path: root/src/cleanup.rs

use anyhow::{Context as _, Result};
use std::path::Path;
use tracing::{debug, info, warn};

/// Result of a cleanup operation.
#[derive(Debug, Default)]
pub struct CleanupResult {
    /// Number of build directories removed.
    pub builds_removed: u32,
    /// Number of log files removed.
    pub logs_removed: u32,
}

/// Clean up old build directories and their corresponding log files.
///
/// Keeps the `max_to_keep` most recent builds and removes older ones.
/// Also removes the corresponding log files for each removed build.
///
/// # Arguments
/// * `base_dir` - Base witryna directory (e.g., /var/lib/witryna)
/// * `log_dir` - Log directory (e.g., /var/log/witryna)
/// * `site_name` - The site name
/// * `max_to_keep` - Maximum number of builds to keep (0 = keep all)
///
/// # Errors
///
/// Returns an error if the builds directory cannot be listed. Individual
/// removal failures are logged as warnings but do not cause the function
/// to return an error.
pub async fn cleanup_old_builds(
    base_dir: &Path,
    log_dir: &Path,
    site_name: &str,
    max_to_keep: u32,
) -> Result<CleanupResult> {
    // If max_to_keep is 0, keep all builds
    if max_to_keep == 0 {
        debug!(%site_name, "max_builds_to_keep is 0, skipping cleanup");
        return Ok(CleanupResult::default());
    }

    let builds_dir = base_dir.join("builds").join(site_name);
    let site_log_dir = log_dir.join(site_name);

    // Check if builds directory exists
    if !builds_dir.exists() {
        debug!(%site_name, "builds directory does not exist, skipping cleanup");
        return Ok(CleanupResult::default());
    }

    // List all build directories (excluding 'current' symlink)
    let mut build_timestamps = list_build_timestamps(&builds_dir).await?;

    // Sort in descending order (newest first)
    build_timestamps.sort_by(|a, b| b.cmp(a));

    let mut result = CleanupResult::default();

    // Calculate how many to remove
    let to_remove = build_timestamps.len().saturating_sub(max_to_keep as usize);
    if to_remove == 0 {
        debug!(%site_name, count = build_timestamps.len(), max = max_to_keep, "no builds to remove");
    }

    // Remove oldest builds (they're at the end after reverse sort)
    for timestamp in build_timestamps.iter().skip(max_to_keep as usize) {
        let build_path = builds_dir.join(timestamp);
        let log_path = site_log_dir.join(format!("{timestamp}.log"));

        // Remove build directory
        match tokio::fs::remove_dir_all(&build_path).await {
            Ok(()) => {
                debug!(path = %build_path.display(), "removed old build");
                result.builds_removed += 1;
            }
            Err(e) => {
                warn!(path = %build_path.display(), error = %e, "failed to remove old build");
            }
        }

        // Remove corresponding log file (if exists)
        if log_path.exists() {
            match tokio::fs::remove_file(&log_path).await {
                Ok(()) => {
                    debug!(path = %log_path.display(), "removed old log");
                    result.logs_removed += 1;
                }
                Err(e) => {
                    warn!(path = %log_path.display(), error = %e, "failed to remove old log");
                }
            }
        }

        // Remove corresponding hook log file (if exists)
        let hook_log_path = site_log_dir.join(format!("{timestamp}-hook.log"));
        match tokio::fs::remove_file(&hook_log_path).await {
            Ok(()) => {
                debug!(path = %hook_log_path.display(), "removed old hook log");
                result.logs_removed += 1;
            }
            Err(e) if e.kind() == std::io::ErrorKind::NotFound => {
                // Not every build has a hook — silently skip
            }
            Err(e) => {
                warn!(path = %hook_log_path.display(), error = %e, "failed to remove old hook log");
            }
        }
    }

    // Remove orphaned temp files (crash recovery)
    if site_log_dir.exists()
        && let Ok(mut entries) = tokio::fs::read_dir(&site_log_dir).await
    {
        while let Ok(Some(entry)) = entries.next_entry().await {
            let name = entry.file_name();
            if name.to_string_lossy().ends_with(".tmp") {
                let path = entry.path();
                match tokio::fs::remove_file(&path).await {
                    Ok(()) => {
                        debug!(path = %path.display(), "removed orphaned temp file");
                    }
                    Err(e) => {
                        warn!(path = %path.display(), error = %e, "failed to remove orphaned temp file");
                    }
                }
            }
        }
    }

    if result.builds_removed > 0 || result.logs_removed > 0 {
        info!(
            %site_name,
            builds_removed = result.builds_removed,
            logs_removed = result.logs_removed,
            "cleanup completed"
        );
    }

    Ok(result)
}

/// List all build timestamps in a builds directory.
///
/// Returns directory names that look like timestamps, excluding 'current' symlink.
async fn list_build_timestamps(builds_dir: &Path) -> Result<Vec<String>> {
    let mut timestamps = Vec::new();

    let mut entries = tokio::fs::read_dir(builds_dir)
        .await
        .with_context(|| format!("failed to read builds directory: {}", builds_dir.display()))?;

    while let Some(entry) = entries.next_entry().await? {
        let name = entry.file_name();
        let name_str = name.to_string_lossy();

        // Skip 'current' symlink and any other non-timestamp entries
        if name_str == "current" {
            continue;
        }

        // Verify it's a directory (not a file or broken symlink)
        let file_type = entry.file_type().await?;
        if !file_type.is_dir() {
            continue;
        }

        // Basic timestamp format validation: YYYYMMDD-HHMMSS-...
        if looks_like_timestamp(&name_str) {
            timestamps.push(name_str.to_string());
        }
    }

    Ok(timestamps)
}

/// Check if a string looks like a valid timestamp format.
///
/// Expected format: YYYYMMDD-HHMMSS-microseconds (e.g., 20260126-143000-123456)
fn looks_like_timestamp(s: &str) -> bool {
    let parts: Vec<&str> = s.split('-').collect();
    let [date, time, micros, ..] = parts.as_slice() else {
        return false;
    };

    // First part should be 8 digits (YYYYMMDD)
    if date.len() != 8 || !date.chars().all(|c| c.is_ascii_digit()) {
        return false;
    }

    // Second part should be 6 digits (HHMMSS)
    if time.len() != 6 || !time.chars().all(|c| c.is_ascii_digit()) {
        return false;
    }

    // Third part should be microseconds (digits)
    if micros.is_empty() || !micros.chars().all(|c| c.is_ascii_digit()) {
        return false;
    }

    true
}

#[cfg(test)]
#[allow(clippy::unwrap_used, clippy::indexing_slicing)]
mod tests {
    use super::*;
    use crate::test_support::{cleanup, temp_dir};
    use tokio::fs;

    async fn create_build_and_log(base_dir: &Path, log_dir: &Path, site: &str, timestamp: &str) {
        let build_dir = base_dir.join("builds").join(site).join(timestamp);
        let site_log_dir = log_dir.join(site);
        let log_file = site_log_dir.join(format!("{timestamp}.log"));

        fs::create_dir_all(&build_dir).await.unwrap();
        fs::create_dir_all(&site_log_dir).await.unwrap();
        fs::write(&log_file, "test log content").await.unwrap();
        fs::write(build_dir.join("index.html"), "<html></html>")
            .await
            .unwrap();
    }

    #[tokio::test]
    async fn cleanup_removes_old_builds_and_logs() {
        let base_dir = temp_dir("cleanup-test").await;
        let log_dir = base_dir.join("logs");
        let site = "test-site";

        // Create 7 builds (keep 5, remove 2)
        let timestamps = [
            "20260126-100000-000001",
            "20260126-100000-000002",
            "20260126-100000-000003",
            "20260126-100000-000004",
            "20260126-100000-000005",
            "20260126-100000-000006",
            "20260126-100000-000007",
        ];

        for ts in &timestamps {
            create_build_and_log(&base_dir, &log_dir, site, ts).await;
        }

        let result = cleanup_old_builds(&base_dir, &log_dir, site, 5).await;
        assert!(result.is_ok(), "cleanup should succeed: {result:?}");
        let result = result.unwrap();

        assert_eq!(result.builds_removed, 2, "should remove 2 builds");
        assert_eq!(result.logs_removed, 2, "should remove 2 logs");

        // Verify oldest 2 are gone
        let builds_dir = base_dir.join("builds").join(site);
        assert!(!builds_dir.join("20260126-100000-000001").exists());
        assert!(!builds_dir.join("20260126-100000-000002").exists());

        // Verify newest 5 remain
        assert!(builds_dir.join("20260126-100000-000003").exists());
        assert!(builds_dir.join("20260126-100000-000007").exists());

        // Verify log cleanup
        let site_logs = log_dir.join(site);
        assert!(!site_logs.join("20260126-100000-000001.log").exists());
        assert!(site_logs.join("20260126-100000-000003.log").exists());

        cleanup(&base_dir).await;
    }

    #[tokio::test]
    async fn cleanup_with_fewer_builds_than_max() {
        let base_dir = temp_dir("cleanup-test").await;
        let log_dir = base_dir.join("logs");
        let site = "test-site";

        // Create only 3 builds (max is 5)
        for ts in &[
            "20260126-100000-000001",
            "20260126-100000-000002",
            "20260126-100000-000003",
        ] {
            create_build_and_log(&base_dir, &log_dir, site, ts).await;
        }

        let result = cleanup_old_builds(&base_dir, &log_dir, site, 5).await;
        assert!(result.is_ok());
        let result = result.unwrap();

        assert_eq!(result.builds_removed, 0, "should not remove any builds");
        assert_eq!(result.logs_removed, 0, "should not remove any logs");

        // Verify all builds remain
        let builds_dir = base_dir.join("builds").join(site);
        assert!(builds_dir.join("20260126-100000-000001").exists());
        assert!(builds_dir.join("20260126-100000-000002").exists());
        assert!(builds_dir.join("20260126-100000-000003").exists());

        cleanup(&base_dir).await;
    }

    #[tokio::test]
    async fn cleanup_preserves_current_symlink() {
        let base_dir = temp_dir("cleanup-test").await;
        let log_dir = base_dir.join("logs");
        let site = "test-site";

        // Create builds
        create_build_and_log(&base_dir, &log_dir, site, "20260126-100000-000001").await;
        create_build_and_log(&base_dir, &log_dir, site, "20260126-100000-000002").await;
        create_build_and_log(&base_dir, &log_dir, site, "20260126-100000-000003").await;

        // Create 'current' symlink
        let builds_dir = base_dir.join("builds").join(site);
        let current = builds_dir.join("current");
        let target = builds_dir.join("20260126-100000-000003");
        tokio::fs::symlink(&target, &current).await.unwrap();

        let result = cleanup_old_builds(&base_dir, &log_dir, site, 2).await;
        assert!(result.is_ok());
        let result = result.unwrap();

        assert_eq!(result.builds_removed, 1, "should remove 1 build");

        // Verify symlink still exists and points correctly
        assert!(current.exists(), "current symlink should exist");
        let link_target = fs::read_link(&current).await.unwrap();
        assert_eq!(link_target, target);

        cleanup(&base_dir).await;
    }

    #[tokio::test]
    async fn cleanup_handles_missing_logs_gracefully() {
        let base_dir = temp_dir("cleanup-test").await;
        let log_dir = base_dir.join("logs");
        let site = "test-site";

        // Create builds but only some logs
        let builds_dir = base_dir.join("builds").join(site);
        fs::create_dir_all(builds_dir.join("20260126-100000-000001"))
            .await
            .unwrap();
        fs::create_dir_all(builds_dir.join("20260126-100000-000002"))
            .await
            .unwrap();
        fs::create_dir_all(builds_dir.join("20260126-100000-000003"))
            .await
            .unwrap();

        // Only create log for one build
        let site_logs = log_dir.join(site);
        fs::create_dir_all(&site_logs).await.unwrap();
        fs::write(site_logs.join("20260126-100000-000001.log"), "log")
            .await
            .unwrap();

        let result = cleanup_old_builds(&base_dir, &log_dir, site, 2).await;
        assert!(result.is_ok(), "should succeed even with missing logs");
        let result = result.unwrap();

        assert_eq!(result.builds_removed, 1, "should remove 1 build");
        assert_eq!(result.logs_removed, 1, "should remove 1 log");

        cleanup(&base_dir).await;
    }

    #[tokio::test]
    async fn cleanup_with_max_zero_keeps_all() {
        let base_dir = temp_dir("cleanup-test").await;
        let log_dir = base_dir.join("logs");
        let site = "test-site";

        // Create builds
        for ts in &[
            "20260126-100000-000001",
            "20260126-100000-000002",
            "20260126-100000-000003",
        ] {
            create_build_and_log(&base_dir, &log_dir, site, ts).await;
        }

        let result = cleanup_old_builds(&base_dir, &log_dir, site, 0).await;
        assert!(result.is_ok());
        let result = result.unwrap();

        assert_eq!(result.builds_removed, 0, "max 0 should keep all");
        assert_eq!(result.logs_removed, 0);

        // Verify all builds remain
        let builds_dir = base_dir.join("builds").join(site);
        assert!(builds_dir.join("20260126-100000-000001").exists());
        assert!(builds_dir.join("20260126-100000-000002").exists());
        assert!(builds_dir.join("20260126-100000-000003").exists());

        cleanup(&base_dir).await;
    }

    #[tokio::test]
    async fn cleanup_nonexistent_builds_dir() {
        let base_dir = temp_dir("cleanup-test").await;
        let site = "nonexistent-site";

        let log_dir = base_dir.join("logs");
        let result = cleanup_old_builds(&base_dir, &log_dir, site, 5).await;
        assert!(result.is_ok(), "should succeed for nonexistent dir");
        let result = result.unwrap();

        assert_eq!(result.builds_removed, 0);
        assert_eq!(result.logs_removed, 0);

        cleanup(&base_dir).await;
    }

    #[tokio::test]
    async fn cleanup_removes_orphaned_tmp_files() {
        let base_dir = temp_dir("cleanup-test").await;
        let log_dir = base_dir.join("logs");
        let site = "test-site";

        // Create a build so cleanup runs
        create_build_and_log(&base_dir, &log_dir, site, "20260126-100000-000001").await;

        // Create orphaned temp files in site log dir
        let site_log_dir = log_dir.join(site);
        fs::write(
            site_log_dir.join("20260126-100000-000001-stdout.tmp"),
            "orphan",
        )
        .await
        .unwrap();
        fs::write(
            site_log_dir.join("20260126-100000-000001-stderr.tmp"),
            "orphan",
        )
        .await
        .unwrap();
        fs::write(site_log_dir.join("random.tmp"), "orphan")
            .await
            .unwrap();

        assert!(
            site_log_dir
                .join("20260126-100000-000001-stdout.tmp")
                .exists()
        );

        // Run cleanup (max_to_keep=5 means no builds removed, but tmp files should go)
        let result = cleanup_old_builds(&base_dir, &log_dir, site, 5).await;
        assert!(result.is_ok());

        // Temp files should be gone
        assert!(
            !site_log_dir
                .join("20260126-100000-000001-stdout.tmp")
                .exists()
        );
        assert!(
            !site_log_dir
                .join("20260126-100000-000001-stderr.tmp")
                .exists()
        );
        assert!(!site_log_dir.join("random.tmp").exists());

        // Log file should still exist
        assert!(site_log_dir.join("20260126-100000-000001.log").exists());

        cleanup(&base_dir).await;
    }
}