| 1 | /* |
| 2 | * AVFoundation input device |
| 3 | * Copyright (c) 2014 Thilo Borgmann <thilo.borgmann@mail.de> |
| 4 | * |
| 5 | * This file is part of FFmpeg. |
| 6 | * |
| 7 | * FFmpeg is free software; you can redistribute it and/or |
| 8 | * modify it under the terms of the GNU Lesser General Public |
| 9 | * License as published by the Free Software Foundation; either |
| 10 | * version 2.1 of the License, or (at your option) any later version. |
| 11 | * |
| 12 | * FFmpeg is distributed in the hope that it will be useful, |
| 13 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 14 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 15 | * Lesser General Public License for more details. |
| 16 | * |
| 17 | * You should have received a copy of the GNU Lesser General Public |
| 18 | * License along with FFmpeg; if not, write to the Free Software |
| 19 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
| 20 | */ |
| 21 | |
| 22 | /** |
| 23 | * @file |
| 24 | * AVFoundation input device |
| 25 | * @author Thilo Borgmann <thilo.borgmann@mail.de> |
| 26 | */ |
| 27 | |
| 28 | #import <AVFoundation/AVFoundation.h> |
| 29 | #include <pthread.h> |
| 30 | |
| 31 | #include "libavutil/pixdesc.h" |
| 32 | #include "libavutil/opt.h" |
| 33 | #include "libavutil/avstring.h" |
| 34 | #include "libavformat/internal.h" |
| 35 | #include "libavutil/internal.h" |
| 36 | #include "libavutil/time.h" |
| 37 | #include "avdevice.h" |
| 38 | |
| 39 | static const int avf_time_base = 1000000; |
| 40 | |
| 41 | static const AVRational avf_time_base_q = { |
| 42 | .num = 1, |
| 43 | .den = avf_time_base |
| 44 | }; |
| 45 | |
| 46 | struct AVFPixelFormatSpec { |
| 47 | enum AVPixelFormat ff_id; |
| 48 | OSType avf_id; |
| 49 | }; |
| 50 | |
| 51 | static const struct AVFPixelFormatSpec avf_pixel_formats[] = { |
| 52 | { AV_PIX_FMT_MONOBLACK, kCVPixelFormatType_1Monochrome }, |
| 53 | { AV_PIX_FMT_RGB555BE, kCVPixelFormatType_16BE555 }, |
| 54 | { AV_PIX_FMT_RGB555LE, kCVPixelFormatType_16LE555 }, |
| 55 | { AV_PIX_FMT_RGB565BE, kCVPixelFormatType_16BE565 }, |
| 56 | { AV_PIX_FMT_RGB565LE, kCVPixelFormatType_16LE565 }, |
| 57 | { AV_PIX_FMT_RGB24, kCVPixelFormatType_24RGB }, |
| 58 | { AV_PIX_FMT_BGR24, kCVPixelFormatType_24BGR }, |
| 59 | { AV_PIX_FMT_0RGB, kCVPixelFormatType_32ARGB }, |
| 60 | { AV_PIX_FMT_BGR0, kCVPixelFormatType_32BGRA }, |
| 61 | { AV_PIX_FMT_0BGR, kCVPixelFormatType_32ABGR }, |
| 62 | { AV_PIX_FMT_RGB0, kCVPixelFormatType_32RGBA }, |
| 63 | { AV_PIX_FMT_BGR48BE, kCVPixelFormatType_48RGB }, |
| 64 | { AV_PIX_FMT_UYVY422, kCVPixelFormatType_422YpCbCr8 }, |
| 65 | { AV_PIX_FMT_YUVA444P, kCVPixelFormatType_4444YpCbCrA8R }, |
| 66 | { AV_PIX_FMT_YUVA444P16LE, kCVPixelFormatType_4444AYpCbCr16 }, |
| 67 | { AV_PIX_FMT_YUV444P, kCVPixelFormatType_444YpCbCr8 }, |
| 68 | { AV_PIX_FMT_YUV422P16, kCVPixelFormatType_422YpCbCr16 }, |
| 69 | { AV_PIX_FMT_YUV422P10, kCVPixelFormatType_422YpCbCr10 }, |
| 70 | { AV_PIX_FMT_YUV444P10, kCVPixelFormatType_444YpCbCr10 }, |
| 71 | { AV_PIX_FMT_YUV420P, kCVPixelFormatType_420YpCbCr8Planar }, |
| 72 | { AV_PIX_FMT_NV12, kCVPixelFormatType_420YpCbCr8BiPlanarVideoRange }, |
| 73 | { AV_PIX_FMT_YUYV422, kCVPixelFormatType_422YpCbCr8_yuvs }, |
| 74 | #if __MAC_OS_X_VERSION_MIN_REQUIRED >= 1080 |
| 75 | { AV_PIX_FMT_GRAY8, kCVPixelFormatType_OneComponent8 }, |
| 76 | #endif |
| 77 | { AV_PIX_FMT_NONE, 0 } |
| 78 | }; |
| 79 | |
| 80 | typedef struct |
| 81 | { |
| 82 | AVClass* class; |
| 83 | |
| 84 | int frames_captured; |
| 85 | int audio_frames_captured; |
| 86 | int64_t first_pts; |
| 87 | int64_t first_audio_pts; |
| 88 | pthread_mutex_t frame_lock; |
| 89 | pthread_cond_t frame_wait_cond; |
| 90 | id avf_delegate; |
| 91 | id avf_audio_delegate; |
| 92 | |
| 93 | int list_devices; |
| 94 | int video_device_index; |
| 95 | int video_stream_index; |
| 96 | int audio_device_index; |
| 97 | int audio_stream_index; |
| 98 | |
| 99 | char *video_filename; |
| 100 | char *audio_filename; |
| 101 | |
| 102 | int num_video_devices; |
| 103 | |
| 104 | int audio_channels; |
| 105 | int audio_bits_per_sample; |
| 106 | int audio_float; |
| 107 | int audio_be; |
| 108 | int audio_signed_integer; |
| 109 | int audio_packed; |
| 110 | int audio_non_interleaved; |
| 111 | |
| 112 | int32_t *audio_buffer; |
| 113 | int audio_buffer_size; |
| 114 | |
| 115 | enum AVPixelFormat pixel_format; |
| 116 | |
| 117 | AVCaptureSession *capture_session; |
| 118 | AVCaptureVideoDataOutput *video_output; |
| 119 | AVCaptureAudioDataOutput *audio_output; |
| 120 | CMSampleBufferRef current_frame; |
| 121 | CMSampleBufferRef current_audio_frame; |
| 122 | } AVFContext; |
| 123 | |
| 124 | static void lock_frames(AVFContext* ctx) |
| 125 | { |
| 126 | pthread_mutex_lock(&ctx->frame_lock); |
| 127 | } |
| 128 | |
| 129 | static void unlock_frames(AVFContext* ctx) |
| 130 | { |
| 131 | pthread_mutex_unlock(&ctx->frame_lock); |
| 132 | } |
| 133 | |
| 134 | /** FrameReciever class - delegate for AVCaptureSession |
| 135 | */ |
| 136 | @interface AVFFrameReceiver : NSObject |
| 137 | { |
| 138 | AVFContext* _context; |
| 139 | } |
| 140 | |
| 141 | - (id)initWithContext:(AVFContext*)context; |
| 142 | |
| 143 | - (void) captureOutput:(AVCaptureOutput *)captureOutput |
| 144 | didOutputSampleBuffer:(CMSampleBufferRef)videoFrame |
| 145 | fromConnection:(AVCaptureConnection *)connection; |
| 146 | |
| 147 | @end |
| 148 | |
| 149 | @implementation AVFFrameReceiver |
| 150 | |
| 151 | - (id)initWithContext:(AVFContext*)context |
| 152 | { |
| 153 | if (self = [super init]) { |
| 154 | _context = context; |
| 155 | } |
| 156 | return self; |
| 157 | } |
| 158 | |
| 159 | - (void) captureOutput:(AVCaptureOutput *)captureOutput |
| 160 | didOutputSampleBuffer:(CMSampleBufferRef)videoFrame |
| 161 | fromConnection:(AVCaptureConnection *)connection |
| 162 | { |
| 163 | lock_frames(_context); |
| 164 | |
| 165 | if (_context->current_frame != nil) { |
| 166 | CFRelease(_context->current_frame); |
| 167 | } |
| 168 | |
| 169 | _context->current_frame = (CMSampleBufferRef)CFRetain(videoFrame); |
| 170 | |
| 171 | pthread_cond_signal(&_context->frame_wait_cond); |
| 172 | |
| 173 | unlock_frames(_context); |
| 174 | |
| 175 | ++_context->frames_captured; |
| 176 | } |
| 177 | |
| 178 | @end |
| 179 | |
| 180 | /** AudioReciever class - delegate for AVCaptureSession |
| 181 | */ |
| 182 | @interface AVFAudioReceiver : NSObject |
| 183 | { |
| 184 | AVFContext* _context; |
| 185 | } |
| 186 | |
| 187 | - (id)initWithContext:(AVFContext*)context; |
| 188 | |
| 189 | - (void) captureOutput:(AVCaptureOutput *)captureOutput |
| 190 | didOutputSampleBuffer:(CMSampleBufferRef)audioFrame |
| 191 | fromConnection:(AVCaptureConnection *)connection; |
| 192 | |
| 193 | @end |
| 194 | |
| 195 | @implementation AVFAudioReceiver |
| 196 | |
| 197 | - (id)initWithContext:(AVFContext*)context |
| 198 | { |
| 199 | if (self = [super init]) { |
| 200 | _context = context; |
| 201 | } |
| 202 | return self; |
| 203 | } |
| 204 | |
| 205 | - (void) captureOutput:(AVCaptureOutput *)captureOutput |
| 206 | didOutputSampleBuffer:(CMSampleBufferRef)audioFrame |
| 207 | fromConnection:(AVCaptureConnection *)connection |
| 208 | { |
| 209 | lock_frames(_context); |
| 210 | |
| 211 | if (_context->current_audio_frame != nil) { |
| 212 | CFRelease(_context->current_audio_frame); |
| 213 | } |
| 214 | |
| 215 | _context->current_audio_frame = (CMSampleBufferRef)CFRetain(audioFrame); |
| 216 | |
| 217 | pthread_cond_signal(&_context->frame_wait_cond); |
| 218 | |
| 219 | unlock_frames(_context); |
| 220 | |
| 221 | ++_context->audio_frames_captured; |
| 222 | } |
| 223 | |
| 224 | @end |
| 225 | |
| 226 | static void destroy_context(AVFContext* ctx) |
| 227 | { |
| 228 | [ctx->capture_session stopRunning]; |
| 229 | |
| 230 | [ctx->capture_session release]; |
| 231 | [ctx->video_output release]; |
| 232 | [ctx->audio_output release]; |
| 233 | [ctx->avf_delegate release]; |
| 234 | [ctx->avf_audio_delegate release]; |
| 235 | |
| 236 | ctx->capture_session = NULL; |
| 237 | ctx->video_output = NULL; |
| 238 | ctx->audio_output = NULL; |
| 239 | ctx->avf_delegate = NULL; |
| 240 | ctx->avf_audio_delegate = NULL; |
| 241 | |
| 242 | av_freep(&ctx->audio_buffer); |
| 243 | |
| 244 | pthread_mutex_destroy(&ctx->frame_lock); |
| 245 | pthread_cond_destroy(&ctx->frame_wait_cond); |
| 246 | |
| 247 | if (ctx->current_frame) { |
| 248 | CFRelease(ctx->current_frame); |
| 249 | } |
| 250 | } |
| 251 | |
| 252 | static void parse_device_name(AVFormatContext *s) |
| 253 | { |
| 254 | AVFContext *ctx = (AVFContext*)s->priv_data; |
| 255 | char *tmp = av_strdup(s->filename); |
| 256 | char *save; |
| 257 | |
| 258 | if (tmp[0] != ':') { |
| 259 | ctx->video_filename = av_strtok(tmp, ":", &save); |
| 260 | ctx->audio_filename = av_strtok(NULL, ":", &save); |
| 261 | } else { |
| 262 | ctx->audio_filename = av_strtok(tmp, ":", &save); |
| 263 | } |
| 264 | } |
| 265 | |
| 266 | static int add_video_device(AVFormatContext *s, AVCaptureDevice *video_device) |
| 267 | { |
| 268 | AVFContext *ctx = (AVFContext*)s->priv_data; |
| 269 | NSError *error = nil; |
| 270 | AVCaptureInput* capture_input = nil; |
| 271 | |
| 272 | if (ctx->video_device_index < ctx->num_video_devices) { |
| 273 | capture_input = (AVCaptureInput*) [[[AVCaptureDeviceInput alloc] initWithDevice:video_device error:&error] autorelease]; |
| 274 | } else { |
| 275 | capture_input = (AVCaptureInput*) video_device; |
| 276 | } |
| 277 | |
| 278 | if (!capture_input) { |
| 279 | av_log(s, AV_LOG_ERROR, "Failed to create AV capture input device: %s\n", |
| 280 | [[error localizedDescription] UTF8String]); |
| 281 | return 1; |
| 282 | } |
| 283 | |
| 284 | if ([ctx->capture_session canAddInput:capture_input]) { |
| 285 | [ctx->capture_session addInput:capture_input]; |
| 286 | } else { |
| 287 | av_log(s, AV_LOG_ERROR, "can't add video input to capture session\n"); |
| 288 | return 1; |
| 289 | } |
| 290 | |
| 291 | // Attaching output |
| 292 | ctx->video_output = [[AVCaptureVideoDataOutput alloc] init]; |
| 293 | |
| 294 | if (!ctx->video_output) { |
| 295 | av_log(s, AV_LOG_ERROR, "Failed to init AV video output\n"); |
| 296 | return 1; |
| 297 | } |
| 298 | |
| 299 | // select pixel format |
| 300 | struct AVFPixelFormatSpec pxl_fmt_spec; |
| 301 | pxl_fmt_spec.ff_id = AV_PIX_FMT_NONE; |
| 302 | |
| 303 | for (int i = 0; avf_pixel_formats[i].ff_id != AV_PIX_FMT_NONE; i++) { |
| 304 | if (ctx->pixel_format == avf_pixel_formats[i].ff_id) { |
| 305 | pxl_fmt_spec = avf_pixel_formats[i]; |
| 306 | break; |
| 307 | } |
| 308 | } |
| 309 | |
| 310 | // check if selected pixel format is supported by AVFoundation |
| 311 | if (pxl_fmt_spec.ff_id == AV_PIX_FMT_NONE) { |
| 312 | av_log(s, AV_LOG_ERROR, "Selected pixel format (%s) is not supported by AVFoundation.\n", |
| 313 | av_get_pix_fmt_name(pxl_fmt_spec.ff_id)); |
| 314 | return 1; |
| 315 | } |
| 316 | |
| 317 | // check if the pixel format is available for this device |
| 318 | if ([[ctx->video_output availableVideoCVPixelFormatTypes] indexOfObject:[NSNumber numberWithInt:pxl_fmt_spec.avf_id]] == NSNotFound) { |
| 319 | av_log(s, AV_LOG_ERROR, "Selected pixel format (%s) is not supported by the input device.\n", |
| 320 | av_get_pix_fmt_name(pxl_fmt_spec.ff_id)); |
| 321 | |
| 322 | pxl_fmt_spec.ff_id = AV_PIX_FMT_NONE; |
| 323 | |
| 324 | av_log(s, AV_LOG_ERROR, "Supported pixel formats:\n"); |
| 325 | for (NSNumber *pxl_fmt in [ctx->video_output availableVideoCVPixelFormatTypes]) { |
| 326 | struct AVFPixelFormatSpec pxl_fmt_dummy; |
| 327 | pxl_fmt_dummy.ff_id = AV_PIX_FMT_NONE; |
| 328 | for (int i = 0; avf_pixel_formats[i].ff_id != AV_PIX_FMT_NONE; i++) { |
| 329 | if ([pxl_fmt intValue] == avf_pixel_formats[i].avf_id) { |
| 330 | pxl_fmt_dummy = avf_pixel_formats[i]; |
| 331 | break; |
| 332 | } |
| 333 | } |
| 334 | |
| 335 | if (pxl_fmt_dummy.ff_id != AV_PIX_FMT_NONE) { |
| 336 | av_log(s, AV_LOG_ERROR, " %s\n", av_get_pix_fmt_name(pxl_fmt_dummy.ff_id)); |
| 337 | |
| 338 | // select first supported pixel format instead of user selected (or default) pixel format |
| 339 | if (pxl_fmt_spec.ff_id == AV_PIX_FMT_NONE) { |
| 340 | pxl_fmt_spec = pxl_fmt_dummy; |
| 341 | } |
| 342 | } |
| 343 | } |
| 344 | |
| 345 | // fail if there is no appropriate pixel format or print a warning about overriding the pixel format |
| 346 | if (pxl_fmt_spec.ff_id == AV_PIX_FMT_NONE) { |
| 347 | return 1; |
| 348 | } else { |
| 349 | av_log(s, AV_LOG_WARNING, "Overriding selected pixel format to use %s instead.\n", |
| 350 | av_get_pix_fmt_name(pxl_fmt_spec.ff_id)); |
| 351 | } |
| 352 | } |
| 353 | |
| 354 | ctx->pixel_format = pxl_fmt_spec.ff_id; |
| 355 | NSNumber *pixel_format = [NSNumber numberWithUnsignedInt:pxl_fmt_spec.avf_id]; |
| 356 | NSDictionary *capture_dict = [NSDictionary dictionaryWithObject:pixel_format |
| 357 | forKey:(id)kCVPixelBufferPixelFormatTypeKey]; |
| 358 | |
| 359 | [ctx->video_output setVideoSettings:capture_dict]; |
| 360 | [ctx->video_output setAlwaysDiscardsLateVideoFrames:YES]; |
| 361 | |
| 362 | ctx->avf_delegate = [[AVFFrameReceiver alloc] initWithContext:ctx]; |
| 363 | |
| 364 | dispatch_queue_t queue = dispatch_queue_create("avf_queue", NULL); |
| 365 | [ctx->video_output setSampleBufferDelegate:ctx->avf_delegate queue:queue]; |
| 366 | dispatch_release(queue); |
| 367 | |
| 368 | if ([ctx->capture_session canAddOutput:ctx->video_output]) { |
| 369 | [ctx->capture_session addOutput:ctx->video_output]; |
| 370 | } else { |
| 371 | av_log(s, AV_LOG_ERROR, "can't add video output to capture session\n"); |
| 372 | return 1; |
| 373 | } |
| 374 | |
| 375 | return 0; |
| 376 | } |
| 377 | |
| 378 | static int add_audio_device(AVFormatContext *s, AVCaptureDevice *audio_device) |
| 379 | { |
| 380 | AVFContext *ctx = (AVFContext*)s->priv_data; |
| 381 | NSError *error = nil; |
| 382 | AVCaptureDeviceInput* audio_dev_input = [[[AVCaptureDeviceInput alloc] initWithDevice:audio_device error:&error] autorelease]; |
| 383 | |
| 384 | if (!audio_dev_input) { |
| 385 | av_log(s, AV_LOG_ERROR, "Failed to create AV capture input device: %s\n", |
| 386 | [[error localizedDescription] UTF8String]); |
| 387 | return 1; |
| 388 | } |
| 389 | |
| 390 | if ([ctx->capture_session canAddInput:audio_dev_input]) { |
| 391 | [ctx->capture_session addInput:audio_dev_input]; |
| 392 | } else { |
| 393 | av_log(s, AV_LOG_ERROR, "can't add audio input to capture session\n"); |
| 394 | return 1; |
| 395 | } |
| 396 | |
| 397 | // Attaching output |
| 398 | ctx->audio_output = [[AVCaptureAudioDataOutput alloc] init]; |
| 399 | |
| 400 | if (!ctx->audio_output) { |
| 401 | av_log(s, AV_LOG_ERROR, "Failed to init AV audio output\n"); |
| 402 | return 1; |
| 403 | } |
| 404 | |
| 405 | ctx->avf_audio_delegate = [[AVFAudioReceiver alloc] initWithContext:ctx]; |
| 406 | |
| 407 | dispatch_queue_t queue = dispatch_queue_create("avf_audio_queue", NULL); |
| 408 | [ctx->audio_output setSampleBufferDelegate:ctx->avf_audio_delegate queue:queue]; |
| 409 | dispatch_release(queue); |
| 410 | |
| 411 | if ([ctx->capture_session canAddOutput:ctx->audio_output]) { |
| 412 | [ctx->capture_session addOutput:ctx->audio_output]; |
| 413 | } else { |
| 414 | av_log(s, AV_LOG_ERROR, "adding audio output to capture session failed\n"); |
| 415 | return 1; |
| 416 | } |
| 417 | |
| 418 | return 0; |
| 419 | } |
| 420 | |
| 421 | static int get_video_config(AVFormatContext *s) |
| 422 | { |
| 423 | AVFContext *ctx = (AVFContext*)s->priv_data; |
| 424 | |
| 425 | // Take stream info from the first frame. |
| 426 | while (ctx->frames_captured < 1) { |
| 427 | CFRunLoopRunInMode(kCFRunLoopDefaultMode, 0.1, YES); |
| 428 | } |
| 429 | |
| 430 | lock_frames(ctx); |
| 431 | |
| 432 | AVStream* stream = avformat_new_stream(s, NULL); |
| 433 | |
| 434 | if (!stream) { |
| 435 | return 1; |
| 436 | } |
| 437 | |
| 438 | ctx->video_stream_index = stream->index; |
| 439 | |
| 440 | avpriv_set_pts_info(stream, 64, 1, avf_time_base); |
| 441 | |
| 442 | CVImageBufferRef image_buffer = CMSampleBufferGetImageBuffer(ctx->current_frame); |
| 443 | CGSize image_buffer_size = CVImageBufferGetEncodedSize(image_buffer); |
| 444 | |
| 445 | stream->codec->codec_id = AV_CODEC_ID_RAWVIDEO; |
| 446 | stream->codec->codec_type = AVMEDIA_TYPE_VIDEO; |
| 447 | stream->codec->width = (int)image_buffer_size.width; |
| 448 | stream->codec->height = (int)image_buffer_size.height; |
| 449 | stream->codec->pix_fmt = ctx->pixel_format; |
| 450 | |
| 451 | CFRelease(ctx->current_frame); |
| 452 | ctx->current_frame = nil; |
| 453 | |
| 454 | unlock_frames(ctx); |
| 455 | |
| 456 | return 0; |
| 457 | } |
| 458 | |
| 459 | static int get_audio_config(AVFormatContext *s) |
| 460 | { |
| 461 | AVFContext *ctx = (AVFContext*)s->priv_data; |
| 462 | |
| 463 | // Take stream info from the first frame. |
| 464 | while (ctx->audio_frames_captured < 1) { |
| 465 | CFRunLoopRunInMode(kCFRunLoopDefaultMode, 0.1, YES); |
| 466 | } |
| 467 | |
| 468 | lock_frames(ctx); |
| 469 | |
| 470 | AVStream* stream = avformat_new_stream(s, NULL); |
| 471 | |
| 472 | if (!stream) { |
| 473 | return 1; |
| 474 | } |
| 475 | |
| 476 | ctx->audio_stream_index = stream->index; |
| 477 | |
| 478 | avpriv_set_pts_info(stream, 64, 1, avf_time_base); |
| 479 | |
| 480 | CMFormatDescriptionRef format_desc = CMSampleBufferGetFormatDescription(ctx->current_audio_frame); |
| 481 | const AudioStreamBasicDescription *basic_desc = CMAudioFormatDescriptionGetStreamBasicDescription(format_desc); |
| 482 | |
| 483 | if (!basic_desc) { |
| 484 | av_log(s, AV_LOG_ERROR, "audio format not available\n"); |
| 485 | return 1; |
| 486 | } |
| 487 | |
| 488 | stream->codec->codec_type = AVMEDIA_TYPE_AUDIO; |
| 489 | stream->codec->sample_rate = basic_desc->mSampleRate; |
| 490 | stream->codec->channels = basic_desc->mChannelsPerFrame; |
| 491 | stream->codec->channel_layout = av_get_default_channel_layout(stream->codec->channels); |
| 492 | |
| 493 | ctx->audio_channels = basic_desc->mChannelsPerFrame; |
| 494 | ctx->audio_bits_per_sample = basic_desc->mBitsPerChannel; |
| 495 | ctx->audio_float = basic_desc->mFormatFlags & kAudioFormatFlagIsFloat; |
| 496 | ctx->audio_be = basic_desc->mFormatFlags & kAudioFormatFlagIsBigEndian; |
| 497 | ctx->audio_signed_integer = basic_desc->mFormatFlags & kAudioFormatFlagIsSignedInteger; |
| 498 | ctx->audio_packed = basic_desc->mFormatFlags & kAudioFormatFlagIsPacked; |
| 499 | ctx->audio_non_interleaved = basic_desc->mFormatFlags & kAudioFormatFlagIsNonInterleaved; |
| 500 | |
| 501 | if (basic_desc->mFormatID == kAudioFormatLinearPCM && |
| 502 | ctx->audio_float && |
| 503 | ctx->audio_packed) { |
| 504 | stream->codec->codec_id = ctx->audio_be ? AV_CODEC_ID_PCM_F32BE : AV_CODEC_ID_PCM_F32LE; |
| 505 | } else { |
| 506 | av_log(s, AV_LOG_ERROR, "audio format is not supported\n"); |
| 507 | return 1; |
| 508 | } |
| 509 | |
| 510 | if (ctx->audio_non_interleaved) { |
| 511 | CMBlockBufferRef block_buffer = CMSampleBufferGetDataBuffer(ctx->current_audio_frame); |
| 512 | ctx->audio_buffer_size = CMBlockBufferGetDataLength(block_buffer); |
| 513 | ctx->audio_buffer = av_malloc(ctx->audio_buffer_size); |
| 514 | if (!ctx->audio_buffer) { |
| 515 | av_log(s, AV_LOG_ERROR, "error allocating audio buffer\n"); |
| 516 | return 1; |
| 517 | } |
| 518 | } |
| 519 | |
| 520 | CFRelease(ctx->current_audio_frame); |
| 521 | ctx->current_audio_frame = nil; |
| 522 | |
| 523 | unlock_frames(ctx); |
| 524 | |
| 525 | return 0; |
| 526 | } |
| 527 | |
| 528 | static int avf_read_header(AVFormatContext *s) |
| 529 | { |
| 530 | NSAutoreleasePool *pool = [[NSAutoreleasePool alloc] init]; |
| 531 | AVFContext *ctx = (AVFContext*)s->priv_data; |
| 532 | ctx->first_pts = av_gettime(); |
| 533 | ctx->first_audio_pts = av_gettime(); |
| 534 | uint32_t num_screens = 0; |
| 535 | |
| 536 | pthread_mutex_init(&ctx->frame_lock, NULL); |
| 537 | pthread_cond_init(&ctx->frame_wait_cond, NULL); |
| 538 | |
| 539 | #if __MAC_OS_X_VERSION_MIN_REQUIRED >= 1070 |
| 540 | CGGetActiveDisplayList(0, NULL, &num_screens); |
| 541 | #endif |
| 542 | |
| 543 | // List devices if requested |
| 544 | if (ctx->list_devices) { |
| 545 | av_log(ctx, AV_LOG_INFO, "AVFoundation video devices:\n"); |
| 546 | NSArray *devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeVideo]; |
| 547 | int index = 0; |
| 548 | for (AVCaptureDevice *device in devices) { |
| 549 | const char *name = [[device localizedName] UTF8String]; |
| 550 | index = [devices indexOfObject:device]; |
| 551 | av_log(ctx, AV_LOG_INFO, "[%d] %s\n", index, name); |
| 552 | index++; |
| 553 | } |
| 554 | #if __MAC_OS_X_VERSION_MIN_REQUIRED >= 1070 |
| 555 | if (num_screens > 0) { |
| 556 | CGDirectDisplayID screens[num_screens]; |
| 557 | CGGetActiveDisplayList(num_screens, screens, &num_screens); |
| 558 | for (int i = 0; i < num_screens; i++) { |
| 559 | av_log(ctx, AV_LOG_INFO, "[%d] Capture screen %d\n", index + i, i); |
| 560 | } |
| 561 | } |
| 562 | #endif |
| 563 | |
| 564 | av_log(ctx, AV_LOG_INFO, "AVFoundation audio devices:\n"); |
| 565 | devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeAudio]; |
| 566 | for (AVCaptureDevice *device in devices) { |
| 567 | const char *name = [[device localizedName] UTF8String]; |
| 568 | int index = [devices indexOfObject:device]; |
| 569 | av_log(ctx, AV_LOG_INFO, "[%d] %s\n", index, name); |
| 570 | } |
| 571 | goto fail; |
| 572 | } |
| 573 | |
| 574 | // Find capture device |
| 575 | AVCaptureDevice *video_device = nil; |
| 576 | AVCaptureDevice *audio_device = nil; |
| 577 | |
| 578 | NSArray *video_devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeVideo]; |
| 579 | ctx->num_video_devices = [video_devices count]; |
| 580 | |
| 581 | // parse input filename for video and audio device |
| 582 | parse_device_name(s); |
| 583 | |
| 584 | // check for device index given in filename |
| 585 | if (ctx->video_device_index == -1 && ctx->video_filename) { |
| 586 | sscanf(ctx->video_filename, "%d", &ctx->video_device_index); |
| 587 | } |
| 588 | if (ctx->audio_device_index == -1 && ctx->audio_filename) { |
| 589 | sscanf(ctx->audio_filename, "%d", &ctx->audio_device_index); |
| 590 | } |
| 591 | |
| 592 | if (ctx->video_device_index >= 0) { |
| 593 | if (ctx->video_device_index < ctx->num_video_devices) { |
| 594 | video_device = [video_devices objectAtIndex:ctx->video_device_index]; |
| 595 | } else if (ctx->video_device_index < ctx->num_video_devices + num_screens) { |
| 596 | #if __MAC_OS_X_VERSION_MIN_REQUIRED >= 1070 |
| 597 | CGDirectDisplayID screens[num_screens]; |
| 598 | CGGetActiveDisplayList(num_screens, screens, &num_screens); |
| 599 | AVCaptureScreenInput* capture_screen_input = [[[AVCaptureScreenInput alloc] initWithDisplayID:screens[ctx->video_device_index - ctx->num_video_devices]] autorelease]; |
| 600 | video_device = (AVCaptureDevice*) capture_screen_input; |
| 601 | #endif |
| 602 | } else { |
| 603 | av_log(ctx, AV_LOG_ERROR, "Invalid device index\n"); |
| 604 | goto fail; |
| 605 | } |
| 606 | } else if (ctx->video_filename && |
| 607 | strncmp(ctx->video_filename, "none", 4)) { |
| 608 | if (!strncmp(ctx->video_filename, "default", 7)) { |
| 609 | video_device = [AVCaptureDevice defaultDeviceWithMediaType:AVMediaTypeVideo]; |
| 610 | } else { |
| 611 | // looking for video inputs |
| 612 | for (AVCaptureDevice *device in video_devices) { |
| 613 | if (!strncmp(ctx->video_filename, [[device localizedName] UTF8String], strlen(ctx->video_filename))) { |
| 614 | video_device = device; |
| 615 | break; |
| 616 | } |
| 617 | } |
| 618 | |
| 619 | #if __MAC_OS_X_VERSION_MIN_REQUIRED >= 1070 |
| 620 | // looking for screen inputs |
| 621 | if (!video_device) { |
| 622 | int idx; |
| 623 | if(sscanf(ctx->video_filename, "Capture screen %d", &idx) && idx < num_screens) { |
| 624 | CGDirectDisplayID screens[num_screens]; |
| 625 | CGGetActiveDisplayList(num_screens, screens, &num_screens); |
| 626 | AVCaptureScreenInput* capture_screen_input = [[[AVCaptureScreenInput alloc] initWithDisplayID:screens[idx]] autorelease]; |
| 627 | video_device = (AVCaptureDevice*) capture_screen_input; |
| 628 | ctx->video_device_index = ctx->num_video_devices + idx; |
| 629 | } |
| 630 | } |
| 631 | #endif |
| 632 | } |
| 633 | |
| 634 | if (!video_device) { |
| 635 | av_log(ctx, AV_LOG_ERROR, "Video device not found\n"); |
| 636 | goto fail; |
| 637 | } |
| 638 | } |
| 639 | |
| 640 | // get audio device |
| 641 | if (ctx->audio_device_index >= 0) { |
| 642 | NSArray *devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeAudio]; |
| 643 | |
| 644 | if (ctx->audio_device_index >= [devices count]) { |
| 645 | av_log(ctx, AV_LOG_ERROR, "Invalid audio device index\n"); |
| 646 | goto fail; |
| 647 | } |
| 648 | |
| 649 | audio_device = [devices objectAtIndex:ctx->audio_device_index]; |
| 650 | } else if (ctx->audio_filename && |
| 651 | strncmp(ctx->audio_filename, "none", 4)) { |
| 652 | if (!strncmp(ctx->audio_filename, "default", 7)) { |
| 653 | audio_device = [AVCaptureDevice defaultDeviceWithMediaType:AVMediaTypeAudio]; |
| 654 | } else { |
| 655 | NSArray *devices = [AVCaptureDevice devicesWithMediaType:AVMediaTypeAudio]; |
| 656 | |
| 657 | for (AVCaptureDevice *device in devices) { |
| 658 | if (!strncmp(ctx->audio_filename, [[device localizedName] UTF8String], strlen(ctx->audio_filename))) { |
| 659 | audio_device = device; |
| 660 | break; |
| 661 | } |
| 662 | } |
| 663 | } |
| 664 | |
| 665 | if (!audio_device) { |
| 666 | av_log(ctx, AV_LOG_ERROR, "Audio device not found\n"); |
| 667 | goto fail; |
| 668 | } |
| 669 | } |
| 670 | |
| 671 | // Video nor Audio capture device not found, looking for AVMediaTypeVideo/Audio |
| 672 | if (!video_device && !audio_device) { |
| 673 | av_log(s, AV_LOG_ERROR, "No AV capture device found\n"); |
| 674 | goto fail; |
| 675 | } |
| 676 | |
| 677 | if (video_device) { |
| 678 | if (ctx->video_device_index < ctx->num_video_devices) { |
| 679 | av_log(s, AV_LOG_DEBUG, "'%s' opened\n", [[video_device localizedName] UTF8String]); |
| 680 | } else { |
| 681 | av_log(s, AV_LOG_DEBUG, "'%s' opened\n", [[video_device description] UTF8String]); |
| 682 | } |
| 683 | } |
| 684 | if (audio_device) { |
| 685 | av_log(s, AV_LOG_DEBUG, "audio device '%s' opened\n", [[audio_device localizedName] UTF8String]); |
| 686 | } |
| 687 | |
| 688 | // Initialize capture session |
| 689 | ctx->capture_session = [[AVCaptureSession alloc] init]; |
| 690 | |
| 691 | if (video_device && add_video_device(s, video_device)) { |
| 692 | goto fail; |
| 693 | } |
| 694 | if (audio_device && add_audio_device(s, audio_device)) { |
| 695 | } |
| 696 | |
| 697 | [ctx->capture_session startRunning]; |
| 698 | |
| 699 | if (video_device && get_video_config(s)) { |
| 700 | goto fail; |
| 701 | } |
| 702 | |
| 703 | // set audio stream |
| 704 | if (audio_device && get_audio_config(s)) { |
| 705 | goto fail; |
| 706 | } |
| 707 | |
| 708 | [pool release]; |
| 709 | return 0; |
| 710 | |
| 711 | fail: |
| 712 | [pool release]; |
| 713 | destroy_context(ctx); |
| 714 | return AVERROR(EIO); |
| 715 | } |
| 716 | |
| 717 | static int avf_read_packet(AVFormatContext *s, AVPacket *pkt) |
| 718 | { |
| 719 | AVFContext* ctx = (AVFContext*)s->priv_data; |
| 720 | |
| 721 | do { |
| 722 | lock_frames(ctx); |
| 723 | |
| 724 | CVImageBufferRef image_buffer = CMSampleBufferGetImageBuffer(ctx->current_frame); |
| 725 | |
| 726 | if (ctx->current_frame != nil) { |
| 727 | if (av_new_packet(pkt, (int)CVPixelBufferGetDataSize(image_buffer)) < 0) { |
| 728 | return AVERROR(EIO); |
| 729 | } |
| 730 | |
| 731 | pkt->pts = pkt->dts = av_rescale_q(av_gettime() - ctx->first_pts, |
| 732 | AV_TIME_BASE_Q, |
| 733 | avf_time_base_q); |
| 734 | pkt->stream_index = ctx->video_stream_index; |
| 735 | pkt->flags |= AV_PKT_FLAG_KEY; |
| 736 | |
| 737 | CVPixelBufferLockBaseAddress(image_buffer, 0); |
| 738 | |
| 739 | void* data = CVPixelBufferGetBaseAddress(image_buffer); |
| 740 | memcpy(pkt->data, data, pkt->size); |
| 741 | |
| 742 | CVPixelBufferUnlockBaseAddress(image_buffer, 0); |
| 743 | CFRelease(ctx->current_frame); |
| 744 | ctx->current_frame = nil; |
| 745 | } else if (ctx->current_audio_frame != nil) { |
| 746 | CMBlockBufferRef block_buffer = CMSampleBufferGetDataBuffer(ctx->current_audio_frame); |
| 747 | int block_buffer_size = CMBlockBufferGetDataLength(block_buffer); |
| 748 | |
| 749 | if (!block_buffer || !block_buffer_size) { |
| 750 | return AVERROR(EIO); |
| 751 | } |
| 752 | |
| 753 | if (ctx->audio_non_interleaved && block_buffer_size > ctx->audio_buffer_size) { |
| 754 | return AVERROR_BUFFER_TOO_SMALL; |
| 755 | } |
| 756 | |
| 757 | if (av_new_packet(pkt, block_buffer_size) < 0) { |
| 758 | return AVERROR(EIO); |
| 759 | } |
| 760 | |
| 761 | pkt->pts = pkt->dts = av_rescale_q(av_gettime() - ctx->first_audio_pts, |
| 762 | AV_TIME_BASE_Q, |
| 763 | avf_time_base_q); |
| 764 | |
| 765 | pkt->stream_index = ctx->audio_stream_index; |
| 766 | pkt->flags |= AV_PKT_FLAG_KEY; |
| 767 | |
| 768 | if (ctx->audio_non_interleaved) { |
| 769 | int sample, c, shift; |
| 770 | |
| 771 | OSStatus ret = CMBlockBufferCopyDataBytes(block_buffer, 0, pkt->size, ctx->audio_buffer); |
| 772 | if (ret != kCMBlockBufferNoErr) { |
| 773 | return AVERROR(EIO); |
| 774 | } |
| 775 | |
| 776 | int num_samples = pkt->size / (ctx->audio_channels * (ctx->audio_bits_per_sample >> 3)); |
| 777 | |
| 778 | // transform decoded frame into output format |
| 779 | #define INTERLEAVE_OUTPUT(bps) \ |
| 780 | { \ |
| 781 | int##bps##_t **src; \ |
| 782 | int##bps##_t *dest; \ |
| 783 | src = av_malloc(ctx->audio_channels * sizeof(int##bps##_t*)); \ |
| 784 | if (!src) return AVERROR(EIO); \ |
| 785 | for (c = 0; c < ctx->audio_channels; c++) { \ |
| 786 | src[c] = ((int##bps##_t*)ctx->audio_buffer) + c * num_samples; \ |
| 787 | } \ |
| 788 | dest = (int##bps##_t*)pkt->data; \ |
| 789 | shift = bps - ctx->audio_bits_per_sample; \ |
| 790 | for (sample = 0; sample < num_samples; sample++) \ |
| 791 | for (c = 0; c < ctx->audio_channels; c++) \ |
| 792 | *dest++ = src[c][sample] << shift; \ |
| 793 | av_freep(&src); \ |
| 794 | } |
| 795 | |
| 796 | if (ctx->audio_bits_per_sample <= 16) { |
| 797 | INTERLEAVE_OUTPUT(16) |
| 798 | } else { |
| 799 | INTERLEAVE_OUTPUT(32) |
| 800 | } |
| 801 | } else { |
| 802 | OSStatus ret = CMBlockBufferCopyDataBytes(block_buffer, 0, pkt->size, pkt->data); |
| 803 | if (ret != kCMBlockBufferNoErr) { |
| 804 | return AVERROR(EIO); |
| 805 | } |
| 806 | } |
| 807 | |
| 808 | CFRelease(ctx->current_audio_frame); |
| 809 | ctx->current_audio_frame = nil; |
| 810 | } else { |
| 811 | pkt->data = NULL; |
| 812 | pthread_cond_wait(&ctx->frame_wait_cond, &ctx->frame_lock); |
| 813 | } |
| 814 | |
| 815 | unlock_frames(ctx); |
| 816 | } while (!pkt->data); |
| 817 | |
| 818 | return 0; |
| 819 | } |
| 820 | |
| 821 | static int avf_close(AVFormatContext *s) |
| 822 | { |
| 823 | AVFContext* ctx = (AVFContext*)s->priv_data; |
| 824 | destroy_context(ctx); |
| 825 | return 0; |
| 826 | } |
| 827 | |
| 828 | static const AVOption options[] = { |
| 829 | { "list_devices", "list available devices", offsetof(AVFContext, list_devices), AV_OPT_TYPE_INT, {.i64=0}, 0, 1, AV_OPT_FLAG_DECODING_PARAM, "list_devices" }, |
| 830 | { "true", "", 0, AV_OPT_TYPE_CONST, {.i64=1}, 0, 0, AV_OPT_FLAG_DECODING_PARAM, "list_devices" }, |
| 831 | { "false", "", 0, AV_OPT_TYPE_CONST, {.i64=0}, 0, 0, AV_OPT_FLAG_DECODING_PARAM, "list_devices" }, |
| 832 | { "video_device_index", "select video device by index for devices with same name (starts at 0)", offsetof(AVFContext, video_device_index), AV_OPT_TYPE_INT, {.i64 = -1}, -1, INT_MAX, AV_OPT_FLAG_DECODING_PARAM }, |
| 833 | { "audio_device_index", "select audio device by index for devices with same name (starts at 0)", offsetof(AVFContext, audio_device_index), AV_OPT_TYPE_INT, {.i64 = -1}, -1, INT_MAX, AV_OPT_FLAG_DECODING_PARAM }, |
| 834 | { "pixel_format", "set pixel format", offsetof(AVFContext, pixel_format), AV_OPT_TYPE_PIXEL_FMT, {.i64 = AV_PIX_FMT_YUV420P}, 0, INT_MAX, AV_OPT_FLAG_DECODING_PARAM}, |
| 835 | { NULL }, |
| 836 | }; |
| 837 | |
| 838 | static const AVClass avf_class = { |
| 839 | .class_name = "AVFoundation input device", |
| 840 | .item_name = av_default_item_name, |
| 841 | .option = options, |
| 842 | .version = LIBAVUTIL_VERSION_INT, |
| 843 | .category = AV_CLASS_CATEGORY_DEVICE_VIDEO_INPUT, |
| 844 | }; |
| 845 | |
| 846 | AVInputFormat ff_avfoundation_demuxer = { |
| 847 | .name = "avfoundation", |
| 848 | .long_name = NULL_IF_CONFIG_SMALL("AVFoundation input device"), |
| 849 | .priv_data_size = sizeof(AVFContext), |
| 850 | .read_header = avf_read_header, |
| 851 | .read_packet = avf_read_packet, |
| 852 | .read_close = avf_close, |
| 853 | .flags = AVFMT_NOFILE, |
| 854 | .priv_class = &avf_class, |
| 855 | }; |