1 /*
2 * Copyright (C) 2009 The Android Open Source Project
3 *
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at
7 *
8 * http://www.apache.org/licenses/LICENSE-2.0
9 *
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
15 */
16
17 //#define LOG_NDEBUG 0
18 #define LOG_TAG "Utils"
19 #include <utils/Log.h>
20 #include <ctype.h>
21 #include <stdio.h>
22 #include <sys/stat.h>
23
24 #include <utility>
25 #include <vector>
26
27 #include "include/ESDS.h"
28 #include "include/HevcUtils.h"
29
30 #include <cutils/properties.h>
31 #include <media/stagefright/CodecBase.h>
32 #include <media/stagefright/foundation/ABuffer.h>
33 #include <media/stagefright/foundation/ADebug.h>
34 #include <media/stagefright/foundation/ALookup.h>
35 #include <media/stagefright/foundation/AMessage.h>
36 #include <media/stagefright/foundation/ByteUtils.h>
37 #include <media/stagefright/foundation/OpusHeader.h>
38 #include <media/stagefright/MetaData.h>
39 #include <media/stagefright/MediaCodecConstants.h>
40 #include <media/stagefright/MediaDefs.h>
41 #include <media/AudioSystem.h>
42 #include <media/MediaPlayerInterface.h>
43 #include <media/stagefright/Utils.h>
44 #include <media/AudioParameter.h>
45 #include <system/audio.h>
46
47 // TODO : Remove the defines once mainline media is built against NDK >= 31.
48 // The mp4 extractor is part of mainline and builds against NDK 29 as of
49 // writing. These keys are available only from NDK 31:
50 #define AMEDIAFORMAT_KEY_MPEGH_PROFILE_LEVEL_INDICATION \
51 "mpegh-profile-level-indication"
52 #define AMEDIAFORMAT_KEY_MPEGH_REFERENCE_CHANNEL_LAYOUT \
53 "mpegh-reference-channel-layout"
54 #define AMEDIAFORMAT_KEY_MPEGH_COMPATIBLE_SETS \
55 "mpegh-compatible-sets"
56
57 namespace {
58 // TODO: this should possibly be handled in an else
59 constexpr static int32_t AACObjectNull = 0;
60
61 // TODO: decide if we should just not transmit the level in this case
62 constexpr static int32_t DolbyVisionLevelUnknown = 0;
63 }
64
65 namespace android {
66
copyNALUToABuffer(sp<ABuffer> * buffer,const uint8_t * ptr,size_t length)67 static status_t copyNALUToABuffer(sp<ABuffer> *buffer, const uint8_t *ptr, size_t length) {
68 if (((*buffer)->size() + 4 + length) > ((*buffer)->capacity() - (*buffer)->offset())) {
69 sp<ABuffer> tmpBuffer = new (std::nothrow) ABuffer((*buffer)->size() + 4 + length + 1024);
70 if (tmpBuffer.get() == NULL || tmpBuffer->base() == NULL) {
71 return NO_MEMORY;
72 }
73 memcpy(tmpBuffer->data(), (*buffer)->data(), (*buffer)->size());
74 tmpBuffer->setRange(0, (*buffer)->size());
75 (*buffer) = tmpBuffer;
76 }
77
78 memcpy((*buffer)->data() + (*buffer)->size(), "\x00\x00\x00\x01", 4);
79 memcpy((*buffer)->data() + (*buffer)->size() + 4, ptr, length);
80 (*buffer)->setRange((*buffer)->offset(), (*buffer)->size() + 4 + length);
81 return OK;
82 }
83
84 #if 0
85 static void convertMetaDataToMessageInt32(
86 const sp<MetaData> &meta, sp<AMessage> &msg, uint32_t key, const char *name) {
87 int32_t value;
88 if (meta->findInt32(key, &value)) {
89 msg->setInt32(name, value);
90 }
91 }
92 #endif
93
convertMetaDataToMessageColorAspects(const MetaDataBase * meta,sp<AMessage> & msg)94 static void convertMetaDataToMessageColorAspects(const MetaDataBase *meta, sp<AMessage> &msg) {
95 // 0 values are unspecified
96 int32_t range = 0;
97 int32_t primaries = 0;
98 int32_t transferFunction = 0;
99 int32_t colorMatrix = 0;
100 meta->findInt32(kKeyColorRange, &range);
101 meta->findInt32(kKeyColorPrimaries, &primaries);
102 meta->findInt32(kKeyTransferFunction, &transferFunction);
103 meta->findInt32(kKeyColorMatrix, &colorMatrix);
104 ColorAspects colorAspects;
105 memset(&colorAspects, 0, sizeof(colorAspects));
106 colorAspects.mRange = (ColorAspects::Range)range;
107 colorAspects.mPrimaries = (ColorAspects::Primaries)primaries;
108 colorAspects.mTransfer = (ColorAspects::Transfer)transferFunction;
109 colorAspects.mMatrixCoeffs = (ColorAspects::MatrixCoeffs)colorMatrix;
110
111 int32_t rangeMsg, standardMsg, transferMsg;
112 if (CodecBase::convertCodecColorAspectsToPlatformAspects(
113 colorAspects, &rangeMsg, &standardMsg, &transferMsg) != OK) {
114 return;
115 }
116
117 // save specified values to msg
118 if (rangeMsg != 0) {
119 msg->setInt32("color-range", rangeMsg);
120 }
121 if (standardMsg != 0) {
122 msg->setInt32("color-standard", standardMsg);
123 }
124 if (transferMsg != 0) {
125 msg->setInt32("color-transfer", transferMsg);
126 }
127 }
128
129 /**
130 * Returns true if, and only if, the given format corresponds to HDR10 or HDR10+.
131 */
isHdr10or10Plus(const sp<AMessage> & format)132 static bool isHdr10or10Plus(const sp<AMessage> &format) {
133
134 // if user/container supplied HDR static info without transfer set, assume true
135 if ((format->contains("hdr-static-info") || format->contains("hdr10-plus-info"))
136 && !format->contains("color-transfer")) {
137 return true;
138 }
139 // otherwise, verify that an HDR transfer function is set
140 int32_t transfer;
141 if (format->findInt32("color-transfer", &transfer)) {
142 return transfer == ColorUtils::kColorTransferST2084;
143 }
144 return false;
145 }
146
parseAacProfileFromCsd(const sp<ABuffer> & csd,sp<AMessage> & format)147 static void parseAacProfileFromCsd(const sp<ABuffer> &csd, sp<AMessage> &format) {
148 if (csd->size() < 2) {
149 return;
150 }
151
152 uint16_t audioObjectType = U16_AT((uint8_t*)csd->data());
153 if ((audioObjectType & 0xF800) == 0xF800) {
154 audioObjectType = 32 + ((audioObjectType >> 5) & 0x3F);
155 } else {
156 audioObjectType >>= 11;
157 }
158
159
160 const static ALookup<uint16_t, int32_t> profiles {
161 { 1, AACObjectMain },
162 { 2, AACObjectLC },
163 { 3, AACObjectSSR },
164 { 4, AACObjectLTP },
165 { 5, AACObjectHE },
166 { 6, AACObjectScalable },
167 { 17, AACObjectERLC },
168 { 23, AACObjectLD },
169 { 29, AACObjectHE_PS },
170 { 39, AACObjectELD },
171 { 42, AACObjectXHE },
172 };
173
174 int32_t profile;
175 if (profiles.map(audioObjectType, &profile)) {
176 format->setInt32("profile", profile);
177 }
178 }
179
parseAvcProfileLevelFromAvcc(const uint8_t * ptr,size_t size,sp<AMessage> & format)180 static void parseAvcProfileLevelFromAvcc(const uint8_t *ptr, size_t size, sp<AMessage> &format) {
181 if (size < 4 || ptr[0] != 1) { // configurationVersion == 1
182 return;
183 }
184 const uint8_t profile = ptr[1];
185 const uint8_t constraints = ptr[2];
186 const uint8_t level = ptr[3];
187
188 const static ALookup<uint8_t, int32_t> levels {
189 { 9, AVCLevel1b }, // technically, 9 is only used for High+ profiles
190 { 10, AVCLevel1 },
191 { 11, AVCLevel11 }, // prefer level 1.1 for the value 11
192 { 11, AVCLevel1b },
193 { 12, AVCLevel12 },
194 { 13, AVCLevel13 },
195 { 20, AVCLevel2 },
196 { 21, AVCLevel21 },
197 { 22, AVCLevel22 },
198 { 30, AVCLevel3 },
199 { 31, AVCLevel31 },
200 { 32, AVCLevel32 },
201 { 40, AVCLevel4 },
202 { 41, AVCLevel41 },
203 { 42, AVCLevel42 },
204 { 50, AVCLevel5 },
205 { 51, AVCLevel51 },
206 { 52, AVCLevel52 },
207 { 60, AVCLevel6 },
208 { 61, AVCLevel61 },
209 { 62, AVCLevel62 },
210 };
211 const static ALookup<uint8_t, int32_t> profiles {
212 { 66, AVCProfileBaseline },
213 { 77, AVCProfileMain },
214 { 88, AVCProfileExtended },
215 { 100, AVCProfileHigh },
216 { 110, AVCProfileHigh10 },
217 { 122, AVCProfileHigh422 },
218 { 244, AVCProfileHigh444 },
219 };
220
221 // set profile & level if they are recognized
222 int32_t codecProfile;
223 int32_t codecLevel;
224 if (profiles.map(profile, &codecProfile)) {
225 if (profile == 66 && (constraints & 0x40)) {
226 codecProfile = AVCProfileConstrainedBaseline;
227 } else if (profile == 100 && (constraints & 0x0C) == 0x0C) {
228 codecProfile = AVCProfileConstrainedHigh;
229 }
230 format->setInt32("profile", codecProfile);
231 if (levels.map(level, &codecLevel)) {
232 // for 9 && 11 decide level based on profile and constraint_set3 flag
233 if (level == 11 && (profile == 66 || profile == 77 || profile == 88)) {
234 codecLevel = (constraints & 0x10) ? AVCLevel1b : AVCLevel11;
235 }
236 format->setInt32("level", codecLevel);
237 }
238 }
239 }
240
getDolbyVisionProfileTable()241 static const ALookup<uint8_t, int32_t>& getDolbyVisionProfileTable() {
242 static const ALookup<uint8_t, int32_t> profileTable = {
243 {1, DolbyVisionProfileDvavPen},
244 {3, DolbyVisionProfileDvheDen},
245 {4, DolbyVisionProfileDvheDtr},
246 {5, DolbyVisionProfileDvheStn},
247 {6, DolbyVisionProfileDvheDth},
248 {7, DolbyVisionProfileDvheDtb},
249 {8, DolbyVisionProfileDvheSt},
250 {9, DolbyVisionProfileDvavSe},
251 {10, DolbyVisionProfileDvav110},
252 };
253 return profileTable;
254 }
255
getDolbyVisionLevelsTable()256 static const ALookup<uint8_t, int32_t>& getDolbyVisionLevelsTable() {
257 static const ALookup<uint8_t, int32_t> levelsTable = {
258 {0, DolbyVisionLevelUnknown},
259 {1, DolbyVisionLevelHd24},
260 {2, DolbyVisionLevelHd30},
261 {3, DolbyVisionLevelFhd24},
262 {4, DolbyVisionLevelFhd30},
263 {5, DolbyVisionLevelFhd60},
264 {6, DolbyVisionLevelUhd24},
265 {7, DolbyVisionLevelUhd30},
266 {8, DolbyVisionLevelUhd48},
267 {9, DolbyVisionLevelUhd60},
268 {10, DolbyVisionLevelUhd120},
269 {11, DolbyVisionLevel8k30},
270 {12, DolbyVisionLevel8k60},
271 };
272 return levelsTable;
273 }
parseDolbyVisionProfileLevelFromDvcc(const uint8_t * ptr,size_t size,sp<AMessage> & format)274 static void parseDolbyVisionProfileLevelFromDvcc(const uint8_t *ptr, size_t size, sp<AMessage> &format) {
275 // dv_major.dv_minor Should be 1.0 or 2.1
276 if (size != 24 || ((ptr[0] != 1 || ptr[1] != 0) && (ptr[0] != 2 || ptr[1] != 1))) {
277 ALOGV("Size %zu, dv_major %d, dv_minor %d", size, ptr[0], ptr[1]);
278 return;
279 }
280
281 const uint8_t profile = ptr[2] >> 1;
282 const uint8_t level = ((ptr[2] & 0x1) << 5) | ((ptr[3] >> 3) & 0x1f);
283 const uint8_t rpu_present_flag = (ptr[3] >> 2) & 0x01;
284 const uint8_t el_present_flag = (ptr[3] >> 1) & 0x01;
285 const uint8_t bl_present_flag = (ptr[3] & 0x01);
286 const int32_t bl_compatibility_id = (int32_t)(ptr[4] >> 4);
287
288 ALOGV("profile-level-compatibility value in dv(c|v)c box %d-%d-%d",
289 profile, level, bl_compatibility_id);
290
291 // All Dolby Profiles will have profile and level info in MediaFormat
292 // Profile 8 and 9 will have bl_compatibility_id too.
293 const ALookup<uint8_t, int32_t> &profiles = getDolbyVisionProfileTable();
294 const ALookup<uint8_t, int32_t> &levels = getDolbyVisionLevelsTable();
295
296 // set rpuAssoc
297 if (rpu_present_flag && el_present_flag && !bl_present_flag) {
298 format->setInt32("rpuAssoc", 1);
299 }
300 // set profile & level if they are recognized
301 int32_t codecProfile;
302 int32_t codecLevel;
303 if (profiles.map(profile, &codecProfile)) {
304 format->setInt32("profile", codecProfile);
305 if (codecProfile == DolbyVisionProfileDvheSt ||
306 codecProfile == DolbyVisionProfileDvavSe) {
307 format->setInt32("bl_compatibility_id", bl_compatibility_id);
308 }
309 if (levels.map(level, &codecLevel)) {
310 format->setInt32("level", codecLevel);
311 }
312 }
313 }
314
parseH263ProfileLevelFromD263(const uint8_t * ptr,size_t size,sp<AMessage> & format)315 static void parseH263ProfileLevelFromD263(const uint8_t *ptr, size_t size, sp<AMessage> &format) {
316 if (size < 7) {
317 return;
318 }
319
320 const uint8_t profile = ptr[6];
321 const uint8_t level = ptr[5];
322
323 const static ALookup<uint8_t, int32_t> profiles {
324 { 0, H263ProfileBaseline },
325 { 1, H263ProfileH320Coding },
326 { 2, H263ProfileBackwardCompatible },
327 { 3, H263ProfileISWV2 },
328 { 4, H263ProfileISWV3 },
329 { 5, H263ProfileHighCompression },
330 { 6, H263ProfileInternet },
331 { 7, H263ProfileInterlace },
332 { 8, H263ProfileHighLatency },
333 };
334
335 const static ALookup<uint8_t, int32_t> levels {
336 { 10, H263Level10 },
337 { 20, H263Level20 },
338 { 30, H263Level30 },
339 { 40, H263Level40 },
340 { 45, H263Level45 },
341 { 50, H263Level50 },
342 { 60, H263Level60 },
343 { 70, H263Level70 },
344 };
345
346 // set profile & level if they are recognized
347 int32_t codecProfile;
348 int32_t codecLevel;
349 if (profiles.map(profile, &codecProfile)) {
350 format->setInt32("profile", codecProfile);
351 if (levels.map(level, &codecLevel)) {
352 format->setInt32("level", codecLevel);
353 }
354 }
355 }
356
parseHevcProfileLevelFromHvcc(const uint8_t * ptr,size_t size,sp<AMessage> & format)357 static void parseHevcProfileLevelFromHvcc(const uint8_t *ptr, size_t size, sp<AMessage> &format) {
358 if (size < 13 || ptr[0] != 1) { // configurationVersion == 1
359 return;
360 }
361
362 const uint8_t profile = ptr[1] & 0x1F;
363 const uint8_t tier = (ptr[1] & 0x20) >> 5;
364 const uint8_t level = ptr[12];
365
366 const static ALookup<std::pair<uint8_t, uint8_t>, int32_t> levels {
367 { { 0, 30 }, HEVCMainTierLevel1 },
368 { { 0, 60 }, HEVCMainTierLevel2 },
369 { { 0, 63 }, HEVCMainTierLevel21 },
370 { { 0, 90 }, HEVCMainTierLevel3 },
371 { { 0, 93 }, HEVCMainTierLevel31 },
372 { { 0, 120 }, HEVCMainTierLevel4 },
373 { { 0, 123 }, HEVCMainTierLevel41 },
374 { { 0, 150 }, HEVCMainTierLevel5 },
375 { { 0, 153 }, HEVCMainTierLevel51 },
376 { { 0, 156 }, HEVCMainTierLevel52 },
377 { { 0, 180 }, HEVCMainTierLevel6 },
378 { { 0, 183 }, HEVCMainTierLevel61 },
379 { { 0, 186 }, HEVCMainTierLevel62 },
380 { { 1, 30 }, HEVCHighTierLevel1 },
381 { { 1, 60 }, HEVCHighTierLevel2 },
382 { { 1, 63 }, HEVCHighTierLevel21 },
383 { { 1, 90 }, HEVCHighTierLevel3 },
384 { { 1, 93 }, HEVCHighTierLevel31 },
385 { { 1, 120 }, HEVCHighTierLevel4 },
386 { { 1, 123 }, HEVCHighTierLevel41 },
387 { { 1, 150 }, HEVCHighTierLevel5 },
388 { { 1, 153 }, HEVCHighTierLevel51 },
389 { { 1, 156 }, HEVCHighTierLevel52 },
390 { { 1, 180 }, HEVCHighTierLevel6 },
391 { { 1, 183 }, HEVCHighTierLevel61 },
392 { { 1, 186 }, HEVCHighTierLevel62 },
393 };
394
395 const static ALookup<uint8_t, int32_t> profiles {
396 { 1, HEVCProfileMain },
397 { 2, HEVCProfileMain10 },
398 // use Main for Main Still Picture decoding
399 { 3, HEVCProfileMain },
400 };
401
402 // set profile & level if they are recognized
403 int32_t codecProfile;
404 int32_t codecLevel;
405 if (!profiles.map(profile, &codecProfile)) {
406 if (ptr[2] & 0x40 /* general compatibility flag 1 */) {
407 // Note that this case covers Main Still Picture too
408 codecProfile = HEVCProfileMain;
409 } else if (ptr[2] & 0x20 /* general compatibility flag 2 */) {
410 codecProfile = HEVCProfileMain10;
411 } else {
412 return;
413 }
414 }
415
416 // bump to HDR profile
417 if (isHdr10or10Plus(format) && codecProfile == HEVCProfileMain10) {
418 if (format->contains("hdr10-plus-info")) {
419 codecProfile = HEVCProfileMain10HDR10Plus;
420 } else {
421 codecProfile = HEVCProfileMain10HDR10;
422 }
423 }
424
425 format->setInt32("profile", codecProfile);
426 if (levels.map(std::make_pair(tier, level), &codecLevel)) {
427 format->setInt32("level", codecLevel);
428 }
429 }
430
parseMpeg2ProfileLevelFromHeader(const uint8_t * data,size_t size,sp<AMessage> & format)431 static void parseMpeg2ProfileLevelFromHeader(
432 const uint8_t *data, size_t size, sp<AMessage> &format) {
433 // find sequence extension
434 const uint8_t *seq = (const uint8_t*)memmem(data, size, "\x00\x00\x01\xB5", 4);
435 if (seq != NULL && seq + 5 < data + size) {
436 const uint8_t start_code = seq[4] >> 4;
437 if (start_code != 1 /* sequence extension ID */) {
438 return;
439 }
440 const uint8_t indication = ((seq[4] & 0xF) << 4) | ((seq[5] & 0xF0) >> 4);
441
442 const static ALookup<uint8_t, int32_t> profiles {
443 { 0x50, MPEG2ProfileSimple },
444 { 0x40, MPEG2ProfileMain },
445 { 0x30, MPEG2ProfileSNR },
446 { 0x20, MPEG2ProfileSpatial },
447 { 0x10, MPEG2ProfileHigh },
448 };
449
450 const static ALookup<uint8_t, int32_t> levels {
451 { 0x0A, MPEG2LevelLL },
452 { 0x08, MPEG2LevelML },
453 { 0x06, MPEG2LevelH14 },
454 { 0x04, MPEG2LevelHL },
455 { 0x02, MPEG2LevelHP },
456 };
457
458 const static ALookup<uint8_t,
459 std::pair<int32_t, int32_t>> escapes {
460 /* unsupported
461 { 0x8E, { XXX_MPEG2ProfileMultiView, MPEG2LevelLL } },
462 { 0x8D, { XXX_MPEG2ProfileMultiView, MPEG2LevelML } },
463 { 0x8B, { XXX_MPEG2ProfileMultiView, MPEG2LevelH14 } },
464 { 0x8A, { XXX_MPEG2ProfileMultiView, MPEG2LevelHL } }, */
465 { 0x85, { MPEG2Profile422, MPEG2LevelML } },
466 { 0x82, { MPEG2Profile422, MPEG2LevelHL } },
467 };
468
469 int32_t profile;
470 int32_t level;
471 std::pair<int32_t, int32_t> profileLevel;
472 if (escapes.map(indication, &profileLevel)) {
473 format->setInt32("profile", profileLevel.first);
474 format->setInt32("level", profileLevel.second);
475 } else if (profiles.map(indication & 0x70, &profile)) {
476 format->setInt32("profile", profile);
477 if (levels.map(indication & 0xF, &level)) {
478 format->setInt32("level", level);
479 }
480 }
481 }
482 }
483
parseMpeg2ProfileLevelFromEsds(ESDS & esds,sp<AMessage> & format)484 static void parseMpeg2ProfileLevelFromEsds(ESDS &esds, sp<AMessage> &format) {
485 // esds seems to only contain the profile for MPEG-2
486 uint8_t objType;
487 if (esds.getObjectTypeIndication(&objType) == OK) {
488 const static ALookup<uint8_t, int32_t> profiles{
489 { 0x60, MPEG2ProfileSimple },
490 { 0x61, MPEG2ProfileMain },
491 { 0x62, MPEG2ProfileSNR },
492 { 0x63, MPEG2ProfileSpatial },
493 { 0x64, MPEG2ProfileHigh },
494 { 0x65, MPEG2Profile422 },
495 };
496
497 int32_t profile;
498 if (profiles.map(objType, &profile)) {
499 format->setInt32("profile", profile);
500 }
501 }
502 }
503
parseMpeg4ProfileLevelFromCsd(const sp<ABuffer> & csd,sp<AMessage> & format)504 static void parseMpeg4ProfileLevelFromCsd(const sp<ABuffer> &csd, sp<AMessage> &format) {
505 const uint8_t *data = csd->data();
506 // find visual object sequence
507 const uint8_t *seq = (const uint8_t*)memmem(data, csd->size(), "\x00\x00\x01\xB0", 4);
508 if (seq != NULL && seq + 4 < data + csd->size()) {
509 const uint8_t indication = seq[4];
510
511 const static ALookup<uint8_t,
512 std::pair<int32_t, int32_t>> table {
513 { 0b00000001, { MPEG4ProfileSimple, MPEG4Level1 } },
514 { 0b00000010, { MPEG4ProfileSimple, MPEG4Level2 } },
515 { 0b00000011, { MPEG4ProfileSimple, MPEG4Level3 } },
516 { 0b00000100, { MPEG4ProfileSimple, MPEG4Level4a } },
517 { 0b00000101, { MPEG4ProfileSimple, MPEG4Level5 } },
518 { 0b00000110, { MPEG4ProfileSimple, MPEG4Level6 } },
519 { 0b00001000, { MPEG4ProfileSimple, MPEG4Level0 } },
520 { 0b00001001, { MPEG4ProfileSimple, MPEG4Level0b } },
521 { 0b00010000, { MPEG4ProfileSimpleScalable, MPEG4Level0 } },
522 { 0b00010001, { MPEG4ProfileSimpleScalable, MPEG4Level1 } },
523 { 0b00010010, { MPEG4ProfileSimpleScalable, MPEG4Level2 } },
524 /* unsupported
525 { 0b00011101, { XXX_MPEG4ProfileSimpleScalableER, MPEG4Level0 } },
526 { 0b00011110, { XXX_MPEG4ProfileSimpleScalableER, MPEG4Level1 } },
527 { 0b00011111, { XXX_MPEG4ProfileSimpleScalableER, MPEG4Level2 } }, */
528 { 0b00100001, { MPEG4ProfileCore, MPEG4Level1 } },
529 { 0b00100010, { MPEG4ProfileCore, MPEG4Level2 } },
530 { 0b00110010, { MPEG4ProfileMain, MPEG4Level2 } },
531 { 0b00110011, { MPEG4ProfileMain, MPEG4Level3 } },
532 { 0b00110100, { MPEG4ProfileMain, MPEG4Level4 } },
533 /* deprecated
534 { 0b01000010, { MPEG4ProfileNbit, MPEG4Level2 } }, */
535 { 0b01010001, { MPEG4ProfileScalableTexture, MPEG4Level1 } },
536 { 0b01100001, { MPEG4ProfileSimpleFace, MPEG4Level1 } },
537 { 0b01100010, { MPEG4ProfileSimpleFace, MPEG4Level2 } },
538 { 0b01100011, { MPEG4ProfileSimpleFBA, MPEG4Level1 } },
539 { 0b01100100, { MPEG4ProfileSimpleFBA, MPEG4Level2 } },
540 { 0b01110001, { MPEG4ProfileBasicAnimated, MPEG4Level1 } },
541 { 0b01110010, { MPEG4ProfileBasicAnimated, MPEG4Level2 } },
542 { 0b10000001, { MPEG4ProfileHybrid, MPEG4Level1 } },
543 { 0b10000010, { MPEG4ProfileHybrid, MPEG4Level2 } },
544 { 0b10010001, { MPEG4ProfileAdvancedRealTime, MPEG4Level1 } },
545 { 0b10010010, { MPEG4ProfileAdvancedRealTime, MPEG4Level2 } },
546 { 0b10010011, { MPEG4ProfileAdvancedRealTime, MPEG4Level3 } },
547 { 0b10010100, { MPEG4ProfileAdvancedRealTime, MPEG4Level4 } },
548 { 0b10100001, { MPEG4ProfileCoreScalable, MPEG4Level1 } },
549 { 0b10100010, { MPEG4ProfileCoreScalable, MPEG4Level2 } },
550 { 0b10100011, { MPEG4ProfileCoreScalable, MPEG4Level3 } },
551 { 0b10110001, { MPEG4ProfileAdvancedCoding, MPEG4Level1 } },
552 { 0b10110010, { MPEG4ProfileAdvancedCoding, MPEG4Level2 } },
553 { 0b10110011, { MPEG4ProfileAdvancedCoding, MPEG4Level3 } },
554 { 0b10110100, { MPEG4ProfileAdvancedCoding, MPEG4Level4 } },
555 { 0b11000001, { MPEG4ProfileAdvancedCore, MPEG4Level1 } },
556 { 0b11000010, { MPEG4ProfileAdvancedCore, MPEG4Level2 } },
557 { 0b11010001, { MPEG4ProfileAdvancedScalable, MPEG4Level1 } },
558 { 0b11010010, { MPEG4ProfileAdvancedScalable, MPEG4Level2 } },
559 { 0b11010011, { MPEG4ProfileAdvancedScalable, MPEG4Level3 } },
560 /* unsupported
561 { 0b11100001, { XXX_MPEG4ProfileSimpleStudio, MPEG4Level1 } },
562 { 0b11100010, { XXX_MPEG4ProfileSimpleStudio, MPEG4Level2 } },
563 { 0b11100011, { XXX_MPEG4ProfileSimpleStudio, MPEG4Level3 } },
564 { 0b11100100, { XXX_MPEG4ProfileSimpleStudio, MPEG4Level4 } },
565 { 0b11100101, { XXX_MPEG4ProfileCoreStudio, MPEG4Level1 } },
566 { 0b11100110, { XXX_MPEG4ProfileCoreStudio, MPEG4Level2 } },
567 { 0b11100111, { XXX_MPEG4ProfileCoreStudio, MPEG4Level3 } },
568 { 0b11101000, { XXX_MPEG4ProfileCoreStudio, MPEG4Level4 } },
569 { 0b11101011, { XXX_MPEG4ProfileSimpleStudio, MPEG4Level5 } },
570 { 0b11101100, { XXX_MPEG4ProfileSimpleStudio, MPEG4Level6 } }, */
571 { 0b11110000, { MPEG4ProfileAdvancedSimple, MPEG4Level0 } },
572 { 0b11110001, { MPEG4ProfileAdvancedSimple, MPEG4Level1 } },
573 { 0b11110010, { MPEG4ProfileAdvancedSimple, MPEG4Level2 } },
574 { 0b11110011, { MPEG4ProfileAdvancedSimple, MPEG4Level3 } },
575 { 0b11110100, { MPEG4ProfileAdvancedSimple, MPEG4Level4 } },
576 { 0b11110101, { MPEG4ProfileAdvancedSimple, MPEG4Level5 } },
577 { 0b11110111, { MPEG4ProfileAdvancedSimple, MPEG4Level3b } },
578 /* deprecated
579 { 0b11111000, { XXX_MPEG4ProfileFineGranularityScalable, MPEG4Level0 } },
580 { 0b11111001, { XXX_MPEG4ProfileFineGranularityScalable, MPEG4Level1 } },
581 { 0b11111010, { XXX_MPEG4ProfileFineGranularityScalable, MPEG4Level2 } },
582 { 0b11111011, { XXX_MPEG4ProfileFineGranularityScalable, MPEG4Level3 } },
583 { 0b11111100, { XXX_MPEG4ProfileFineGranularityScalable, MPEG4Level4 } },
584 { 0b11111101, { XXX_MPEG4ProfileFineGranularityScalable, MPEG4Level5 } }, */
585 };
586
587 std::pair<int32_t, int32_t> profileLevel;
588 if (table.map(indication, &profileLevel)) {
589 format->setInt32("profile", profileLevel.first);
590 format->setInt32("level", profileLevel.second);
591 }
592 }
593 }
594
parseVp9ProfileLevelFromCsd(const sp<ABuffer> & csd,sp<AMessage> & format)595 static void parseVp9ProfileLevelFromCsd(const sp<ABuffer> &csd, sp<AMessage> &format) {
596 const uint8_t *data = csd->data();
597 size_t remaining = csd->size();
598
599 while (remaining >= 2) {
600 const uint8_t id = data[0];
601 const uint8_t length = data[1];
602 remaining -= 2;
603 data += 2;
604 if (length > remaining) {
605 break;
606 }
607 switch (id) {
608 case 1 /* profileId */:
609 if (length >= 1) {
610 const static ALookup<uint8_t, int32_t> profiles {
611 { 0, VP9Profile0 },
612 { 1, VP9Profile1 },
613 { 2, VP9Profile2 },
614 { 3, VP9Profile3 },
615 };
616
617 const static ALookup<int32_t, int32_t> toHdr10 {
618 { VP9Profile2, VP9Profile2HDR },
619 { VP9Profile3, VP9Profile3HDR },
620 };
621
622 const static ALookup<int32_t, int32_t> toHdr10Plus {
623 { VP9Profile2, VP9Profile2HDR10Plus },
624 { VP9Profile3, VP9Profile3HDR10Plus },
625 };
626
627 int32_t profile;
628 if (profiles.map(data[0], &profile)) {
629 // convert to HDR profile
630 if (isHdr10or10Plus(format)) {
631 if (format->contains("hdr10-plus-info")) {
632 toHdr10Plus.lookup(profile, &profile);
633 } else {
634 toHdr10.lookup(profile, &profile);
635 }
636 }
637
638 format->setInt32("profile", profile);
639 }
640 }
641 break;
642 case 2 /* levelId */:
643 if (length >= 1) {
644 const static ALookup<uint8_t, int32_t> levels {
645 { 10, VP9Level1 },
646 { 11, VP9Level11 },
647 { 20, VP9Level2 },
648 { 21, VP9Level21 },
649 { 30, VP9Level3 },
650 { 31, VP9Level31 },
651 { 40, VP9Level4 },
652 { 41, VP9Level41 },
653 { 50, VP9Level5 },
654 { 51, VP9Level51 },
655 { 52, VP9Level52 },
656 { 60, VP9Level6 },
657 { 61, VP9Level61 },
658 { 62, VP9Level62 },
659 };
660
661 int32_t level;
662 if (levels.map(data[0], &level)) {
663 format->setInt32("level", level);
664 }
665 }
666 break;
667 default:
668 break;
669 }
670 remaining -= length;
671 data += length;
672 }
673 }
674
parseAV1ProfileLevelFromCsd(const sp<ABuffer> & csd,sp<AMessage> & format)675 static void parseAV1ProfileLevelFromCsd(const sp<ABuffer> &csd, sp<AMessage> &format) {
676 // Parse CSD structure to extract profile level information
677 // https://aomediacodec.github.io/av1-isobmff/#av1codecconfigurationbox
678 const uint8_t *data = csd->data();
679 size_t remaining = csd->size();
680 if (remaining < 4 || data[0] != 0x81) { // configurationVersion == 1
681 return;
682 }
683 uint8_t profileData = (data[1] & 0xE0) >> 5;
684 uint8_t levelData = data[1] & 0x1F;
685 uint8_t highBitDepth = (data[2] & 0x40) >> 6;
686
687 const static ALookup<std::pair<uint8_t, uint8_t>, int32_t> profiles {
688 { { 0, 0 }, AV1ProfileMain8 },
689 { { 1, 0 }, AV1ProfileMain10 },
690 };
691
692 int32_t profile;
693 if (profiles.map(std::make_pair(highBitDepth, profileData), &profile)) {
694 // bump to HDR profile
695 if (isHdr10or10Plus(format) && profile == AV1ProfileMain10) {
696 if (format->contains("hdr10-plus-info")) {
697 profile = AV1ProfileMain10HDR10Plus;
698 } else {
699 profile = AV1ProfileMain10HDR10;
700 }
701 }
702 format->setInt32("profile", profile);
703 }
704 const static ALookup<uint8_t, int32_t> levels {
705 { 0, AV1Level2 },
706 { 1, AV1Level21 },
707 { 2, AV1Level22 },
708 { 3, AV1Level23 },
709 { 4, AV1Level3 },
710 { 5, AV1Level31 },
711 { 6, AV1Level32 },
712 { 7, AV1Level33 },
713 { 8, AV1Level4 },
714 { 9, AV1Level41 },
715 { 10, AV1Level42 },
716 { 11, AV1Level43 },
717 { 12, AV1Level5 },
718 { 13, AV1Level51 },
719 { 14, AV1Level52 },
720 { 15, AV1Level53 },
721 { 16, AV1Level6 },
722 { 17, AV1Level61 },
723 { 18, AV1Level62 },
724 { 19, AV1Level63 },
725 { 20, AV1Level7 },
726 { 21, AV1Level71 },
727 { 22, AV1Level72 },
728 { 23, AV1Level73 },
729 };
730
731 int32_t level;
732 if (levels.map(levelData, &level)) {
733 format->setInt32("level", level);
734 }
735 }
736
737
738 static std::vector<std::pair<const char *, uint32_t>> stringMappings {
739 {
740 { "album", kKeyAlbum },
741 { "albumartist", kKeyAlbumArtist },
742 { "artist", kKeyArtist },
743 { "author", kKeyAuthor },
744 { "cdtracknum", kKeyCDTrackNumber },
745 { "compilation", kKeyCompilation },
746 { "composer", kKeyComposer },
747 { "date", kKeyDate },
748 { "discnum", kKeyDiscNumber },
749 { "genre", kKeyGenre },
750 { "location", kKeyLocation },
751 { "lyricist", kKeyWriter },
752 { "manufacturer", kKeyManufacturer },
753 { "title", kKeyTitle },
754 { "year", kKeyYear },
755 }
756 };
757
758 static std::vector<std::pair<const char *, uint32_t>> floatMappings {
759 {
760 { "capture-rate", kKeyCaptureFramerate },
761 }
762 };
763
764 static std::vector<std::pair<const char*, uint32_t>> int64Mappings {
765 {
766 { "exif-offset", kKeyExifOffset},
767 { "exif-size", kKeyExifSize},
768 { "xmp-offset", kKeyXmpOffset},
769 { "xmp-size", kKeyXmpSize},
770 { "target-time", kKeyTargetTime},
771 { "thumbnail-time", kKeyThumbnailTime},
772 { "timeUs", kKeyTime},
773 { "durationUs", kKeyDuration},
774 { "sample-file-offset", kKeySampleFileOffset},
775 { "last-sample-index-in-chunk", kKeyLastSampleIndexInChunk},
776 { "sample-time-before-append", kKeySampleTimeBeforeAppend},
777 }
778 };
779
780 static std::vector<std::pair<const char *, uint32_t>> int32Mappings {
781 {
782 { "loop", kKeyAutoLoop },
783 { "time-scale", kKeyTimeScale },
784 { "crypto-mode", kKeyCryptoMode },
785 { "crypto-default-iv-size", kKeyCryptoDefaultIVSize },
786 { "crypto-encrypted-byte-block", kKeyEncryptedByteBlock },
787 { "crypto-skip-byte-block", kKeySkipByteBlock },
788 { "frame-count", kKeyFrameCount },
789 { "max-bitrate", kKeyMaxBitRate },
790 { "pcm-big-endian", kKeyPcmBigEndian },
791 { "temporal-layer-count", kKeyTemporalLayerCount },
792 { "temporal-layer-id", kKeyTemporalLayerId },
793 { "thumbnail-width", kKeyThumbnailWidth },
794 { "thumbnail-height", kKeyThumbnailHeight },
795 { "track-id", kKeyTrackID },
796 { "valid-samples", kKeyValidSamples },
797 }
798 };
799
800 static std::vector<std::pair<const char *, uint32_t>> bufferMappings {
801 {
802 { "albumart", kKeyAlbumArt },
803 { "audio-presentation-info", kKeyAudioPresentationInfo },
804 { "pssh", kKeyPssh },
805 { "crypto-iv", kKeyCryptoIV },
806 { "crypto-key", kKeyCryptoKey },
807 { "crypto-encrypted-sizes", kKeyEncryptedSizes },
808 { "crypto-plain-sizes", kKeyPlainSizes },
809 { "icc-profile", kKeyIccProfile },
810 { "sei", kKeySEI },
811 { "text-format-data", kKeyTextFormatData },
812 { "thumbnail-csd-hevc", kKeyThumbnailHVCC },
813 { "slow-motion-markers", kKeySlowMotionMarkers },
814 { "thumbnail-csd-av1c", kKeyThumbnailAV1C },
815 }
816 };
817
818 static std::vector<std::pair<const char *, uint32_t>> CSDMappings {
819 {
820 { "csd-0", kKeyOpaqueCSD0 },
821 { "csd-1", kKeyOpaqueCSD1 },
822 { "csd-2", kKeyOpaqueCSD2 },
823 }
824 };
825
convertMessageToMetaDataFromMappings(const sp<AMessage> & msg,sp<MetaData> & meta)826 void convertMessageToMetaDataFromMappings(const sp<AMessage> &msg, sp<MetaData> &meta) {
827 for (auto elem : stringMappings) {
828 AString value;
829 if (msg->findString(elem.first, &value)) {
830 meta->setCString(elem.second, value.c_str());
831 }
832 }
833
834 for (auto elem : floatMappings) {
835 float value;
836 if (msg->findFloat(elem.first, &value)) {
837 meta->setFloat(elem.second, value);
838 }
839 }
840
841 for (auto elem : int64Mappings) {
842 int64_t value;
843 if (msg->findInt64(elem.first, &value)) {
844 meta->setInt64(elem.second, value);
845 }
846 }
847
848 for (auto elem : int32Mappings) {
849 int32_t value;
850 if (msg->findInt32(elem.first, &value)) {
851 meta->setInt32(elem.second, value);
852 }
853 }
854
855 for (auto elem : bufferMappings) {
856 sp<ABuffer> value;
857 if (msg->findBuffer(elem.first, &value)) {
858 meta->setData(elem.second,
859 MetaDataBase::Type::TYPE_NONE, value->data(), value->size());
860 }
861 }
862
863 for (auto elem : CSDMappings) {
864 sp<ABuffer> value;
865 if (msg->findBuffer(elem.first, &value)) {
866 meta->setData(elem.second,
867 MetaDataBase::Type::TYPE_NONE, value->data(), value->size());
868 }
869 }
870 }
871
convertMetaDataToMessageFromMappings(const MetaDataBase * meta,sp<AMessage> format)872 void convertMetaDataToMessageFromMappings(const MetaDataBase *meta, sp<AMessage> format) {
873 for (auto elem : stringMappings) {
874 const char *value;
875 if (meta->findCString(elem.second, &value)) {
876 format->setString(elem.first, value, strlen(value));
877 }
878 }
879
880 for (auto elem : floatMappings) {
881 float value;
882 if (meta->findFloat(elem.second, &value)) {
883 format->setFloat(elem.first, value);
884 }
885 }
886
887 for (auto elem : int64Mappings) {
888 int64_t value;
889 if (meta->findInt64(elem.second, &value)) {
890 format->setInt64(elem.first, value);
891 }
892 }
893
894 for (auto elem : int32Mappings) {
895 int32_t value;
896 if (meta->findInt32(elem.second, &value)) {
897 format->setInt32(elem.first, value);
898 }
899 }
900
901 for (auto elem : bufferMappings) {
902 uint32_t type;
903 const void* data;
904 size_t size;
905 if (meta->findData(elem.second, &type, &data, &size)) {
906 sp<ABuffer> buf = ABuffer::CreateAsCopy(data, size);
907 format->setBuffer(elem.first, buf);
908 }
909 }
910
911 for (auto elem : CSDMappings) {
912 uint32_t type;
913 const void* data;
914 size_t size;
915 if (meta->findData(elem.second, &type, &data, &size)) {
916 sp<ABuffer> buf = ABuffer::CreateAsCopy(data, size);
917 buf->meta()->setInt32("csd", true);
918 buf->meta()->setInt64("timeUs", 0);
919 format->setBuffer(elem.first, buf);
920 }
921 }
922 }
923
convertMetaDataToMessage(const sp<MetaData> & meta,sp<AMessage> * format)924 status_t convertMetaDataToMessage(
925 const sp<MetaData> &meta, sp<AMessage> *format) {
926 return convertMetaDataToMessage(meta.get(), format);
927 }
928
convertMetaDataToMessage(const MetaDataBase * meta,sp<AMessage> * format)929 status_t convertMetaDataToMessage(
930 const MetaDataBase *meta, sp<AMessage> *format) {
931
932 format->clear();
933
934 if (meta == NULL) {
935 ALOGE("convertMetaDataToMessage: NULL input");
936 return BAD_VALUE;
937 }
938
939 const char *mime;
940 if (!meta->findCString(kKeyMIMEType, &mime)) {
941 return BAD_VALUE;
942 }
943
944 sp<AMessage> msg = new AMessage;
945 msg->setString("mime", mime);
946
947 convertMetaDataToMessageFromMappings(meta, msg);
948
949 uint32_t type;
950 const void *data;
951 size_t size;
952 if (meta->findData(kKeyCASessionID, &type, &data, &size)) {
953 sp<ABuffer> buffer = new (std::nothrow) ABuffer(size);
954 if (buffer.get() == NULL || buffer->base() == NULL) {
955 return NO_MEMORY;
956 }
957
958 msg->setBuffer("ca-session-id", buffer);
959 memcpy(buffer->data(), data, size);
960 }
961
962 if (meta->findData(kKeyCAPrivateData, &type, &data, &size)) {
963 sp<ABuffer> buffer = new (std::nothrow) ABuffer(size);
964 if (buffer.get() == NULL || buffer->base() == NULL) {
965 return NO_MEMORY;
966 }
967
968 msg->setBuffer("ca-private-data", buffer);
969 memcpy(buffer->data(), data, size);
970 }
971
972 int32_t systemId;
973 if (meta->findInt32(kKeyCASystemID, &systemId)) {
974 msg->setInt32("ca-system-id", systemId);
975 }
976
977 if (!strncasecmp("video/scrambled", mime, 15)
978 || !strncasecmp("audio/scrambled", mime, 15)) {
979
980 *format = msg;
981 return OK;
982 }
983
984 int64_t durationUs;
985 if (meta->findInt64(kKeyDuration, &durationUs)) {
986 msg->setInt64("durationUs", durationUs);
987 }
988
989 int32_t avgBitRate = 0;
990 if (meta->findInt32(kKeyBitRate, &avgBitRate) && avgBitRate > 0) {
991 msg->setInt32("bitrate", avgBitRate);
992 }
993
994 int32_t maxBitRate;
995 if (meta->findInt32(kKeyMaxBitRate, &maxBitRate)
996 && maxBitRate > 0 && maxBitRate >= avgBitRate) {
997 msg->setInt32("max-bitrate", maxBitRate);
998 }
999
1000 int32_t isSync;
1001 if (meta->findInt32(kKeyIsSyncFrame, &isSync) && isSync != 0) {
1002 msg->setInt32("is-sync-frame", 1);
1003 }
1004
1005 const char *lang;
1006 if (meta->findCString(kKeyMediaLanguage, &lang)) {
1007 msg->setString("language", lang);
1008 }
1009
1010 if (!strncasecmp("video/", mime, 6) ||
1011 !strncasecmp("image/", mime, 6)) {
1012 int32_t width, height;
1013 if (!meta->findInt32(kKeyWidth, &width)
1014 || !meta->findInt32(kKeyHeight, &height)) {
1015 return BAD_VALUE;
1016 }
1017
1018 msg->setInt32("width", width);
1019 msg->setInt32("height", height);
1020
1021 int32_t displayWidth, displayHeight;
1022 if (meta->findInt32(kKeyDisplayWidth, &displayWidth)
1023 && meta->findInt32(kKeyDisplayHeight, &displayHeight)) {
1024 msg->setInt32("display-width", displayWidth);
1025 msg->setInt32("display-height", displayHeight);
1026 }
1027
1028 int32_t sarWidth, sarHeight;
1029 if (meta->findInt32(kKeySARWidth, &sarWidth)
1030 && meta->findInt32(kKeySARHeight, &sarHeight)) {
1031 msg->setInt32("sar-width", sarWidth);
1032 msg->setInt32("sar-height", sarHeight);
1033 }
1034
1035 if (!strncasecmp("image/", mime, 6)) {
1036 int32_t tileWidth, tileHeight, gridRows, gridCols;
1037 if (meta->findInt32(kKeyTileWidth, &tileWidth)
1038 && meta->findInt32(kKeyTileHeight, &tileHeight)
1039 && meta->findInt32(kKeyGridRows, &gridRows)
1040 && meta->findInt32(kKeyGridCols, &gridCols)) {
1041 msg->setInt32("tile-width", tileWidth);
1042 msg->setInt32("tile-height", tileHeight);
1043 msg->setInt32("grid-rows", gridRows);
1044 msg->setInt32("grid-cols", gridCols);
1045 }
1046 int32_t isPrimary;
1047 if (meta->findInt32(kKeyTrackIsDefault, &isPrimary) && isPrimary) {
1048 msg->setInt32("is-default", 1);
1049 }
1050 }
1051
1052 int32_t colorFormat;
1053 if (meta->findInt32(kKeyColorFormat, &colorFormat)) {
1054 msg->setInt32("color-format", colorFormat);
1055 }
1056
1057 int32_t cropLeft, cropTop, cropRight, cropBottom;
1058 if (meta->findRect(kKeyCropRect,
1059 &cropLeft,
1060 &cropTop,
1061 &cropRight,
1062 &cropBottom)) {
1063 msg->setRect("crop", cropLeft, cropTop, cropRight, cropBottom);
1064 }
1065
1066 int32_t rotationDegrees;
1067 if (meta->findInt32(kKeyRotation, &rotationDegrees)) {
1068 msg->setInt32("rotation-degrees", rotationDegrees);
1069 }
1070
1071 uint32_t type;
1072 const void *data;
1073 size_t size;
1074 if (meta->findData(kKeyHdrStaticInfo, &type, &data, &size)
1075 && type == 'hdrS' && size == sizeof(HDRStaticInfo)) {
1076 ColorUtils::setHDRStaticInfoIntoFormat(*(HDRStaticInfo*)data, msg);
1077 }
1078
1079 if (meta->findData(kKeyHdr10PlusInfo, &type, &data, &size)
1080 && size > 0) {
1081 sp<ABuffer> buffer = new (std::nothrow) ABuffer(size);
1082 if (buffer.get() == NULL || buffer->base() == NULL) {
1083 return NO_MEMORY;
1084 }
1085 memcpy(buffer->data(), data, size);
1086 msg->setBuffer("hdr10-plus-info", buffer);
1087 }
1088
1089 convertMetaDataToMessageColorAspects(meta, msg);
1090 } else if (!strncasecmp("audio/", mime, 6)) {
1091 int32_t numChannels, sampleRate;
1092 if (!meta->findInt32(kKeyChannelCount, &numChannels)
1093 || !meta->findInt32(kKeySampleRate, &sampleRate)) {
1094 return BAD_VALUE;
1095 }
1096
1097 msg->setInt32("channel-count", numChannels);
1098 msg->setInt32("sample-rate", sampleRate);
1099
1100 int32_t bitsPerSample;
1101 if (meta->findInt32(kKeyBitsPerSample, &bitsPerSample)) {
1102 msg->setInt32("bits-per-sample", bitsPerSample);
1103 }
1104
1105 int32_t channelMask;
1106 if (meta->findInt32(kKeyChannelMask, &channelMask)) {
1107 msg->setInt32("channel-mask", channelMask);
1108 }
1109
1110 int32_t delay = 0;
1111 if (meta->findInt32(kKeyEncoderDelay, &delay)) {
1112 msg->setInt32("encoder-delay", delay);
1113 }
1114 int32_t padding = 0;
1115 if (meta->findInt32(kKeyEncoderPadding, &padding)) {
1116 msg->setInt32("encoder-padding", padding);
1117 }
1118
1119 int32_t isADTS;
1120 if (meta->findInt32(kKeyIsADTS, &isADTS)) {
1121 msg->setInt32("is-adts", isADTS);
1122 }
1123
1124 int32_t mpeghProfileLevelIndication;
1125 if (meta->findInt32(kKeyMpeghProfileLevelIndication, &mpeghProfileLevelIndication)) {
1126 msg->setInt32(AMEDIAFORMAT_KEY_MPEGH_PROFILE_LEVEL_INDICATION,
1127 mpeghProfileLevelIndication);
1128 }
1129 int32_t mpeghReferenceChannelLayout;
1130 if (meta->findInt32(kKeyMpeghReferenceChannelLayout, &mpeghReferenceChannelLayout)) {
1131 msg->setInt32(AMEDIAFORMAT_KEY_MPEGH_REFERENCE_CHANNEL_LAYOUT,
1132 mpeghReferenceChannelLayout);
1133 }
1134 if (meta->findData(kKeyMpeghCompatibleSets, &type, &data, &size)) {
1135 sp<ABuffer> buffer = new (std::nothrow) ABuffer(size);
1136 if (buffer.get() == NULL || buffer->base() == NULL) {
1137 return NO_MEMORY;
1138 }
1139 msg->setBuffer(AMEDIAFORMAT_KEY_MPEGH_COMPATIBLE_SETS, buffer);
1140 memcpy(buffer->data(), data, size);
1141 }
1142
1143 int32_t aacProfile = -1;
1144 if (meta->findInt32(kKeyAACAOT, &aacProfile)) {
1145 msg->setInt32("aac-profile", aacProfile);
1146 }
1147
1148 int32_t pcmEncoding;
1149 if (meta->findInt32(kKeyPcmEncoding, &pcmEncoding)) {
1150 msg->setInt32("pcm-encoding", pcmEncoding);
1151 }
1152
1153 int32_t hapticChannelCount;
1154 if (meta->findInt32(kKeyHapticChannelCount, &hapticChannelCount)) {
1155 msg->setInt32("haptic-channel-count", hapticChannelCount);
1156 }
1157 }
1158
1159 int32_t maxInputSize;
1160 if (meta->findInt32(kKeyMaxInputSize, &maxInputSize)) {
1161 msg->setInt32("max-input-size", maxInputSize);
1162 }
1163
1164 int32_t maxWidth;
1165 if (meta->findInt32(kKeyMaxWidth, &maxWidth)) {
1166 msg->setInt32("max-width", maxWidth);
1167 }
1168
1169 int32_t maxHeight;
1170 if (meta->findInt32(kKeyMaxHeight, &maxHeight)) {
1171 msg->setInt32("max-height", maxHeight);
1172 }
1173
1174 int32_t rotationDegrees;
1175 if (meta->findInt32(kKeyRotation, &rotationDegrees)) {
1176 msg->setInt32("rotation-degrees", rotationDegrees);
1177 }
1178
1179 int32_t fps;
1180 if (meta->findInt32(kKeyFrameRate, &fps) && fps > 0) {
1181 msg->setInt32("frame-rate", fps);
1182 }
1183
1184 if (meta->findData(kKeyAVCC, &type, &data, &size)) {
1185 // Parse the AVCDecoderConfigurationRecord
1186
1187 const uint8_t *ptr = (const uint8_t *)data;
1188
1189 if (size < 7 || ptr[0] != 1) { // configurationVersion == 1
1190 ALOGE("b/23680780");
1191 return BAD_VALUE;
1192 }
1193
1194 parseAvcProfileLevelFromAvcc(ptr, size, msg);
1195
1196 // There is decodable content out there that fails the following
1197 // assertion, let's be lenient for now...
1198 // CHECK((ptr[4] >> 2) == 0x3f); // reserved
1199
1200 // we can get lengthSize value from 1 + (ptr[4] & 3)
1201
1202 // commented out check below as H264_QVGA_500_NO_AUDIO.3gp
1203 // violates it...
1204 // CHECK((ptr[5] >> 5) == 7); // reserved
1205
1206 size_t numSeqParameterSets = ptr[5] & 31;
1207
1208 ptr += 6;
1209 size -= 6;
1210
1211 sp<ABuffer> buffer = new (std::nothrow) ABuffer(1024);
1212 if (buffer.get() == NULL || buffer->base() == NULL) {
1213 return NO_MEMORY;
1214 }
1215 buffer->setRange(0, 0);
1216
1217 for (size_t i = 0; i < numSeqParameterSets; ++i) {
1218 if (size < 2) {
1219 ALOGE("b/23680780");
1220 return BAD_VALUE;
1221 }
1222 size_t length = U16_AT(ptr);
1223
1224 ptr += 2;
1225 size -= 2;
1226
1227 if (size < length) {
1228 return BAD_VALUE;
1229 }
1230 status_t err = copyNALUToABuffer(&buffer, ptr, length);
1231 if (err != OK) {
1232 return err;
1233 }
1234
1235 ptr += length;
1236 size -= length;
1237 }
1238
1239 buffer->meta()->setInt32("csd", true);
1240 buffer->meta()->setInt64("timeUs", 0);
1241
1242 msg->setBuffer("csd-0", buffer);
1243
1244 buffer = new (std::nothrow) ABuffer(1024);
1245 if (buffer.get() == NULL || buffer->base() == NULL) {
1246 return NO_MEMORY;
1247 }
1248 buffer->setRange(0, 0);
1249
1250 if (size < 1) {
1251 ALOGE("b/23680780");
1252 return BAD_VALUE;
1253 }
1254 size_t numPictureParameterSets = *ptr;
1255 ++ptr;
1256 --size;
1257
1258 for (size_t i = 0; i < numPictureParameterSets; ++i) {
1259 if (size < 2) {
1260 ALOGE("b/23680780");
1261 return BAD_VALUE;
1262 }
1263 size_t length = U16_AT(ptr);
1264
1265 ptr += 2;
1266 size -= 2;
1267
1268 if (size < length) {
1269 return BAD_VALUE;
1270 }
1271 status_t err = copyNALUToABuffer(&buffer, ptr, length);
1272 if (err != OK) {
1273 return err;
1274 }
1275
1276 ptr += length;
1277 size -= length;
1278 }
1279
1280 buffer->meta()->setInt32("csd", true);
1281 buffer->meta()->setInt64("timeUs", 0);
1282 msg->setBuffer("csd-1", buffer);
1283 } else if (meta->findData(kKeyHVCC, &type, &data, &size)) {
1284 const uint8_t *ptr = (const uint8_t *)data;
1285
1286 if (size < 23 || (ptr[0] != 1 && ptr[0] != 0)) {
1287 // configurationVersion == 1 or 0
1288 // 1 is what the standard dictates, but some old muxers may have used 0.
1289 ALOGE("b/23680780");
1290 return BAD_VALUE;
1291 }
1292
1293 const size_t dataSize = size; // save for later
1294 ptr += 22;
1295 size -= 22;
1296
1297 size_t numofArrays = (char)ptr[0];
1298 ptr += 1;
1299 size -= 1;
1300 size_t j = 0, i = 0;
1301
1302 sp<ABuffer> buffer = new (std::nothrow) ABuffer(1024);
1303 if (buffer.get() == NULL || buffer->base() == NULL) {
1304 return NO_MEMORY;
1305 }
1306 buffer->setRange(0, 0);
1307
1308 HevcParameterSets hvcc;
1309
1310 for (i = 0; i < numofArrays; i++) {
1311 if (size < 3) {
1312 ALOGE("b/23680780");
1313 return BAD_VALUE;
1314 }
1315 ptr += 1;
1316 size -= 1;
1317
1318 //Num of nals
1319 size_t numofNals = U16_AT(ptr);
1320
1321 ptr += 2;
1322 size -= 2;
1323
1324 for (j = 0; j < numofNals; j++) {
1325 if (size < 2) {
1326 ALOGE("b/23680780");
1327 return BAD_VALUE;
1328 }
1329 size_t length = U16_AT(ptr);
1330
1331 ptr += 2;
1332 size -= 2;
1333
1334 if (size < length) {
1335 return BAD_VALUE;
1336 }
1337 status_t err = copyNALUToABuffer(&buffer, ptr, length);
1338 if (err != OK) {
1339 return err;
1340 }
1341 (void)hvcc.addNalUnit(ptr, length);
1342
1343 ptr += length;
1344 size -= length;
1345 }
1346 }
1347 buffer->meta()->setInt32("csd", true);
1348 buffer->meta()->setInt64("timeUs", 0);
1349 msg->setBuffer("csd-0", buffer);
1350
1351 // if we saw VUI color information we know whether this is HDR because VUI trumps other
1352 // format parameters for HEVC.
1353 HevcParameterSets::Info info = hvcc.getInfo();
1354 if (info & hvcc.kInfoHasColorDescription) {
1355 msg->setInt32("android._is-hdr", (info & hvcc.kInfoIsHdr) != 0);
1356 }
1357
1358 uint32_t isoPrimaries, isoTransfer, isoMatrix, isoRange;
1359 if (hvcc.findParam32(kColourPrimaries, &isoPrimaries)
1360 && hvcc.findParam32(kTransferCharacteristics, &isoTransfer)
1361 && hvcc.findParam32(kMatrixCoeffs, &isoMatrix)
1362 && hvcc.findParam32(kVideoFullRangeFlag, &isoRange)) {
1363 ALOGV("found iso color aspects : primaris=%d, transfer=%d, matrix=%d, range=%d",
1364 isoPrimaries, isoTransfer, isoMatrix, isoRange);
1365
1366 ColorAspects aspects;
1367 ColorUtils::convertIsoColorAspectsToCodecAspects(
1368 isoPrimaries, isoTransfer, isoMatrix, isoRange, aspects);
1369
1370 if (aspects.mPrimaries == ColorAspects::PrimariesUnspecified) {
1371 int32_t primaries;
1372 if (meta->findInt32(kKeyColorPrimaries, &primaries)) {
1373 ALOGV("unspecified primaries found, replaced to %d", primaries);
1374 aspects.mPrimaries = static_cast<ColorAspects::Primaries>(primaries);
1375 }
1376 }
1377 if (aspects.mTransfer == ColorAspects::TransferUnspecified) {
1378 int32_t transferFunction;
1379 if (meta->findInt32(kKeyTransferFunction, &transferFunction)) {
1380 ALOGV("unspecified transfer found, replaced to %d", transferFunction);
1381 aspects.mTransfer = static_cast<ColorAspects::Transfer>(transferFunction);
1382 }
1383 }
1384 if (aspects.mMatrixCoeffs == ColorAspects::MatrixUnspecified) {
1385 int32_t colorMatrix;
1386 if (meta->findInt32(kKeyColorMatrix, &colorMatrix)) {
1387 ALOGV("unspecified matrix found, replaced to %d", colorMatrix);
1388 aspects.mMatrixCoeffs = static_cast<ColorAspects::MatrixCoeffs>(colorMatrix);
1389 }
1390 }
1391 if (aspects.mRange == ColorAspects::RangeUnspecified) {
1392 int32_t range;
1393 if (meta->findInt32(kKeyColorRange, &range)) {
1394 ALOGV("unspecified range found, replaced to %d", range);
1395 aspects.mRange = static_cast<ColorAspects::Range>(range);
1396 }
1397 }
1398
1399 int32_t standard, transfer, range;
1400 if (ColorUtils::convertCodecColorAspectsToPlatformAspects(
1401 aspects, &range, &standard, &transfer) == OK) {
1402 msg->setInt32("color-standard", standard);
1403 msg->setInt32("color-transfer", transfer);
1404 msg->setInt32("color-range", range);
1405 }
1406 }
1407
1408 parseHevcProfileLevelFromHvcc((const uint8_t *)data, dataSize, msg);
1409 } else if (meta->findData(kKeyAV1C, &type, &data, &size)) {
1410 sp<ABuffer> buffer = new (std::nothrow) ABuffer(size);
1411 if (buffer.get() == NULL || buffer->base() == NULL) {
1412 return NO_MEMORY;
1413 }
1414 memcpy(buffer->data(), data, size);
1415
1416 buffer->meta()->setInt32("csd", true);
1417 buffer->meta()->setInt64("timeUs", 0);
1418 msg->setBuffer("csd-0", buffer);
1419 parseAV1ProfileLevelFromCsd(buffer, msg);
1420 } else if (meta->findData(kKeyESDS, &type, &data, &size)) {
1421 ESDS esds((const char *)data, size);
1422 if (esds.InitCheck() != (status_t)OK) {
1423 return BAD_VALUE;
1424 }
1425
1426 const void *codec_specific_data;
1427 size_t codec_specific_data_size;
1428 esds.getCodecSpecificInfo(
1429 &codec_specific_data, &codec_specific_data_size);
1430
1431 sp<ABuffer> buffer = new (std::nothrow) ABuffer(codec_specific_data_size);
1432 if (buffer.get() == NULL || buffer->base() == NULL) {
1433 return NO_MEMORY;
1434 }
1435
1436 memcpy(buffer->data(), codec_specific_data,
1437 codec_specific_data_size);
1438
1439 buffer->meta()->setInt32("csd", true);
1440 buffer->meta()->setInt64("timeUs", 0);
1441 msg->setBuffer("csd-0", buffer);
1442
1443 if (!strcasecmp(mime, MEDIA_MIMETYPE_VIDEO_MPEG4)) {
1444 parseMpeg4ProfileLevelFromCsd(buffer, msg);
1445 } else if (!strcasecmp(mime, MEDIA_MIMETYPE_VIDEO_MPEG2)) {
1446 parseMpeg2ProfileLevelFromEsds(esds, msg);
1447 if (meta->findData(kKeyStreamHeader, &type, &data, &size)) {
1448 parseMpeg2ProfileLevelFromHeader((uint8_t*)data, size, msg);
1449 }
1450 } else if (!strcasecmp(mime, MEDIA_MIMETYPE_AUDIO_AAC)) {
1451 parseAacProfileFromCsd(buffer, msg);
1452 }
1453
1454 uint32_t maxBitrate, avgBitrate;
1455 if (esds.getBitRate(&maxBitrate, &avgBitrate) == OK) {
1456 if (!meta->hasData(kKeyBitRate)
1457 && avgBitrate > 0 && avgBitrate <= INT32_MAX) {
1458 msg->setInt32("bitrate", (int32_t)avgBitrate);
1459 } else {
1460 (void)msg->findInt32("bitrate", (int32_t*)&avgBitrate);
1461 }
1462 if (!meta->hasData(kKeyMaxBitRate)
1463 && maxBitrate > 0 && maxBitrate <= INT32_MAX && maxBitrate >= avgBitrate) {
1464 msg->setInt32("max-bitrate", (int32_t)maxBitrate);
1465 }
1466 }
1467 } else if (meta->findData(kKeyD263, &type, &data, &size)) {
1468 const uint8_t *ptr = (const uint8_t *)data;
1469 parseH263ProfileLevelFromD263(ptr, size, msg);
1470 } else if (meta->findData(kKeyOpusHeader, &type, &data, &size)) {
1471 sp<ABuffer> buffer = new (std::nothrow) ABuffer(size);
1472 if (buffer.get() == NULL || buffer->base() == NULL) {
1473 return NO_MEMORY;
1474 }
1475 memcpy(buffer->data(), data, size);
1476
1477 buffer->meta()->setInt32("csd", true);
1478 buffer->meta()->setInt64("timeUs", 0);
1479 msg->setBuffer("csd-0", buffer);
1480
1481 if (!meta->findData(kKeyOpusCodecDelay, &type, &data, &size)) {
1482 return -EINVAL;
1483 }
1484
1485 buffer = new (std::nothrow) ABuffer(size);
1486 if (buffer.get() == NULL || buffer->base() == NULL) {
1487 return NO_MEMORY;
1488 }
1489 memcpy(buffer->data(), data, size);
1490
1491 buffer->meta()->setInt32("csd", true);
1492 buffer->meta()->setInt64("timeUs", 0);
1493 msg->setBuffer("csd-1", buffer);
1494
1495 if (!meta->findData(kKeyOpusSeekPreRoll, &type, &data, &size)) {
1496 return -EINVAL;
1497 }
1498
1499 buffer = new (std::nothrow) ABuffer(size);
1500 if (buffer.get() == NULL || buffer->base() == NULL) {
1501 return NO_MEMORY;
1502 }
1503 memcpy(buffer->data(), data, size);
1504
1505 buffer->meta()->setInt32("csd", true);
1506 buffer->meta()->setInt64("timeUs", 0);
1507 msg->setBuffer("csd-2", buffer);
1508 } else if (meta->findData(kKeyVp9CodecPrivate, &type, &data, &size)) {
1509 sp<ABuffer> buffer = new (std::nothrow) ABuffer(size);
1510 if (buffer.get() == NULL || buffer->base() == NULL) {
1511 return NO_MEMORY;
1512 }
1513 memcpy(buffer->data(), data, size);
1514
1515 buffer->meta()->setInt32("csd", true);
1516 buffer->meta()->setInt64("timeUs", 0);
1517 msg->setBuffer("csd-0", buffer);
1518
1519 parseVp9ProfileLevelFromCsd(buffer, msg);
1520 } else if (meta->findData(kKeyAlacMagicCookie, &type, &data, &size)) {
1521 ALOGV("convertMetaDataToMessage found kKeyAlacMagicCookie of size %zu\n", size);
1522 sp<ABuffer> buffer = new (std::nothrow) ABuffer(size);
1523 if (buffer.get() == NULL || buffer->base() == NULL) {
1524 return NO_MEMORY;
1525 }
1526 memcpy(buffer->data(), data, size);
1527
1528 buffer->meta()->setInt32("csd", true);
1529 buffer->meta()->setInt64("timeUs", 0);
1530 msg->setBuffer("csd-0", buffer);
1531 }
1532
1533 if (meta->findData(kKeyDVCC, &type, &data, &size)
1534 || meta->findData(kKeyDVVC, &type, &data, &size)
1535 || meta->findData(kKeyDVWC, &type, &data, &size)) {
1536 const uint8_t *ptr = (const uint8_t *)data;
1537 ALOGV("DV: calling parseDolbyVisionProfileLevelFromDvcc with data size %zu", size);
1538 parseDolbyVisionProfileLevelFromDvcc(ptr, size, msg);
1539 sp<ABuffer> buffer = new (std::nothrow) ABuffer(size);
1540 if (buffer.get() == nullptr || buffer->base() == nullptr) {
1541 return NO_MEMORY;
1542 }
1543 memcpy(buffer->data(), data, size);
1544
1545 buffer->meta()->setInt32("csd", true);
1546 buffer->meta()->setInt64("timeUs", 0);
1547 msg->setBuffer("csd-2", buffer);
1548 }
1549
1550 *format = msg;
1551
1552 return OK;
1553 }
1554
findNextNalStartCode(const uint8_t * data,size_t length)1555 const uint8_t *findNextNalStartCode(const uint8_t *data, size_t length) {
1556 uint8_t *res = NULL;
1557 if (length > 4) {
1558 // minus 1 as to not match NAL start code at end
1559 res = (uint8_t *)memmem(data, length - 1, "\x00\x00\x00\x01", 4);
1560 }
1561 return res != NULL && res < data + length - 4 ? res : &data[length];
1562 }
1563
reassembleAVCC(const sp<ABuffer> & csd0,const sp<ABuffer> & csd1,char * avcc)1564 static size_t reassembleAVCC(const sp<ABuffer> &csd0, const sp<ABuffer> &csd1, char *avcc) {
1565 avcc[0] = 1; // version
1566 avcc[1] = 0x64; // profile (default to high)
1567 avcc[2] = 0; // constraints (default to none)
1568 avcc[3] = 0xd; // level (default to 1.3)
1569 avcc[4] = 0xff; // reserved+size
1570
1571 size_t i = 0;
1572 int numparams = 0;
1573 int lastparamoffset = 0;
1574 int avccidx = 6;
1575 do {
1576 i = findNextNalStartCode(csd0->data() + i, csd0->size() - i) - csd0->data();
1577 ALOGV("block at %zu, last was %d", i, lastparamoffset);
1578 if (lastparamoffset > 0) {
1579 const uint8_t *lastparam = csd0->data() + lastparamoffset;
1580 int size = i - lastparamoffset;
1581 if (size > 3) {
1582 if (numparams && memcmp(avcc + 1, lastparam + 1, 3)) {
1583 ALOGW("Inconsisted profile/level found in SPS: %x,%x,%x vs %x,%x,%x",
1584 avcc[1], avcc[2], avcc[3], lastparam[1], lastparam[2], lastparam[3]);
1585 } else if (!numparams) {
1586 // fill in profile, constraints and level
1587 memcpy(avcc + 1, lastparam + 1, 3);
1588 }
1589 }
1590 avcc[avccidx++] = size >> 8;
1591 avcc[avccidx++] = size & 0xff;
1592 memcpy(avcc+avccidx, lastparam, size);
1593 avccidx += size;
1594 numparams++;
1595 }
1596 i += 4;
1597 lastparamoffset = i;
1598 } while(i < csd0->size());
1599 ALOGV("csd0 contains %d params", numparams);
1600
1601 avcc[5] = 0xe0 | numparams;
1602 //and now csd-1
1603 i = 0;
1604 numparams = 0;
1605 lastparamoffset = 0;
1606 int numpicparamsoffset = avccidx;
1607 avccidx++;
1608 do {
1609 i = findNextNalStartCode(csd1->data() + i, csd1->size() - i) - csd1->data();
1610 ALOGV("block at %zu, last was %d", i, lastparamoffset);
1611 if (lastparamoffset > 0) {
1612 int size = i - lastparamoffset;
1613 avcc[avccidx++] = size >> 8;
1614 avcc[avccidx++] = size & 0xff;
1615 memcpy(avcc+avccidx, csd1->data() + lastparamoffset, size);
1616 avccidx += size;
1617 numparams++;
1618 }
1619 i += 4;
1620 lastparamoffset = i;
1621 } while(i < csd1->size());
1622 avcc[numpicparamsoffset] = numparams;
1623 return avccidx;
1624 }
1625
reassembleESDS(const sp<ABuffer> & csd0,char * esds)1626 static void reassembleESDS(const sp<ABuffer> &csd0, char *esds) {
1627 int csd0size = csd0->size();
1628 esds[0] = 3; // kTag_ESDescriptor;
1629 int esdescriptorsize = 26 + csd0size;
1630 CHECK(esdescriptorsize < 268435456); // 7 bits per byte, so max is 2^28-1
1631 esds[1] = 0x80 | (esdescriptorsize >> 21);
1632 esds[2] = 0x80 | ((esdescriptorsize >> 14) & 0x7f);
1633 esds[3] = 0x80 | ((esdescriptorsize >> 7) & 0x7f);
1634 esds[4] = (esdescriptorsize & 0x7f);
1635 esds[5] = esds[6] = 0; // es id
1636 esds[7] = 0; // flags
1637 esds[8] = 4; // kTag_DecoderConfigDescriptor
1638 int configdescriptorsize = 18 + csd0size;
1639 esds[9] = 0x80 | (configdescriptorsize >> 21);
1640 esds[10] = 0x80 | ((configdescriptorsize >> 14) & 0x7f);
1641 esds[11] = 0x80 | ((configdescriptorsize >> 7) & 0x7f);
1642 esds[12] = (configdescriptorsize & 0x7f);
1643 esds[13] = 0x40; // objectTypeIndication
1644 // bytes 14-25 are examples from a real file. they are unused/overwritten by muxers.
1645 esds[14] = 0x15; // streamType(5), upStream(0),
1646 esds[15] = 0x00; // 15-17: bufferSizeDB (6KB)
1647 esds[16] = 0x18;
1648 esds[17] = 0x00;
1649 esds[18] = 0x00; // 18-21: maxBitrate (64kbps)
1650 esds[19] = 0x00;
1651 esds[20] = 0xfa;
1652 esds[21] = 0x00;
1653 esds[22] = 0x00; // 22-25: avgBitrate (64kbps)
1654 esds[23] = 0x00;
1655 esds[24] = 0xfa;
1656 esds[25] = 0x00;
1657 esds[26] = 5; // kTag_DecoderSpecificInfo;
1658 esds[27] = 0x80 | (csd0size >> 21);
1659 esds[28] = 0x80 | ((csd0size >> 14) & 0x7f);
1660 esds[29] = 0x80 | ((csd0size >> 7) & 0x7f);
1661 esds[30] = (csd0size & 0x7f);
1662 memcpy((void*)&esds[31], csd0->data(), csd0size);
1663 // data following this is ignored, so don't bother appending it
1664 }
1665
reassembleHVCC(const sp<ABuffer> & csd0,uint8_t * hvcc,size_t hvccSize,size_t nalSizeLength)1666 static size_t reassembleHVCC(const sp<ABuffer> &csd0, uint8_t *hvcc, size_t hvccSize, size_t nalSizeLength) {
1667 HevcParameterSets paramSets;
1668 uint8_t* data = csd0->data();
1669 if (csd0->size() < 4) {
1670 ALOGE("csd0 too small");
1671 return 0;
1672 }
1673 if (memcmp(data, "\x00\x00\x00\x01", 4) != 0) {
1674 ALOGE("csd0 doesn't start with a start code");
1675 return 0;
1676 }
1677 size_t prevNalOffset = 4;
1678 status_t err = OK;
1679 for (size_t i = 1; i < csd0->size() - 4; ++i) {
1680 if (memcmp(&data[i], "\x00\x00\x00\x01", 4) != 0) {
1681 continue;
1682 }
1683 err = paramSets.addNalUnit(&data[prevNalOffset], i - prevNalOffset);
1684 if (err != OK) {
1685 return 0;
1686 }
1687 prevNalOffset = i + 4;
1688 }
1689 err = paramSets.addNalUnit(&data[prevNalOffset], csd0->size() - prevNalOffset);
1690 if (err != OK) {
1691 return 0;
1692 }
1693 size_t size = hvccSize;
1694 err = paramSets.makeHvcc(hvcc, &size, nalSizeLength);
1695 if (err != OK) {
1696 return 0;
1697 }
1698 return size;
1699 }
1700
1701 #if 0
1702 static void convertMessageToMetaDataInt32(
1703 const sp<AMessage> &msg, sp<MetaData> &meta, uint32_t key, const char *name) {
1704 int32_t value;
1705 if (msg->findInt32(name, &value)) {
1706 meta->setInt32(key, value);
1707 }
1708 }
1709 #endif
1710
convertMessageToMetaDataColorAspects(const sp<AMessage> & msg,sp<MetaData> & meta)1711 static void convertMessageToMetaDataColorAspects(const sp<AMessage> &msg, sp<MetaData> &meta) {
1712 // 0 values are unspecified
1713 int32_t range = 0, standard = 0, transfer = 0;
1714 (void)msg->findInt32("color-range", &range);
1715 (void)msg->findInt32("color-standard", &standard);
1716 (void)msg->findInt32("color-transfer", &transfer);
1717
1718 ColorAspects colorAspects;
1719 memset(&colorAspects, 0, sizeof(colorAspects));
1720 if (CodecBase::convertPlatformColorAspectsToCodecAspects(
1721 range, standard, transfer, colorAspects) != OK) {
1722 return;
1723 }
1724
1725 // save specified values to meta
1726 if (colorAspects.mRange != 0) {
1727 meta->setInt32(kKeyColorRange, colorAspects.mRange);
1728 }
1729 if (colorAspects.mPrimaries != 0) {
1730 meta->setInt32(kKeyColorPrimaries, colorAspects.mPrimaries);
1731 }
1732 if (colorAspects.mTransfer != 0) {
1733 meta->setInt32(kKeyTransferFunction, colorAspects.mTransfer);
1734 }
1735 if (colorAspects.mMatrixCoeffs != 0) {
1736 meta->setInt32(kKeyColorMatrix, colorAspects.mMatrixCoeffs);
1737 }
1738 }
1739 /* Converts key and value pairs in AMessage format to MetaData format.
1740 * Also checks for the presence of required keys.
1741 */
convertMessageToMetaData(const sp<AMessage> & msg,sp<MetaData> & meta)1742 status_t convertMessageToMetaData(const sp<AMessage> &msg, sp<MetaData> &meta) {
1743 AString mime;
1744 if (msg->findString("mime", &mime)) {
1745 meta->setCString(kKeyMIMEType, mime.c_str());
1746 } else {
1747 ALOGV("did not find mime type");
1748 return BAD_VALUE;
1749 }
1750
1751 convertMessageToMetaDataFromMappings(msg, meta);
1752
1753 int32_t systemId;
1754 if (msg->findInt32("ca-system-id", &systemId)) {
1755 meta->setInt32(kKeyCASystemID, systemId);
1756
1757 sp<ABuffer> caSessionId, caPvtData;
1758 if (msg->findBuffer("ca-session-id", &caSessionId)) {
1759 meta->setData(kKeyCASessionID, 0, caSessionId->data(), caSessionId->size());
1760 }
1761 if (msg->findBuffer("ca-private-data", &caPvtData)) {
1762 meta->setData(kKeyCAPrivateData, 0, caPvtData->data(), caPvtData->size());
1763 }
1764 }
1765
1766 int64_t durationUs;
1767 if (msg->findInt64("durationUs", &durationUs)) {
1768 meta->setInt64(kKeyDuration, durationUs);
1769 }
1770
1771 int32_t isSync;
1772 if (msg->findInt32("is-sync-frame", &isSync) && isSync != 0) {
1773 meta->setInt32(kKeyIsSyncFrame, 1);
1774 }
1775
1776 // Mode for media transcoding.
1777 int32_t isBackgroundMode;
1778 if (msg->findInt32("android._background-mode", &isBackgroundMode) && isBackgroundMode != 0) {
1779 meta->setInt32(isBackgroundMode, 1);
1780 }
1781
1782 int32_t avgBitrate = 0;
1783 int32_t maxBitrate;
1784 if (msg->findInt32("bitrate", &avgBitrate) && avgBitrate > 0) {
1785 meta->setInt32(kKeyBitRate, avgBitrate);
1786 }
1787 if (msg->findInt32("max-bitrate", &maxBitrate) && maxBitrate > 0 && maxBitrate >= avgBitrate) {
1788 meta->setInt32(kKeyMaxBitRate, maxBitrate);
1789 }
1790
1791 AString lang;
1792 if (msg->findString("language", &lang)) {
1793 meta->setCString(kKeyMediaLanguage, lang.c_str());
1794 }
1795
1796 if (mime.startsWith("video/") || mime.startsWith("image/")) {
1797 int32_t width;
1798 int32_t height;
1799 if (!msg->findInt32("width", &width) || !msg->findInt32("height", &height)) {
1800 ALOGV("did not find width and/or height");
1801 return BAD_VALUE;
1802 }
1803 if (width <= 0 || height <= 0) {
1804 ALOGE("Invalid value of width: %d and/or height: %d", width, height);
1805 return BAD_VALUE;
1806 }
1807 meta->setInt32(kKeyWidth, width);
1808 meta->setInt32(kKeyHeight, height);
1809
1810 int32_t sarWidth = -1, sarHeight = -1;
1811 bool foundWidth, foundHeight;
1812 foundWidth = msg->findInt32("sar-width", &sarWidth);
1813 foundHeight = msg->findInt32("sar-height", &sarHeight);
1814 if (foundWidth || foundHeight) {
1815 if (sarWidth <= 0 || sarHeight <= 0) {
1816 ALOGE("Invalid value of sarWidth: %d and/or sarHeight: %d", sarWidth, sarHeight);
1817 return BAD_VALUE;
1818 }
1819 meta->setInt32(kKeySARWidth, sarWidth);
1820 meta->setInt32(kKeySARHeight, sarHeight);
1821 }
1822
1823 int32_t displayWidth = -1, displayHeight = -1;
1824 foundWidth = msg->findInt32("display-width", &displayWidth);
1825 foundHeight = msg->findInt32("display-height", &displayHeight);
1826 if (foundWidth || foundHeight) {
1827 if (displayWidth <= 0 || displayHeight <= 0) {
1828 ALOGE("Invalid value of displayWidth: %d and/or displayHeight: %d",
1829 displayWidth, displayHeight);
1830 return BAD_VALUE;
1831 }
1832 meta->setInt32(kKeyDisplayWidth, displayWidth);
1833 meta->setInt32(kKeyDisplayHeight, displayHeight);
1834 }
1835
1836 if (mime.startsWith("image/")){
1837 int32_t isPrimary;
1838 if (msg->findInt32("is-default", &isPrimary) && isPrimary) {
1839 meta->setInt32(kKeyTrackIsDefault, 1);
1840 }
1841 int32_t tileWidth = -1, tileHeight = -1;
1842 foundWidth = msg->findInt32("tile-width", &tileWidth);
1843 foundHeight = msg->findInt32("tile-height", &tileHeight);
1844 if (foundWidth || foundHeight) {
1845 if (tileWidth <= 0 || tileHeight <= 0) {
1846 ALOGE("Invalid value of tileWidth: %d and/or tileHeight: %d",
1847 tileWidth, tileHeight);
1848 return BAD_VALUE;
1849 }
1850 meta->setInt32(kKeyTileWidth, tileWidth);
1851 meta->setInt32(kKeyTileHeight, tileHeight);
1852 }
1853 int32_t gridRows = -1, gridCols = -1;
1854 bool foundRows, foundCols;
1855 foundRows = msg->findInt32("grid-rows", &gridRows);
1856 foundCols = msg->findInt32("grid-cols", &gridCols);
1857 if (foundRows || foundCols) {
1858 if (gridRows <= 0 || gridCols <= 0) {
1859 ALOGE("Invalid value of gridRows: %d and/or gridCols: %d",
1860 gridRows, gridCols);
1861 return BAD_VALUE;
1862 }
1863 meta->setInt32(kKeyGridRows, gridRows);
1864 meta->setInt32(kKeyGridCols, gridCols);
1865 }
1866 }
1867
1868 int32_t colorFormat;
1869 if (msg->findInt32("color-format", &colorFormat)) {
1870 meta->setInt32(kKeyColorFormat, colorFormat);
1871 }
1872
1873 int32_t cropLeft, cropTop, cropRight, cropBottom;
1874 if (msg->findRect("crop",
1875 &cropLeft,
1876 &cropTop,
1877 &cropRight,
1878 &cropBottom)) {
1879 if (cropLeft < 0 || cropLeft > cropRight || cropRight >= width) {
1880 ALOGE("Invalid value of cropLeft: %d and/or cropRight: %d", cropLeft, cropRight);
1881 return BAD_VALUE;
1882 }
1883 if (cropTop < 0 || cropTop > cropBottom || cropBottom >= height) {
1884 ALOGE("Invalid value of cropTop: %d and/or cropBottom: %d", cropTop, cropBottom);
1885 return BAD_VALUE;
1886 }
1887 meta->setRect(kKeyCropRect, cropLeft, cropTop, cropRight, cropBottom);
1888 }
1889
1890 int32_t rotationDegrees;
1891 if (msg->findInt32("rotation-degrees", &rotationDegrees)) {
1892 meta->setInt32(kKeyRotation, rotationDegrees);
1893 }
1894
1895 if (msg->contains("hdr-static-info")) {
1896 HDRStaticInfo info;
1897 if (ColorUtils::getHDRStaticInfoFromFormat(msg, &info)) {
1898 meta->setData(kKeyHdrStaticInfo, 'hdrS', &info, sizeof(info));
1899 }
1900 }
1901
1902 sp<ABuffer> hdr10PlusInfo;
1903 if (msg->findBuffer("hdr10-plus-info", &hdr10PlusInfo)) {
1904 meta->setData(kKeyHdr10PlusInfo, 0,
1905 hdr10PlusInfo->data(), hdr10PlusInfo->size());
1906 }
1907
1908 convertMessageToMetaDataColorAspects(msg, meta);
1909
1910 AString tsSchema;
1911 if (msg->findString("ts-schema", &tsSchema)) {
1912 unsigned int numLayers = 0;
1913 unsigned int numBLayers = 0;
1914 char dummy;
1915 int tags = sscanf(tsSchema.c_str(), "android.generic.%u%c%u%c",
1916 &numLayers, &dummy, &numBLayers, &dummy);
1917 if ((tags == 1 || (tags == 3 && dummy == '+'))
1918 && numLayers > 0 && numLayers < UINT32_MAX - numBLayers
1919 && numLayers + numBLayers <= INT32_MAX) {
1920 meta->setInt32(kKeyTemporalLayerCount, numLayers + numBLayers);
1921 }
1922 }
1923 } else if (mime.startsWith("audio/")) {
1924 int32_t numChannels, sampleRate;
1925 if (!msg->findInt32("channel-count", &numChannels) ||
1926 !msg->findInt32("sample-rate", &sampleRate)) {
1927 ALOGV("did not find channel-count and/or sample-rate");
1928 return BAD_VALUE;
1929 }
1930 // channel count can be zero in some cases like mpeg h
1931 if (sampleRate <= 0 || numChannels < 0) {
1932 ALOGE("Invalid value of channel-count: %d and/or sample-rate: %d",
1933 numChannels, sampleRate);
1934 return BAD_VALUE;
1935 }
1936 meta->setInt32(kKeyChannelCount, numChannels);
1937 meta->setInt32(kKeySampleRate, sampleRate);
1938 int32_t bitsPerSample;
1939 // TODO:(b/204430952) add appropriate bound check for bitsPerSample
1940 if (msg->findInt32("bits-per-sample", &bitsPerSample)) {
1941 meta->setInt32(kKeyBitsPerSample, bitsPerSample);
1942 }
1943 int32_t channelMask;
1944 if (msg->findInt32("channel-mask", &channelMask)) {
1945 meta->setInt32(kKeyChannelMask, channelMask);
1946 }
1947 int32_t delay = 0;
1948 if (msg->findInt32("encoder-delay", &delay)) {
1949 meta->setInt32(kKeyEncoderDelay, delay);
1950 }
1951 int32_t padding = 0;
1952 if (msg->findInt32("encoder-padding", &padding)) {
1953 meta->setInt32(kKeyEncoderPadding, padding);
1954 }
1955
1956 int32_t isADTS;
1957 if (msg->findInt32("is-adts", &isADTS)) {
1958 meta->setInt32(kKeyIsADTS, isADTS);
1959 }
1960
1961 int32_t mpeghProfileLevelIndication = -1;
1962 if (msg->findInt32(AMEDIAFORMAT_KEY_MPEGH_PROFILE_LEVEL_INDICATION,
1963 &mpeghProfileLevelIndication)) {
1964 meta->setInt32(kKeyMpeghProfileLevelIndication, mpeghProfileLevelIndication);
1965 }
1966 int32_t mpeghReferenceChannelLayout = -1;
1967 if (msg->findInt32(AMEDIAFORMAT_KEY_MPEGH_REFERENCE_CHANNEL_LAYOUT,
1968 &mpeghReferenceChannelLayout)) {
1969 meta->setInt32(kKeyMpeghReferenceChannelLayout, mpeghReferenceChannelLayout);
1970 }
1971 sp<ABuffer> mpeghCompatibleSets;
1972 if (msg->findBuffer(AMEDIAFORMAT_KEY_MPEGH_COMPATIBLE_SETS,
1973 &mpeghCompatibleSets)) {
1974 meta->setData(kKeyMpeghCompatibleSets, kTypeHCOS,
1975 mpeghCompatibleSets->data(), mpeghCompatibleSets->size());
1976 }
1977
1978 int32_t aacProfile = -1;
1979 if (msg->findInt32("aac-profile", &aacProfile)) {
1980 meta->setInt32(kKeyAACAOT, aacProfile);
1981 }
1982
1983 int32_t pcmEncoding;
1984 if (msg->findInt32("pcm-encoding", &pcmEncoding)) {
1985 meta->setInt32(kKeyPcmEncoding, pcmEncoding);
1986 }
1987
1988 int32_t hapticChannelCount;
1989 if (msg->findInt32("haptic-channel-count", &hapticChannelCount)) {
1990 meta->setInt32(kKeyHapticChannelCount, hapticChannelCount);
1991 }
1992 }
1993
1994 int32_t maxInputSize;
1995 if (msg->findInt32("max-input-size", &maxInputSize)) {
1996 meta->setInt32(kKeyMaxInputSize, maxInputSize);
1997 }
1998
1999 int32_t maxWidth;
2000 if (msg->findInt32("max-width", &maxWidth)) {
2001 meta->setInt32(kKeyMaxWidth, maxWidth);
2002 }
2003
2004 int32_t maxHeight;
2005 if (msg->findInt32("max-height", &maxHeight)) {
2006 meta->setInt32(kKeyMaxHeight, maxHeight);
2007 }
2008
2009 int32_t fps;
2010 float fpsFloat;
2011 if (msg->findInt32("frame-rate", &fps) && fps > 0) {
2012 meta->setInt32(kKeyFrameRate, fps);
2013 } else if (msg->findFloat("frame-rate", &fpsFloat)
2014 && fpsFloat >= 1 && fpsFloat <= (float)INT32_MAX) {
2015 // truncate values to distinguish between e.g. 24 vs 23.976 fps
2016 meta->setInt32(kKeyFrameRate, (int32_t)fpsFloat);
2017 }
2018
2019 // reassemble the csd data into its original form
2020 sp<ABuffer> csd0, csd1, csd2;
2021 if (msg->findBuffer("csd-0", &csd0)) {
2022 int csd0size = csd0->size();
2023 if (mime == MEDIA_MIMETYPE_VIDEO_AVC) {
2024 sp<ABuffer> csd1;
2025 if (msg->findBuffer("csd-1", &csd1)) {
2026 std::vector<char> avcc(csd0size + csd1->size() + 1024);
2027 size_t outsize = reassembleAVCC(csd0, csd1, avcc.data());
2028 meta->setData(kKeyAVCC, kTypeAVCC, avcc.data(), outsize);
2029 }
2030 } else if (mime == MEDIA_MIMETYPE_AUDIO_AAC ||
2031 mime == MEDIA_MIMETYPE_VIDEO_MPEG4 ||
2032 mime == MEDIA_MIMETYPE_AUDIO_WMA ||
2033 mime == MEDIA_MIMETYPE_AUDIO_MS_ADPCM ||
2034 mime == MEDIA_MIMETYPE_AUDIO_DVI_IMA_ADPCM) {
2035 std::vector<char> esds(csd0size + 31);
2036 // The written ESDS is actually for an audio stream, but it's enough
2037 // for transporting the CSD to muxers.
2038 reassembleESDS(csd0, esds.data());
2039 meta->setData(kKeyESDS, kTypeESDS, esds.data(), esds.size());
2040 } else if (mime == MEDIA_MIMETYPE_VIDEO_HEVC ||
2041 mime == MEDIA_MIMETYPE_IMAGE_ANDROID_HEIC) {
2042 std::vector<uint8_t> hvcc(csd0size + 1024);
2043 size_t outsize = reassembleHVCC(csd0, hvcc.data(), hvcc.size(), 4);
2044 meta->setData(kKeyHVCC, kTypeHVCC, hvcc.data(), outsize);
2045 } else if (mime == MEDIA_MIMETYPE_VIDEO_AV1 ||
2046 mime == MEDIA_MIMETYPE_IMAGE_AVIF) {
2047 meta->setData(kKeyAV1C, 0, csd0->data(), csd0->size());
2048 } else if (mime == MEDIA_MIMETYPE_VIDEO_DOLBY_VISION) {
2049 int32_t profile = -1;
2050 uint8_t blCompatibilityId = -1;
2051 int32_t level = 0;
2052 uint8_t profileVal = -1;
2053 uint8_t profileVal1 = -1;
2054 uint8_t profileVal2 = -1;
2055 constexpr size_t dvccSize = 24;
2056
2057 const ALookup<uint8_t, int32_t> &profiles =
2058 getDolbyVisionProfileTable();
2059 const ALookup<uint8_t, int32_t> &levels =
2060 getDolbyVisionLevelsTable();
2061
2062 if (!msg->findBuffer("csd-2", &csd2)) {
2063 // MP4 extractors are expected to generate csd buffer
2064 // some encoders might not be generating it, in which
2065 // case we populate the track metadata dv (cc|vc|wc)
2066 // from the 'profile' and 'level' info.
2067 // This is done according to Dolby Vision ISOBMFF spec
2068
2069 if (!msg->findInt32("profile", &profile)) {
2070 ALOGE("Dolby Vision profile not found");
2071 return BAD_VALUE;
2072 }
2073 msg->findInt32("level", &level);
2074
2075 if (profile == DolbyVisionProfileDvheSt) {
2076 if (!profiles.rlookup(DolbyVisionProfileDvheSt, &profileVal)) { // dvhe.08
2077 ALOGE("Dolby Vision profile lookup error");
2078 return BAD_VALUE;
2079 }
2080 blCompatibilityId = 4;
2081 } else if (profile == DolbyVisionProfileDvavSe) {
2082 if (!profiles.rlookup(DolbyVisionProfileDvavSe, &profileVal)) { // dvav.09
2083 ALOGE("Dolby Vision profile lookup error");
2084 return BAD_VALUE;
2085 }
2086 blCompatibilityId = 2;
2087 } else {
2088 ALOGE("Dolby Vision profile look up error");
2089 return BAD_VALUE;
2090 }
2091
2092 profile = (int32_t) profileVal;
2093
2094 uint8_t level_val = 0;
2095 if (!levels.map(level, &level_val)) {
2096 ALOGE("Dolby Vision level lookup error");
2097 return BAD_VALUE;
2098 }
2099
2100 std::vector<uint8_t> dvcc(dvccSize);
2101
2102 dvcc[0] = 1; // major version
2103 dvcc[1] = 0; // minor version
2104 dvcc[2] = (uint8_t)((profile & 0x7f) << 1); // dolby vision profile
2105 dvcc[2] = (uint8_t)((dvcc[2] | (uint8_t)((level_val >> 5) & 0x1)) & 0xff);
2106 dvcc[3] = (uint8_t)((level_val & 0x1f) << 3); // dolby vision level
2107 dvcc[3] = (uint8_t)(dvcc[3] | (1 << 2)); // rpu_present_flag
2108 dvcc[3] = (uint8_t)(dvcc[3] | (1)); // bl_present_flag
2109 dvcc[4] = (uint8_t)(blCompatibilityId << 4); // bl_compatibility id
2110
2111 profiles.rlookup(DolbyVisionProfileDvav110, &profileVal);
2112 profiles.rlookup(DolbyVisionProfileDvheDtb, &profileVal1);
2113 if (profile > (int32_t) profileVal) {
2114 meta->setData(kKeyDVWC, kTypeDVWC, dvcc.data(), dvccSize);
2115 } else if (profile > (int32_t) profileVal1) {
2116 meta->setData(kKeyDVVC, kTypeDVVC, dvcc.data(), dvccSize);
2117 } else {
2118 meta->setData(kKeyDVCC, kTypeDVCC, dvcc.data(), dvccSize);
2119 }
2120
2121 } else {
2122 // we have csd-2, just use that to populate dvcc
2123 if (csd2->size() == dvccSize) {
2124 uint8_t *dvcc = csd2->data();
2125 profile = dvcc[2] >> 1;
2126
2127 profiles.rlookup(DolbyVisionProfileDvav110, &profileVal);
2128 profiles.rlookup(DolbyVisionProfileDvheDtb, &profileVal1);
2129 if (profile > (int32_t) profileVal) {
2130 meta->setData(kKeyDVWC, kTypeDVWC, csd2->data(), csd2->size());
2131 } else if (profile > (int32_t) profileVal1) {
2132 meta->setData(kKeyDVVC, kTypeDVVC, csd2->data(), csd2->size());
2133 } else {
2134 meta->setData(kKeyDVCC, kTypeDVCC, csd2->data(), csd2->size());
2135 }
2136
2137 } else {
2138 ALOGE("Convert MessageToMetadata csd-2 is present but not valid");
2139 return BAD_VALUE;
2140 }
2141 }
2142 profiles.rlookup(DolbyVisionProfileDvavPen, &profileVal);
2143 profiles.rlookup(DolbyVisionProfileDvavSe, &profileVal1);
2144 profiles.rlookup(DolbyVisionProfileDvav110, &profileVal2);
2145 if ((profile > (int32_t) profileVal) && (profile < (int32_t) profileVal1)) {
2146 std::vector<uint8_t> hvcc(csd0size + 1024);
2147 size_t outsize = reassembleHVCC(csd0, hvcc.data(), hvcc.size(), 4);
2148 meta->setData(kKeyHVCC, kTypeHVCC, hvcc.data(), outsize);
2149 } else if (profile == (int32_t) profileVal2) {
2150 meta->setData(kKeyAV1C, 0, csd0->data(), csd0->size());
2151 } else {
2152 sp<ABuffer> csd1;
2153 if (msg->findBuffer("csd-1", &csd1)) {
2154 std::vector<char> avcc(csd0size + csd1->size() + 1024);
2155 size_t outsize = reassembleAVCC(csd0, csd1, avcc.data());
2156 meta->setData(kKeyAVCC, kTypeAVCC, avcc.data(), outsize);
2157 }
2158 else {
2159 // for dolby vision avc, csd0 also holds csd1
2160 size_t i = 0;
2161 int csd0realsize = 0;
2162 do {
2163 i = findNextNalStartCode(csd0->data() + i,
2164 csd0->size() - i) - csd0->data();
2165 if (i > 0) {
2166 csd0realsize = i;
2167 break;
2168 }
2169 i += 4;
2170 } while(i < csd0->size());
2171 // buffer0 -> csd0
2172 sp<ABuffer> buffer0 = new (std::nothrow) ABuffer(csd0realsize);
2173 if (buffer0.get() == NULL || buffer0->base() == NULL) {
2174 return NO_MEMORY;
2175 }
2176 memcpy(buffer0->data(), csd0->data(), csd0realsize);
2177 // buffer1 -> csd1
2178 sp<ABuffer> buffer1 = new (std::nothrow)
2179 ABuffer(csd0->size() - csd0realsize);
2180 if (buffer1.get() == NULL || buffer1->base() == NULL) {
2181 return NO_MEMORY;
2182 }
2183 memcpy(buffer1->data(), csd0->data()+csd0realsize,
2184 csd0->size() - csd0realsize);
2185
2186 std::vector<char> avcc(csd0->size() + 1024);
2187 size_t outsize = reassembleAVCC(buffer0, buffer1, avcc.data());
2188 meta->setData(kKeyAVCC, kTypeAVCC, avcc.data(), outsize);
2189 }
2190 }
2191 } else if (mime == MEDIA_MIMETYPE_VIDEO_VP9) {
2192 meta->setData(kKeyVp9CodecPrivate, 0, csd0->data(), csd0->size());
2193 } else if (mime == MEDIA_MIMETYPE_AUDIO_OPUS) {
2194 size_t opusHeadSize = csd0->size();
2195 size_t codecDelayBufSize = 0;
2196 size_t seekPreRollBufSize = 0;
2197 void *opusHeadBuf = csd0->data();
2198 void *codecDelayBuf = NULL;
2199 void *seekPreRollBuf = NULL;
2200 if (msg->findBuffer("csd-1", &csd1)) {
2201 codecDelayBufSize = csd1->size();
2202 codecDelayBuf = csd1->data();
2203 }
2204 if (msg->findBuffer("csd-2", &csd2)) {
2205 seekPreRollBufSize = csd2->size();
2206 seekPreRollBuf = csd2->data();
2207 }
2208 /* Extract codec delay and seek pre roll from csd-0,
2209 * if csd-1 and csd-2 are not present */
2210 if (!codecDelayBuf && !seekPreRollBuf) {
2211 GetOpusHeaderBuffers(csd0->data(), csd0->size(), &opusHeadBuf,
2212 &opusHeadSize, &codecDelayBuf,
2213 &codecDelayBufSize, &seekPreRollBuf,
2214 &seekPreRollBufSize);
2215 }
2216 meta->setData(kKeyOpusHeader, 0, opusHeadBuf, opusHeadSize);
2217 if (codecDelayBuf) {
2218 meta->setData(kKeyOpusCodecDelay, 0, codecDelayBuf, codecDelayBufSize);
2219 }
2220 if (seekPreRollBuf) {
2221 meta->setData(kKeyOpusSeekPreRoll, 0, seekPreRollBuf, seekPreRollBufSize);
2222 }
2223 } else if (mime == MEDIA_MIMETYPE_AUDIO_ALAC) {
2224 meta->setData(kKeyAlacMagicCookie, 0, csd0->data(), csd0->size());
2225 }
2226 } else if (mime == MEDIA_MIMETYPE_VIDEO_AVC && msg->findBuffer("csd-avc", &csd0)) {
2227 meta->setData(kKeyAVCC, kTypeAVCC, csd0->data(), csd0->size());
2228 } else if ((mime == MEDIA_MIMETYPE_VIDEO_HEVC || mime == MEDIA_MIMETYPE_IMAGE_ANDROID_HEIC)
2229 && msg->findBuffer("csd-hevc", &csd0)) {
2230 meta->setData(kKeyHVCC, kTypeHVCC, csd0->data(), csd0->size());
2231 } else if (msg->findBuffer("esds", &csd0)) {
2232 meta->setData(kKeyESDS, kTypeESDS, csd0->data(), csd0->size());
2233 } else if (msg->findBuffer("mpeg2-stream-header", &csd0)) {
2234 meta->setData(kKeyStreamHeader, 'mdat', csd0->data(), csd0->size());
2235 } else if (msg->findBuffer("d263", &csd0)) {
2236 meta->setData(kKeyD263, kTypeD263, csd0->data(), csd0->size());
2237 } else if (mime == MEDIA_MIMETYPE_VIDEO_DOLBY_VISION && msg->findBuffer("csd-2", &csd2)) {
2238 meta->setData(kKeyDVCC, kTypeDVCC, csd2->data(), csd2->size());
2239
2240 // Remove CSD-2 from the data here to avoid duplicate data in meta
2241 meta->remove(kKeyOpaqueCSD2);
2242
2243 if (msg->findBuffer("csd-avc", &csd0)) {
2244 meta->setData(kKeyAVCC, kTypeAVCC, csd0->data(), csd0->size());
2245 } else if (msg->findBuffer("csd-hevc", &csd0)) {
2246 meta->setData(kKeyHVCC, kTypeHVCC, csd0->data(), csd0->size());
2247 }
2248 }
2249 // XXX TODO add whatever other keys there are
2250
2251 #if 0
2252 ALOGI("converted %s to:", msg->debugString(0).c_str());
2253 meta->dumpToLog();
2254 #endif
2255 return OK;
2256 }
2257
sendMetaDataToHal(sp<MediaPlayerBase::AudioSink> & sink,const sp<MetaData> & meta)2258 status_t sendMetaDataToHal(sp<MediaPlayerBase::AudioSink>& sink,
2259 const sp<MetaData>& meta)
2260 {
2261 int32_t sampleRate = 0;
2262 int32_t bitRate = 0;
2263 int32_t channelMask = 0;
2264 int32_t delaySamples = 0;
2265 int32_t paddingSamples = 0;
2266
2267 AudioParameter param = AudioParameter();
2268
2269 if (meta->findInt32(kKeySampleRate, &sampleRate)) {
2270 param.addInt(String8(AUDIO_OFFLOAD_CODEC_SAMPLE_RATE), sampleRate);
2271 }
2272 if (meta->findInt32(kKeyChannelMask, &channelMask)) {
2273 param.addInt(String8(AUDIO_OFFLOAD_CODEC_NUM_CHANNEL), channelMask);
2274 }
2275 if (meta->findInt32(kKeyBitRate, &bitRate)) {
2276 param.addInt(String8(AUDIO_OFFLOAD_CODEC_AVG_BIT_RATE), bitRate);
2277 }
2278 if (meta->findInt32(kKeyEncoderDelay, &delaySamples)) {
2279 param.addInt(String8(AUDIO_OFFLOAD_CODEC_DELAY_SAMPLES), delaySamples);
2280 }
2281 if (meta->findInt32(kKeyEncoderPadding, &paddingSamples)) {
2282 param.addInt(String8(AUDIO_OFFLOAD_CODEC_PADDING_SAMPLES), paddingSamples);
2283 }
2284
2285 ALOGV("sendMetaDataToHal: bitRate %d, sampleRate %d, chanMask %d,"
2286 "delaySample %d, paddingSample %d", bitRate, sampleRate,
2287 channelMask, delaySamples, paddingSamples);
2288
2289 sink->setParameters(param.toString());
2290 return OK;
2291 }
2292
2293 struct mime_conv_t {
2294 const char* mime;
2295 audio_format_t format;
2296 };
2297
2298 static const struct mime_conv_t mimeLookup[] = {
2299 { MEDIA_MIMETYPE_AUDIO_MPEG, AUDIO_FORMAT_MP3 },
2300 { MEDIA_MIMETYPE_AUDIO_RAW, AUDIO_FORMAT_PCM_16_BIT },
2301 { MEDIA_MIMETYPE_AUDIO_AMR_NB, AUDIO_FORMAT_AMR_NB },
2302 { MEDIA_MIMETYPE_AUDIO_AMR_WB, AUDIO_FORMAT_AMR_WB },
2303 { MEDIA_MIMETYPE_AUDIO_AAC, AUDIO_FORMAT_AAC },
2304 { MEDIA_MIMETYPE_AUDIO_VORBIS, AUDIO_FORMAT_VORBIS },
2305 { MEDIA_MIMETYPE_AUDIO_OPUS, AUDIO_FORMAT_OPUS},
2306 { MEDIA_MIMETYPE_AUDIO_AC3, AUDIO_FORMAT_AC3},
2307 { MEDIA_MIMETYPE_AUDIO_EAC3, AUDIO_FORMAT_E_AC3},
2308 { MEDIA_MIMETYPE_AUDIO_EAC3_JOC, AUDIO_FORMAT_E_AC3_JOC},
2309 { MEDIA_MIMETYPE_AUDIO_AC4, AUDIO_FORMAT_AC4},
2310 { MEDIA_MIMETYPE_AUDIO_FLAC, AUDIO_FORMAT_FLAC},
2311 { MEDIA_MIMETYPE_AUDIO_ALAC, AUDIO_FORMAT_ALAC },
2312 { 0, AUDIO_FORMAT_INVALID }
2313 };
2314
mapMimeToAudioFormat(audio_format_t & format,const char * mime)2315 status_t mapMimeToAudioFormat( audio_format_t& format, const char* mime )
2316 {
2317 const struct mime_conv_t* p = &mimeLookup[0];
2318 while (p->mime != NULL) {
2319 if (0 == strcasecmp(mime, p->mime)) {
2320 format = p->format;
2321 return OK;
2322 }
2323 ++p;
2324 }
2325
2326 return BAD_VALUE;
2327 }
2328
2329 struct aac_format_conv_t {
2330 int32_t eAacProfileType;
2331 audio_format_t format;
2332 };
2333
2334 static const struct aac_format_conv_t profileLookup[] = {
2335 { AACObjectMain, AUDIO_FORMAT_AAC_MAIN},
2336 { AACObjectLC, AUDIO_FORMAT_AAC_LC},
2337 { AACObjectSSR, AUDIO_FORMAT_AAC_SSR},
2338 { AACObjectLTP, AUDIO_FORMAT_AAC_LTP},
2339 { AACObjectHE, AUDIO_FORMAT_AAC_HE_V1},
2340 { AACObjectScalable, AUDIO_FORMAT_AAC_SCALABLE},
2341 { AACObjectERLC, AUDIO_FORMAT_AAC_ERLC},
2342 { AACObjectLD, AUDIO_FORMAT_AAC_LD},
2343 { AACObjectHE_PS, AUDIO_FORMAT_AAC_HE_V2},
2344 { AACObjectELD, AUDIO_FORMAT_AAC_ELD},
2345 { AACObjectXHE, AUDIO_FORMAT_AAC_XHE},
2346 { AACObjectNull, AUDIO_FORMAT_AAC},
2347 };
2348
mapAACProfileToAudioFormat(audio_format_t & format,uint64_t eAacProfile)2349 void mapAACProfileToAudioFormat( audio_format_t& format, uint64_t eAacProfile)
2350 {
2351 const struct aac_format_conv_t* p = &profileLookup[0];
2352 while (p->eAacProfileType != AACObjectNull) {
2353 if (eAacProfile == p->eAacProfileType) {
2354 format = p->format;
2355 return;
2356 }
2357 ++p;
2358 }
2359 format = AUDIO_FORMAT_AAC;
2360 return;
2361 }
2362
getAudioOffloadInfo(const sp<MetaData> & meta,bool hasVideo,bool isStreaming,audio_stream_type_t streamType,audio_offload_info_t * info)2363 status_t getAudioOffloadInfo(const sp<MetaData>& meta, bool hasVideo,
2364 bool isStreaming, audio_stream_type_t streamType, audio_offload_info_t *info)
2365 {
2366 const char *mime;
2367 if (meta == NULL) {
2368 return BAD_VALUE;
2369 }
2370 CHECK(meta->findCString(kKeyMIMEType, &mime));
2371
2372 (*info) = AUDIO_INFO_INITIALIZER;
2373
2374 info->format = AUDIO_FORMAT_INVALID;
2375 if (mapMimeToAudioFormat(info->format, mime) != OK) {
2376 ALOGE(" Couldn't map mime type \"%s\" to a valid AudioSystem::audio_format !", mime);
2377 return BAD_VALUE;
2378 } else {
2379 ALOGV("Mime type \"%s\" mapped to audio_format %d", mime, info->format);
2380 }
2381
2382 if (AUDIO_FORMAT_INVALID == info->format) {
2383 // can't offload if we don't know what the source format is
2384 ALOGE("mime type \"%s\" not a known audio format", mime);
2385 return BAD_VALUE;
2386 }
2387
2388 // Redefine aac format according to its profile
2389 // Offloading depends on audio DSP capabilities.
2390 int32_t aacaot = -1;
2391 if (meta->findInt32(kKeyAACAOT, &aacaot)) {
2392 mapAACProfileToAudioFormat(info->format, aacaot);
2393 }
2394
2395 int32_t srate = -1;
2396 if (!meta->findInt32(kKeySampleRate, &srate)) {
2397 ALOGV("track of type '%s' does not publish sample rate", mime);
2398 }
2399 info->sample_rate = srate;
2400
2401 int32_t rawChannelMask;
2402 audio_channel_mask_t cmask = meta->findInt32(kKeyChannelMask, &rawChannelMask) ?
2403 static_cast<audio_channel_mask_t>(rawChannelMask) : CHANNEL_MASK_USE_CHANNEL_ORDER;
2404 if (cmask == CHANNEL_MASK_USE_CHANNEL_ORDER) {
2405 ALOGV("track of type '%s' does not publish channel mask", mime);
2406
2407 // Try a channel count instead
2408 int32_t channelCount;
2409 if (!meta->findInt32(kKeyChannelCount, &channelCount)) {
2410 ALOGV("track of type '%s' does not publish channel count", mime);
2411 } else {
2412 cmask = audio_channel_out_mask_from_count(channelCount);
2413 }
2414 }
2415 info->channel_mask = cmask;
2416
2417 int64_t duration = 0;
2418 if (!meta->findInt64(kKeyDuration, &duration)) {
2419 ALOGV("track of type '%s' does not publish duration", mime);
2420 }
2421 info->duration_us = duration;
2422
2423 int32_t brate = 0;
2424 if (!meta->findInt32(kKeyBitRate, &brate)) {
2425 ALOGV("track of type '%s' does not publish bitrate", mime);
2426 }
2427 info->bit_rate = brate;
2428
2429
2430 info->stream_type = streamType;
2431 info->has_video = hasVideo;
2432 info->is_streaming = isStreaming;
2433 return OK;
2434 }
2435
canOffloadStream(const sp<MetaData> & meta,bool hasVideo,bool isStreaming,audio_stream_type_t streamType)2436 bool canOffloadStream(const sp<MetaData>& meta, bool hasVideo,
2437 bool isStreaming, audio_stream_type_t streamType)
2438 {
2439 audio_offload_info_t info = AUDIO_INFO_INITIALIZER;
2440 if (OK != getAudioOffloadInfo(meta, hasVideo, isStreaming, streamType, &info)) {
2441 return false;
2442 }
2443 // Check if offload is possible for given format, stream type, sample rate,
2444 // bit rate, duration, video and streaming
2445 #ifdef DISABLE_AUDIO_SYSTEM_OFFLOAD
2446 return false;
2447 #else
2448 return AudioSystem::getOffloadSupport(info) != AUDIO_OFFLOAD_NOT_SUPPORTED;
2449 #endif
2450 }
2451
HLSTime(const sp<AMessage> & meta)2452 HLSTime::HLSTime(const sp<AMessage>& meta) :
2453 mSeq(-1),
2454 mTimeUs(-1LL),
2455 mMeta(meta) {
2456 if (meta != NULL) {
2457 CHECK(meta->findInt32("discontinuitySeq", &mSeq));
2458 CHECK(meta->findInt64("timeUs", &mTimeUs));
2459 }
2460 }
2461
getSegmentTimeUs() const2462 int64_t HLSTime::getSegmentTimeUs() const {
2463 int64_t segmentStartTimeUs = -1LL;
2464 if (mMeta != NULL) {
2465 CHECK(mMeta->findInt64("segmentStartTimeUs", &segmentStartTimeUs));
2466
2467 int64_t segmentFirstTimeUs;
2468 if (mMeta->findInt64("segmentFirstTimeUs", &segmentFirstTimeUs)) {
2469 segmentStartTimeUs += mTimeUs - segmentFirstTimeUs;
2470 }
2471
2472 // adjust segment time by playlist age (for live streaming)
2473 int64_t playlistTimeUs;
2474 if (mMeta->findInt64("playlistTimeUs", &playlistTimeUs)) {
2475 int64_t playlistAgeUs = ALooper::GetNowUs() - playlistTimeUs;
2476
2477 int64_t durationUs;
2478 CHECK(mMeta->findInt64("segmentDurationUs", &durationUs));
2479
2480 // round to nearest whole segment
2481 playlistAgeUs = (playlistAgeUs + durationUs / 2)
2482 / durationUs * durationUs;
2483
2484 segmentStartTimeUs -= playlistAgeUs;
2485 if (segmentStartTimeUs < 0) {
2486 segmentStartTimeUs = 0;
2487 }
2488 }
2489 }
2490 return segmentStartTimeUs;
2491 }
2492
operator <(const HLSTime & t0,const HLSTime & t1)2493 bool operator <(const HLSTime &t0, const HLSTime &t1) {
2494 // we can only compare discontinuity sequence and timestamp.
2495 // (mSegmentTimeUs is not reliable in live streaming case, it's the
2496 // time starting from beginning of playlist but playlist could change.)
2497 return t0.mSeq < t1.mSeq
2498 || (t0.mSeq == t1.mSeq && t0.mTimeUs < t1.mTimeUs);
2499 }
2500
writeToAMessage(const sp<AMessage> & msg,const AudioPlaybackRate & rate)2501 void writeToAMessage(const sp<AMessage> &msg, const AudioPlaybackRate &rate) {
2502 msg->setFloat("speed", rate.mSpeed);
2503 msg->setFloat("pitch", rate.mPitch);
2504 msg->setInt32("audio-fallback-mode", rate.mFallbackMode);
2505 msg->setInt32("audio-stretch-mode", rate.mStretchMode);
2506 }
2507
readFromAMessage(const sp<AMessage> & msg,AudioPlaybackRate * rate)2508 void readFromAMessage(const sp<AMessage> &msg, AudioPlaybackRate *rate /* nonnull */) {
2509 *rate = AUDIO_PLAYBACK_RATE_DEFAULT;
2510 CHECK(msg->findFloat("speed", &rate->mSpeed));
2511 CHECK(msg->findFloat("pitch", &rate->mPitch));
2512 CHECK(msg->findInt32("audio-fallback-mode", (int32_t *)&rate->mFallbackMode));
2513 CHECK(msg->findInt32("audio-stretch-mode", (int32_t *)&rate->mStretchMode));
2514 }
2515
writeToAMessage(const sp<AMessage> & msg,const AVSyncSettings & sync,float videoFpsHint)2516 void writeToAMessage(const sp<AMessage> &msg, const AVSyncSettings &sync, float videoFpsHint) {
2517 msg->setInt32("sync-source", sync.mSource);
2518 msg->setInt32("audio-adjust-mode", sync.mAudioAdjustMode);
2519 msg->setFloat("tolerance", sync.mTolerance);
2520 msg->setFloat("video-fps", videoFpsHint);
2521 }
2522
readFromAMessage(const sp<AMessage> & msg,AVSyncSettings * sync,float * videoFps)2523 void readFromAMessage(
2524 const sp<AMessage> &msg,
2525 AVSyncSettings *sync /* nonnull */,
2526 float *videoFps /* nonnull */) {
2527 AVSyncSettings settings;
2528 CHECK(msg->findInt32("sync-source", (int32_t *)&settings.mSource));
2529 CHECK(msg->findInt32("audio-adjust-mode", (int32_t *)&settings.mAudioAdjustMode));
2530 CHECK(msg->findFloat("tolerance", &settings.mTolerance));
2531 CHECK(msg->findFloat("video-fps", videoFps));
2532 *sync = settings;
2533 }
2534
writeToAMessage(const sp<AMessage> & msg,const BufferingSettings & buffering)2535 void writeToAMessage(const sp<AMessage> &msg, const BufferingSettings &buffering) {
2536 msg->setInt32("init-ms", buffering.mInitialMarkMs);
2537 msg->setInt32("resume-playback-ms", buffering.mResumePlaybackMarkMs);
2538 }
2539
readFromAMessage(const sp<AMessage> & msg,BufferingSettings * buffering)2540 void readFromAMessage(const sp<AMessage> &msg, BufferingSettings *buffering /* nonnull */) {
2541 int32_t value;
2542 if (msg->findInt32("init-ms", &value)) {
2543 buffering->mInitialMarkMs = value;
2544 }
2545 if (msg->findInt32("resume-playback-ms", &value)) {
2546 buffering->mResumePlaybackMarkMs = value;
2547 }
2548 }
2549
2550 } // namespace android
2551