 + (n_blocks * 1 * bitpool) / 8;
4371 else if (ch_mode == 1)
4372 return 4 + (n_subbands * 2) / 2 + (n_blocks * 2 * bitpool) / 8;
4373 else if (ch_mode == 2)
4374 return 4 + (n_subbands * 2) / 2 + (n_blocks * bitpool) / 8;
4375 else if (ch_mode == 3)
4376 return 4 + (n_subbands * 2) / 2 + (n_subbands + n_blocks * bitpool) / 8;
4382 sbc_type_find (GstTypeFind * tf, gpointer unused)
4386 guint i, rate, channels, offset = 0;
4388 for (i = 0; i < 10; ++i) {
4389 data = gst_type_find_peek (tf, offset, 8);
4393 frame_len = sbc_check_header (data, 8, &rate, &channels);
4397 offset += frame_len;
4399 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_POSSIBLE, "audio/x-sbc",
4400 "rate", G_TYPE_INT, rate, "channels", G_TYPE_INT, channels,
4401 "parsed", G_TYPE_BOOLEAN, FALSE, NULL);
4404 /*** audio/iLBC-sh ***/
4405 /* NOTE: do not replace this function with two TYPE_FIND_REGISTER_START_WITH */
4406 static GstStaticCaps ilbc_caps = GST_STATIC_CAPS ("audio/iLBC-sh");
4408 #define ILBC_CAPS (gst_static_caps_get(&ilbc_caps))
4410 ilbc_type_find (GstTypeFind * tf, gpointer unused)
4412 const guint8 *data = gst_type_find_peek (tf, 0, 8);
4415 if (memcmp (data, "#!iLBC30", 8) == 0 || memcmp (data, "#!iLBC20", 8) == 0) {
4416 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY, ILBC_CAPS);
4421 /*** application/x-ms-dos-executable ***/
4423 static GstStaticCaps msdos_caps =
4424 GST_STATIC_CAPS ("application/x-ms-dos-executable");
4425 #define MSDOS_CAPS (gst_static_caps_get(&msdos_caps))
4426 /* see http://www.madchat.org/vxdevl/papers/winsys/pefile/pefile.htm */
4428 msdos_type_find (GstTypeFind * tf, gpointer unused)
4430 const guint8 *data = gst_type_find_peek (tf, 0, 64);
4432 if (data && data[0] == 'M' && data[1] == 'Z' &&
4433 GST_READ_UINT16_LE (data + 8) == 4) {
4434 guint32 pe_offset = GST_READ_UINT32_LE (data + 60);
4436 data = gst_type_find_peek (tf, pe_offset, 2);
4437 if (data && data[0] == 'P' && data[1] == 'E') {
4438 gst_type_find_suggest (tf, GST_TYPE_FIND_NEARLY_CERTAIN, MSDOS_CAPS);
4443 /*** application/x-mmsh ***/
4445 static GstStaticCaps mmsh_caps = GST_STATIC_CAPS ("application/x-mmsh");
4447 #define MMSH_CAPS gst_static_caps_get(&mmsh_caps)
4449 /* This is to recognise mssh-over-http */
4451 mmsh_type_find (GstTypeFind * tf, gpointer unused)
4453 static const guint8 asf_marker[16] = { 0x30, 0x26, 0xb2, 0x75, 0x8e, 0x66,
4454 0xcf, 0x11, 0xa6, 0xd9, 0x00, 0xaa, 0x00, 0x62, 0xce, 0x6c
4459 data = gst_type_find_peek (tf, 0, 2 + 2 + 4 + 2 + 2 + 16);
4460 if (data && data[0] == 0x24 && data[1] == 0x48 &&
4461 GST_READ_UINT16_LE (data + 2) > 2 + 2 + 4 + 2 + 2 + 16 &&
4462 memcmp (data + 2 + 2 + 4 + 2 + 2, asf_marker, 16) == 0) {
4463 gst_type_find_suggest (tf, GST_TYPE_FIND_LIKELY, MMSH_CAPS);
4467 /*** video/x-dirac ***/
4469 /* NOTE: we cannot replace this function with TYPE_FIND_REGISTER_START_WITH,
4470 * as it is only possible to register one typefind factory per 'name'
4471 * (which is in this case the caps), and the first one would be replaced by
4472 * the second one. */
4473 static GstStaticCaps dirac_caps = GST_STATIC_CAPS ("video/x-dirac");
4475 #define DIRAC_CAPS (gst_static_caps_get(&dirac_caps))
4477 dirac_type_find (GstTypeFind * tf, gpointer unused)
4479 const guint8 *data = gst_type_find_peek (tf, 0, 8);
4482 if (memcmp (data, "BBCD", 4) == 0 || memcmp (data, "KW-DIRAC", 8) == 0) {
4483 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, DIRAC_CAPS);
4488 /*** video/vivo ***/
4490 static GstStaticCaps vivo_caps = GST_STATIC_CAPS ("video/vivo");
4492 #define VIVO_CAPS gst_static_caps_get(&vivo_caps)
4495 vivo_type_find (GstTypeFind * tf, gpointer unused)
4497 static const guint8 vivo_marker[] = { 'V', 'e', 'r', 's', 'i', 'o', 'n',
4498 ':', 'V', 'i', 'v', 'o', '/'
4503 data = gst_type_find_peek (tf, 0, 1024);
4504 if (data == NULL || data[0] != 0x00)
4507 if ((data[1] & 0x80)) {
4508 if ((data[2] & 0x80))
4510 hdr_len = ((guint) (data[1] & 0x7f)) << 7;
4520 /* 1008 = 1022 - strlen ("Version:Vivo/") - 1 */
4521 while (pos < 1008 && data[pos] == '\r' && data[pos + 1] == '\n')
4524 if (memcmp (data + pos, vivo_marker, sizeof (vivo_marker)) == 0) {
4525 gst_type_find_suggest (tf, GST_TYPE_FIND_MAXIMUM, VIVO_CAPS);
4529 /*** XDG MIME typefinder (to avoid false positives mostly) ***/
4533 xdgmime_typefind (GstTypeFind * find, gpointer user_data)
4536 gsize length = 16384;
4541 if ((tf_length = gst_type_find_get_length (find)) > 0)
4542 length = MIN (length, tf_length);
4544 if ((data = gst_type_find_peek (find, 0, length)) == NULL)
4547 tmp = g_content_type_guess (NULL, data, length, NULL);
4548 if (tmp == NULL || g_content_type_is_unknown (tmp)) {
4553 mimetype = g_content_type_get_mime_type (tmp);
4556 if (mimetype == NULL)
4559 GST_DEBUG ("Got mimetype '%s'", mimetype);
4561 /* Ignore audio/video types:
4562 * - our own typefinders in -base are likely to be better at this
4563 * (and if they're not, we really want to fix them, that's why we don't
4564 * report xdg-detected audio/video types at all, not even with a low
4566 * - we want to detect GStreamer media types and not MIME types
4567 * - the purpose of this xdg mime finder is mainly to prevent false
4568 * positives of non-media formats, not to typefind audio/video formats */
4569 if (g_str_has_prefix (mimetype, "audio/") ||
4570 g_str_has_prefix (mimetype, "video/")) {
4571 GST_LOG ("Ignoring audio/video mime type");
4576 /* Again, we mainly want the xdg typefinding to prevent false-positives on
4577 * non-media formats, so suggest the type with a probability that trumps
4578 * uncertain results of our typefinders, but not more than that. */
4579 GST_LOG ("Suggesting '%s' with probability POSSIBLE", mimetype);
4580 gst_type_find_suggest_simple (find, GST_TYPE_FIND_POSSIBLE, mimetype, NULL);
4583 #endif /* USE_GIO */
4585 /*** Windows icon typefinder (to avoid false positives mostly) ***/
4588 windows_icon_typefind (GstTypeFind * find, gpointer user_data)
4592 guint16 type, nimages;
4593 gint32 size, offset;
4595 datalen = gst_type_find_get_length (find);
4596 if ((data = gst_type_find_peek (find, 0, 6)) == NULL)
4599 /* header - simple and not enough to rely on it alone */
4600 if (GST_READ_UINT16_LE (data) != 0)
4602 type = GST_READ_UINT16_LE (data + 2);
4603 if (type != 1 && type != 2)
4605 nimages = GST_READ_UINT16_LE (data + 4);
4606 if (nimages == 0) /* we can assume we can't have an empty image file ? */
4610 if (data[6 + 3] != 0)
4613 guint16 planes = GST_READ_UINT16_LE (data + 6 + 4);
4617 size = GST_READ_UINT32_LE (data + 6 + 8);
4618 offset = GST_READ_UINT32_LE (data + 6 + 12);
4619 if (offset < 0 || size <= 0 || size >= datalen || offset >= datalen
4620 || size + offset > datalen)
4623 gst_type_find_suggest_simple (find, GST_TYPE_FIND_NEARLY_CERTAIN,
4624 "image/x-icon", NULL);
4627 /*** WAP WBMP typefinder ***/
4630 wbmp_typefind (GstTypeFind * find, gpointer user_data)
4636 /* http://en.wikipedia.org/wiki/Wireless_Application_Protocol_Bitmap_Format */
4637 datalen = gst_type_find_get_length (find);
4641 data = gst_type_find_peek (find, 0, 5);
4645 /* want 0x00 0x00 at start */
4646 if (*data++ != 0 || *data++ != 0)
4649 /* min header size */
4652 /* let's assume max width/height is 65536 */
4655 w = (w << 8) | *data++;
4659 data = gst_type_find_peek (find, 4, 2);
4665 h = (h << 8) | *data++;
4671 if (w == 0 || h == 0)
4674 /* now add bitmap size */
4675 size += h * (GST_ROUND_UP_8 (w) / 8);
4677 if (datalen == size) {
4678 gst_type_find_suggest_simple (find, GST_TYPE_FIND_POSSIBLE - 10,
4679 "image/vnd.wap.wbmp", NULL);
4683 /*** DEGAS Atari images (also to avoid false positives, see #625129) ***/
4685 degas_type_find (GstTypeFind * tf, gpointer private)
4687 /* No magic, but it should have a fixed size and a few invalid values */
4688 /* http://www.fileformat.info/format/atari/spec/6ecf9f6eb5be494284a47feb8a214687/view.htm */
4694 len = gst_type_find_get_length (tf);
4695 if (len < 34) /* smallest header of the lot */
4697 data = gst_type_find_peek (tf, 0, 4);
4698 if (G_UNLIKELY (data == NULL))
4700 resolution = GST_READ_UINT16_BE (data);
4702 /* could be DEGAS */
4703 if (resolution <= 2)
4704 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_POSSIBLE + 5,
4705 "image/x-degas", NULL);
4706 } else if (len == 32066) {
4707 /* could be DEGAS Elite */
4708 if (resolution <= 2) {
4709 data = gst_type_find_peek (tf, len - 16, 8);
4710 if (G_UNLIKELY (data == NULL))
4712 for (n = 0; n < 4; n++) {
4713 if (GST_READ_UINT16_BE (data + n * 2) > 2)
4716 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_POSSIBLE + 5,
4717 "image/x-degas", NULL);
4719 } else if (len >= 66 && len < 32066) {
4720 /* could be compressed DEGAS Elite, but it's compressed and so we can't rely on size,
4721 it does have 4 16 bytes values near the end that are 0-2 though. */
4722 if ((resolution & 0x8000) && (resolution & 0x7fff) <= 2) {
4723 data = gst_type_find_peek (tf, len - 16, 8);
4724 if (G_UNLIKELY (data == NULL))
4726 for (n = 0; n < 4; n++) {
4727 if (GST_READ_UINT16_BE (data + n * 2) > 2)
4730 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_POSSIBLE + 5,
4731 "image/x-degas", NULL);
4736 /*** DVD ISO images (looks like H.264, see #674069) ***/
4738 dvdiso_type_find (GstTypeFind * tf, gpointer private)
4740 /* 0x8000 bytes of zeros, then "\001CD001" */
4744 len = gst_type_find_get_length (tf);
4747 data = gst_type_find_peek (tf, 0, 0x8006);
4748 if (G_UNLIKELY (data == NULL))
4750 for (len = 0; len < 0x8000; len++)
4753 /* Can the '1' be anything else ? My three samples all have '1'. */
4754 if (memcmp (data + 0x8000, "\001CD001", 6))
4757 /* May need more inspection, we may be able to demux some of them */
4758 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_LIKELY,
4759 "application/octet-stream", NULL);
4762 /* SSA/ASS subtitles
4764 * http://en.wikipedia.org/wiki/SubStation_Alpha
4765 * http://matroska.org/technical/specs/subtitles/ssa.html
4768 ssa_type_find (GstTypeFind * tf, gpointer private)
4770 const gchar *start, *end, *ver_str, *media_type = NULL;
4772 gchar *str, *script_type, *p = NULL;
4775 data = gst_type_find_peek (tf, 0, 32);
4780 /* there might be a BOM at the beginning */
4781 if (memcmp (data, "[Script Info]", 13) != 0 &&
4782 memcmp (data + 2, "[Script Info]", 13) != 0 &&
4783 memcmp (data + 3, "[Script Info]", 13) != 0 &&
4784 memcmp (data + 4, "[Script Info]", 13) != 0) {
4788 /* now check if we have SSA or ASS */
4789 len = gst_type_find_get_length (tf);
4793 data = gst_type_find_peek (tf, 0, len);
4798 start = (gchar *) memchr (data, '[', 5);
4800 len -= (start - (gchar *) data);
4802 /* ignore anything non-UTF8 for now, in future we might at least allow
4803 * other UTF variants that are clearly prefixed with the appropriate BOM */
4804 if (!g_utf8_validate (start, len, &end) && (len - (end - start)) > 6) {
4805 GST_FIXME ("non-UTF8 SSA/ASS file");
4809 /* something at start, but not a UTF-8 BOM? */
4810 if (data[0] != '[' && (data[0] != 0xEF || data[1] != 0xBB || data[2] != 0xBF))
4813 /* ignore any partial UTF-8 characters at the end */
4816 /* create a NUL-terminated string so it's easier to process it safely */
4817 str = g_strndup (start, len - 1);
4818 script_type = strstr (str, "ScriptType:");
4819 if (script_type != NULL) {
4822 ver_str = script_type + 11;
4823 while (*ver_str == ' ' || *ver_str == 'v' || *ver_str == 'V')
4825 version = g_ascii_strtod (ver_str, &p);
4826 if (version == 4.0 && p != NULL && *p == '+')
4827 media_type = "application/x-ass";
4828 else if (version >= 1.0 && version <= 4.0)
4829 media_type = "application/x-ssa";
4832 if (media_type == NULL) {
4833 if (strstr (str, "[v4+ Styles]") || strstr (str, "[V4+ Styles]"))
4834 media_type = "application/x-ass";
4835 else if (strstr (str, "[v4 Styles]") || strstr (str, "[V4 Styles]"))
4836 media_type = "application/x-ssa";
4839 if (media_type != NULL) {
4840 gst_type_find_suggest_simple (tf, GST_TYPE_FIND_MAXIMUM,
4841 media_type, "parsed", G_TYPE_BOOLEAN, FALSE, NULL);
4843 GST_WARNING ("could not detect SSA/ASS variant");
4849 /*** generic typefind for streams that have some data at a specific position***/
4860 start_with_type_find (GstTypeFind * tf, gpointer private)
4862 GstTypeFindData *start_with = (GstTypeFindData *) private;
4865 GST_LOG ("trying to find mime type %s with the first %u bytes of data",
4866 gst_structure_get_name (gst_caps_get_structure (start_with->caps, 0)),
4868 data = gst_type_find_peek (tf, 0, start_with->size);
4869 if (data && memcmp (data, start_with->data, start_with->size) == 0) {
4870 gst_type_find_suggest (tf, start_with->probability, start_with->caps);
4875 sw_data_destroy (GstTypeFindData * sw_data)
4877 if (G_LIKELY (sw_data->caps != NULL))
4878 gst_caps_unref (sw_data->caps);
4882 #define TYPE_FIND_REGISTER_START_WITH(plugin,name,rank,ext,_data,_size,_probability)\
4884 GstTypeFindData *sw_data = g_new (GstTypeFindData, 1); \
4885 sw_data->data = (const guint8 *)_data; \
4886 sw_data->size = _size; \
4887 sw_data->probability = _probability; \
4888 sw_data->caps = gst_caps_new_empty_simple (name); \
4889 if (!gst_type_find_register (plugin, name, rank, start_with_type_find,\
4890 ext, sw_data->caps, sw_data, \
4891 (GDestroyNotify) (sw_data_destroy))) { \
4892 gst_caps_unref (sw_data->caps); \
4897 /*** same for riff types ***/
4900 riff_type_find (GstTypeFind * tf, gpointer private)
4902 GstTypeFindData *riff_data = (GstTypeFindData *) private;
4903 const guint8 *data = gst_type_find_peek (tf, 0, 12);
4905 if (data && (memcmp (data, "RIFF", 4) == 0 || memcmp (data, "AVF0", 4) == 0)) {
4907 if (memcmp (data, riff_data->data, 4) == 0)
4908 gst_type_find_suggest (tf, riff_data->probability, riff_data->caps);
4912 #define TYPE_FIND_REGISTER_RIFF(plugin,name,rank,ext,_data) \
4914 GstTypeFindData *sw_data = g_new (GstTypeFindData, 1); \
4915 sw_data->data = (gpointer)_data; \
4916 sw_data->size = 4; \
4917 sw_data->probability = GST_TYPE_FIND_MAXIMUM; \
4918 sw_data->caps = gst_caps_new_empty_simple (name); \
4919 if (!gst_type_find_register (plugin, name, rank, riff_type_find, \
4920 ext, sw_data->caps, sw_data, \
4921 (GDestroyNotify) (sw_data_destroy))) { \
4922 gst_caps_unref (sw_data->caps); \
4928 /*** plugin initialization ***/
4930 #define TYPE_FIND_REGISTER(plugin,name,rank,func,ext,caps,priv,notify) \
4932 if (!gst_type_find_register (plugin, name, rank, func, ext, caps, priv, notify))\
4938 plugin_init (GstPlugin * plugin)
4940 /* can't initialize this via a struct as caps can't be statically initialized */
4942 GST_DEBUG_CATEGORY_INIT (type_find_debug, "typefindfunctions",
4943 GST_DEBUG_FG_GREEN | GST_DEBUG_BG_RED, "generic type find functions");
4945 /* note: asx/wax/wmx are XML files, asf doesn't handle them */
4946 /* must use strings, macros don't accept initializers */
4947 TYPE_FIND_REGISTER_START_WITH (plugin, "video/x-ms-asf", GST_RANK_SECONDARY,
4949 "\060\046\262\165\216\146\317\021\246\331\000\252\000\142\316\154", 16,
4950 GST_TYPE_FIND_MAXIMUM);
4951 TYPE_FIND_REGISTER (plugin, "audio/x-musepack", GST_RANK_PRIMARY,
4952 musepack_type_find, "mpc,mpp,mp+", MUSEPACK_CAPS, NULL, NULL);
4953 TYPE_FIND_REGISTER (plugin, "audio/x-au", GST_RANK_MARGINAL,
4954 au_type_find, "au,snd", AU_CAPS, NULL, NULL);
4955 TYPE_FIND_REGISTER_RIFF (plugin, "video/x-msvideo", GST_RANK_PRIMARY,
4957 TYPE_FIND_REGISTER_RIFF (plugin, "audio/qcelp", GST_RANK_PRIMARY,
4959 TYPE_FIND_REGISTER_RIFF (plugin, "video/x-cdxa", GST_RANK_PRIMARY,
4961 TYPE_FIND_REGISTER_START_WITH (plugin, "video/x-vcd", GST_RANK_PRIMARY,
4962 "dat", "\000\377\377\377\377\377\377\377\377\377\377\000", 12,
4963 GST_TYPE_FIND_MAXIMUM);
4964 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-imelody", GST_RANK_PRIMARY,
4965 "imy,ime,imelody", "BEGIN:IMELODY", 13, GST_TYPE_FIND_MAXIMUM);
4967 TYPE_FIND_REGISTER_START_WITH (plugin, "video/x-smoke", GST_RANK_PRIMARY,
4968 NULL, "\x80smoke\x00\x01\x00", 6, GST_TYPE_FIND_MAXIMUM);
4970 TYPE_FIND_REGISTER (plugin, "audio/midi", GST_RANK_PRIMARY, mid_type_find,
4971 "mid,midi", MID_CAPS, NULL, NULL);
4972 TYPE_FIND_REGISTER_RIFF (plugin, "audio/riff-midi", GST_RANK_PRIMARY,
4973 "mid,midi", "RMID");
4974 TYPE_FIND_REGISTER (plugin, "audio/mobile-xmf", GST_RANK_PRIMARY,
4975 mxmf_type_find, "mxmf", MXMF_CAPS, NULL, NULL);
4976 TYPE_FIND_REGISTER (plugin, "video/x-fli", GST_RANK_MARGINAL, flx_type_find,
4977 "flc,fli", FLX_CAPS, NULL, NULL);
4978 TYPE_FIND_REGISTER (plugin, "application/x-id3v2", GST_RANK_PRIMARY + 103,
4979 id3v2_type_find, "mp3,mp2,mp1,mpga,ogg,flac,tta", ID3_CAPS, NULL, NULL);
4980 TYPE_FIND_REGISTER (plugin, "application/x-id3v1", GST_RANK_PRIMARY + 101,
4981 id3v1_type_find, "mp3,mp2,mp1,mpga,ogg,flac,tta", ID3_CAPS, NULL, NULL);
4982 TYPE_FIND_REGISTER (plugin, "application/x-apetag", GST_RANK_PRIMARY + 102,
4983 apetag_type_find, "mp3,ape,mpc,wv", APETAG_CAPS, NULL, NULL);
4984 TYPE_FIND_REGISTER (plugin, "audio/x-ttafile", GST_RANK_PRIMARY,
4985 tta_type_find, "tta", TTA_CAPS, NULL, NULL);
4986 TYPE_FIND_REGISTER (plugin, "audio/x-mod", GST_RANK_SECONDARY, mod_type_find,
4987 "669,amf,dsm,gdm,far,imf,it,med,mod,mtm,okt,sam,s3m,stm,stx,ult,xm",
4988 MOD_CAPS, NULL, NULL);
4989 TYPE_FIND_REGISTER (plugin, "audio/mpeg", GST_RANK_PRIMARY, mp3_type_find,
4990 "mp3,mp2,mp1,mpga", MP3_CAPS, NULL, NULL);
4991 TYPE_FIND_REGISTER (plugin, "audio/x-ac3", GST_RANK_PRIMARY, ac3_type_find,
4992 "ac3,eac3", AC3_CAPS, NULL, NULL);
4993 TYPE_FIND_REGISTER (plugin, "audio/x-dts", GST_RANK_SECONDARY, dts_type_find,
4994 "dts", DTS_CAPS, NULL, NULL);
4995 TYPE_FIND_REGISTER (plugin, "audio/x-gsm", GST_RANK_PRIMARY, NULL, "gsm",
4996 GSM_CAPS, NULL, NULL);
4997 TYPE_FIND_REGISTER (plugin, "video/mpeg-sys", GST_RANK_PRIMARY,
4998 mpeg_sys_type_find, "mpe,mpeg,mpg", MPEG_SYS_CAPS, NULL, NULL);
4999 TYPE_FIND_REGISTER (plugin, "video/mpegts", GST_RANK_PRIMARY,
5000 mpeg_ts_type_find, "ts,mts", MPEGTS_CAPS, NULL, NULL);
5001 TYPE_FIND_REGISTER (plugin, "application/ogg", GST_RANK_PRIMARY,
5002 ogganx_type_find, "ogg,oga,ogv,ogm,ogx,spx,anx,axa,axv", OGG_CAPS,
5004 TYPE_FIND_REGISTER (plugin, "video/mpeg-elementary", GST_RANK_MARGINAL,
5005 mpeg_video_stream_type_find, "mpv,mpeg,mpg", MPEG_VIDEO_CAPS, NULL, NULL);
5006 TYPE_FIND_REGISTER (plugin, "video/mpeg4", GST_RANK_PRIMARY,
5007 mpeg4_video_type_find, "m4v", MPEG_VIDEO_CAPS, NULL, NULL);
5008 TYPE_FIND_REGISTER (plugin, "video/x-h263", GST_RANK_SECONDARY,
5009 h263_video_type_find, "h263,263", H263_VIDEO_CAPS, NULL, NULL);
5010 TYPE_FIND_REGISTER (plugin, "video/x-h264", GST_RANK_PRIMARY,
5011 h264_video_type_find, "h264,x264,264", H264_VIDEO_CAPS, NULL, NULL);
5012 TYPE_FIND_REGISTER (plugin, "video/x-nuv", GST_RANK_SECONDARY, nuv_type_find,
5013 "nuv", NUV_CAPS, NULL, NULL);
5016 TYPE_FIND_REGISTER (plugin, "audio/x-m4a", GST_RANK_PRIMARY, m4a_type_find,
5017 "m4a", M4A_CAPS, NULL, NULL);
5018 TYPE_FIND_REGISTER (plugin, "application/x-3gp", GST_RANK_PRIMARY,
5019 q3gp_type_find, "3gp", Q3GP_CAPS, NULL, NULL);
5020 TYPE_FIND_REGISTER (plugin, "video/quicktime", GST_RANK_SECONDARY,
5021 qt_type_find, "mov", QT_CAPS, NULL, NULL);
5022 TYPE_FIND_REGISTER (plugin, "image/x-quicktime", GST_RANK_SECONDARY,
5023 qtif_type_find, "qif,qtif,qti", QTIF_CAPS, NULL, NULL);
5024 TYPE_FIND_REGISTER (plugin, "image/jp2", GST_RANK_PRIMARY,
5025 jp2_type_find, "jp2", JP2_CAPS, NULL, NULL);
5026 TYPE_FIND_REGISTER (plugin, "video/mj2", GST_RANK_PRIMARY,
5027 jp2_type_find, "mj2", MJ2_CAPS, NULL, NULL);
5029 TYPE_FIND_REGISTER (plugin, "text/html", GST_RANK_SECONDARY, html_type_find,
5030 "htm,html", HTML_CAPS, NULL, NULL);
5031 TYPE_FIND_REGISTER_START_WITH (plugin, "application/vnd.rn-realmedia",
5032 GST_RANK_SECONDARY, "ra,ram,rm,rmvb", ".RMF", 4, GST_TYPE_FIND_MAXIMUM);
5033 TYPE_FIND_REGISTER_START_WITH (plugin, "application/x-pn-realaudio",
5034 GST_RANK_SECONDARY, "ra,ram,rm,rmvb", ".ra\375", 4,
5035 GST_TYPE_FIND_MAXIMUM);
5036 TYPE_FIND_REGISTER (plugin, "application/x-shockwave-flash",
5037 GST_RANK_SECONDARY, swf_type_find, "swf,swfl", SWF_CAPS, NULL, NULL);
5038 TYPE_FIND_REGISTER_START_WITH (plugin, "video/x-flv", GST_RANK_SECONDARY,
5039 "flv", "FLV", 3, GST_TYPE_FIND_MAXIMUM);
5040 TYPE_FIND_REGISTER (plugin, "text/plain", GST_RANK_MARGINAL, utf8_type_find,
5041 "txt", UTF8_CAPS, NULL, NULL);
5042 TYPE_FIND_REGISTER (plugin, "text/utf-16", GST_RANK_MARGINAL, utf16_type_find,
5043 "txt", UTF16_CAPS, NULL, NULL);
5044 TYPE_FIND_REGISTER (plugin, "text/utf-32", GST_RANK_MARGINAL, utf32_type_find,
5045 "txt", UTF32_CAPS, NULL, NULL);
5046 TYPE_FIND_REGISTER (plugin, "text/uri-list", GST_RANK_MARGINAL, uri_type_find,
5047 "ram", URI_CAPS, NULL, NULL);
5048 TYPE_FIND_REGISTER (plugin, "application/x-hls", GST_RANK_MARGINAL,
5049 hls_type_find, "m3u8", HLS_CAPS, NULL, NULL);
5050 TYPE_FIND_REGISTER (plugin, "application/sdp", GST_RANK_SECONDARY,
5051 sdp_type_find, "sdp", SDP_CAPS, NULL, NULL);
5052 TYPE_FIND_REGISTER (plugin, "application/smil", GST_RANK_SECONDARY,
5053 smil_type_find, "smil", SMIL_CAPS, NULL, NULL);
5054 TYPE_FIND_REGISTER (plugin, "application/xml", GST_RANK_MARGINAL,
5055 xml_type_find, "xml", GENERIC_XML_CAPS, NULL, NULL);
5056 TYPE_FIND_REGISTER_RIFF (plugin, "audio/x-wav", GST_RANK_PRIMARY, "wav",
5058 TYPE_FIND_REGISTER (plugin, "audio/x-aiff", GST_RANK_SECONDARY,
5059 aiff_type_find, "aiff,aif,aifc", AIFF_CAPS, NULL, NULL);
5060 TYPE_FIND_REGISTER (plugin, "audio/x-svx", GST_RANK_SECONDARY, svx_type_find,
5061 "iff,svx", SVX_CAPS, NULL, NULL);
5062 TYPE_FIND_REGISTER (plugin, "audio/x-paris", GST_RANK_SECONDARY,
5063 paris_type_find, "paf", PARIS_CAPS, NULL, NULL);
5064 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-nist", GST_RANK_SECONDARY,
5065 "nist", "NIST", 4, GST_TYPE_FIND_MAXIMUM);
5066 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-voc", GST_RANK_SECONDARY,
5067 "voc", "Creative", 8, GST_TYPE_FIND_MAXIMUM);
5068 TYPE_FIND_REGISTER (plugin, "audio/x-sds", GST_RANK_SECONDARY, sds_type_find,
5069 "sds", SDS_CAPS, NULL, NULL);
5070 TYPE_FIND_REGISTER (plugin, "audio/x-ircam", GST_RANK_SECONDARY,
5071 ircam_type_find, "sf", IRCAM_CAPS, NULL, NULL);
5072 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-w64", GST_RANK_SECONDARY,
5073 "w64", "riff", 4, GST_TYPE_FIND_MAXIMUM);
5074 TYPE_FIND_REGISTER (plugin, "audio/x-shorten", GST_RANK_SECONDARY,
5075 shn_type_find, "shn", SHN_CAPS, NULL, NULL);
5076 TYPE_FIND_REGISTER (plugin, "application/x-ape", GST_RANK_SECONDARY,
5077 ape_type_find, "ape", APE_CAPS, NULL, NULL);
5078 TYPE_FIND_REGISTER (plugin, "image/jpeg", GST_RANK_PRIMARY + 15,
5079 jpeg_type_find, "jpg,jpe,jpeg", JPEG_CAPS, NULL, NULL);
5080 TYPE_FIND_REGISTER_START_WITH (plugin, "image/gif", GST_RANK_PRIMARY, "gif",
5081 "GIF8", 4, GST_TYPE_FIND_MAXIMUM);
5082 TYPE_FIND_REGISTER_START_WITH (plugin, "image/png", GST_RANK_PRIMARY + 14,
5083 "png", "\211PNG\015\012\032\012", 8, GST_TYPE_FIND_MAXIMUM);
5084 TYPE_FIND_REGISTER (plugin, "image/bmp", GST_RANK_PRIMARY, bmp_type_find,
5085 "bmp", BMP_CAPS, NULL, NULL);
5086 TYPE_FIND_REGISTER (plugin, "image/tiff", GST_RANK_PRIMARY, tiff_type_find,
5087 "tif,tiff", TIFF_CAPS, NULL, NULL);
5088 TYPE_FIND_REGISTER (plugin, "image/x-portable-pixmap", GST_RANK_SECONDARY,
5089 pnm_type_find, "pnm,ppm,pgm,pbm", PNM_CAPS, NULL, NULL);
5090 TYPE_FIND_REGISTER (plugin, "video/x-matroska", GST_RANK_PRIMARY,
5091 matroska_type_find, "mkv,mka,mk3d,webm", MATROSKA_CAPS, NULL, NULL);
5092 TYPE_FIND_REGISTER (plugin, "application/mxf", GST_RANK_PRIMARY,
5093 mxf_type_find, "mxf", MXF_CAPS, NULL, NULL);
5094 TYPE_FIND_REGISTER_START_WITH (plugin, "video/x-mve", GST_RANK_SECONDARY,
5095 "mve", "Interplay MVE File\032\000\032\000\000\001\063\021", 26,
5096 GST_TYPE_FIND_MAXIMUM);
5097 TYPE_FIND_REGISTER (plugin, "video/x-dv", GST_RANK_SECONDARY, dv_type_find,
5098 "dv,dif", DV_CAPS, NULL, NULL);
5099 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-amr-nb-sh", GST_RANK_PRIMARY,
5100 "amr", "#!AMR", 5, GST_TYPE_FIND_LIKELY);
5101 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-amr-wb-sh", GST_RANK_PRIMARY,
5102 "amr", "#!AMR-WB", 7, GST_TYPE_FIND_MAXIMUM);
5103 TYPE_FIND_REGISTER (plugin, "audio/iLBC-sh", GST_RANK_PRIMARY, ilbc_type_find,
5104 "ilbc", ILBC_CAPS, NULL, NULL);
5105 TYPE_FIND_REGISTER (plugin, "audio/x-sbc", GST_RANK_MARGINAL, sbc_type_find,
5106 "sbc", SBC_CAPS, NULL, NULL);
5107 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-sid", GST_RANK_MARGINAL,
5108 "sid", "PSID", 4, GST_TYPE_FIND_MAXIMUM);
5109 TYPE_FIND_REGISTER_START_WITH (plugin, "image/x-xcf", GST_RANK_SECONDARY,
5110 "xcf", "gimp xcf", 8, GST_TYPE_FIND_MAXIMUM);
5111 TYPE_FIND_REGISTER_START_WITH (plugin, "video/x-mng", GST_RANK_SECONDARY,
5112 "mng", "\212MNG\015\012\032\012", 8, GST_TYPE_FIND_MAXIMUM);
5113 TYPE_FIND_REGISTER_START_WITH (plugin, "image/x-jng", GST_RANK_SECONDARY,
5114 "jng", "\213JNG\015\012\032\012", 8, GST_TYPE_FIND_MAXIMUM);
5115 TYPE_FIND_REGISTER_START_WITH (plugin, "image/x-xpixmap", GST_RANK_SECONDARY,
5116 "xpm", "/* XPM */", 9, GST_TYPE_FIND_MAXIMUM);
5117 TYPE_FIND_REGISTER_START_WITH (plugin, "image/x-sun-raster",
5118 GST_RANK_SECONDARY, "ras", "\131\246\152\225", 4, GST_TYPE_FIND_MAXIMUM);
5119 TYPE_FIND_REGISTER_START_WITH (plugin, "application/x-bzip",
5120 GST_RANK_SECONDARY, "bz2", "BZh", 3, GST_TYPE_FIND_LIKELY);
5121 TYPE_FIND_REGISTER_START_WITH (plugin, "application/x-gzip",
5122 GST_RANK_SECONDARY, "gz", "\037\213", 2, GST_TYPE_FIND_LIKELY);
5123 TYPE_FIND_REGISTER_START_WITH (plugin, "application/zip", GST_RANK_SECONDARY,
5124 "zip", "PK\003\004", 4, GST_TYPE_FIND_LIKELY);
5125 TYPE_FIND_REGISTER_START_WITH (plugin, "application/x-compress",
5126 GST_RANK_SECONDARY, "Z", "\037\235", 2, GST_TYPE_FIND_LIKELY);
5127 TYPE_FIND_REGISTER (plugin, "subtitle/x-kate", GST_RANK_MARGINAL,
5128 kate_type_find, NULL, NULL, NULL, NULL);
5129 TYPE_FIND_REGISTER (plugin, "audio/x-flac", GST_RANK_PRIMARY, flac_type_find,
5130 "flac", FLAC_CAPS, NULL, NULL);
5131 TYPE_FIND_REGISTER (plugin, "audio/x-vorbis", GST_RANK_PRIMARY,
5132 vorbis_type_find, NULL, VORBIS_CAPS, NULL, NULL);
5133 TYPE_FIND_REGISTER (plugin, "video/x-theora", GST_RANK_PRIMARY,
5134 theora_type_find, NULL, THEORA_CAPS, NULL, NULL);
5135 TYPE_FIND_REGISTER (plugin, "application/x-ogm-video", GST_RANK_PRIMARY,
5136 ogmvideo_type_find, NULL, OGMVIDEO_CAPS, NULL, NULL);
5137 TYPE_FIND_REGISTER (plugin, "application/x-ogm-audio", GST_RANK_PRIMARY,
5138 ogmaudio_type_find, NULL, OGMAUDIO_CAPS, NULL, NULL);
5139 TYPE_FIND_REGISTER (plugin, "application/x-ogm-text", GST_RANK_PRIMARY,
5140 ogmtext_type_find, NULL, OGMTEXT_CAPS, NULL, NULL);
5141 TYPE_FIND_REGISTER (plugin, "audio/x-speex", GST_RANK_PRIMARY,
5142 speex_type_find, NULL, SPEEX_CAPS, NULL, NULL);
5143 TYPE_FIND_REGISTER (plugin, "audio/x-celt", GST_RANK_PRIMARY, celt_type_find,
5144 NULL, CELT_CAPS, NULL, NULL);
5145 TYPE_FIND_REGISTER (plugin, "application/x-ogg-skeleton", GST_RANK_PRIMARY,
5146 oggskel_type_find, NULL, OGG_SKELETON_CAPS, NULL, NULL);
5147 TYPE_FIND_REGISTER (plugin, "text/x-cmml", GST_RANK_PRIMARY, cmml_type_find,
5148 NULL, CMML_CAPS, NULL, NULL);
5149 TYPE_FIND_REGISTER_START_WITH (plugin, "application/x-executable",
5150 GST_RANK_MARGINAL, NULL, "\177ELF", 4, GST_TYPE_FIND_MAXIMUM);
5151 TYPE_FIND_REGISTER (plugin, "audio/aac", GST_RANK_SECONDARY, aac_type_find,
5152 "aac,adts,adif,loas", AAC_CAPS, NULL, NULL);
5153 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-spc", GST_RANK_SECONDARY,
5154 "spc", "SNES-SPC700 Sound File Data", 27, GST_TYPE_FIND_MAXIMUM);
5155 TYPE_FIND_REGISTER (plugin, "audio/x-wavpack", GST_RANK_SECONDARY,
5156 wavpack_type_find, "wv,wvp", WAVPACK_CAPS, NULL, NULL);
5157 TYPE_FIND_REGISTER (plugin, "audio/x-wavpack-correction", GST_RANK_SECONDARY,
5158 wavpack_type_find, "wvc", WAVPACK_CORRECTION_CAPS, NULL, NULL);
5159 TYPE_FIND_REGISTER (plugin, "application/postscript", GST_RANK_SECONDARY,
5160 postscript_type_find, "ps", POSTSCRIPT_CAPS, NULL, NULL);
5161 TYPE_FIND_REGISTER (plugin, "image/svg+xml", GST_RANK_SECONDARY,
5162 svg_type_find, "svg", SVG_CAPS, NULL, NULL);
5163 TYPE_FIND_REGISTER_START_WITH (plugin, "application/x-rar",
5164 GST_RANK_SECONDARY, "rar", "Rar!", 4, GST_TYPE_FIND_LIKELY);
5165 TYPE_FIND_REGISTER (plugin, "application/x-tar", GST_RANK_SECONDARY,
5166 tar_type_find, "tar", TAR_CAPS, NULL, NULL);
5167 TYPE_FIND_REGISTER (plugin, "application/x-ar", GST_RANK_SECONDARY,
5168 ar_type_find, "a", AR_CAPS, NULL, NULL);
5169 TYPE_FIND_REGISTER (plugin, "application/x-ms-dos-executable",
5170 GST_RANK_SECONDARY, msdos_type_find, "dll,exe,ocx,sys,scr,msstyles,cpl",
5171 MSDOS_CAPS, NULL, NULL);
5172 TYPE_FIND_REGISTER (plugin, "video/x-dirac", GST_RANK_PRIMARY,
5173 dirac_type_find, NULL, DIRAC_CAPS, NULL, NULL);
5174 TYPE_FIND_REGISTER (plugin, "multipart/x-mixed-replace", GST_RANK_SECONDARY,
5175 multipart_type_find, NULL, MULTIPART_CAPS, NULL, NULL);
5176 TYPE_FIND_REGISTER (plugin, "application/x-mmsh", GST_RANK_SECONDARY,
5177 mmsh_type_find, NULL, MMSH_CAPS, NULL, NULL);
5178 TYPE_FIND_REGISTER (plugin, "video/vivo", GST_RANK_SECONDARY, vivo_type_find,
5179 "viv", VIVO_CAPS, NULL, NULL);
5180 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-nsf", GST_RANK_SECONDARY,
5181 "nsf", "NESM\x1a", 5, GST_TYPE_FIND_MAXIMUM);
5182 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-gym", GST_RANK_SECONDARY,
5183 "gym", "GYMX", 4, GST_TYPE_FIND_MAXIMUM);
5184 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-ay", GST_RANK_SECONDARY, "ay",
5185 "ZXAYEMUL", 8, GST_TYPE_FIND_MAXIMUM);
5186 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-gbs", GST_RANK_SECONDARY,
5187 "gbs", "GBS\x01", 4, GST_TYPE_FIND_MAXIMUM);
5188 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-vgm", GST_RANK_SECONDARY,
5189 "vgm", "Vgm\x20", 4, GST_TYPE_FIND_MAXIMUM);
5190 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-sap", GST_RANK_SECONDARY,
5191 "sap", "SAP\x0d\x0a" "AUTHOR\x20", 12, GST_TYPE_FIND_MAXIMUM);
5192 TYPE_FIND_REGISTER_START_WITH (plugin, "video/x-ivf", GST_RANK_SECONDARY,
5193 "ivf", "DKIF", 4, GST_TYPE_FIND_NEARLY_CERTAIN);
5194 TYPE_FIND_REGISTER_START_WITH (plugin, "audio/x-kss", GST_RANK_SECONDARY,
5195 "kss", "KSSX\0", 5, GST_TYPE_FIND_MAXIMUM);
5196 TYPE_FIND_REGISTER_START_WITH (plugin, "application/pdf", GST_RANK_SECONDARY,
5197 "pdf", "%PDF-", 5, GST_TYPE_FIND_LIKELY);
5198 TYPE_FIND_REGISTER_START_WITH (plugin, "application/msword",
5199 GST_RANK_SECONDARY, "doc", "\320\317\021\340\241\261\032\341", 8,
5200 GST_TYPE_FIND_LIKELY);
5201 /* Mac OS X .DS_Store files tend to be taken for video/mpeg */
5202 TYPE_FIND_REGISTER_START_WITH (plugin, "application/octet-stream",
5203 GST_RANK_SECONDARY, "DS_Store", "\000\000\000\001Bud1", 8,
5204 GST_TYPE_FIND_LIKELY);
5205 TYPE_FIND_REGISTER_START_WITH (plugin, "image/vnd.adobe.photoshop",
5206 GST_RANK_SECONDARY, "psd", "8BPS\000\001\000\000\000\000", 10,
5207 GST_TYPE_FIND_LIKELY);
5208 TYPE_FIND_REGISTER (plugin, "image/vnd.wap.wbmp", GST_RANK_MARGINAL,
5209 wbmp_typefind, NULL, NULL, NULL, NULL);
5210 TYPE_FIND_REGISTER_START_WITH (plugin, "application/x-yuv4mpeg",
5211 GST_RANK_SECONDARY, "y4m", "YUV4MPEG2 ", 10, GST_TYPE_FIND_LIKELY);
5212 TYPE_FIND_REGISTER (plugin, "image/x-icon", GST_RANK_MARGINAL,
5213 windows_icon_typefind, NULL, NULL, NULL, NULL);
5216 TYPE_FIND_REGISTER (plugin, "xdgmime-base", GST_RANK_MARGINAL,
5217 xdgmime_typefind, NULL, NULL, NULL, NULL);
5220 TYPE_FIND_REGISTER (plugin, "image/x-degas", GST_RANK_MARGINAL,
5221 degas_type_find, NULL, NULL, NULL, NULL);
5222 TYPE_FIND_REGISTER (plugin, "application/octet-stream", GST_RANK_MARGINAL,
5223 dvdiso_type_find, NULL, NULL, NULL, NULL);
5225 TYPE_FIND_REGISTER (plugin, "application/x-ssa", GST_RANK_SECONDARY,
5226 ssa_type_find, "ssa,ass", NULL, NULL, NULL);
5231 GST_PLUGIN_DEFINE (GST_VERSION_MAJOR,
5234 "default typefind functions",
5235 plugin_init, VERSION, GST_LICENSE, GST_PACKAGE_NAME, GST_PACKAGE_ORIGIN)