Refactor build script for Linux and MacOS in CI workflow

cmake/BuildSDL.cmake

@@ -18,7 +18,7 @@ ExternalProject_Add(
         -DSDL_STATIC=ON
         -DSDL_SHARED=OFF
     BUILD_BYPRODUCTS
-        "${SDL_INSTALL_DIR}/lib/${CMAKE_STATIC_LIBRARY_PREFIX}SDL2${CMAKE_STATIC_LIBRARY_SUFFIX}"
+        "${SDL_INSTALL_DIR}/lib/${CMAKE_STATIC_LIBRARY_PREFIX}SDL2-static${CMAKE_STATIC_LIBRARY_SUFFIX}"
         "${SDL_INSTALL_DIR}/lib/${CMAKE_STATIC_LIBRARY_PREFIX}SDL2main${CMAKE_STATIC_LIBRARY_SUFFIX}"
 )
 

examples/audio_capture.h

@@ -1,33 +1,35 @@
 #pragma once
 
+#define SDL_MAIN_HANDLED
 #include <SDL.h>
+
 #include <atomic>
 #include <cstdint>
 #include <vector>
 #include <mutex>
 
 class AudioCapture {
 public:
-    AudioCapture(int buffer_duration_ms);
-    ~AudioCapture();
+	AudioCapture(int buffer_duration_ms);
+	~AudioCapture();
 
-    bool initialize(int device_index, int sample_rate);
-    bool startCapture();
-    bool stopCapture();
-    bool resetBuffer();
+	bool initialize(int device_index, int sample_rate);
+	bool startCapture();
+	bool stopCapture();
+	bool resetBuffer();
 
-    void processAudio(uint8_t* stream, int length);
-    void getAudioData(int duration_ms, std::vector<float>& output);
+	void processAudio(uint8_t *stream, int length);
+	void getAudioData(int duration_ms, std::vector<float> &output);
 
 private:
-    SDL_AudioDeviceID device_id = 0;
-    int buffer_duration_ms = 0;
-    int sample_rate = 0;
-    std::atomic_bool is_capturing;
-    std::mutex buffer_mutex;
-    std::vector<float> audio_buffer;
-    size_t write_position = 0;
-    size_t buffer_size = 0;
+	SDL_AudioDeviceID device_id = 0;
+	int buffer_duration_ms = 0;
+	int sample_rate = 0;
+	std::atomic_bool is_capturing;
+	std::mutex buffer_mutex;
+	std::vector<float> audio_buffer;
+	size_t write_position = 0;
+	size_t buffer_size = 0;
 };
 
 bool handleSDLEvents();

examples/realtime_transcription.cpp

@@ -3,55 +3,65 @@
 
 #include <iostream>
 
-int main() {
-    // Initialize the library
-    locaal::Transcription tt;
-
-    // Set the transcription parameters (language, model, etc.)
-    tt.setTranscriptionParams("en-US");
-
-    tt.setModelDownloadCallbacks([](const std::string &model_name, const std::string &model_path) {
-        std::cout << "Model downloaded: " << model_name << " at " << model_path << std::endl;
-    }, [](const std::string &model_name, const std::string &model_path) {
-        std::cout << "Model download failed: " << model_name << " at " << model_path << std::endl;
-    }, [](const std::string &model_name, const std::string &model_path) {
-        std::cout << "Model download progress: " << model_name << " at " << model_path << std::endl;
-    });
-
-    // Set the callbacks for the transcription
-    tt.setTranscriptionCallback([](const locaal::TranscriptionResult &result) {
-        // Print the transcription result
-        std::cout << "Transcription: " << result.text << std::endl;
-    });
-
-    // Start real-time transcription background thread
-    tt.startTranscription();
-
-    // Start capturing audio from the microphone
-    AudioCapture audio_capture(1000);
-    if (!audio_capture.initialize(0, 16000)) {
-        std::cerr << "Failed to initialize audio capture" << std::endl;
-        return 1;
-    }
-    if (!audio_capture.startCapture()) {
-        std::cerr << "Failed to start audio capture" << std::endl;
-        return 1;
-    }
-
-    // Main loop
-    while (true) {
-        // Handle SDL events
-        if (!handleSDLEvents()) {
-            break;
-        }
-
-        // Get audio data from the audio capture
-        std::vector<float> audio_data;
-        audio_capture.getAudioData(1000, audio_data);
-
-        // Process the audio data for transcription
-        tt.processAudio(audio_data);
-    }
-
-    return 0;
+int main()
+{
+	SDL_SetMainReady();
+
+	// Initialize the library
+	locaal::Transcription tt;
+
+	// Set the transcription parameters (language, model, etc.)
+	tt.setTranscriptionParams("en-US");
+
+	tt.setModelDownloadCallbacks(
+		[](const std::string &model_name, const std::string &model_path) {
+			std::cout << "Model downloaded: " << model_name << " at " << model_path
+				  << std::endl;
+		},
+		[](const std::string &model_name, const std::string &model_path) {
+			std::cout << "Model download failed: " << model_name << " at " << model_path
+				  << std::endl;
+		},
+		[](const std::string &model_name, const std::string &model_path) {
+			std::cout << "Model download progress: " << model_name << " at "
+				  << model_path << std::endl;
+		});
+
+	// Set the callbacks for the transcription
+	tt.setTranscriptionCallback([](const locaal::TranscriptionResult &result) {
+		// Print the transcription result
+		std::cout << "Transcription:" << (result.is_partial ? " (partial) " : " ")
+			  << result.text << std::endl;
+	});
+
+	// Start real-time transcription background thread
+	tt.startTranscription();
+
+	// Start capturing audio from the microphone
+	AudioCapture audio_capture(1000);
+	if (!audio_capture.initialize(0, 16000)) {
+		std::cerr << "Failed to initialize audio capture" << std::endl;
+		return 1;
+	}
+	if (!audio_capture.startCapture()) {
+		std::cerr << "Failed to start audio capture" << std::endl;
+		return 1;
+	}
+
+	// Main loop
+	while (true) {
+		// Handle SDL events
+		if (!handleSDLEvents()) {
+			break;
+		}
+
+		// Get audio data from the audio capture
+		std::vector<float> audio_data;
+		audio_capture.getAudioData(1000, audio_data);
+
+		// Process the audio data for transcription
+		tt.processAudio(audio_data);
+	}
+
+	return 0;
 }

src/modules/core/CMakeLists.txt

@@ -2,18 +2,19 @@ add_library(Core
     src/model-downloader.cpp
     src/model-infos.cpp
     src/model-find-utils.cpp
+    src/logger.cpp
 )
 
 target_link_libraries(Core PUBLIC sago_platform_folders_lib libcurl)
 target_include_directories(Core
     PRIVATE
-        ${SAGO_INSTALL_DIR}/include
+    ${SAGO_INSTALL_DIR}/include
 )
 
 target_include_directories(Core
     PUBLIC
-        $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/include>
-        $<INSTALL_INTERFACE:include>
+    $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/include>
+    $<INSTALL_INTERFACE:include>
 )
 
 set_target_properties(Core PROPERTIES

src/modules/core/include/logger.h

@@ -12,7 +12,7 @@ class Logger {
 
 	static void setLogCallback(LogCallback callback);
 	static void setLogLevel(Level level);
-	static void Logger::log(Level level, const std::string &format, ...);
+	static void Logger::log(Level level, const std::string format, ...);
 	// set log level
 
 private:

src/modules/core/src/logger.cpp

@@ -18,7 +18,7 @@ void Logger::setLogLevel(Logger::Level level)
 	s_logLevel = level;
 }
 
-void Logger::log(Level level, const std::string &format, ...)
+void Logger::log(Level level, const std::string format, ...)
 {
 	if (level < s_logLevel) {
 		return;

src/modules/core/src/model-downloader.cpp

@@ -5,6 +5,7 @@
 #include <sago/platform_folders.h>
 
 #include <filesystem>
+#include <fstream>
 
 #include <curl/curl.h>
 
@@ -67,10 +68,11 @@ std::string find_model_ext_file(const ModelInfo &model_info, const std::string &
 	return find_ext_file_in_folder(model_local_folder_path, ext);
 }
 
-size_t write_data(void *ptr, size_t size, size_t nmemb, FILE *stream)
+size_t write_data(void *ptr, size_t size, size_t nmemb, void *userdata)
 {
-	size_t written = fwrite(ptr, size, nmemb, stream);
-	return written;
+	std::ofstream *fp = static_cast<std::ofstream *>(userdata);
+	fp->write((const char *)ptr, size * nmemb);
+	return size * nmemb;
 }
 
 std::string get_filename_from_url(const std::string &url)
@@ -123,8 +125,8 @@ void download_model(const ModelInfo &model_info, download_finished_callback_t fi
 				continue;
 			}
 
-			FILE *fp = fopen(model_file_save_path.c_str(), "wb");
-			if (fp == nullptr) {
+			std::ofstream fp(model_file_save_path, std::ios::binary);
+			if (!fp.is_open()) {
 				Logger::log(Logger::Level::ERROR_LOG,
 					    "Failed to open model file for writing %s.",
 					    model_file_save_path.c_str());
@@ -134,7 +136,7 @@ void download_model(const ModelInfo &model_info, download_finished_callback_t fi
 			}
 			curl_easy_setopt(curl, CURLOPT_URL, model_download_file.url.c_str());
 			curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, write_data);
-			curl_easy_setopt(curl, CURLOPT_WRITEDATA, fp);
+			curl_easy_setopt(curl, CURLOPT_WRITEDATA, &fp);
 			curl_easy_setopt(curl, CURLOPT_NOPROGRESS, 0L);
 			curl_easy_setopt(curl, CURLOPT_XFERINFOFUNCTION,
 					 [progress_callback](void *clientp, curl_off_t dltotal,
@@ -158,7 +160,7 @@ void download_model(const ModelInfo &model_info, download_finished_callback_t fi
 				error_callback(DownloadError::DOWNLOAD_ERROR_NETWORK,
 					       "Failed to download model file.");
 			}
-			fclose(fp);
+			fp.close();
 		}
 		curl_easy_cleanup(curl);
 		finished_callback(DownloadStatus::DOWNLOAD_STATUS_OK, model_local_config_path);

src/modules/transcription/include/whisper-processing.h

@@ -28,10 +28,8 @@ struct DetectionResultWithText {
 	std::vector<whisper_token_data> tokens;
 	std::string language;
 };
-whats it
 
-	void
-	whisper_loop(void *data);
+void whisper_loop(void *data);
 struct whisper_context *init_whisper_context(const std::string &model_path,
 					     struct transcription_context *gf);
 void run_inference_and_callbacks(transcription_context *gf, uint64_t start_offset_ms,