aprender-orchestrate 0.31.2

//! Deploy command implementations
//!
//! This module contains all deployment-related CLI commands extracted from main.rs.

#![cfg(feature = "native")]

use crate::ansi_colors::Colorize;
use std::path::{Path, PathBuf};

/// Deploy subcommand
#[derive(Debug, Clone, clap::Subcommand)]
pub enum DeployCommand {
    /// Generate Docker deployment
    Docker {
        /// Model reference
        model: String,

        /// Output directory for Dockerfile
        #[arg(long, short = 'o', default_value = ".")]
        output: PathBuf,

        /// Base image
        #[arg(long, default_value = "rust:slim")]
        base_image: String,

        /// Expose port
        #[arg(long, default_value = "8080")]
        port: u16,

        /// Build multi-stage (smaller image)
        #[arg(long, default_value = "true")]
        multi_stage: bool,
    },

    /// Generate AWS Lambda deployment
    Lambda {
        /// Model reference
        model: String,

        /// Output directory for Lambda package
        #[arg(long, short = 'o', default_value = ".")]
        output: PathBuf,

        /// Memory size (MB)
        #[arg(long, default_value = "1024")]
        memory: u32,

        /// Timeout (seconds)
        #[arg(long, default_value = "30")]
        timeout: u32,

        /// Create SAM template
        #[arg(long)]
        sam: bool,
    },

    /// Generate Kubernetes deployment
    K8s {
        /// Model reference
        model: String,

        /// Output directory for manifests
        #[arg(long, short = 'o', default_value = ".")]
        output: PathBuf,

        /// Number of replicas
        #[arg(long, default_value = "1")]
        replicas: u32,

        /// Namespace
        #[arg(long, default_value = "default")]
        namespace: String,

        /// Create Helm chart
        #[arg(long)]
        helm: bool,

        /// Enable autoscaling
        #[arg(long)]
        hpa: bool,
    },

    /// Generate Fly.io deployment
    Fly {
        /// Model reference
        model: String,

        /// Output directory for fly.toml
        #[arg(long, short = 'o', default_value = ".")]
        output: PathBuf,

        /// App name
        #[arg(long)]
        app: Option<String>,

        /// Region
        #[arg(long, default_value = "iad")]
        region: String,
    },

    /// Generate Cloudflare Workers deployment
    Cloudflare {
        /// Model reference
        model: String,

        /// Output directory for wrangler.toml
        #[arg(long, short = 'o', default_value = ".")]
        output: PathBuf,

        /// Worker name
        #[arg(long)]
        name: Option<String>,
    },
}

/// Main deploy command dispatcher
pub fn cmd_deploy(command: DeployCommand) -> anyhow::Result<()> {
    match command {
        DeployCommand::Docker { model, output, base_image, port, multi_stage } => {
            cmd_deploy_docker(&model, &output, &base_image, port, multi_stage)?;
        }
        DeployCommand::Lambda { model, output, memory, timeout, sam } => {
            cmd_deploy_lambda(&model, &output, memory, timeout, sam)?;
        }
        DeployCommand::K8s { model, output, replicas, namespace, helm, hpa } => {
            cmd_deploy_k8s(&model, &output, replicas, &namespace, helm, hpa)?;
        }
        DeployCommand::Fly { model, output, app, region } => {
            cmd_deploy_fly(&model, &output, app.as_deref(), &region)?;
        }
        DeployCommand::Cloudflare { model, output, name } => {
            cmd_deploy_cloudflare(&model, &output, name.as_deref())?;
        }
    }
    Ok(())
}

fn cmd_deploy_docker(
    model: &str,
    output: &Path,
    base_image: &str,
    port: u16,
    multi_stage: bool,
) -> anyhow::Result<()> {
    println!("{}", "🐳 Generating Docker Deployment".bright_cyan().bold());
    println!();
    println!("{} Model: {}", "•".bright_blue(), model.cyan());
    println!("{} Output: {}", "•".bright_blue(), output.display());
    println!("{} Base image: {}", "•".bright_blue(), base_image.cyan());
    println!("{} Port: {}", "•".bright_blue(), port);
    println!(
        "{} Multi-stage: {}",
        "•".bright_blue(),
        if multi_stage { "yes".green() } else { "no".dimmed() }
    );
    println!();

    // Generate Dockerfile
    let dockerfile = if multi_stage {
        format!(
            r#"# Multi-stage Dockerfile for Realizar model serving
# Generated by batuta deploy docker

# Build stage
FROM rust:latest AS builder
WORKDIR /app

# Install dependencies
RUN cargo install realizar

# Runtime stage
FROM {base_image}
WORKDIR /app

# Copy binary from builder
COPY --from=builder /usr/local/cargo/bin/realizar /usr/local/bin/

# Copy model (if local)
# COPY model.gguf /app/model.gguf

# Expose port
EXPOSE {port}

# Health check
HEALTHCHECK --interval=30s --timeout=3s \
  CMD curl -f http://localhost:{port}/health || exit 1

# Run server
ENV MODEL_REF="{model}"
CMD ["realizar", "serve", "--host", "0.0.0.0", "--port", "{port}"]
"#
        )
    } else {
        format!(
            r#"# Dockerfile for Realizar model serving
# Generated by batuta deploy docker

FROM {base_image}
WORKDIR /app

# Install realizar
RUN cargo install realizar

# Copy model (if local)
# COPY model.gguf /app/model.gguf

EXPOSE {port}

ENV MODEL_REF="{model}"
CMD ["realizar", "serve", "--host", "0.0.0.0", "--port", "{port}"]
"#
        )
    };

    let dockerfile_path = output.join("Dockerfile");
    std::fs::write(&dockerfile_path, dockerfile)?;

    println!("{} Generated: {}", "✓".bright_green(), dockerfile_path.display());
    println!();
    println!("{}", "Build and run:".bright_yellow());
    println!("  docker build -t my-model-server .");
    println!("  docker run -p {}:{} my-model-server", port, port);

    Ok(())
}

fn cmd_deploy_lambda(
    model: &str,
    output: &Path,
    memory: u32,
    timeout: u32,
    sam: bool,
) -> anyhow::Result<()> {
    println!("{}", "λ Generating Lambda Deployment".bright_cyan().bold());
    println!();
    println!("{} Model: {}", "•".bright_blue(), model.cyan());
    println!("{} Output: {}", "•".bright_blue(), output.display());
    println!("{} Memory: {} MB", "•".bright_blue(), memory);
    println!("{} Timeout: {} seconds", "•".bright_blue(), timeout);
    println!(
        "{} SAM template: {}",
        "•".bright_blue(),
        if sam { "yes".green() } else { "no".dimmed() }
    );
    println!();

    if sam {
        let template = format!(
            r#"AWSTemplateFormatVersion: '2010-09-09'
Transform: AWS::Serverless-2016-10-31
Description: Realizar model serving Lambda
# Generated by batuta deploy lambda

Globals:
  Function:
    Timeout: {timeout}
    MemorySize: {memory}
    Runtime: provided.al2
    Architectures:
      - arm64

Resources:
  ModelFunction:
    Type: AWS::Serverless::Function
    Properties:
      Handler: bootstrap
      CodeUri: .
      Description: Realizar model inference
      Events:
        Inference:
          Type: Api
          Properties:
            Path: /v1/chat/completions
            Method: post
        Health:
          Type: Api
          Properties:
            Path: /health
            Method: get
      Environment:
        Variables:
          MODEL_REF: "{model}"

Outputs:
  ApiEndpoint:
    Description: API Gateway endpoint URL
    Value: !Sub "https://${{ServerlessRestApi}}.execute-api.${{AWS::Region}}.amazonaws.com/Prod/"
"#
        );

        let template_path = output.join("template.yaml");
        std::fs::write(&template_path, template)?;
        println!("{} Generated: {}", "✓".bright_green(), template_path.display());
    }

    println!();
    println!("{}", "Build for Lambda:".bright_yellow());
    println!("  cargo lambda build --release --arm64");
    if sam {
        println!();
        println!("{}", "Deploy with SAM:".bright_yellow());
        println!("  sam deploy --guided");
    }

    Ok(())
}

fn cmd_deploy_k8s(
    model: &str,
    output: &Path,
    replicas: u32,
    namespace: &str,
    _helm: bool,
    hpa: bool,
) -> anyhow::Result<()> {
    println!("{}", "☸ Generating Kubernetes Deployment".bright_cyan().bold());
    println!();
    println!("{} Model: {}", "•".bright_blue(), model.cyan());
    println!("{} Output: {}", "•".bright_blue(), output.display());
    println!("{} Replicas: {}", "•".bright_blue(), replicas);
    println!("{} Namespace: {}", "•".bright_blue(), namespace.cyan());
    println!(
        "{} HPA: {}",
        "•".bright_blue(),
        if hpa { "enabled".green() } else { "disabled".dimmed() }
    );
    println!();

    let deployment = format!(
        r#"apiVersion: apps/v1
kind: Deployment
metadata:
  name: realizar-model-server
  namespace: {namespace}
spec:
  replicas: {replicas}
  selector:
    matchLabels:
      app: realizar-model-server
  template:
    metadata:
      labels:
        app: realizar-model-server
    spec:
      containers:
      - name: realizar
        image: realizar:latest
        ports:
        - containerPort: 8080
        env:
        - name: MODEL_REF
          value: "{model}"
        resources:
          requests:
            memory: "1Gi"
            cpu: "500m"
          limits:
            memory: "4Gi"
            cpu: "2"
        livenessProbe:
          httpGet:
            path: /health
            port: 8080
          initialDelaySeconds: 30
          periodSeconds: 10
        readinessProbe:
          httpGet:
            path: /health
            port: 8080
          initialDelaySeconds: 5
          periodSeconds: 5
---
apiVersion: v1
kind: Service
metadata:
  name: realizar-model-server
  namespace: {namespace}
spec:
  selector:
    app: realizar-model-server
  ports:
  - port: 80
    targetPort: 8080
  type: ClusterIP
"#
    );

    let deployment_path = output.join("deployment.yaml");
    std::fs::write(&deployment_path, deployment)?;
    println!("{} Generated: {}", "✓".bright_green(), deployment_path.display());

    if hpa {
        let hpa_manifest = format!(
            r"apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: realizar-model-server-hpa
  namespace: {namespace}
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: realizar-model-server
  minReplicas: {replicas}
  maxReplicas: 10
  metrics:
  - type: Resource
    resource:
      name: cpu
      target:
        type: Utilization
        averageUtilization: 70
"
        );
        let hpa_path = output.join("hpa.yaml");
        std::fs::write(&hpa_path, hpa_manifest)?;
        println!("{} Generated: {}", "✓".bright_green(), hpa_path.display());
    }

    println!();
    println!("{}", "Deploy:".bright_yellow());
    println!("  kubectl apply -f {}", deployment_path.display());
    if hpa {
        println!("  kubectl apply -f {}", output.join("hpa.yaml").display());
    }

    Ok(())
}

fn cmd_deploy_fly(
    model: &str,
    output: &Path,
    app: Option<&str>,
    region: &str,
) -> anyhow::Result<()> {
    println!("{}", "✈️ Generating Fly.io Deployment".bright_cyan().bold());
    println!();
    println!("{} Model: {}", "•".bright_blue(), model.cyan());
    println!("{} Output: {}", "•".bright_blue(), output.display());
    println!("{} App: {}", "•".bright_blue(), app.unwrap_or("auto-generated").cyan());
    println!("{} Region: {}", "•".bright_blue(), region.cyan());
    println!();

    let app_name = app.unwrap_or("realizar-model-server");
    let fly_toml = format!(
        r#"# fly.toml - Fly.io configuration
# Generated by batuta deploy fly

app = "{app_name}"
primary_region = "{region}"

[env]
  MODEL_REF = "{model}"

[http_service]
  internal_port = 8080
  force_https = true
  auto_stop_machines = true
  auto_start_machines = true
  min_machines_running = 0

[[services]]
  internal_port = 8080
  protocol = "tcp"

  [[services.ports]]
    handlers = ["http"]
    port = 80

  [[services.ports]]
    handlers = ["tls", "http"]
    port = 443

  [[services.tcp_checks]]
    grace_period = "30s"
    interval = "15s"
    restart_limit = 0
    timeout = "2s"

  [[services.http_checks]]
    grace_period = "30s"
    interval = "10s"
    method = "get"
    path = "/health"
    protocol = "http"
    timeout = "2s"
"#
    );

    let fly_path = output.join("fly.toml");
    std::fs::write(&fly_path, fly_toml)?;
    println!("{} Generated: {}", "✓".bright_green(), fly_path.display());

    println!();
    println!("{}", "Deploy:".bright_yellow());
    println!("  fly launch");
    println!("  fly deploy");

    Ok(())
}

fn cmd_deploy_cloudflare(model: &str, output: &Path, name: Option<&str>) -> anyhow::Result<()> {
    println!("{}", "☁️ Generating Cloudflare Workers Deployment".bright_cyan().bold());
    println!();
    println!("{} Model: {}", "•".bright_blue(), model.cyan());
    println!("{} Output: {}", "•".bright_blue(), output.display());
    println!("{} Worker name: {}", "•".bright_blue(), name.unwrap_or("realizar-worker").cyan());
    println!();

    let worker_name = name.unwrap_or("realizar-worker");
    let wrangler_toml = format!(
        r#"# wrangler.toml - Cloudflare Workers configuration
# Generated by batuta deploy cloudflare

name = "{worker_name}"
main = "src/index.js"
compatibility_date = "2024-01-01"

[vars]
MODEL_REF = "{model}"

# Note: Cloudflare Workers have limited compute resources
# Consider using Cloudflare Pages with Functions for larger models
# or Cloudflare Workers with Durable Objects for persistent state
"#
    );

    let wrangler_path = output.join("wrangler.toml");
    std::fs::write(&wrangler_path, wrangler_toml)?;
    println!("{} Generated: {}", "✓".bright_green(), wrangler_path.display());

    println!();
    println!("{}", "Note:".bright_yellow());
    println!("  Cloudflare Workers have limited compute resources.");
    println!("  For ML inference, consider using:");
    println!("  - Cloudflare Workers AI (built-in LLM support)");
    println!("  - Edge proxy to Realizar server");
    println!();
    println!("{}", "Deploy:".bright_yellow());
    println!("  wrangler deploy");

    Ok(())
}