update-posts-urls.php

<?php

// Update new posts to have the correct publication URL

add_filter('wp_insert_post_data', 'update_post_file_to_append_proxy', 10, 1);

function update_post_file_to_append_proxy(array $data): array
{
    // We don't need to take actions for publications, any other post type we will proceeed.
    if($data['post_type'] == 'publications'){
        return $data;
    }

    // Extract the PDF URLs from the post content
    $pdfs = extract_pdf_urls($data['post_content']);

    if(empty($pdfs)){
        return $data;
    }

    // Loop through the PDF URLs and update the post content with the correct publication URL
    foreach ($pdfs as $pdf) {
        $attachment_id = lookup_pdf_id_by_url($pdf);

        if(!$attachment_id){
            continue;
        }

        // Lookup the publication ID by the attachment ID
        $publication_id = lookup_publication_id_by_attachment_id($attachment_id);

        if(!$publication_id){
            continue;
        }

        $new_url = "/download/publication/". $publication_id;

        $data['post_content'] = str_replace($pdf, $new_url, $data['post_content']);

    }

    return $data;

}

function extract_pdf_urls(string $post_content): array {

    $pdf_urls = [];

    $pattern = '/https?:\/\/[^\s<>"\'()]+?\.pdf\b/i';

    if (preg_match_all($pattern, $post_content, $matches)) {
        $pdf_urls = array_unique($matches[0]); // Remove duplicates
    }

    return array_values($pdf_urls);
}

function lookup_publication_id_by_attachment_id(int $attachment_id): ?int {
    $posts = get_posts([
        'post_type'      => 'publications',
        'posts_per_page' => 1,
        'meta_query'     => [
            [
                'key'   => 'pdf_file', // ACF field key
                'value' => $attachment_id,
                'type'  => 'NUMERIC',
            ],
        ],
        'fields'         => 'ids',
    ]);

    return !empty($posts) ? (int) $posts[0] : null;
}

function lookup_pdf_id_by_url(string $url): ?int {
    $url = esc_url_raw($url);

    $upload_dir = wp_upload_dir();
    $file_path = str_replace($upload_dir['baseurl'], '', $url);
    $file_path = trim($file_path, '/');

    $posts = get_posts([
        'post_type'      => 'attachment',
        'posts_per_page' => 1,
        'meta_query'     => [
            [
                'key'   => '_wp_attached_file',
                'value' => $file_path,
            ],
        ],
        'fields'         => 'ids',
    ]);

    return !empty($posts) ? (int) $posts[0] : null;
}


// handle existing ones, it could be a fresh mu-plugin to run once only, use ?update-posts to trigger it
add_action('admin_init', 'handle_update_posts_request');

function handle_update_posts_request() {
    if (!current_user_can('manage_options')) {
        return;
    }
    
    if (isset($_GET['update-posts'])) {
        // Show loading state and AJAX script
        add_action('admin_notices', function() {
            echo '<div id="pdf-update-notice" class="notice notice-info"><p><span class="spinner is-active" style="float: none; margin-right: 10px;"></span>Processing PDF URL updates... This may take a moment.</p></div>';
        });
        
        add_action('admin_footer', function() {
            ?>
            <script>
            jQuery(document).ready(function($) {
                // Start the update process
                $.ajax({
                    url: ajaxurl,
                    type: 'POST',
                    data: {
                        action: 'update_pdf_urls_ajax',
                        nonce: '<?php echo wp_create_nonce('update_pdf_urls_nonce'); ?>'
                    },
                    success: function(response) {
                        if (response.success) {
                            $('#pdf-update-notice').removeClass('notice-info').addClass('notice-success').html('<p><strong>Complete!</strong> Updated ' + response.data.total_updated + ' posts with PDF URL replacements.</p>');
                        } else {
                            $('#pdf-update-notice').removeClass('notice-info').addClass('notice-error').html('<p><strong>Error:</strong> ' + response.data + '</p>');
                        }
                    },
                    error: function() {
                        $('#pdf-update-notice').removeClass('notice-info').addClass('notice-error').html('<p><strong>Error:</strong> Failed to process PDF URL updates.</p>');
                    }
                });
            });
            </script>
            <?php
        });
    }
}

// AJAX handler for PDF URL updates
add_action('wp_ajax_update_pdf_urls_ajax', 'handle_pdf_urls_ajax');

function handle_pdf_urls_ajax() {
    // Verify nonce
    if (!wp_verify_nonce($_POST['nonce'], 'update_pdf_urls_nonce')) {
        wp_die('Security check failed');
    }
    
    if (!current_user_can('manage_options')) {
        wp_die('Insufficient permissions');
    }
    
    // Process updates for each post type
    $post_types = ['post', 'page', 'projects', 'research_theme'];
    $total_updated = 0;
    
    foreach ($post_types as $post_type) {
        $posts = get_posts([
            'post_type' => $post_type,
            'posts_per_page' => -1,
        ]);
        
        foreach ($posts as $post) {
            $pdfs = extract_pdf_urls($post->post_content);
            
            if (!empty($pdfs)) {
                $updated_content = $post->post_content;
                
                foreach ($pdfs as $pdf) {
                    $attachment_id = lookup_pdf_id_by_url($pdf);
                    
                    if ($attachment_id) {
                        $publication_id = lookup_publication_id_by_attachment_id($attachment_id);
                        
                        if ($publication_id) {
                            $new_url = "/download/publication/" . $publication_id;
                            $updated_content = str_replace($pdf, $new_url, $updated_content);
                        }
                    }
                }
                
                if ($updated_content !== $post->post_content) {
                    wp_update_post([
                        'ID' => $post->ID,
                        'post_content' => $updated_content,
                    ]);
                    $total_updated++;
                }
            }
        }
    }
    
    wp_send_json_success(['total_updated' => $total_updated]);
}