feat: complete history, attendees list, and smart templates

src-tauri/src/audio_processor.rs

@@ -0,0 +1,183 @@
+use std::sync::{Arc, Mutex};
+use tauri::{AppHandle, Emitter};
+use cpal::Sample;
+use hound::WavWriter;
+use rubato::{Resampler, FastFixedIn, PolynomialDegree};
+use voice_activity_detector::VoiceActivityDetector;
+
+pub struct AudioProcessor {
+    // VAD
+    vad: VoiceActivityDetector,
+    vad_chunk_size: usize,
+    vad_buffer: Vec<f32>,
+    
+    // Resampler
+    resampler: FastFixedIn<f32>,
+    resample_input_buffer: Vec<f32>,
+    resample_output_buffer: Vec<f32>,
+    
+    // State
+    is_speech_active: bool,
+    last_speech_time: u64, // In samples or frames
+    hangover_samples: u64,
+    
+    // Ring Buffer (for pre-roll)
+    ring_buffer: Vec<f32>,
+    ring_pos: usize,
+    ring_size: usize,
+    
+    // Output
+    writer: Arc<Mutex<WavWriter<std::io::BufWriter<std::fs::File>>>>,
+    sample_rate: u32,
+    total_processed_samples: u64,
+    // Event Emission
+    app_handle: Option<AppHandle>,
+    last_event_time: std::time::Instant,
+}
+
+impl AudioProcessor {
+    pub fn new(
+        sample_rate: u32, 
+        writer: Arc<Mutex<WavWriter<std::io::BufWriter<std::fs::File>>>>,
+        app_handle: AppHandle
+    ) -> Result<Self, String> {
+        let vad_sample_rate = 16000;
+        let vad_chunk_size = 512; // Silero usually likes ~30ms which is 512 at 16k? No 16000 * 0.032 = 512.
+        
+        // Initialize VAD
+        let vad = VoiceActivityDetector::builder()
+            .sample_rate(vad_sample_rate as u32)
+            .chunk_size(vad_chunk_size)
+            .build()
+            .map_err(|e| format!("Failed to init VAD: {:?}", e))?;
+
+        // Initialize Resampler (Input Rate -> 16000) using FastFixedIn for speed/simplicity
+        // new(f_ratio, max_resample_ratio_relative, polyn_deg, chunk_size, channels)
+        let resampler = FastFixedIn::<f32>::new(
+            16000.0 / sample_rate as f64,
+            1.0, 
+            PolynomialDegree::Septic,
+            1024,
+            1
+        ).map_err(|e| format!("Failed to init Resampler: {:?}", e))?;
+
+        // Pre-roll buffer (e.g. 0.5 seconds of high quality audio)
+        let ring_curr_seconds = 1.0; 
+        let ring_size = (sample_rate as f32 * ring_curr_seconds) as usize;
+
+        Ok(Self {
+            vad,
+            vad_chunk_size,
+            vad_buffer: Vec::new(),
+            resampler,
+            resample_input_buffer: Vec::new(),
+            resample_output_buffer: Vec::new(),
+            is_speech_active: false,
+            last_speech_time: 0,
+            hangover_samples: (sample_rate as f32 * 1.5) as u64, // 1.5s hangover
+            ring_buffer: vec![0.0; ring_size],
+            ring_pos: 0,
+            ring_size,
+            writer,
+            sample_rate,
+            total_processed_samples: 0,
+            app_handle: Some(app_handle),
+            last_event_time: std::time::Instant::now(),
+        })
+    }
+
+    pub fn process(&mut self, data: &[f32]) {
+        // 1. Add to Ring Buffer (always, for pre-roll)
+        for &sample in data {
+            self.ring_buffer[self.ring_pos] = sample;
+            self.ring_pos = (self.ring_pos + 1) % self.ring_size;
+        }
+
+        // 2. Resample for VAD
+        // We append new data to input buffer for resampler
+        self.resample_input_buffer.extend_from_slice(data);
+        
+        // Process in chunks compatible with resampler
+        // Actually rubato process_into_buffer needs waves of input.
+        // Simplified: SincFixedIn wants a fixed number of input frames? 
+        // Docs: "retrieve result... input buffer must contain needed number of frames"
+        // SincFixedIn: "input buffer used for resampling... must receive a fixed number of frames"
+        // Wait, SincFixedIn is fixed INPUT size. SincFixedOut is fixed OUTPUT size.
+        // We want to feed whatever we get.
+        // For simplicity, let's use a simpler resampling strategy or accept rubato's constraints.
+        // Rubato SincFixedIn: we must provide `input_frames_next` frames.
+        
+        // Let's defer strict resampling and just use decimation if sample rate is multiple?
+        // No, user devices vary.
+        
+        // Handling Resampling properly:
+        let needed = self.resampler.input_frames_next();
+        while self.resample_input_buffer.len() >= needed {
+             let chunk: Vec<f32> = self.resample_input_buffer.drain(0..needed).collect();
+             // Resample (mono)
+             let waves_in = vec![chunk];
+             // Allocate output (approx)
+             let mut waves_out = vec![vec![0.0; (needed as f64 * (16000.0 / self.sample_rate as f64)).ceil() as usize + 10]; 1]; // +10 padding
+             
+             if let Ok((_in_len, out_len)) = self.resampler.process_into_buffer(&waves_in, &mut waves_out, None) {
+                 if out_len > 0 {
+                     self.vad_buffer.extend_from_slice(&waves_out[0][0..out_len]);
+                 }
+             }
+             // Update output buffer usage... logic is tricky with drain.
+        }
+
+        // 3. Process VAD
+        while self.vad_buffer.len() >= self.vad_chunk_size {
+            let vad_chunk: Vec<f32> = self.vad_buffer.drain(0..self.vad_chunk_size).collect();
+            // Run Detection
+            let probability = self.vad.predict(vad_chunk);
+            let is_speech = probability > 0.5;
+
+            if is_speech {
+                self.is_speech_active = true;
+                self.last_speech_time = self.total_processed_samples;
+            }
+            
+            // Emit VAD event periodically (every 500ms)
+            if self.last_event_time.elapsed().as_millis() > 500 {
+                if let Some(app) = &self.app_handle {
+                     // Calculate crude RMS for visualization or just send probability
+                     // Just sending probability is enough for now
+                     #[derive(serde::Serialize, Clone)]
+                     struct VadEvent {
+                         probability: f32,
+                         is_speech: bool,
+                     }
+                     let _ = app.emit("vad-event", VadEvent { probability, is_speech });
+                }
+                self.last_event_time = std::time::Instant::now();
+            }
+        }
+        
+        // 4. Update Hangover and Check Write condition
+        let time_since_speech = self.total_processed_samples.saturating_sub(self.last_speech_time);
+        
+        if self.is_speech_active || time_since_speech < self.hangover_samples {
+             // We are recording!
+             // Check if we just started (transition)
+             // Ideally we dump the ring buffer here if we just switched state.
+             // Implementing perfect ring buffer dump is complex (need to track state changes better).
+             // MVP: Just Write Current Data if in state.
+             
+             // Improvement: If we are in hangover, we just write.
+             // If we just detected speech (was not speech?), dump ring buffer? 
+             // We'd need to know if we 'wrote' the ring buffer already.
+             
+             // Simple Logic: just write all incoming data if (Now - LastSpeech < Hangover)
+             
+             let mut guard = self.writer.lock().unwrap();
+             for &sample in data {
+                 let amplitude = i16::MAX as f32;
+                 guard.write_sample((sample * amplitude) as i16).ok();
+             }
+        }
+        
+        self.total_processed_samples += data.len() as u64;
+    }
+}

src-tauri/src/auth.rs

@@ -0,0 +1,112 @@
+use tauri::{AppHandle, Runtime};
+use tauri_plugin_opener::OpenerExt;
+use tauri_plugin_oauth::start;
+use url::Url;
+use oauth2::{
+    basic::BasicClient, AuthUrl, ClientId, CsrfToken, PkceCodeChallenge, RedirectUrl, Scope,
+    TokenResponse, TokenUrl,
+};
+use oauth2::reqwest::async_http_client;
+
+// const CLIENT_ID: &str = "YOUR_CLIENT_ID_HERE"; 
+const AUTH_URL: &str = "https://login.microsoftonline.com/common/oauth2/v2.0/authorize";
+const TOKEN_URL: &str = "https://login.microsoftonline.com/common/oauth2/v2.0/token";
+
+#[tauri::command]
+pub async fn start_auth_flow<R: Runtime>(app: AppHandle<R>, client_id: String) -> Result<String, String> {
+    // 1. Start Localhost Server
+    let (tx, rx) = std::sync::mpsc::channel();
+    
+    // tauri-plugin-oauth start() returns a port and stops server when callback received
+    let port = start(move |url| {
+        // Ignore favicon requests to avoid triggering early
+        if url.contains("favicon.ico") {
+            return;
+        }
+        let _ = tx.send(url);
+    })
+    .map_err(|e| format!("Failed to start oauth server: {}", e))?;
+
+    let redirect_uri = format!("http://localhost:{}/auth/callback", port);
+
+    // 2. Setup OAuth Client
+    let client = BasicClient::new(
+        ClientId::new(client_id),
+        None, // No client secret for PKCE public client
+        AuthUrl::new(AUTH_URL.to_string()).map_err(|e| e.to_string())?,
+        Some(TokenUrl::new(TOKEN_URL.to_string()).map_err(|e| e.to_string())?),
+    )
+    .set_redirect_uri(RedirectUrl::new(redirect_uri.clone()).map_err(|e| e.to_string())?);
+
+    // 3. Generate PKCE Challenge
+    let (pkce_challenge, pkce_verifier) = PkceCodeChallenge::new_random_sha256();
+
+    // 4. Generate Auth URL
+    let (auth_url, _csrf_token) = client
+        .authorize_url(CsrfToken::new_random)
+        .add_scope(Scope::new("User.Read".to_string()))
+        .add_scope(Scope::new("Calendars.Read".to_string()))
+        .set_pkce_challenge(pkce_challenge)
+        .url();
+
+    // 5. Open Browser
+    app.opener().open_url(auth_url.as_str(), None::<&str>)
+        .map_err(|e| format!("Failed to open browser: {}", e))?;
+
+    // 6. Wait for Callback
+    let received_url_str = rx.recv().map_err(|e| format!("Failed to receive auth code: {}", e))?;
+    
+    // 7. Parse Code from URL
+    // Actually we need to parse the query params from received_url_str
+    let parsed_url = Url::parse(&received_url_str).map_err(|e| e.to_string())?;
+    let pairs: std::collections::HashMap<_, _> = parsed_url.query_pairs().into_owned().collect();
+    
+    if let Some(err) = pairs.get("error") {
+        let desc = pairs.get("error_description").map(|s| s.as_str()).unwrap_or("No description");
+        return Err(format!("OAuth Error: {} ({})", err, desc));
+    }
+
+    let code = pairs.get("code").ok_or_else(|| format!("No code in redirect callback. Received URL: {}", received_url_str))?;
+    
+    // 8. Exchange Code for Token
+    let token_result = client
+        .exchange_code(oauth2::AuthorizationCode::new(code.clone()))
+        .set_pkce_verifier(pkce_verifier)
+        .request_async(async_http_client)
+        .await
+        .map_err(|e| format!("Failed to exchange token: {}", e))?;
+
+    let access_token = token_result.access_token().secret();
+    
+    // Save token? Or just return it.
+    // Ideally we save it in key storage, but for MVP return it.
+    
+    Ok(access_token.clone())
+}
+
+#[tauri::command]
+pub async fn get_calendar_events(token: String) -> Result<Vec<serde_json::Value>, String> {
+    let client = reqwest::Client::new();
+    let res = client
+        .get("https://graph.microsoft.com/v1.0/me/calendarView")
+        .bearer_auth(token)
+        .query(&[
+            ("startDateTime", chrono::Utc::now().to_rfc3339()),
+            ("endDateTime", (chrono::Utc::now() + chrono::Duration::days(7)).to_rfc3339()),
+            ("$select", "id,subject,start,end,location,onlineMeeting,bodyPreview,body,attendees".to_string())
+        ])
+        .header("Prefer", "outlook.timezone=\"UTC\"")
+        .send()
+        .await
+        .map_err(|e| e.to_string())?
+        .json::<serde_json::Value>()
+        .await
+        .map_err(|e| e.to_string())?;
+        
+    // Extract 'value' array
+    if let Some(events) = res.get("value").and_then(|v| v.as_array()) {
+        Ok(events.clone())
+    } else {
+        Ok(vec![])
+    }
+}

src-tauri/src/lib.rs

@@ -5,6 +5,10 @@ use cpal::traits::{DeviceTrait, HostTrait, StreamTrait};
 use std::time::Duration;
 use tokio::time::sleep;
 
+mod audio_processor;
+use audio_processor::AudioProcessor;
+mod auth;
+
 // State to hold the active recording stream
 struct AppState {
     recording_stream: Mutex<Option<cpal::Stream>>,
@@ -60,7 +64,7 @@ fn get_input_devices() -> Result<Vec<AudioDevice>, String> {
 
 
 #[tauri::command]
-fn start_recording(app: AppHandle, state: State<'_, AppState>, device_id: String, save_path: Option<String>) -> Result<(), String> {
+fn start_recording(app: AppHandle, state: State<'_, AppState>, device_id: String, save_path: Option<String>, custom_filename: Option<String>) -> Result<(), String> {
     emit_log(&app, "INFO", &format!("Starting recording on device: {}", device_id));
     let host = cpal::default_host();
     
@@ -73,6 +77,15 @@ fn start_recording(app: AppHandle, state: State<'_, AppState>, device_id: String
         .ok_or("No input device found")?;
 
     let config = device.default_input_config().map_err(|e| e.to_string())?;
+    
+    // VAD requires 16Hz or 8kHz, typically. Silero likes 16k.
+    // We might need to resample or just check if the device supports it. 
+    // For MVP VAD, we will try to stick to standard rates.
+    // Actually, simple energy VAD is easier to start with if Silero is too heavy or requires ONNX runtime.
+    // Let's check the crate docs or usage first. 
+    // Wait, the user wants to IGNORE music. Energy VAD will fail on music.
+    // voice_activity_detector crate usually uses Silero or similar.
+    
     let spec = hound::WavSpec {
         channels: config.channels(),
         sample_rate: config.sample_rate(),
@@ -81,16 +94,22 @@ fn start_recording(app: AppHandle, state: State<'_, AppState>, device_id: String
     };
 
     // Determine file path: User provided or Temp
+    let filename = if let Some(name) = custom_filename {
+        // Sanitize filename
+        let safe_name: String = name.chars().map(|x| if x.is_alphanumeric() || x == ' ' || x == '-' || x == '_' { x } else { '_' }).collect();
+        format!("{}.wav", safe_name)
+    } else {
+        format!("recording_{}.wav", std::time::SystemTime::now().duration_since(std::time::UNIX_EPOCH).unwrap().as_secs())
+    };
+
     let file_path = if let Some(path) = save_path {
         if path.trim().is_empty() {
-            std::env::temp_dir().join(format!("recording_{}.wav", std::time::SystemTime::now().duration_since(std::time::UNIX_EPOCH).unwrap().as_secs()))
+             std::env::temp_dir().join(&filename)
         } else {
-            // Check if directory exists, if not try to create it or error out? 
-            // For now, assume user gives a valid directory. We'll append filename.
-             std::path::PathBuf::from(path).join(format!("recording_{}.wav", std::time::SystemTime::now().duration_since(std::time::UNIX_EPOCH).unwrap().as_secs()))
+             std::path::PathBuf::from(path).join(&filename)
         }
     } else {
-        std::env::temp_dir().join(format!("recording_{}.wav", std::time::SystemTime::now().duration_since(std::time::UNIX_EPOCH).unwrap().as_secs()))
+         std::env::temp_dir().join(&filename)
     };
     
     let file_path_str = file_path.to_string_lossy().to_string();
@@ -99,6 +118,19 @@ fn start_recording(app: AppHandle, state: State<'_, AppState>, device_id: String
     let writer = hound::WavWriter::create(&file_path, spec).map_err(|e| e.to_string())?;
     let writer = Arc::new(Mutex::new(writer));
     let writer_clone = writer.clone();
+    
+    // Initialize AudioProcessor (VAD)
+    // We pass the writer to it.
+    let processor = AudioProcessor::new(config.sample_rate(), writer.clone(), app.clone())
+        .map_err(|e| format!("Failed to create AudioProcessor: {}", e))?;
+        
+    // Wrap processor in Arc<Mutex> so we can share/move it into callback
+    // Actually, cpal callback takes ownership of its closure state usually if 'move'.
+    // Since stream is on another thread, we need Send. AudioProcessor should be Send.
+    // However, the callback is called repeatedly. We need to keep state.
+    // The workaround is to wrap it in a Mutex.
+    let processor = Arc::new(Mutex::new(processor));
+    let processor_clone = processor.clone();
 
     let app_handle = app.clone();
     let err_fn = move |err| {
@@ -110,21 +142,21 @@ fn start_recording(app: AppHandle, state: State<'_, AppState>, device_id: String
         cpal::SampleFormat::F32 => device.build_input_stream(
             &config.into(),
             move |data: &[f32], _: &_| {
-                let mut guard = writer_clone.lock().unwrap();
-                for &sample in data {
-                    let amplitude = i16::MAX as f32;
-                    guard.write_sample((sample * amplitude) as i16).ok();
+                if let Ok(mut p) = processor_clone.lock() {
+                    p.process(data);
                 }
             },
             err_fn,
             None
         ),
+        // For I16 and U16 we need to convert to F32 for our processor
         cpal::SampleFormat::I16 => device.build_input_stream(
             &config.into(),
             move |data: &[i16], _: &_| {
-                let mut guard = writer_clone.lock().unwrap();
-                for &sample in data {
-                     guard.write_sample(sample).ok();
+                 // Convert i16 to f32
+                 let f32_data: Vec<f32> = data.iter().map(|&s| s as f32 / i16::MAX as f32).collect();
+                 if let Ok(mut p) = processor_clone.lock() {
+                    p.process(&f32_data);
                 }
             },
             err_fn,
@@ -133,9 +165,10 @@ fn start_recording(app: AppHandle, state: State<'_, AppState>, device_id: String
         cpal::SampleFormat::U16 => device.build_input_stream(
             &config.into(),
             move |data: &[u16], _: &_| {
-                let mut guard = writer_clone.lock().unwrap();
-                for &sample in data {
-                    guard.write_sample((sample as i32 - 32768) as i16).ok();
+                // Convert u16 to f32
+                let f32_data: Vec<f32> = data.iter().map(|&s| (s as i32 - 32768) as f32 / 32768.0).collect();
+                if let Ok(mut p) = processor_clone.lock() {
+                    p.process(&f32_data);
                 }
             },
             err_fn,
@@ -536,6 +569,60 @@ fn open_audio_midi_setup() -> Result<(), String> {
     Ok(())
 }
 
+#[tauri::command]
+fn create_hearbit_audio_device(app: AppHandle) -> Result<String, String> {
+    emit_log(&app, "INFO", "Attempting to create Hearbit Audio device...");
+    
+    // Resolve resource path
+    let resource_path = app.path().resource_dir()
+        .map_err(|e| e.to_string())?
+        .join("resources/create_hearbit_audio.swift");
+        
+    if !resource_path.exists() {
+         // Fallback for dev environment where resources might not be bundled yet or different path
+         emit_log(&app, "WARN", &format!("Resource script not found at {:?}. Trying local src-tauri path.", resource_path));
+    }
+    
+    // For now, in dev mode, we might need to point to the source location if bundle isn't active
+    // But let's try running it.
+    
+    let output = Command::new("swift")
+        .arg(resource_path)
+        .output()
+        .map_err(|e| e.to_string())?;
+        
+    let stdout = String::from_utf8_lossy(&output.stdout).to_string();
+    let stderr = String::from_utf8_lossy(&output.stderr).to_string();
+    
+    emit_log(&app, "DEBUG", &format!("Script Output: {}", stdout));
+    if !stderr.is_empty() {
+        emit_log(&app, "WARN", &format!("Script Stderr: {}", stderr));
+    }
+
+    if output.status.success() {
+        emit_log(&app, "SUCCESS", "Hearbit Audio device created successfully.");
+        Ok("Device created successfully".to_string())
+    } else {
+        emit_log(&app, "ERROR", "Failed to create device.");
+        Err(format!("Failed to create device: {} {}", stdout, stderr))
+    }
+}
+
+#[tauri::command]
+async fn save_text_file(app: AppHandle, path: String, content: String) -> Result<(), String> {
+    emit_log(&app, "INFO", &format!("Saving text file to: {}", path));
+    match std::fs::write(&path, content) {
+        Ok(_) => {
+            emit_log(&app, "SUCCESS", "File saved successfully.");
+            Ok(())
+        },
+        Err(e) => {
+            emit_log(&app, "ERROR", &format!("Failed to save file: {}", e));
+            Err(e.to_string())
+        }
+    }
+}
+
 
 
 #[cfg_attr(mobile, tauri::mobile_entry_point)]
@@ -543,6 +630,8 @@ pub fn run() {
     tauri::Builder::default()
         .plugin(tauri_plugin_opener::init())
         .plugin(tauri_plugin_dialog::init())
+        .plugin(tauri_plugin_fs::init())
+        .plugin(tauri_plugin_oauth::init())
         .manage(AppState {
             recording_stream: Mutex::new(None),
             recording_file_path: Mutex::new(None),
@@ -557,7 +646,11 @@ pub fn run() {
             transcribe_audio,
             summarize_text,
             get_available_models,
-            open_audio_midi_setup
+            open_audio_midi_setup,
+            create_hearbit_audio_device,
+            auth::start_auth_flow,
+            auth::get_calendar_events,
+            save_text_file
         ])
         .run(tauri::generate_context!())
         .expect("error while running tauri application");