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"
28 #include "libavcodec/raw.h"
36 int video_device_number;
37 int audio_device_number;
41 int audio_buffer_size;
43 IBaseFilter *device_filter[2];
45 libAVFilter *capture_filter[2];
46 libAVPin *capture_pin[2];
49 HANDLE event[2]; /* event[0] is set by DirectShow
50 * event[1] is set by callback() */
56 unsigned int video_frame_num;
58 IMediaControl *control;
59 IMediaEvent *media_event;
61 enum AVPixelFormat pixel_format;
62 enum AVCodecID video_codec_id;
67 AVRational requested_framerate;
74 static enum AVPixelFormat dshow_pixfmt(DWORD biCompression, WORD biBitCount)
76 switch(biCompression) {
79 switch(biBitCount) { /* 1-8 are untested */
81 return AV_PIX_FMT_MONOWHITE;
83 return AV_PIX_FMT_RGB4;
85 return AV_PIX_FMT_RGB8;
87 return AV_PIX_FMT_RGB555;
89 return AV_PIX_FMT_BGR24;
91 return AV_PIX_FMT_RGB32;
94 return avpriv_find_pix_fmt(ff_raw_pix_fmt_tags, biCompression); // all others
97 static enum AVCodecID dshow_codecid(DWORD biCompression)
99 switch(biCompression) {
100 case MKTAG('d', 'v', 's', 'd'):
101 return AV_CODEC_ID_DVVIDEO;
102 case MKTAG('M', 'J', 'P', 'G'):
103 case MKTAG('m', 'j', 'p', 'g'):
104 return AV_CODEC_ID_MJPEG;
106 return AV_CODEC_ID_NONE;
110 dshow_read_close(AVFormatContext *s)
112 struct dshow_ctx *ctx = s->priv_data;
116 IMediaControl_Stop(ctx->control);
117 IMediaControl_Release(ctx->control);
120 if (ctx->media_event)
121 IMediaEvent_Release(ctx->media_event);
126 r = IGraphBuilder_EnumFilters(ctx->graph, &fenum);
129 IEnumFilters_Reset(fenum);
130 while (IEnumFilters_Next(fenum, 1, &f, NULL) == S_OK) {
131 if (IGraphBuilder_RemoveFilter(ctx->graph, f) == S_OK)
132 IEnumFilters_Reset(fenum); /* When a filter is removed,
133 * the list must be reset. */
134 IBaseFilter_Release(f);
136 IEnumFilters_Release(fenum);
138 IGraphBuilder_Release(ctx->graph);
141 if (ctx->capture_pin[VideoDevice])
142 libAVPin_Release(ctx->capture_pin[VideoDevice]);
143 if (ctx->capture_pin[AudioDevice])
144 libAVPin_Release(ctx->capture_pin[AudioDevice]);
145 if (ctx->capture_filter[VideoDevice])
146 libAVFilter_Release(ctx->capture_filter[VideoDevice]);
147 if (ctx->capture_filter[AudioDevice])
148 libAVFilter_Release(ctx->capture_filter[AudioDevice]);
150 if (ctx->device_pin[VideoDevice])
151 IPin_Release(ctx->device_pin[VideoDevice]);
152 if (ctx->device_pin[AudioDevice])
153 IPin_Release(ctx->device_pin[AudioDevice]);
154 if (ctx->device_filter[VideoDevice])
155 IBaseFilter_Release(ctx->device_filter[VideoDevice]);
156 if (ctx->device_filter[AudioDevice])
157 IBaseFilter_Release(ctx->device_filter[AudioDevice]);
159 if (ctx->device_name[0])
160 av_free(ctx->device_name[0]);
161 if (ctx->device_name[1])
162 av_free(ctx->device_name[1]);
165 CloseHandle(ctx->mutex);
167 CloseHandle(ctx->event[0]);
169 CloseHandle(ctx->event[1]);
173 AVPacketList *next = pktl->next;
174 av_destruct_packet(&pktl->pkt);
184 static char *dup_wchar_to_utf8(wchar_t *w)
187 int l = WideCharToMultiByte(CP_UTF8, 0, w, -1, 0, 0, 0, 0);
190 WideCharToMultiByte(CP_UTF8, 0, w, -1, s, l, 0, 0);
194 static int shall_we_drop(AVFormatContext *s)
196 struct dshow_ctx *ctx = s->priv_data;
197 const uint8_t dropscore[] = {62, 75, 87, 100};
198 const int ndropscores = FF_ARRAY_ELEMS(dropscore);
199 unsigned int buffer_fullness = (ctx->curbufsize*100)/s->max_picture_buffer;
201 if(dropscore[++ctx->video_frame_num%ndropscores] <= buffer_fullness) {
202 av_log(s, AV_LOG_ERROR,
203 "real-time buffer %d%% full! frame dropped!\n", buffer_fullness);
211 callback(void *priv_data, int index, uint8_t *buf, int buf_size, int64_t time)
213 AVFormatContext *s = priv_data;
214 struct dshow_ctx *ctx = s->priv_data;
215 AVPacketList **ppktl, *pktl_next;
217 // dump_videohdr(s, vdhdr);
219 WaitForSingleObject(ctx->mutex, INFINITE);
224 pktl_next = av_mallocz(sizeof(AVPacketList));
228 if(av_new_packet(&pktl_next->pkt, buf_size) < 0) {
233 pktl_next->pkt.stream_index = index;
234 pktl_next->pkt.pts = time;
235 memcpy(pktl_next->pkt.data, buf, buf_size);
237 for(ppktl = &ctx->pktl ; *ppktl ; ppktl = &(*ppktl)->next);
240 ctx->curbufsize += buf_size;
242 SetEvent(ctx->event[1]);
243 ReleaseMutex(ctx->mutex);
247 ReleaseMutex(ctx->mutex);
252 * Cycle through available devices using the device enumerator devenum,
253 * retrieve the device with type specified by devtype and return the
254 * pointer to the object found in *pfilter.
255 * If pfilter is NULL, list all device names.
258 dshow_cycle_devices(AVFormatContext *avctx, ICreateDevEnum *devenum,
259 enum dshowDeviceType devtype, IBaseFilter **pfilter)
261 struct dshow_ctx *ctx = avctx->priv_data;
262 IBaseFilter *device_filter = NULL;
263 IEnumMoniker *classenum = NULL;
265 const char *device_name = ctx->device_name[devtype];
266 int skip = (devtype == VideoDevice) ? ctx->video_device_number
267 : ctx->audio_device_number;
270 const GUID *device_guid[2] = { &CLSID_VideoInputDeviceCategory,
271 &CLSID_AudioInputDeviceCategory };
272 const char *devtypename = (devtype == VideoDevice) ? "video" : "audio";
274 r = ICreateDevEnum_CreateClassEnumerator(devenum, device_guid[devtype],
275 (IEnumMoniker **) &classenum, 0);
277 av_log(avctx, AV_LOG_ERROR, "Could not enumerate %s devices.\n",
282 while (!device_filter && IEnumMoniker_Next(classenum, 1, &m, NULL) == S_OK) {
283 IPropertyBag *bag = NULL;
287 r = IMoniker_BindToStorage(m, 0, 0, &IID_IPropertyBag, (void *) &bag);
292 r = IPropertyBag_Read(bag, L"FriendlyName", &var, NULL);
296 buf = dup_wchar_to_utf8(var.bstrVal);
299 if (strcmp(device_name, buf))
303 IMoniker_BindToObject(m, 0, 0, &IID_IBaseFilter, (void *) &device_filter);
305 av_log(avctx, AV_LOG_INFO, " \"%s\"\n", buf);
312 IPropertyBag_Release(bag);
316 IEnumMoniker_Release(classenum);
319 if (!device_filter) {
320 av_log(avctx, AV_LOG_ERROR, "Could not find %s device.\n",
324 *pfilter = device_filter;
331 * Cycle through available formats using the specified pin,
332 * try to set parameters specified through AVOptions and if successful
333 * return 1 in *pformat_set.
334 * If pformat_set is NULL, list all pin capabilities.
337 dshow_cycle_formats(AVFormatContext *avctx, enum dshowDeviceType devtype,
338 IPin *pin, int *pformat_set)
340 struct dshow_ctx *ctx = avctx->priv_data;
341 IAMStreamConfig *config = NULL;
342 AM_MEDIA_TYPE *type = NULL;
347 if (IPin_QueryInterface(pin, &IID_IAMStreamConfig, (void **) &config) != S_OK)
349 if (IAMStreamConfig_GetNumberOfCapabilities(config, &n, &size) != S_OK)
352 caps = av_malloc(size);
356 for (i = 0; i < n && !format_set; i++) {
357 IAMStreamConfig_GetStreamCaps(config, i, &type, (void *) caps);
360 ff_print_AM_MEDIA_TYPE(type);
363 if (devtype == VideoDevice) {
364 VIDEO_STREAM_CONFIG_CAPS *vcaps = caps;
365 BITMAPINFOHEADER *bih;
368 ff_print_VIDEO_STREAM_CONFIG_CAPS(vcaps);
370 if (IsEqualGUID(&type->formattype, &FORMAT_VideoInfo)) {
371 VIDEOINFOHEADER *v = (void *) type->pbFormat;
372 fr = &v->AvgTimePerFrame;
374 } else if (IsEqualGUID(&type->formattype, &FORMAT_VideoInfo2)) {
375 VIDEOINFOHEADER2 *v = (void *) type->pbFormat;
376 fr = &v->AvgTimePerFrame;
382 enum AVPixelFormat pix_fmt = dshow_pixfmt(bih->biCompression, bih->biBitCount);
383 if (pix_fmt == AV_PIX_FMT_NONE) {
384 enum AVCodecID codec_id = dshow_codecid(bih->biCompression);
385 AVCodec *codec = avcodec_find_decoder(codec_id);
386 if (codec_id == AV_CODEC_ID_NONE || !codec) {
387 av_log(avctx, AV_LOG_INFO, " unknown compression type 0x%X", (int) bih->biCompression);
389 av_log(avctx, AV_LOG_INFO, " vcodec=%s", codec->name);
392 av_log(avctx, AV_LOG_INFO, " pixel_format=%s", av_get_pix_fmt_name(pix_fmt));
394 av_log(avctx, AV_LOG_INFO, " min s=%ldx%ld fps=%g max s=%ldx%ld fps=%g\n",
395 vcaps->MinOutputSize.cx, vcaps->MinOutputSize.cy,
396 1e7 / vcaps->MaxFrameInterval,
397 vcaps->MaxOutputSize.cx, vcaps->MaxOutputSize.cy,
398 1e7 / vcaps->MinFrameInterval);
401 if (ctx->video_codec_id != AV_CODEC_ID_RAWVIDEO) {
402 if (ctx->video_codec_id != dshow_codecid(bih->biCompression))
405 if (ctx->pixel_format != AV_PIX_FMT_NONE &&
406 ctx->pixel_format != dshow_pixfmt(bih->biCompression, bih->biBitCount)) {
409 if (ctx->framerate) {
410 int64_t framerate = ((int64_t) ctx->requested_framerate.den*10000000)
411 / ctx->requested_framerate.num;
412 if (framerate > vcaps->MaxFrameInterval ||
413 framerate < vcaps->MinFrameInterval)
417 if (ctx->requested_width && ctx->requested_height) {
418 if (ctx->requested_width > vcaps->MaxOutputSize.cx ||
419 ctx->requested_width < vcaps->MinOutputSize.cx ||
420 ctx->requested_height > vcaps->MaxOutputSize.cy ||
421 ctx->requested_height < vcaps->MinOutputSize.cy)
423 bih->biWidth = ctx->requested_width;
424 bih->biHeight = ctx->requested_height;
427 AUDIO_STREAM_CONFIG_CAPS *acaps = caps;
430 ff_print_AUDIO_STREAM_CONFIG_CAPS(acaps);
432 if (IsEqualGUID(&type->formattype, &FORMAT_WaveFormatEx)) {
433 fx = (void *) type->pbFormat;
438 av_log(avctx, AV_LOG_INFO, " min ch=%lu bits=%lu rate=%6lu max ch=%lu bits=%lu rate=%6lu\n",
439 acaps->MinimumChannels, acaps->MinimumBitsPerSample, acaps->MinimumSampleFrequency,
440 acaps->MaximumChannels, acaps->MaximumBitsPerSample, acaps->MaximumSampleFrequency);
443 if (ctx->sample_rate) {
444 if (ctx->sample_rate > acaps->MaximumSampleFrequency ||
445 ctx->sample_rate < acaps->MinimumSampleFrequency)
447 fx->nSamplesPerSec = ctx->sample_rate;
449 if (ctx->sample_size) {
450 if (ctx->sample_size > acaps->MaximumBitsPerSample ||
451 ctx->sample_size < acaps->MinimumBitsPerSample)
453 fx->wBitsPerSample = ctx->sample_size;
456 if (ctx->channels > acaps->MaximumChannels ||
457 ctx->channels < acaps->MinimumChannels)
459 fx->nChannels = ctx->channels;
462 if (IAMStreamConfig_SetFormat(config, type) != S_OK)
467 CoTaskMemFree(type->pbFormat);
471 IAMStreamConfig_Release(config);
475 *pformat_set = format_set;
479 * Set audio device buffer size in milliseconds (which can directly impact
480 * latency, depending on the device).
483 dshow_set_audio_buffer_size(AVFormatContext *avctx, IPin *pin)
485 struct dshow_ctx *ctx = avctx->priv_data;
486 IAMBufferNegotiation *buffer_negotiation = NULL;
487 ALLOCATOR_PROPERTIES props = { -1, -1, -1, -1 };
488 IAMStreamConfig *config = NULL;
489 AM_MEDIA_TYPE *type = NULL;
490 int ret = AVERROR(EIO);
492 if (IPin_QueryInterface(pin, &IID_IAMStreamConfig, (void **) &config) != S_OK)
494 if (IAMStreamConfig_GetFormat(config, &type) != S_OK)
496 if (!IsEqualGUID(&type->formattype, &FORMAT_WaveFormatEx))
499 props.cbBuffer = (((WAVEFORMATEX *) type->pbFormat)->nAvgBytesPerSec)
500 * ctx->audio_buffer_size / 1000;
502 if (IPin_QueryInterface(pin, &IID_IAMBufferNegotiation, (void **) &buffer_negotiation) != S_OK)
504 if (IAMBufferNegotiation_SuggestAllocatorProperties(buffer_negotiation, &props) != S_OK)
510 if (buffer_negotiation)
511 IAMBufferNegotiation_Release(buffer_negotiation);
514 CoTaskMemFree(type->pbFormat);
518 IAMStreamConfig_Release(config);
524 * Cycle through available pins using the device_filter device, of type
525 * devtype, retrieve the first output pin and return the pointer to the
526 * object found in *ppin.
527 * If ppin is NULL, cycle through all pins listing audio/video capabilities.
530 dshow_cycle_pins(AVFormatContext *avctx, enum dshowDeviceType devtype,
531 IBaseFilter *device_filter, IPin **ppin)
533 struct dshow_ctx *ctx = avctx->priv_data;
535 IPin *device_pin = NULL;
539 const GUID *mediatype[2] = { &MEDIATYPE_Video, &MEDIATYPE_Audio };
540 const char *devtypename = (devtype == VideoDevice) ? "video" : "audio";
542 int set_format = (devtype == VideoDevice && (ctx->framerate ||
543 (ctx->requested_width && ctx->requested_height) ||
544 ctx->pixel_format != AV_PIX_FMT_NONE ||
545 ctx->video_codec_id != AV_CODEC_ID_RAWVIDEO))
546 || (devtype == AudioDevice && (ctx->channels || ctx->sample_rate));
549 r = IBaseFilter_EnumPins(device_filter, &pins);
551 av_log(avctx, AV_LOG_ERROR, "Could not enumerate pins.\n");
556 av_log(avctx, AV_LOG_INFO, "DirectShow %s device options\n",
559 while (!device_pin && IEnumPins_Next(pins, 1, &pin, NULL) == S_OK) {
560 IKsPropertySet *p = NULL;
561 IEnumMediaTypes *types = NULL;
567 IPin_QueryPinInfo(pin, &info);
568 IBaseFilter_Release(info.pFilter);
570 if (info.dir != PINDIR_OUTPUT)
572 if (IPin_QueryInterface(pin, &IID_IKsPropertySet, (void **) &p) != S_OK)
574 if (IKsPropertySet_Get(p, &ROPSETID_Pin, AMPROPERTY_PIN_CATEGORY,
575 NULL, 0, &category, sizeof(GUID), &r2) != S_OK)
577 if (!IsEqualGUID(&category, &PIN_CATEGORY_CAPTURE))
581 char *buf = dup_wchar_to_utf8(info.achName);
582 av_log(avctx, AV_LOG_INFO, " Pin \"%s\"\n", buf);
584 dshow_cycle_formats(avctx, devtype, pin, NULL);
588 dshow_cycle_formats(avctx, devtype, pin, &format_set);
593 if (devtype == AudioDevice && ctx->audio_buffer_size) {
594 if (dshow_set_audio_buffer_size(avctx, pin) < 0)
598 if (IPin_EnumMediaTypes(pin, &types) != S_OK)
601 IEnumMediaTypes_Reset(types);
602 while (!device_pin && IEnumMediaTypes_Next(types, 1, &type, NULL) == S_OK) {
603 if (IsEqualGUID(&type->majortype, mediatype[devtype])) {
612 IEnumMediaTypes_Release(types);
614 IKsPropertySet_Release(p);
615 if (device_pin != pin)
619 IEnumPins_Release(pins);
622 if (set_format && !format_set) {
623 av_log(avctx, AV_LOG_ERROR, "Could not set %s options\n", devtypename);
627 av_log(avctx, AV_LOG_ERROR,
628 "Could not find output pin from %s capture device.\n", devtypename);
638 * List options for device with type devtype.
640 * @param devenum device enumerator used for accessing the device
643 dshow_list_device_options(AVFormatContext *avctx, ICreateDevEnum *devenum,
644 enum dshowDeviceType devtype)
646 struct dshow_ctx *ctx = avctx->priv_data;
647 IBaseFilter *device_filter = NULL;
650 if ((r = dshow_cycle_devices(avctx, devenum, devtype, &device_filter)) < 0)
652 ctx->device_filter[devtype] = device_filter;
653 if ((r = dshow_cycle_pins(avctx, devtype, device_filter, NULL)) < 0)
660 dshow_open_device(AVFormatContext *avctx, ICreateDevEnum *devenum,
661 enum dshowDeviceType devtype)
663 struct dshow_ctx *ctx = avctx->priv_data;
664 IBaseFilter *device_filter = NULL;
665 IGraphBuilder *graph = ctx->graph;
666 IPin *device_pin = NULL;
667 libAVPin *capture_pin = NULL;
668 libAVFilter *capture_filter = NULL;
669 int ret = AVERROR(EIO);
672 const wchar_t *filter_name[2] = { L"Audio capture filter", L"Video capture filter" };
674 if ((r = dshow_cycle_devices(avctx, devenum, devtype, &device_filter)) < 0) {
679 ctx->device_filter [devtype] = device_filter;
681 r = IGraphBuilder_AddFilter(graph, device_filter, NULL);
683 av_log(avctx, AV_LOG_ERROR, "Could not add device filter to graph.\n");
687 if ((r = dshow_cycle_pins(avctx, devtype, device_filter, &device_pin)) < 0) {
691 ctx->device_pin[devtype] = device_pin;
693 capture_filter = libAVFilter_Create(avctx, callback, devtype);
694 if (!capture_filter) {
695 av_log(avctx, AV_LOG_ERROR, "Could not create grabber filter.\n");
698 ctx->capture_filter[devtype] = capture_filter;
700 r = IGraphBuilder_AddFilter(graph, (IBaseFilter *) capture_filter,
701 filter_name[devtype]);
703 av_log(avctx, AV_LOG_ERROR, "Could not add capture filter to graph\n");
707 libAVPin_AddRef(capture_filter->pin);
708 capture_pin = capture_filter->pin;
709 ctx->capture_pin[devtype] = capture_pin;
711 r = IGraphBuilder_ConnectDirect(graph, device_pin, (IPin *) capture_pin, NULL);
713 av_log(avctx, AV_LOG_ERROR, "Could not connect pins\n");
723 static enum AVCodecID waveform_codec_id(enum AVSampleFormat sample_fmt)
725 switch (sample_fmt) {
726 case AV_SAMPLE_FMT_U8: return AV_CODEC_ID_PCM_U8;
727 case AV_SAMPLE_FMT_S16: return AV_CODEC_ID_PCM_S16LE;
728 case AV_SAMPLE_FMT_S32: return AV_CODEC_ID_PCM_S32LE;
729 default: return AV_CODEC_ID_NONE; /* Should never happen. */
733 static enum AVSampleFormat sample_fmt_bits_per_sample(int bits)
736 case 8: return AV_SAMPLE_FMT_U8;
737 case 16: return AV_SAMPLE_FMT_S16;
738 case 32: return AV_SAMPLE_FMT_S32;
739 default: return AV_SAMPLE_FMT_NONE; /* Should never happen. */
744 dshow_add_device(AVFormatContext *avctx,
745 enum dshowDeviceType devtype)
747 struct dshow_ctx *ctx = avctx->priv_data;
749 AVCodecContext *codec;
751 int ret = AVERROR(EIO);
753 st = avformat_new_stream(avctx, NULL);
755 ret = AVERROR(ENOMEM);
760 ctx->capture_filter[devtype]->stream_index = st->index;
762 libAVPin_ConnectionMediaType(ctx->capture_pin[devtype], &type);
765 if (devtype == VideoDevice) {
766 BITMAPINFOHEADER *bih = NULL;
767 AVRational time_base;
769 if (IsEqualGUID(&type.formattype, &FORMAT_VideoInfo)) {
770 VIDEOINFOHEADER *v = (void *) type.pbFormat;
771 time_base = (AVRational) { v->AvgTimePerFrame, 10000000 };
773 } else if (IsEqualGUID(&type.formattype, &FORMAT_VideoInfo2)) {
774 VIDEOINFOHEADER2 *v = (void *) type.pbFormat;
775 time_base = (AVRational) { v->AvgTimePerFrame, 10000000 };
779 av_log(avctx, AV_LOG_ERROR, "Could not get media type.\n");
783 codec->time_base = time_base;
784 codec->codec_type = AVMEDIA_TYPE_VIDEO;
785 codec->width = bih->biWidth;
786 codec->height = bih->biHeight;
787 codec->pix_fmt = dshow_pixfmt(bih->biCompression, bih->biBitCount);
788 if(bih->biCompression == MKTAG('H', 'D', 'Y', 'C')) {
789 av_log(avctx, AV_LOG_ERROR, "attempt use full range for HDYC...");
790 codec->color_range = AVCOL_RANGE_MPEG;
792 if (codec->pix_fmt == AV_PIX_FMT_NONE) {
793 codec->codec_id = dshow_codecid(bih->biCompression);
794 if (codec->codec_id == AV_CODEC_ID_NONE) {
795 av_log(avctx, AV_LOG_ERROR, "Unknown compression type. "
796 "Please report verbose (-v 9) debug information.\n");
797 return AVERROR_PATCHWELCOME;
799 codec->bits_per_coded_sample = bih->biBitCount;
801 codec->codec_id = AV_CODEC_ID_RAWVIDEO;
802 if (bih->biCompression == BI_RGB || bih->biCompression == BI_BITFIELDS) {
803 codec->bits_per_coded_sample = bih->biBitCount;
804 codec->extradata = av_malloc(9 + FF_INPUT_BUFFER_PADDING_SIZE);
805 if (codec->extradata) {
806 codec->extradata_size = 9;
807 memcpy(codec->extradata, "BottomUp", 9);
812 WAVEFORMATEX *fx = NULL;
814 if (IsEqualGUID(&type.formattype, &FORMAT_WaveFormatEx)) {
815 fx = (void *) type.pbFormat;
818 av_log(avctx, AV_LOG_ERROR, "Could not get media type.\n");
822 codec->codec_type = AVMEDIA_TYPE_AUDIO;
823 codec->sample_fmt = sample_fmt_bits_per_sample(fx->wBitsPerSample);
824 codec->codec_id = waveform_codec_id(codec->sample_fmt);
825 codec->sample_rate = fx->nSamplesPerSec;
826 codec->channels = fx->nChannels;
829 avpriv_set_pts_info(st, 64, 1, 10000000);
837 static int parse_device_name(AVFormatContext *avctx)
839 struct dshow_ctx *ctx = avctx->priv_data;
840 char **device_name = ctx->device_name;
841 char *name = av_strdup(avctx->filename);
846 while ((type = strtok(tmp, "="))) {
847 char *token = strtok(NULL, ":");
850 if (!strcmp(type, "video")) {
851 device_name[0] = token;
852 } else if (!strcmp(type, "audio")) {
853 device_name[1] = token;
855 device_name[0] = NULL;
856 device_name[1] = NULL;
861 if (!device_name[0] && !device_name[1]) {
865 device_name[0] = av_strdup(device_name[0]);
867 device_name[1] = av_strdup(device_name[1]);
874 static int dshow_read_header(AVFormatContext *avctx)
876 struct dshow_ctx *ctx = avctx->priv_data;
877 IGraphBuilder *graph = NULL;
878 ICreateDevEnum *devenum = NULL;
879 IMediaControl *control = NULL;
880 IMediaEvent *media_event = NULL;
881 HANDLE media_event_handle;
883 int ret = AVERROR(EIO);
888 if (!ctx->list_devices && !parse_device_name(avctx)) {
889 av_log(avctx, AV_LOG_ERROR, "Malformed dshow input string.\n");
893 ctx->video_codec_id = avctx->video_codec_id ? avctx->video_codec_id
894 : AV_CODEC_ID_RAWVIDEO;
895 if (ctx->pixel_format != AV_PIX_FMT_NONE) {
896 if (ctx->video_codec_id != AV_CODEC_ID_RAWVIDEO) {
897 av_log(avctx, AV_LOG_ERROR, "Pixel format may only be set when "
898 "video codec is not set or set to rawvideo\n");
899 ret = AVERROR(EINVAL);
903 if (ctx->framerate) {
904 r = av_parse_video_rate(&ctx->requested_framerate, ctx->framerate);
906 av_log(avctx, AV_LOG_ERROR, "Could not parse framerate '%s'.\n", ctx->framerate);
911 r = CoCreateInstance(&CLSID_FilterGraph, NULL, CLSCTX_INPROC_SERVER,
912 &IID_IGraphBuilder, (void **) &graph);
914 av_log(avctx, AV_LOG_ERROR, "Could not create capture graph.\n");
919 r = CoCreateInstance(&CLSID_SystemDeviceEnum, NULL, CLSCTX_INPROC_SERVER,
920 &IID_ICreateDevEnum, (void **) &devenum);
922 av_log(avctx, AV_LOG_ERROR, "Could not enumerate system devices.\n");
926 if (ctx->list_devices) {
927 av_log(avctx, AV_LOG_INFO, "DirectShow video devices\n");
928 dshow_cycle_devices(avctx, devenum, VideoDevice, NULL);
929 av_log(avctx, AV_LOG_INFO, "DirectShow audio devices\n");
930 dshow_cycle_devices(avctx, devenum, AudioDevice, NULL);
934 if (ctx->list_options) {
935 if (ctx->device_name[VideoDevice])
936 dshow_list_device_options(avctx, devenum, VideoDevice);
937 if (ctx->device_name[AudioDevice])
938 dshow_list_device_options(avctx, devenum, AudioDevice);
943 if (ctx->device_name[VideoDevice]) {
944 if ((r = dshow_open_device(avctx, devenum, VideoDevice)) < 0 ||
945 (r = dshow_add_device(avctx, VideoDevice)) < 0) {
950 if (ctx->device_name[AudioDevice]) {
951 if ((r = dshow_open_device(avctx, devenum, AudioDevice)) < 0 ||
952 (r = dshow_add_device(avctx, AudioDevice)) < 0) {
958 ctx->mutex = CreateMutex(NULL, 0, NULL);
960 av_log(avctx, AV_LOG_ERROR, "Could not create Mutex\n");
963 ctx->event[1] = CreateEvent(NULL, 1, 0, NULL);
964 if (!ctx->event[1]) {
965 av_log(avctx, AV_LOG_ERROR, "Could not create Event\n");
969 r = IGraphBuilder_QueryInterface(graph, &IID_IMediaControl, (void **) &control);
971 av_log(avctx, AV_LOG_ERROR, "Could not get media control.\n");
974 ctx->control = control;
976 r = IGraphBuilder_QueryInterface(graph, &IID_IMediaEvent, (void **) &media_event);
978 av_log(avctx, AV_LOG_ERROR, "Could not get media event.\n");
981 ctx->media_event = media_event;
983 r = IMediaEvent_GetEventHandle(media_event, (void *) &media_event_handle);
985 av_log(avctx, AV_LOG_ERROR, "Could not get media event handle.\n");
988 proc = GetCurrentProcess();
989 r = DuplicateHandle(proc, media_event_handle, proc, &ctx->event[0],
990 0, 0, DUPLICATE_SAME_ACCESS);
992 av_log(avctx, AV_LOG_ERROR, "Could not duplicate media event handle.\n");
996 r = IMediaControl_Run(control);
999 r = IMediaControl_GetState(control, 0, &pfs);
1002 av_log(avctx, AV_LOG_ERROR, "Could not run filter\n");
1011 ICreateDevEnum_Release(devenum);
1014 dshow_read_close(avctx);
1020 * Checks media events from DirectShow and returns -1 on error or EOF. Also
1021 * purges all events that might be in the event queue to stop the trigger
1022 * of event notification.
1024 static int dshow_check_event_queue(IMediaEvent *media_event)
1030 while (IMediaEvent_GetEvent(media_event, &code, &p1, &p2, 0) != E_ABORT) {
1031 if (code == EC_COMPLETE || code == EC_DEVICE_LOST || code == EC_ERRORABORT)
1033 IMediaEvent_FreeEventParams(media_event, code, p1, p2);
1039 static int dshow_read_packet(AVFormatContext *s, AVPacket *pkt)
1041 struct dshow_ctx *ctx = s->priv_data;
1042 AVPacketList *pktl = NULL;
1044 while (!ctx->eof && !pktl) {
1045 WaitForSingleObject(ctx->mutex, INFINITE);
1049 ctx->pktl = ctx->pktl->next;
1051 ctx->curbufsize -= pkt->size;
1053 ResetEvent(ctx->event[1]);
1054 ReleaseMutex(ctx->mutex);
1056 if (dshow_check_event_queue(ctx->media_event) < 0) {
1058 } else if (s->flags & AVFMT_FLAG_NONBLOCK) {
1059 return AVERROR(EAGAIN);
1061 WaitForMultipleObjects(2, ctx->event, 0, INFINITE);
1066 return ctx->eof ? AVERROR(EIO) : pkt->size;
1069 #define OFFSET(x) offsetof(struct dshow_ctx, x)
1070 #define DEC AV_OPT_FLAG_DECODING_PARAM
1071 static const AVOption options[] = {
1072 { "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 },
1073 { "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 },
1074 { "framerate", "set video frame rate", OFFSET(framerate), AV_OPT_TYPE_STRING, {.str = NULL}, 0, 0, DEC },
1075 { "sample_rate", "set audio sample rate", OFFSET(sample_rate), AV_OPT_TYPE_INT, {.i64 = 0}, 0, INT_MAX, DEC },
1076 { "sample_size", "set audio sample size", OFFSET(sample_size), AV_OPT_TYPE_INT, {.i64 = 0}, 0, 16, DEC },
1077 { "channels", "set number of audio channels, such as 1 or 2", OFFSET(channels), AV_OPT_TYPE_INT, {.i64 = 0}, 0, INT_MAX, DEC },
1078 { "list_devices", "list available devices", OFFSET(list_devices), AV_OPT_TYPE_INT, {.i64=0}, 0, 1, DEC, "list_devices" },
1079 { "true", "", 0, AV_OPT_TYPE_CONST, {.i64=1}, 0, 0, DEC, "list_devices" },
1080 { "false", "", 0, AV_OPT_TYPE_CONST, {.i64=0}, 0, 0, DEC, "list_devices" },
1081 { "list_options", "list available options for specified device", OFFSET(list_options), AV_OPT_TYPE_INT, {.i64=0}, 0, 1, DEC, "list_options" },
1082 { "true", "", 0, AV_OPT_TYPE_CONST, {.i64=1}, 0, 0, DEC, "list_options" },
1083 { "false", "", 0, AV_OPT_TYPE_CONST, {.i64=0}, 0, 0, DEC, "list_options" },
1084 { "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 },
1085 { "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 },
1086 { "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 },
1090 static const AVClass dshow_class = {
1091 .class_name = "dshow indev",
1092 .item_name = av_default_item_name,
1094 .version = LIBAVUTIL_VERSION_INT,
1097 AVInputFormat ff_dshow_demuxer = {
1099 .long_name = NULL_IF_CONFIG_SMALL("DirectShow capture"),
1100 .priv_data_size = sizeof(struct dshow_ctx),
1101 .read_header = dshow_read_header,
1102 .read_packet = dshow_read_packet,
1103 .read_close = dshow_read_close,
1104 .flags = AVFMT_NOFILE,
1105 .priv_class = &dshow_class,