From 01088f4f1915e36a7d0d8d8756f62f8207a48911 Mon Sep 17 00:00:00 2001
From: soryu <soryu@soryu.co>
Date: Sat, 20 Dec 2025 15:36:04 +0000
Subject: Implement makima listen websockets server

---
 makima/src/server/messages.rs | 92 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 92 insertions(+)
 create mode 100644 makima/src/server/messages.rs

(limited to 'makima/src/server/messages.rs')

diff --git a/makima/src/server/messages.rs b/makima/src/server/messages.rs
new file mode 100644
index 0000000..0c92447
--- /dev/null
+++ b/makima/src/server/messages.rs
@@ -0,0 +1,92 @@
+//! WebSocket and API message types for the makima server.
+
+use serde::{Deserialize, Serialize};
+use utoipa::ToSchema;
+
+/// Audio encoding format for WebSocket streaming.
+#[derive(Debug, Clone, Copy, Deserialize, Serialize, ToSchema)]
+#[serde(rename_all = "lowercase")]
+pub enum AudioEncoding {
+    /// 32-bit floating point PCM samples
+    Pcm32f,
+    /// 16-bit signed integer PCM samples
+    Pcm16,
+    /// Raw bytes (will be interpreted as PCM16)
+    Raw,
+}
+
+/// Initial handshake message from client specifying audio format.
+#[derive(Debug, Clone, Deserialize, Serialize, ToSchema)]
+#[serde(rename_all = "camelCase")]
+pub struct StartMessage {
+    /// Audio sample rate in Hz (e.g., 16000, 44100, 48000)
+    pub sample_rate: u32,
+    /// Number of audio channels (1 for mono, 2 for stereo)
+    pub channels: u16,
+    /// Audio encoding format
+    pub encoding: AudioEncoding,
+}
+
+/// Stop message to terminate the session.
+#[derive(Debug, Clone, Deserialize, Serialize, ToSchema)]
+#[serde(rename_all = "camelCase")]
+pub struct StopMessage {
+    /// Optional reason for stopping
+    pub reason: Option<String>,
+}
+
+/// Wrapper for all WebSocket messages from client to server.
+#[derive(Debug, Clone, Deserialize)]
+#[serde(tag = "type", rename_all = "camelCase")]
+pub enum ClientMessage {
+    Start(StartMessage),
+    Stop(StopMessage),
+}
+
+/// Transcription result message sent from server to client.
+#[derive(Debug, Clone, Serialize, ToSchema)]
+#[serde(rename_all = "camelCase")]
+pub struct TranscriptMessage {
+    /// Speaker identifier (e.g., "Speaker 0", "Speaker 1")
+    pub speaker: String,
+    /// Segment start time in seconds
+    pub start: f32,
+    /// Segment end time in seconds
+    pub end: f32,
+    /// Transcribed text
+    pub text: String,
+    /// Whether this is a final or interim result
+    pub is_final: bool,
+}
+
+/// Wrapper for all WebSocket messages from server to client.
+#[derive(Debug, Clone, Serialize)]
+#[serde(tag = "type", rename_all = "camelCase")]
+pub enum ServerMessage {
+    /// Session is ready for audio streaming
+    Ready { session_id: String },
+    /// Transcription result
+    Transcript(TranscriptMessage),
+    /// Error occurred during processing
+    Error { code: String, message: String },
+    /// Session has been stopped
+    Stopped { reason: String },
+}
+
+/// Error response for HTTP API endpoints.
+#[derive(Debug, Clone, Serialize, ToSchema)]
+pub struct ApiError {
+    /// Error code for programmatic handling
+    pub code: String,
+    /// Human-readable error message
+    pub message: String,
+}
+
+impl ApiError {
+    pub fn new(code: impl Into<String>, message: impl Into<String>) -> Self {
+        Self {
+            code: code.into(),
+            message: message.into(),
+        }
+    }
+}
-- 
cgit v1.2.3