Media Deduper Pro CLI commands

media-dedup-cli.php

<?php
/*
Plugin Name: Media Deduper Pro CLI
Plugin URI: https://avu.nu/
Description: Provides WP-CLI commands for Media Deduper Pro.
Version: 1.0
Author: Avunu LLC
Author URI: https://avu.nu/
*/

if (!class_exists('WP_CLI')) {
    return;
}

/**
 * Commands for the Media Deduper Pro Plugin.
 */
class Media_Deduper_Pro_Command extends WP_CLI_Command
{
    /**
     * Run the media deduplication process.
     */
    public function run_indexer()
    {

        // Fetch all media post IDs that are not already indexed
        $media_query_args = array(
            'post_type' => 'attachment',
            'post_status' => 'inherit',
            'posts_per_page' => -1,
            'fields' => 'ids',
            'meta_query' => array(
                array(
                    'key' => 'mdd_hash',
                    'compare' => 'NOT EXISTS' // This will only get media without an 'mdd_hash' meta key.
                ),
            ),
        );

        $media_query = new WP_Query($media_query_args);
        $media_ids = $media_query->posts;

        // check if the number of media items is greater than 0
        if (count($media_ids) == 0) {
            WP_CLI::success("No media items to process.");
            return;
        }

        // Provide feedback
        WP_CLI::log("Found " . count($media_ids) . " media items to process.");

        // Initialize the MDD Pro class
        $media_deduper_pro = new Media_Deduper_Pro();

        // Progress bar
        $progress = WP_CLI\Utils\make_progress_bar("Processing", count($media_ids));

        // Loop through media IDs and process
        foreach ($media_ids as $media_id) {

            $result = $media_deduper_pro->calc_media_meta($media_id);

            if (is_wp_error($result)) {
                WP_CLI::warning($result->get_error_message());
            } elseif ($result !== true) {
                WP_CLI::warning("Unexpected result for media ID: " . $media_id);
            }

            // Update progress
            $progress->tick();
        }

        // Finish the progress bar
        $progress->finish();
        WP_CLI::success("Finished processing all media items.");
    }

    /**
     * Run the smart bulk delete process.
     */
    public function smart_bulk_delete()
    {

        // Initialize your Media_Deduper_Pro instance if not already done.
        $media_deduper = new Media_Deduper_Pro();

        // Check if bulk processing is already running.
        if ($media_deduper->is_bulk_processing()) {
            WP_CLI::error(__('There is another Bulk Process already running. Please wait until that process is complete or stop it!', 'media-deduper'));
            return;
        }

        // Fetch duplicate IDs.
        $attachments = $this->get_duplicate_ids();
        $total_count = count($attachments);

        // If there are no duplicates.
        if ($total_count < 1) {
            WP_CLI::error(__('There are no duplicates on this site!', 'media-deduper'));
            return;
        }

        // Provide feedback
        WP_CLI::log("Found " . $total_count . " duplicate media items to process.");

        // Progress bar
        $progress = WP_CLI\Utils\make_progress_bar("Processing", $total_count);

        // Loop through media IDs and process
        foreach ($attachments as $attachment) {

            // Delete the duplicate.
            $media_deduper->bulk_delete_unused_item($attachment);

            // Update progress
            $progress->tick();
        }

        // Finish the progress bar
        $progress->finish();

        // Provide feedback
        WP_CLI::success("Finished processing all duplicate media items.");

    }

    /**
     * Retrieves an array of post ids that have duplicate hashes, 
     * returning a single ID for each group of duplicates.
     *
     * @return array
     */
    private function get_duplicate_ids()
    {
        global $wpdb;

        $sql = "SELECT DISTINCT p.post_id
            FROM $wpdb->postmeta AS p
            JOIN (
                SELECT count(*) AS dupe_count, meta_value
                FROM $wpdb->postmeta
                WHERE meta_key = 'mdd_hash'
                AND meta_value != '" . Media_Deduper_Pro::NOT_FOUND_HASH . "'
                GROUP BY meta_value
                HAVING dupe_count > 1
            ) AS p2
            ON p.meta_value = p2.meta_value
            GROUP BY p.meta_value;";

        // debug, write the query to a file
        file_put_contents( dirname( __FILE__ ) . '/query.txt', $sql );

        $duplicate_ids = $wpdb->get_col($sql);

        // If the query returns an empty array, replace it with an array containing a single 0.
        if (!count($duplicate_ids)) {
            $duplicate_ids = array('0');
        }

        return $duplicate_ids;
    }


}

WP_CLI::add_command('media-deduper-pro', 'Media_Deduper_Pro_Command');