Files
dmf-studio-rnd/shared/NDIReceiver.hpp
T
JohannesItten 31b57f7a33 fix: cast NDI p_data to float* before copying audio samples
NDIlib_audio_frame_v3_t.p_data is uint8_t*, not float*. Calling
vector<float>::assign(uint8_t*, uint8_t*+N) converted each individual
byte (0-255) into a separate float instead of reinterpreting 4 bytes
as one float sample — producing complete noise on playback.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-07-01 13:08:04 +03:00

228 lines
8.9 KiB
C++

#pragma once
#include <stdexcept>
#include <string>
#include <vector>
#include <Processing.NDI.Lib.h>
#include "Signal.hpp"
#include "V210.hpp"
namespace dmf {
class NDIReceiver {
public:
struct SourceInfo {
int width = 0;
int height = 0;
int fps_num = 0;
int fps_den = 0;
int stride = 0;
NDIlib_FourCC_type_e fourcc{};
};
NDIReceiver() {
if (!NDIlib_is_supported_CPU())
throw std::runtime_error("CPU is not sufficient for NDI");
if (!NDIlib_initialize())
throw std::runtime_error("NDI lib init failed");
}
~NDIReceiver() {
if (recv_) NDIlib_recv_destroy(recv_);
if (find_) NDIlib_find_destroy(find_);
NDIlib_destroy();
}
// Discovers NDI sources on the network. Polls up to max_attempts times,
// each waiting timeout_ms milliseconds. Respects g_running.
std::vector<std::string> find_sources(uint32_t timeout_ms, int max_attempts = 10) {
find_ = NDIlib_find_create_v2();
if (!find_)
throw std::runtime_error("Cannot create NDI finder");
uint32_t count = 0;
const NDIlib_source_t* p_sources = nullptr;
for (int i = 0; !count && i < max_attempts; ++i) {
if (!dmf::g_running.load(std::memory_order_relaxed)) {
NDIlib_find_destroy(find_);
find_ = nullptr;
throw std::runtime_error("Interrupted while searching for NDI sources");
}
NDIlib_find_wait_for_sources(find_, timeout_ms);
p_sources = NDIlib_find_get_current_sources(find_, &count);
}
if (!count) {
NDIlib_find_destroy(find_);
find_ = nullptr;
throw std::runtime_error("No NDI sources found after timeout");
}
// Copy while finder is alive: p_sources points into finder-owned memory
std::vector<std::string> names;
for (uint32_t i = 0; i < count; ++i) {
names.emplace_back(p_sources[i].p_ndi_name);
source_names_.emplace_back(p_sources[i].p_ndi_name);
source_urls_.emplace_back(p_sources[i].p_url_address);
}
sources_.reserve(source_names_.size());
for (size_t i = 0; i < source_names_.size(); ++i)
sources_.push_back({source_names_[i].c_str(), source_urls_[i].c_str()});
NDIlib_find_destroy(find_);
find_ = nullptr;
return names;
}
// Connects to a discovered source by index.
void connect(uint32_t source_num) {
if (sources_.empty())
throw std::runtime_error("No sources available — call find_sources() first");
if (source_num >= static_cast<uint32_t>(sources_.size()))
throw std::runtime_error("source_num exceeds available source count");
recv_ = NDIlib_recv_create_v3();
if (!recv_)
throw std::runtime_error("Cannot create NDI receive instance");
NDIlib_recv_connect(recv_, &sources_[source_num]);
}
// Captures the first video frame to determine resolution, frame rate, and format.
// Stores the result internally for use by capture_v210(). Respects g_running.
SourceInfo probe() {
while (dmf::g_running.load(std::memory_order_relaxed)) {
NDIlib_video_frame_v2_t frame;
auto type = NDIlib_recv_capture_v3(recv_, &frame, nullptr, nullptr, 1000);
if (type == NDIlib_frame_type_error)
throw std::runtime_error("NDI source lost during probe");
if (type != NDIlib_frame_type_video)
continue;
info_.width = frame.xres;
info_.height = frame.yres;
info_.fps_num = frame.frame_rate_N;
info_.fps_den = frame.frame_rate_D;
info_.fourcc = static_cast<NDIlib_FourCC_type_e>(frame.FourCC);
info_.stride = frame.line_stride_in_bytes;
if (info_.stride == 0)
info_.stride = info_.width * bytes_per_pixel(info_.fourcc);
NDIlib_recv_free_video_v2(recv_, &frame);
return info_;
}
throw std::runtime_error("Interrupted during probe");
}
enum class FrameKind { None, Video, Audio };
struct AudioInfo {
int sample_rate = 0;
int channels = 0;
int samples = 0;
int channel_stride = 0; // floats between channel planes (NDI planar layout)
};
// Receives one NDI frame. On video: converts to V210 in frame_buffer/frame_stride.
// On audio: copies float32 planar samples into audio_out and fills audio_info.
// Returns FrameKind::None on timeout or non-A/V frames.
// Throws on source lost or video format change.
FrameKind capture(uint8_t* frame_buffer, uint32_t frame_stride,
std::vector<float>& audio_out, AudioInfo& audio_info) {
NDIlib_video_frame_v2_t video_frame{};
NDIlib_audio_frame_v3_t audio_frame{};
auto type = NDIlib_recv_capture_v3(recv_, &video_frame, &audio_frame, nullptr, 5);
if (type == NDIlib_frame_type_error)
throw std::runtime_error("NDI source lost");
if (type == NDIlib_frame_type_status_change) {
const SourceInfo old = info_;
info_ = probe();
if (info_.width != old.width || info_.height != old.height ||
info_.fps_num != old.fps_num || info_.fps_den != old.fps_den) {
throw std::runtime_error(
"source format changed: " +
std::to_string(old.width) + "x" + std::to_string(old.height) +
" @" + std::to_string(old.fps_num) + "/" + std::to_string(old.fps_den) +
" -> " +
std::to_string(info_.width) + "x" + std::to_string(info_.height) +
" @" + std::to_string(info_.fps_num) + "/" + std::to_string(info_.fps_den));
}
return FrameKind::None;
}
if (type == NDIlib_frame_type_video) {
switch (info_.fourcc) {
case NDIlib_FourCC_type_UYVY:
v210::UYVYtoV210(video_frame.p_data, frame_buffer,
info_.width, info_.height, info_.stride, frame_stride);
break;
case NDIlib_FourCC_type_P216: {
NDIlib_video_frame_v2_t dst{};
dst.p_data = frame_buffer;
dst.line_stride_in_bytes = frame_stride;
NDIlib_util_P216_to_V210(&video_frame, &dst);
break;
}
default:
NDIlib_recv_free_video_v2(recv_, &video_frame);
throw std::runtime_error("Unsupported NDI color format: " + fourcc_str(info_.fourcc));
}
NDIlib_recv_free_video_v2(recv_, &video_frame);
return FrameKind::Video;
}
if (type == NDIlib_frame_type_audio) {
audio_info.sample_rate = audio_frame.sample_rate;
audio_info.channels = audio_frame.no_channels;
audio_info.samples = audio_frame.no_samples;
audio_info.channel_stride = audio_frame.channel_stride_in_bytes / sizeof(float);
const int total = audio_frame.no_channels * audio_info.channel_stride;
const auto* fdata = reinterpret_cast<const float*>(audio_frame.p_data);
audio_out.assign(fdata, fdata + total);
NDIlib_recv_free_audio_v3(recv_, &audio_frame);
return FrameKind::Audio;
}
return FrameKind::None;
}
private:
NDIlib_find_instance_t find_ = nullptr;
NDIlib_recv_instance_t recv_ = nullptr;
SourceInfo info_;
std::vector<std::string> source_names_;
std::vector<std::string> source_urls_;
std::vector<NDIlib_source_t> sources_;
static int bytes_per_pixel(NDIlib_FourCC_type_e fc) {
switch (fc) {
case NDIlib_FourCC_video_type_UYVY:
case NDIlib_FourCC_video_type_YV12:
case NDIlib_FourCC_video_type_I420:
case NDIlib_FourCC_video_type_NV12: return 2;
case NDIlib_FourCC_video_type_BGRA:
case NDIlib_FourCC_video_type_RGBA:
case NDIlib_FourCC_video_type_BGRX:
case NDIlib_FourCC_video_type_RGBX: return 4;
case NDIlib_FourCC_video_type_UYVA: return 3;
case NDIlib_FourCC_video_type_P216: return 4;
case NDIlib_FourCC_video_type_PA16: return 6;
default: return 2;
}
}
static std::string fourcc_str(NDIlib_FourCC_type_e fc) {
uint32_t v = static_cast<uint32_t>(fc);
char s[5] = {
static_cast<char>((v >> 0) & 0xFF),
static_cast<char>((v >> 8) & 0xFF),
static_cast<char>((v >> 16) & 0xFF),
static_cast<char>((v >> 24) & 0xFF),
'\0'
};
return s;
}
};
} // namespace dmf