cm-dashboard/agent/src/collectors/smart.rs

use async_trait::async_trait;
use chrono::Utc;
use serde::{Deserialize, Serialize};
use serde_json::json;
use std::io::ErrorKind;
use std::process::Stdio;
use std::time::Duration;
use tokio::process::Command;
use tokio::time::timeout;

use super::{AgentType, Collector, CollectorError, CollectorOutput};

#[derive(Debug, Clone)]
pub struct SmartCollector {
    pub enabled: bool,
    pub interval: Duration,
    pub devices: Vec<String>,
    pub timeout_ms: u64,
}

impl SmartCollector {
    pub fn new(enabled: bool, interval_ms: u64, devices: Vec<String>) -> Self {
        Self {
            enabled,
            interval: Duration::from_millis(interval_ms),
            devices,
            timeout_ms: 30000, // 30 second timeout for smartctl
        }
    }

    async fn get_smart_data(&self, device: &str) -> Result<SmartDeviceData, CollectorError> {
        let timeout_duration = Duration::from_millis(self.timeout_ms);

        let command_result = timeout(
            timeout_duration,
            Command::new("smartctl")
                .args(["-a", "-j", &format!("/dev/{}", device)])
                .stdout(Stdio::piped())
                .stderr(Stdio::piped())
                .output(),
        )
        .await
        .map_err(|_| CollectorError::Timeout {
            duration_ms: self.timeout_ms,
        })?;

        let output = command_result.map_err(|e| match e.kind() {
            ErrorKind::NotFound => CollectorError::ExternalDependency {
                dependency: "smartctl".to_string(),
                message: e.to_string(),
            },
            ErrorKind::PermissionDenied => CollectorError::PermissionDenied {
                message: e.to_string(),
            },
            _ => CollectorError::CommandFailed {
                command: format!("smartctl -a -j /dev/{}", device),
                message: e.to_string(),
            },
        })?;

        if !output.status.success() {
            let stderr = String::from_utf8_lossy(&output.stderr);
            let stderr_lower = stderr.to_lowercase();

            if stderr_lower.contains("permission denied") {
                return Err(CollectorError::PermissionDenied {
                    message: stderr.to_string(),
                });
            }

            if stderr_lower.contains("no such device") || stderr_lower.contains("cannot open") {
                return Err(CollectorError::DeviceNotFound {
                    device: device.to_string(),
                });
            }

            return Err(CollectorError::CommandFailed {
                command: format!("smartctl -a -j /dev/{}", device),
                message: stderr.to_string(),
            });
        }

        let stdout = String::from_utf8_lossy(&output.stdout);
        let smart_output: SmartCtlOutput =
            serde_json::from_str(&stdout).map_err(|e| CollectorError::ParseError {
                message: format!("Failed to parse smartctl output for {}: {}", device, e),
            })?;

        Ok(SmartDeviceData::from_smartctl_output(device, smart_output))
    }

    async fn get_drive_usage(
        &self,
        device: &str,
    ) -> Result<(Option<f32>, Option<f32>), CollectorError> {
        // Get capacity first
        let capacity = match self.get_drive_capacity(device).await {
            Ok(cap) => Some(cap),
            Err(_) => None,
        };

        // Try to get usage information
        // For simplicity, we'll use the root filesystem usage for now
        // In the future, this could be enhanced to map drives to specific mount points
        let usage = if device.contains("nvme0n1") || device.contains("sda") {
            // This is likely the main system drive, use root filesystem usage
            match self.get_disk_usage().await {
                Ok(disk_usage) => Some(disk_usage.used_gb),
                Err(_) => None,
            }
        } else {
            // For other drives, we don't have usage info yet
            None
        };

        Ok((capacity, usage))
    }

    async fn get_drive_capacity(&self, device: &str) -> Result<f32, CollectorError> {
        let output = Command::new("lsblk")
            .args(["-J", "-o", "NAME,SIZE", &format!("/dev/{}", device)])
            .stdout(Stdio::piped())
            .stderr(Stdio::piped())
            .output()
            .await
            .map_err(|e| CollectorError::CommandFailed {
                command: format!("lsblk -J -o NAME,SIZE /dev/{}", device),
                message: e.to_string(),
            })?;

        if !output.status.success() {
            let stderr = String::from_utf8_lossy(&output.stderr);
            return Err(CollectorError::CommandFailed {
                command: format!("lsblk -J -o NAME,SIZE /dev/{}", device),
                message: stderr.to_string(),
            });
        }

        let stdout = String::from_utf8_lossy(&output.stdout);
        let lsblk_output: serde_json::Value =
            serde_json::from_str(&stdout).map_err(|e| CollectorError::ParseError {
                message: format!("Failed to parse lsblk JSON: {}", e),
            })?;

        // Extract size from the first blockdevice
        if let Some(blockdevices) = lsblk_output["blockdevices"].as_array() {
            if let Some(device_info) = blockdevices.first() {
                if let Some(size_str) = device_info["size"].as_str() {
                    return self.parse_lsblk_size(size_str);
                }
            }
        }

        Err(CollectorError::ParseError {
            message: format!("No size information found for device {}", device),
        })
    }

    fn parse_lsblk_size(&self, size_str: &str) -> Result<f32, CollectorError> {
        // Parse sizes like "953,9G", "1T", "512M"
        let size_str = size_str.replace(',', "."); // Handle European decimal separator

        if let Some(pos) = size_str.find(|c: char| c.is_alphabetic()) {
            let (number_part, unit_part) = size_str.split_at(pos);
            let number: f32 = number_part
                .parse()
                .map_err(|e| CollectorError::ParseError {
                    message: format!("Failed to parse size number '{}': {}", number_part, e),
                })?;

            let multiplier = match unit_part.to_uppercase().as_str() {
                "T" | "TB" => 1024.0,
                "G" | "GB" => 1.0,
                "M" | "MB" => 1.0 / 1024.0,
                "K" | "KB" => 1.0 / (1024.0 * 1024.0),
                _ => {
                    return Err(CollectorError::ParseError {
                        message: format!("Unknown size unit: {}", unit_part),
                    })
                }
            };

            Ok(number * multiplier)
        } else {
            Err(CollectorError::ParseError {
                message: format!("Invalid size format: {}", size_str),
            })
        }
    }

    async fn get_disk_usage(&self) -> Result<DiskUsage, CollectorError> {
        let output = Command::new("df")
            .args(["-BG", "--output=size,used,avail", "/"])
            .stdout(Stdio::piped())
            .stderr(Stdio::piped())
            .output()
            .await
            .map_err(|e| CollectorError::CommandFailed {
                command: "df -BG --output=size,used,avail /".to_string(),
                message: e.to_string(),
            })?;

        if !output.status.success() {
            let stderr = String::from_utf8_lossy(&output.stderr);
            return Err(CollectorError::CommandFailed {
                command: "df -BG --output=size,used,avail /".to_string(),
                message: stderr.to_string(),
            });
        }

        let stdout = String::from_utf8_lossy(&output.stdout);
        let lines: Vec<&str> = stdout.lines().collect();

        if lines.len() < 2 {
            return Err(CollectorError::ParseError {
                message: "Unexpected df output format".to_string(),
            });
        }

        // Skip header line, parse data line
        let data_line = lines[1].trim();
        let parts: Vec<&str> = data_line.split_whitespace().collect();

        if parts.len() < 3 {
            return Err(CollectorError::ParseError {
                message: format!("Unexpected df data format: {}", data_line),
            });
        }

        let parse_size = |s: &str| -> Result<f32, CollectorError> {
            s.trim_end_matches('G')
                .parse::<f32>()
                .map_err(|e| CollectorError::ParseError {
                    message: format!("Failed to parse disk size '{}': {}", s, e),
                })
        };

        Ok(DiskUsage {
            total_gb: parse_size(parts[0])?,
            used_gb: parse_size(parts[1])?,
            available_gb: parse_size(parts[2])?,
        })
    }
}

#[async_trait]
impl Collector for SmartCollector {
    fn name(&self) -> &str {
        "smart"
    }

    fn agent_type(&self) -> AgentType {
        AgentType::Smart
    }

    fn collect_interval(&self) -> Duration {
        self.interval
    }

    fn is_enabled(&self) -> bool {
        self.enabled
    }

    fn requires_root(&self) -> bool {
        true // smartctl typically requires root access
    }

    async fn collect(&self) -> Result<CollectorOutput, CollectorError> {
        let mut drives = Vec::new();
        let mut issues = Vec::new();
        let mut healthy = 0;
        let mut warning = 0;
        let mut critical = 0;

        // Collect data from all configured devices
        for device in &self.devices {
            match self.get_smart_data(device).await {
                Ok(mut drive_data) => {
                    // Try to get capacity and usage for this drive
                    if let Ok((capacity, usage)) = self.get_drive_usage(device).await {
                        drive_data.capacity_gb = capacity;
                        drive_data.used_gb = usage;
                    }
                    match drive_data.health_status.as_str() {
                        "PASSED" => healthy += 1,
                        "FAILED" => {
                            critical += 1;
                            issues.push(format!("{}: SMART status FAILED", device));
                        }
                        _ => {
                            warning += 1;
                            issues.push(format!("{}: Unknown SMART status", device));
                        }
                    }
                    drives.push(drive_data);
                }
                Err(e) => {
                    warning += 1;
                    issues.push(format!("{}: {}", device, e));
                }
            }
        }

        // Get disk usage information
        let disk_usage = self.get_disk_usage().await?;

        let status = if critical > 0 {
            "CRITICAL"
        } else if warning > 0 {
            "WARNING"
        } else {
            "HEALTHY"
        };

        let smart_metrics = json!({
            "status": status,
            "drives": drives,
            "summary": {
                "healthy": healthy,
                "warning": warning,
                "critical": critical,
                "capacity_total_gb": disk_usage.total_gb,
                "capacity_used_gb": disk_usage.used_gb,
                "capacity_available_gb": disk_usage.available_gb
            },
            "issues": issues,
            "timestamp": Utc::now()
        });

        Ok(CollectorOutput {
            agent_type: AgentType::Smart,
            data: smart_metrics,
            timestamp: Utc::now(),
        })
    }
}

#[derive(Debug, Clone, Serialize)]
struct SmartDeviceData {
    name: String,
    temperature_c: f32,
    wear_level: f32,
    power_on_hours: u64,
    available_spare: f32,
    health_status: String,
    capacity_gb: Option<f32>,
    used_gb: Option<f32>,
}

impl SmartDeviceData {
    fn from_smartctl_output(device: &str, output: SmartCtlOutput) -> Self {
        let temperature_c = output.temperature.and_then(|t| t.current).unwrap_or(0.0);

        let wear_level = output
            .nvme_smart_health_information_log
            .as_ref()
            .and_then(|nvme| nvme.percentage_used)
            .unwrap_or(0.0);

        let power_on_hours = output.power_on_time.and_then(|p| p.hours).unwrap_or(0);

        let available_spare = output
            .nvme_smart_health_information_log
            .as_ref()
            .and_then(|nvme| nvme.available_spare)
            .unwrap_or(100.0);

        let health_status = output
            .smart_status
            .and_then(|s| s.passed)
            .map(|passed| {
                if passed {
                    "PASSED".to_string()
                } else {
                    "FAILED".to_string()
                }
            })
            .unwrap_or_else(|| "UNKNOWN".to_string());

        Self {
            name: device.to_string(),
            temperature_c,
            wear_level,
            power_on_hours,
            available_spare,
            health_status,
            capacity_gb: None, // Will be set later by the collector
            used_gb: None,     // Will be set later by the collector
        }
    }
}

#[derive(Debug, Clone)]
struct DiskUsage {
    total_gb: f32,
    used_gb: f32,
    available_gb: f32,
}

// Minimal smartctl JSON output structure - only the fields we need
#[derive(Debug, Deserialize)]
struct SmartCtlOutput {
    temperature: Option<Temperature>,
    power_on_time: Option<PowerOnTime>,
    smart_status: Option<SmartStatus>,
    nvme_smart_health_information_log: Option<NvmeSmartLog>,
}

#[derive(Debug, Deserialize)]
struct Temperature {
    current: Option<f32>,
}

#[derive(Debug, Deserialize)]
struct PowerOnTime {
    hours: Option<u64>,
}

#[derive(Debug, Deserialize)]
struct SmartStatus {
    passed: Option<bool>,
}

#[derive(Debug, Deserialize)]
struct NvmeSmartLog {
    percentage_used: Option<f32>,
    available_spare: Option<f32>,
}

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

    #[test]
    fn test_parse_lsblk_size() {
        let collector = SmartCollector::new(true, 5000, vec![]);

        // Test gigabyte sizes
        assert!((collector.parse_lsblk_size("953,9G").unwrap() - 953.9).abs() < 0.1);
        assert!((collector.parse_lsblk_size("1G").unwrap() - 1.0).abs() < 0.1);

        // Test terabyte sizes
        assert!((collector.parse_lsblk_size("1T").unwrap() - 1024.0).abs() < 0.1);
        assert!((collector.parse_lsblk_size("2,5T").unwrap() - 2560.0).abs() < 0.1);

        // Test megabyte sizes
        assert!((collector.parse_lsblk_size("512M").unwrap() - 0.5).abs() < 0.1);

        // Test error cases
        assert!(collector.parse_lsblk_size("invalid").is_err());
        assert!(collector.parse_lsblk_size("1X").is_err());
    }
}