• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 // Copyright (c) 2013 The Chromium Authors. All rights reserved.
2 // Use of this source code is governed by a BSD-style license that can be
3 // found in the LICENSE file.
4 
5 #include "media/base/container_names.h"
6 
7 #include <cctype>
8 #include <limits>
9 
10 #include "base/basictypes.h"
11 #include "base/logging.h"
12 #include "media/base/bit_reader.h"
13 
14 namespace media {
15 
16 namespace container_names {
17 
18 #define TAG(a, b, c, d) \
19     ((static_cast<uint8>(a) << 24) | (static_cast<uint8>(b) << 16) | \
20      (static_cast<uint8>(c) << 8) | (static_cast<uint8>(d)))
21 
22 #define RCHECK(x)     \
23     do {              \
24       if (!(x))       \
25         return false; \
26     } while (0)
27 
28 #define UTF8_BYTE_ORDER_MARK "\xef\xbb\xbf"
29 
30 // Helper function to read 2 bytes (16 bits, big endian) from a buffer.
Read16(const uint8 * p)31 static int Read16(const uint8* p) {
32   return p[0] << 8 | p[1];
33 }
34 
35 // Helper function to read 3 bytes (24 bits, big endian) from a buffer.
Read24(const uint8 * p)36 static uint32 Read24(const uint8* p) {
37   return p[0] << 16 | p[1] << 8 | p[2];
38 }
39 
40 // Helper function to read 4 bytes (32 bits, big endian) from a buffer.
Read32(const uint8 * p)41 static uint32 Read32(const uint8* p) {
42   return p[0] << 24 | p[1] << 16 | p[2] << 8 | p[3];
43 }
44 
45 // Helper function to read 4 bytes (32 bits, little endian) from a buffer.
Read32LE(const uint8 * p)46 static uint32 Read32LE(const uint8* p) {
47   return p[3] << 24 | p[2] << 16 | p[1] << 8 | p[0];
48 }
49 
50 // Helper function to do buffer comparisons with a string without going off the
51 // end of the buffer.
StartsWith(const uint8 * buffer,size_t buffer_size,const char * prefix)52 static bool StartsWith(const uint8* buffer,
53                        size_t buffer_size,
54                        const char* prefix) {
55   size_t prefix_size = strlen(prefix);
56   return (prefix_size <= buffer_size &&
57           memcmp(buffer, prefix, prefix_size) == 0);
58 }
59 
60 // Helper function to do buffer comparisons with another buffer (to allow for
61 // embedded \0 in the comparison) without going off the end of the buffer.
StartsWith(const uint8 * buffer,size_t buffer_size,const uint8 * prefix,size_t prefix_size)62 static bool StartsWith(const uint8* buffer,
63                        size_t buffer_size,
64                        const uint8* prefix,
65                        size_t prefix_size) {
66   return (prefix_size <= buffer_size &&
67           memcmp(buffer, prefix, prefix_size) == 0);
68 }
69 
70 // Helper function to read up to 64 bits from a bit stream.
ReadBits(BitReader * reader,int num_bits)71 static uint64 ReadBits(BitReader* reader, int num_bits) {
72   DCHECK_GE(reader->bits_available(), num_bits);
73   DCHECK((num_bits > 0) && (num_bits <= 64));
74   uint64 value;
75   reader->ReadBits(num_bits, &value);
76   return value;
77 }
78 
79 const int kAc3FrameSizeTable[38][3] = {
80   { 128, 138, 192 }, { 128, 140, 192 }, { 160, 174, 240 }, { 160, 176, 240 },
81   { 192, 208, 288 }, { 192, 210, 288 }, { 224, 242, 336 }, { 224, 244, 336 },
82   { 256, 278, 384 }, { 256, 280, 384 }, { 320, 348, 480 }, { 320, 350, 480 },
83   { 384, 416, 576 }, { 384, 418, 576 }, { 448, 486, 672 }, { 448, 488, 672 },
84   { 512, 556, 768 }, { 512, 558, 768 }, { 640, 696, 960 }, { 640, 698, 960 },
85   { 768, 834, 1152 }, { 768, 836, 1152 }, { 896, 974, 1344 },
86   { 896, 976, 1344 }, { 1024, 1114, 1536 }, { 1024, 1116, 1536 },
87   { 1280, 1392, 1920 }, { 1280, 1394, 1920 }, { 1536, 1670, 2304 },
88   { 1536, 1672, 2304 }, { 1792, 1950, 2688 }, { 1792, 1952, 2688 },
89   { 2048, 2228, 3072 }, { 2048, 2230, 3072 }, { 2304, 2506, 3456 },
90   { 2304, 2508, 3456 }, { 2560, 2768, 3840 }, { 2560, 2770, 3840 }
91 };
92 
93 // Checks for an ADTS AAC container.
CheckAac(const uint8 * buffer,int buffer_size)94 static bool CheckAac(const uint8* buffer, int buffer_size) {
95   // Audio Data Transport Stream (ADTS) header is 7 or 9 bytes
96   // (from http://wiki.multimedia.cx/index.php?title=ADTS)
97   RCHECK(buffer_size > 6);
98 
99   int offset = 0;
100   while (offset + 6 < buffer_size) {
101     BitReader reader(buffer + offset, 6);
102 
103     // Syncword must be 0xfff.
104     RCHECK(ReadBits(&reader, 12) == 0xfff);
105 
106     // Skip MPEG version.
107     reader.SkipBits(1);
108 
109     // Layer is always 0.
110     RCHECK(ReadBits(&reader, 2) == 0);
111 
112     // Skip protection + profile.
113     reader.SkipBits(1 + 2);
114 
115     // Check sampling frequency index.
116     RCHECK(ReadBits(&reader, 4) != 15);  // Forbidden.
117 
118     // Skip private stream, channel configuration, originality, home,
119     // copyrighted stream, and copyright_start.
120     reader.SkipBits(1 + 3 + 1 + 1 + 1 + 1);
121 
122     // Get frame length (includes header).
123     int size = ReadBits(&reader, 13);
124     RCHECK(size > 0);
125     offset += size;
126   }
127   return true;
128 }
129 
130 const uint16 kAc3SyncWord = 0x0b77;
131 
132 // Checks for an AC3 container.
CheckAc3(const uint8 * buffer,int buffer_size)133 static bool CheckAc3(const uint8* buffer, int buffer_size) {
134   // Reference: ATSC Standard: Digital Audio Compression (AC-3, E-AC-3)
135   //            Doc. A/52:2012
136   // (http://www.atsc.org/cms/standards/A52-2012(12-17).pdf)
137 
138   // AC3 container looks like syncinfo | bsi | audblk * 6 | aux | check.
139   RCHECK(buffer_size > 6);
140 
141   int offset = 0;
142   while (offset + 6 < buffer_size) {
143     BitReader reader(buffer + offset, 6);
144 
145     // Check syncinfo.
146     RCHECK(ReadBits(&reader, 16) == kAc3SyncWord);
147 
148     // Skip crc1.
149     reader.SkipBits(16);
150 
151     // Verify fscod.
152     int sample_rate_code = ReadBits(&reader, 2);
153     RCHECK(sample_rate_code != 3);  // Reserved.
154 
155     // Verify frmsizecod.
156     int frame_size_code = ReadBits(&reader, 6);
157     RCHECK(frame_size_code < 38);  // Undefined.
158 
159     // Verify bsid.
160     RCHECK(ReadBits(&reader, 5) < 10);  // Normally 8 or 6, 16 used by EAC3.
161 
162     offset += kAc3FrameSizeTable[frame_size_code][sample_rate_code];
163   }
164   return true;
165 }
166 
167 // Checks for an EAC3 container (very similar to AC3)
CheckEac3(const uint8 * buffer,int buffer_size)168 static bool CheckEac3(const uint8* buffer, int buffer_size) {
169   // Reference: ATSC Standard: Digital Audio Compression (AC-3, E-AC-3)
170   //            Doc. A/52:2012
171   // (http://www.atsc.org/cms/standards/A52-2012(12-17).pdf)
172 
173   // EAC3 container looks like syncinfo | bsi | audfrm | audblk* | aux | check.
174   RCHECK(buffer_size > 6);
175 
176   int offset = 0;
177   while (offset + 6 < buffer_size) {
178     BitReader reader(buffer + offset, 6);
179 
180     // Check syncinfo.
181     RCHECK(ReadBits(&reader, 16) == kAc3SyncWord);
182 
183     // Verify strmtyp.
184     RCHECK(ReadBits(&reader, 2) != 3);
185 
186     // Skip substreamid.
187     reader.SkipBits(3);
188 
189     // Get frmsize. Include syncinfo size and convert to bytes.
190     int frame_size = (ReadBits(&reader, 11) + 1) * 2;
191     RCHECK(frame_size >= 7);
192 
193     // Skip fscod, fscod2, acmod, and lfeon.
194     reader.SkipBits(2 + 2 + 3 + 1);
195 
196     // Verify bsid.
197     int bit_stream_id = ReadBits(&reader, 5);
198     RCHECK(bit_stream_id >= 11 && bit_stream_id <= 16);
199 
200     offset += frame_size;
201   }
202   return true;
203 }
204 
205 // Additional checks for a BINK container.
CheckBink(const uint8 * buffer,int buffer_size)206 static bool CheckBink(const uint8* buffer, int buffer_size) {
207   // Reference: http://wiki.multimedia.cx/index.php?title=Bink_Container
208   RCHECK(buffer_size >= 44);
209 
210   // Verify number of frames specified.
211   RCHECK(Read32LE(buffer + 8) > 0);
212 
213   // Verify width in range.
214   int width = Read32LE(buffer + 20);
215   RCHECK(width > 0 && width <= 32767);
216 
217   // Verify height in range.
218   int height = Read32LE(buffer + 24);
219   RCHECK(height > 0 && height <= 32767);
220 
221   // Verify frames per second specified.
222   RCHECK(Read32LE(buffer + 28) > 0);
223 
224   // Verify video frames per second specified.
225   RCHECK(Read32LE(buffer + 32) > 0);
226 
227   // Number of audio tracks must be 256 or less.
228   return (Read32LE(buffer + 40) <= 256);
229 }
230 
231 // Additional checks for a CAF container.
CheckCaf(const uint8 * buffer,int buffer_size)232 static bool CheckCaf(const uint8* buffer, int buffer_size) {
233   // Reference: Apple Core Audio Format Specification 1.0
234   // (https://developer.apple.com/library/mac/#documentation/MusicAudio/Reference/CAFSpec/CAF_spec/CAF_spec.html)
235   RCHECK(buffer_size >= 52);
236   BitReader reader(buffer, buffer_size);
237 
238   // mFileType should be "caff".
239   RCHECK(ReadBits(&reader, 32) == TAG('c', 'a', 'f', 'f'));
240 
241   // mFileVersion should be 1.
242   RCHECK(ReadBits(&reader, 16) == 1);
243 
244   // Skip mFileFlags.
245   reader.SkipBits(16);
246 
247   // First chunk should be Audio Description chunk, size 32l.
248   RCHECK(ReadBits(&reader, 32) == TAG('d', 'e', 's', 'c'));
249   RCHECK(ReadBits(&reader, 64) == 32);
250 
251   // CAFAudioFormat.mSampleRate(float64) not 0
252   RCHECK(ReadBits(&reader, 64) != 0);
253 
254   // CAFAudioFormat.mFormatID not 0
255   RCHECK(ReadBits(&reader, 32) != 0);
256 
257   // Skip CAFAudioFormat.mBytesPerPacket and mFramesPerPacket.
258   reader.SkipBits(32 + 32);
259 
260   // CAFAudioFormat.mChannelsPerFrame not 0
261   RCHECK(ReadBits(&reader, 32) != 0);
262   return true;
263 }
264 
265 static bool kSamplingFrequencyValid[16] = { false, true, true, true, false,
266                                             false, true, true, true, false,
267                                             false, true, true, true, false,
268                                             false };
269 static bool kExtAudioIdValid[8] = { true, false, true, false, false, false,
270                                     true, false };
271 
272 // Additional checks for a DTS container.
CheckDts(const uint8 * buffer,int buffer_size)273 static bool CheckDts(const uint8* buffer, int buffer_size) {
274   // Reference: ETSI TS 102 114 V1.3.1 (2011-08)
275   // (http://www.etsi.org/deliver/etsi_ts/102100_102199/102114/01.03.01_60/ts_102114v010301p.pdf)
276   RCHECK(buffer_size > 11);
277 
278   int offset = 0;
279   while (offset + 11 < buffer_size) {
280     BitReader reader(buffer + offset, 11);
281 
282     // Verify sync word.
283     RCHECK(ReadBits(&reader, 32) == 0x7ffe8001);
284 
285     // Skip frame type and deficit sample count.
286     reader.SkipBits(1 + 5);
287 
288     // Verify CRC present flag.
289     RCHECK(ReadBits(&reader, 1) == 0);  // CPF must be 0.
290 
291     // Verify number of PCM sample blocks.
292     RCHECK(ReadBits(&reader, 7) >= 5);
293 
294     // Verify primary frame byte size.
295     int frame_size = ReadBits(&reader, 14);
296     RCHECK(frame_size >= 95);
297 
298     // Skip audio channel arrangement.
299     reader.SkipBits(6);
300 
301     // Verify core audio sampling frequency is an allowed value.
302     RCHECK(kSamplingFrequencyValid[ReadBits(&reader, 4)]);
303 
304     // Verify transmission bit rate is valid.
305     RCHECK(ReadBits(&reader, 5) <= 25);
306 
307     // Verify reserved field is 0.
308     RCHECK(ReadBits(&reader, 1) == 0);
309 
310     // Skip dynamic range flag, time stamp flag, auxiliary data flag, and HDCD.
311     reader.SkipBits(1 + 1 + 1 + 1);
312 
313     // Verify extension audio descriptor flag is an allowed value.
314     RCHECK(kExtAudioIdValid[ReadBits(&reader, 3)]);
315 
316     // Skip extended coding flag and audio sync word insertion flag.
317     reader.SkipBits(1 + 1);
318 
319     // Verify low frequency effects flag is an allowed value.
320     RCHECK(ReadBits(&reader, 2) != 3);
321 
322     offset += frame_size + 1;
323   }
324   return true;
325 }
326 
327 // Checks for a DV container.
CheckDV(const uint8 * buffer,int buffer_size)328 static bool CheckDV(const uint8* buffer, int buffer_size) {
329   // Reference: SMPTE 314M (Annex A has differences with IEC 61834).
330   // (http://standards.smpte.org/content/978-1-61482-454-1/st-314-2005/SEC1.body.pdf)
331   RCHECK(buffer_size > 11);
332 
333   int offset = 0;
334   int current_sequence_number = -1;
335   int last_block_number[6];
336   while (offset + 11 < buffer_size) {
337     BitReader reader(buffer + offset, 11);
338 
339     // Decode ID data. Sections 5, 6, and 7 are reserved.
340     int section = ReadBits(&reader, 3);
341     RCHECK(section < 5);
342 
343     // Next bit must be 1.
344     RCHECK(ReadBits(&reader, 1) == 1);
345 
346     // Skip arbitrary bits.
347     reader.SkipBits(4);
348 
349     int sequence_number = ReadBits(&reader, 4);
350 
351     // Skip FSC.
352     reader.SkipBits(1);
353 
354     // Next 3 bits must be 1.
355     RCHECK(ReadBits(&reader, 3) == 7);
356 
357     int block_number = ReadBits(&reader, 8);
358 
359     if (section == 0) {  // Header.
360       // Validate the reserved bits in the next 8 bytes.
361       reader.SkipBits(1);
362       RCHECK(ReadBits(&reader, 1) == 0);
363       RCHECK(ReadBits(&reader, 11) == 0x7ff);
364       reader.SkipBits(4);
365       RCHECK(ReadBits(&reader, 4) == 0xf);
366       reader.SkipBits(4);
367       RCHECK(ReadBits(&reader, 4) == 0xf);
368       reader.SkipBits(4);
369       RCHECK(ReadBits(&reader, 4) == 0xf);
370       reader.SkipBits(3);
371       RCHECK(ReadBits(&reader, 24) == 0xffffff);
372       current_sequence_number = sequence_number;
373       for (size_t i = 0; i < arraysize(last_block_number); ++i)
374         last_block_number[i] = -1;
375     } else {
376       // Sequence number must match (this will also fail if no header seen).
377       RCHECK(sequence_number == current_sequence_number);
378       // Block number should be increasing.
379       RCHECK(block_number > last_block_number[section]);
380       last_block_number[section] = block_number;
381     }
382 
383     // Move to next block.
384     offset += 80;
385   }
386   return true;
387 }
388 
389 
390 // Checks for a GSM container.
CheckGsm(const uint8 * buffer,int buffer_size)391 static bool CheckGsm(const uint8* buffer, int buffer_size) {
392   // Reference: ETSI EN 300 961 V8.1.1
393   // (http://www.etsi.org/deliver/etsi_en/300900_300999/300961/08.01.01_60/en_300961v080101p.pdf)
394   // also http://tools.ietf.org/html/rfc3551#page-24
395   // GSM files have a 33 byte block, only first 4 bits are fixed.
396   RCHECK(buffer_size >= 1024);  // Need enough data to do a decent check.
397 
398   int offset = 0;
399   while (offset < buffer_size) {
400     // First 4 bits of each block are xD.
401     RCHECK((buffer[offset] & 0xf0) == 0xd0);
402     offset += 33;
403   }
404   return true;
405 }
406 
407 // Advance to the first set of |num_bits| bits that match |start_code|. |offset|
408 // is the current location in the buffer, and is updated. |bytes_needed| is the
409 // number of bytes that must remain in the buffer when |start_code| is found.
410 // Returns true if start_code found (and enough space in the buffer after it),
411 // false otherwise.
AdvanceToStartCode(const uint8 * buffer,int buffer_size,int * offset,int bytes_needed,int num_bits,uint32 start_code)412 static bool AdvanceToStartCode(const uint8* buffer,
413                                int buffer_size,
414                                int* offset,
415                                int bytes_needed,
416                                int num_bits,
417                                uint32 start_code) {
418   DCHECK_GE(bytes_needed, 3);
419   DCHECK_LE(num_bits, 24);  // Only supports up to 24 bits.
420 
421   // Create a mask to isolate |num_bits| bits, once shifted over.
422   uint32 bits_to_shift = 24 - num_bits;
423   uint32 mask = (1 << num_bits) - 1;
424   while (*offset + bytes_needed < buffer_size) {
425     uint32 next = Read24(buffer + *offset);
426     if (((next >> bits_to_shift) & mask) == start_code)
427       return true;
428     ++(*offset);
429   }
430   return false;
431 }
432 
433 // Checks for an H.261 container.
CheckH261(const uint8 * buffer,int buffer_size)434 static bool CheckH261(const uint8* buffer, int buffer_size) {
435   // Reference: ITU-T Recommendation H.261 (03/1993)
436   // (http://www.itu.int/rec/T-REC-H.261-199303-I/en)
437   RCHECK(buffer_size > 16);
438 
439   int offset = 0;
440   bool seen_start_code = false;
441   while (true) {
442     // Advance to picture_start_code, if there is one.
443     if (!AdvanceToStartCode(buffer, buffer_size, &offset, 4, 20, 0x10)) {
444       // No start code found (or off end of buffer), so success if
445       // there was at least one valid header.
446       return seen_start_code;
447     }
448 
449     // Now verify the block. AdvanceToStartCode() made sure that there are
450     // at least 4 bytes remaining in the buffer.
451     BitReader reader(buffer + offset, buffer_size - offset);
452     RCHECK(ReadBits(&reader, 20) == 0x10);
453 
454     // Skip the temporal reference and PTYPE.
455     reader.SkipBits(5 + 6);
456 
457     // Skip any extra insertion information. Since this is open-ended, if we run
458     // out of bits assume that the buffer is correctly formatted.
459     int extra = ReadBits(&reader, 1);
460     while (extra == 1) {
461       if (!reader.SkipBits(8))
462         return seen_start_code;
463       if (!reader.ReadBits(1, &extra))
464         return seen_start_code;
465     }
466 
467     // Next should be a Group of Blocks start code. Again, if we run out of
468     // bits, then assume that the buffer up to here is correct, and the buffer
469     // just happened to end in the middle of a header.
470     int next;
471     if (!reader.ReadBits(16, &next))
472       return seen_start_code;
473     RCHECK(next == 1);
474 
475     // Move to the next block.
476     seen_start_code = true;
477     offset += 4;
478   }
479 }
480 
481 // Checks for an H.263 container.
CheckH263(const uint8 * buffer,int buffer_size)482 static bool CheckH263(const uint8* buffer, int buffer_size) {
483   // Reference: ITU-T Recommendation H.263 (01/2005)
484   // (http://www.itu.int/rec/T-REC-H.263-200501-I/en)
485   // header is PSC(22b) + TR(8b) + PTYPE(8+b).
486   RCHECK(buffer_size > 16);
487 
488   int offset = 0;
489   bool seen_start_code = false;
490   while (true) {
491     // Advance to picture_start_code, if there is one.
492     if (!AdvanceToStartCode(buffer, buffer_size, &offset, 9, 22, 0x20)) {
493       // No start code found (or off end of buffer), so success if
494       // there was at least one valid header.
495       return seen_start_code;
496     }
497 
498     // Now verify the block. AdvanceToStartCode() made sure that there are
499     // at least 9 bytes remaining in the buffer.
500     BitReader reader(buffer + offset, 9);
501     RCHECK(ReadBits(&reader, 22) == 0x20);
502 
503     // Skip the temporal reference.
504     reader.SkipBits(8);
505 
506     // Verify that the first 2 bits of PTYPE are 10b.
507     RCHECK(ReadBits(&reader, 2) == 2);
508 
509     // Skip the split screen indicator, document camera indicator, and full
510     // picture freeze release.
511     reader.SkipBits(1 + 1 + 1);
512 
513     // Verify Source Format.
514     int format = ReadBits(&reader, 3);
515     RCHECK(format != 0 && format != 6);  // Forbidden or reserved.
516 
517     if (format == 7) {
518       // Verify full extended PTYPE.
519       int ufep = ReadBits(&reader, 3);
520       if (ufep == 1) {
521         // Verify the optional part of PLUSPTYPE.
522         format = ReadBits(&reader, 3);
523         RCHECK(format != 0 && format != 7);  // Reserved.
524         reader.SkipBits(11);
525         // Next 4 bits should be b1000.
526         RCHECK(ReadBits(&reader, 4) == 8);  // Not allowed.
527       } else {
528         RCHECK(ufep == 0);  // Only 0 and 1 allowed.
529       }
530 
531       // Verify picture type code is not a reserved value.
532       int picture_type_code = ReadBits(&reader, 3);
533       RCHECK(picture_type_code != 6 && picture_type_code != 7);  // Reserved.
534 
535       // Skip picture resampling mode, reduced resolution mode,
536       // and rounding type.
537       reader.SkipBits(1 + 1 + 1);
538 
539       // Next 3 bits should be b001.
540       RCHECK(ReadBits(&reader, 3) == 1);  // Not allowed.
541     }
542 
543     // Move to the next block.
544     seen_start_code = true;
545     offset += 9;
546   }
547 }
548 
549 // Checks for an H.264 container.
CheckH264(const uint8 * buffer,int buffer_size)550 static bool CheckH264(const uint8* buffer, int buffer_size) {
551   // Reference: ITU-T Recommendation H.264 (01/2012)
552   // (http://www.itu.int/rec/T-REC-H.264)
553   // Section B.1: Byte stream NAL unit syntax and semantics.
554   RCHECK(buffer_size > 4);
555 
556   int offset = 0;
557   int parameter_count = 0;
558   while (true) {
559     // Advance to picture_start_code, if there is one.
560     if (!AdvanceToStartCode(buffer, buffer_size, &offset, 4, 24, 1)) {
561       // No start code found (or off end of buffer), so success if
562       // there was at least one valid header.
563       return parameter_count > 0;
564     }
565 
566     // Now verify the block. AdvanceToStartCode() made sure that there are
567     // at least 4 bytes remaining in the buffer.
568     BitReader reader(buffer + offset, 4);
569     RCHECK(ReadBits(&reader, 24) == 1);
570 
571     // Verify forbidden_zero_bit.
572     RCHECK(ReadBits(&reader, 1) == 0);
573 
574     // Extract nal_ref_idc and nal_unit_type.
575     int nal_ref_idc = ReadBits(&reader, 2);
576     int nal_unit_type = ReadBits(&reader, 5);
577 
578     switch (nal_unit_type) {
579       case 5:  // Coded slice of an IDR picture.
580         RCHECK(nal_ref_idc != 0);
581         break;
582       case 6:   // Supplemental enhancement information (SEI).
583       case 9:   // Access unit delimiter.
584       case 10:  // End of sequence.
585       case 11:  // End of stream.
586       case 12:  // Filler data.
587         RCHECK(nal_ref_idc == 0);
588         break;
589       case 7:  // Sequence parameter set.
590       case 8:  // Picture parameter set.
591         ++parameter_count;
592         break;
593     }
594 
595     // Skip the current start_code_prefix and move to the next.
596     offset += 4;
597   }
598 }
599 
600 static const char kHlsSignature[] = "#EXTM3U";
601 static const char kHls1[] = "#EXT-X-STREAM-INF:";
602 static const char kHls2[] = "#EXT-X-TARGETDURATION:";
603 static const char kHls3[] = "#EXT-X-MEDIA-SEQUENCE:";
604 
605 // Additional checks for a HLS container.
CheckHls(const uint8 * buffer,int buffer_size)606 static bool CheckHls(const uint8* buffer, int buffer_size) {
607   // HLS is simply a play list used for Apple HTTP Live Streaming.
608   // Reference: Apple HTTP Live Streaming Overview
609   // (http://goo.gl/MIwxj)
610 
611   if (StartsWith(buffer, buffer_size, kHlsSignature)) {
612     // Need to find "#EXT-X-STREAM-INF:", "#EXT-X-TARGETDURATION:", or
613     // "#EXT-X-MEDIA-SEQUENCE:" somewhere in the buffer. Other playlists (like
614     // WinAmp) only have additional lines with #EXTINF
615     // (http://en.wikipedia.org/wiki/M3U).
616     int offset = strlen(kHlsSignature);
617     while (offset < buffer_size) {
618       if (buffer[offset] == '#') {
619         if (StartsWith(buffer + offset, buffer_size - offset, kHls1) ||
620             StartsWith(buffer + offset, buffer_size - offset, kHls2) ||
621             StartsWith(buffer + offset, buffer_size - offset, kHls3)) {
622           return true;
623         }
624       }
625       ++offset;
626     }
627   }
628   return false;
629 }
630 
631 // Checks for a MJPEG stream.
CheckMJpeg(const uint8 * buffer,int buffer_size)632 static bool CheckMJpeg(const uint8* buffer, int buffer_size) {
633   // Reference: ISO/IEC 10918-1 : 1993(E), Annex B
634   // (http://www.w3.org/Graphics/JPEG/itu-t81.pdf)
635   RCHECK(buffer_size >= 16);
636 
637   int offset = 0;
638   int last_restart = -1;
639   int num_codes = 0;
640   while (offset + 5 < buffer_size) {
641     // Marker codes are always a two byte code with the first byte xFF.
642     RCHECK(buffer[offset] == 0xff);
643     uint8 code = buffer[offset + 1];
644     RCHECK(code >= 0xc0 || code == 1);
645 
646     // Skip sequences of xFF.
647     if (code == 0xff) {
648       ++offset;
649       continue;
650     }
651 
652     // Success if the next marker code is EOI (end of image)
653     if (code == 0xd9)
654       return true;
655 
656     // Check remaining codes.
657     if (code == 0xd8 || code == 1) {
658       // SOI (start of image) / TEM (private use). No other data with header.
659       offset += 2;
660     } else if (code >= 0xd0 && code <= 0xd7) {
661       // RST (restart) codes must be in sequence. No other data with header.
662       int restart = code & 0x07;
663       if (last_restart >= 0)
664         RCHECK(restart == (last_restart + 1) % 8);
665       last_restart = restart;
666       offset += 2;
667     } else {
668       // All remaining marker codes are followed by a length of the header.
669       int length = Read16(buffer + offset + 2) + 2;
670 
671       // Special handling of SOS (start of scan) marker since the entropy
672       // coded data follows the SOS. Any xFF byte in the data block must be
673       // followed by x00 in the data.
674       if (code == 0xda) {
675         int number_components = buffer[offset + 4];
676         RCHECK(length == 8 + 2 * number_components);
677 
678         // Advance to the next marker.
679         offset += length;
680         while (offset + 2 < buffer_size) {
681           if (buffer[offset] == 0xff && buffer[offset + 1] != 0)
682             break;
683           ++offset;
684         }
685       } else {
686         // Skip over the marker data for the other marker codes.
687         offset += length;
688       }
689     }
690     ++num_codes;
691   }
692   return (num_codes > 1);
693 }
694 
695 enum Mpeg2StartCodes {
696   PROGRAM_END_CODE = 0xb9,
697   PACK_START_CODE = 0xba
698 };
699 
700 // Checks for a MPEG2 Program Stream.
CheckMpeg2ProgramStream(const uint8 * buffer,int buffer_size)701 static bool CheckMpeg2ProgramStream(const uint8* buffer, int buffer_size) {
702   // Reference: ISO/IEC 13818-1 : 2000 (E) / ITU-T Rec. H.222.0 (2000 E).
703   RCHECK(buffer_size > 14);
704 
705   int offset = 0;
706   while (offset + 14 < buffer_size) {
707     BitReader reader(buffer + offset, 14);
708 
709     // Must start with pack_start_code.
710     RCHECK(ReadBits(&reader, 24) == 1);
711     RCHECK(ReadBits(&reader, 8) == PACK_START_CODE);
712 
713     // Determine MPEG version (MPEG1 has b0010, while MPEG2 has b01).
714     int mpeg_version = ReadBits(&reader, 2);
715     if (mpeg_version == 0) {
716       // MPEG1, 10 byte header
717       // Validate rest of version code
718       RCHECK(ReadBits(&reader, 2) == 2);
719     } else {
720       RCHECK(mpeg_version == 1);
721     }
722 
723     // Skip system_clock_reference_base [32..30].
724     reader.SkipBits(3);
725 
726     // Verify marker bit.
727     RCHECK(ReadBits(&reader, 1) == 1);
728 
729     // Skip system_clock_reference_base [29..15].
730     reader.SkipBits(15);
731 
732     // Verify next marker bit.
733     RCHECK(ReadBits(&reader, 1) == 1);
734 
735     // Skip system_clock_reference_base [14..0].
736     reader.SkipBits(15);
737 
738     // Verify next marker bit.
739     RCHECK(ReadBits(&reader, 1) == 1);
740 
741     if (mpeg_version == 0) {
742       // Verify second marker bit.
743       RCHECK(ReadBits(&reader, 1) == 1);
744 
745       // Skip mux_rate.
746       reader.SkipBits(22);
747 
748       // Verify next marker bit.
749       RCHECK(ReadBits(&reader, 1) == 1);
750 
751       // Update offset to be after this header.
752       offset += 12;
753     } else {
754       // Must be MPEG2.
755       // Skip program_mux_rate.
756       reader.SkipBits(22);
757 
758       // Verify pair of marker bits.
759       RCHECK(ReadBits(&reader, 2) == 3);
760 
761       // Skip reserved.
762       reader.SkipBits(5);
763 
764       // Update offset to be after this header.
765       int pack_stuffing_length = ReadBits(&reader, 3);
766       offset += 14 + pack_stuffing_length;
767     }
768 
769     // Check for system headers and PES_packets.
770     while (offset + 6 < buffer_size && Read24(buffer + offset) == 1) {
771       // Next 8 bits determine stream type.
772       int stream_id = buffer[offset + 3];
773 
774       // Some stream types are reserved and shouldn't occur.
775       if (mpeg_version == 0)
776         RCHECK(stream_id != 0xbc && stream_id < 0xf0);
777       else
778         RCHECK(stream_id != 0xfc && stream_id != 0xfd && stream_id != 0xfe);
779 
780       // Some stream types are used for pack headers.
781       if (stream_id == PACK_START_CODE)  // back to outer loop.
782         break;
783       if (stream_id == PROGRAM_END_CODE)  // end of stream.
784         return true;
785 
786       int pes_length = Read16(buffer + offset + 4);
787       RCHECK(pes_length > 0);
788       offset = offset + 6 + pes_length;
789     }
790   }
791   // Success as we are off the end of the buffer and liked everything
792   // in the buffer.
793   return true;
794 }
795 
796 const uint8 kMpeg2SyncWord = 0x47;
797 
798 // Checks for a MPEG2 Transport Stream.
CheckMpeg2TransportStream(const uint8 * buffer,int buffer_size)799 static bool CheckMpeg2TransportStream(const uint8* buffer, int buffer_size) {
800   // Spec: ISO/IEC 13818-1 : 2000 (E) / ITU-T Rec. H.222.0 (2000 E).
801   // Normal packet size is 188 bytes. However, some systems add various error
802   // correction data at the end, resulting in packet of length 192/204/208
803   // (https://en.wikipedia.org/wiki/MPEG_transport_stream). Determine the
804   // length with the first packet.
805   RCHECK(buffer_size >= 250);  // Want more than 1 packet to check.
806 
807   int offset = 0;
808   int packet_length = -1;
809   while (buffer[offset] != kMpeg2SyncWord && offset < 20) {
810     // Skip over any header in the first 20 bytes.
811     ++offset;
812   }
813 
814   while (offset + 6 < buffer_size) {
815     BitReader reader(buffer + offset, 6);
816 
817     // Must start with sync byte.
818     RCHECK(ReadBits(&reader, 8) == kMpeg2SyncWord);
819 
820     // Skip transport_error_indicator, payload_unit_start_indicator, and
821     // transport_priority.
822     reader.SkipBits(1 + 1 + 1);
823 
824     // Verify the pid is not a reserved value.
825     int pid = ReadBits(&reader, 13);
826     RCHECK(pid < 3 || pid > 15);
827 
828     // Skip transport_scrambling_control.
829     reader.SkipBits(2);
830 
831     // Adaptation_field_control can not be 0.
832     int adaptation_field_control = ReadBits(&reader, 2);
833     RCHECK(adaptation_field_control != 0);
834 
835     // If there is an adaptation_field, verify it.
836     if (adaptation_field_control >= 2) {
837       // Skip continuity_counter.
838       reader.SkipBits(4);
839 
840       // Get adaptation_field_length and verify it.
841       int adaptation_field_length = ReadBits(&reader, 8);
842       if (adaptation_field_control == 2)
843         RCHECK(adaptation_field_length == 183);
844       else
845         RCHECK(adaptation_field_length <= 182);
846     }
847 
848     // Attempt to determine the packet length on the first packet.
849     if (packet_length < 0) {
850       if (buffer[offset + 188] == kMpeg2SyncWord)
851         packet_length = 188;
852       else if (buffer[offset + 192] == kMpeg2SyncWord)
853         packet_length = 192;
854       else if (buffer[offset + 204] == kMpeg2SyncWord)
855         packet_length = 204;
856       else
857         packet_length = 208;
858     }
859     offset += packet_length;
860   }
861   return true;
862 }
863 
864 enum Mpeg4StartCodes {
865   VISUAL_OBJECT_SEQUENCE_START_CODE = 0xb0,
866   VISUAL_OBJECT_SEQUENCE_END_CODE = 0xb1,
867   VISUAL_OBJECT_START_CODE = 0xb5,
868   VOP_START_CODE = 0xb6
869 };
870 
871 // Checks for a raw MPEG4 bitstream container.
CheckMpeg4BitStream(const uint8 * buffer,int buffer_size)872 static bool CheckMpeg4BitStream(const uint8* buffer, int buffer_size) {
873   // Defined in ISO/IEC 14496-2:2001.
874   // However, no length ... simply scan for start code values.
875   // Note tags are very similar to H.264.
876   RCHECK(buffer_size > 4);
877 
878   int offset = 0;
879   int sequence_start_count = 0;
880   int sequence_end_count = 0;
881   int visual_object_count = 0;
882   int vop_count = 0;
883   while (true) {
884     // Advance to start_code, if there is one.
885     if (!AdvanceToStartCode(buffer, buffer_size, &offset, 6, 24, 1)) {
886       // Not a complete sequence in memory, so return true if we've seen a
887       // visual_object_sequence_start_code and a visual_object_start_code.
888       return (sequence_start_count > 0 && visual_object_count > 0);
889     }
890 
891     // Now verify the block. AdvanceToStartCode() made sure that there are
892     // at least 6 bytes remaining in the buffer.
893     BitReader reader(buffer + offset, 6);
894     RCHECK(ReadBits(&reader, 24) == 1);
895 
896     int start_code = ReadBits(&reader, 8);
897     RCHECK(start_code < 0x30 || start_code > 0xaf);  // 30..AF and
898     RCHECK(start_code < 0xb7 || start_code > 0xb9);  // B7..B9 reserved
899 
900     switch (start_code) {
901       case VISUAL_OBJECT_SEQUENCE_START_CODE: {
902         ++sequence_start_count;
903         // Verify profile in not one of many reserved values.
904         int profile = ReadBits(&reader, 8);
905         RCHECK(profile > 0);
906         RCHECK(profile < 0x04 || profile > 0x10);
907         RCHECK(profile < 0x13 || profile > 0x20);
908         RCHECK(profile < 0x23 || profile > 0x31);
909         RCHECK(profile < 0x35 || profile > 0x41);
910         RCHECK(profile < 0x43 || profile > 0x60);
911         RCHECK(profile < 0x65 || profile > 0x70);
912         RCHECK(profile < 0x73 || profile > 0x80);
913         RCHECK(profile < 0x83 || profile > 0x90);
914         RCHECK(profile < 0x95 || profile > 0xa0);
915         RCHECK(profile < 0xa4 || profile > 0xb0);
916         RCHECK(profile < 0xb5 || profile > 0xc0);
917         RCHECK(profile < 0xc3 || profile > 0xd0);
918         RCHECK(profile < 0xe4);
919         break;
920       }
921 
922       case VISUAL_OBJECT_SEQUENCE_END_CODE:
923         RCHECK(++sequence_end_count == sequence_start_count);
924         break;
925 
926       case VISUAL_OBJECT_START_CODE: {
927         ++visual_object_count;
928         if (ReadBits(&reader, 1) == 1) {
929           int visual_object_verid = ReadBits(&reader, 4);
930           RCHECK(visual_object_verid > 0 && visual_object_verid < 3);
931           RCHECK(ReadBits(&reader, 3) != 0);
932         }
933         int visual_object_type = ReadBits(&reader, 4);
934         RCHECK(visual_object_type > 0 && visual_object_type < 6);
935         break;
936       }
937 
938       case VOP_START_CODE:
939         RCHECK(++vop_count <= visual_object_count);
940         break;
941     }
942     // Skip this block.
943     offset += 6;
944   }
945 }
946 
947 // Additional checks for a MOV/QuickTime/MPEG4 container.
CheckMov(const uint8 * buffer,int buffer_size)948 static bool CheckMov(const uint8* buffer, int buffer_size) {
949   // Reference: ISO/IEC 14496-12:2005(E).
950   // (http://standards.iso.org/ittf/PubliclyAvailableStandards/c061988_ISO_IEC_14496-12_2012.zip)
951   RCHECK(buffer_size > 8);
952 
953   int offset = 0;
954   while (offset + 8 < buffer_size) {
955     int atomsize = Read32(buffer + offset);
956     uint32 atomtype = Read32(buffer + offset + 4);
957     // Only need to check for ones that are valid at the top level.
958     switch (atomtype) {
959       case TAG('f','t','y','p'):
960       case TAG('p','d','i','n'):
961       case TAG('m','o','o','v'):
962       case TAG('m','o','o','f'):
963       case TAG('m','f','r','a'):
964       case TAG('m','d','a','t'):
965       case TAG('f','r','e','e'):
966       case TAG('s','k','i','p'):
967       case TAG('m','e','t','a'):
968       case TAG('m','e','c','o'):
969       case TAG('s','t','y','p'):
970       case TAG('s','i','d','x'):
971       case TAG('s','s','i','x'):
972       case TAG('p','r','f','t'):
973       case TAG('b','l','o','c'):
974         break;
975       default:
976         return false;
977     }
978     if (atomsize == 1) {
979       // Indicates that the length is the next 64bits.
980       if (offset + 16 > buffer_size)
981         break;
982       if (Read32(buffer + offset + 8) != 0)
983         break;  // Offset is way past buffer size.
984       atomsize = Read32(buffer + offset + 12);
985     }
986     if (atomsize <= 0)
987       break;  // Indicates the last atom or length too big.
988     offset += atomsize;
989   }
990   return true;
991 }
992 
993 enum MPEGVersion {
994   VERSION_25 = 0,
995   VERSION_RESERVED,
996   VERSION_2,
997   VERSION_1
998 };
999 enum MPEGLayer {
1000   L_RESERVED = 0,
1001   LAYER_3,
1002   LAYER_2,
1003   LAYER_1
1004 };
1005 
1006 static int kSampleRateTable[4][4] = { { 11025, 12000, 8000, 0 },   // v2.5
1007                                       { 0, 0, 0, 0 },              // not used
1008                                       { 22050, 24000, 16000, 0 },  // v2
1009                                       { 44100, 48000, 32000, 0 }   // v1
1010 };
1011 
1012 static int kBitRateTableV1L1[16] = { 0, 32, 64, 96, 128, 160, 192, 224, 256,
1013                                      288, 320, 352, 384, 416, 448, 0 };
1014 static int kBitRateTableV1L2[16] = { 0, 32, 48, 56, 64, 80, 96, 112, 128, 160,
1015                                      192, 224, 256, 320, 384, 0 };
1016 static int kBitRateTableV1L3[16] = { 0, 32, 40, 48, 56, 64, 80, 96, 112, 128,
1017                                      160, 192, 224, 256, 320, 0 };
1018 static int kBitRateTableV2L1[16] = { 0, 32, 48, 56, 64, 80, 96, 112, 128, 144,
1019                                      160, 176, 192, 224, 256, 0 };
1020 static int kBitRateTableV2L23[16] = { 0, 8, 16, 24, 32, 40, 48, 56, 64, 80, 96,
1021                                       112, 128, 144, 160, 0 };
1022 
ValidMpegAudioFrameHeader(const uint8 * header,int header_size,int * framesize)1023 static bool ValidMpegAudioFrameHeader(const uint8* header,
1024                                       int header_size,
1025                                       int* framesize) {
1026   // Reference: http://mpgedit.org/mpgedit/mpeg_format/mpeghdr.htm.
1027   DCHECK_GE(header_size, 4);
1028   *framesize = 0;
1029   BitReader reader(header, 4);  // Header can only be 4 bytes long.
1030 
1031   // Verify frame sync (11 bits) are all set.
1032   RCHECK(ReadBits(&reader, 11) == 0x7ff);
1033 
1034   // Verify MPEG audio version id.
1035   int version = ReadBits(&reader, 2);
1036   RCHECK(version != 1);  // Reserved.
1037 
1038   // Verify layer.
1039   int layer = ReadBits(&reader, 2);
1040   RCHECK(layer != 0);
1041 
1042   // Skip protection bit.
1043   reader.SkipBits(1);
1044 
1045   // Verify bitrate index.
1046   int bitrate_index = ReadBits(&reader, 4);
1047   RCHECK(bitrate_index != 0xf);
1048 
1049   // Verify sampling rate frequency index.
1050   int sampling_index = ReadBits(&reader, 2);
1051   RCHECK(sampling_index != 3);
1052 
1053   // Get padding bit.
1054   int padding = ReadBits(&reader, 1);
1055 
1056   // Frame size:
1057   // For Layer I files = (12 * BitRate / SampleRate + Padding) * 4
1058   // For others = 144 * BitRate / SampleRate + Padding
1059   // Unfortunately, BitRate and SampleRate are coded.
1060   int sampling_rate = kSampleRateTable[version][sampling_index];
1061   int bitrate;
1062   if (version == VERSION_1) {
1063     if (layer == LAYER_1)
1064       bitrate = kBitRateTableV1L1[bitrate_index];
1065     else if (layer == LAYER_2)
1066       bitrate = kBitRateTableV1L2[bitrate_index];
1067     else
1068       bitrate = kBitRateTableV1L3[bitrate_index];
1069   } else {
1070     if (layer == LAYER_1)
1071       bitrate = kBitRateTableV2L1[bitrate_index];
1072     else
1073       bitrate = kBitRateTableV2L23[bitrate_index];
1074   }
1075   if (layer == LAYER_1)
1076     *framesize = ((12000 * bitrate) / sampling_rate + padding) * 4;
1077   else
1078     *framesize = (144000 * bitrate) / sampling_rate + padding;
1079   return (bitrate > 0 && sampling_rate > 0);
1080 }
1081 
1082 // Extract a size encoded the MP3 way.
GetMp3HeaderSize(const uint8 * buffer,int buffer_size)1083 static int GetMp3HeaderSize(const uint8* buffer, int buffer_size) {
1084   DCHECK_GE(buffer_size, 9);
1085   int size = ((buffer[6] & 0x7f) << 21) + ((buffer[7] & 0x7f) << 14) +
1086              ((buffer[8] & 0x7f) << 7) + (buffer[9] & 0x7f) + 10;
1087   if (buffer[5] & 0x10)  // Footer added?
1088     size += 10;
1089   return size;
1090 }
1091 
1092 // Additional checks for a MP3 container.
CheckMp3(const uint8 * buffer,int buffer_size,bool seenHeader)1093 static bool CheckMp3(const uint8* buffer, int buffer_size, bool seenHeader) {
1094   RCHECK(buffer_size >= 10);  // Must be enough to read the initial header.
1095 
1096   int framesize;
1097   int numSeen = 0;
1098   int offset = 0;
1099   if (seenHeader) {
1100     offset = GetMp3HeaderSize(buffer, buffer_size);
1101   } else {
1102     // Skip over leading 0's.
1103     while (offset < buffer_size && buffer[offset] == 0)
1104       ++offset;
1105   }
1106 
1107   while (offset + 3 < buffer_size) {
1108     RCHECK(ValidMpegAudioFrameHeader(
1109         buffer + offset, buffer_size - offset, &framesize));
1110 
1111     // Have we seen enough valid headers?
1112     if (++numSeen > 10)
1113       return true;
1114     offset += framesize;
1115   }
1116   // Off the end of the buffer, return success if a few valid headers seen.
1117   return numSeen > 2;
1118 }
1119 
1120 // Check that the next characters in |buffer| represent a number. The format
1121 // accepted is optional whitespace followed by 1 or more digits. |max_digits|
1122 // specifies the maximum number of digits to process. Returns true if a valid
1123 // number is found, false otherwise.
VerifyNumber(const uint8 * buffer,int buffer_size,int * offset,int max_digits)1124 static bool VerifyNumber(const uint8* buffer,
1125                          int buffer_size,
1126                          int* offset,
1127                          int max_digits) {
1128   RCHECK(*offset < buffer_size);
1129 
1130   // Skip over any leading space.
1131   while (isspace(buffer[*offset])) {
1132     ++(*offset);
1133     RCHECK(*offset < buffer_size);
1134   }
1135 
1136   // Need to process up to max_digits digits.
1137   int numSeen = 0;
1138   while (--max_digits >= 0 && isdigit(buffer[*offset])) {
1139     ++numSeen;
1140     ++(*offset);
1141     if (*offset >= buffer_size)
1142       return true;  // Out of space but seen a digit.
1143   }
1144 
1145   // Success if at least one digit seen.
1146   return (numSeen > 0);
1147 }
1148 
1149 // Check that the next character in |buffer| is one of |c1| or |c2|. |c2| is
1150 // optional. Returns true if there is a match, false if no match or out of
1151 // space.
VerifyCharacters(const uint8 * buffer,int buffer_size,int * offset,char c1,char c2)1152 static inline bool VerifyCharacters(const uint8* buffer,
1153                                     int buffer_size,
1154                                     int* offset,
1155                                     char c1,
1156                                     char c2) {
1157   RCHECK(*offset < buffer_size);
1158   char c = static_cast<char>(buffer[(*offset)++]);
1159   return (c == c1 || (c == c2 && c2 != 0));
1160 }
1161 
1162 // Checks for a SRT container.
CheckSrt(const uint8 * buffer,int buffer_size)1163 static bool CheckSrt(const uint8* buffer, int buffer_size) {
1164   // Reference: http://en.wikipedia.org/wiki/SubRip
1165   RCHECK(buffer_size > 20);
1166 
1167   // First line should just be the subtitle sequence number.
1168   int offset = StartsWith(buffer, buffer_size, UTF8_BYTE_ORDER_MARK) ? 3 : 0;
1169   RCHECK(VerifyNumber(buffer, buffer_size, &offset, 100));
1170   RCHECK(VerifyCharacters(buffer, buffer_size, &offset, '\n', '\r'));
1171 
1172   // Skip any additional \n\r.
1173   while (VerifyCharacters(buffer, buffer_size, &offset, '\n', '\r')) {}
1174   --offset;  // Since VerifyCharacters() gobbled up the next non-CR/LF.
1175 
1176   // Second line should look like the following:
1177   //   00:00:10,500 --> 00:00:13,000
1178   // Units separator can be , or .
1179   RCHECK(VerifyNumber(buffer, buffer_size, &offset, 100));
1180   RCHECK(VerifyCharacters(buffer, buffer_size, &offset, ':', 0));
1181   RCHECK(VerifyNumber(buffer, buffer_size, &offset, 2));
1182   RCHECK(VerifyCharacters(buffer, buffer_size, &offset, ':', 0));
1183   RCHECK(VerifyNumber(buffer, buffer_size, &offset, 2));
1184   RCHECK(VerifyCharacters(buffer, buffer_size, &offset, ',', '.'));
1185   RCHECK(VerifyNumber(buffer, buffer_size, &offset, 3));
1186   RCHECK(VerifyCharacters(buffer, buffer_size, &offset, ' ', 0));
1187   RCHECK(VerifyCharacters(buffer, buffer_size, &offset, '-', 0));
1188   RCHECK(VerifyCharacters(buffer, buffer_size, &offset, '-', 0));
1189   RCHECK(VerifyCharacters(buffer, buffer_size, &offset, '>', 0));
1190   RCHECK(VerifyCharacters(buffer, buffer_size, &offset, ' ', 0));
1191   RCHECK(VerifyNumber(buffer, buffer_size, &offset, 100));
1192   RCHECK(VerifyCharacters(buffer, buffer_size, &offset, ':', 0));
1193   RCHECK(VerifyNumber(buffer, buffer_size, &offset, 2));
1194   RCHECK(VerifyCharacters(buffer, buffer_size, &offset, ':', 0));
1195   RCHECK(VerifyNumber(buffer, buffer_size, &offset, 2));
1196   RCHECK(VerifyCharacters(buffer, buffer_size, &offset, ',', '.'));
1197   RCHECK(VerifyNumber(buffer, buffer_size, &offset, 3));
1198   return true;
1199 }
1200 
1201 // Read a Matroska Element Id.
GetElementId(BitReader * reader)1202 static int GetElementId(BitReader* reader) {
1203   // Element ID is coded with the leading zero bits (max 3) determining size.
1204   // If it is an invalid encoding or the end of the buffer is reached,
1205   // return -1 as a tag that won't be expected.
1206   if (reader->bits_available() >= 8) {
1207     int num_bits_to_read = 0;
1208     static int prefix[] = { 0x80, 0x4000, 0x200000, 0x10000000 };
1209     for (int i = 0; i < 4; ++i) {
1210       num_bits_to_read += 7;
1211       if (ReadBits(reader, 1) == 1) {
1212         if (reader->bits_available() < num_bits_to_read)
1213           break;
1214         // prefix[] adds back the bits read individually.
1215         return ReadBits(reader, num_bits_to_read) | prefix[i];
1216       }
1217     }
1218   }
1219   // Invalid encoding, return something not expected.
1220   return -1;
1221 }
1222 
1223 // Read a Matroska Unsigned Integer (VINT).
GetVint(BitReader * reader)1224 static uint64 GetVint(BitReader* reader) {
1225   // Values are coded with the leading zero bits (max 7) determining size.
1226   // If it is an invalid coding or the end of the buffer is reached,
1227   // return something that will go off the end of the buffer.
1228   if (reader->bits_available() >= 8) {
1229     int num_bits_to_read = 0;
1230     for (int i = 0; i < 8; ++i) {
1231       num_bits_to_read += 7;
1232       if (ReadBits(reader, 1) == 1) {
1233         if (reader->bits_available() < num_bits_to_read)
1234           break;
1235         return ReadBits(reader, num_bits_to_read);
1236       }
1237     }
1238   }
1239   // Incorrect format (more than 7 leading 0's) or off the end of the buffer.
1240   // Since the return value is used as a byte size, return a value that will
1241   // cause a failure when used.
1242   return (reader->bits_available() / 8) + 2;
1243 }
1244 
1245 // Additional checks for a WEBM container.
CheckWebm(const uint8 * buffer,int buffer_size)1246 static bool CheckWebm(const uint8* buffer, int buffer_size) {
1247   // Reference: http://www.matroska.org/technical/specs/index.html
1248   RCHECK(buffer_size > 12);
1249 
1250   BitReader reader(buffer, buffer_size);
1251 
1252   // Verify starting Element Id.
1253   RCHECK(GetElementId(&reader) == 0x1a45dfa3);
1254 
1255   // Get the header size, and ensure there are enough bits to check.
1256   int header_size = GetVint(&reader);
1257   RCHECK(reader.bits_available() / 8 >= header_size);
1258 
1259   // Loop through the header.
1260   while (reader.bits_available() > 0) {
1261     int tag = GetElementId(&reader);
1262     int tagsize = GetVint(&reader);
1263     switch (tag) {
1264       case 0x4286:  // EBMLVersion
1265       case 0x42f7:  // EBMLReadVersion
1266       case 0x42f2:  // EBMLMaxIdLength
1267       case 0x42f3:  // EBMLMaxSizeLength
1268       case 0x4287:  // DocTypeVersion
1269       case 0x4285:  // DocTypeReadVersion
1270       case 0xec:    // void
1271       case 0xbf:    // CRC32
1272         RCHECK(reader.SkipBits(tagsize * 8));
1273         break;
1274 
1275       case 0x4282:  // EBMLDocType
1276         // Need to see "webm" or "matroska" next.
1277         switch (ReadBits(&reader, 32)) {
1278           case TAG('w', 'e', 'b', 'm') :
1279             return true;
1280           case TAG('m', 'a', 't', 'r') :
1281             return (ReadBits(&reader, 32) == TAG('o', 's', 'k', 'a'));
1282         }
1283         return false;
1284 
1285       default:  // Unrecognized tag
1286         return false;
1287     }
1288   }
1289   return false;
1290 }
1291 
1292 enum VC1StartCodes {
1293   VC1_FRAME_START_CODE = 0x0d,
1294   VC1_ENTRY_POINT_START_CODE = 0x0e,
1295   VC1_SEQUENCE_START_CODE = 0x0f
1296 };
1297 
1298 // Checks for a VC1 bitstream container.
CheckVC1(const uint8 * buffer,int buffer_size)1299 static bool CheckVC1(const uint8* buffer, int buffer_size) {
1300   // Reference: SMPTE 421M
1301   // (http://standards.smpte.org/content/978-1-61482-555-5/st-421-2006/SEC1.body.pdf)
1302   // However, no length ... simply scan for start code values.
1303   // Expect to see SEQ | [ [ ENTRY ] PIC* ]*
1304   // Note tags are very similar to H.264.
1305 
1306   RCHECK(buffer_size >= 24);
1307 
1308   // First check for Bitstream Metadata Serialization (Annex L)
1309   if (buffer[0] == 0xc5 &&
1310       Read32(buffer + 4) == 0x04 &&
1311       Read32(buffer + 20) == 0x0c) {
1312     // Verify settings in STRUCT_C and STRUCT_A
1313     BitReader reader(buffer + 8, 12);
1314 
1315     int profile = ReadBits(&reader, 4);
1316     if (profile == 0 || profile == 4) {  // simple or main
1317       // Skip FRMRTQ_POSTPROC, BITRTQ_POSTPROC, and LOOPFILTER.
1318       reader.SkipBits(3 + 5 + 1);
1319 
1320       // Next bit must be 0.
1321       RCHECK(ReadBits(&reader, 1) == 0);
1322 
1323       // Skip MULTIRES.
1324       reader.SkipBits(1);
1325 
1326       // Next bit must be 1.
1327       RCHECK(ReadBits(&reader, 1) == 1);
1328 
1329       // Skip FASTUVMC, EXTENDED_MV, DQUANT, and VSTRANSFORM.
1330       reader.SkipBits(1 + 1 + 2 + 1);
1331 
1332       // Next bit must be 0.
1333       RCHECK(ReadBits(&reader, 1) == 0);
1334 
1335       // Skip OVERLAP, SYNCMARKER, RANGERED, MAXBFRAMES, QUANTIZER, and
1336       // FINTERPFLAG.
1337       reader.SkipBits(1 + 1 + 1 + 3 + 2 + 1);
1338 
1339       // Next bit must be 1.
1340       RCHECK(ReadBits(&reader, 1) == 1);
1341 
1342     } else {
1343       RCHECK(profile == 12);  // Other profile values not allowed.
1344       RCHECK(ReadBits(&reader, 28) == 0);
1345     }
1346 
1347     // Now check HORIZ_SIZE and VERT_SIZE, which must be 8192 or less.
1348     RCHECK(ReadBits(&reader, 32) <= 8192);
1349     RCHECK(ReadBits(&reader, 32) <= 8192);
1350     return true;
1351   }
1352 
1353   // Buffer isn't Bitstream Metadata, so scan for start codes.
1354   int offset = 0;
1355   int sequence_start_code = 0;
1356   int frame_start_code = 0;
1357   while (true) {
1358     // Advance to start_code, if there is one.
1359     if (!AdvanceToStartCode(buffer, buffer_size, &offset, 5, 24, 1)) {
1360       // Not a complete sequence in memory, so return true if we've seen a
1361       // sequence start and a frame start (not checking entry points since
1362       // they only occur in advanced profiles).
1363       return (sequence_start_code > 0 && frame_start_code > 0);
1364     }
1365 
1366     // Now verify the block. AdvanceToStartCode() made sure that there are
1367     // at least 5 bytes remaining in the buffer.
1368     BitReader reader(buffer + offset, 5);
1369     RCHECK(ReadBits(&reader, 24) == 1);
1370 
1371     // Keep track of the number of certain types received.
1372     switch (ReadBits(&reader, 8)) {
1373       case VC1_SEQUENCE_START_CODE: {
1374         ++sequence_start_code;
1375         switch (ReadBits(&reader, 2)) {
1376           case 0:  // simple
1377           case 1:  // main
1378             RCHECK(ReadBits(&reader, 2) == 0);
1379             break;
1380           case 2:  // complex
1381             return false;
1382           case 3:  // advanced
1383             RCHECK(ReadBits(&reader, 3) <= 4);  // Verify level = 0..4
1384             RCHECK(ReadBits(&reader, 2) == 1);  // Verify colordiff_format = 1
1385             break;
1386         }
1387         break;
1388       }
1389 
1390       case VC1_ENTRY_POINT_START_CODE:
1391         // No fields in entry data to check. However, it must occur after
1392         // sequence header.
1393         RCHECK(sequence_start_code > 0);
1394         break;
1395 
1396       case VC1_FRAME_START_CODE:
1397         ++frame_start_code;
1398         break;
1399     }
1400     offset += 5;
1401   }
1402 }
1403 
1404 // For some formats the signature is a bunch of characters. They are defined
1405 // below. Note that the first 4 characters of the string may be used as a TAG
1406 // in LookupContainerByFirst4. For signatures that contain embedded \0, use
1407 // uint8[].
1408 static const char kAmrSignature[] = "#!AMR";
1409 static const uint8 kAsfSignature[] = { 0x30, 0x26, 0xb2, 0x75, 0x8e, 0x66, 0xcf,
1410                                        0x11, 0xa6, 0xd9, 0x00, 0xaa, 0x00, 0x62,
1411                                        0xce, 0x6c };
1412 static const char kAssSignature[] = "[Script Info]";
1413 static const char kAssBomSignature[] = UTF8_BYTE_ORDER_MARK "[Script Info]";
1414 static const uint8 kWtvSignature[] = { 0xb7, 0xd8, 0x00, 0x20, 0x37, 0x49, 0xda,
1415                                        0x11, 0xa6, 0x4e, 0x00, 0x07, 0xe9, 0x5e,
1416                                        0xad, 0x8d };
1417 
1418 // Attempt to determine the container type from the buffer provided. This is
1419 // a simple pass, that uses the first 4 bytes of the buffer as an index to get
1420 // a rough idea of the container format.
LookupContainerByFirst4(const uint8 * buffer,int buffer_size)1421 static MediaContainerName LookupContainerByFirst4(const uint8* buffer,
1422                                                   int buffer_size) {
1423   // Minimum size that the code expects to exist without checking size.
1424   if (buffer_size < 12)
1425     return CONTAINER_UNKNOWN;
1426 
1427   uint32 first4 = Read32(buffer);
1428   switch (first4) {
1429     case 0x1a45dfa3:
1430       if (CheckWebm(buffer, buffer_size))
1431         return CONTAINER_WEBM;
1432       break;
1433 
1434     case 0x3026b275:
1435       if (StartsWith(buffer,
1436                      buffer_size,
1437                      kAsfSignature,
1438                      sizeof(kAsfSignature))) {
1439         return CONTAINER_ASF;
1440       }
1441       break;
1442 
1443     case TAG('#','!','A','M'):
1444       if (StartsWith(buffer, buffer_size, kAmrSignature))
1445         return CONTAINER_AMR;
1446       break;
1447 
1448     case TAG('#','E','X','T'):
1449       if (CheckHls(buffer, buffer_size))
1450         return CONTAINER_HLS;
1451       break;
1452 
1453     case TAG('.','R','M','F'):
1454       if (buffer[4] == 0 && buffer[5] == 0)
1455         return CONTAINER_RM;
1456       break;
1457 
1458     case TAG('.','r','a','\xfd'):
1459       return CONTAINER_RM;
1460 
1461     case TAG('B','I','K','b'):
1462     case TAG('B','I','K','d'):
1463     case TAG('B','I','K','f'):
1464     case TAG('B','I','K','g'):
1465     case TAG('B','I','K','h'):
1466     case TAG('B','I','K','i'):
1467       if (CheckBink(buffer, buffer_size))
1468         return CONTAINER_BINK;
1469       break;
1470 
1471     case TAG('c','a','f','f'):
1472       if (CheckCaf(buffer, buffer_size))
1473         return CONTAINER_CAF;
1474       break;
1475 
1476     case TAG('D','E','X','A'):
1477       if (buffer_size > 15 &&
1478           Read16(buffer + 11) <= 2048 &&
1479           Read16(buffer + 13) <= 2048) {
1480         return CONTAINER_DXA;
1481       }
1482       break;
1483 
1484     case TAG('D','T','S','H'):
1485       if (Read32(buffer + 4) == TAG('D','H','D','R'))
1486         return CONTAINER_DTSHD;
1487       break;
1488 
1489     case 0x64a30100:
1490     case 0x64a30200:
1491     case 0x64a30300:
1492     case 0x64a30400:
1493     case 0x0001a364:
1494     case 0x0002a364:
1495     case 0x0003a364:
1496       if (Read32(buffer + 4) != 0 && Read32(buffer + 8) != 0)
1497         return CONTAINER_IRCAM;
1498       break;
1499 
1500     case TAG('f','L','a','C'):
1501       return CONTAINER_FLAC;
1502 
1503     case TAG('F','L','V',0):
1504     case TAG('F','L','V',1):
1505     case TAG('F','L','V',2):
1506     case TAG('F','L','V',3):
1507     case TAG('F','L','V',4):
1508       if (buffer[5] == 0 && Read32(buffer + 5) > 8)
1509         return CONTAINER_FLV;
1510       break;
1511 
1512     case TAG('F','O','R','M'):
1513       switch (Read32(buffer + 8)) {
1514         case TAG('A','I','F','F'):
1515         case TAG('A','I','F','C'):
1516           return CONTAINER_AIFF;
1517       }
1518       break;
1519 
1520     case TAG('M','A','C',' '):
1521       return CONTAINER_APE;
1522 
1523     case TAG('O','N','2',' '):
1524       if (Read32(buffer + 8) == TAG('O','N','2','f'))
1525         return CONTAINER_AVI;
1526       break;
1527 
1528     case TAG('O','g','g','S'):
1529       if (buffer[5] <= 7)
1530         return CONTAINER_OGG;
1531       break;
1532 
1533     case TAG('R','F','6','4'):
1534       if (buffer_size > 16 && Read32(buffer + 12) == TAG('d','s','6','4'))
1535         return CONTAINER_WAV;
1536       break;
1537 
1538     case TAG('R','I','F','F'):
1539       switch (Read32(buffer + 8)) {
1540         case TAG('A','V','I',' '):
1541         case TAG('A','V','I','X'):
1542         case TAG('A','V','I','\x19'):
1543         case TAG('A','M','V',' '):
1544           return CONTAINER_AVI;
1545         case TAG('W','A','V','E'):
1546           return CONTAINER_WAV;
1547       }
1548       break;
1549 
1550     case TAG('[','S','c','r'):
1551       if (StartsWith(buffer, buffer_size, kAssSignature))
1552         return CONTAINER_ASS;
1553       break;
1554 
1555     case TAG('\xef','\xbb','\xbf','['):
1556       if (StartsWith(buffer, buffer_size, kAssBomSignature))
1557         return CONTAINER_ASS;
1558       break;
1559 
1560     case 0x7ffe8001:
1561     case 0xfe7f0180:
1562     case 0x1fffe800:
1563     case 0xff1f00e8:
1564       if (CheckDts(buffer, buffer_size))
1565         return CONTAINER_DTS;
1566       break;
1567 
1568     case 0xb7d80020:
1569       if (StartsWith(buffer,
1570                      buffer_size,
1571                      kWtvSignature,
1572                      sizeof(kWtvSignature))) {
1573         return CONTAINER_WTV;
1574       }
1575       break;
1576   }
1577 
1578   // Now try a few different ones that look at something other
1579   // than the first 4 bytes.
1580   uint32 first3 = first4 & 0xffffff00;
1581   switch (first3) {
1582     case TAG('C','W','S',0):
1583     case TAG('F','W','S',0):
1584       return CONTAINER_SWF;
1585 
1586     case TAG('I','D','3',0):
1587       if (CheckMp3(buffer, buffer_size, true))
1588         return CONTAINER_MP3;
1589       break;
1590   }
1591 
1592   // Maybe the first 2 characters are something we can use.
1593   uint32 first2 = Read16(buffer);
1594   switch (first2) {
1595     case kAc3SyncWord:
1596       if (CheckAc3(buffer, buffer_size))
1597         return CONTAINER_AC3;
1598       if (CheckEac3(buffer, buffer_size))
1599         return CONTAINER_EAC3;
1600       break;
1601 
1602     case 0xfff0:
1603     case 0xfff1:
1604     case 0xfff8:
1605     case 0xfff9:
1606       if (CheckAac(buffer, buffer_size))
1607         return CONTAINER_AAC;
1608       break;
1609   }
1610 
1611   // Check if the file is in MP3 format without the header.
1612   if (CheckMp3(buffer, buffer_size, false))
1613     return CONTAINER_MP3;
1614 
1615   return CONTAINER_UNKNOWN;
1616 }
1617 
1618 // Attempt to determine the container name from the buffer provided.
DetermineContainer(const uint8 * buffer,int buffer_size)1619 MediaContainerName DetermineContainer(const uint8* buffer, int buffer_size) {
1620   DCHECK(buffer);
1621 
1622   // Since MOV/QuickTime/MPEG4 streams are common, check for them first.
1623   if (CheckMov(buffer, buffer_size))
1624     return CONTAINER_MOV;
1625 
1626   // Next attempt the simple checks, that typically look at just the
1627   // first few bytes of the file.
1628   MediaContainerName result = LookupContainerByFirst4(buffer, buffer_size);
1629   if (result != CONTAINER_UNKNOWN)
1630     return result;
1631 
1632   // Additional checks that may scan a portion of the buffer.
1633   if (CheckMpeg2ProgramStream(buffer, buffer_size))
1634     return CONTAINER_MPEG2PS;
1635   if (CheckMpeg2TransportStream(buffer, buffer_size))
1636     return CONTAINER_MPEG2TS;
1637   if (CheckMJpeg(buffer, buffer_size))
1638     return CONTAINER_MJPEG;
1639   if (CheckDV(buffer, buffer_size))
1640     return CONTAINER_DV;
1641   if (CheckH261(buffer, buffer_size))
1642     return CONTAINER_H261;
1643   if (CheckH263(buffer, buffer_size))
1644     return CONTAINER_H263;
1645   if (CheckH264(buffer, buffer_size))
1646     return CONTAINER_H264;
1647   if (CheckMpeg4BitStream(buffer, buffer_size))
1648     return CONTAINER_MPEG4BS;
1649   if (CheckVC1(buffer, buffer_size))
1650     return CONTAINER_VC1;
1651   if (CheckSrt(buffer, buffer_size))
1652     return CONTAINER_SRT;
1653   if (CheckGsm(buffer, buffer_size))
1654     return CONTAINER_GSM;
1655 
1656   // AC3/EAC3 might not start at the beginning of the stream,
1657   // so scan for a start code.
1658   int offset = 1;  // No need to start at byte 0 due to First4 check.
1659   if (AdvanceToStartCode(buffer, buffer_size, &offset, 4, 16, kAc3SyncWord)) {
1660     if (CheckAc3(buffer + offset, buffer_size - offset))
1661       return CONTAINER_AC3;
1662     if (CheckEac3(buffer + offset, buffer_size - offset))
1663       return CONTAINER_EAC3;
1664   }
1665 
1666   return CONTAINER_UNKNOWN;
1667 }
1668 
1669 }  // namespace container_names
1670 
1671 }  // namespace media
1672