extern "C" {
#include "../include/capture/nvfbc.h"
#include "../include/capture/xcomposite_cuda.h"
#include "../include/capture/xcomposite_drm.h"
#include "../include/egl.h"
#include "../include/time.h"
}

#include <assert.h>
#include <stdio.h>
#include <stdlib.h>
#include <string>
#include <vector>
#include <unordered_map>
#include <thread>
#include <mutex>
#include <map>
#include <signal.h>
#include <sys/stat.h>

#include <unistd.h>
#include <fcntl.h>

#include "../include/sound.hpp"

#include <X11/extensions/Xrandr.h>

extern "C" {
#include <libavutil/pixfmt.h>
#include <libavcodec/avcodec.h>
#include <libavformat/avformat.h>
#include <libavutil/opt.h>
#include <libswresample/swresample.h>
#include <libavutil/avutil.h>
#include <libavutil/time.h>
#include <libavfilter/avfilter.h>
#include <libavfilter/buffersink.h>
#include <libavfilter/buffersrc.h>
}

#include <deque>
#include <future>

// TODO: If options are not supported then they are returned (allocated) in the options. This should be free'd.

typedef enum {
    GPU_VENDOR_AMD,
    GPU_VENDOR_INTEL,
    GPU_VENDOR_NVIDIA
} gpu_vendor;

// TODO: Remove LIBAVUTIL_VERSION_MAJOR checks in the future when ubuntu, pop os LTS etc update ffmpeg to >= 5.0

static const int VIDEO_STREAM_INDEX = 0;

static thread_local char av_error_buffer[AV_ERROR_MAX_STRING_SIZE];

static const XRRModeInfo* get_mode_info(const XRRScreenResources *sr, RRMode id) {
    for(int i = 0; i < sr->nmode; ++i) {
        if(sr->modes[i].id == id)
            return &sr->modes[i];
    }    
    return nullptr;
}

typedef void (*active_monitor_callback)(const XRROutputInfo *output_info, const XRRCrtcInfo *crt_info, const XRRModeInfo *mode_info, void *userdata);

static void for_each_active_monitor_output(Display *display, active_monitor_callback callback, void *userdata) {
    XRRScreenResources *screen_res = XRRGetScreenResources(display, DefaultRootWindow(display));
    if(!screen_res)
        return;

    for(int i = 0; i < screen_res->noutput; ++i) {
        XRROutputInfo *out_info = XRRGetOutputInfo(display, screen_res, screen_res->outputs[i]);
        if(out_info && out_info->crtc && out_info->connection == RR_Connected) {
            XRRCrtcInfo *crt_info = XRRGetCrtcInfo(display, screen_res, out_info->crtc);
            if(crt_info && crt_info->mode) {
                const XRRModeInfo *mode_info = get_mode_info(screen_res, crt_info->mode);
                if(mode_info)
                    callback(out_info, crt_info, mode_info, userdata);
            }
            if(crt_info)
                XRRFreeCrtcInfo(crt_info);
        }
        if(out_info)
            XRRFreeOutputInfo(out_info);
    }    

    XRRFreeScreenResources(screen_res);
}

typedef struct {
    vec2i pos;
    vec2i size;
} gsr_monitor;

typedef struct {
    const char *name;
    int name_len;
    gsr_monitor *monitor;
    bool found_monitor;
} get_monitor_by_name_userdata;

static void get_monitor_by_name_callback(const XRROutputInfo *output_info, const XRRCrtcInfo *crt_info, const XRRModeInfo *mode_info, void *userdata) {
    get_monitor_by_name_userdata *data = (get_monitor_by_name_userdata*)userdata;
    if(!data->found_monitor && data->name_len == output_info->nameLen && memcmp(data->name, output_info->name, data->name_len) == 0) {
        data->monitor->pos = { crt_info->x, crt_info->y };
        data->monitor->size = { (int)crt_info->width, (int)crt_info->height };
        data->found_monitor = true;
    }
}

static bool get_monitor_by_name(Display *display, const char *name, gsr_monitor *monitor) {
    get_monitor_by_name_userdata userdata;
    userdata.name = name;
    userdata.name_len = strlen(name);
    userdata.monitor = monitor;
    userdata.found_monitor = false;
    for_each_active_monitor_output(display, get_monitor_by_name_callback, &userdata);
    return userdata.found_monitor;
}

static void monitor_output_callback_print(const XRROutputInfo *output_info, const XRRCrtcInfo *crt_info, const XRRModeInfo *mode_info, void *userdata) {
    fprintf(stderr, "    \"%.*s\"    (%dx%d+%d+%d)\n", output_info->nameLen, output_info->name, (int)crt_info->width, (int)crt_info->height, crt_info->x, crt_info->y);
}

static char* av_error_to_string(int err) {
    if(av_strerror(err, av_error_buffer, sizeof(av_error_buffer)) < 0)
        strcpy(av_error_buffer, "Unknown error");
    return av_error_buffer;
}

enum class VideoQuality {
    MEDIUM,
    HIGH,
    VERY_HIGH,
    ULTRA
};

enum class VideoCodec {
    H264,
    H265
};

enum class AudioCodec {
    AAC,
    OPUS,
    FLAC
};

enum class PixelFormat {
    YUV420,
    YUV444
};

static int x11_error_handler(Display *dpy, XErrorEvent *ev) {
    return 0;
}

static int x11_io_error_handler(Display *dpy) {
    return 0;
}

// |stream| is only required for non-replay mode
static void receive_frames(AVCodecContext *av_codec_context, int stream_index, AVStream *stream, int64_t pts,
                           AVFormatContext *av_format_context,
                           double replay_start_time,
                           std::deque<AVPacket> &frame_data_queue,
                           int replay_buffer_size_secs,
                           bool &frames_erased,
						   std::mutex &write_output_mutex) {
    for (;;) {
        // TODO: Use av_packet_alloc instead because sizeof(av_packet) might not be future proof(?)
        AVPacket av_packet;
        memset(&av_packet, 0, sizeof(av_packet));
        av_packet.data = NULL;
        av_packet.size = 0;
        int res = avcodec_receive_packet(av_codec_context, &av_packet);
        if (res == 0) { // we have a packet, send the packet to the muxer
            av_packet.stream_index = stream_index;
            av_packet.pts = pts;
            av_packet.dts = pts;

            std::lock_guard<std::mutex> lock(write_output_mutex);
            if(replay_buffer_size_secs != -1) {
                double time_now = clock_get_monotonic_seconds();
                double replay_time_elapsed = time_now - replay_start_time;

                AVPacket new_pack;
                av_packet_move_ref(&new_pack, &av_packet);
                frame_data_queue.push_back(std::move(new_pack));
                if(replay_time_elapsed >= replay_buffer_size_secs) {
                    av_packet_unref(&frame_data_queue.front());
                    frame_data_queue.pop_front();
                    frames_erased = true;
                }
                av_packet_unref(&av_packet);
            } else {
                av_packet_rescale_ts(&av_packet, av_codec_context->time_base, stream->time_base);
                av_packet.stream_index = stream->index;
                // TODO: Is av_interleaved_write_frame needed?
                int ret = av_interleaved_write_frame(av_format_context, &av_packet);
                if(ret < 0) {
                    fprintf(stderr, "Error: Failed to write frame index %d to muxer, reason: %s (%d)\n", av_packet.stream_index, av_error_to_string(ret), ret);
                }
            }
        } else if (res == AVERROR(EAGAIN)) { // we have no packet
                                             // fprintf(stderr, "No packet!\n");
            av_packet_unref(&av_packet);
            break;
        } else if (res == AVERROR_EOF) { // this is the end of the stream
            fprintf(stderr, "End of stream!\n");
            av_packet_unref(&av_packet);
            break;
        } else {
            fprintf(stderr, "Unexpected error: %d\n", res);
            av_packet_unref(&av_packet);
            break;
        }
    }
}

static const char* audio_codec_get_name(AudioCodec audio_codec) {
    switch(audio_codec) {
        case AudioCodec::AAC:  return "aac";
        case AudioCodec::OPUS: return "opus";
        case AudioCodec::FLAC: return "flac";
    }
    assert(false);
    return "";
}

static AVCodecID audio_codec_get_id(AudioCodec audio_codec) {
    switch(audio_codec) {
        case AudioCodec::AAC:  return AV_CODEC_ID_AAC;
        case AudioCodec::OPUS: return AV_CODEC_ID_OPUS;
        case AudioCodec::FLAC: return AV_CODEC_ID_FLAC;
    }
    assert(false);
    return AV_CODEC_ID_AAC;
}

static AVSampleFormat audio_codec_get_sample_format(AudioCodec audio_codec, const AVCodec *codec) {
    switch(audio_codec) {
        case AudioCodec::AAC: {
            return AV_SAMPLE_FMT_FLTP;
        }
        case AudioCodec::OPUS: {
            bool supports_s16 = false;
            bool supports_flt = false;

            for(size_t i = 0; codec->sample_fmts && codec->sample_fmts[i] != -1; ++i) {
                if(codec->sample_fmts[i] == AV_SAMPLE_FMT_S16) {
                    supports_s16 = true;
                } else if(codec->sample_fmts[i] == AV_SAMPLE_FMT_FLT) {
                    supports_flt = true;
                }
            }

            if(!supports_s16 && !supports_flt) {
                fprintf(stderr, "Warning: opus audio codec is chosen but your ffmpeg version does not support s16/flt sample format and performance might be slightly worse. You can either rebuild ffmpeg with libopus instead of the built-in opus, use the flatpak version of gpu screen recorder or record with flac audio codec instead (-ac flac). Falling back to fltp audio sample format instead.\n");
            }

            if(supports_s16)
                return AV_SAMPLE_FMT_S16;
            else if(supports_flt)
                return AV_SAMPLE_FMT_FLT;
            else
                return AV_SAMPLE_FMT_FLTP;
        }
        case AudioCodec::FLAC: {
            return AV_SAMPLE_FMT_S32;
        }
    }
    assert(false);
    return AV_SAMPLE_FMT_FLTP;
}

static int64_t audio_codec_get_get_bitrate(AudioCodec audio_codec) {
    switch(audio_codec) {
        case AudioCodec::AAC:  return 128000;
        case AudioCodec::OPUS: return 96000;
        case AudioCodec::FLAC: return 96000;
    }
    assert(false);
    return 96000;
}

static AudioFormat audio_codec_context_get_audio_format(const AVCodecContext *audio_codec_context) {
    switch(audio_codec_context->sample_fmt) {
        case AV_SAMPLE_FMT_FLT:   return F32;
        case AV_SAMPLE_FMT_FLTP:  return S32;
        case AV_SAMPLE_FMT_S16:   return S16;
        case AV_SAMPLE_FMT_S32:   return S32;
        default:                  return S16;
    }
}

static AVSampleFormat audio_format_to_sample_format(const AudioFormat audio_format) {
    switch(audio_format) {
        case S16:   return AV_SAMPLE_FMT_S16;
        case S32:   return AV_SAMPLE_FMT_S32;
        case F32:   return AV_SAMPLE_FMT_FLT;
    }
    assert(false);
    return AV_SAMPLE_FMT_S16;
}

static AVCodecContext* create_audio_codec_context(int fps, AudioCodec audio_codec) {
    const AVCodec *codec = avcodec_find_encoder(audio_codec_get_id(audio_codec));
    if (!codec) {
        fprintf(stderr, "Error: Could not find %s audio encoder\n", audio_codec_get_name(audio_codec));
        exit(1);
    }

    AVCodecContext *codec_context = avcodec_alloc_context3(codec);

    assert(codec->type == AVMEDIA_TYPE_AUDIO);
	codec_context->codec_id = codec->id;
    codec_context->sample_fmt = audio_codec_get_sample_format(audio_codec, codec);
    codec_context->bit_rate = audio_codec_get_get_bitrate(audio_codec);
    codec_context->sample_rate = 48000;
    if(audio_codec == AudioCodec::AAC)
        codec_context->profile = FF_PROFILE_AAC_LOW;
#if LIBAVCODEC_VERSION_MAJOR < 60
    codec_context->channel_layout = AV_CH_LAYOUT_STEREO;
    codec_context->channels = 2;
#else
    av_channel_layout_default(&codec_context->ch_layout, 2);
#endif

    codec_context->time_base.num = 1;
    codec_context->time_base.den = codec_context->sample_rate;
    codec_context->framerate.num = fps;
    codec_context->framerate.den = 1;
    codec_context->flags |= AV_CODEC_FLAG_GLOBAL_HEADER;

    return codec_context;
}

static AVCodecContext *create_video_codec_context(AVPixelFormat pix_fmt,
                            VideoQuality video_quality,
                            int fps, const AVCodec *codec, bool is_livestream, gpu_vendor vendor) {

    AVCodecContext *codec_context = avcodec_alloc_context3(codec);

    //double fps_ratio = (double)fps / 30.0;

    assert(codec->type == AVMEDIA_TYPE_VIDEO);
    codec_context->codec_id = codec->id;
    // Timebase: This is the fundamental unit of time (in seconds) in terms
    // of which frame timestamps are represented. For fixed-fps content,
    // timebase should be 1/framerate and timestamp increments should be
    // identical to 1
    codec_context->time_base.num = 1;
    codec_context->time_base.den = fps;
    codec_context->framerate.num = fps;
    codec_context->framerate.den = 1;
    codec_context->sample_aspect_ratio.num = 0;
    codec_context->sample_aspect_ratio.den = 0;
    // High values reeduce file size but increases time it takes to seek
    if(is_livestream) {
        codec_context->flags |= (AV_CODEC_FLAG_CLOSED_GOP | AV_CODEC_FLAG_LOW_DELAY);
        codec_context->flags2 |= AV_CODEC_FLAG2_FAST;
        //codec_context->gop_size = std::numeric_limits<int>::max();
        //codec_context->keyint_min = std::numeric_limits<int>::max();
        codec_context->gop_size = fps * 2;
    } else {
        codec_context->gop_size = fps * 2;
    }
    codec_context->max_b_frames = 0;
    codec_context->pix_fmt = pix_fmt;
    //codec_context->color_range = AVCOL_RANGE_JPEG;
    //codec_context->color_primaries = AVCOL_PRI_BT709;
    //codec_context->color_trc = AVCOL_TRC_BT709;
    //codec_context->colorspace = AVCOL_SPC_BT709;
    if(codec->id == AV_CODEC_ID_HEVC)
        codec_context->codec_tag = MKTAG('h', 'v', 'c', '1');
    switch(video_quality) {
        case VideoQuality::MEDIUM:
            //codec_context->qmin = 35;
            //codec_context->qmax = 35;
            codec_context->bit_rate = 100000;//4500000 + (codec_context->width * codec_context->height)*0.75;
            break;
        case VideoQuality::HIGH:
            //codec_context->qmin = 34;
            //codec_context->qmax = 34;
            codec_context->bit_rate = 100000;//10000000-9000000 + (codec_context->width * codec_context->height)*0.75;
            break;
        case VideoQuality::VERY_HIGH:
            //codec_context->qmin = 28;
            //codec_context->qmax = 28;
            codec_context->bit_rate = 100000;//10000000-9000000 + (codec_context->width * codec_context->height)*0.75;
            break;
        case VideoQuality::ULTRA:
            //codec_context->qmin = 22;
            //codec_context->qmax = 22;
            codec_context->bit_rate = 100000;//10000000-9000000 + (codec_context->width * codec_context->height)*0.75;
            break;
    }
    //codec_context->profile = FF_PROFILE_H264_MAIN;
    if (codec_context->codec_id == AV_CODEC_ID_MPEG1VIDEO)
        codec_context->mb_decision = 2;

    // stream->time_base = codec_context->time_base;
    // codec_context->ticks_per_frame = 30;
    //av_opt_set(codec_context->priv_data, "tune", "hq", 0);
    // TODO: Do this for better file size? also allows setting qmin, qmax per frame? which can then be used to dynamically set bitrate to reduce quality
    // if live streaming is slow or if the users harddrive is cant handle writing megabytes of data per second.
    #if 0
    char qmin_str[32];
    snprintf(qmin_str, sizeof(qmin_str), "%d", codec_context->qmin);

    char qmax_str[32];
    snprintf(qmax_str, sizeof(qmax_str), "%d", codec_context->qmax);

    av_opt_set(codec_context->priv_data, "cq", qmax_str, 0);
    av_opt_set(codec_context->priv_data, "rc", "vbr", 0);
    av_opt_set(codec_context->priv_data, "qmin", qmin_str, 0);
    av_opt_set(codec_context->priv_data, "qmax", qmax_str, 0);
    codec_context->bit_rate = 0;
    #endif

    av_opt_set_int(codec_context->priv_data, "b_ref_mode", 0, 0);

    if(vendor != GPU_VENDOR_NVIDIA) {
        // TODO: More options, better options
        //codec_context->bit_rate = codec_context->width * codec_context->height;
        av_opt_set(codec_context->priv_data, "rc_mode", "CQP", 0);
        codec_context->global_quality = 4;
    }

    //codec_context->rc_max_rate = codec_context->bit_rate;
    //codec_context->rc_min_rate = codec_context->bit_rate;
    //codec_context->rc_buffer_size = codec_context->bit_rate / 10;

    codec_context->flags |= AV_CODEC_FLAG_GLOBAL_HEADER;

    return codec_context;
}

static bool check_if_codec_valid_for_hardware(const AVCodec *codec, gpu_vendor vendor) {
    // TODO: For now we assume that amd and intel always support h264 and hevc, but we default to h264
    if(vendor != GPU_VENDOR_NVIDIA)
        return true;

    bool success = false;
    // Do not use AV_PIX_FMT_CUDA because we dont want to do full check with hardware context
    AVCodecContext *codec_context = create_video_codec_context(AV_PIX_FMT_YUV420P, VideoQuality::VERY_HIGH, 60, codec, false, vendor);
    codec_context->width = 1920;
    codec_context->height = 1080;
    if(codec_context) {
        success = avcodec_open2(codec_context, codec_context->codec, NULL) == 0;
        avcodec_free_context(&codec_context);
    }
    return success;
}

static const AVCodec* find_h264_encoder(gpu_vendor vendor) {
    const AVCodec *codec = avcodec_find_encoder_by_name(vendor == GPU_VENDOR_NVIDIA ? "h264_nvenc" : "h264_vaapi");
    if(!codec)
        codec = avcodec_find_encoder_by_name(vendor == GPU_VENDOR_NVIDIA ? "nvenc_h264" : "vaapi_h264");

    static bool checked = false;
    static bool checked_success = true;
    if(!checked) {
        checked = true;
        if(!check_if_codec_valid_for_hardware(codec, vendor))
            checked_success = false;
    }
    return checked_success ? codec : nullptr;
}

// TODO: Disable under intel/amd?

static const AVCodec* find_h265_encoder(gpu_vendor vendor) {
    const AVCodec *codec = avcodec_find_encoder_by_name(vendor == GPU_VENDOR_NVIDIA ? "hevc_nvenc" : "hevc_vaapi");
    if(!codec)
        codec = avcodec_find_encoder_by_name(vendor == GPU_VENDOR_NVIDIA ? "nvenc_hevc" : "vaapi_hevc");

    if(!codec)
        return nullptr;

    static bool checked = false;
    static bool checked_success = true;
    if(!checked) {
        checked = true;
        if(!check_if_codec_valid_for_hardware(codec, vendor))
            checked_success = false;
    }
    return checked_success ? codec : nullptr;
}

static AVFrame* open_audio(AVCodecContext *audio_codec_context) {
    AVDictionary *options = nullptr;
    av_dict_set(&options, "strict", "experimental", 0);

    int ret;
    ret = avcodec_open2(audio_codec_context, audio_codec_context->codec, &options);
    if(ret < 0) {
        fprintf(stderr, "failed to open codec, reason: %s\n", av_error_to_string(ret));
        exit(1);
    }

    AVFrame *frame = av_frame_alloc();
    if(!frame) {
        fprintf(stderr, "failed to allocate audio frame\n");
        exit(1);
    }

    frame->sample_rate = audio_codec_context->sample_rate;
    frame->nb_samples = audio_codec_context->frame_size;
    frame->format = audio_codec_context->sample_fmt;
#if LIBAVCODEC_VERSION_MAJOR < 60
    frame->channels = audio_codec_context->channels;
    frame->channel_layout = audio_codec_context->channel_layout;
#else
    av_channel_layout_copy(&frame->ch_layout, &audio_codec_context->ch_layout);
#endif

    ret = av_frame_get_buffer(frame, 0);
    if(ret < 0) {
        fprintf(stderr, "failed to allocate audio data buffers, reason: %s\n", av_error_to_string(ret));
        exit(1);
    }

    return frame;
}

static void open_video(AVCodecContext *codec_context, VideoQuality video_quality, bool very_old_gpu, gpu_vendor vendor, PixelFormat pixel_format) {
    AVDictionary *options = nullptr;
    if(vendor == GPU_VENDOR_NVIDIA) {
        bool supports_p4 = false;
        bool supports_p6 = false;

        const AVOption *opt = nullptr;
        while((opt = av_opt_next(codec_context->priv_data, opt))) {
            if(opt->type == AV_OPT_TYPE_CONST) {
                if(strcmp(opt->name, "p4") == 0)
                    supports_p4 = true;
                else if(strcmp(opt->name, "p6") == 0)
                    supports_p6 = true;
            }
        }

        if(very_old_gpu) {
            switch(video_quality) {
                case VideoQuality::MEDIUM:
                    av_dict_set_int(&options, "qp", 37, 0);
                    break;
                case VideoQuality::HIGH:
                    av_dict_set_int(&options, "qp", 32, 0);
                    break;
                case VideoQuality::VERY_HIGH:
                    av_dict_set_int(&options, "qp", 27, 0);
                    break;
                case VideoQuality::ULTRA:
                    av_dict_set_int(&options, "qp", 21, 0);
                    break;
            }
        } else {
            switch(video_quality) {
                case VideoQuality::MEDIUM:
                    av_dict_set_int(&options, "qp", 40, 0);
                    break;
                case VideoQuality::HIGH:
                    av_dict_set_int(&options, "qp", 35, 0);
                    break;
                case VideoQuality::VERY_HIGH:
                    av_dict_set_int(&options, "qp", 30, 0);
                    break;
                case VideoQuality::ULTRA:
                    av_dict_set_int(&options, "qp", 24, 0);
                    break;
            }
        }

        if(!supports_p4 && !supports_p6)
            fprintf(stderr, "Info: your ffmpeg version is outdated. It's recommended that you use the flatpak version of gpu-screen-recorder version instead, which you can find at https://flathub.org/apps/details/com.dec05eba.gpu_screen_recorder\n");

        //if(is_livestream) {
        //    av_dict_set_int(&options, "zerolatency", 1, 0);
        //    //av_dict_set(&options, "preset", "llhq", 0);
        //}

        // Fuck nvidia and ffmpeg, I want to use a good preset for the gpu but all gpus prefer different
        // presets. Nvidia and ffmpeg used to support "hq" preset that chose the best preset for the gpu
        // with pretty good performance but you now have to choose p1-p7, which are gpu agnostic and on
        // older gpus p5-p7 slow the gpu down to a crawl...
        // "hq" is now just an alias for p7 in ffmpeg :(
        // TODO: Temporary disable because of stuttering?
        if(very_old_gpu)
            av_dict_set(&options, "preset", supports_p4 ? "p4" : "medium", 0);
        else
            av_dict_set(&options, "preset", supports_p6 ? "p6" : "slow", 0);

        av_dict_set(&options, "tune", "hq", 0);
        av_dict_set(&options, "rc", "constqp", 0);

        if(codec_context->codec_id == AV_CODEC_ID_H264) {
            switch(pixel_format) {
                case PixelFormat::YUV420:
                    av_dict_set(&options, "profile", "high", 0);
                    break;
                case PixelFormat::YUV444:
                    av_dict_set(&options, "profile", "high444p", 0);
                    break;
            }
        }
    } else {
        switch(video_quality) {
            case VideoQuality::MEDIUM:
                av_dict_set_int(&options, "qp", 40, 0);
                break;
            case VideoQuality::HIGH:
                av_dict_set_int(&options, "qp", 35, 0);
                break;
            case VideoQuality::VERY_HIGH:
                av_dict_set_int(&options, "qp", 30, 0);
                break;
            case VideoQuality::ULTRA:
                av_dict_set_int(&options, "qp", 24, 0);
                break;
        }

        // TODO: More quality options
        av_dict_set(&options, "rc_mode", "CQP", 0);
        //av_dict_set_int(&options, "low_power", 1, 0);

        if(codec_context->codec_id == AV_CODEC_ID_H264) {
            av_dict_set(&options, "profile", "high", 0);
            av_dict_set_int(&options, "quality", 32, 0);
        } else {
            av_dict_set(&options, "profile", "main", 0);
        }
    }

    if(codec_context->codec_id == AV_CODEC_ID_H264) {
        av_dict_set(&options, "coder", "cabac", 0); // TODO: cavlc is faster than cabac but worse compression. Which to use?
    }

    av_dict_set(&options, "strict", "experimental", 0);

    int ret = avcodec_open2(codec_context, codec_context->codec, &options);
    if (ret < 0) {
        fprintf(stderr, "Error: Could not open video codec: %s\n", av_error_to_string(ret));
        exit(1);
    }
}

static void usage() {
    fprintf(stderr, "usage: gpu-screen-recorder -w <window_id|monitor|focused> [-c <container_format>] [-s WxH] -f <fps> [-a <audio_input>...] [-q <quality>] [-r <replay_buffer_size_sec>] [-k h264|h265] [-ac aac|opus|flac] [-oc yes|no] [-pixfmt yuv420|yuv444] [-o <output_file>]\n");
    fprintf(stderr, "OPTIONS:\n");
    fprintf(stderr, "  -w       Window to record, a display, \"screen\", \"screen-direct\", \"screen-direct-force\" or \"focused\". The display is the display (monitor) name in xrandr and if \"screen\" or \"screen-direct\" is selected then all displays are recorded. If this is \"focused\" then the currently focused window is recorded. When recording the focused window then the -s option has to be used as well.\n"
        "        \"screen-direct\"/\"screen-direct-force\" skips one texture copy for fullscreen applications so it may lead to better performance and it works with VRR monitors when recording fullscreen application but may break some applications, such as mpv in fullscreen mode. Direct mode doesn't capture cursor either. \"screen-direct-force\" is not recommended unless you use a VRR monitor because there might be driver issues that cause the video to stutter or record a black screen.\n");
    fprintf(stderr, "  -c       Container format for output file, for example mp4, or flv. Only required if no output file is specified or if recording in replay buffer mode. If an output file is specified and -c is not used then the container format is determined from the output filename extension.\n");
    fprintf(stderr, "  -s       The size (area) to record at in the format WxH, for example 1920x1080. This option is only supported (and required) when -w is \"focused\".\n");
    fprintf(stderr, "  -f       Framerate to record at.\n");
    fprintf(stderr, "  -a       Audio device to record from (pulse audio device). Can be specified multiple times. Each time this is specified a new audio track is added for the specified audio device. A name can be given to the audio input device by prefixing the audio input with <name>/, for example \"dummy/alsa_output.pci-0000_00_1b.0.analog-stereo.monitor\". Multiple audio devices can be merged into one audio track by using \"|\" as a separator into one -a argument, for example: -a \"alsa_output1|alsa_output2\". Optional, no audio track is added by default.\n");
    fprintf(stderr, "  -q       Video quality. Should be either 'medium', 'high', 'very_high' or 'ultra'. 'high' is the recommended option when live streaming or when you have a slower harddrive. Optional, set to 'very_high' be default.\n");
    fprintf(stderr, "  -r       Replay buffer size in seconds. If this is set, then only the last seconds as set by this option will be stored"
        " and the video will only be saved when the gpu-screen-recorder is closed. This feature is similar to Nvidia's instant replay feature."
        " This option has be between 5 and 1200. Note that the replay buffer size will not always be precise, because of keyframes. Optional, disabled by default.\n");
    fprintf(stderr, "  -k       Video codec to use. Should be either 'auto', 'h264' or 'h265'. Defaults to 'auto' which defaults to 'h265' on nvidia unless recording at a higher resolution than 3840x2160. On AMD/Intel this defaults to 'auto' which defaults to 'h264'. Forcefully set to 'h264' if -c is 'flv'.\n");
    fprintf(stderr, "  -ac      Audio codec to use. Should be either 'aac', 'opus' or 'flac'. Defaults to 'opus' for .mp4/.mkv files, otherwise defaults to 'aac'. 'opus' and 'flac' is only supported by .mp4/.mkv files. 'opus' is recommended for best performance and smallest audio size.\n");
    fprintf(stderr, "  -oc      Overclock memory transfer rate to the maximum performance level. This only applies to NVIDIA and exists to overcome a bug in NVIDIA driver where performance level is dropped when you record a game. Only needed if you are recording a game that is bottlenecked by GPU. Works only if your have \"Coolbits\" set to \"12\" in NVIDIA X settings, see README for more information. Note! use at your own risk! Optional, disabled by default\n");
    fprintf(stderr, "  -pixfmt  The pixel format to use for the output video. yuv420 is the most common format and is best supported, but the color is compressed, so colors can look washed outandr certain colors of text can look bad. Use yuv444 for no color compression, but the video may not work everywhere and it may not work with hardware video decoding. Optional, defaults to yuv420\n");
    fprintf(stderr, "  -o       The output file path. If omitted then the encoded data is sent to stdout. Required in replay mode (when using -r). In replay mode this has to be an existing directory instead of a file.\n");
    fprintf(stderr, "NOTES:\n");
    fprintf(stderr, "  Send signal SIGINT (Ctrl+C) to gpu-screen-recorder to stop and save the recording (when not using replay mode).\n");
    fprintf(stderr, "  Send signal SIGUSR1 (killall -SIGUSR1 gpu-screen-recorder) to gpu-screen-recorder to save a replay.\n");
    fprintf(stderr, "EXAMPLES\n");
    fprintf(stderr, "  gpu-screen-recorder -w screen -f 60 -a \"$(pactl get-default-sink).monitor\" -o video.mp4\n");
    fprintf(stderr, "  gpu-screen-recorder -w screen -f 60 -q ultra -pixfmt yuv444 -o video.mp4\n");
    exit(1);
}

static sig_atomic_t running = 1;
static sig_atomic_t save_replay = 0;

static void int_handler(int) {
    running = 0;
}

static void save_replay_handler(int) {
    save_replay = 1;
}

struct Arg {
    std::vector<const char*> values;
    bool optional = false;
    bool list = false;

    const char* value() const {
        if(values.empty())
            return nullptr;
        return values.front();
    }
};

static bool is_hex_num(char c) {
    return (c >= 'A' && c <= 'F') || (c >= 'a' && c <= 'f') || (c >= '0' && c <= '9');
}

static bool contains_non_hex_number(const char *str) {
    size_t len = strlen(str);
    if(len >= 2 && memcmp(str, "0x", 2) == 0) {
        str += 2;
        len -= 2;
    }

    for(size_t i = 0; i < len; ++i) {
        char c = str[i];
        if(c == '\0')
            return false;
        if(!is_hex_num(c))
            return true;
    }
    return false;
}

static std::string get_date_str() {
    char str[128];
    time_t now = time(NULL);
    struct tm *t = localtime(&now);
    strftime(str, sizeof(str)-1, "%Y-%m-%d_%H-%M-%S", t);
    return str; 
}

static AVStream* create_stream(AVFormatContext *av_format_context, AVCodecContext *codec_context) {
    AVStream *stream = avformat_new_stream(av_format_context, nullptr);
    if (!stream) {
        fprintf(stderr, "Error: Could not allocate stream\n");
        exit(1);
    }
    stream->id = av_format_context->nb_streams - 1;
    stream->time_base = codec_context->time_base;
    stream->avg_frame_rate = codec_context->framerate;
    return stream;
}

struct AudioDevice {
    SoundDevice sound_device;
    AudioInput audio_input;
    AVFilterContext *src_filter_ctx = nullptr;
    std::thread thread; // TODO: Instead of having a thread for each track, have one thread for all threads and read the data with non-blocking read
};

struct AudioTrack {
    AVCodecContext *codec_context = nullptr;
    AVFrame *frame = nullptr;
    AVStream *stream = nullptr;

    std::vector<AudioDevice> audio_devices;
    AVFilterGraph *graph = nullptr;
    AVFilterContext *sink = nullptr;
    int64_t pts = 0;
    int stream_index = 0;
};

static std::future<void> save_replay_thread;
static std::vector<AVPacket> save_replay_packets;
static std::string save_replay_output_filepath;

static void save_replay_async(AVCodecContext *video_codec_context, int video_stream_index, std::vector<AudioTrack> &audio_tracks, const std::deque<AVPacket> &frame_data_queue, bool frames_erased, std::string output_dir, const char *container_format, const std::string &file_extension, std::mutex &write_output_mutex) {
    if(save_replay_thread.valid())
        return;
    
    size_t start_index = (size_t)-1;
    int64_t video_pts_offset = 0;
    int64_t audio_pts_offset = 0;

    {
        std::lock_guard<std::mutex> lock(write_output_mutex);
        start_index = (size_t)-1;
        for(size_t i = 0; i < frame_data_queue.size(); ++i) {
            const AVPacket &av_packet = frame_data_queue[i];
            if((av_packet.flags & AV_PKT_FLAG_KEY) && av_packet.stream_index == video_stream_index) {
                start_index = i;
                break;
            }
        }

        if(start_index == (size_t)-1)
            return;

        if(frames_erased) {
            video_pts_offset = frame_data_queue[start_index].pts;
            
            // Find the next audio packet to use as audio pts offset
            for(size_t i = start_index; i < frame_data_queue.size(); ++i) {
                const AVPacket &av_packet = frame_data_queue[i];
                if(av_packet.stream_index != video_stream_index) {
                    audio_pts_offset = av_packet.pts;
                    break;
                }
            }
        } else {
            start_index = 0;
        }

        save_replay_packets.resize(frame_data_queue.size());
        for(size_t i = 0; i < frame_data_queue.size(); ++i) {
            av_packet_ref(&save_replay_packets[i], &frame_data_queue[i]);
        }
    }

    save_replay_output_filepath = output_dir + "/Replay_" + get_date_str() + "." + file_extension;
    save_replay_thread = std::async(std::launch::async, [video_stream_index, container_format, start_index, video_pts_offset, audio_pts_offset, video_codec_context, &audio_tracks]() mutable {
        AVFormatContext *av_format_context;
        avformat_alloc_output_context2(&av_format_context, nullptr, container_format, nullptr);

        AVStream *video_stream = create_stream(av_format_context, video_codec_context);
        avcodec_parameters_from_context(video_stream->codecpar, video_codec_context);

        std::unordered_map<int, AudioTrack*> stream_index_to_audio_track_map;
        for(AudioTrack &audio_track : audio_tracks) {
            stream_index_to_audio_track_map[audio_track.stream_index] = &audio_track;
            AVStream *audio_stream = create_stream(av_format_context, audio_track.codec_context);
            avcodec_parameters_from_context(audio_stream->codecpar, audio_track.codec_context);
            audio_track.stream = audio_stream;
        }

        int ret = avio_open(&av_format_context->pb, save_replay_output_filepath.c_str(), AVIO_FLAG_WRITE);
        if (ret < 0) {
            fprintf(stderr, "Error: Could not open '%s': %s. Make sure %s is an existing directory with write access\n", save_replay_output_filepath.c_str(), av_error_to_string(ret), save_replay_output_filepath.c_str());
            return;
        }

        AVDictionary *options = nullptr;
        av_dict_set(&options, "strict", "experimental", 0);

        ret = avformat_write_header(av_format_context, &options);
        if (ret < 0) {
            fprintf(stderr, "Error occurred when writing header to output file: %s\n", av_error_to_string(ret));
            return;
        }

        for(size_t i = start_index; i < save_replay_packets.size(); ++i) {
            AVPacket &av_packet = save_replay_packets[i];

            AVStream *stream = video_stream;
            AVCodecContext *codec_context = video_codec_context;

            if(av_packet.stream_index == video_stream_index) {
                av_packet.pts -= video_pts_offset;
                av_packet.dts -= video_pts_offset;
            } else {
                AudioTrack *audio_track = stream_index_to_audio_track_map[av_packet.stream_index];
                stream = audio_track->stream;
                codec_context = audio_track->codec_context;

                av_packet.pts -= audio_pts_offset;
                av_packet.dts -= audio_pts_offset;
            }

            av_packet.stream_index = stream->index;
            av_packet_rescale_ts(&av_packet, codec_context->time_base, stream->time_base);

            int ret = av_interleaved_write_frame(av_format_context, &av_packet);
            if(ret < 0)
                fprintf(stderr, "Error: Failed to write frame index %d to muxer, reason: %s (%d)\n", stream->index, av_error_to_string(ret), ret);
        }

        if (av_write_trailer(av_format_context) != 0)
            fprintf(stderr, "Failed to write trailer\n");

        avio_close(av_format_context->pb);
        avformat_free_context(av_format_context);
        av_dict_free(&options);

        for(AudioTrack &audio_track : audio_tracks) {
            audio_track.stream = nullptr;
        }
    });
}

static void split_string(const std::string &str, char delimiter, std::function<bool(const char*,size_t)> callback) {
    size_t index = 0;
    while(index < str.size()) {
        size_t end_index = str.find(delimiter, index);
        if(end_index == std::string::npos)
            end_index = str.size();

        if(!callback(&str[index], end_index - index))
            break;

        index = end_index + 1;
    }
}

static std::vector<AudioInput> parse_audio_input_arg(const char *str) {
    std::vector<AudioInput> audio_inputs;
    split_string(str, '|', [&audio_inputs](const char *sub, size_t size) {
        AudioInput audio_input;
        audio_input.name.assign(sub, size);
        const size_t index = audio_input.name.find('/');
        if(index != std::string::npos) {
            audio_input.description = audio_input.name.substr(0, index);
            audio_input.name.erase(audio_input.name.begin(), audio_input.name.begin() + index + 1);
        }
        audio_inputs.push_back(std::move(audio_input));
        return true;
    });
    return audio_inputs;
}

// TODO: Does this match all livestreaming cases?
static bool is_livestream_path(const char *str) {
    const int len = strlen(str);
    if((len >= 7 && memcmp(str, "http://", 7) == 0) || (len >= 8 && memcmp(str, "https://", 8) == 0))
        return true;
    else if((len >= 7 && memcmp(str, "rtmp://", 7) == 0) || (len >= 8 && memcmp(str, "rtmps://", 8) == 0))
        return true;
    else
        return false;
}

typedef struct {
    gpu_vendor vendor;
    int gpu_version; /* 0 if unknown */
} gpu_info;

static bool gl_get_gpu_info(Display *dpy, gpu_info *info) {
    gsr_egl gl;
    if(!gsr_egl_load(&gl, dpy)) {
        fprintf(stderr, "Error: failed to load opengl\n");
        return false;
    }

    bool supported = true;
    const unsigned char *gl_vendor = gl.glGetString(GL_VENDOR);
    const unsigned char *gl_renderer = gl.glGetString(GL_RENDERER);

    info->gpu_version = 0;

    if(!gl_vendor) {
        fprintf(stderr, "Error: failed to get gpu vendor\n");
        supported = false;
        goto end;
    }

    if(strstr((const char*)gl_vendor, "AMD"))
        info->vendor = GPU_VENDOR_AMD;
    else if(strstr((const char*)gl_vendor, "Intel"))
        info->vendor = GPU_VENDOR_INTEL;
    else if(strstr((const char*)gl_vendor, "NVIDIA"))
        info->vendor = GPU_VENDOR_NVIDIA;
    else {
        fprintf(stderr, "Error: unknown gpu vendor: %s\n", gl_vendor);
        supported = false;
        goto end;
    }

    if(gl_renderer) {
        if(info->vendor == GPU_VENDOR_NVIDIA)
            sscanf((const char*)gl_renderer, "%*s %*s %*s %d", &info->gpu_version);
    }

    end:
    gsr_egl_unload(&gl);
    return supported;
}

// TODO: Proper cleanup
static int init_filter_graph(AVCodecContext *audio_codec_context, AVFilterGraph **graph, AVFilterContext **sink, std::vector<AVFilterContext*> &src_filter_ctx, size_t num_sources)
{
    char ch_layout[64];
    int err = 0;
 
    AVFilterGraph *filter_graph = avfilter_graph_alloc();
    if (!filter_graph) {
        fprintf(stderr, "Unable to create filter graph.\n");
        return AVERROR(ENOMEM);
    }
 
    for(size_t i = 0; i < num_sources; ++i) {
        const AVFilter *abuffer = avfilter_get_by_name("abuffer");
        if (!abuffer) {
            fprintf(stderr, "Could not find the abuffer filter.\n");
            return AVERROR_FILTER_NOT_FOUND;
        }
    
        AVFilterContext *abuffer_ctx = avfilter_graph_alloc_filter(filter_graph, abuffer, NULL);
        if (!abuffer_ctx) {
            fprintf(stderr, "Could not allocate the abuffer instance.\n");
            return AVERROR(ENOMEM);
        }
    
        #if LIBAVCODEC_VERSION_MAJOR < 60
        av_get_channel_layout_string(ch_layout, sizeof(ch_layout), 0, AV_CH_LAYOUT_STEREO);
        #else
        av_channel_layout_describe(&audio_codec_context->ch_layout, ch_layout, sizeof(ch_layout));
        #endif
        av_opt_set    (abuffer_ctx, "channel_layout", ch_layout,                            AV_OPT_SEARCH_CHILDREN);
        av_opt_set    (abuffer_ctx, "sample_fmt",     av_get_sample_fmt_name(audio_codec_context->sample_fmt), AV_OPT_SEARCH_CHILDREN);
        av_opt_set_q  (abuffer_ctx, "time_base",      { 1, audio_codec_context->sample_rate },  AV_OPT_SEARCH_CHILDREN);
        av_opt_set_int(abuffer_ctx, "sample_rate",    audio_codec_context->sample_rate,                     AV_OPT_SEARCH_CHILDREN);
    
        err = avfilter_init_str(abuffer_ctx, NULL);
        if (err < 0) {
            fprintf(stderr, "Could not initialize the abuffer filter.\n");
            return err;
        }

        src_filter_ctx.push_back(abuffer_ctx);
    }

    const AVFilter *mix_filter = avfilter_get_by_name("amix");
    if (!mix_filter) {
        av_log(NULL, AV_LOG_ERROR, "Could not find the mix filter.\n");
        return AVERROR_FILTER_NOT_FOUND;
    }
    
    char args[512];
    snprintf(args, sizeof(args), "inputs=%d", (int)num_sources);
	
    AVFilterContext *mix_ctx;
	err = avfilter_graph_create_filter(&mix_ctx, mix_filter, "amix",
                                       args, NULL, filter_graph);

    if (err < 0) {
        av_log(NULL, AV_LOG_ERROR, "Cannot create audio amix filter\n");
        return err;
    }
 
    const AVFilter *abuffersink = avfilter_get_by_name("abuffersink");
    if (!abuffersink) {
        fprintf(stderr, "Could not find the abuffersink filter.\n");
        return AVERROR_FILTER_NOT_FOUND;
    }
 
    AVFilterContext *abuffersink_ctx = avfilter_graph_alloc_filter(filter_graph, abuffersink, "sink");
    if (!abuffersink_ctx) {
        fprintf(stderr, "Could not allocate the abuffersink instance.\n");
        return AVERROR(ENOMEM);
    }
 
    err = avfilter_init_str(abuffersink_ctx, NULL);
    if (err < 0) {
        fprintf(stderr, "Could not initialize the abuffersink instance.\n");
        return err;
    }
 
    err = 0;
    for(size_t i = 0; i < src_filter_ctx.size(); ++i) {
        AVFilterContext *src_ctx = src_filter_ctx[i];
        if (err >= 0)
            err = avfilter_link(src_ctx, 0, mix_ctx, i);
    }
    if (err >= 0)
        err = avfilter_link(mix_ctx, 0, abuffersink_ctx, 0);
    if (err < 0) {
        av_log(NULL, AV_LOG_ERROR, "Error connecting filters\n");
        return err;
    }
 
    err = avfilter_graph_config(filter_graph, NULL);
    if (err < 0) {
        av_log(NULL, AV_LOG_ERROR, "Error configuring the filter graph\n");
        return err;
    }
 
    *graph = filter_graph;
    *sink  = abuffersink_ctx;
 
    return 0;
}

static void xwayland_check_callback(const XRROutputInfo *output_info, const XRRCrtcInfo*, const XRRModeInfo*, void *userdata) {
    bool *xwayland_found = (bool*)userdata;
    if(output_info->nameLen >= 8 && strncmp(output_info->name, "XWAYLAND", 8) == 0)
        *xwayland_found = true;
}

static bool is_xwayland(Display *display) {
    int opcode, event, error;
    if(XQueryExtension(display, "XWAYLAND", &opcode, &event, &error))
        return true;

    bool xwayland_found = false;
    for_each_active_monitor_output(display, xwayland_check_callback, &xwayland_found);
    return xwayland_found;
}

int main(int argc, char **argv) {
    signal(SIGINT, int_handler);
    signal(SIGUSR1, save_replay_handler);

    //av_log_set_level(AV_LOG_TRACE);

    std::map<std::string, Arg> args = {
        { "-w", Arg { {}, false, false } },
        { "-c", Arg { {}, true, false } },
        { "-f", Arg { {}, false, false } },
        { "-s", Arg { {}, true, false } },
        { "-a", Arg { {}, true, true } },
        { "-q", Arg { {}, true, false } },
        { "-o", Arg { {}, true, false } },
        { "-r", Arg { {}, true, false } },
        { "-k", Arg { {}, true, false } },
        { "-ac", Arg { {}, true, false } },
        { "-oc", Arg { {}, true, false } },
        { "-pixfmt", Arg { {}, true, false } }
    };

    for(int i = 1; i < argc; i += 2) {
        auto it = args.find(argv[i]);
        if(it == args.end()) {
            fprintf(stderr, "Invalid argument '%s'\n", argv[i]);
            usage();
        }

        if(!it->second.values.empty() && !it->second.list) {
            fprintf(stderr, "Expected argument '%s' to only be specified once\n", argv[i]);
            usage();
        }

        if(i + 1 >= argc) {
            fprintf(stderr, "Missing value for argument '%s'\n", argv[i]);
            usage();
        }

        it->second.values.push_back(argv[i + 1]);
    }

    for(auto &it : args) {
        if(!it.second.optional && !it.second.value()) {
            fprintf(stderr, "Missing argument '%s'\n", it.first.c_str());
            usage();
        }
    }

    VideoCodec video_codec = VideoCodec::H265;
    const char *video_codec_to_use = args["-k"].value();
    if(!video_codec_to_use)
        video_codec_to_use = "auto";

    if(strcmp(video_codec_to_use, "h264") == 0) {
        video_codec = VideoCodec::H264;
    } else if(strcmp(video_codec_to_use, "h265") == 0) {
        video_codec = VideoCodec::H265;
    } else if(strcmp(video_codec_to_use, "auto") != 0) {
        fprintf(stderr, "Error: -k should either be either 'auto', 'h264' or 'h265', got: '%s'\n", video_codec_to_use);
        usage();
    }

    AudioCodec audio_codec = AudioCodec::OPUS;
    const char *audio_codec_to_use = args["-ac"].value();
    if(!audio_codec_to_use)
        audio_codec_to_use = "opus";

    if(strcmp(audio_codec_to_use, "aac") == 0) {
        audio_codec = AudioCodec::AAC;
    } else if(strcmp(audio_codec_to_use, "opus") == 0) {
        audio_codec = AudioCodec::OPUS;
    } else if(strcmp(audio_codec_to_use, "flac") == 0) {
        audio_codec = AudioCodec::FLAC;
    } else {
        fprintf(stderr, "Error: -ac should either be either 'aac', 'opus' or 'flac', got: '%s'\n", audio_codec_to_use);
        usage();
    }

    bool overclock = false;
    const char *overclock_str = args["-oc"].value();
    if(!overclock_str)
        overclock_str = "no";

    if(strcmp(overclock_str, "yes") == 0) {
        overclock = true;
    } else if(strcmp(overclock_str, "no") == 0) {
        overclock = false;
    } else {
        fprintf(stderr, "Error: -oc should either be either 'yes' or 'no', got: '%s'\n", overclock_str);
        usage();
    }

    PixelFormat pixel_format = PixelFormat::YUV420;
    const char *pixfmt = args["-pixfmt"].value();
    if(!pixfmt)
        pixfmt = "yuv420";

    if(strcmp(pixfmt, "yuv420") == 0) {
        pixel_format = PixelFormat::YUV420;
    } else if(strcmp(pixfmt, "yuv444") == 0) {
        pixel_format = PixelFormat::YUV444;
    } else {
        fprintf(stderr, "Error: -pixfmt should either be either 'yuv420', or 'yuv444', got: '%s'\n", pixfmt);
        usage();
    }

    const Arg &audio_input_arg = args["-a"];
    const std::vector<AudioInput> audio_inputs = get_pulseaudio_inputs();
    std::vector<MergedAudioInputs> requested_audio_inputs;

    // Manually check if the audio inputs we give exist. This is only needed for pipewire, not pulseaudio.
    // Pipewire instead DEFAULTS TO THE DEFAULT AUDIO INPUT. THAT'S RETARDED.
    // OH, YOU MISSPELLED THE AUDIO INPUT? FUCK YOU
    for(const char *audio_input : audio_input_arg.values) {
        requested_audio_inputs.push_back({parse_audio_input_arg(audio_input)});
        for(AudioInput &request_audio_input : requested_audio_inputs.back().audio_inputs) {
            bool match = false;
            for(const auto &existing_audio_input : audio_inputs) {
                if(strcmp(request_audio_input.name.c_str(), existing_audio_input.name.c_str()) == 0) {
                    if(request_audio_input.description.empty())
                        request_audio_input.description = "gsr-" + existing_audio_input.description;

                    match = true;
                    break;
                }
            }

            if(!match) {
                fprintf(stderr, "Error: Audio input device '%s' is not a valid audio device, expected one of:\n", request_audio_input.name.c_str());
                for(const auto &existing_audio_input : audio_inputs) {
                    fprintf(stderr, "    %s\n", existing_audio_input.name.c_str());
                }
                exit(2);
            }
        }
    }

    const char *container_format = args["-c"].value();
    int fps = atoi(args["-f"].value());
    if(fps == 0) {
        fprintf(stderr, "Invalid fps argument: %s\n", args["-f"].value());
        return 1;
    }
    if(fps < 1)
        fps = 1;

    const char *quality_str = args["-q"].value();
    if(!quality_str)
        quality_str = "very_high";

    VideoQuality quality;
    if(strcmp(quality_str, "medium") == 0) {
        quality = VideoQuality::MEDIUM;
    } else if(strcmp(quality_str, "high") == 0) {
        quality = VideoQuality::HIGH;
    } else if(strcmp(quality_str, "very_high") == 0) {
        quality = VideoQuality::VERY_HIGH;
    } else if(strcmp(quality_str, "ultra") == 0) {
        quality = VideoQuality::ULTRA;
    } else {
        fprintf(stderr, "Error: -q should either be either 'medium', 'high', 'very_high' or 'ultra', got: '%s'\n", quality_str);
        usage();
    }

    int replay_buffer_size_secs = -1;
    const char *replay_buffer_size_secs_str = args["-r"].value();
    if(replay_buffer_size_secs_str) {
        replay_buffer_size_secs = atoi(replay_buffer_size_secs_str);
        if(replay_buffer_size_secs < 5 || replay_buffer_size_secs > 1200) {
            fprintf(stderr, "Error: option -r has to be between 5 and 1200, was: %s\n", replay_buffer_size_secs_str);
            return 1;
        }
        replay_buffer_size_secs += 5; // Add a few seconds to account of lost packets because of non-keyframe packets skipped
    }

    Display *dpy = XOpenDisplay(nullptr);
    if (!dpy) {
        fprintf(stderr, "Error: Failed to open display. Make sure you are running x11\n");
        return 2;
    }

    XSetErrorHandler(x11_error_handler);
    XSetIOErrorHandler(x11_io_error_handler);

    if(is_xwayland(dpy)) {
        fprintf(stderr, "Error: GPU Screen Recorder only works in a pure X11 session. Xwayland is not supported\n");
        return 2;
    }

    gpu_info gpu_inf;
    bool very_old_gpu = false;
    if(!gl_get_gpu_info(dpy, &gpu_inf))
        return 2;

    if(gpu_inf.vendor == GPU_VENDOR_NVIDIA && gpu_inf.gpu_version != 0 && gpu_inf.gpu_version < 900) {
        fprintf(stderr, "Info: your gpu appears to be very old (older than maxwell architecture). Switching to lower preset\n");
        very_old_gpu = true;
    }

    // TODO: Remove once gpu screen recorder supports amd and intel properly
    if(gpu_inf.vendor != GPU_VENDOR_NVIDIA) {
        fprintf(stderr, "Error: gpu-screen-recorder does currently only support nvidia gpus\n");
        return 2;
    }

    const char *screen_region = args["-s"].value();
    const char *window_str = args["-w"].value();

    if(screen_region && strcmp(window_str, "focused") != 0) {
        fprintf(stderr, "Error: option -s is only available when using -w focused\n");
        usage();
    }

    vec2i region_size = { 0, 0 };
    Window src_window_id = None;
    bool follow_focused = false;

    gsr_capture *capture = nullptr;
    if(strcmp(window_str, "focused") == 0) {
        if(!screen_region) {
            fprintf(stderr, "Error: option -s is required when using -w focused\n");
            usage();
        }

        if(sscanf(screen_region, "%dx%d", &region_size.x, &region_size.y) != 2) {
            fprintf(stderr, "Error: invalid value for option -s '%s', expected a value in format WxH\n", screen_region);
            usage();
        }

        if(region_size.x <= 0 || region_size.y <= 0) {
            fprintf(stderr, "Error: invalud value for option -s '%s', expected width and height to be greater than 0\n", screen_region);
            usage();
        }

        follow_focused = true;
    } else if(contains_non_hex_number(window_str)) {
        if(gpu_inf.vendor != GPU_VENDOR_NVIDIA) {
            fprintf(stderr, "Error: recording a monitor is only supported on NVIDIA right now. Record \"focused\" instead for convenient fullscreen window recording\n");
            return 2;
        }

        if(strcmp(window_str, "screen") != 0 && strcmp(window_str, "screen-direct") != 0 && strcmp(window_str, "screen-direct-force") != 0) {
            gsr_monitor gmon;
            if(!get_monitor_by_name(dpy, window_str, &gmon)) {
                fprintf(stderr, "gsr error: display \"%s\" not found, expected one of:\n", window_str);
                fprintf(stderr, "    \"screen\"    (%dx%d+%d+%d)\n", XWidthOfScreen(DefaultScreenOfDisplay(dpy)), XHeightOfScreen(DefaultScreenOfDisplay(dpy)), 0, 0);
                fprintf(stderr, "    \"screen-direct\"    (%dx%d+%d+%d)\n", XWidthOfScreen(DefaultScreenOfDisplay(dpy)), XHeightOfScreen(DefaultScreenOfDisplay(dpy)), 0, 0);
                fprintf(stderr, "    \"screen-direct-force\"    (%dx%d+%d+%d)\n", XWidthOfScreen(DefaultScreenOfDisplay(dpy)), XHeightOfScreen(DefaultScreenOfDisplay(dpy)), 0, 0);
                for_each_active_monitor_output(dpy, monitor_output_callback_print, NULL);
                return 1;
            }
        }

        const char *capture_target = window_str;
        bool direct_capture = strcmp(window_str, "screen-direct") == 0;
        if(direct_capture) {
            capture_target = "screen";
            // TODO: Temporary disable direct capture because push model causes stuttering when it's direct capturing. This might be a nvfbc bug. This does not happen when using a compositor.
            direct_capture = false;
            fprintf(stderr, "Warning: screen-direct has temporary been disabled as it causes stuttering. This is likely a NvFBC bug. Falling back to \"screen\".\n");
        }

        if(strcmp(window_str, "screen-direct-force") == 0) {
            direct_capture = true;
            capture_target = "screen";
        }

        gsr_capture_nvfbc_params nvfbc_params;
        nvfbc_params.dpy = dpy;
        nvfbc_params.display_to_capture = capture_target;
        nvfbc_params.fps = fps;
        nvfbc_params.pos = { 0, 0 };
        nvfbc_params.size = { 0, 0 };
        nvfbc_params.direct_capture = direct_capture;
        nvfbc_params.overclock = overclock;
        capture = gsr_capture_nvfbc_create(&nvfbc_params);
        if(!capture)
            return 1;
    } else {
        errno = 0;
        src_window_id = strtol(window_str, nullptr, 0);
        if(src_window_id == None || errno == EINVAL) {
            fprintf(stderr, "Invalid window number %s\n", window_str);
            usage();
        }
    }

    if(!capture) {
        switch(gpu_inf.vendor) {
            case GPU_VENDOR_AMD: {
                gsr_capture_xcomposite_drm_params xcomposite_params;
                xcomposite_params.window = src_window_id;
                xcomposite_params.follow_focused = follow_focused;
                xcomposite_params.region_size = region_size;
                capture = gsr_capture_xcomposite_drm_create(&xcomposite_params);
                if(!capture)
                    return 1;
                break;
            }
            case GPU_VENDOR_INTEL: {
                gsr_capture_xcomposite_drm_params xcomposite_params;
                xcomposite_params.window = src_window_id;
                xcomposite_params.follow_focused = follow_focused;
                xcomposite_params.region_size = region_size;
                capture = gsr_capture_xcomposite_drm_create(&xcomposite_params);
                if(!capture)
                    return 1;
                break;
            }
            case GPU_VENDOR_NVIDIA: {
                gsr_capture_xcomposite_cuda_params xcomposite_params;
                xcomposite_params.window = src_window_id;
                xcomposite_params.follow_focused = follow_focused;
                xcomposite_params.region_size = region_size;
                xcomposite_params.overclock = overclock;
                capture = gsr_capture_xcomposite_cuda_create(&xcomposite_params);
                if(!capture)
                    return 1;
                break;
            }
        }
    }

    const char *filename = args["-o"].value();
    if(filename) {
        if(replay_buffer_size_secs != -1) {
            if(!container_format) {
                fprintf(stderr, "Error: option -c is required when using option -r\n");
                usage();
            }

            struct stat buf;
            if(stat(filename, &buf) == -1 || !S_ISDIR(buf.st_mode)) {
                fprintf(stderr, "Error: directory \"%s\" does not exist or is not a directory\n", filename);
                usage();
            }
        }
    } else {
        if(replay_buffer_size_secs == -1) {
            filename = "/dev/stdout";
        } else {
            fprintf(stderr, "Error: Option -o is required when using option -r\n");
            usage();
        }

        if(!container_format) {
            fprintf(stderr, "Error: option -c is required when not using option -o\n");
            usage();
        }
    }

    AVFormatContext *av_format_context;
    // The output format is automatically guessed by the file extension
    avformat_alloc_output_context2(&av_format_context, nullptr, container_format, filename);
    if (!av_format_context) {
        fprintf(stderr, "Error: Failed to deduce container format from file extension\n");
        return 1;
    }

    const AVOutputFormat *output_format = av_format_context->oformat;

    std::string file_extension = output_format->extensions;
    {
        size_t comma_index = file_extension.find(',');
        if(comma_index != std::string::npos)
            file_extension = file_extension.substr(0, comma_index);
    }

    switch(audio_codec) {
        case AudioCodec::AAC: {
            break;
        }
        case AudioCodec::OPUS: {
            if(file_extension != "mp4" && file_extension != "mkv") {
                audio_codec_to_use = "aac";
                audio_codec = AudioCodec::AAC;
                fprintf(stderr, "Warning: opus audio codec is only supported by .mp4 and .mkv files, falling back to aac instead\n");
            }
            break;
        }
        case AudioCodec::FLAC: {
            if(file_extension != "mp4" && file_extension != "mkv") {
                audio_codec_to_use = "aac";
                audio_codec = AudioCodec::AAC;
                fprintf(stderr, "Warning: flac audio codec is only supported by .mp4 and .mkv files, falling back to aac instead\n");
            }
            break;
        }
    }

    const double target_fps = 1.0 / (double)fps;

    if(strcmp(video_codec_to_use, "auto") == 0) {
        if(gpu_inf.vendor == GPU_VENDOR_NVIDIA) {
            const AVCodec *h265_codec = find_h265_encoder(gpu_inf.vendor);

            // h265 generally allows recording at a higher resolution than h264 on nvidia cards. On a gtx 1080 4k is the max resolution for h264 but for h265 it's 8k.
            // Another important info is that when recording at a higher fps than.. 60? h265 has very bad performance. For example when recording at 144 fps the fps drops to 1
            // while with h264 the fps doesn't drop.
            if(!h265_codec) {
                fprintf(stderr, "Info: using h264 encoder because a codec was not specified and your gpu does not support h265\n");
                video_codec_to_use = "h264";
                video_codec = VideoCodec::H264;
            } else if(fps > 60) {
                fprintf(stderr, "Info: using h264 encoder because a codec was not specified and fps is more than 60\n");
                video_codec_to_use = "h264";
                video_codec = VideoCodec::H264;
            } else {
                fprintf(stderr, "Info: using h265 encoder because a codec was not specified\n");
                video_codec_to_use = "h265";
                video_codec = VideoCodec::H265;
            }
        } else {
            const AVCodec *h264_codec = find_h264_encoder(gpu_inf.vendor);

            if(!h264_codec) {
                fprintf(stderr, "Info: using h265 encoder because a codec was not specified and your gpu does not support h264\n");
                video_codec_to_use = "h265";
                video_codec = VideoCodec::H265;
            //} else if(fps > 60) {
            //    fprintf(stderr, "Info: using h264 encoder because a codec was not specified and fps is more than 60\n");
            //    video_codec_to_use = "h264";
            //    video_codec = VideoCodec::H264;
            } else {
                fprintf(stderr, "Info: using h264 encoder because a codec was not specified\n");
                video_codec_to_use = "h264";
                video_codec = VideoCodec::H264;
            }
        }
    }

    //bool use_hevc = strcmp(window_str, "screen") == 0 || strcmp(window_str, "screen-direct") == 0;
    if(video_codec != VideoCodec::H264 && strcmp(file_extension.c_str(), "flv") == 0) {
        video_codec_to_use = "h264";
        video_codec = VideoCodec::H264;
        fprintf(stderr, "Warning: h265 is not compatible with flv, falling back to h264 instead.\n");
    }

    const AVCodec *video_codec_f = nullptr;
    switch(video_codec) {
        case VideoCodec::H264:
            video_codec_f = find_h264_encoder(gpu_inf.vendor);
            break;
        case VideoCodec::H265:
            video_codec_f = find_h265_encoder(gpu_inf.vendor);
            break;
    }

    if(!video_codec_f) {
        fprintf(stderr, "Error: your gpu does not support '%s' video codec\n", video_codec == VideoCodec::H264 ? "h264" : "h265");
        exit(2);
    }

    const bool is_livestream = is_livestream_path(filename);
    // (Some?) livestreaming services require at least one audio track to work.
    // If not audio is provided then create one silent audio track.
    if(is_livestream && requested_audio_inputs.empty()) {
        fprintf(stderr, "Info: live streaming but no audio track was added. Adding a silent audio track\n");
        MergedAudioInputs mai;
        mai.audio_inputs.push_back({ "", "gsr-silent" });
        requested_audio_inputs.push_back(std::move(mai));
    }

    AVStream *video_stream = nullptr;
    std::vector<AudioTrack> audio_tracks;

    AVCodecContext *video_codec_context = create_video_codec_context(gpu_inf.vendor == GPU_VENDOR_NVIDIA ? AV_PIX_FMT_CUDA : AV_PIX_FMT_VAAPI, quality, fps, video_codec_f, is_livestream, gpu_inf.vendor);
    if(replay_buffer_size_secs == -1)
        video_stream = create_stream(av_format_context, video_codec_context);

    if(gsr_capture_start(capture, video_codec_context) != 0) {
        fprintf(stderr, "gsr error: gsr_capture_start failed\n");
        return 1;
    }

    open_video(video_codec_context, quality, very_old_gpu, gpu_inf.vendor, pixel_format);
    if(video_stream)
        avcodec_parameters_from_context(video_stream->codecpar, video_codec_context);

    int audio_stream_index = VIDEO_STREAM_INDEX + 1;
    for(const MergedAudioInputs &merged_audio_inputs : requested_audio_inputs) {
        AVCodecContext *audio_codec_context = create_audio_codec_context(fps, audio_codec);

        AVStream *audio_stream = nullptr;
        if(replay_buffer_size_secs == -1)
            audio_stream = create_stream(av_format_context, audio_codec_context);

        AVFrame *audio_frame = open_audio(audio_codec_context);
        if(audio_stream)
            avcodec_parameters_from_context(audio_stream->codecpar, audio_codec_context);

        #if LIBAVCODEC_VERSION_MAJOR < 60
        const int num_channels = audio_codec_context->channels;
        #else
        const int num_channels = audio_codec_context->ch_layout.nb_channels;
        #endif

        //audio_frame->sample_rate = audio_codec_context->sample_rate;

        std::vector<AVFilterContext*> src_filter_ctx;
        AVFilterGraph *graph = nullptr;
        AVFilterContext *sink = nullptr;
        bool use_amix = merged_audio_inputs.audio_inputs.size() > 1;
        if(use_amix) {
            int err = init_filter_graph(audio_codec_context, &graph, &sink, src_filter_ctx, merged_audio_inputs.audio_inputs.size());
            if(err < 0) {
                fprintf(stderr, "Error: failed to create audio filter\n");
                exit(1);
            }
        }

        // TODO: Cleanup above

        std::vector<AudioDevice> audio_devices;
        for(size_t i = 0; i < merged_audio_inputs.audio_inputs.size(); ++i) {
            auto &audio_input = merged_audio_inputs.audio_inputs[i];
            AVFilterContext *src_ctx = nullptr;
            if(use_amix)
                src_ctx = src_filter_ctx[i];

            AudioDevice audio_device;
            audio_device.audio_input = audio_input;
            audio_device.src_filter_ctx = src_ctx;

            if(audio_input.name.empty()) {
                audio_device.sound_device.handle = NULL;
                audio_device.sound_device.frames = 0;
            } else {
                if(sound_device_get_by_name(&audio_device.sound_device, audio_input.name.c_str(), audio_input.description.c_str(), num_channels, audio_codec_context->frame_size, audio_codec_context_get_audio_format(audio_codec_context)) != 0) {
                    fprintf(stderr, "Error: failed to get \"%s\" sound device\n", audio_input.name.c_str());
                    exit(1);
                }
            }

            audio_devices.push_back(std::move(audio_device));
        }

        AudioTrack audio_track;
        audio_track.codec_context = audio_codec_context;
        audio_track.frame = audio_frame;
        audio_track.stream = audio_stream;
        audio_track.audio_devices = std::move(audio_devices);
        audio_track.graph = graph;
        audio_track.sink = sink;
        audio_track.pts = 0;
        audio_track.stream_index = audio_stream_index;
        audio_tracks.push_back(std::move(audio_track));
        ++audio_stream_index;
    }

    //av_dump_format(av_format_context, 0, filename, 1);

    if (replay_buffer_size_secs == -1 && !(output_format->flags & AVFMT_NOFILE)) {
        int ret = avio_open(&av_format_context->pb, filename, AVIO_FLAG_WRITE);
        if (ret < 0) {
            fprintf(stderr, "Error: Could not open '%s': %s\n", filename, av_error_to_string(ret));
            return 1;
        }
    }

    if(replay_buffer_size_secs == -1) {
        AVDictionary *options = nullptr;
        av_dict_set(&options, "strict", "experimental", 0);

        int ret = avformat_write_header(av_format_context, &options);
        if (ret < 0) {
            fprintf(stderr, "Error occurred when writing header to output file: %s\n", av_error_to_string(ret));
            return 1;
        }

        av_dict_free(&options);
    }

    const double start_time_pts = clock_get_monotonic_seconds();

    double start_time = clock_get_monotonic_seconds();
    double frame_timer_start = start_time;
    int fps_counter = 0;

    AVFrame *frame = av_frame_alloc();
    if (!frame) {
        fprintf(stderr, "Error: Failed to allocate frame\n");
        exit(1);
    }
    frame->format = video_codec_context->pix_fmt;
    frame->width = video_codec_context->width;
    frame->height = video_codec_context->height;
    frame->color_range = video_codec_context->color_range;
    frame->color_primaries = video_codec_context->color_primaries;
    frame->color_trc = video_codec_context->color_trc;
    frame->colorspace = video_codec_context->colorspace;

    std::mutex write_output_mutex;
    std::mutex audio_filter_mutex;

    const double record_start_time = clock_get_monotonic_seconds();
    std::deque<AVPacket> frame_data_queue;
    bool frames_erased = false;

    const size_t audio_buffer_size = 1024 * 4 * 2; // max 4 bytes/sample, 2 channels
    uint8_t *empty_audio = (uint8_t*)malloc(audio_buffer_size);
    if(!empty_audio) {
        fprintf(stderr, "Error: failed to create empty audio\n");
        exit(1);
    }
    memset(empty_audio, 0, audio_buffer_size);

    for(AudioTrack &audio_track : audio_tracks) {
        for(AudioDevice &audio_device : audio_track.audio_devices) {
            audio_device.thread = std::thread([record_start_time, replay_buffer_size_secs, &frame_data_queue, &frames_erased, &audio_track, empty_audio, &audio_device, &audio_filter_mutex, &write_output_mutex](AVFormatContext *av_format_context) mutable {
                const AVSampleFormat sound_device_sample_format = audio_format_to_sample_format(audio_codec_context_get_audio_format(audio_track.codec_context));
                const bool needs_audio_conversion = audio_track.codec_context->sample_fmt != sound_device_sample_format;
                SwrContext *swr = nullptr;
                if(needs_audio_conversion) {
                    swr = swr_alloc();
                    if(!swr) {
                        fprintf(stderr, "Failed to create SwrContext\n");
                        exit(1);
                    }
                    av_opt_set_int(swr, "in_channel_layout", AV_CH_LAYOUT_STEREO, 0);
                    av_opt_set_int(swr, "out_channel_layout", AV_CH_LAYOUT_STEREO, 0);
                    av_opt_set_int(swr, "in_sample_rate", audio_track.codec_context->sample_rate, 0);
                    av_opt_set_int(swr, "out_sample_rate", audio_track.codec_context->sample_rate, 0);
                    av_opt_set_sample_fmt(swr, "in_sample_fmt", sound_device_sample_format, 0);
                    av_opt_set_sample_fmt(swr, "out_sample_fmt", audio_track.codec_context->sample_fmt, 0);
                    swr_init(swr);
                }

                const double target_audio_hz = 1.0 / (double)audio_track.codec_context->sample_rate;
                double received_audio_time = clock_get_monotonic_seconds();
                const int64_t timeout_ms = std::round((1000.0 / (double)audio_track.codec_context->sample_rate) * 1000.0);

                while(running) {
                    void *sound_buffer;
                    int sound_buffer_size = -1;
                    if(audio_device.sound_device.handle)
                        sound_buffer_size = sound_device_read_next_chunk(&audio_device.sound_device, &sound_buffer);
                    const bool got_audio_data = sound_buffer_size >= 0;

                    const double this_audio_frame_time = clock_get_monotonic_seconds();
                    if(got_audio_data)
                        received_audio_time = this_audio_frame_time;

                    int ret = av_frame_make_writable(audio_track.frame);
                    if (ret < 0) {
                        fprintf(stderr, "Failed to make audio frame writable\n");
                        break;
                    }

                    int64_t num_missing_frames = std::round((this_audio_frame_time - received_audio_time) / target_audio_hz / (int64_t)audio_track.frame->nb_samples);
                    if(got_audio_data)
                        num_missing_frames = std::max((int64_t)0, num_missing_frames - 1);

                    if(!audio_device.sound_device.handle)
                        num_missing_frames = std::max((int64_t)1, num_missing_frames);

                    // Jesus is there a better way to do this? I JUST WANT TO KEEP VIDEO AND AUDIO SYNCED HOLY FUCK I WANT TO KILL MYSELF NOW.
                    // THIS PIECE OF SHIT WANTS EMPTY FRAMES OTHERWISE VIDEO PLAYS TOO FAST TO KEEP UP WITH AUDIO OR THE AUDIO PLAYS TOO EARLY.
                    // BUT WE CANT USE DELAYS TO GIVE DUMMY DATA BECAUSE PULSEAUDIO MIGHT GIVE AUDIO A BIG DELAYED!!!
                    // This garbage is needed because we want to produce constant frame rate videos instead of variable frame rate
                    // videos because bad software such as video editing software and VLC do not support variable frame rate software,
                    // despite nvidia shadowplay and xbox game bar producing variable frame rate videos.
                    // So we have to make sure we produce frames at the same relative rate as the video.
                    if(num_missing_frames >= 5 || !audio_device.sound_device.handle) {
                        // TODO:
                        //audio_track.frame->data[0] = empty_audio;
                        received_audio_time = this_audio_frame_time;
                        if(needs_audio_conversion)
                            swr_convert(swr, &audio_track.frame->data[0], audio_track.frame->nb_samples, (const uint8_t**)&empty_audio, audio_track.codec_context->frame_size);
                        else
                            audio_track.frame->data[0] = empty_audio;

                        // TODO: Check if duplicate frame can be saved just by writing it with a different pts instead of sending it again
                        std::lock_guard<std::mutex> lock(audio_filter_mutex);
                        for(int i = 0; i < num_missing_frames; ++i) {
                            if(audio_track.graph) {
                                // TODO: av_buffersrc_add_frame
                                if(av_buffersrc_write_frame(audio_device.src_filter_ctx, audio_track.frame) < 0) {
                                    fprintf(stderr, "Error: failed to add audio frame to filter\n");
                                }
                            } else {
                                audio_track.frame->pts = audio_track.pts;
                                audio_track.pts += audio_track.frame->nb_samples;
                                ret = avcodec_send_frame(audio_track.codec_context, audio_track.frame);
                                if(ret >= 0) {
                                    // TODO: Move to separate thread because this could write to network (for example when livestreaming)
                                    receive_frames(audio_track.codec_context, audio_track.stream_index, audio_track.stream, audio_track.frame->pts, av_format_context, record_start_time, frame_data_queue, replay_buffer_size_secs, frames_erased, write_output_mutex);
                                } else {
                                    fprintf(stderr, "Failed to encode audio!\n");
                                }
                            }
                        }
                    }

                    if(!audio_device.sound_device.handle)
                        usleep(timeout_ms * 1000);

                    if(got_audio_data) {
                        // TODO: Instead of converting audio, get float audio from alsa. Or does alsa do conversion internally to get this format?
                        if(needs_audio_conversion)
                            swr_convert(swr, &audio_track.frame->data[0], audio_track.frame->nb_samples, (const uint8_t**)&sound_buffer, audio_track.codec_context->frame_size);
                        else
                            audio_track.frame->data[0] = (uint8_t*)sound_buffer;

                        if(audio_track.graph) {
                            std::lock_guard<std::mutex> lock(audio_filter_mutex);
                            // TODO: av_buffersrc_add_frame
                            if(av_buffersrc_write_frame(audio_device.src_filter_ctx, audio_track.frame) < 0) {
                                fprintf(stderr, "Error: failed to add audio frame to filter\n");
                            }
                        } else {
                            audio_track.frame->pts = audio_track.pts;
                            audio_track.pts += audio_track.frame->nb_samples;
                            ret = avcodec_send_frame(audio_track.codec_context, audio_track.frame);
                            if(ret >= 0) {
                                // TODO: Move to separate thread because this could write to network (for example when livestreaming)
                                receive_frames(audio_track.codec_context, audio_track.stream_index, audio_track.stream, audio_track.frame->pts, av_format_context, record_start_time, frame_data_queue, replay_buffer_size_secs, frames_erased, write_output_mutex);
                            } else {
                                fprintf(stderr, "Failed to encode audio!\n");
                            }
                        }
                    }
                }

                if(swr)
                    swr_free(&swr);
            }, av_format_context);
        }
    }

    // Set update_fps to 24 to test if duplicate/delayed frames cause video/audio desync or too fast/slow video.
    const double update_fps = fps + 190;
    bool should_stop_error = false;

    AVFrame *aframe = av_frame_alloc();

    // Separate video encoding from frame capture because on amd/intel the frame capture can be very very slow
    // if we are hitting the graphical processing limit, in which case all applications will run at the same framerate
    // as the game framerate. This performance seems to be artificially limited.
    // This garbage is needed because we want to produce constant frame rate videos instead of variable frame rate
    // videos because bad software such as video editing software and VLC do not support variable frame rate software,
    // despite nvidia shadowplay and xbox game bar producing variable frame rate videos.
    // So we have to encode a frame multiple times (duplicate) if we dont produce exactly 1000/fps frames a second.
    AVFrame *latest_video_frame = nullptr;
    std::condition_variable video_frame_cv;
    std::mutex video_frame_mutex;
    std::thread video_send_encode_thread([&]() {
        int64_t video_pts_counter = 0;
        AVFrame *video_frame = nullptr;
        while(running) {
            {
                std::unique_lock<std::mutex> lock(video_frame_mutex);
                video_frame_cv.wait(lock, [&]{ return latest_video_frame || !running; });
                if(!running)
                    break;

                if(!latest_video_frame)
                    continue;

                video_frame = latest_video_frame;
                latest_video_frame = nullptr;
            }

            const double this_video_frame_time = clock_get_monotonic_seconds();
            const int64_t expected_frames = std::round((this_video_frame_time - start_time_pts) / target_fps);

            const int num_frames = std::max(0L, expected_frames - video_pts_counter);

            // TODO: Check if duplicate frame can be saved just by writing it with a different pts instead of sending it again
            for(int i = 0; i < num_frames; ++i) {
                video_frame->pts = video_pts_counter + i;
                int ret = avcodec_send_frame(video_codec_context, video_frame);
                if(ret == 0) {
                    // TODO: Move to separate thread because this could write to network (for example when livestreaming)
                    receive_frames(video_codec_context, VIDEO_STREAM_INDEX, video_stream, video_frame->pts, av_format_context,
                           record_start_time, frame_data_queue, replay_buffer_size_secs, frames_erased, write_output_mutex);
                } else {
                    fprintf(stderr, "Error: avcodec_send_frame failed, error: %s\n", av_error_to_string(ret));
                }
            }
            video_pts_counter += num_frames;

            av_frame_free(&video_frame);
            video_frame = nullptr;
        }
    });

    while(running) {
        double frame_start = clock_get_monotonic_seconds();

        gsr_capture_tick(capture, video_codec_context, &frame);
        should_stop_error = false;
        if(gsr_capture_should_stop(capture, &should_stop_error)) {
            running = 0;
            break;
        }
        ++fps_counter;

        {
            std::lock_guard<std::mutex> lock(audio_filter_mutex);
            for(AudioTrack &audio_track : audio_tracks) {
                if(!audio_track.sink)
                    continue;

                int err = 0;
                while ((err = av_buffersink_get_frame(audio_track.sink, aframe)) >= 0) {
                    aframe->pts = audio_track.pts;
                    audio_track.pts += audio_track.codec_context->frame_size;
                    err = avcodec_send_frame(audio_track.codec_context, aframe);
                    if(err >= 0){
                        // TODO: Move to separate thread because this could write to network (for example when livestreaming)
                        receive_frames(audio_track.codec_context, audio_track.stream_index, audio_track.stream, aframe->pts, av_format_context, record_start_time, frame_data_queue, replay_buffer_size_secs, frames_erased, write_output_mutex);
                    } else {
                        fprintf(stderr, "Failed to encode audio!\n");
                    }
                    av_frame_unref(aframe);
                }
            }
        }

        double time_now = clock_get_monotonic_seconds();
        double frame_timer_elapsed = time_now - frame_timer_start;
        double elapsed = time_now - start_time;
        if (elapsed >= 1.0) {
            fprintf(stderr, "update fps: %d\n", fps_counter);
            start_time = time_now;
            fps_counter = 0;
        }

        double frame_time_overflow = frame_timer_elapsed - target_fps;
        if (frame_time_overflow >= 0.0) {
            frame_timer_start = time_now - frame_time_overflow;
            gsr_capture_capture(capture, frame);
            std::lock_guard<std::mutex> lock(video_frame_mutex);
            if(latest_video_frame) {
                av_frame_free(&latest_video_frame);
                latest_video_frame = nullptr;
            }
            latest_video_frame = av_frame_clone(frame);
            video_frame_cv.notify_one();
        }

        if(save_replay_thread.valid() && save_replay_thread.wait_for(std::chrono::seconds(0)) == std::future_status::ready) {
            save_replay_thread.get();
            puts(save_replay_output_filepath.c_str());
            save_replay_packets.clear();
        }

        if(save_replay == 1 && !save_replay_thread.valid() && replay_buffer_size_secs != -1) {
            save_replay = 0;
            save_replay_async(video_codec_context, VIDEO_STREAM_INDEX, audio_tracks, frame_data_queue, frames_erased, filename, container_format, file_extension, write_output_mutex);
        }

        double frame_end = clock_get_monotonic_seconds();
        double frame_sleep_fps = 1.0 / update_fps;
        double sleep_time = frame_sleep_fps - (frame_end - frame_start);
        if(sleep_time > 0.0)
            usleep(sleep_time * 1000.0 * 1000.0);
    }

	running = 0;
    av_frame_free(&aframe);

    if(save_replay_thread.valid()) {
        save_replay_thread.get();
        puts(save_replay_output_filepath.c_str());
    }

    for(AudioTrack &audio_track : audio_tracks) {
        for(AudioDevice &audio_device : audio_track.audio_devices) {
            audio_device.thread.join();
            sound_device_close(&audio_device.sound_device);
        }
    }

    {
        std::lock_guard<std::mutex> lock(video_frame_mutex);
        video_frame_cv.notify_one();
    }
    video_send_encode_thread.join();
    //video_packet_save_thread.join();

    if(latest_video_frame) {
        av_frame_free(&latest_video_frame);
        latest_video_frame = nullptr;
    }

    if (replay_buffer_size_secs == -1 && av_write_trailer(av_format_context) != 0) {
        fprintf(stderr, "Failed to write trailer\n");
    }

    if(replay_buffer_size_secs == -1 && !(output_format->flags & AVFMT_NOFILE))
        avio_close(av_format_context->pb);

    gsr_capture_destroy(capture, video_codec_context);

    if(dpy) {
        // TODO: This causes a crash, why? maybe some other library dlclose xlib and that also happened to unload this???
        //XCloseDisplay(dpy);
    }

    free(empty_audio);
    return should_stop_error ? 3 : 0;
}