ferrum_server/

axum_server.rs

//! Axum-based HTTP server implementation for Ferrum
//!
//! This module provides a concrete implementation of the HttpServer trait
//! using the Axum web framework, with full OpenAI API compatibility.

use crate::{openai::*, traits::HttpServer, types::*};
use async_trait::async_trait;
use axum::{
    extract::State,
    http::StatusCode as AxumStatusCode,
    response::{sse::Event, IntoResponse, Response, Sse},
    routing::{get, post},
    Json, Router,
};
use ferrum_interfaces::engine::InferenceEngine;
use ferrum_types::{
    FerrumError as Error, FinishReason, InferenceRequest, ModelId, Priority, RequestId,
    SamplingParams,
};
use std::sync::Arc;
use tokio::sync::mpsc;
use tokio_stream::StreamExt;
use tower::ServiceBuilder;
use tower_http::{cors::CorsLayer, trace::TraceLayer};
use tracing::{debug, error, info, span, Level};
use uuid::Uuid;

/// Shared Prometheus recorder handle for rendering metrics.
static PROM_HANDLE: std::sync::OnceLock<metrics_exporter_prometheus::PrometheusHandle> =
    std::sync::OnceLock::new();

/// Initialize the Prometheus metrics recorder.
///
/// Must be called once before any `metrics::counter!()` / `histogram!()` calls.
/// Safe to call multiple times — subsequent calls are no-ops.
pub fn init_prometheus_recorder() {
    PROM_HANDLE.get_or_init(|| {
        let builder = metrics_exporter_prometheus::PrometheusBuilder::new();
        let handle = builder
            .install_recorder()
            .expect("Failed to install Prometheus recorder");
        info!("Prometheus metrics recorder installed");
        handle
    });
}

/// Axum-based server implementation
pub struct AxumServer {
    engine: Arc<dyn InferenceEngine + Send + Sync>,
    config: ServerConfig,
}

impl AxumServer {
    /// Create a new Axum server
    pub fn new(engine: Arc<dyn InferenceEngine + Send + Sync>) -> Self {
        Self {
            engine,
            config: ServerConfig::default(),
        }
    }

    /// Build the router with all routes
    fn build_router(&self) -> Router {
        let app_state = AppState {
            engine: self.engine.clone(),
        };

        Router::new()
            // OpenAI API routes
            .route("/v1/chat/completions", post(chat_completions_handler))
            .route("/v1/completions", post(completions_handler))
            .route("/v1/embeddings", post(embeddings_handler))
            .route("/v1/audio/transcriptions", post(transcriptions_handler))
            .route("/v1/audio/speech", post(speech_handler))
            .route("/v1/models", get(models_handler))
            // Health & observability
            .route("/health", get(health_handler))
            .route("/metrics", get(metrics_handler))
            .route("/", get(root_handler))
            // Apply middleware
            .layer(
                ServiceBuilder::new()
                    .layer(TraceLayer::new_for_http())
                    .layer(CorsLayer::permissive()), // For MVP, allow all origins
            )
            .with_state(app_state)
    }
}

/// Application state shared across handlers
#[derive(Clone)]
struct AppState {
    engine: Arc<dyn InferenceEngine + Send + Sync>,
}

#[async_trait]
impl HttpServer for AxumServer {
    async fn start(&self, config: &ServerConfig) -> ferrum_types::Result<()> {
        let addr = format!("{}:{}", config.host, config.port);
        info!("Starting Axum server on {}", addr);

        let app = self.build_router();
        let listener = tokio::net::TcpListener::bind(&addr)
            .await
            .map_err(|e| Error::internal(format!("Failed to bind to {}: {}", addr, e)))?;

        info!("Server listening on {}", addr);

        axum::serve(listener, app)
            .await
            .map_err(|e| Error::internal(format!("Server error: {}", e)))?;

        Ok(())
    }

    async fn stop(&self, _timeout: std::time::Duration) -> ferrum_types::Result<()> {
        info!("Stopping Axum server");
        // Axum doesn't have explicit stop - server stops when task is cancelled
        Ok(())
    }

    fn is_running(&self) -> bool {
        // For MVP, always return true when server object exists
        true
    }

    fn address(&self) -> Option<std::net::SocketAddr> {
        // For MVP, return configured address
        format!("{}:{}", self.config.host, self.config.port)
            .parse()
            .ok()
    }

    fn register_handler(
        &mut self,
        _path: &str,
        _method: HttpMethod,
        _handler: Box<dyn crate::traits::RequestHandler>,
    ) {
        // For MVP, routes are static
        unimplemented!("Dynamic handler registration not implemented in MVP")
    }

    fn register_middleware(&mut self, _middleware: Box<dyn crate::traits::Middleware>) {
        // For MVP, middleware is static
        unimplemented!("Dynamic middleware registration not implemented in MVP")
    }

    fn get_metrics(&self) -> ServerMetrics {
        // Return empty metrics for MVP
        ServerMetrics {
            total_requests: 0,
            requests_by_endpoint: std::collections::HashMap::new(),
            requests_by_status: std::collections::HashMap::new(),
            avg_response_time_ms: 0.0,
            p95_response_time_ms: 0.0,
            p99_response_time_ms: 0.0,
            active_connections: 0,
            bytes_sent: 0,
            bytes_received: 0,
            error_rate: 0.0,
            uptime_seconds: 0,
        }
    }

    async fn health_check(&self) -> HealthStatus {
        HealthStatus::Healthy
    }
}

/// Main chat completions handler
async fn chat_completions_handler(
    State(state): State<AppState>,
    Json(request): Json<ChatCompletionsRequest>,
) -> std::result::Result<Response, ServerError> {
    let span = span!(Level::INFO, "chat_completions", model = %request.model);
    let _enter = span.enter();

    info!(
        "Received chat completions request for model: {}",
        request.model
    );
    debug!("Request: {:?}", request);

    // Convert OpenAI request to internal format
    let inference_request =
        convert_chat_request(&request).map_err(|e| ServerError::BadRequest(e.to_string()))?;

    // Check if streaming is requested
    if request.stream.unwrap_or(false) {
        handle_chat_completions_stream(state, request, inference_request).await
    } else {
        handle_chat_completions_sync(state, request, inference_request).await
    }
}

/// Handle streaming chat completions
async fn handle_chat_completions_stream(
    state: AppState,
    openai_request: ChatCompletionsRequest,
    inference_request: InferenceRequest,
) -> std::result::Result<Response, ServerError> {
    let (tx, rx) = mpsc::unbounded_channel::<std::result::Result<Event, axum::Error>>();

    // Spawn task to generate tokens
    let engine = state.engine.clone();
    let request_id = Uuid::new_v4().to_string();
    let prompt_tokens = inference_request.prompt.split_whitespace().count() as u32;

    tokio::spawn(async move {
        let mut current_text = String::new();
        let mut token_count = 0;
        let max_tokens = openai_request.max_tokens.unwrap_or(100);

        match engine.infer_stream(inference_request).await {
            Ok(mut stream) => {
                while let Some(result) = stream.next().await {
                    match result {
                        Ok(chunk) => {
                            current_text.push_str(&chunk.text);
                            token_count += 1;

                            // Create streaming response chunk
                            let response_chunk = ChatCompletionsResponse {
                                id: request_id.clone(),
                                object: "chat.completion.chunk".to_string(),
                                created: chrono::Utc::now().timestamp() as u64,
                                model: openai_request.model.clone(),
                                choices: vec![ChatChoice {
                                    index: 0,
                                    message: None,
                                    delta: Some(ChatMessage {
                                        role: MessageRole::Assistant,
                                        content: chunk.text.clone(),
                                        name: None,
                                    }),
                                    finish_reason: None,
                                }],
                                usage: None,
                            };

                            let sse_event = Event::default()
                                .json_data(&response_chunk)
                                .unwrap_or_else(|_| Event::default().data("error"));
                            if tx.send(Ok(sse_event)).is_err() {
                                break;
                            }

                            // Check stopping conditions
                            if token_count >= max_tokens || chunk.finish_reason.is_some() {
                                // Send final chunk
                                let final_chunk = ChatCompletionsResponse {
                                    id: request_id.clone(),
                                    object: "chat.completion.chunk".to_string(),
                                    created: chrono::Utc::now().timestamp() as u64,
                                    model: openai_request.model.clone(),
                                    choices: vec![ChatChoice {
                                        index: 0,
                                        message: None,
                                        delta: None,
                                        finish_reason: chunk
                                            .finish_reason
                                            .as_ref()
                                            .map(finish_reason_to_string),
                                    }],
                                    usage: Some(Usage {
                                        prompt_tokens,
                                        completion_tokens: token_count,
                                        total_tokens: prompt_tokens + token_count,
                                    }),
                                };

                                let final_event = Event::default()
                                    .json_data(&final_chunk)
                                    .unwrap_or_else(|_| Event::default().data("error"));
                                let _ = tx.send(Ok(final_event));
                                let _ = tx.send(Ok(Event::default().data("[DONE]")));
                                break;
                            }
                        }
                        Err(e) => {
                            error!("Stream generation error: {}", e);
                            let _ = tx.send(Ok(
                                Event::default().data(format!("{{\"error\": \"{}\"}}", e))
                            ));
                            break;
                        }
                    }
                }
            }
            Err(e) => {
                error!("Failed to start streaming: {}", e);
                let _ = tx.send(Ok(
                    Event::default().data(format!("{{\"error\": \"{}\"}}", e))
                ));
            }
        }
    });

    let stream = tokio_stream::wrappers::UnboundedReceiverStream::new(rx);
    let sse_stream = Sse::new(stream);

    Ok(sse_stream.into_response())
}

/// Handle non-streaming chat completions
async fn handle_chat_completions_sync(
    state: AppState,
    openai_request: ChatCompletionsRequest,
    inference_request: InferenceRequest,
) -> std::result::Result<Response, ServerError> {
    info!("Processing non-streaming chat completion");

    let prompt_tokens = inference_request.prompt.split_whitespace().count() as u32;
    match state.engine.infer(inference_request).await {
        Ok(output) => {
            let response = ChatCompletionsResponse {
                id: Uuid::new_v4().to_string(),
                object: "chat.completion".to_string(),
                created: chrono::Utc::now().timestamp() as u64,
                model: openai_request.model,
                choices: vec![ChatChoice {
                    index: 0,
                    message: Some(ChatMessage {
                        role: MessageRole::Assistant,
                        content: output.text,
                        name: None,
                    }),
                    delta: None,
                    finish_reason: Some(finish_reason_to_string(&output.finish_reason)),
                }],
                usage: Some(Usage {
                    prompt_tokens,
                    completion_tokens: output.tokens.len() as u32,
                    total_tokens: prompt_tokens + output.tokens.len() as u32,
                }),
            };

            Ok(Json(response).into_response())
        }
        Err(e) => {
            error!("Generation failed: {}", e);
            Err(ServerError::InternalError(e.to_string()))
        }
    }
}

/// Convert OpenAI chat request to internal inference request
fn convert_chat_request(
    request: &ChatCompletionsRequest,
) -> ferrum_types::Result<InferenceRequest> {
    let prompt = apply_chat_template(&request.messages, &request.model);

    Ok(InferenceRequest {
        id: RequestId(Uuid::new_v4()),
        model_id: ModelId(request.model.clone()),
        prompt,
        sampling_params: SamplingParams {
            max_tokens: request.max_tokens.unwrap_or(100) as usize,
            temperature: request.temperature.unwrap_or(1.0),
            top_p: request.top_p.unwrap_or(1.0),
            top_k: None, // OpenAI doesn't use top-k
            repetition_penalty: 1.0,
            presence_penalty: request.presence_penalty.unwrap_or(0.0),
            frequency_penalty: request.frequency_penalty.unwrap_or(0.0),
            stop_sequences: request.stop.clone().unwrap_or_default(),
            seed: request.seed,
            min_p: None,
            tfs: None,
            typical_p: None,
            mirostat: None,
            response_format: match &request.response_format {
                Some(rf) if rf.format_type == "json_object" => {
                    ferrum_types::ResponseFormat::JsonObject
                }
                Some(rf) if rf.format_type == "json_schema" => {
                    // OpenAI wraps the actual schema one level deeper.
                    match rf.json_schema.as_ref().map(|js| &js.schema) {
                        Some(schema) => match serde_json::to_string(schema) {
                            Ok(s) => ferrum_types::ResponseFormat::JsonSchema(s),
                            Err(_) => ferrum_types::ResponseFormat::Text,
                        },
                        None => ferrum_types::ResponseFormat::JsonObject, // fallback
                    }
                }
                _ => ferrum_types::ResponseFormat::Text,
            },
        },
        stream: request.stream.unwrap_or(false),
        priority: Priority::Normal, // Default priority
        client_id: None,
        session_id: None,
        created_at: chrono::Utc::now(),
        metadata: std::collections::HashMap::new(),
    })
}

/// Other handlers
async fn completions_handler(
    State(_state): State<AppState>,
    Json(_request): Json<CompletionsRequest>,
) -> std::result::Result<Response, ServerError> {
    // TODO: Implement legacy completions endpoint
    Err(ServerError::NotImplemented(
        "Legacy completions not implemented in MVP".to_string(),
    ))
}

/// Embeddings handler — text and image embedding via OpenAI-compatible API.
async fn embeddings_handler(
    State(state): State<AppState>,
    Json(request): Json<EmbeddingsRequest>,
) -> std::result::Result<Response, ServerError> {
    let span = span!(Level::INFO, "embeddings", model = %request.model);
    let _enter = span.enter();

    // Flatten input into individual items
    let items: Vec<EmbeddingItem> = match request.input {
        EmbeddingInput::Single(text) => vec![EmbeddingItem {
            text: Some(text),
            image: None,
        }],
        EmbeddingInput::Batch(texts) => texts
            .into_iter()
            .map(|t| EmbeddingItem {
                text: Some(t),
                image: None,
            })
            .collect(),
        EmbeddingInput::SingleObject(item) => vec![item],
        EmbeddingInput::BatchObjects(items) => items,
    };

    if items.is_empty() {
        return Err(ServerError::BadRequest("Empty input".to_string()));
    }

    let mut data = Vec::with_capacity(items.len());
    let mut total_tokens = 0u32;

    for (idx, item) in items.iter().enumerate() {
        let embedding = if let Some(ref image) = item.image {
            state
                .engine
                .embed_image(image)
                .await
                .map_err(|e| ServerError::InternalError(format!("embed_image: {e}")))?
        } else if let Some(ref text) = item.text {
            total_tokens += text.len() as u32;
            state
                .engine
                .embed_text(text)
                .await
                .map_err(|e| ServerError::InternalError(format!("embed_text: {e}")))?
        } else {
            return Err(ServerError::BadRequest(
                "Each input must have either 'text' or 'image'".to_string(),
            ));
        };

        data.push(EmbeddingData {
            object: "embedding".to_string(),
            embedding,
            index: idx,
        });
    }

    let response = EmbeddingsResponse {
        object: "list".to_string(),
        data,
        model: request.model,
        usage: EmbeddingUsage {
            prompt_tokens: total_tokens,
            total_tokens,
        },
    };

    Ok(Json(response).into_response())
}

/// Audio transcription handler (OpenAI-compatible multipart form).
async fn transcriptions_handler(
    State(state): State<AppState>,
    mut multipart: axum::extract::Multipart,
) -> std::result::Result<Response, ServerError> {
    let span = span!(Level::INFO, "transcription");
    let _enter = span.enter();

    let mut file_data: Option<Vec<u8>> = None;
    let mut language: Option<String> = None;

    while let Some(field) = multipart
        .next_field()
        .await
        .map_err(|e| ServerError::BadRequest(format!("multipart: {e}")))?
    {
        let name = field.name().unwrap_or("").to_string();
        match name.as_str() {
            "file" => {
                file_data = Some(
                    field
                        .bytes()
                        .await
                        .map_err(|e| ServerError::BadRequest(format!("read file: {e}")))?
                        .to_vec(),
                );
            }
            "language" => {
                language = field.text().await.ok().filter(|s| !s.is_empty());
            }
            _ => {} // ignore model, response_format, etc. for now
        }
    }

    let data = file_data.ok_or_else(|| ServerError::BadRequest("missing 'file' field".into()))?;

    let text = state
        .engine
        .transcribe_bytes(&data, language.as_deref())
        .await
        .map_err(|e| ServerError::InternalError(format!("transcribe: {e}")))?;

    Ok(Json(TranscriptionResponse { text }).into_response())
}

/// TTS speech synthesis handler (OpenAI-compatible /v1/audio/speech)
async fn speech_handler(
    State(state): State<AppState>,
    Json(request): Json<SpeechRequest>,
) -> std::result::Result<Response, ServerError> {
    let span = span!(Level::INFO, "speech");
    let _guard = span.enter();

    let language = if request.language.is_empty() || request.language == "auto" {
        None
    } else {
        Some(request.language.as_str())
    };

    let chunk_frames = 10usize;
    let sample_rate = state.engine.tts_sample_rate();

    if request.stream {
        // Streaming: chunked transfer encoding with WAV audio
        let (tx, rx) =
            mpsc::unbounded_channel::<std::result::Result<axum::body::Bytes, std::io::Error>>();

        let engine = state.engine.clone();
        let text = request.input.clone();
        let lang = request.language.clone();

        tokio::task::spawn_blocking(move || {
            let lang_opt = if lang.is_empty() || lang == "auto" {
                None
            } else {
                Some(lang.as_str())
            };
            let rt = tokio::runtime::Handle::current();

            match rt.block_on(engine.synthesize_speech(&text, lang_opt, chunk_frames)) {
                Ok(chunks) => {
                    for chunk in &chunks {
                        let wav_bytes = pcm_to_wav_bytes(chunk, sample_rate);
                        let _ = tx.send(Ok(axum::body::Bytes::from(wav_bytes)));
                    }
                }
                Err(e) => {
                    error!("TTS error: {e}");
                }
            }
        });

        let stream = tokio_stream::wrappers::UnboundedReceiverStream::new(rx);
        let body = axum::body::Body::from_stream(stream);
        Ok(Response::builder()
            .status(200)
            .header("content-type", "audio/wav")
            .header("transfer-encoding", "chunked")
            .body(body)
            .unwrap())
    } else {
        // Non-streaming: return complete WAV
        let chunks = state
            .engine
            .synthesize_speech(&request.input, language, chunk_frames)
            .await
            .map_err(|e| ServerError::InternalError(format!("TTS: {e}")))?;

        let all_samples: Vec<f32> = chunks.into_iter().flatten().collect();
        let wav_bytes = pcm_to_wav_bytes(&all_samples, sample_rate);

        Ok(Response::builder()
            .status(200)
            .header("content-type", "audio/wav")
            .header("content-length", wav_bytes.len().to_string())
            .body(axum::body::Body::from(wav_bytes))
            .unwrap())
    }
}

/// Convert PCM f32 samples to WAV bytes (16-bit, mono).
fn pcm_to_wav_bytes(samples: &[f32], sample_rate: u32) -> Vec<u8> {
    let num_samples = samples.len();
    let data_size = num_samples * 2; // 16-bit = 2 bytes per sample
    let file_size = 44 + data_size;

    let mut buf = Vec::with_capacity(file_size);
    // RIFF header
    buf.extend_from_slice(b"RIFF");
    buf.extend_from_slice(&((file_size - 8) as u32).to_le_bytes());
    buf.extend_from_slice(b"WAVE");
    // fmt chunk
    buf.extend_from_slice(b"fmt ");
    buf.extend_from_slice(&16u32.to_le_bytes()); // chunk size
    buf.extend_from_slice(&1u16.to_le_bytes()); // PCM
    buf.extend_from_slice(&1u16.to_le_bytes()); // mono
    buf.extend_from_slice(&sample_rate.to_le_bytes());
    buf.extend_from_slice(&(sample_rate * 2).to_le_bytes()); // byte rate
    buf.extend_from_slice(&2u16.to_le_bytes()); // block align
    buf.extend_from_slice(&16u16.to_le_bytes()); // bits per sample
                                                 // data chunk
    buf.extend_from_slice(b"data");
    buf.extend_from_slice(&(data_size as u32).to_le_bytes());
    for &s in samples {
        let i16_val = (s.clamp(-1.0, 1.0) * 32767.0) as i16;
        buf.extend_from_slice(&i16_val.to_le_bytes());
    }
    buf
}

async fn models_handler(
    State(state): State<AppState>,
) -> std::result::Result<Response, ServerError> {
    let status = state.engine.status().await;
    let now = chrono::Utc::now().timestamp() as u64;
    let data = status
        .loaded_models
        .into_iter()
        .map(|model_id| crate::openai::ModelInfo {
            id: model_id.to_string(),
            object: "model".to_string(),
            created: now,
            owned_by: "ferrum".to_string(),
            permission: vec![],
            root: None,
            parent: None,
        })
        .collect();

    let models = ModelListResponse {
        object: "list".to_string(),
        data,
    };

    Ok(Json(models).into_response())
}

async fn health_handler(
    State(state): State<AppState>,
) -> std::result::Result<Response, ServerError> {
    let engine_status = state.engine.status().await;
    let scheduler_metrics = state.engine.metrics();

    let health = serde_json::json!({
        "status": "healthy",
        "timestamp": chrono::Utc::now().to_rfc3339(),
        "version": env!("CARGO_PKG_VERSION"),
        "engine": {
            "active_requests": engine_status.active_requests,
            "queued_requests": engine_status.queued_requests,
        },
        "scheduler": {
            "total_requests": scheduler_metrics.total_requests,
            "successful_requests": scheduler_metrics.successful_requests,
            "failed_requests": scheduler_metrics.failed_requests,
            "throughput_rps": scheduler_metrics.throughput_rps,
        }
    });

    Ok(Json(health).into_response())
}

/// Prometheus metrics endpoint — returns metrics in Prometheus text format.
async fn metrics_handler() -> std::result::Result<Response, ServerError> {
    let body = match PROM_HANDLE.get() {
        Some(handle) => handle.render(),
        None => "# Prometheus recorder not initialized\n".to_string(),
    };

    Ok((
        [(
            axum::http::header::CONTENT_TYPE,
            "text/plain; version=0.0.4; charset=utf-8",
        )],
        body,
    )
        .into_response())
}

async fn root_handler() -> std::result::Result<Response, ServerError> {
    let info = serde_json::json!({
        "name": "Ferrum Inference Server",
        "version": env!("CARGO_PKG_VERSION"),
        "api_version": "v1",
        "status": "running"
    });

    Ok(Json(info).into_response())
}

/// Server error type for HTTP responses
#[derive(Debug)]
enum ServerError {
    BadRequest(String),
    InternalError(String),
    NotImplemented(String),
}

impl IntoResponse for ServerError {
    fn into_response(self) -> Response {
        let (status, message) = match self {
            ServerError::BadRequest(msg) => (AxumStatusCode::BAD_REQUEST, msg),
            ServerError::InternalError(msg) => (AxumStatusCode::INTERNAL_SERVER_ERROR, msg),
            ServerError::NotImplemented(msg) => (AxumStatusCode::NOT_IMPLEMENTED, msg),
        };

        let error = OpenAiError {
            error: OpenAiErrorDetail {
                message,
                error_type: "server_error".to_string(),
                param: None,
                code: None,
            },
        };

        (status, Json(error)).into_response()
    }
}

impl std::fmt::Display for MessageRole {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        match self {
            MessageRole::System => write!(f, "system"),
            MessageRole::User => write!(f, "user"),
            MessageRole::Assistant => write!(f, "assistant"),
            MessageRole::Function => write!(f, "function"),
        }
    }
}

/// Render OpenAI-style chat messages into the prompt string the model was
/// trained on. Mirrors the templates `ferrum-cli::commands::run` uses so
/// `/v1/chat/completions` produces the same behaviour as the interactive CLI.
///
/// Detects model family from the request's `model` field:
///   - qwen (Qwen2 / Qwen2.5 / Qwen3): ChatML with `<|im_start|>` / `<|im_end|>`
///     (Qwen3 adds the empty `<think></think>` marker to disable reasoning)
///   - llama 3: `<|start_header_id|>…<|end_header_id|>` + `<|eot_id|>`
///   - fallback: TinyLlama-style `<|system|>` / `<|user|>` / `<|assistant|>`
///     with `</s>` separators
///
/// All templates end with the assistant header so the first generated token
/// becomes the reply content (no extra role prefix).
fn apply_chat_template(messages: &[ChatMessage], model_id: &str) -> String {
    let model_lower = model_id.to_lowercase();

    if model_lower.contains("qwen") {
        let mut prompt = String::new();
        for msg in messages {
            let role = match msg.role {
                MessageRole::System => "system",
                MessageRole::User => "user",
                MessageRole::Assistant => "assistant",
                MessageRole::Function => "function",
            };
            prompt.push_str(&format!(
                "<|im_start|>{}\n{}<|im_end|>\n",
                role, msg.content
            ));
        }
        prompt.push_str("<|im_start|>assistant\n");
        if model_lower.contains("qwen3") {
            // Qwen3: disable thinking mode by inserting an empty think block.
            prompt.push_str("<think>\n\n</think>\n\n");
        }
        prompt
    } else if model_lower.contains("llama") && model_lower.contains("3") {
        let mut prompt = String::from("<|begin_of_text|>");
        for msg in messages {
            let role = match msg.role {
                MessageRole::System => "system",
                MessageRole::User => "user",
                MessageRole::Assistant => "assistant",
                MessageRole::Function => "function",
            };
            prompt.push_str(&format!(
                "<|start_header_id|>{}<|end_header_id|>\n\n{}<|eot_id|>",
                role, msg.content
            ));
        }
        prompt.push_str("<|start_header_id|>assistant<|end_header_id|>\n\n");
        prompt
    } else {
        // TinyLlama / generic chat format. Promote the first system message
        // to the top; subsequent ones (rare) are emitted inline.
        let has_system = messages
            .iter()
            .any(|m| matches!(m.role, MessageRole::System));
        let mut prompt = String::new();
        if !has_system {
            prompt.push_str("<|system|>\nYou are a helpful assistant.</s>\n");
        }
        for msg in messages {
            let tag = match msg.role {
                MessageRole::System => "system",
                MessageRole::User => "user",
                MessageRole::Assistant => "assistant",
                MessageRole::Function => "assistant",
            };
            prompt.push_str(&format!("<|{}|>\n{}</s>\n", tag, msg.content));
        }
        prompt.push_str("<|assistant|>\n");
        prompt
    }
}

/// Convert FinishReason to OpenAI API string
fn finish_reason_to_string(reason: &FinishReason) -> String {
    match reason {
        FinishReason::Length => "length".to_string(),
        FinishReason::Stop => "stop".to_string(),
        FinishReason::EOS => "stop".to_string(),
        FinishReason::Cancelled => "cancelled".to_string(),
        FinishReason::Error => "error".to_string(),
        FinishReason::ContentFilter => "content_filter".to_string(),
    }
}

#[cfg(test)]
mod chat_template_tests {
    use super::*;

    fn msg(role: MessageRole, content: &str) -> ChatMessage {
        ChatMessage {
            role,
            content: content.to_string(),
            name: None,
        }
    }

    #[test]
    fn qwen3_renders_chatml_with_think_marker() {
        let out = apply_chat_template(
            &[
                msg(MessageRole::System, "You are helpful."),
                msg(MessageRole::User, "Hi"),
            ],
            "qwen3:0.6b",
        );
        assert!(out.contains("<|im_start|>system\nYou are helpful.<|im_end|>"));
        assert!(out.contains("<|im_start|>user\nHi<|im_end|>"));
        assert!(out.ends_with("<|im_start|>assistant\n<think>\n\n</think>\n\n"));
    }

    #[test]
    fn qwen2_renders_chatml_without_think() {
        let out = apply_chat_template(&[msg(MessageRole::User, "Hi")], "Qwen/Qwen2.5-7B-Instruct");
        assert!(out.ends_with("<|im_start|>assistant\n"));
        assert!(!out.contains("<think>"));
    }

    #[test]
    fn multi_turn_preserves_order() {
        let out = apply_chat_template(
            &[
                msg(MessageRole::User, "A"),
                msg(MessageRole::Assistant, "B"),
                msg(MessageRole::User, "C"),
            ],
            "qwen3",
        );
        let a_idx = out.find("A").unwrap();
        let b_idx = out.find("B").unwrap();
        let c_idx = out.find("C").unwrap();
        assert!(a_idx < b_idx && b_idx < c_idx);
    }

    #[test]
    fn llama3_renders_header_format() {
        let out = apply_chat_template(
            &[
                msg(MessageRole::System, "sys"),
                msg(MessageRole::User, "hi"),
            ],
            "meta-llama/Llama-3.2-1B-Instruct",
        );
        assert!(out.starts_with("<|begin_of_text|>"));
        assert!(out.contains("<|start_header_id|>system<|end_header_id|>\n\nsys<|eot_id|>"));
        assert!(out.contains("<|start_header_id|>user<|end_header_id|>\n\nhi<|eot_id|>"));
        assert!(out.ends_with("<|start_header_id|>assistant<|end_header_id|>\n\n"));
    }

    #[test]
    fn unknown_model_uses_tinyllama_fallback() {
        let out = apply_chat_template(&[msg(MessageRole::User, "hi")], "mystery-model");
        assert!(out.contains("<|system|>"));
        assert!(out.contains("<|user|>\nhi</s>"));
        assert!(out.ends_with("<|assistant|>\n"));
    }
}