RustAI_Full/src/tts.rs

use anyhow::{anyhow, Context, Result};
use rodio::{Decoder, OutputStream, Sink};
use std::io::Cursor;
use std::process::{Child, Command};
use std::sync::Arc;
use tokio::sync::Mutex;

pub struct TtsEngine {
    process: Arc<Mutex<Option<Child>>>,
    pub server_url: String,
    client: reqwest::Client,
    voice_path: String,
}

impl TtsEngine {
    pub async fn start(voice_dir: &str, port: u16) -> Result<Self> {
        let server_url = format!("http://127.0.0.1:{}", port);
        let client = reqwest::Client::new();

        // Check if server is already running
        let already_running = client
            .get(&format!("{}/", server_url))
            .send()
            .await
            .is_ok();

        let process = if !already_running {
            println!("Starting XTTS TTS server on port {}...", port);
            match Command::new("python")
                .args([
                    "-m",
                    "xtts_api_server",
                    "--port",
                    &port.to_string(),
                    "--speaker-folder",
                    voice_dir,
                ])
                .spawn()
            {
                Ok(child) => {
                    println!("XTTS server starting (model loading may take a minute)...");
                    Some(child)
                }
                Err(e) => {
                    eprintln!("Could not start XTTS server: {}. Voice will be unavailable.", e);
                    eprintln!("Install with: pip install xtts-api-server");
                    None
                }
            }
        } else {
            println!("XTTS server already running on {}", server_url);
            None
        };

        Ok(Self {
            process: Arc::new(Mutex::new(process)),
            server_url,
            client,
            voice_path: voice_dir.to_string(),
        })
    }

    pub async fn speak(&self, text: &str, speaker: &str) -> Result<()> {
        let speaker_path = format!("{}/{}.wav", self.voice_path, speaker);
        let response = self
            .client
            .post(format!("{}/tts", self.server_url))
            .json(&serde_json::json!({
                "text": text,
                "speaker_wav": speaker_path,
                "language": "en"
            }))
            .send()
            .await
            .context("Failed to reach XTTS server")?;

        if !response.status().is_success() {
            let status = response.status();
            let body = response.text().await.unwrap_or_default();
            return Err(anyhow!("XTTS error {}: {}", status, body));
        }

        let wav_bytes = response.bytes().await?;

        // Play audio in a blocking thread (rodio requires sync)
        let bytes = wav_bytes.to_vec();
        std::thread::spawn(move || {
            if let Ok((_stream, stream_handle)) = OutputStream::try_default() {
                if let Ok(sink) = Sink::try_new(&stream_handle) {
                    if let Ok(source) = Decoder::new(Cursor::new(bytes)) {
                        sink.append(source);
                        sink.sleep_until_end();
                    }
                }
            }
        });

        Ok(())
    }

    pub async fn stop(&self) {
        if let Ok(mut guard) = self.process.lock().await {
            if let Some(ref mut child) = *guard {
                let _ = child.kill();
                let _ = child.wait();
            }
        }
    }

    pub async fn is_available(&self) -> bool {
        self.client
            .get(&format!("{}/", self.server_url))
            .send()
            .await
            .is_ok()
    }
}

impl Drop for TtsEngine {
    fn drop(&mut self) {
        if let Ok(mut guard) = self.process.lock() {
            if let Some(ref mut child) = *guard {
                let _ = child.kill();
                let _ = child.wait();
            }
        }
    }
}