homeproz/wp-content/plugins/mls-by-hansonxyz/includes/class-mls-garbage-collector.php

<?php
/**
 * MLS Garbage Collector
 *
 * Cleans up old MLS image directories when disk space is low.
 * Runs after sync to prevent disk from filling up.
 *
 * IMPORTANT: Only cleans the standard cache directory (mls-listings).
 * The persistent cache directory (mls-listings-persistent) is NEVER touched.
 * HomeProz listing images are stored in persistent cache and preserved
 * even after listings are sold or removed from MLS.
 *
 * Configuration (wp-config.php):
 * - MLS_GC_DISK_THRESHOLD: Minimum free disk space in bytes before cleanup triggers
 *   Example: define('MLS_GC_DISK_THRESHOLD', 5 * 1024 * 1024 * 1024); // 5GB
 *
 * Behavior:
 * - Only runs if MLS_GC_DISK_THRESHOLD is defined
 * - Only cleans standard cache (mls-listings), never persistent cache
 * - Skips directories modified within the last 24 hours
 * - Deletes oldest directories first
 * - Stops when free space >= 5GB or 2GB deleted per run
 */

if (!defined('ABSPATH')) {
    exit;
}

class MLS_Garbage_Collector {

    /**
     * Target free space to achieve (5GB)
     */
    const TARGET_FREE_SPACE = 5368709120; // 5 * 1024 * 1024 * 1024

    /**
     * Maximum bytes to delete per run (2GB)
     */
    const MAX_DELETE_PER_RUN = 2147483648; // 2 * 1024 * 1024 * 1024

    /**
     * Minimum age of directories to delete (24 hours)
     */
    const MIN_AGE_SECONDS = 86400; // 24 * 60 * 60

    /**
     * Logger instance
     */
    private $logger;

    /**
     * Database instance
     */
    private $db;

    /**
     * Constructor
     *
     * @param MLS_Logger $logger Logger instance
     * @param MLS_DB|null $db Database instance (optional for backwards compatibility)
     */
    public function __construct(MLS_Logger $logger, MLS_DB $db = null) {
        $this->logger = $logger;
        $this->db = $db;
    }

    /**
     * Check if garbage collection is enabled
     *
     * @return bool True if MLS_GC_DISK_THRESHOLD is defined
     */
    public function is_enabled() {
        return defined('MLS_GC_DISK_THRESHOLD') && MLS_GC_DISK_THRESHOLD > 0;
    }

    /**
     * Get the disk threshold from config
     *
     * @return int Threshold in bytes, or 0 if not defined
     */
    public function get_threshold() {
        return defined('MLS_GC_DISK_THRESHOLD') ? (int) MLS_GC_DISK_THRESHOLD : 0;
    }

    /**
     * Get the MLS images upload directory (standard cache only)
     *
     * Returns the standard cache directory that is subject to garbage collection.
     * The persistent cache (mls-listings-persistent) is intentionally excluded
     * to preserve HomeProz listing images indefinitely.
     *
     * @return string Absolute path to MLS images directory
     */
    public function get_images_dir() {
        $upload_dir = wp_upload_dir();
        return $upload_dir['basedir'] . '/mls-listings';
    }

    /**
     * Get free disk space for the volume hosting MLS images
     *
     * @return int|false Free space in bytes, or false on error
     */
    public function get_free_space() {
        $dir = $this->get_images_dir();

        // Create directory if it doesn't exist
        if (!file_exists($dir)) {
            wp_mkdir_p($dir);
        }

        return disk_free_space($dir);
    }

    /**
     * Check if cleanup is needed
     *
     * @return bool True if free space is below threshold
     */
    public function needs_cleanup() {
        if (!$this->is_enabled()) {
            return false;
        }

        $free_space = $this->get_free_space();
        if ($free_space === false) {
            return false;
        }

        return $free_space < $this->get_threshold();
    }

    /**
     * Get listing directories sorted by modification time (oldest first)
     *
     * Only returns directories older than MIN_AGE_SECONDS.
     *
     * @return array Array of ['path' => string, 'mtime' => int, 'size' => int]
     */
    public function get_old_directories() {
        $base_dir = $this->get_images_dir();
        $cutoff_time = time() - self::MIN_AGE_SECONDS;
        $directories = array();

        if (!is_dir($base_dir)) {
            return $directories;
        }

        // Iterate through prefix directories (2-char subdirs like "AB", "CD")
        $prefix_dirs = glob($base_dir . '/*', GLOB_ONLYDIR);
        if (!$prefix_dirs) {
            return $directories;
        }

        foreach ($prefix_dirs as $prefix_dir) {
            // Iterate through listing directories within each prefix
            $listing_dirs = glob($prefix_dir . '/*', GLOB_ONLYDIR);
            if (!$listing_dirs) {
                continue;
            }

            foreach ($listing_dirs as $listing_dir) {
                $mtime = filemtime($listing_dir);

                // Skip if modified within the last 24 hours
                if ($mtime >= $cutoff_time) {
                    continue;
                }

                $directories[] = array(
                    'path' => $listing_dir,
                    'mtime' => $mtime,
                    'size' => $this->get_directory_size($listing_dir),
                );
            }
        }

        // Sort by modification time (oldest first)
        usort($directories, function($a, $b) {
            return $a['mtime'] - $b['mtime'];
        });

        return $directories;
    }

    /**
     * Get total size of a directory
     *
     * @param string $dir Directory path
     * @return int Size in bytes
     */
    private function get_directory_size($dir) {
        $size = 0;
        $files = new RecursiveIteratorIterator(
            new RecursiveDirectoryIterator($dir, RecursiveDirectoryIterator::SKIP_DOTS),
            RecursiveIteratorIterator::LEAVES_ONLY
        );

        foreach ($files as $file) {
            if ($file->isFile()) {
                $size += $file->getSize();
            }
        }

        return $size;
    }

    /**
     * Delete a directory and all its contents
     *
     * @param string $dir Directory path
     * @return bool True on success
     */
    private function delete_directory($dir) {
        if (!is_dir($dir)) {
            return false;
        }

        $files = new RecursiveIteratorIterator(
            new RecursiveDirectoryIterator($dir, RecursiveDirectoryIterator::SKIP_DOTS),
            RecursiveIteratorIterator::CHILD_FIRST
        );

        foreach ($files as $file) {
            if ($file->isDir()) {
                @rmdir($file->getRealPath());
            } else {
                @unlink($file->getRealPath());
            }
        }

        return @rmdir($dir);
    }

    /**
     * Run garbage collection
     *
     * Deletes old directories until free space >= TARGET_FREE_SPACE
     * or MAX_DELETE_PER_RUN bytes have been deleted.
     *
     * @param callable|null $status_callback Optional callback for status messages
     * @return array Results with 'deleted_count', 'deleted_bytes', 'free_space_before', 'free_space_after'
     */
    public function run($status_callback = null) {
        $result = array(
            'enabled' => $this->is_enabled(),
            'ran' => false,
            'deleted_count' => 0,
            'deleted_bytes' => 0,
            'free_space_before' => 0,
            'free_space_after' => 0,
            'threshold' => $this->get_threshold(),
        );

        // Check if enabled
        if (!$this->is_enabled()) {
            if ($status_callback) {
                $status_callback('Garbage collection disabled (MLS_GC_DISK_THRESHOLD not defined)', 'info');
            }
            return $result;
        }

        $free_space = $this->get_free_space();
        if ($free_space === false) {
            if ($status_callback) {
                $status_callback('Could not determine free disk space', 'warning');
            }
            return $result;
        }

        $result['free_space_before'] = $free_space;
        $threshold = $this->get_threshold();

        // Check if cleanup needed
        if ($free_space >= $threshold) {
            if ($status_callback) {
                $status_callback(sprintf(
                    'Disk space OK: %s free (threshold: %s)',
                    $this->format_bytes($free_space),
                    $this->format_bytes($threshold)
                ), 'info');
            }
            $result['free_space_after'] = $free_space;
            return $result;
        }

        $result['ran'] = true;

        if ($status_callback) {
            $status_callback(sprintf(
                'Disk space low: %s free (threshold: %s). Starting cleanup...',
                $this->format_bytes($free_space),
                $this->format_bytes($threshold)
            ), 'warning');
        }

        // Get old directories
        $directories = $this->get_old_directories();

        if (empty($directories)) {
            if ($status_callback) {
                $status_callback('No directories older than 24 hours found for cleanup', 'info');
            }
            $result['free_space_after'] = $free_space;
            return $result;
        }

        $deleted_bytes = 0;
        $deleted_count = 0;

        foreach ($directories as $dir_info) {
            // Stop if we've reached target free space
            $current_free = $this->get_free_space();
            if ($current_free !== false && $current_free >= self::TARGET_FREE_SPACE) {
                if ($status_callback) {
                    $status_callback(sprintf(
                        'Target free space reached: %s',
                        $this->format_bytes($current_free)
                    ), 'info');
                }
                break;
            }

            // Stop if we've deleted enough this run
            if ($deleted_bytes >= self::MAX_DELETE_PER_RUN) {
                if ($status_callback) {
                    $status_callback(sprintf(
                        'Max deletion limit reached: %s',
                        $this->format_bytes($deleted_bytes)
                    ), 'info');
                }
                break;
            }

            // Delete the directory
            $path = $dir_info['path'];
            $size = $dir_info['size'];
            $listing_key = basename($path);

            if ($this->delete_directory($path)) {
                $deleted_bytes += $size;
                $deleted_count++;

                // Reset download_status to 'pending' for this listing's media
                // so images can be re-downloaded on demand later
                $this->reset_media_download_status($listing_key);

                $this->logger->info('Garbage collection deleted directory', array(
                    'listing_key' => $listing_key,
                    'size' => $size,
                    'age_days' => round((time() - $dir_info['mtime']) / 86400, 1),
                ));

                if ($status_callback) {
                    $status_callback(sprintf(
                        'Deleted: %s (%s)',
                        $listing_key,
                        $this->format_bytes($size)
                    ), 'info');
                }
            }
        }

        // Clean up empty prefix directories
        $this->cleanup_empty_prefix_dirs();

        $result['deleted_count'] = $deleted_count;
        $result['deleted_bytes'] = $deleted_bytes;
        $result['free_space_after'] = $this->get_free_space();

        if ($status_callback) {
            $status_callback(sprintf(
                'Cleanup complete: Deleted %d directories (%s). Free space now: %s',
                $deleted_count,
                $this->format_bytes($deleted_bytes),
                $this->format_bytes($result['free_space_after'])
            ), 'info');
        }

        $this->logger->info('Garbage collection completed', array(
            'deleted_count' => $deleted_count,
            'deleted_bytes' => $deleted_bytes,
            'free_space_before' => $result['free_space_before'],
            'free_space_after' => $result['free_space_after'],
        ));

        return $result;
    }

    /**
     * Reset download_status to 'pending' for a listing's media records
     *
     * Called after deleting cached files so images can be re-downloaded on demand.
     *
     * @param string $listing_key Listing key
     */
    private function reset_media_download_status($listing_key) {
        global $wpdb;

        // Get the media table name
        $media_table = $this->db ? $this->db->media_table() : $wpdb->prefix . 'mls_media';

        $updated = $wpdb->update(
            $media_table,
            array(
                'download_status' => 'pending',
                'local_path' => null,
                'local_url' => null,
                'downloaded_at' => null,
            ),
            array('listing_key' => $listing_key),
            array('%s', null, null, null),
            array('%s')
        );

        if ($updated > 0) {
            $this->logger->debug('Reset media download status for garbage collected listing', array(
                'listing_key' => $listing_key,
                'records_updated' => $updated,
            ));
        }
    }

    /**
     * Clean up empty prefix directories
     */
    private function cleanup_empty_prefix_dirs() {
        $base_dir = $this->get_images_dir();
        $prefix_dirs = glob($base_dir . '/*', GLOB_ONLYDIR);

        if (!$prefix_dirs) {
            return;
        }

        foreach ($prefix_dirs as $prefix_dir) {
            // Check if directory is empty
            $contents = glob($prefix_dir . '/*');
            if (empty($contents)) {
                @rmdir($prefix_dir);
            }
        }
    }

    /**
     * Format bytes to human readable string
     *
     * @param int $bytes Bytes
     * @return string Formatted string (e.g., "1.5 GB")
     */
    private function format_bytes($bytes) {
        if ($bytes >= 1073741824) {
            return number_format($bytes / 1073741824, 2) . ' GB';
        } elseif ($bytes >= 1048576) {
            return number_format($bytes / 1048576, 2) . ' MB';
        } elseif ($bytes >= 1024) {
            return number_format($bytes / 1024, 2) . ' KB';
        }
        return $bytes . ' bytes';
    }

    /**
     * Get statistics about the image cache
     *
     * @return array Stats including total_size, directory_count, oldest_mtime
     */
    public function get_stats() {
        $base_dir = $this->get_images_dir();
        $stats = array(
            'total_size' => 0,
            'directory_count' => 0,
            'oldest_mtime' => null,
            'newest_mtime' => null,
            'free_space' => $this->get_free_space(),
            'threshold' => $this->get_threshold(),
            'needs_cleanup' => $this->needs_cleanup(),
        );

        if (!is_dir($base_dir)) {
            return $stats;
        }

        $prefix_dirs = glob($base_dir . '/*', GLOB_ONLYDIR);
        if (!$prefix_dirs) {
            return $stats;
        }

        foreach ($prefix_dirs as $prefix_dir) {
            $listing_dirs = glob($prefix_dir . '/*', GLOB_ONLYDIR);
            if (!$listing_dirs) {
                continue;
            }

            foreach ($listing_dirs as $listing_dir) {
                $stats['directory_count']++;
                $stats['total_size'] += $this->get_directory_size($listing_dir);

                $mtime = filemtime($listing_dir);
                if ($stats['oldest_mtime'] === null || $mtime < $stats['oldest_mtime']) {
                    $stats['oldest_mtime'] = $mtime;
                }
                if ($stats['newest_mtime'] === null || $mtime > $stats['newest_mtime']) {
                    $stats['newest_mtime'] = $mtime;
                }
            }
        }

        return $stats;
    }
}