]> git.sesse.net Git - casparcg/blobdiff - modules/ffmpeg/producer/util/util.cpp
[ffmpeg] Copied flush logic when seeking from 2.0, as well as current frame in clip...
[casparcg] / modules / ffmpeg / producer / util / util.cpp
index 865b96bcb3f19c950c6d2f5c3063b945d587d502..eb773c6c115f5818572bdee16ddd1072e409a4ba 100644 (file)
@@ -40,6 +40,7 @@
 #include <common/except.h>
 #include <common/array.h>
 #include <common/os/filesystem.h>
+#include <common/memcpy.h>
 
 #include <tbb/parallel_for.h>
 
@@ -148,7 +149,7 @@ core::pixel_format_desc pixel_format_desc(PixelFormat pix_fmt, int width, int he
        }
 }
 
-core::mutable_frame make_frame(const void* tag, const spl::shared_ptr<AVFrame>& decoded_frame, double fps, core::frame_factory& frame_factory, const core::audio_channel_layout& channel_layout)
+core::mutable_frame make_frame(const void* tag, const spl::shared_ptr<AVFrame>& decoded_frame, core::frame_factory& frame_factory, const core::audio_channel_layout& channel_layout)
 {                      
        static tbb::concurrent_unordered_map<int64_t, tbb::concurrent_queue<std::shared_ptr<SwsContext>>> sws_contvalid_exts_;
        
@@ -204,8 +205,7 @@ core::mutable_frame make_frame(const void* tag, const spl::shared_ptr<AVFrame>&
                                                                        boost::errinfo_api_function("sws_getContext"));
                }       
                
-               spl::shared_ptr<AVFrame> av_frame(avcodec_alloc_frame(), av_free);      
-               avcodec_get_frame_defaults(av_frame.get());                     
+               auto av_frame = create_frame();
                if(target_pix_fmt == PIX_FMT_BGRA)
                {
                        auto size = avpicture_fill(reinterpret_cast<AVPicture*>(av_frame.get()), write.image_data(0).begin(), PIX_FMT_BGRA, width, height);
@@ -241,13 +241,20 @@ core::mutable_frame make_frame(const void* tag, const spl::shared_ptr<AVFrame>&
                        CASPAR_ASSERT(decoded);
                        CASPAR_ASSERT(write.image_data(n).begin());
 
-                       // Copy line by line since ffmpeg sometimes pads each line.
-                       tbb::affinity_partitioner ap;
-                       tbb::parallel_for(tbb::blocked_range<int>(0, desc.planes[n].height), [&](const tbb::blocked_range<int>& r)
+                       if (decoded_linesize != plane.linesize)
                        {
-                               for(int y = r.begin(); y != r.end(); ++y)
-                                       A_memcpy(result + y*plane.linesize, decoded + y*decoded_linesize, plane.linesize);
-                       }, ap);
+                               // Copy line by line since ffmpeg sometimes pads each line.
+                               tbb::affinity_partitioner ap;
+                               tbb::parallel_for(tbb::blocked_range<int>(0, desc.planes[n].height), [&](const tbb::blocked_range<int>& r)
+                               {
+                                       for (int y = r.begin(); y != r.end(); ++y)
+                                               A_memcpy(result + y*plane.linesize, decoded + y*decoded_linesize, plane.linesize);
+                               }, ap);
+                       }
+                       else
+                       {
+                               fast_memcpy(result, decoded, plane.size);
+                       }
                }
        
                return std::move(write);
@@ -265,8 +272,7 @@ spl::shared_ptr<AVFrame> make_av_frame(core::mutable_frame& frame)
 
 spl::shared_ptr<AVFrame> make_av_frame(std::array<uint8_t*, 4> data, const core::pixel_format_desc& pix_desc)
 {
-       spl::shared_ptr<AVFrame> av_frame(avcodec_alloc_frame(), av_free);      
-       avcodec_get_frame_defaults(av_frame.get());
+       auto av_frame = create_frame();
        
        auto planes              = pix_desc.planes;
        auto format              = pix_desc.format;
@@ -352,77 +358,83 @@ AVRational fix_time_base(AVRational time_base)
 }
 
 double read_fps(AVFormatContext& context, double fail_value)
-{                                              
+{
+       auto framerate = read_framerate(context, boost::rational<int>(static_cast<int>(fail_value * 1000000.0), 1000000));
+       
+       return static_cast<double>(framerate.numerator()) / static_cast<double>(framerate.denominator());
+}
+
+boost::rational<int> read_framerate(AVFormatContext& context, const boost::rational<int>& fail_value)
+{
        auto video_index = av_find_best_stream(&context, AVMEDIA_TYPE_VIDEO, -1, -1, 0, 0);
        auto audio_index = av_find_best_stream(&context, AVMEDIA_TYPE_AUDIO, -1, -1, 0, 0);
-       
-       if(video_index > -1)
+
+       if (video_index > -1)
        {
                const auto video_context = context.streams[video_index]->codec;
-               const auto video_stream  = context.streams[video_index];
+               const auto video_stream = context.streams[video_index];
 
                auto frame_rate_time_base = video_stream->avg_frame_rate;
                std::swap(frame_rate_time_base.num, frame_rate_time_base.den);
 
                if (is_sane_fps(frame_rate_time_base))
                {
-                       return static_cast<double>(frame_rate_time_base.den) / static_cast<double>(frame_rate_time_base.num);
+                       return boost::rational<int>(frame_rate_time_base.den, frame_rate_time_base.num);
                }
 
                AVRational time_base = video_context->time_base;
 
-               if(boost::filesystem::path(context.filename).extension().string() == ".flv")
+               if (boost::filesystem::path(context.filename).extension().string() == ".flv")
                {
                        try
                        {
                                auto meta = read_flv_meta_info(context.filename);
-                               return boost::lexical_cast<double>(meta["framerate"]);
+                               return boost::rational<int>(static_cast<int>(boost::lexical_cast<double>(meta["framerate"]) * 1000000.0), 1000000);
                        }
-                       catch(...)
+                       catch (...)
                        {
-                               return 0.0;
+                               return fail_value;
                        }
                }
                else
                {
                        time_base.num *= video_context->ticks_per_frame;
 
-                       if(!is_sane_fps(time_base))
-                       {                       
+                       if (!is_sane_fps(time_base))
+                       {
                                time_base = fix_time_base(time_base);
 
-                               if(!is_sane_fps(time_base) && audio_index > -1)
+                               if (!is_sane_fps(time_base) && audio_index > -1)
                                {
                                        auto& audio_context = *context.streams[audio_index]->codec;
-                                       auto& audio_stream  = *context.streams[audio_index];
+                                       auto& audio_stream = *context.streams[audio_index];
 
                                        double duration_sec = audio_stream.duration / static_cast<double>(audio_context.sample_rate);
-                                                               
+
                                        time_base.num = static_cast<int>(duration_sec*100000.0);
-                                       time_base.den = static_cast<int>(video_stream->nb_frames*100000);
+                                       time_base.den = static_cast<int>(video_stream->nb_frames * 100000);
                                }
                        }
                }
-               
-               double fps = static_cast<double>(time_base.den) / static_cast<double>(time_base.num);
 
-               double closest_fps = 0.0;
+               boost::rational<int> fps(time_base.den, time_base.num);
+               boost::rational<int> closest_fps(0);
 
                for (auto video_mode : enum_constants<core::video_format>())
                {
                        auto format = core::video_format_desc(core::video_format(video_mode));
 
-                       double diff1 = std::abs(format.fps - fps);
-                       double diff2 = std::abs(closest_fps - fps);
+                       auto diff1 = boost::abs(boost::rational<int>(format.time_scale, format.duration) - fps);
+                       auto diff2 = boost::abs(closest_fps - fps);
 
-                       if(diff1 < diff2)
-                               closest_fps = format.fps;
+                       if (diff1 < diff2)
+                               closest_fps = boost::rational<int>(format.time_scale, format.duration);
                }
-       
+
                return closest_fps;
        }
 
-       return fail_value;      
+       return fail_value;
 }
 
 void fix_meta_data(AVFormatContext& context)
@@ -471,19 +483,32 @@ spl::shared_ptr<AVPacket> create_packet()
 
 spl::shared_ptr<AVFrame> create_frame()
 {      
-       spl::shared_ptr<AVFrame> frame(avcodec_alloc_frame(), av_free);
+       spl::shared_ptr<AVFrame> frame(av_frame_alloc(), [](AVFrame* p)
+       {
+               av_frame_free(&p);
+       });
        avcodec_get_frame_defaults(frame.get());
        return frame;
 }
 
-spl::shared_ptr<AVCodecContext> open_codec(AVFormatContext& context, enum AVMediaType type, int& index)
+std::shared_ptr<AVFrame> flush()
+{
+       static std::shared_ptr<AVFrame> dummy(av_frame_alloc(), [](AVFrame* p)
+       {
+               av_frame_free(&p);
+       });
+
+       return dummy;
+}
+
+spl::shared_ptr<AVCodecContext> open_codec(AVFormatContext& context, enum AVMediaType type, int& index, bool single_threaded)
 {      
        AVCodec* decoder;
-       index = THROW_ON_ERROR2(av_find_best_stream(&context, type, -1, -1, &decoder, 0), "");
+       index = THROW_ON_ERROR2(av_find_best_stream(&context, type, index, -1, &decoder, 0), "");
        //if(strcmp(decoder->name, "prores") == 0 && decoder->next && strcmp(decoder->next->name, "prores_lgpl") == 0)
        //      decoder = decoder->next;
 
-       THROW_ON_ERROR2(tbb_avcodec_open(context.streams[index]->codec, decoder), "");
+       THROW_ON_ERROR2(tbb_avcodec_open(context.streams[index]->codec, decoder, single_threaded), "");
        return spl::shared_ptr<AVCodecContext>(context.streams[index]->codec, tbb_avcodec_close);
 }
 
@@ -508,7 +533,7 @@ std::wstring print_mode(int width, int height, double fps, bool interlaced)
        return boost::lexical_cast<std::wstring>(width) + L"x" + boost::lexical_cast<std::wstring>(height) + (!interlaced ? L"p" : L"i") + fps_ss.str();
 }
 
-bool is_valid_file(const std::wstring& filename)
+bool is_valid_file(const std::wstring& filename, bool only_video)
 {                              
        static const auto invalid_exts = {
                L".png",
@@ -526,6 +551,11 @@ bool is_valid_file(const std::wstring& filename)
                L".swf",
                L".ct"
        };
+       static const auto only_audio = {
+               L".mp3",
+               L".wav",
+               L".wma"
+       };
        static const auto valid_exts = {
                L".m2t",
                L".mov",
@@ -533,8 +563,6 @@ bool is_valid_file(const std::wstring& filename)
                L".dv",
                L".flv",
                L".mpg",
-               L".wav",
-               L".mp3",
                L".dnxhd",
                L".h264",
                L".prores"
@@ -543,11 +571,17 @@ bool is_valid_file(const std::wstring& filename)
        auto ext = boost::to_lower_copy(boost::filesystem::path(filename).extension().wstring());
                
        if(std::find(valid_exts.begin(), valid_exts.end(), ext) != valid_exts.end())
-               return true;    
+               return true;
+
+       if (!only_video && std::find(only_audio.begin(), only_audio.end(), ext) != only_audio.end())
+               return true;
        
        if(std::find(invalid_exts.begin(), invalid_exts.end(), ext) != invalid_exts.end())
                return false;   
 
+       if (only_video && std::find(only_audio.begin(), only_audio.end(), ext) != only_audio.end())
+               return false;
+
        auto u8filename = u8(filename);
        
        int score = 0;
@@ -564,7 +598,7 @@ bool is_valid_file(const std::wstring& filename)
                buf.push_back(*file_it);
 
        if(buf.empty())
-               return nullptr;
+               return false;
 
        pb.buf          = buf.data();
        pb.buf_size = static_cast<int>(buf.size());
@@ -603,7 +637,7 @@ bool try_get_duration(const std::wstring filename, std::int64_t& duration, boost
        return true;
 }
 
-std::wstring probe_stem(const std::wstring& stem)
+std::wstring probe_stem(const std::wstring& stem, bool only_video)
 {
        auto stem2 = boost::filesystem::path(stem);
        auto parent = find_case_insensitive(stem2.parent_path().wstring());
@@ -615,16 +649,14 @@ std::wstring probe_stem(const std::wstring& stem)
 
        for(auto it = boost::filesystem::directory_iterator(dir); it != boost::filesystem::directory_iterator(); ++it)
        {
-               if(boost::iequals(it->path().stem().wstring(), stem2.filename().wstring()) && is_valid_file(it->path().wstring()))
+               if(boost::iequals(it->path().stem().wstring(), stem2.filename().wstring()) && is_valid_file(it->path().wstring(), only_video))
                        return it->path().wstring();
        }
        return L"";
 }
 
-core::audio_channel_layout get_audio_channel_layout(const AVCodecContext& codec_context, const std::wstring& channel_layout_spec)
+core::audio_channel_layout get_audio_channel_layout(int num_channels, std::uint64_t layout, const std::wstring& channel_layout_spec)
 {
-       auto num_channels = codec_context.channels;
-
        if (!channel_layout_spec.empty())
        {
                if (boost::contains(channel_layout_spec, L":")) // Custom on the fly layout specified.
@@ -638,18 +670,18 @@ core::audio_channel_layout get_audio_channel_layout(const AVCodecContext& codec_
                }
                else // Preconfigured named channel layout selected.
                {
-                       auto layout = core::audio_channel_layout_repository::get_default()->get_layout(channel_layout_spec);
+                       auto channel_layout = core::audio_channel_layout_repository::get_default()->get_layout(channel_layout_spec);
 
-                       if (!layout)
-                               CASPAR_THROW_EXCEPTION(invalid_argument() << msg_info(L"No channel layout with name " + channel_layout_spec + L" registered"));
+                       if (!channel_layout)
+                               CASPAR_THROW_EXCEPTION(user_error() << msg_info(L"No channel layout with name " + channel_layout_spec + L" registered"));
 
-                       layout->num_channels = num_channels;
+                       channel_layout->num_channels = num_channels;
 
-                       return *layout;
+                       return *channel_layout;
                }
        }
 
-       if (!codec_context.channel_layout)
+       if (!layout)
        {
                if (num_channels == 1)
                        return core::audio_channel_layout(num_channels, L"mono", L"FC");
@@ -659,13 +691,13 @@ core::audio_channel_layout get_audio_channel_layout(const AVCodecContext& codec_
                        return core::audio_channel_layout(num_channels, L"", L""); // Passthru without named channels as is.
        }
 
-       // What FFMpeg calls "channel layout" is only the "layout type" of a channel layout in
+       // What FFmpeg calls "channel layout" is only the "layout type" of a channel layout in
        // CasparCG where the channel layout supports different orders as well.
        // The user needs to provide additional mix-configs in casparcg.config to support more
        // than the most common (5.1, mono and stereo) types.
 
        // Based on information in https://ffmpeg.org/ffmpeg-utils.html#Channel-Layout
-       switch (codec_context.channel_layout)
+       switch (layout)
        {
        case AV_CH_LAYOUT_MONO:
                return core::audio_channel_layout(num_channels, L"mono",                        L"FC");
@@ -729,7 +761,7 @@ core::audio_channel_layout get_audio_channel_layout(const AVCodecContext& codec_
 std::int64_t create_channel_layout_bitmask(int num_channels)
 {
        if (num_channels > 63)
-               CASPAR_THROW_EXCEPTION(invalid_argument() << msg_info(L"FFMpeg cannot handle more than 63 audio channels"));
+               CASPAR_THROW_EXCEPTION(invalid_argument() << msg_info(L"FFmpeg cannot handle more than 63 audio channels"));
 
        const auto ALL_63_CHANNELS = 0x7FFFFFFFFFFFFFFFULL;