scar-chat7/client/media/screen_capture.cpp

932 lines
33 KiB
C++
Raw Normal View History

2025-12-07 12:00:44 -07:00
#include "screen_capture.h"
#include "video_encoder.h"
2025-12-07 12:00:44 -07:00
#include <iostream>
#include <cstring>
#include <random>
#include <sstream>
#include <iomanip>
#include <unistd.h>
#include <sys/mman.h>
#include <cerrno>
2025-12-07 12:00:44 -07:00
#ifdef __linux__
#include <sdbus-c++/sdbus-c++.h>
#include <pipewire/pipewire.h>
#include <spa/param/video/format-utils.h>
#include <spa/param/video/type-info.h>
#include <spa/param/format-utils.h>
#include <spa/param/video/format.h>
#include <spa/utils/hook.h>
#include <spa/pod/builder.h>
#include <spa/buffer/buffer.h>
#include <QImage>
#endif
2025-12-07 12:00:44 -07:00
namespace scar {
ScreenCapture::ScreenCapture()
: capturing_(false),
pw_loop_(nullptr),
pw_stream_(nullptr),
pw_context_(nullptr),
frame_width_(0),
frame_height_(0),
portal_session_ready_(false),
portal_sources_selected_(false),
portal_started_(false),
encoder_(nullptr) {
2025-12-07 12:00:44 -07:00
backend_ = detectBestBackend();
#ifdef __linux__
memset(&stream_listener_, 0, sizeof(stream_listener_));
#endif
2025-12-07 12:00:44 -07:00
}
ScreenCapture::~ScreenCapture() {
stop();
cleanupPipewire();
cleanupPortalConnection();
2025-12-07 12:00:44 -07:00
}
ScreenCaptureBackend ScreenCapture::detectBestBackend() {
#ifdef _WIN32
return ScreenCaptureBackend::FFMPEG_WINDOWS;
#else
// Check for Wayland/Hyprland
const char* waylandDisplay = std::getenv("WAYLAND_DISPLAY");
const char* hyprlandInstance = std::getenv("HYPRLAND_INSTANCE_SIGNATURE");
if (waylandDisplay || hyprlandInstance) {
std::cout << "Detected Wayland/Hyprland environment" << std::endl;
return ScreenCaptureBackend::PORTAL_PIPEWIRE;
}
// Fallback to X11
std::cout << "Detected X11 environment" << std::endl;
return ScreenCaptureBackend::FFMPEG_X11;
#endif
}
bool ScreenCapture::start() {
return start(backend_);
}
bool ScreenCapture::start(ScreenCaptureBackend backend) {
if (capturing_) {
return true;
}
backend_ = backend;
switch (backend_) {
case ScreenCaptureBackend::PORTAL_PIPEWIRE:
return startPortalCapture();
2025-12-07 12:00:44 -07:00
case ScreenCaptureBackend::FFMPEG_X11:
return startX11Capture();
case ScreenCaptureBackend::FFMPEG_WAYLAND:
return startWaylandCapture();
case ScreenCaptureBackend::FFMPEG_WINDOWS:
return startWindowsCapture();
default:
std::cerr << "Unknown backend" << std::endl;
return false;
2025-12-07 12:00:44 -07:00
}
}
void ScreenCapture::stop() {
if (!capturing_) {
return;
}
std::cout << "Stopping screen capture..." << std::endl;
cleanupPipewire();
cleanupPortalConnection();
2025-12-07 12:00:44 -07:00
capturing_ = false;
}
void ScreenCapture::setFrameCallback(FrameCallback callback) {
frameCallback_ = std::move(callback);
}
bool ScreenCapture::startX11Capture() {
std::cout << "X11 screen capture not yet implemented" << std::endl;
// TODO: Implement FFmpeg x11grab
return false;
}
bool ScreenCapture::startWaylandCapture() {
std::cout << "FFmpeg Wayland capture not yet implemented" << std::endl;
// TODO: Implement FFmpeg Wayland
return false;
}
bool ScreenCapture::startWindowsCapture() {
std::cout << "Windows screen capture not yet implemented" << std::endl;
// TODO: Implement FFmpeg GDI
return false;
}
bool ScreenCapture::startPortalCapture() {
#ifndef __linux__
std::cerr << "Portal capture only supported on Linux" << std::endl;
return false;
#else
std::cout << "Starting Portal + Pipewire screen capture..." << std::endl;
// Initialize state
portal_session_ready_ = false;
portal_sources_selected_ = false;
portal_started_ = false;
// Step 1: Initialize portal connection
if (!initPortalConnection()) {
std::cerr << "Failed to connect to xdg-desktop-portal" << std::endl;
return false;
}
// Step 2: Create portal session (async - will trigger onCreateSessionResponse)
std::string session_handle = createPortalSession();
if (session_handle.empty()) {
std::cerr << "Failed to create portal session" << std::endl;
cleanupPortalConnection();
return false;
}
session_path_ = session_handle;
2025-12-07 12:00:44 -07:00
// The rest of the workflow continues in async callbacks
// We return true here as the request was initiated successfully
std::cout << "Portal workflow initiated, waiting for user interaction..." << std::endl;
2025-12-07 12:00:44 -07:00
return true;
#endif
2025-12-07 12:00:44 -07:00
}
#ifdef __linux__
bool ScreenCapture::initPortalConnection() {
try {
// Create session bus connection with event loop processing
portal_connection_ = sdbus::createSessionBusConnection();
// Enter event loop in async mode to process signals
portal_connection_->enterEventLoopAsync();
// Create proxy for ScreenCast portal interface
// Using standard org.freedesktop.portal.Desktop interface
// xdg-desktop-portal-hyprland implements this standard interface
screencast_proxy_ = sdbus::createProxy(
*portal_connection_,
sdbus::ServiceName{"org.freedesktop.portal.Desktop"},
sdbus::ObjectPath{"/org/freedesktop/portal/desktop"}
);
std::cout << "Connected to xdg-desktop-portal (Hyprland implementation)" << std::endl;
return true;
}
catch (const sdbus::Error& e) {
std::cerr << "Failed to connect to portal: " << e.what() << std::endl;
return false;
}
}
void ScreenCapture::cleanupPortalConnection() {
request_proxies_.clear(); // Clear signal handlers
screencast_proxy_.reset();
portal_connection_.reset();
}
std::string ScreenCapture::createPortalSession() {
try {
// Generate unique handle for this request
std::random_device rd;
std::mt19937 gen(rd());
std::uniform_int_distribution<> dis(0, 999999);
std::stringstream ss;
ss << "/org/freedesktop/portal/desktop/request/"
<< getpid() << "_" << dis(gen);
std::string request_handle = ss.str();
ss.str("");
ss << "/org/freedesktop/portal/desktop/session/"
<< getpid() << "_" << dis(gen);
std::string session_handle = ss.str();
std::cout << "Creating portal session..." << std::endl;
std::cout << " Predicted request handle: " << request_handle << std::endl;
std::cout << " Session handle: " << session_handle << std::endl;
// Build options dictionary
std::map<std::string, sdbus::Variant> options;
options["handle_token"] = sdbus::Variant{request_handle.substr(request_handle.rfind('/') + 1)};
options["session_handle_token"] = sdbus::Variant{session_handle.substr(session_handle.rfind('/') + 1)};
// Call CreateSession
sdbus::ObjectPath response_path;
screencast_proxy_->callMethod("CreateSession")
.onInterface("org.freedesktop.portal.ScreenCast")
.withArguments(options)
.storeResultsTo(response_path);
std::cout << "CreateSession returned request path: " << response_path << std::endl;
std::cout << " Expected path: " << request_handle << std::endl;
// Setup signal handler with the actual path returned by portal
setupResponseSignal(response_path, [this](uint32_t response, const std::map<std::string, sdbus::Variant>& results) {
onCreateSessionResponse(response, results);
});
std::cout << "CreateSession request sent, waiting for response..." << std::endl;
return session_handle;
}
catch (const sdbus::Error& e) {
std::cerr << "Failed to create portal session: " << e.what() << std::endl;
return "";
}
}
bool ScreenCapture::selectPortalSources(const std::string& session_handle) {
try {
// Generate unique request handle
std::random_device rd;
std::mt19937 gen(rd());
std::uniform_int_distribution<> dis(0, 999999);
std::stringstream ss;
ss << "/org/freedesktop/portal/desktop/request/"
<< getpid() << "_" << dis(gen);
std::string request_handle = ss.str();
std::cout << "Selecting portal sources..." << std::endl;
// Build options dictionary
std::map<std::string, sdbus::Variant> options;
options["handle_token"] = sdbus::Variant{request_handle.substr(request_handle.rfind('/') + 1)};
options["types"] = sdbus::Variant{uint32_t(1 | 2)}; // MONITOR (1) | WINDOW (2)
options["multiple"] = sdbus::Variant{false};
options["cursor_mode"] = sdbus::Variant{uint32_t(2)}; // Embedded cursor
// Call SelectSources
sdbus::ObjectPath response_path;
screencast_proxy_->callMethod("SelectSources")
.onInterface("org.freedesktop.portal.ScreenCast")
.withArguments(sdbus::ObjectPath{session_handle}, options)
.storeResultsTo(response_path);
std::cout << "SelectSources returned request path: " << response_path << std::endl;
// Set up Response signal handler with actual path
setupResponseSignal(response_path,
[this](uint32_t response, const std::map<std::string, sdbus::Variant>& results) {
onSelectSourcesResponse(response, results);
});
std::cout << "SelectSources request sent (user will see hyprland-share-picker)..." << std::endl;
return true;
}
catch (const sdbus::Error& e) {
std::cerr << "Failed to select portal sources: " << e.what() << std::endl;
return false;
}
}
bool ScreenCapture::startPortalSession(const std::string& session_handle, uint32_t& node_id) {
try {
// Generate unique request handle
std::random_device rd;
std::mt19937 gen(rd());
std::uniform_int_distribution<> dis(0, 999999);
std::stringstream ss;
ss << "/org/freedesktop/portal/desktop/request/"
<< getpid() << "_" << dis(gen);
std::string request_handle = ss.str();
std::cout << "Starting portal session..." << std::endl;
// Build options dictionary
std::map<std::string, sdbus::Variant> options;
options["handle_token"] = sdbus::Variant{request_handle.substr(request_handle.rfind('/') + 1)};
// Call Start
sdbus::ObjectPath response_path;
screencast_proxy_->callMethod("Start")
.onInterface("org.freedesktop.portal.ScreenCast")
.withArguments(sdbus::ObjectPath{session_handle}, std::string(""), options)
.storeResultsTo(response_path);
std::cout << "Start returned request path: " << response_path << std::endl;
// Set up Response signal handler with actual path
setupResponseSignal(response_path,
[this](uint32_t response, const std::map<std::string, sdbus::Variant>& results) {
onStartSessionResponse(response, results);
});
std::cout << "Start request sent, waiting for response..." << std::endl;
node_id = 0; // Will be set in response handler
return true;
}
catch (const sdbus::Error& e) {
std::cerr << "Failed to start portal session: " << e.what() << std::endl;
return false;
}
}
int ScreenCapture::openPipeWireRemote(const std::string& session_handle) {
try {
std::cout << "Opening PipeWire remote..." << std::endl;
// Build options dictionary (empty for now)
std::map<std::string, sdbus::Variant> options;
// Call OpenPipeWireRemote
sdbus::UnixFd pipewire_fd;
screencast_proxy_->callMethod("OpenPipeWireRemote")
.onInterface("org.freedesktop.portal.ScreenCast")
.withArguments(sdbus::ObjectPath{session_handle}, options)
.storeResultsTo(pipewire_fd);
// Extract the actual file descriptor from UnixFd
int fd = dup(pipewire_fd.get());
if (fd < 0) {
std::cerr << "Failed to duplicate PipeWire file descriptor" << std::endl;
return -1;
}
std::cout << "PipeWire remote opened, fd: " << fd << std::endl;
return fd;
}
catch (const sdbus::Error& e) {
std::cerr << "Failed to open PipeWire remote: " << e.what() << std::endl;
return -1;
}
}
uint32_t ScreenCapture::getStreamsNodeId(const std::string& session_handle) {
try {
std::cout << "Getting streams from session..." << std::endl;
// Create a proxy for the session object
auto session_proxy = sdbus::createProxy(
*portal_connection_,
sdbus::ServiceName{"org.freedesktop.portal.Desktop"},
sdbus::ObjectPath{session_handle}
);
// Get the Streams property from org.freedesktop.portal.Session interface
auto streams_variant = session_proxy->getProperty("Streams")
.onInterface("org.freedesktop.portal.Session");
// Streams is a(ua{sv}) - array of (node_id, properties)
auto streams = streams_variant.get<std::vector<sdbus::Struct<uint32_t, std::map<std::string, sdbus::Variant>>>>();
if (streams.empty()) {
std::cerr << "No streams available in session" << std::endl;
return 0;
}
// Get the first stream's node_id
uint32_t node_id = std::get<0>(streams[0]);
auto& properties = std::get<1>(streams[0]);
std::cout << "Found stream node_id: " << node_id << std::endl;
// Print stream properties if available
if (properties.count("size")) {
auto size = properties["size"].get<sdbus::Struct<int32_t, int32_t>>();
std::cout << " Stream size: " << std::get<0>(size) << "x" << std::get<1>(size) << std::endl;
}
return node_id;
}
catch (const sdbus::Error& e) {
std::cerr << "Failed to get streams: " << e.what() << std::endl;
return 0;
}
}
void ScreenCapture::setupResponseSignal(const std::string& request_handle,
std::function<void(uint32_t, const std::map<std::string, sdbus::Variant>&)> callback) {
try {
// Create a proxy for the Request object to receive the Response signal
auto request_proxy = sdbus::createProxy(
*portal_connection_,
sdbus::ServiceName{"org.freedesktop.portal.Desktop"},
sdbus::ObjectPath{request_handle}
);
// Register signal handler for Response
// Capture callback by value to extend its lifetime
request_proxy->uponSignal("Response")
.onInterface("org.freedesktop.portal.Request")
.call([this, callback](uint32_t response, std::map<std::string, sdbus::Variant> results) {
std::cout << "Response signal received!" << std::endl;
callback(response, results);
});
// Store the proxy to keep signal handler alive
request_proxies_.push_back(std::move(request_proxy));
std::cout << "Signal handler registered for: " << request_handle << std::endl;
}
catch (const sdbus::Error& e) {
std::cerr << "Failed to setup response signal: " << e.what() << std::endl;
}
}
void ScreenCapture::onCreateSessionResponse(uint32_t response, const std::map<std::string, sdbus::Variant>& results) {
std::cout << "CreateSession Response received: " << response << std::endl;
if (response != 0) {
std::cerr << "CreateSession failed or was cancelled by user" << std::endl;
cleanupPortalConnection();
return;
}
// Extract the actual session handle from the response
std::cout << "Response contains " << results.size() << " fields:" << std::endl;
for (const auto& [key, value] : results) {
std::cout << " Field: " << key << ", ContainsValueOfType: " << value.peekValueType() << std::endl;
}
// The session handle should be in the results
try {
if (results.find("session_handle") != results.end()) {
const auto& variant = results.at("session_handle");
std::cout << "Attempting to extract session_handle..." << std::endl;
std::cout << " Variant type signature: " << variant.peekValueType() << std::endl;
// Try as ObjectPath
try {
auto obj_path = variant.get<sdbus::ObjectPath>();
session_path_ = obj_path;
std::cout << " Extracted as ObjectPath: " << session_path_ << std::endl;
} catch (...) {
// Try as string
std::cout << " ObjectPath extraction failed, trying as string..." << std::endl;
auto str_path = variant.get<std::string>();
session_path_ = str_path;
std::cout << " Extracted as string: " << session_path_ << std::endl;
}
} else {
std::cerr << "No session_handle in response!" << std::endl;
cleanupPortalConnection();
return;
}
} catch (const std::exception& e) {
std::cerr << "Failed to extract session_handle: " << e.what() << std::endl;
std::cerr << "Exception type: " << typeid(e).name() << std::endl;
cleanupPortalConnection();
return;
}
portal_session_ready_ = true;
std::cout << "Session ready, proceeding to SelectSources with path: " << session_path_ << std::endl;
// Now we can call SelectSources
if (!selectPortalSources(session_path_)) {
std::cerr << "Failed to call SelectSources" << std::endl;
cleanupPortalConnection();
}
}
void ScreenCapture::onSelectSourcesResponse(uint32_t response, const std::map<std::string, sdbus::Variant>& results) {
std::cout << "SelectSources Response received: " << response << std::endl;
if (response != 0) {
std::cerr << "SelectSources failed or was cancelled by user" << std::endl;
cleanupPortalConnection();
return;
}
portal_sources_selected_ = true;
std::cout << "Sources selected, proceeding to Start..." << std::endl;
// Now we can call Start
uint32_t node_id = 0;
if (!startPortalSession(session_path_, node_id)) {
std::cerr << "Failed to call Start" << std::endl;
cleanupPortalConnection();
}
}
void ScreenCapture::onStartSessionResponse(uint32_t response, const std::map<std::string, sdbus::Variant>& results) {
std::cout << "Start Response received: " << response << std::endl;
if (response != 0) {
std::cerr << "Start failed or was cancelled by user" << std::endl;
cleanupPortalConnection();
return;
}
portal_started_ = true;
std::cout << "Session started, extracting streams from response..." << std::endl;
// List all fields in response
std::cout << "Start Response contains " << results.size() << " fields:" << std::endl;
for (const auto& [key, value] : results) {
std::cout << " Field: " << key << ", Type: " << value.peekValueType() << std::endl;
}
uint32_t node_id = 0;
try {
// Extract streams from the response - format is a(ua{sv})
if (results.find("streams") != results.end()) {
const auto& streams_variant = results.at("streams");
auto streams = streams_variant.get<std::vector<sdbus::Struct<uint32_t, std::map<std::string, sdbus::Variant>>>>();
if (streams.empty()) {
std::cerr << "No streams in Start response" << std::endl;
cleanupPortalConnection();
return;
}
// Get the first stream's node_id
node_id = std::get<0>(streams[0]);
std::cout << "Extracted PipeWire node ID from response: " << node_id << std::endl;
} else {
std::cerr << "No 'streams' field in Start response" << std::endl;
cleanupPortalConnection();
return;
}
} catch (const std::exception& e) {
std::cerr << "Failed to extract streams: " << e.what() << std::endl;
cleanupPortalConnection();
return;
}
if (node_id == 0) {
std::cerr << "Invalid node_id" << std::endl;
cleanupPortalConnection();
return;
}
std::cout << "Got Pipewire node ID: " << node_id << std::endl;
// Open PipeWire remote
int pw_fd = openPipeWireRemote(session_path_);
if (pw_fd < 0) {
std::cerr << "Failed to open PipeWire remote" << std::endl;
cleanupPortalConnection();
return;
}
// Initialize Pipewire and connect to stream
if (!initPipewire(pw_fd, node_id)) {
std::cerr << "Failed to initialize Pipewire" << std::endl;
cleanupPortalConnection();
return;
}
capturing_ = true;
std::cout << "Screen capture started successfully!" << std::endl;
}
bool ScreenCapture::initPipewire(int fd, uint32_t node_id) {
pw_init(nullptr, nullptr);
2025-12-07 12:00:44 -07:00
std::cout << "Initializing PipeWire with fd: " << fd << ", node_id: " << node_id << std::endl;
2025-12-07 12:00:44 -07:00
pw_loop_ = pw_thread_loop_new("screen-capture", nullptr);
if (!pw_loop_) {
std::cerr << "Failed to create Pipewire loop" << std::endl;
return false;
}
2025-12-07 12:00:44 -07:00
pw_context_ = pw_context_new(
pw_thread_loop_get_loop(pw_loop_),
nullptr, 0
);
if (!pw_context_) {
std::cerr << "Failed to create Pipewire context" << std::endl;
pw_thread_loop_destroy(pw_loop_);
pw_loop_ = nullptr;
return false;
}
// Connect to the PipeWire remote using the file descriptor
pw_core* core = pw_context_connect_fd(
pw_context_,
fd,
nullptr, 0
);
if (!core) {
std::cerr << "Failed to connect to PipeWire remote" << std::endl;
cleanupPipewire();
return false;
}
// Create stream properties
pw_properties* props = pw_properties_new(
PW_KEY_MEDIA_TYPE, "Video",
PW_KEY_MEDIA_CATEGORY, "Capture",
PW_KEY_MEDIA_ROLE, "Screen",
nullptr
);
// Create the stream
pw_stream_ = pw_stream_new(
core,
"screen-capture-stream",
props
);
if (!pw_stream_) {
std::cerr << "Failed to create PipeWire stream" << std::endl;
pw_core_disconnect(core);
cleanupPipewire();
return false;
}
// Set up stream events
static const pw_stream_events stream_events = {
PW_VERSION_STREAM_EVENTS,
nullptr, // destroy
nullptr, // state_changed
nullptr, // control_info
nullptr, // io_changed
onStreamParamChanged, // param_changed
nullptr, // add_buffer
nullptr, // remove_buffer
onStreamProcess, // process
nullptr, // drained
nullptr, // command
nullptr, // trigger_done
};
pw_stream_add_listener(
pw_stream_,
&stream_listener_,
&stream_events,
this // userdata pointer
);
// Lock before connecting
pw_thread_loop_lock(pw_loop_);
// Connect to the node
uint8_t buffer[1024];
spa_pod_builder pod_builder = SPA_POD_BUILDER_INIT(buffer, sizeof(buffer));
// Request any video format (we'll adapt to what we get)
const struct spa_pod* params[1];
params[0] = reinterpret_cast<const struct spa_pod*>(spa_pod_builder_add_object(
&pod_builder,
SPA_TYPE_OBJECT_Format, SPA_PARAM_EnumFormat,
SPA_FORMAT_mediaType, SPA_POD_Id(SPA_MEDIA_TYPE_video),
SPA_FORMAT_mediaSubtype, SPA_POD_Id(SPA_MEDIA_SUBTYPE_raw)
));
int result = pw_stream_connect(
pw_stream_,
PW_DIRECTION_INPUT,
node_id,
PW_STREAM_FLAG_AUTOCONNECT, // Removed MAP_BUFFERS flag
params, 1
);
if (result < 0) {
std::cerr << "Failed to connect stream to node " << node_id << ": " << strerror(-result) << std::endl;
pw_thread_loop_unlock(pw_loop_);
pw_core_disconnect(core);
cleanupPipewire();
return false;
}
// Unlock after connecting
pw_thread_loop_unlock(pw_loop_);
// Start the thread loop
if (pw_thread_loop_start(pw_loop_) < 0) {
std::cerr << "Failed to start PipeWire thread loop" << std::endl;
pw_core_disconnect(core);
cleanupPipewire();
return false;
}
std::cout << "PipeWire stream connected and running" << std::endl;
2025-12-07 12:00:44 -07:00
return true;
}
void ScreenCapture::cleanupPipewire() {
if (pw_loop_) {
pw_thread_loop_stop(pw_loop_);
}
2025-12-07 12:00:44 -07:00
if (pw_stream_) {
pw_stream_destroy(pw_stream_);
pw_stream_ = nullptr;
}
2025-12-07 12:00:44 -07:00
if (pw_context_) {
pw_context_destroy(pw_context_);
pw_context_ = nullptr;
}
if (pw_loop_) {
pw_thread_loop_destroy(pw_loop_);
pw_loop_ = nullptr;
}
pw_deinit();
2025-12-07 12:00:44 -07:00
}
void ScreenCapture::onStreamProcess(void* userdata) {
std::cout << "onStreamProcess called" << std::endl;
auto* self = static_cast<ScreenCapture*>(userdata);
std::cout << " userdata cast successful" << std::endl;
try {
pw_buffer* buffer = pw_stream_dequeue_buffer(self->pw_stream_);
std::cout << " dequeue_buffer returned: " << (void*)buffer << std::endl;
if (!buffer) {
std::cerr << "No buffer available" << std::endl;
return;
}
spa_buffer* spa_buf = buffer->buffer;
std::cout << " spa_buf: " << (void*)spa_buf << std::endl;
if (!spa_buf || !spa_buf->datas) {
std::cout << " Invalid spa_buf structure, requeueing" << std::endl;
pw_stream_queue_buffer(self->pw_stream_, buffer);
return;
}
std::cout << " spa_buf valid, getting frame data" << std::endl;
// Get frame data
spa_data& data = spa_buf->datas[0];
std::cout << " Buffer type: " << data.type << " (MemFd=" << SPA_DATA_MemFd
<< ", MemPtr=" << SPA_DATA_MemPtr << ", DmaBuf=" << SPA_DATA_DmaBuf << ")" << std::endl;
std::cout << " data.data: " << data.data << ", data.fd: " << data.fd
<< ", data.maxsize: " << data.maxsize << std::endl;
// ALWAYS use mmap for DMA-BUF and MemFd types (ignore data.data pointer)
void* frame_data = nullptr;
bool needs_unmap = false;
if (data.type == SPA_DATA_MemFd || data.type == SPA_DATA_DmaBuf) {
// Map the file descriptor
std::cout << " Mapping buffer from fd..." << std::endl;
frame_data = mmap(NULL, data.maxsize, PROT_READ, MAP_PRIVATE, data.fd, data.mapoffset);
if (frame_data == MAP_FAILED) {
std::cerr << " Failed to mmap buffer, errno: " << errno << std::endl;
pw_stream_queue_buffer(self->pw_stream_, buffer);
return;
}
needs_unmap = true;
std::cout << " Successfully mapped buffer at " << frame_data << std::endl;
} else if (data.type == SPA_DATA_MemPtr) {
// Use the pointer directly
frame_data = data.data;
std::cout << " Using direct pointer" << std::endl;
}
if (!frame_data) {
std::cout << " No frame data available, requeueing" << std::endl;
pw_stream_queue_buffer(self->pw_stream_, buffer);
return;
}
// Validate chunk
if (!data.chunk) {
std::cerr << "No chunk in buffer" << std::endl;
if (needs_unmap) {
munmap(frame_data, data.maxsize);
}
pw_stream_queue_buffer(self->pw_stream_, buffer);
return;
}
// Apply chunk offset to get actual frame data
uint8_t* frame_ptr = static_cast<uint8_t*>(frame_data) + data.chunk->offset;
uint32_t size = data.chunk->size;
std::cout << " chunk->offset: " << data.chunk->offset << ", chunk->size: " << size << std::endl;
// Validate that chunk doesn't exceed mapped buffer
if (data.chunk->offset + size > data.maxsize) {
std::cerr << "Chunk exceeds buffer bounds: offset=" << data.chunk->offset
<< " size=" << size << " maxsize=" << data.maxsize << std::endl;
if (needs_unmap) {
munmap(frame_data, data.maxsize);
}
pw_stream_queue_buffer(self->pw_stream_, buffer);
return;
}
// Validate data pointer and size
if (!frame_ptr || size == 0 || size > 100 * 1024 * 1024) { // Max 100MB sanity check
std::cerr << "Invalid frame data: ptr=" << (void*)frame_ptr << " size=" << size << std::endl;
if (needs_unmap) {
munmap(frame_data, data.maxsize);
}
pw_stream_queue_buffer(self->pw_stream_, buffer);
return;
}
static int frame_count = 0;
if (++frame_count % 30 == 0) {
std::cout << "Frame received: " << frame_count << " (size: " << size << " bytes)" << std::endl;
}
// Encode and send frame if encoder is available
if (self->encoder_ && self->frameCallback_) {
try {
// Encode the raw frame to H.264
std::vector<uint8_t> encoded_frame = self->encoder_->encode(frame_ptr, size);
if (!encoded_frame.empty()) {
// Send encoded frame via callback
self->frameCallback_(encoded_frame, self->frame_width_, self->frame_height_);
if (frame_count % 30 == 0) {
std::cout << " Encoded frame size: " << encoded_frame.size() << " bytes (compression: "
<< (100.0f * encoded_frame.size() / size) << "%)" << std::endl;
}
}
} catch (const std::exception& e) {
std::cerr << "Exception encoding/sending frame: " << e.what() << std::endl;
}
} else {
if (frame_count < 5) {
if (!self->encoder_) {
std::cout << " Encoder not initialized yet" << std::endl;
} else if (!self->frameCallback_) {
std::cout << " Frame callback not set" << std::endl;
}
}
}
// Clean up mapped memory if we mapped it ourselves
if (needs_unmap) {
munmap(frame_data, data.maxsize);
}
// Return buffer to stream
pw_stream_queue_buffer(self->pw_stream_, buffer);
} catch (const std::exception& e) {
std::cerr << "Exception in onStreamProcess: " << e.what() << std::endl;
} catch (...) {
std::cerr << "Unknown exception in onStreamProcess" << std::endl;
}
}
void ScreenCapture::onStreamParamChanged(void* userdata, uint32_t id, const struct spa_pod* param) {
auto* self = static_cast<ScreenCapture*>(userdata);
if (!param || id != SPA_PARAM_Format) {
return;
}
// Parse video format
spa_video_info_raw video_info;
if (spa_format_video_raw_parse(param, &video_info) < 0) {
std::cerr << "Failed to parse video format" << std::endl;
return;
}
// Update our stored dimensions
self->frame_width_ = video_info.size.width;
self->frame_height_ = video_info.size.height;
std::cout << "Stream format changed:" << std::endl;
std::cout << " Resolution: " << self->frame_width_ << "x" << self->frame_height_ << std::endl;
std::cout << " Format: " << video_info.format << std::endl;
std::cout << " Framerate: " << video_info.framerate.num << "/" << video_info.framerate.denom << std::endl;
// Initialize video encoder now that we know the resolution
int fps = (video_info.framerate.denom > 0) ? (video_info.framerate.num / video_info.framerate.denom) : 30;
if (fps <= 0) fps = 30;
self->encoder_ = std::make_unique<VideoEncoder>();
if (!self->encoder_->initialize(self->frame_width_, self->frame_height_, fps)) {
std::cerr << "Failed to initialize video encoder" << std::endl;
self->encoder_.reset();
}
2025-12-07 12:00:44 -07:00
}
#else
// Stub implementations for non-Linux platforms
bool ScreenCapture::initPortalConnection() { return false; }
void ScreenCapture::cleanupPortalConnection() {}
std::string ScreenCapture::createPortalSession() { return ""; }
bool ScreenCapture::selectPortalSources(const std::string&) { return false; }
bool ScreenCapture::startPortalSession(const std::string&, uint32_t&) { return false; }
int ScreenCapture::openPipeWireRemote(const std::string&) { return -1; }
uint32_t ScreenCapture::getStreamsNodeId(const std::string&) { return 0; }
void ScreenCapture::setupResponseSignal(const std::string&, std::function<void(uint32_t, const std::map<std::string, sdbus::Variant>&)>) {}
void ScreenCapture::onCreateSessionResponse(uint32_t, const std::map<std::string, sdbus::Variant>&) {}
void ScreenCapture::onSelectSourcesResponse(uint32_t, const std::map<std::string, sdbus::Variant>&) {}
void ScreenCapture::onStartSessionResponse(uint32_t, const std::map<std::string, sdbus::Variant>&) {}
bool ScreenCapture::initPipewire(int, uint32_t) { return false; }
void ScreenCapture::cleanupPipewire() {}
void ScreenCapture::onStreamProcess(void*) {}
void ScreenCapture::onStreamParamChanged(void*, uint32_t, const struct spa_pod*) {}
#endif
2025-12-07 12:00:44 -07:00
} // namespace scar