zclaw_openfang/desktop/src-tauri/src/health_check.rs

use serde::Serialize;
use serde_json::Value;
use std::net::{TcpStream, ToSocketAddrs};
use std::time::{Duration, Instant, SystemTime, UNIX_EPOCH};
use tauri::AppHandle;

use crate::gateway::io::{parse_json_output, read_gateway_status, run_zclaw, LocalGatewayStatus};
use crate::gateway::runtime::{resolve_zclaw_runtime, ZCLAW_DEFAULT_PORT};

/// Health status enum
#[derive(Debug, Clone, Serialize)]
#[serde(rename_all = "lowercase")]
pub(crate) enum HealthStatus {
    Healthy,
    Unhealthy,
}

/// Port check result
#[derive(Debug, Clone, Serialize)]
#[serde(rename_all = "camelCase")]
pub(crate) struct PortCheckResult {
    port: u16,
    accessible: bool,
    latency_ms: Option<u64>,
    error: Option<String>,
}

/// Process health details
#[derive(Debug, Clone, Serialize)]
#[serde(rename_all = "camelCase")]
pub(crate) struct ProcessHealthDetails {
    pid: Option<u32>,
    name: Option<String>,
    status: Option<String>,
    uptime_seconds: Option<u64>,
    cpu_percent: Option<f64>,
    memory_mb: Option<f64>,
}

/// Health check response
#[derive(Debug, Clone, Serialize)]
#[serde(rename_all = "camelCase")]
pub(crate) struct HealthCheckResponse {
    status: HealthStatus,
    process: ProcessHealthDetails,
    port_check: PortCheckResult,
    last_check_timestamp: u64,
    checks_performed: Vec<String>,
    issues: Vec<String>,
    runtime_source: Option<String>,
}

/// Check if a TCP port is accessible
fn check_port_accessibility(host: &str, port: u16, timeout_ms: u64) -> PortCheckResult {
    let addr = format!("{}:{}", host, port);

    // Resolve the address
    let socket_addr = match addr.to_socket_addrs() {
        Ok(mut addrs) => addrs.next(),
        Err(e) => {
            return PortCheckResult {
                port,
                accessible: false,
                latency_ms: None,
                error: Some(format!("Failed to resolve address: {}", e)),
            };
        }
    };

    let Some(socket_addr) = socket_addr else {
        return PortCheckResult {
            port,
            accessible: false,
            latency_ms: None,
            error: Some("Failed to resolve address".to_string()),
        };
    };

    // Try to connect with timeout
    let start = Instant::now();

    // Use a simple TCP connect with timeout simulation
    let result = TcpStream::connect_timeout(&socket_addr, Duration::from_millis(timeout_ms));

    match result {
        Ok(_) => {
            let latency = start.elapsed().as_millis() as u64;
            PortCheckResult {
                port,
                accessible: true,
                latency_ms: Some(latency),
                error: None,
            }
        }
        Err(e) => PortCheckResult {
            port,
            accessible: false,
            latency_ms: None,
            error: Some(format!("Connection failed: {}", e)),
        },
    }
}

/// Get process uptime from status command
fn get_process_uptime(status: &LocalGatewayStatus) -> Option<u64> {
    // Try to extract uptime from raw status data
    status
        .raw
        .get("process")
        .and_then(|p| p.get("uptimeSeconds"))
        .and_then(Value::as_u64)
}

/// Perform comprehensive health check on ZCLAW Kernel
// @connected
#[tauri::command]
pub fn zclaw_health_check(
    app: AppHandle,
    port: Option<u16>,
    timeout_ms: Option<u64>,
) -> Result<HealthCheckResponse, String> {
    let check_port = port.unwrap_or(ZCLAW_DEFAULT_PORT);
    let timeout = timeout_ms.unwrap_or(3000);
    let mut checks_performed = Vec::new();
    let mut issues = Vec::new();

    // Get current timestamp
    let last_check_timestamp = SystemTime::now()
        .duration_since(UNIX_EPOCH)
        .map(|d| d.as_secs())
        .unwrap_or(0);

    // 1. Check if ZCLAW CLI is available
    let runtime = resolve_zclaw_runtime(&app);
    let cli_available = runtime.executable.is_file();

    if !cli_available {
        return Ok(HealthCheckResponse {
            status: HealthStatus::Unhealthy,
            process: ProcessHealthDetails {
                pid: None,
                name: None,
                status: None,
                uptime_seconds: None,
                cpu_percent: None,
                memory_mb: None,
            },
            port_check: PortCheckResult {
                port: check_port,
                accessible: false,
                latency_ms: None,
                error: Some("ZCLAW CLI not available".to_string()),
            },
            last_check_timestamp,
            checks_performed: vec!["cli_availability".to_string()],
            issues: vec![format!(
                "ZCLAW runtime not found at: {}",
                runtime.display_path.display()
            )],
            runtime_source: Some(runtime.source),
        });
    }
    checks_performed.push("cli_availability".to_string());

    // 2. Get gateway status
    let gateway_status = read_gateway_status(&app)?;
    checks_performed.push("gateway_status".to_string());

    // Check for configuration issues
    if !gateway_status.config_ok {
        issues.push("Gateway configuration has issues".to_string());
    }

    // 3. Check port accessibility
    let port_check = check_port_accessibility("127.0.0.1", check_port, timeout);
    checks_performed.push("port_accessibility".to_string());

    if !port_check.accessible {
        issues.push(format!(
            "Port {} is not accessible: {}",
            check_port,
            port_check.error.as_deref().unwrap_or("unknown error")
        ));
    }

    // 4. Extract process information
    let process_health = if !gateway_status.listener_pids.is_empty() {
        // Get the first listener PID
        let pid = gateway_status.listener_pids[0];

        // Try to get detailed process info from process list
        let process_info = run_zclaw(&app, &["process", "list", "--json"])
            .ok()
            .and_then(|result| parse_json_output(&result.stdout).ok())
            .and_then(|json| json.get("processes").and_then(Value::as_array).cloned());

        let (cpu, memory, uptime) = if let Some(ref processes) = process_info {
            let matching = processes
                .iter()
                .find(|p| p.get("pid").and_then(Value::as_u64) == Some(pid as u64));

            matching.map_or((None, None, None), |p| {
                (
                    p.get("cpuPercent").and_then(Value::as_f64),
                    p.get("memoryMb").and_then(Value::as_f64),
                    p.get("uptimeSeconds").and_then(Value::as_u64),
                )
            })
        } else {
            (None, None, get_process_uptime(&gateway_status))
        };

        ProcessHealthDetails {
            pid: Some(pid),
            name: Some("zclaw".to_string()),
            status: Some(
                gateway_status
                    .service_status
                    .clone()
                    .unwrap_or_else(|| "running".to_string()),
            ),
            uptime_seconds: uptime,
            cpu_percent: cpu,
            memory_mb: memory,
        }
    } else {
        ProcessHealthDetails {
            pid: None,
            name: None,
            status: gateway_status.service_status.clone(),
            uptime_seconds: None,
            cpu_percent: None,
            memory_mb: None,
        }
    };

    // Check if process is running but no listeners
    if gateway_status.service_status.as_deref() == Some("running")
        && gateway_status.listener_pids.is_empty()
    {
        issues.push("Service reports running but no listener processes found".to_string());
    }

    // 5. Determine overall health status
    let status = if !cli_available {
        HealthStatus::Unhealthy
    } else if !port_check.accessible {
        HealthStatus::Unhealthy
    } else if gateway_status.listener_pids.is_empty() {
        HealthStatus::Unhealthy
    } else if !issues.is_empty() {
        // Has some issues but core functionality is working
        HealthStatus::Healthy
    } else {
        HealthStatus::Healthy
    };

    Ok(HealthCheckResponse {
        status,
        process: process_health,
        port_check,
        last_check_timestamp,
        checks_performed,
        issues,
        runtime_source: Some(runtime.source),
    })
}

/// @reserved — no frontend UI yet
/// Quick ping to check if ZCLAW is alive (lightweight check)
#[tauri::command]
pub fn zclaw_ping(app: AppHandle) -> Result<bool, String> {
    let port_check = check_port_accessibility("127.0.0.1", ZCLAW_DEFAULT_PORT, 1000);

    if port_check.accessible {
        return Ok(true);
    }

    // Fallback: check via status command
    match run_zclaw(&app, &["gateway", "status", "--json", "--no-probe"]) {
        Ok(result) => {
            if let Ok(status) = parse_json_output(&result.stdout) {
                // Check if there are any listener PIDs
                let has_listeners = status
                    .get("port")
                    .and_then(|p| p.get("listeners"))
                    .and_then(Value::as_array)
                    .map(|arr| !arr.is_empty())
                    .unwrap_or(false);

                Ok(has_listeners)
            } else {
                Ok(false)
            }
        }
        Err(_) => Ok(false),
    }
}