2 * Directshow capture interface
3 * Copyright (c) 2010 Ramiro Polla
5 * This file is part of FFmpeg.
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.
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.
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
22 #include "libavutil/parseutils.h"
23 #include "libavutil/pixdesc.h"
24 #include "libavutil/opt.h"
25 #include "libavformat/internal.h"
27 #include "dshow_capture.h"
35 int video_device_number;
36 int audio_device_number;
40 int audio_buffer_size;
42 IBaseFilter *device_filter[2];
44 libAVFilter *capture_filter[2];
45 libAVPin *capture_pin[2];
48 HANDLE event[2]; /* event[0] is set by DirectShow
49 * event[1] is set by callback() */
55 unsigned int video_frame_num;
57 IMediaControl *control;
58 IMediaEvent *media_event;
60 enum AVPixelFormat pixel_format;
61 enum AVCodecID video_codec_id;
66 AVRational requested_framerate;
73 static enum AVPixelFormat dshow_pixfmt(DWORD biCompression, WORD biBitCount)
75 switch(biCompression) {
76 case MKTAG('U', 'Y', 'V', 'Y'):
77 return AV_PIX_FMT_UYVY422;
78 case MKTAG('Y', 'U', 'Y', '2'):
79 return AV_PIX_FMT_YUYV422;
80 case MKTAG('I', '4', '2', '0'):
81 return AV_PIX_FMT_YUV420P;
84 switch(biBitCount) { /* 1-8 are untested */
86 return AV_PIX_FMT_MONOWHITE;
88 return AV_PIX_FMT_RGB4;
90 return AV_PIX_FMT_RGB8;
92 return AV_PIX_FMT_RGB555;
94 return AV_PIX_FMT_BGR24;
96 return AV_PIX_FMT_RGB32;
99 return AV_PIX_FMT_NONE;
102 static enum AVCodecID dshow_codecid(DWORD biCompression)
104 switch(biCompression) {
105 case MKTAG('d', 'v', 's', 'd'):
106 return AV_CODEC_ID_DVVIDEO;
107 case MKTAG('M', 'J', 'P', 'G'):
108 case MKTAG('m', 'j', 'p', 'g'):
109 return AV_CODEC_ID_MJPEG;
111 return AV_CODEC_ID_NONE;
115 dshow_read_close(AVFormatContext *s)
117 struct dshow_ctx *ctx = s->priv_data;
121 IMediaControl_Stop(ctx->control);
122 IMediaControl_Release(ctx->control);
125 if (ctx->media_event)
126 IMediaEvent_Release(ctx->media_event);
131 r = IGraphBuilder_EnumFilters(ctx->graph, &fenum);
134 IEnumFilters_Reset(fenum);
135 while (IEnumFilters_Next(fenum, 1, &f, NULL) == S_OK) {
136 if (IGraphBuilder_RemoveFilter(ctx->graph, f) == S_OK)
137 IEnumFilters_Reset(fenum); /* When a filter is removed,
138 * the list must be reset. */
139 IBaseFilter_Release(f);
141 IEnumFilters_Release(fenum);
143 IGraphBuilder_Release(ctx->graph);
146 if (ctx->capture_pin[VideoDevice])
147 libAVPin_Release(ctx->capture_pin[VideoDevice]);
148 if (ctx->capture_pin[AudioDevice])
149 libAVPin_Release(ctx->capture_pin[AudioDevice]);
150 if (ctx->capture_filter[VideoDevice])
151 libAVFilter_Release(ctx->capture_filter[VideoDevice]);
152 if (ctx->capture_filter[AudioDevice])
153 libAVFilter_Release(ctx->capture_filter[AudioDevice]);
155 if (ctx->device_pin[VideoDevice])
156 IPin_Release(ctx->device_pin[VideoDevice]);
157 if (ctx->device_pin[AudioDevice])
158 IPin_Release(ctx->device_pin[AudioDevice]);
159 if (ctx->device_filter[VideoDevice])
160 IBaseFilter_Release(ctx->device_filter[VideoDevice]);
161 if (ctx->device_filter[AudioDevice])
162 IBaseFilter_Release(ctx->device_filter[AudioDevice]);
164 if (ctx->device_name[0])
165 av_free(ctx->device_name[0]);
166 if (ctx->device_name[1])
167 av_free(ctx->device_name[1]);
170 CloseHandle(ctx->mutex);
172 CloseHandle(ctx->event[0]);
174 CloseHandle(ctx->event[1]);
178 AVPacketList *next = pktl->next;
179 av_destruct_packet(&pktl->pkt);
189 static char *dup_wchar_to_utf8(wchar_t *w)
192 int l = WideCharToMultiByte(CP_UTF8, 0, w, -1, 0, 0, 0, 0);
195 WideCharToMultiByte(CP_UTF8, 0, w, -1, s, l, 0, 0);
199 static int shall_we_drop(AVFormatContext *s)
201 struct dshow_ctx *ctx = s->priv_data;
202 const uint8_t dropscore[] = {62, 75, 87, 100};
203 const int ndropscores = FF_ARRAY_ELEMS(dropscore);
204 unsigned int buffer_fullness = (ctx->curbufsize*100)/s->max_picture_buffer;
206 if(dropscore[++ctx->video_frame_num%ndropscores] <= buffer_fullness) {
207 av_log(s, AV_LOG_ERROR,
208 "real-time buffer %d%% full! frame dropped!\n", buffer_fullness);
216 callback(void *priv_data, int index, uint8_t *buf, int buf_size, int64_t time)
218 AVFormatContext *s = priv_data;
219 struct dshow_ctx *ctx = s->priv_data;
220 AVPacketList **ppktl, *pktl_next;
222 // dump_videohdr(s, vdhdr);
224 WaitForSingleObject(ctx->mutex, INFINITE);
229 pktl_next = av_mallocz(sizeof(AVPacketList));
233 if(av_new_packet(&pktl_next->pkt, buf_size) < 0) {
238 pktl_next->pkt.stream_index = index;
239 pktl_next->pkt.pts = time;
240 memcpy(pktl_next->pkt.data, buf, buf_size);
242 for(ppktl = &ctx->pktl ; *ppktl ; ppktl = &(*ppktl)->next);
245 ctx->curbufsize += buf_size;
247 SetEvent(ctx->event[1]);
248 ReleaseMutex(ctx->mutex);
252 ReleaseMutex(ctx->mutex);
257 * Cycle through available devices using the device enumerator devenum,
258 * retrieve the device with type specified by devtype and return the
259 * pointer to the object found in *pfilter.
260 * If pfilter is NULL, list all device names.
263 dshow_cycle_devices(AVFormatContext *avctx, ICreateDevEnum *devenum,
264 enum dshowDeviceType devtype, IBaseFilter **pfilter)
266 struct dshow_ctx *ctx = avctx->priv_data;
267 IBaseFilter *device_filter = NULL;
268 IEnumMoniker *classenum = NULL;
270 const char *device_name = ctx->device_name[devtype];
271 int skip = (devtype == VideoDevice) ? ctx->video_device_number
272 : ctx->audio_device_number;
275 const GUID *device_guid[2] = { &CLSID_VideoInputDeviceCategory,
276 &CLSID_AudioInputDeviceCategory };
277 const char *devtypename = (devtype == VideoDevice) ? "video" : "audio";
279 r = ICreateDevEnum_CreateClassEnumerator(devenum, device_guid[devtype],
280 (IEnumMoniker **) &classenum, 0);
282 av_log(avctx, AV_LOG_ERROR, "Could not enumerate %s devices.\n",
287 while (!device_filter && IEnumMoniker_Next(classenum, 1, &m, NULL) == S_OK) {
288 IPropertyBag *bag = NULL;
292 r = IMoniker_BindToStorage(m, 0, 0, &IID_IPropertyBag, (void *) &bag);
297 r = IPropertyBag_Read(bag, L"FriendlyName", &var, NULL);
301 buf = dup_wchar_to_utf8(var.bstrVal);
304 if (strcmp(device_name, buf))
308 IMoniker_BindToObject(m, 0, 0, &IID_IBaseFilter, (void *) &device_filter);
310 av_log(avctx, AV_LOG_INFO, " \"%s\"\n", buf);
317 IPropertyBag_Release(bag);
321 IEnumMoniker_Release(classenum);
324 if (!device_filter) {
325 av_log(avctx, AV_LOG_ERROR, "Could not find %s device.\n",
329 *pfilter = device_filter;
336 * Cycle through available formats using the specified pin,
337 * try to set parameters specified through AVOptions and if successful
338 * return 1 in *pformat_set.
339 * If pformat_set is NULL, list all pin capabilities.
342 dshow_cycle_formats(AVFormatContext *avctx, enum dshowDeviceType devtype,
343 IPin *pin, int *pformat_set)
345 struct dshow_ctx *ctx = avctx->priv_data;
346 IAMStreamConfig *config = NULL;
347 AM_MEDIA_TYPE *type = NULL;
352 if (IPin_QueryInterface(pin, &IID_IAMStreamConfig, (void **) &config) != S_OK)
354 if (IAMStreamConfig_GetNumberOfCapabilities(config, &n, &size) != S_OK)
357 caps = av_malloc(size);
361 for (i = 0; i < n && !format_set; i++) {
362 IAMStreamConfig_GetStreamCaps(config, i, &type, (void *) caps);
365 ff_print_AM_MEDIA_TYPE(type);
368 if (devtype == VideoDevice) {
369 VIDEO_STREAM_CONFIG_CAPS *vcaps = caps;
370 BITMAPINFOHEADER *bih;
373 ff_print_VIDEO_STREAM_CONFIG_CAPS(vcaps);
375 if (IsEqualGUID(&type->formattype, &FORMAT_VideoInfo)) {
376 VIDEOINFOHEADER *v = (void *) type->pbFormat;
377 fr = &v->AvgTimePerFrame;
379 } else if (IsEqualGUID(&type->formattype, &FORMAT_VideoInfo2)) {
380 VIDEOINFOHEADER2 *v = (void *) type->pbFormat;
381 fr = &v->AvgTimePerFrame;
387 enum AVPixelFormat pix_fmt = dshow_pixfmt(bih->biCompression, bih->biBitCount);
388 if (pix_fmt == AV_PIX_FMT_NONE) {
389 enum AVCodecID codec_id = dshow_codecid(bih->biCompression);
390 AVCodec *codec = avcodec_find_decoder(codec_id);
391 if (codec_id == AV_CODEC_ID_NONE || !codec) {
392 av_log(avctx, AV_LOG_INFO, " unknown compression type 0x%X", (int) bih->biCompression);
394 av_log(avctx, AV_LOG_INFO, " vcodec=%s", codec->name);
397 av_log(avctx, AV_LOG_INFO, " pixel_format=%s", av_get_pix_fmt_name(pix_fmt));
399 av_log(avctx, AV_LOG_INFO, " min s=%ldx%ld fps=%g max s=%ldx%ld fps=%g\n",
400 vcaps->MinOutputSize.cx, vcaps->MinOutputSize.cy,
401 1e7 / vcaps->MaxFrameInterval,
402 vcaps->MaxOutputSize.cx, vcaps->MaxOutputSize.cy,
403 1e7 / vcaps->MinFrameInterval);
406 if (ctx->video_codec_id != AV_CODEC_ID_RAWVIDEO) {
407 if (ctx->video_codec_id != dshow_codecid(bih->biCompression))
410 if (ctx->pixel_format != AV_PIX_FMT_NONE &&
411 ctx->pixel_format != dshow_pixfmt(bih->biCompression, bih->biBitCount)) {
414 if (ctx->framerate) {
415 int64_t framerate = ((int64_t) ctx->requested_framerate.den*10000000)
416 / ctx->requested_framerate.num;
417 if (framerate > vcaps->MaxFrameInterval ||
418 framerate < vcaps->MinFrameInterval)
422 if (ctx->requested_width && ctx->requested_height) {
423 if (ctx->requested_width > vcaps->MaxOutputSize.cx ||
424 ctx->requested_width < vcaps->MinOutputSize.cx ||
425 ctx->requested_height > vcaps->MaxOutputSize.cy ||
426 ctx->requested_height < vcaps->MinOutputSize.cy)
428 bih->biWidth = ctx->requested_width;
429 bih->biHeight = ctx->requested_height;
432 AUDIO_STREAM_CONFIG_CAPS *acaps = caps;
435 ff_print_AUDIO_STREAM_CONFIG_CAPS(acaps);
437 if (IsEqualGUID(&type->formattype, &FORMAT_WaveFormatEx)) {
438 fx = (void *) type->pbFormat;
443 av_log(avctx, AV_LOG_INFO, " min ch=%lu bits=%lu rate=%6lu max ch=%lu bits=%lu rate=%6lu\n",
444 acaps->MinimumChannels, acaps->MinimumBitsPerSample, acaps->MinimumSampleFrequency,
445 acaps->MaximumChannels, acaps->MaximumBitsPerSample, acaps->MaximumSampleFrequency);
448 if (ctx->sample_rate) {
449 if (ctx->sample_rate > acaps->MaximumSampleFrequency ||
450 ctx->sample_rate < acaps->MinimumSampleFrequency)
452 fx->nSamplesPerSec = ctx->sample_rate;
454 if (ctx->sample_size) {
455 if (ctx->sample_size > acaps->MaximumBitsPerSample ||
456 ctx->sample_size < acaps->MinimumBitsPerSample)
458 fx->wBitsPerSample = ctx->sample_size;
461 if (ctx->channels > acaps->MaximumChannels ||
462 ctx->channels < acaps->MinimumChannels)
464 fx->nChannels = ctx->channels;
467 if (IAMStreamConfig_SetFormat(config, type) != S_OK)
472 CoTaskMemFree(type->pbFormat);
476 IAMStreamConfig_Release(config);
480 *pformat_set = format_set;
484 * Set audio device buffer size in milliseconds (which can directly impact
485 * latency, depending on the device).
488 dshow_set_audio_buffer_size(AVFormatContext *avctx, IPin *pin)
490 struct dshow_ctx *ctx = avctx->priv_data;
491 IAMBufferNegotiation *buffer_negotiation = NULL;
492 ALLOCATOR_PROPERTIES props = { -1, -1, -1, -1 };
493 IAMStreamConfig *config = NULL;
494 AM_MEDIA_TYPE *type = NULL;
495 int ret = AVERROR(EIO);
497 if (IPin_QueryInterface(pin, &IID_IAMStreamConfig, (void **) &config) != S_OK)
499 if (IAMStreamConfig_GetFormat(config, &type) != S_OK)
501 if (!IsEqualGUID(&type->formattype, &FORMAT_WaveFormatEx))
504 props.cbBuffer = (((WAVEFORMATEX *) type->pbFormat)->nAvgBytesPerSec)
505 * ctx->audio_buffer_size / 1000;
507 if (IPin_QueryInterface(pin, &IID_IAMBufferNegotiation, (void **) &buffer_negotiation) != S_OK)
509 if (IAMBufferNegotiation_SuggestAllocatorProperties(buffer_negotiation, &props) != S_OK)
515 if (buffer_negotiation)
516 IAMBufferNegotiation_Release(buffer_negotiation);
519 CoTaskMemFree(type->pbFormat);
523 IAMStreamConfig_Release(config);
529 * Cycle through available pins using the device_filter device, of type
530 * devtype, retrieve the first output pin and return the pointer to the
531 * object found in *ppin.
532 * If ppin is NULL, cycle through all pins listing audio/video capabilities.
535 dshow_cycle_pins(AVFormatContext *avctx, enum dshowDeviceType devtype,
536 IBaseFilter *device_filter, IPin **ppin)
538 struct dshow_ctx *ctx = avctx->priv_data;
540 IPin *device_pin = NULL;
544 const GUID *mediatype[2] = { &MEDIATYPE_Video, &MEDIATYPE_Audio };
545 const char *devtypename = (devtype == VideoDevice) ? "video" : "audio";
547 int set_format = (devtype == VideoDevice && (ctx->framerate ||
548 (ctx->requested_width && ctx->requested_height) ||
549 ctx->pixel_format != AV_PIX_FMT_NONE ||
550 ctx->video_codec_id != AV_CODEC_ID_RAWVIDEO))
551 || (devtype == AudioDevice && (ctx->channels || ctx->sample_rate));
554 r = IBaseFilter_EnumPins(device_filter, &pins);
556 av_log(avctx, AV_LOG_ERROR, "Could not enumerate pins.\n");
561 av_log(avctx, AV_LOG_INFO, "DirectShow %s device options\n",
564 while (!device_pin && IEnumPins_Next(pins, 1, &pin, NULL) == S_OK) {
565 IKsPropertySet *p = NULL;
566 IEnumMediaTypes *types = NULL;
572 IPin_QueryPinInfo(pin, &info);
573 IBaseFilter_Release(info.pFilter);
575 if (info.dir != PINDIR_OUTPUT)
577 if (IPin_QueryInterface(pin, &IID_IKsPropertySet, (void **) &p) != S_OK)
579 if (IKsPropertySet_Get(p, &ROPSETID_Pin, AMPROPERTY_PIN_CATEGORY,
580 NULL, 0, &category, sizeof(GUID), &r2) != S_OK)
582 if (!IsEqualGUID(&category, &PIN_CATEGORY_CAPTURE))
586 char *buf = dup_wchar_to_utf8(info.achName);
587 av_log(avctx, AV_LOG_INFO, " Pin \"%s\"\n", buf);
589 dshow_cycle_formats(avctx, devtype, pin, NULL);
593 dshow_cycle_formats(avctx, devtype, pin, &format_set);
598 if (devtype == AudioDevice && ctx->audio_buffer_size) {
599 if (dshow_set_audio_buffer_size(avctx, pin) < 0)
603 if (IPin_EnumMediaTypes(pin, &types) != S_OK)
606 IEnumMediaTypes_Reset(types);
607 while (!device_pin && IEnumMediaTypes_Next(types, 1, &type, NULL) == S_OK) {
608 if (IsEqualGUID(&type->majortype, mediatype[devtype])) {
617 IEnumMediaTypes_Release(types);
619 IKsPropertySet_Release(p);
620 if (device_pin != pin)
624 IEnumPins_Release(pins);
627 if (set_format && !format_set) {
628 av_log(avctx, AV_LOG_ERROR, "Could not set %s options\n", devtypename);
632 av_log(avctx, AV_LOG_ERROR,
633 "Could not find output pin from %s capture device.\n", devtypename);
643 * List options for device with type devtype.
645 * @param devenum device enumerator used for accessing the device
648 dshow_list_device_options(AVFormatContext *avctx, ICreateDevEnum *devenum,
649 enum dshowDeviceType devtype)
651 struct dshow_ctx *ctx = avctx->priv_data;
652 IBaseFilter *device_filter = NULL;
655 if ((r = dshow_cycle_devices(avctx, devenum, devtype, &device_filter)) < 0)
657 ctx->device_filter[devtype] = device_filter;
658 if ((r = dshow_cycle_pins(avctx, devtype, device_filter, NULL)) < 0)
665 dshow_open_device(AVFormatContext *avctx, ICreateDevEnum *devenum,
666 enum dshowDeviceType devtype)
668 struct dshow_ctx *ctx = avctx->priv_data;
669 IBaseFilter *device_filter = NULL;
670 IGraphBuilder *graph = ctx->graph;
671 IPin *device_pin = NULL;
672 libAVPin *capture_pin = NULL;
673 libAVFilter *capture_filter = NULL;
674 int ret = AVERROR(EIO);
677 const wchar_t *filter_name[2] = { L"Audio capture filter", L"Video capture filter" };
679 if ((r = dshow_cycle_devices(avctx, devenum, devtype, &device_filter)) < 0) {
684 ctx->device_filter [devtype] = device_filter;
686 r = IGraphBuilder_AddFilter(graph, device_filter, NULL);
688 av_log(avctx, AV_LOG_ERROR, "Could not add device filter to graph.\n");
692 if ((r = dshow_cycle_pins(avctx, devtype, device_filter, &device_pin)) < 0) {
696 ctx->device_pin[devtype] = device_pin;
698 capture_filter = libAVFilter_Create(avctx, callback, devtype);
699 if (!capture_filter) {
700 av_log(avctx, AV_LOG_ERROR, "Could not create grabber filter.\n");
703 ctx->capture_filter[devtype] = capture_filter;
705 r = IGraphBuilder_AddFilter(graph, (IBaseFilter *) capture_filter,
706 filter_name[devtype]);
708 av_log(avctx, AV_LOG_ERROR, "Could not add capture filter to graph\n");
712 libAVPin_AddRef(capture_filter->pin);
713 capture_pin = capture_filter->pin;
714 ctx->capture_pin[devtype] = capture_pin;
716 r = IGraphBuilder_ConnectDirect(graph, device_pin, (IPin *) capture_pin, NULL);
718 av_log(avctx, AV_LOG_ERROR, "Could not connect pins\n");
728 static enum AVCodecID waveform_codec_id(enum AVSampleFormat sample_fmt)
730 switch (sample_fmt) {
731 case AV_SAMPLE_FMT_U8: return AV_CODEC_ID_PCM_U8;
732 case AV_SAMPLE_FMT_S16: return AV_CODEC_ID_PCM_S16LE;
733 case AV_SAMPLE_FMT_S32: return AV_CODEC_ID_PCM_S32LE;
734 default: return AV_CODEC_ID_NONE; /* Should never happen. */
738 static enum AVSampleFormat sample_fmt_bits_per_sample(int bits)
741 case 8: return AV_SAMPLE_FMT_U8;
742 case 16: return AV_SAMPLE_FMT_S16;
743 case 32: return AV_SAMPLE_FMT_S32;
744 default: return AV_SAMPLE_FMT_NONE; /* Should never happen. */
749 dshow_add_device(AVFormatContext *avctx,
750 enum dshowDeviceType devtype)
752 struct dshow_ctx *ctx = avctx->priv_data;
754 AVCodecContext *codec;
756 int ret = AVERROR(EIO);
758 st = avformat_new_stream(avctx, NULL);
760 ret = AVERROR(ENOMEM);
765 ctx->capture_filter[devtype]->stream_index = st->index;
767 libAVPin_ConnectionMediaType(ctx->capture_pin[devtype], &type);
770 if (devtype == VideoDevice) {
771 BITMAPINFOHEADER *bih = NULL;
772 AVRational time_base;
774 if (IsEqualGUID(&type.formattype, &FORMAT_VideoInfo)) {
775 VIDEOINFOHEADER *v = (void *) type.pbFormat;
776 time_base = (AVRational) { v->AvgTimePerFrame, 10000000 };
778 } else if (IsEqualGUID(&type.formattype, &FORMAT_VideoInfo2)) {
779 VIDEOINFOHEADER2 *v = (void *) type.pbFormat;
780 time_base = (AVRational) { v->AvgTimePerFrame, 10000000 };
784 av_log(avctx, AV_LOG_ERROR, "Could not get media type.\n");
788 codec->time_base = time_base;
789 codec->codec_type = AVMEDIA_TYPE_VIDEO;
790 codec->width = bih->biWidth;
791 codec->height = bih->biHeight;
792 codec->pix_fmt = dshow_pixfmt(bih->biCompression, bih->biBitCount);
793 if (codec->pix_fmt == AV_PIX_FMT_NONE) {
794 codec->codec_id = dshow_codecid(bih->biCompression);
795 if (codec->codec_id == AV_CODEC_ID_NONE) {
796 av_log(avctx, AV_LOG_ERROR, "Unknown compression type. "
797 "Please report verbose (-v 9) debug information.\n");
798 return AVERROR_PATCHWELCOME;
800 codec->bits_per_coded_sample = bih->biBitCount;
802 codec->codec_id = AV_CODEC_ID_RAWVIDEO;
803 if (bih->biCompression == BI_RGB || bih->biCompression == BI_BITFIELDS) {
804 codec->bits_per_coded_sample = bih->biBitCount;
805 codec->extradata = av_malloc(9 + FF_INPUT_BUFFER_PADDING_SIZE);
806 if (codec->extradata) {
807 codec->extradata_size = 9;
808 memcpy(codec->extradata, "BottomUp", 9);
813 WAVEFORMATEX *fx = NULL;
815 if (IsEqualGUID(&type.formattype, &FORMAT_WaveFormatEx)) {
816 fx = (void *) type.pbFormat;
819 av_log(avctx, AV_LOG_ERROR, "Could not get media type.\n");
823 codec->codec_type = AVMEDIA_TYPE_AUDIO;
824 codec->sample_fmt = sample_fmt_bits_per_sample(fx->wBitsPerSample);
825 codec->codec_id = waveform_codec_id(codec->sample_fmt);
826 codec->sample_rate = fx->nSamplesPerSec;
827 codec->channels = fx->nChannels;
830 avpriv_set_pts_info(st, 64, 1, 10000000);
838 static int parse_device_name(AVFormatContext *avctx)
840 struct dshow_ctx *ctx = avctx->priv_data;
841 char **device_name = ctx->device_name;
842 char *name = av_strdup(avctx->filename);
847 while ((type = strtok(tmp, "="))) {
848 char *token = strtok(NULL, ":");
851 if (!strcmp(type, "video")) {
852 device_name[0] = token;
853 } else if (!strcmp(type, "audio")) {
854 device_name[1] = token;
856 device_name[0] = NULL;
857 device_name[1] = NULL;
862 if (!device_name[0] && !device_name[1]) {
866 device_name[0] = av_strdup(device_name[0]);
868 device_name[1] = av_strdup(device_name[1]);
875 static int dshow_read_header(AVFormatContext *avctx)
877 struct dshow_ctx *ctx = avctx->priv_data;
878 IGraphBuilder *graph = NULL;
879 ICreateDevEnum *devenum = NULL;
880 IMediaControl *control = NULL;
881 IMediaEvent *media_event = NULL;
882 HANDLE media_event_handle;
884 int ret = AVERROR(EIO);
889 if (!ctx->list_devices && !parse_device_name(avctx)) {
890 av_log(avctx, AV_LOG_ERROR, "Malformed dshow input string.\n");
894 ctx->video_codec_id = avctx->video_codec_id ? avctx->video_codec_id
895 : AV_CODEC_ID_RAWVIDEO;
896 if (ctx->pixel_format != AV_PIX_FMT_NONE) {
897 if (ctx->video_codec_id != AV_CODEC_ID_RAWVIDEO) {
898 av_log(avctx, AV_LOG_ERROR, "Pixel format may only be set when "
899 "video codec is not set or set to rawvideo\n");
900 ret = AVERROR(EINVAL);
904 if (ctx->framerate) {
905 r = av_parse_video_rate(&ctx->requested_framerate, ctx->framerate);
907 av_log(avctx, AV_LOG_ERROR, "Could not parse framerate '%s'.\n", ctx->framerate);
912 r = CoCreateInstance(&CLSID_FilterGraph, NULL, CLSCTX_INPROC_SERVER,
913 &IID_IGraphBuilder, (void **) &graph);
915 av_log(avctx, AV_LOG_ERROR, "Could not create capture graph.\n");
920 r = CoCreateInstance(&CLSID_SystemDeviceEnum, NULL, CLSCTX_INPROC_SERVER,
921 &IID_ICreateDevEnum, (void **) &devenum);
923 av_log(avctx, AV_LOG_ERROR, "Could not enumerate system devices.\n");
927 if (ctx->list_devices) {
928 av_log(avctx, AV_LOG_INFO, "DirectShow video devices\n");
929 dshow_cycle_devices(avctx, devenum, VideoDevice, NULL);
930 av_log(avctx, AV_LOG_INFO, "DirectShow audio devices\n");
931 dshow_cycle_devices(avctx, devenum, AudioDevice, NULL);
935 if (ctx->list_options) {
936 if (ctx->device_name[VideoDevice])
937 dshow_list_device_options(avctx, devenum, VideoDevice);
938 if (ctx->device_name[AudioDevice])
939 dshow_list_device_options(avctx, devenum, AudioDevice);
944 if (ctx->device_name[VideoDevice]) {
945 if ((r = dshow_open_device(avctx, devenum, VideoDevice)) < 0 ||
946 (r = dshow_add_device(avctx, VideoDevice)) < 0) {
951 if (ctx->device_name[AudioDevice]) {
952 if ((r = dshow_open_device(avctx, devenum, AudioDevice)) < 0 ||
953 (r = dshow_add_device(avctx, AudioDevice)) < 0) {
959 ctx->mutex = CreateMutex(NULL, 0, NULL);
961 av_log(avctx, AV_LOG_ERROR, "Could not create Mutex\n");
964 ctx->event[1] = CreateEvent(NULL, 1, 0, NULL);
965 if (!ctx->event[1]) {
966 av_log(avctx, AV_LOG_ERROR, "Could not create Event\n");
970 r = IGraphBuilder_QueryInterface(graph, &IID_IMediaControl, (void **) &control);
972 av_log(avctx, AV_LOG_ERROR, "Could not get media control.\n");
975 ctx->control = control;
977 r = IGraphBuilder_QueryInterface(graph, &IID_IMediaEvent, (void **) &media_event);
979 av_log(avctx, AV_LOG_ERROR, "Could not get media event.\n");
982 ctx->media_event = media_event;
984 r = IMediaEvent_GetEventHandle(media_event, (void *) &media_event_handle);
986 av_log(avctx, AV_LOG_ERROR, "Could not get media event handle.\n");
989 proc = GetCurrentProcess();
990 r = DuplicateHandle(proc, media_event_handle, proc, &ctx->event[0],
991 0, 0, DUPLICATE_SAME_ACCESS);
993 av_log(avctx, AV_LOG_ERROR, "Could not duplicate media event handle.\n");
997 r = IMediaControl_Run(control);
1000 r = IMediaControl_GetState(control, 0, &pfs);
1003 av_log(avctx, AV_LOG_ERROR, "Could not run filter\n");
1012 ICreateDevEnum_Release(devenum);
1015 dshow_read_close(avctx);
1021 * Checks media events from DirectShow and returns -1 on error or EOF. Also
1022 * purges all events that might be in the event queue to stop the trigger
1023 * of event notification.
1025 static int dshow_check_event_queue(IMediaEvent *media_event)
1031 while (IMediaEvent_GetEvent(media_event, &code, &p1, &p2, 0) != E_ABORT) {
1032 if (code == EC_COMPLETE || code == EC_DEVICE_LOST || code == EC_ERRORABORT)
1034 IMediaEvent_FreeEventParams(media_event, code, p1, p2);
1040 static int dshow_read_packet(AVFormatContext *s, AVPacket *pkt)
1042 struct dshow_ctx *ctx = s->priv_data;
1043 AVPacketList *pktl = NULL;
1045 while (!ctx->eof && !pktl) {
1046 WaitForSingleObject(ctx->mutex, INFINITE);
1050 ctx->pktl = ctx->pktl->next;
1052 ctx->curbufsize -= pkt->size;
1054 ResetEvent(ctx->event[1]);
1055 ReleaseMutex(ctx->mutex);
1057 if (dshow_check_event_queue(ctx->media_event) < 0) {
1059 } else if (s->flags & AVFMT_FLAG_NONBLOCK) {
1060 return AVERROR(EAGAIN);
1062 WaitForMultipleObjects(2, ctx->event, 0, INFINITE);
1067 return ctx->eof ? AVERROR(EIO) : pkt->size;
1070 #define OFFSET(x) offsetof(struct dshow_ctx, x)
1071 #define DEC AV_OPT_FLAG_DECODING_PARAM
1072 static const AVOption options[] = {
1073 { "video_size", "set video size given a string such as 640x480 or hd720.", OFFSET(requested_width), AV_OPT_TYPE_IMAGE_SIZE, {.str = NULL}, 0, 0, DEC },
1074 { "pixel_format", "set video pixel format", OFFSET(pixel_format), AV_OPT_TYPE_PIXEL_FMT, {.i64 = AV_PIX_FMT_NONE}, -1, AV_PIX_FMT_NB-1, DEC },
1075 { "framerate", "set video frame rate", OFFSET(framerate), AV_OPT_TYPE_STRING, {.str = NULL}, 0, 0, DEC },
1076 { "sample_rate", "set audio sample rate", OFFSET(sample_rate), AV_OPT_TYPE_INT, {.i64 = 0}, 0, INT_MAX, DEC },
1077 { "sample_size", "set audio sample size", OFFSET(sample_size), AV_OPT_TYPE_INT, {.i64 = 0}, 0, 16, DEC },
1078 { "channels", "set number of audio channels, such as 1 or 2", OFFSET(channels), AV_OPT_TYPE_INT, {.i64 = 0}, 0, INT_MAX, DEC },
1079 { "list_devices", "list available devices", OFFSET(list_devices), AV_OPT_TYPE_INT, {.i64=0}, 0, 1, DEC, "list_devices" },
1080 { "true", "", 0, AV_OPT_TYPE_CONST, {.i64=1}, 0, 0, DEC, "list_devices" },
1081 { "false", "", 0, AV_OPT_TYPE_CONST, {.i64=0}, 0, 0, DEC, "list_devices" },
1082 { "list_options", "list available options for specified device", OFFSET(list_options), AV_OPT_TYPE_INT, {.i64=0}, 0, 1, DEC, "list_options" },
1083 { "true", "", 0, AV_OPT_TYPE_CONST, {.i64=1}, 0, 0, DEC, "list_options" },
1084 { "false", "", 0, AV_OPT_TYPE_CONST, {.i64=0}, 0, 0, DEC, "list_options" },
1085 { "video_device_number", "set video device number for devices with same name (starts at 0)", OFFSET(video_device_number), AV_OPT_TYPE_INT, {.i64 = 0}, 0, INT_MAX, DEC },
1086 { "audio_device_number", "set audio device number for devices with same name (starts at 0)", OFFSET(audio_device_number), AV_OPT_TYPE_INT, {.i64 = 0}, 0, INT_MAX, DEC },
1087 { "audio_buffer_size", "set audio device buffer latency size in milliseconds (default is the device's default)", OFFSET(audio_buffer_size), AV_OPT_TYPE_INT, {.i64 = 0}, 0, INT_MAX, DEC },
1091 static const AVClass dshow_class = {
1092 .class_name = "dshow indev",
1093 .item_name = av_default_item_name,
1095 .version = LIBAVUTIL_VERSION_INT,
1098 AVInputFormat ff_dshow_demuxer = {
1100 .long_name = NULL_IF_CONFIG_SMALL("DirectShow capture"),
1101 .priv_data_size = sizeof(struct dshow_ctx),
1102 .read_header = dshow_read_header,
1103 .read_packet = dshow_read_packet,
1104 .read_close = dshow_read_close,
1105 .flags = AVFMT_NOFILE,
1106 .priv_class = &dshow_class,