parsers
9 months ago
admin.php
7 months ago
core.php
6 months ago
engine.php
6 months ago
init.php
7 months ago
parsers.php
10 months ago
rest.php
6 months ago
support.php
6 months ago
ui.php
3 years ago
core.php
2473 lines
| 1 | <?php |
| 2 | |
| 3 | class Meow_WPMC_Core { |
| 4 | |
| 5 | |
| 6 | public $admin = null; |
| 7 | public $is_rest = false; |
| 8 | public $is_cli = false; |
| 9 | public $is_pro = false; |
| 10 | public $engine = null; |
| 11 | public $catch_timeout = true; // This will halt the plugin before reaching the PHP timeout. |
| 12 | public $types = "jpg|jpeg|jpe|gif|png|tiff|bmp|csv|svg|pdf|xls|xlsx|doc|docx|odt|wpd|rtf|tiff|mp3|mp4|mov|wav|lua|webp|avif|ico"; |
| 13 | public $current_method = 'media'; |
| 14 | public $servername = null; // meowapps.com (site URL without http/https) |
| 15 | public $site_url = null; // https://meowapps.com |
| 16 | public $upload_path = null; // /www/wp-content/uploads (path to uploads) |
| 17 | public $upload_url = null; // wp-content/uploads (uploads without domain) |
| 18 | private $option_name = 'wpmc_options'; |
| 19 | private $nonce = null; // Nonce for the REST API |
| 20 | |
| 21 | private $regex_file = '/[A-Za-z0-9-_,.\(\)\s]+[.]{1}(MIMETYPES)/'; |
| 22 | |
| 23 | private $refcache = array(); |
| 24 | private $use_cached_references = false; |
| 25 | private $progress_key = 'wpmc_progress'; |
| 26 | private $cached_ids_key = 'wpmc_cached_ids'; |
| 27 | private $cached_urls_key = 'wpmc_cached_urls'; |
| 28 | |
| 29 | private $cached_ids_cli = array(); |
| 30 | private $cached_urls_cli = array(); |
| 31 | |
| 32 | private $check_content = null; |
| 33 | private $debug_logs = null; |
| 34 | private $multilingual = false; |
| 35 | private $languages = array(); |
| 36 | private $shortcode_analysis = false; |
| 37 | |
| 38 | public function get_shortcode_analysis() { |
| 39 | return $this->shortcode_analysis; |
| 40 | } |
| 41 | |
| 42 | public function __construct() { |
| 43 | add_action( 'plugins_loaded', array( $this, 'plugins_loaded' ) ); |
| 44 | add_action( 'init', array( $this, 'init' ) ); |
| 45 | add_action( 'delete_attachment', array( $this, 'delete_attachment_related_data' ), 10, 1 ); |
| 46 | add_action( 'trashed_post', array( $this, 'delete_attachment_related_data' ), 10, 1 ); |
| 47 | } |
| 48 | |
| 49 | function plugins_loaded() { |
| 50 | |
| 51 | |
| 52 | if ( is_admin() ) { |
| 53 | new Meow_WPMC_UI( $this ); |
| 54 | } |
| 55 | |
| 56 | // Admin |
| 57 | $this->admin = new Meow_WPMC_Admin( $this ); |
| 58 | |
| 59 | // Advanced core |
| 60 | if ( class_exists( 'MeowPro_WPMC_Core' ) ) { |
| 61 | new MeowPro_WPMC_Core( $this ); |
| 62 | } |
| 63 | |
| 64 | // Only initialize variables if we are on a relevant screen |
| 65 | $pages = [ 'wpmc_dashboard', 'wpmc_settings' ]; |
| 66 | $page = isset( $_GET["page"] ) ? sanitize_text_field( $_GET["page"] ) : null; |
| 67 | $is_wpmc_screen = in_array( $page, $pages ); |
| 68 | |
| 69 | // Check if this is a REST request specifically for Media Cleaner |
| 70 | $is_wpmc_rest = false; |
| 71 | if ( MeowKit_WPMC_Helpers::is_rest() ) { |
| 72 | $request_uri = isset( $_SERVER['REQUEST_URI'] ) ? $_SERVER['REQUEST_URI'] : ''; |
| 73 | $is_wpmc_rest = strpos( $request_uri, '/media-cleaner/v1' ) !== false; |
| 74 | } |
| 75 | |
| 76 | $shouldLoad = ( defined( 'WP_CLI' ) && WP_CLI ) || $is_wpmc_screen || $is_wpmc_rest; |
| 77 | |
| 78 | if ( ! $shouldLoad ) { |
| 79 | return; |
| 80 | } |
| 81 | |
| 82 | |
| 83 | // Variables |
| 84 | $this->site_url = get_site_url(); |
| 85 | $this->multilingual = $this->is_multilingual(); |
| 86 | $this->languages = $this->get_languages(); |
| 87 | $this->current_method = $this->get_option( 'method' ); |
| 88 | $this->regex_file = str_replace( "MIMETYPES", $this->types, $this->regex_file ); |
| 89 | $this->servername = str_replace( 'http://', '', str_replace( 'https://', '', $this->site_url ) ); |
| 90 | $uploaddir = wp_upload_dir(); |
| 91 | $this->upload_path = $uploaddir['basedir']; |
| 92 | $this->upload_url = substr( $uploaddir['baseurl'], strlen( $this->site_url ) ); |
| 93 | $this->check_content = $this->get_option( 'content' ); |
| 94 | $this->debug_logs = $this->get_option( 'debuglogs' ); |
| 95 | $this->is_rest = MeowKit_WPMC_Helpers::is_rest(); |
| 96 | $this->is_cli = defined( 'WP_CLI' ) && WP_CLI; |
| 97 | $this->shortcode_analysis = !$this->get_option( 'shortcodes_disabled' ); |
| 98 | $this->use_cached_references = $this->get_option( 'use_cached_references' ); |
| 99 | |
| 100 | global $wpmc; |
| 101 | $wpmc = $this; |
| 102 | |
| 103 | // Language |
| 104 | load_plugin_textdomain( WPMC_DOMAIN, false, basename( WPMC_PATH ) . '/languages' ); |
| 105 | |
| 106 | // Install hooks and engine only if they might be used |
| 107 | if ( is_admin() || $this->is_rest || $this->is_cli ) { |
| 108 | add_action( 'wpmc_initialize_parsers', array( $this, 'initialize_parsers' ), 10, 0 ); |
| 109 | add_filter( 'wp_unique_filename', array( $this, 'wp_unique_filename' ), 10, 3 ); |
| 110 | $this->engine = new Meow_WPMC_Engine( $this, $this->admin ); |
| 111 | } |
| 112 | |
| 113 | // Only for REST |
| 114 | if ( $this->is_rest ) { |
| 115 | new Meow_WPMC_Rest( $this, $this->admin ); |
| 116 | } |
| 117 | |
| 118 | |
| 119 | } |
| 120 | |
| 121 | function init() { |
| 122 | remove_action( 'wp_scheduled_delete', 'wp_scheduled_delete' ); |
| 123 | } |
| 124 | |
| 125 | public function get_nonce( $force = false ) { |
| 126 | if ( !$force && !is_user_logged_in() ) { |
| 127 | return null; |
| 128 | } |
| 129 | if ( isset( $this->nonce ) ) { |
| 130 | return $this->nonce; |
| 131 | } |
| 132 | |
| 133 | $this->nonce = wp_create_nonce( 'wp_rest' ); |
| 134 | return $this->nonce; |
| 135 | } |
| 136 | |
| 137 | function initialize_parsers() { |
| 138 | include_once( 'parsers.php' ); |
| 139 | new Meow_WPMC_Parsers(); |
| 140 | } |
| 141 | |
| 142 | function deepsleep( $seconds ) { |
| 143 | $start_time = time(); |
| 144 | while( true ) { |
| 145 | if ( ( time() - $start_time ) > $seconds ) { |
| 146 | return false; |
| 147 | } |
| 148 | get_post( array( 'posts_per_page' => 50 ) ); |
| 149 | } |
| 150 | } |
| 151 | |
| 152 | private $start_time; |
| 153 | private $time_elapsed = 0; |
| 154 | private $time_remaining = 0; |
| 155 | private $item_scan_avg_time = 0; |
| 156 | private $wordpress_init_time = 0.5; |
| 157 | private $max_execution_time; |
| 158 | private $items_checked = 0; |
| 159 | private $items_count = 0; |
| 160 | |
| 161 | function get_max_execution_time() { |
| 162 | if ( isset( $this->max_execution_time ) ) |
| 163 | return $this->max_execution_time; |
| 164 | |
| 165 | $this->max_execution_time = ini_get( "max_execution_time" ); |
| 166 | if ( empty( $this->max_execution_time ) || $this->max_execution_time < 5 ) |
| 167 | $this->max_execution_time = 30; |
| 168 | |
| 169 | return $this->max_execution_time; |
| 170 | } |
| 171 | |
| 172 | function timeout_check_start( $count ) { |
| 173 | $this->start_time = time(); |
| 174 | $this->items_count = $count; |
| 175 | $this->get_max_execution_time(); |
| 176 | } |
| 177 | |
| 178 | function timeout_get_elapsed() { |
| 179 | return $this->time_elapsed . 'ms'; |
| 180 | } |
| 181 | |
| 182 | function timeout_check() { |
| 183 | $this->time_elapsed = time() - $this->start_time; |
| 184 | $this->time_remaining = $this->max_execution_time - $this->wordpress_init_time - $this->time_elapsed; |
| 185 | if ( $this->catch_timeout ) { |
| 186 | if ( $this->time_remaining - $this->item_scan_avg_time < 0 ) { |
| 187 | error_log("Media Cleaner Timeout! Check the Media Cleaner logs for more info."); |
| 188 | $this->log( "😵 Timeout! Some info for debug:" ); |
| 189 | $this->log( "🍀 Elapsed time: $this->time_elapsed" ); |
| 190 | $this->log( "🍀 WP init time: $this->wordpress_init_time" ); |
| 191 | $this->log( "🍀 Remaining time: $this->time_remaining" ); |
| 192 | $this->log( "🍀 Scan time per item: $this->item_scan_avg_time" ); |
| 193 | $this->log( "🍀 PHP max_execution_time: $this->max_execution_time" ); |
| 194 | header("HTTP/1.0 408 Request Timeout"); |
| 195 | exit; |
| 196 | } |
| 197 | } |
| 198 | } |
| 199 | |
| 200 | function delete_attachment_related_data( $post_id ) { |
| 201 | global $wpdb; |
| 202 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 203 | $wpdb->query( $wpdb->prepare( "DELETE FROM $table_name WHERE postId = %d", $post_id ) ); |
| 204 | } |
| 205 | |
| 206 | function timeout_check_additem() { |
| 207 | $this->items_checked++; |
| 208 | $this->time_elapsed = time() - $this->start_time; |
| 209 | $this->item_scan_avg_time = ceil( ( $this->time_elapsed / $this->items_checked ) * 10 ) / 10; |
| 210 | } |
| 211 | |
| 212 | // This checks if a new uploaded filename isn't the same one as a currently |
| 213 | // filename in the trash (that would cause issues) |
| 214 | function wp_unique_filename( $filename, $ext, $dir ) { |
| 215 | $fullpath = trailingslashit( $dir ) . $filename; |
| 216 | $relativepath = $this->clean_uploaded_filename( $fullpath ); |
| 217 | $trashfilepath = trailingslashit( $this->get_trashdir() ) . $relativepath; |
| 218 | if ( file_exists( $trashfilepath ) ) { |
| 219 | $path_parts = pathinfo( $fullpath ); |
| 220 | $filename_noext = $path_parts['filename']; |
| 221 | $new_filename = $filename_noext . '-' . date('Ymd-His', time()) . '.' . $path_parts['extension']; |
| 222 | //error_log( 'POTENTIALLY TRASH PATH: ' . $trashfilepath ); |
| 223 | //error_log( 'POTENTIALLY NEW FILE: ' . $new_filename ); |
| 224 | return $new_filename; |
| 225 | } |
| 226 | return $filename; |
| 227 | } |
| 228 | |
| 229 | function array_to_ids_or_urls( $meta, &$ids, &$urls, $recursive = false, $filters = array() ) { |
| 230 | foreach ( $meta as $k => $m ) { |
| 231 | |
| 232 | if ( is_numeric( $m ) ) { |
| 233 | |
| 234 | if ( !empty( $filters ) && is_array( $filters ) && !in_array( $k, $filters ) ) { |
| 235 | continue; |
| 236 | } |
| 237 | |
| 238 | // Probably a Media ID |
| 239 | if ( $m > 0 ) |
| 240 | { |
| 241 | array_push( $ids, $m ); |
| 242 | } |
| 243 | } |
| 244 | |
| 245 | else if ( is_array( $m ) ) { |
| 246 | |
| 247 | |
| 248 | if ( $recursive ) { |
| 249 | // If it's an array, we need to go deeper |
| 250 | $this->array_to_ids_or_urls( $m, $ids, $urls, true, $filters ); |
| 251 | } |
| 252 | |
| 253 | } |
| 254 | else if ( !empty( $m ) ) { |
| 255 | |
| 256 | if ( !empty( $filters ) && is_array( $filters ) && !in_array( $k, $filters ) ) { |
| 257 | continue; |
| 258 | } |
| 259 | |
| 260 | if ( is_string( $m ) && preg_match( '/^[\d\s,]+$/', $m ) && strpos( $m, ',' ) !== false ) { |
| 261 | // If this is a string that contains only digits, spaces, and commas, and contains at least one comma |
| 262 | // it is probably a list of IDs. So we should explode it to make an array |
| 263 | // Remove any spaces |
| 264 | |
| 265 | $m = str_replace( ' ', '', $m ); |
| 266 | $m = explode( ',', $m ); |
| 267 | |
| 268 | foreach ( $m as $mv ) { |
| 269 | if ( is_numeric( $mv ) && !in_array( (int)$mv, $ids ) ) { |
| 270 | array_push( $ids, (int)$mv ); |
| 271 | } |
| 272 | } |
| 273 | |
| 274 | continue; |
| 275 | } |
| 276 | |
| 277 | // If it's a string, maybe it's a file (with an extension) |
| 278 | if ( preg_match( $this->regex_file, $m ) ) |
| 279 | { |
| 280 | $clean_url = $this->clean_url( $m ); |
| 281 | array_push( $urls, $clean_url ); |
| 282 | } |
| 283 | } |
| 284 | } |
| 285 | } |
| 286 | |
| 287 | function get_favicon() { |
| 288 | // Yoast SEO plugin |
| 289 | $vals = get_option( 'wpseo_titles' ); |
| 290 | if ( !empty( $vals ) && isset( $vals['company_logo'] ) ) { |
| 291 | $url = $vals['company_logo']; |
| 292 | if ( $this->is_url( $url ) ) |
| 293 | return $this->clean_url( $url ); |
| 294 | } |
| 295 | } |
| 296 | |
| 297 | function get_all_shortcodes_attributes( $html, $ids_attr = array(), $urls_attr = array() ) { |
| 298 | // Get all the shortcodes from html, and check for each attributes of the shortcode if it is an ID or a URL and add the value in an array to return |
| 299 | $urls_values = array(); |
| 300 | $ids_values = array(); |
| 301 | |
| 302 | $pattern = get_shortcode_regex(); |
| 303 | if ( preg_match_all( '/'. $pattern .'/s', $html, $matches ) ) |
| 304 | { |
| 305 | foreach( $matches[0] as $key => $value) { |
| 306 | // $matches[3] return the shortcode attribute as string |
| 307 | // replace space with '&' for parse_str() function |
| 308 | $get = str_replace(" ", "&" , trim( $matches[3][$key] ) ); |
| 309 | $get = str_replace('"', '' , $get ); |
| 310 | parse_str( $get, $sub_output ); |
| 311 | |
| 312 | foreach ( $sub_output as $attr_key => $attr_value ) { |
| 313 | |
| 314 | if ( in_array( $attr_key, $ids_attr ) ) { |
| 315 | if ( is_numeric( $attr_value ) && !in_array( (int)$attr_value, $ids_values ) ) { |
| 316 | array_push( $ids_values, (int)$attr_value ); |
| 317 | } |
| 318 | |
| 319 | // In case of separated by commas |
| 320 | else if ( strpos( $attr_value, ',' ) !== false ) { |
| 321 | $attr_value = str_replace(' ', '', $attr_value ); |
| 322 | $pieces = explode( ',', $attr_value ); |
| 323 | foreach ( $pieces as $pval ) { |
| 324 | if ( is_numeric( $pval ) && !in_array( (int)$pval, $ids_values ) ) { |
| 325 | array_push( $ids_values, (int)$pval ); |
| 326 | } |
| 327 | } |
| 328 | } |
| 329 | } |
| 330 | |
| 331 | else if ( in_array( $attr_key, $urls_attr ) ) { |
| 332 | if ( !empty( trim( $attr_value ) ) && !in_array( trim( $attr_value ), $urls_values ) && !is_numeric( trim( $attr_value ) ) && strpos( trim( $attr_value ), 'http' ) !== false ) { |
| 333 | array_push( $urls_values, trim( $this->clean_url( $attr_value ) ) ); |
| 334 | } |
| 335 | } |
| 336 | } |
| 337 | } |
| 338 | } |
| 339 | |
| 340 | // Remove duplicates |
| 341 | $urls_values = array_unique( $urls_values ); |
| 342 | $ids_values = array_unique( $ids_values ); |
| 343 | |
| 344 | // Return the values |
| 345 | $values = array( |
| 346 | 'urls' => $urls_values, |
| 347 | 'ids' => $ids_values |
| 348 | ); |
| 349 | |
| 350 | return $values; |
| 351 | |
| 352 | } |
| 353 | |
| 354 | |
| 355 | |
| 356 | /** |
| 357 | * Recursively transforms a string with WordPress shortcodes into a |
| 358 | * hierarchical tree structure (an Abstract Syntax Tree). |
| 359 | * |
| 360 | * @param string $content The string containing the shortcodes. |
| 361 | * @return array An array of nodes, where each node can be a shortcode with its |
| 362 | * own 'children' array, or a simple text node. |
| 363 | */ |
| 364 | function nested_shortcodes_to_array(string $content): array |
| 365 | { |
| 366 | $nodes = []; |
| 367 | $last_pos = 0; |
| 368 | |
| 369 | $pattern = '/\\[' . '(\\[?)' . '([\w-]+)' . '(?![\\w-])' . '(' . '[^\\]\\/]*' . '(?:' . '\\/(?!\\])' . '[^\\]\\/]*' . ')*?' . ')' . '(?:' . '(\\/)' . '\\]' . '|' . '\\]' . '(?:' . '(' . '[^\\[]*+' . '(?:' . '\\[(?!\\/\\2\\])' . '[^\\[]*+' . ')*+' . ')' . '\\[\\/\\2\\]' . ')?' . ')' . '(\\]?)/s'; |
| 370 | |
| 371 | // preg_match_all with PREG_OFFSET_CAPTURE is key to tracking positions. |
| 372 | if (preg_match_all($pattern, $content, $matches, PREG_SET_ORDER | PREG_OFFSET_CAPTURE)) { |
| 373 | foreach ($matches as $match) { |
| 374 | // Get the position and content of the full shortcode match |
| 375 | $match_start_pos = $match[0][1]; |
| 376 | $match_full_string = $match[0][0]; |
| 377 | $match_end_pos = $match_start_pos + strlen($match_full_string); |
| 378 | |
| 379 | // 1. Capture any text that appeared *before* this shortcode |
| 380 | if ($match_start_pos > $last_pos) { |
| 381 | $text_content = substr($content, $last_pos, $match_start_pos - $last_pos); |
| 382 | if (trim($text_content) !== '') { |
| 383 | $nodes[] = [ |
| 384 | 'type' => 'text', |
| 385 | 'content' => $text_content |
| 386 | ]; |
| 387 | } |
| 388 | } |
| 389 | |
| 390 | // 2. Process the shortcode match itself |
| 391 | $tag = $match[2][0]; |
| 392 | $attributes_string = $match[3][0]; |
| 393 | // Use isset since self-closing tags won't have inner content (group 5) |
| 394 | $inner_content = isset($match[5]) ? $match[5][0] : null; |
| 395 | |
| 396 | // Parse attributes from the attribute string |
| 397 | $parsed_attributes = []; |
| 398 | if (preg_match_all('/([\w-]+)\s*=\s*(["\'])([^"\']*?)\2/', $attributes_string, $attr_matches)) { |
| 399 | foreach ($attr_matches[1] as $attr_index => $key) { |
| 400 | $parsed_attributes[$key] = $attr_matches[3][$attr_index]; |
| 401 | } |
| 402 | } |
| 403 | |
| 404 | $shortcode_node = [ |
| 405 | 'type' => 'shortcode', |
| 406 | 'tag' => $tag, |
| 407 | 'attributes' => $parsed_attributes, |
| 408 | ]; |
| 409 | |
| 410 | // 3. This is the recursion! |
| 411 | // If there is inner content, parse it with the same function. |
| 412 | if ($inner_content !== null) { |
| 413 | $children = $this->nested_shortcodes_to_array($inner_content); |
| 414 | if (!empty($children)) { |
| 415 | $shortcode_node['children'] = $children; |
| 416 | } |
| 417 | } |
| 418 | |
| 419 | $nodes[] = $shortcode_node; |
| 420 | |
| 421 | // Update the last position to the end of the current match |
| 422 | $last_pos = $match_end_pos; |
| 423 | } |
| 424 | } |
| 425 | |
| 426 | // 4. Capture any remaining text after the very last shortcode |
| 427 | if ($last_pos < strlen($content)) { |
| 428 | $text_content = substr($content, $last_pos); |
| 429 | if (trim($text_content) !== '') { |
| 430 | $nodes[] = [ |
| 431 | 'type' => 'text', |
| 432 | 'content' => $text_content |
| 433 | ]; |
| 434 | } |
| 435 | } |
| 436 | |
| 437 | return $nodes; |
| 438 | } |
| 439 | |
| 440 | |
| 441 | |
| 442 | |
| 443 | function get_shortcode_attributes( $shortcode_tag, $post ) { |
| 444 | if ( has_shortcode( $post->post_content, $shortcode_tag ) ) { |
| 445 | $output = array(); |
| 446 | //get shortcode regex pattern wordpress function |
| 447 | $pattern = get_shortcode_regex( [ $shortcode_tag ] ); |
| 448 | if ( preg_match_all( '/'. $pattern .'/s', $post->post_content, $matches ) ) |
| 449 | { |
| 450 | $keys = array(); |
| 451 | $output = array(); |
| 452 | foreach( $matches[0] as $key => $value) { |
| 453 | // $matches[3] return the shortcode attribute as string |
| 454 | // replace space with '&' for parse_str() function |
| 455 | $get = str_replace(" ", "&" , trim( $matches[3][$key] ) ); |
| 456 | $get = str_replace('"', '' , $get ); |
| 457 | parse_str( $get, $sub_output ); |
| 458 | |
| 459 | //get all shortcode attribute keys |
| 460 | $keys = array_unique( array_merge( $keys, array_keys( $sub_output )) ); |
| 461 | $output[] = $sub_output; |
| 462 | } |
| 463 | if ( $keys && $output ) { |
| 464 | // Loop the output array and add the missing shortcode attribute key |
| 465 | foreach ($output as $key => $value) { |
| 466 | // Loop the shortcode attribute key |
| 467 | foreach ($keys as $attr_key) { |
| 468 | $output[$key][$attr_key] = isset( $output[$key] ) && isset( $output[$key] ) ? $output[$key][$attr_key] : NULL; |
| 469 | } |
| 470 | //sort the array key |
| 471 | ksort( $output[$key]); |
| 472 | } |
| 473 | } |
| 474 | } |
| 475 | return $output; |
| 476 | } |
| 477 | else { |
| 478 | return false; |
| 479 | } |
| 480 | } |
| 481 | |
| 482 | // SImply use regex to get URLs from a string return an array of URLs |
| 483 | function get_urls_from_string( $string ) { |
| 484 | $urls = array(); |
| 485 | $pattern = '/https?:\/\/[^\s"\']+/i'; |
| 486 | if ( preg_match_all( $pattern, $string, $matches ) ) { |
| 487 | foreach ( $matches[0] as $match ) { |
| 488 | $clean_url = $this->clean_url( $match ); |
| 489 | array_push( $urls, $clean_url ); |
| 490 | } |
| 491 | } |
| 492 | return $urls; |
| 493 | } |
| 494 | |
| 495 | function get_urls_from_html( $html ) { |
| 496 | if ( empty( $html ) ) { |
| 497 | return array(); |
| 498 | } |
| 499 | |
| 500 | |
| 501 | // Proposal/fix by @copytrans |
| 502 | // Discussion: https://wordpress.org/support/topic/bug-in-core-php/#post-11647775 |
| 503 | // Modified by Jordy again in 2021 for those who don't have MB enabled |
| 504 | if ( function_exists( 'mb_encode_numericentity' ) ) { |
| 505 | $convmap = [0x80, 0xffff, 0, 0xffff]; |
| 506 | $html = mb_encode_numericentity( $html, $convmap, 'UTF-8' ); |
| 507 | } else { |
| 508 | $html = preg_replace_callback( |
| 509 | '/[\x80-\xFF]/', |
| 510 | function( $match ) { |
| 511 | return '&#' . ord( $match[0] ) . ';'; |
| 512 | }, |
| 513 | $html |
| 514 | ); |
| 515 | } |
| 516 | |
| 517 | // Resolve src-set and shortcodes |
| 518 | if ( $this->get_shortcode_analysis() ) { |
| 519 | $html = do_shortcode( $html ); |
| 520 | } |
| 521 | |
| 522 | // TODO: Since WP 5.5, wp_filter_content_tags should be used instead of wp_make_content_images_responsive. |
| 523 | $html = function_exists( 'wp_filter_content_tags' ) ? wp_filter_content_tags( $html ) : |
| 524 | wp_make_content_images_responsive( $html ); |
| 525 | |
| 526 | // Create the DOM Document |
| 527 | if ( !class_exists("DOMDocument") ) { |
| 528 | error_log( 'Media Cleaner: The DOM extension for PHP is not installed.' ); |
| 529 | throw new Error( 'The DOM extension for PHP is not installed.' ); |
| 530 | } |
| 531 | |
| 532 | |
| 533 | if ( empty( $html ) ) { |
| 534 | return array(); |
| 535 | } |
| 536 | |
| 537 | libxml_use_internal_errors(true); |
| 538 | $dom = new DOMDocument(); |
| 539 | @$dom->loadHTML( $html ); |
| 540 | libxml_clear_errors(); |
| 541 | $results = array(); |
| 542 | |
| 543 | // <meta> tags in <head> area |
| 544 | $metas = $dom->getElementsByTagName( 'meta' ); |
| 545 | foreach ( $metas as $meta ) { |
| 546 | $property = $meta->getAttribute( 'property' ); |
| 547 | if ( $property == 'og:image' || $property == 'og:image:secure_url' || $property == 'twitter:image' ) { |
| 548 | $url = $meta->getAttribute( 'content' ); |
| 549 | if ( $this->is_url( $url ) ) { |
| 550 | $src = $this->clean_url( $url ); |
| 551 | if ( !empty( $src ) ) { |
| 552 | array_push( $results, $src ); |
| 553 | } |
| 554 | } |
| 555 | } |
| 556 | } |
| 557 | |
| 558 | |
| 559 | |
| 560 | // IFrames (by Mike Meinz) |
| 561 | $iframes = $dom->getElementsByTagName( 'iframe' ); |
| 562 | foreach( $iframes as $iframe ) { |
| 563 | $iframe_src = $iframe->getAttribute( 'src' ); |
| 564 | // Ignore if the iframe src is not on this server |
| 565 | if ( ( strpos( $iframe_src, $this->servername ) !== false) || ( substr( $iframe_src, 0, 1 ) == "/" ) ) { |
| 566 | // Create a new DOM Document to hold iframe |
| 567 | $iframe_doc = new DOMDocument(); |
| 568 | // Load the url's contents into the DOM |
| 569 | libxml_use_internal_errors( true ); // ignore html formatting problems |
| 570 | $rslt = @$iframe_doc->loadHTMLFile( $iframe_src ); |
| 571 | libxml_clear_errors(); |
| 572 | libxml_use_internal_errors( false ); |
| 573 | if ( $rslt ) { |
| 574 | // Get the resulting html |
| 575 | $iframe_html = $iframe_doc->saveHTML(); |
| 576 | if ( $iframe_html !== false ) { |
| 577 | // Scan for links in the iframe |
| 578 | $iframe_urls = $this->get_urls_from_html( $iframe_html ); // Recursion |
| 579 | if ( !empty( $iframe_urls ) ) { |
| 580 | $results = array_merge( $results, $iframe_urls ); |
| 581 | } |
| 582 | } |
| 583 | } |
| 584 | else { |
| 585 | $this->log( '🚫 Failed to load iframe: ' . $iframe_src ); |
| 586 | } |
| 587 | } |
| 588 | } |
| 589 | |
| 590 | |
| 591 | // Images: src, srcset |
| 592 | $imgs = $dom->getElementsByTagName( 'img' ); |
| 593 | foreach ( $imgs as $img ) { |
| 594 | //error_log($img->getAttribute('src')); |
| 595 | $src = $this->clean_url( $img->getAttribute('src') ); |
| 596 | array_push( $results, $src ); |
| 597 | $srcset = $img->getAttribute('srcset'); |
| 598 | if ( !empty( $srcset ) ) { |
| 599 | $setImgs = explode( ',', trim( $srcset ) ); |
| 600 | foreach ( $setImgs as $setImg ) { |
| 601 | $finalSetImg = explode( ' ', trim( $setImg ) ); |
| 602 | if ( is_array( $finalSetImg ) ) { |
| 603 | array_push( $results, $this->clean_url( $finalSetImg[0] ) ); |
| 604 | } |
| 605 | } |
| 606 | } |
| 607 | } |
| 608 | |
| 609 | // Videos: src, poster, and attached file |
| 610 | $videos = $dom->getElementsByTagName( 'video' ); |
| 611 | foreach ($videos as $video) { |
| 612 | // Get src attribute |
| 613 | $raw_video_src = $video->getAttribute( 'src' ); |
| 614 | $src = $this->clean_url( $raw_video_src ); |
| 615 | if ( !empty( $src ) ) { |
| 616 | $video_id = $this->custom_attachment_url_to_postid( $raw_video_src ); |
| 617 | |
| 618 | $attached_file = get_post_meta( $video_id, '_wp_attached_file', true ); |
| 619 | if ( !empty( $attached_file ) ) { |
| 620 | array_push( $results, $attached_file ); |
| 621 | } |
| 622 | } |
| 623 | |
| 624 | // Get poster attribute |
| 625 | $raw_poster_src = $video->getAttribute( 'poster' ); |
| 626 | $poster = $this->clean_url( $raw_poster_src ); |
| 627 | if ( !empty( $poster ) ) { |
| 628 | $poster_id = $this->custom_attachment_url_to_postid( $raw_poster_src ); |
| 629 | |
| 630 | $attached_file = get_post_meta( $poster_id, '_wp_attached_file', true ); |
| 631 | if ( !empty( $attached_file ) ) { |
| 632 | array_push( $results, $attached_file ); |
| 633 | } |
| 634 | } |
| 635 | |
| 636 | } |
| 637 | |
| 638 | // Audios: src |
| 639 | $audios = $dom->getElementsByTagName( 'audio' ); |
| 640 | foreach ( $audios as $audio ) { |
| 641 | //error_log($audio->getAttribute('src')); |
| 642 | $src = $this->clean_url( $audio->getAttribute('src') ); |
| 643 | array_push( $results, $src ); |
| 644 | } |
| 645 | |
| 646 | // Sources: src |
| 647 | $audios = $dom->getElementsByTagName( 'source' ); |
| 648 | foreach ( $audios as $audio ) { |
| 649 | //error_log($audio->getAttribute('src')); |
| 650 | $src = $this->clean_url( $audio->getAttribute('src') ); |
| 651 | array_push( $results, $src ); |
| 652 | } |
| 653 | |
| 654 | // Links, href |
| 655 | $urls = $dom->getElementsByTagName( 'a' ); |
| 656 | foreach ( $urls as $url ) { |
| 657 | $url_href = $url->getAttribute('href'); // mm change |
| 658 | if ( $this->is_url( $url_href ) ) { // mm change |
| 659 | $src = $this->clean_url( $url_href ); // mm change |
| 660 | if ( !empty( $src ) ) |
| 661 | array_push( $results, $src ); |
| 662 | } |
| 663 | } |
| 664 | |
| 665 | // <link> tags in <head> area |
| 666 | $urls = $dom->getElementsByTagName( 'link' ); |
| 667 | foreach ( $urls as $url ) { |
| 668 | $url_href = $url->getAttribute( 'href' ); |
| 669 | if ( $this->is_url( $url_href ) ) { |
| 670 | $src = $this->clean_url( $url_href ); |
| 671 | if ( !empty( $src ) ) { |
| 672 | array_push( $results, $src ); |
| 673 | } |
| 674 | } |
| 675 | } |
| 676 | |
| 677 | |
| 678 | preg_match_all( "/((https?:\/\/)?[^\\&\#\[\] \"\?]+\.pdf)/", $html, $res ); |
| 679 | if ( !empty( $res ) && isset( $res[1] ) && count( $res[1] ) > 0 ) { |
| 680 | foreach ( $res[1] as $url ) { |
| 681 | if ( $this->is_url( $url ) ) |
| 682 | array_push( $results, $this->clean_url( $url ) ); |
| 683 | } |
| 684 | } |
| 685 | |
| 686 | // Background images |
| 687 | preg_match_all( "/url\(\'?\"?((https?:\/\/)?[^\\&\#\[\] \"\?]+\.(jpe?g|gif|png))\'?\"?/", $html, $res ); |
| 688 | if ( !empty( $res ) && isset( $res[1] ) && count( $res[1] ) > 0 ) { |
| 689 | foreach ( $res[1] as $url ) { |
| 690 | if ( $this->is_url( $url ) ) |
| 691 | array_push( $results, $this->clean_url( $url ) ); |
| 692 | } |
| 693 | } |
| 694 | |
| 695 | return $results; |
| 696 | } |
| 697 | |
| 698 | /** |
| 699 | * |
| 700 | * Get the IDs and URLs from the blocks of a post. |
| 701 | * |
| 702 | * @param string $html The HTML content of the post. |
| 703 | * @param string $prefix The prefix of the blocks to look for. |
| 704 | * @param array $keys The keys to look for in the blocks. |
| 705 | * @param array $urls The array to fill with the URLs. |
| 706 | * @param array $ids The array to fill with the IDs. |
| 707 | * |
| 708 | */ |
| 709 | function get_from_blocks( $html, $prefix, $keys, &$urls, &$ids ) { |
| 710 | |
| 711 | $blocks = parse_blocks( $html ); |
| 712 | |
| 713 | if ( ! is_array( $blocks ) || ! isset( $blocks[0] ) ) { |
| 714 | return; |
| 715 | } |
| 716 | |
| 717 | |
| 718 | foreach ( $blocks as $block ) { |
| 719 | |
| 720 | if ( strpos( $block['blockName'], $prefix ) === false ) { |
| 721 | continue; |
| 722 | } |
| 723 | |
| 724 | $this->array_to_ids_or_urls( $block, $ids, $urls, true, $keys ); |
| 725 | |
| 726 | } |
| 727 | |
| 728 | |
| 729 | } |
| 730 | // Parse a meta, visit all the arrays, look for the attributes, fill $ids and $urls arrays |
| 731 | // If rawMode is enabled, it will not check if the value is an ID or an URL, it will just returns it in URLs |
| 732 | function get_from_meta( $meta, $lookFor, &$ids, &$urls, $rawMode = false ) { |
| 733 | if ( !is_array( $meta ) && !is_object( $meta) ) { |
| 734 | return; |
| 735 | } |
| 736 | foreach ( $meta as $key => $value ) { |
| 737 | if ( is_object( $value ) || is_array( $value ) ) |
| 738 | $this->get_from_meta( $value, $lookFor, $ids, $urls, $rawMode ); |
| 739 | else if ( in_array( $key, $lookFor ) ) { |
| 740 | if ( empty( $value ) ) { |
| 741 | continue; |
| 742 | } |
| 743 | else if ( $rawMode ) { |
| 744 | array_push( $urls, $value ); |
| 745 | } |
| 746 | else if ( is_numeric( $value ) ) { |
| 747 | // It this an ID? |
| 748 | array_push( $ids, $value ); |
| 749 | } |
| 750 | else { |
| 751 | if ( $this->is_url( $value ) ) { |
| 752 | // Is this an URL? |
| 753 | array_push( $urls, $this->clean_url( $value ) ); |
| 754 | } |
| 755 | else { |
| 756 | // Is this an array of IDs, encoded as a string? (like "20,13") |
| 757 | $pieces = explode( ',', $value ); |
| 758 | foreach ( $pieces as $pval ) { |
| 759 | if ( is_numeric( $pval ) ) { |
| 760 | array_push( $ids, $pval ); |
| 761 | } |
| 762 | } |
| 763 | } |
| 764 | } |
| 765 | } |
| 766 | } |
| 767 | } |
| 768 | |
| 769 | function get_images_from_themes( &$ids, &$urls ) { |
| 770 | // USE CURRENT THEME AND WP API |
| 771 | $ch = get_custom_header(); |
| 772 | if ( !empty( $ch ) && !empty( $ch->url ) ) { |
| 773 | array_push( $urls, $this->clean_url( $ch->url ) ); |
| 774 | } |
| 775 | if ( $this->is_url( $ch->thumbnail_url ) ) { |
| 776 | array_push( $urls, $this->clean_url( $ch->thumbnail_url ) ); |
| 777 | } |
| 778 | if ( !empty( $ch ) && !empty( $ch->attachment_id ) ) { |
| 779 | array_push( $ids, $ch->attachment_id ); |
| 780 | } |
| 781 | $cl = get_custom_logo(); |
| 782 | if ( $this->is_url( $cl ) ) { |
| 783 | $urls = array_merge( $this->get_urls_from_html( $cl ), $urls ); |
| 784 | } |
| 785 | $custom_logo = get_theme_mod( 'custom_logo' ); |
| 786 | if ( !empty( $custom_logo ) && is_numeric( $custom_logo ) ) { |
| 787 | array_push( $ids, (int)$custom_logo ); |
| 788 | } |
| 789 | $si = get_site_icon_url(); |
| 790 | if ( $this->is_url( $si ) ) { |
| 791 | array_push( $urls, $this->clean_url( $si ) ); |
| 792 | } |
| 793 | $si_id = get_option( 'site_icon' ); |
| 794 | if ( !empty( $si_id ) && is_numeric( $si_id ) ) { |
| 795 | array_push( $ids, (int)$si_id ); |
| 796 | } |
| 797 | $cd = get_background_image(); |
| 798 | if ( $this->is_url( $cd ) ) { |
| 799 | array_push( $urls, $this->clean_url( $cd ) ); |
| 800 | } |
| 801 | $photography_hero_image = get_theme_mod( 'photography_hero_image' ); |
| 802 | if ( !empty( $photography_hero_image ) ) { |
| 803 | array_push( $ids, $photography_hero_image ); |
| 804 | } |
| 805 | $author_profile_picture = get_theme_mod( 'author_profile_picture' ); |
| 806 | if ( !empty( $author_profile_picture ) ) { |
| 807 | array_push( $ids, $author_profile_picture ); |
| 808 | } |
| 809 | if ( function_exists ( 'get_uploaded_header_images' ) ) { |
| 810 | $header_images = get_uploaded_header_images(); |
| 811 | if ( !empty( $header_images ) ) { |
| 812 | foreach ( $header_images as $hi ) { |
| 813 | if ( !empty ( $hi['attachment_id'] ) ) { |
| 814 | array_push( $ids, $hi['attachment_id'] ); |
| 815 | } |
| 816 | } |
| 817 | } |
| 818 | } |
| 819 | } |
| 820 | |
| 821 | #region LOGS |
| 822 | |
| 823 | function log( $data = null, $force = false ) { |
| 824 | if ( !$this->debug_logs && !$force ) |
| 825 | return; |
| 826 | |
| 827 | $php_logs = $this->get_option( 'php_error_logs' ); |
| 828 | $log_file_path = $this->get_logs_path(); |
| 829 | |
| 830 | $fh = @fopen( $log_file_path, 'a' ); |
| 831 | if ( !$fh ) { return false; } |
| 832 | $date = date( "Y-m-d H:i:s" ); |
| 833 | if ( is_null( $data ) ) { |
| 834 | fwrite( $fh, "\n" ); |
| 835 | } |
| 836 | else { |
| 837 | fwrite( $fh, "$date: {$data}\n" ); |
| 838 | if ( $php_logs ) { |
| 839 | error_log( "[MEDIA CLEANER] " . $data ); |
| 840 | } |
| 841 | } |
| 842 | fclose( $fh ); |
| 843 | return true; |
| 844 | } |
| 845 | |
| 846 | //WPMC_PREFIX |
| 847 | |
| 848 | function get_logs_path() { |
| 849 | $uploads_dir = wp_upload_dir(); |
| 850 | $uploads_dir_path = trailingslashit( $uploads_dir['basedir'] ); |
| 851 | |
| 852 | $path = $this->get_option( 'logs_path' ); |
| 853 | |
| 854 | if ( $path && file_exists( $path ) ) { |
| 855 | // make sure the path is legal (within the uploads directory with the WPMC_PREFIX prefix and log extension) |
| 856 | if ( strpos( $path, $uploads_dir_path ) !== 0 || strpos( $path, WPMC_PREFIX ) === false || substr( $path, -4 ) !== '.log' ) { |
| 857 | $path = null; |
| 858 | } else { |
| 859 | return $path; |
| 860 | } |
| 861 | } |
| 862 | |
| 863 | if ( !$path ) { |
| 864 | $path = $uploads_dir_path . WPMC_PREFIX . "_" . $this->random_ascii_chars() . ".log"; |
| 865 | if ( !file_exists( $path ) ) { |
| 866 | touch( $path ); |
| 867 | } |
| 868 | |
| 869 | $options = $this->get_all_options(); |
| 870 | $options['logs_path'] = $path; |
| 871 | $this->update_options( $options ); |
| 872 | } |
| 873 | |
| 874 | return $path; |
| 875 | } |
| 876 | |
| 877 | |
| 878 | function get_logs() { |
| 879 | $log_file_path = $this->get_logs_path(); |
| 880 | |
| 881 | if ( !file_exists( $log_file_path ) ) { |
| 882 | return "No logs found."; |
| 883 | } |
| 884 | |
| 885 | $content = file_get_contents( $log_file_path ); |
| 886 | $lines = explode( "\n", $content ); |
| 887 | $lines = array_filter( $lines ); |
| 888 | $lines = array_reverse( $lines ); |
| 889 | $content = implode( "\n", $lines ); |
| 890 | return $content; |
| 891 | } |
| 892 | |
| 893 | function clear_logs() { |
| 894 | $logPath = $this->get_logs_path(); |
| 895 | if ( file_exists( $logPath ) ) { |
| 896 | unlink( $logPath ); |
| 897 | } |
| 898 | |
| 899 | $options = $this->get_all_options(); |
| 900 | $options['logs_path'] = null; |
| 901 | $this->update_options( $options ); |
| 902 | } |
| 903 | |
| 904 | #endregion |
| 905 | |
| 906 | /** |
| 907 | * |
| 908 | * HELPERS |
| 909 | * |
| 910 | */ |
| 911 | |
| 912 | private function random_ascii_chars($length = 8) |
| 913 | { |
| 914 | $characters = array_merge(range('A', 'Z'), range('a', 'z'), range('0', '9')); |
| 915 | $characters_length = count($characters); |
| 916 | $random_string = ''; |
| 917 | |
| 918 | for ($i = 0; $i < $length; $i++) { |
| 919 | $random_string .= $characters[rand(0, $characters_length - 1)]; |
| 920 | } |
| 921 | |
| 922 | return $random_string; |
| 923 | } |
| 924 | |
| 925 | function get_trashdir() { |
| 926 | return trailingslashit( $this->upload_path ) . 'wpmc-trash'; |
| 927 | } |
| 928 | |
| 929 | function get_trashurl() { |
| 930 | return trailingslashit( $this->upload_url ) . 'wpmc-trash'; |
| 931 | } |
| 932 | |
| 933 | function clean_ob(){ |
| 934 | $disabled = $this->get_option( 'output_buffer_cleaning_disabled' ); |
| 935 | $ob_content = ob_get_contents(); |
| 936 | if ( !empty( trim( $ob_content ) ) ) { |
| 937 | |
| 938 | if ( $disabled ) { |
| 939 | $this->log( "🚨 If the server's response was broken, try to let Output Buffer Cleaning enabled." ); |
| 940 | return; |
| 941 | } |
| 942 | |
| 943 | $this->log( "🧹 The response is broken due to output buffering, it will be cleaned." ); |
| 944 | $this->log( "📄 Output buffer content: " . $ob_content ); |
| 945 | |
| 946 | ob_end_clean(); |
| 947 | } |
| 948 | } |
| 949 | |
| 950 | /** |
| 951 | * |
| 952 | * I18N RELATED HELPERS |
| 953 | * |
| 954 | */ |
| 955 | |
| 956 | function is_multilingual() { |
| 957 | return function_exists( 'icl_get_languages' ); |
| 958 | } |
| 959 | |
| 960 | function get_languages() { |
| 961 | $results = array(); |
| 962 | if ( $this->is_multilingual() ) { |
| 963 | $languages = icl_get_languages(); |
| 964 | foreach ( $languages as $language ) { |
| 965 | if ( isset( $language['code'] ) ) { |
| 966 | array_push( $results, $language['code'] ); |
| 967 | } |
| 968 | else if ( isset( $language['language_code'] ) ) { |
| 969 | array_push( $results, $language['language_code'] ); |
| 970 | } |
| 971 | } |
| 972 | } |
| 973 | return $results; |
| 974 | } |
| 975 | |
| 976 | function get_translated_media_ids( $mediaId ) { |
| 977 | $translated_ids = array(); |
| 978 | foreach ( $this->languages as $language ) { |
| 979 | $id = apply_filters( 'wpml_object_id', $mediaId, 'attachment', false, $language ); |
| 980 | if ( !empty( $id ) ) { |
| 981 | array_push( $translated_ids, $id ); |
| 982 | } |
| 983 | } |
| 984 | return $translated_ids; |
| 985 | } |
| 986 | |
| 987 | /** |
| 988 | * |
| 989 | * DELETE / SCANNING / RESET |
| 990 | * |
| 991 | */ |
| 992 | |
| 993 | function recover_file( $path ) { |
| 994 | $originalPath = trailingslashit( $this->upload_path ) . $path; |
| 995 | $trashPath = trailingslashit( $this->get_trashdir() ) . $path; |
| 996 | if ( !file_exists( $trashPath ) ) { |
| 997 | $this->log( "🚫 The file $originalPath actually does not exist in the trash." ); |
| 998 | return true; |
| 999 | } |
| 1000 | $path_parts = pathinfo( $originalPath ); |
| 1001 | if ( !file_exists( $path_parts['dirname'] ) && !wp_mkdir_p( $path_parts['dirname'] ) ) { |
| 1002 | die( 'Failed to create folder.' ); |
| 1003 | } |
| 1004 | if ( !rename( $trashPath, $originalPath ) ) { |
| 1005 | die( 'Failed to move the file.' ); |
| 1006 | } |
| 1007 | return true; |
| 1008 | } |
| 1009 | |
| 1010 | function recover( $id ) { |
| 1011 | global $wpdb; |
| 1012 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 1013 | $issue = $this->get_issue( $id ); |
| 1014 | |
| 1015 | if ( empty( $issue ) ) { |
| 1016 | $this->log( "🚫 Issue #{$id} does not exist. Cannot recover this." ); |
| 1017 | return false; |
| 1018 | } |
| 1019 | |
| 1020 | // Files |
| 1021 | if ( $issue->type === 0 ) { |
| 1022 | $this->recover_file( $issue->path ); |
| 1023 | $wpdb->query( $wpdb->prepare( "UPDATE $table_name SET deleted = 0 WHERE id = %d", $id ) ); |
| 1024 | $this->log( "� |
| 1025 | Recovered {$issue->path}." ); |
| 1026 | return true; |
| 1027 | } |
| 1028 | // Media |
| 1029 | else if ( $issue->type === 1 ) { |
| 1030 | |
| 1031 | // If there is no file attached, doesn't handle the files |
| 1032 | $fullpath = get_attached_file( $issue->postId ); |
| 1033 | if ( empty( $fullpath ) ) { |
| 1034 | $this->log( "🚫 Media #{$issue->postId} does not have attached file anymore." ); |
| 1035 | error_log( "Media #{$issue->postId} does not have attached file anymore." ); |
| 1036 | return false; |
| 1037 | } |
| 1038 | |
| 1039 | $paths = $this->get_paths_from_attachment( $issue->postId ); |
| 1040 | foreach ( $paths as $path ) { |
| 1041 | if ( !$this->recover_file( $path ) ) { |
| 1042 | $this->log( "🚫 Could not recover $path." ); |
| 1043 | error_log( "Media Cleaner: Could not recover $path." ); |
| 1044 | } |
| 1045 | } |
| 1046 | if ( !wp_update_post( array( 'ID' => $issue->postId, 'post_type' => 'attachment' ) ) ) { |
| 1047 | $this->log( "🚫 Failed to Untrash Post {$issue->postId} (but deleted it from Cleaner DB)." ); |
| 1048 | error_log( "Media Cleaner: Failed to Untrash Post {$issue->postId} (but deleted it from Cleaner DB)." ); |
| 1049 | return false; |
| 1050 | } |
| 1051 | $wpdb->query( $wpdb->prepare( "UPDATE $table_name SET deleted = 0 WHERE id = %d", $id ) ); |
| 1052 | $this->log( "� |
| 1053 | Recovered Media #{$issue->postId}." ); |
| 1054 | return true; |
| 1055 | } |
| 1056 | } |
| 1057 | |
| 1058 | function trash_file( $fileIssuePath ) { |
| 1059 | $originalPath = trailingslashit( $this->upload_path ) . $fileIssuePath; |
| 1060 | $trashPath = trailingslashit( $this->get_trashdir() ) . $fileIssuePath; |
| 1061 | $path_parts = pathinfo( $trashPath ); |
| 1062 | |
| 1063 | try { |
| 1064 | if ( !file_exists( $path_parts['dirname'] ) && !wp_mkdir_p( $path_parts['dirname'] ) ) { |
| 1065 | $this->log( "🚫 Could not create the trash directory for Media Cleaner." ); |
| 1066 | error_log( "Media Cleaner: Could not create the trash directory." ); |
| 1067 | return false; |
| 1068 | } |
| 1069 | // Rename the file (move). 'is_dir' is just there for security (no way we should move a whole directory) |
| 1070 | if ( is_dir( $originalPath ) ) { |
| 1071 | $this->log( "🚫 Attempted to delete a directory instead of a file ($originalPath). Can't do that." ); |
| 1072 | error_log( "Media Cleaner: Attempted to delete a directory instead of a file ($originalPath). Can't do that." ); |
| 1073 | return false; |
| 1074 | } |
| 1075 | if ( !file_exists( $originalPath ) ) { |
| 1076 | $this->log( "🚫 The file $originalPath actually does not exist." ); |
| 1077 | error_log( "Media Cleaner: The file $originalPath actually does not exist." ); |
| 1078 | return true; |
| 1079 | } |
| 1080 | if ( !@rename( $originalPath, $trashPath ) ) { |
| 1081 | error_log( "Media Cleaner: Unknown error occured while trying to delete a file ($originalPath)." ); |
| 1082 | return false; |
| 1083 | } |
| 1084 | } |
| 1085 | catch ( Exception $e ) { |
| 1086 | return false; |
| 1087 | } |
| 1088 | $this->clean_dir( dirname( $originalPath ) ); |
| 1089 | return true; |
| 1090 | } |
| 1091 | |
| 1092 | function repair( $id ) { |
| 1093 | $repair = $this->get_repair( $id ); |
| 1094 | if ( empty( $repair ) ) { |
| 1095 | $this->log( "🚫 Repair #{$id} does not exist. Cannot repair this." ); |
| 1096 | return false; |
| 1097 | } |
| 1098 | foreach ( $repair->child_ids as $child_id ) { |
| 1099 | if ( !$this->delete( $child_id ) ) { |
| 1100 | $this->log( "🚫 Failed to repair the file." ); |
| 1101 | return false; |
| 1102 | } |
| 1103 | } |
| 1104 | $full_path = $this->get_full_upload_path( $repair->path ); |
| 1105 | $filetype = wp_check_filetype( basename( $full_path ), null ); |
| 1106 | $wp_upload_dir = wp_upload_dir(); |
| 1107 | $attachment = array( |
| 1108 | 'guid' => $wp_upload_dir['url'] . '/' . basename( $full_path ), |
| 1109 | 'post_mime_type' => $filetype['type'], |
| 1110 | 'post_title' => preg_replace( '/\.[^.]+$/', '', basename( $full_path ) ), |
| 1111 | 'post_content' => '', |
| 1112 | 'post_status' => 'inherit' |
| 1113 | ); |
| 1114 | |
| 1115 | $attach_id = wp_insert_attachment( $attachment, $full_path ); |
| 1116 | |
| 1117 | require_once( ABSPATH . 'wp-admin/includes/image.php' ); |
| 1118 | $attach_data = wp_generate_attachment_metadata( $attach_id, $full_path ); |
| 1119 | wp_update_attachment_metadata( $attach_id, $attach_data ); |
| 1120 | |
| 1121 | global $wpdb; |
| 1122 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 1123 | $wpdb->query( $wpdb->prepare( "DELETE FROM $table_name WHERE id = %d OR parentId = %d", $id, $id ) ); |
| 1124 | $this->log( "� |
| 1125 | Repaired {$repair->path}." ); |
| 1126 | return true; |
| 1127 | } |
| 1128 | |
| 1129 | function ignore( $id, $ignore ) { |
| 1130 | global $wpdb; |
| 1131 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 1132 | $issue = $this->get_issue( $id ); |
| 1133 | |
| 1134 | if ( empty( $issue ) ) { |
| 1135 | $this->log( "🚫 Issue #{$id} does not exist. Cannot ignore this." ); |
| 1136 | return false; |
| 1137 | } |
| 1138 | |
| 1139 | if ( !$ignore ) { |
| 1140 | $wpdb->query( $wpdb->prepare( "UPDATE $table_name SET ignored = 0 WHERE id = %d", $id ) ); |
| 1141 | } |
| 1142 | else { |
| 1143 | // If it is in trash, recover it |
| 1144 | if ( $issue->deleted ) { |
| 1145 | $this->recover( $id ); |
| 1146 | } |
| 1147 | $wpdb->query( $wpdb->prepare( "UPDATE $table_name SET ignored = 1 WHERE id = %d", $id ) ); |
| 1148 | } |
| 1149 | return true; |
| 1150 | } |
| 1151 | |
| 1152 | function endsWith( $haystack, $needle ) |
| 1153 | { |
| 1154 | $length = strlen( $needle ); |
| 1155 | if ( $length == 0 ) |
| 1156 | return true; |
| 1157 | return ( substr( $haystack, -$length ) === $needle ); |
| 1158 | } |
| 1159 | |
| 1160 | function clean_dir( $dir ) { |
| 1161 | if ( !file_exists( $dir ) ) |
| 1162 | return; |
| 1163 | else if ( $this->endsWith( $dir, 'uploads' ) ) |
| 1164 | return; |
| 1165 | $found = array_diff( scandir( $dir ), array( '.', '..' ) ); |
| 1166 | if ( count( $found ) < 1 ) { |
| 1167 | if ( rmdir( $dir ) ) { |
| 1168 | $this->clean_dir( dirname( $dir ) ); |
| 1169 | } |
| 1170 | } |
| 1171 | } |
| 1172 | |
| 1173 | function get_issue( $id ) { |
| 1174 | global $wpdb; |
| 1175 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 1176 | $issue = $wpdb->get_row( $wpdb->prepare( "SELECT * FROM $table_name WHERE id = %d", $id ), OBJECT ); |
| 1177 | if ( empty( $issue ) ) { |
| 1178 | return false; |
| 1179 | } |
| 1180 | $issue->id = (int)$issue->id; |
| 1181 | $issue->postId = (int)$issue->postId; |
| 1182 | $issue->type = (int)$issue->type; |
| 1183 | $issue->deleted = (int)$issue->deleted; |
| 1184 | $issue->ignored = (int)$issue->ignored; |
| 1185 | $issue->path = stripslashes( $issue->path ); |
| 1186 | return $issue; |
| 1187 | } |
| 1188 | |
| 1189 | function get_repair( $id ) { |
| 1190 | global $wpdb; |
| 1191 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 1192 | $repair = $wpdb->get_row( $wpdb->prepare( "SELECT |
| 1193 | main.id AS id, |
| 1194 | main.path AS path, |
| 1195 | GROUP_CONCAT(child.id) AS child_ids |
| 1196 | FROM |
| 1197 | $table_name AS main |
| 1198 | LEFT JOIN |
| 1199 | $table_name AS child ON main.id = child.parentId |
| 1200 | WHERE main.id = %d", $id |
| 1201 | ), OBJECT ); |
| 1202 | if ( empty( $repair ) ) { |
| 1203 | return false; |
| 1204 | } |
| 1205 | |
| 1206 | // If $repair->path is null or empty return false |
| 1207 | if ( empty( $repair->path ) ) { |
| 1208 | $this->log( "🚫 Repair #{$id} does not have a path. Cannot repair this." ); |
| 1209 | return false; |
| 1210 | } |
| 1211 | |
| 1212 | |
| 1213 | $repair->id = (int)$repair->id; |
| 1214 | $regex = "^(.*)(\\s\\(\\+.*)$"; |
| 1215 | $repair->path = preg_replace( '/' . $regex . '/i', '$1', stripslashes( $repair->path ) ); |
| 1216 | $repair->child_ids = $repair->child_ids ? explode( ',', $repair->child_ids ) : []; |
| 1217 | return $repair; |
| 1218 | } |
| 1219 | |
| 1220 | function get_issues_to_repair( $order_by = 'id', $order = 'asc', $search = '', $skip = 0, $limit = 10 ) { |
| 1221 | global $wpdb; |
| 1222 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 1223 | |
| 1224 | $search_clause = ''; |
| 1225 | if ( !empty( $search ) ) { |
| 1226 | $search_clause = $wpdb->prepare("AND main.path LIKE %s", ( '%' . $search . '%' )); |
| 1227 | } |
| 1228 | |
| 1229 | $order_clause = 'ORDER BY main.id ASC'; |
| 1230 | if ( $order_by === 'path' ) { |
| 1231 | $order_clause = 'ORDER BY main.path ' . ( $order === 'asc' ? 'ASC' : 'DESC' ); |
| 1232 | } |
| 1233 | else if ( $order_by === 'issue' ) { |
| 1234 | $order_clause = 'ORDER BY main.issue ' . ( $order === 'asc' ? 'ASC' : 'DESC' ); |
| 1235 | } |
| 1236 | else if ( $order_by === 'size' ) { |
| 1237 | $order_clause = 'ORDER BY main.size ' . ( $order === 'asc' ? 'ASC' : 'DESC' ); |
| 1238 | } |
| 1239 | |
| 1240 | $result = $wpdb->get_results( $wpdb->prepare( "SELECT |
| 1241 | main.id AS id, |
| 1242 | main.path AS path, |
| 1243 | GROUP_CONCAT(child.id) AS child_ids, |
| 1244 | GROUP_CONCAT(child.path) AS child_paths, |
| 1245 | main.type AS type, |
| 1246 | main.postId AS postId, |
| 1247 | main.size AS size, |
| 1248 | main.ignored AS ignored, |
| 1249 | main.deleted AS deleted, |
| 1250 | main.issue AS issue |
| 1251 | FROM |
| 1252 | $table_name AS main |
| 1253 | LEFT JOIN |
| 1254 | $table_name AS child ON main.id = child.parentId |
| 1255 | WHERE |
| 1256 | main.path IS NOT NULL AND main.parentId IS NULL |
| 1257 | AND main.deleted = 0 AND main.ignored = 0 |
| 1258 | AND main.type = 0 |
| 1259 | $search_clause |
| 1260 | GROUP BY main.id |
| 1261 | $order_clause |
| 1262 | LIMIT %d, %d; |
| 1263 | ", $skip, $limit ) ); |
| 1264 | |
| 1265 | return $result; |
| 1266 | } |
| 1267 | |
| 1268 | function get_repair_ids ( $search = '' ) { |
| 1269 | global $wpdb; |
| 1270 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 1271 | |
| 1272 | $search_clause = ''; |
| 1273 | if ( !empty( $search ) ) { |
| 1274 | $search_clause = $wpdb->prepare("AND main.path LIKE %s", ( '%' . $search . '%' )); |
| 1275 | } |
| 1276 | |
| 1277 | return $wpdb->get_col( "SELECT DISTINCT main.id |
| 1278 | FROM |
| 1279 | $table_name AS main |
| 1280 | LEFT JOIN $table_name AS child ON main.id = child.parentId |
| 1281 | WHERE |
| 1282 | main.path IS NOT NULL |
| 1283 | AND main.parentId IS NULL |
| 1284 | $search_clause |
| 1285 | GROUP BY |
| 1286 | main.id |
| 1287 | ;" |
| 1288 | ); |
| 1289 | } |
| 1290 | |
| 1291 | function get_stats_of_issues_to_repair( $search = '' ) { |
| 1292 | global $wpdb; |
| 1293 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 1294 | |
| 1295 | $search_clause = ''; |
| 1296 | if ( !empty( $search ) ) { |
| 1297 | $search_clause = $wpdb->prepare("AND main.path LIKE %s", ( '%' . $search . '%' )); |
| 1298 | } |
| 1299 | |
| 1300 | return $wpdb->get_row( "SELECT |
| 1301 | COUNT(id) AS entries, |
| 1302 | SUM(size) AS size |
| 1303 | FROM ( |
| 1304 | SELECT |
| 1305 | COUNT(DISTINCT main.id) as id, |
| 1306 | main.size as size |
| 1307 | FROM |
| 1308 | $table_name AS main |
| 1309 | LEFT JOIN |
| 1310 | $table_name AS child ON main.id = child.parentId |
| 1311 | WHERE |
| 1312 | main.path IS NOT NULL AND main.parentId IS NULL AND main.deleted = 0 AND main.ignored = 0 |
| 1313 | $search_clause |
| 1314 | GROUP BY main.id |
| 1315 | ) t; |
| 1316 | " ); |
| 1317 | } |
| 1318 | |
| 1319 | function get_count_of_issues_to_repair( $search ) { |
| 1320 | $stats = $this->get_stats_of_issues_to_repair( $search ); |
| 1321 | return $stats->entries; |
| 1322 | } |
| 1323 | |
| 1324 | function delete( $id ) { |
| 1325 | global $wpdb; |
| 1326 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 1327 | $issue = $this->get_issue( $id ); |
| 1328 | |
| 1329 | if ( empty( $issue ) ) { |
| 1330 | $this->log( "🚫 Issue #{$id} does not exist. Cannot delete this." ); |
| 1331 | return false; |
| 1332 | } |
| 1333 | |
| 1334 | $regex = "^(.*)(\\s\\(\\+.*)$"; |
| 1335 | $issue->path = preg_replace( '/' . $regex . '/i', '$1', $issue->path ); // remove " (+ 6 files)" from path |
| 1336 | $skip_trash = $this->get_option( 'skip_trash' ); |
| 1337 | |
| 1338 | if ( $issue->type === 0 ) { |
| 1339 | |
| 1340 | // Delete file from the trash |
| 1341 | if ( $issue->deleted === 1 ) { |
| 1342 | $trashPath = trailingslashit( $this->get_trashdir() ) . $issue->path; |
| 1343 | if ( unlink( $trashPath ) ) { |
| 1344 | $wpdb->query( $wpdb->prepare( "DELETE FROM $table_name WHERE id = %d", $id ) ); |
| 1345 | $this->clean_dir( dirname( $trashPath ) ); |
| 1346 | return true; |
| 1347 | } |
| 1348 | } |
| 1349 | // Delete file without using trash |
| 1350 | else if ( $skip_trash ) { |
| 1351 | $originalPath = trailingslashit( $this->upload_path ) . $issue->path; |
| 1352 | if ( unlink( $originalPath ) ) { |
| 1353 | $wpdb->query( $wpdb->prepare( "DELETE FROM $table_name WHERE id = %d", $id ) ); |
| 1354 | $this->clean_dir( dirname( $originalPath ) ); |
| 1355 | return true; |
| 1356 | } |
| 1357 | } |
| 1358 | // Move file to the trash |
| 1359 | else if ( $this->trash_file( $issue->path ) ) { |
| 1360 | $wpdb->query( $wpdb->prepare( "UPDATE $table_name SET deleted = 1, ignored = 0, time = NOW() WHERE id = %d", $id ) ); |
| 1361 | return true; |
| 1362 | } |
| 1363 | |
| 1364 | $this->log( "🚫 Failed to delete/trash the file." ); |
| 1365 | error_log( "Media Cleaner: Failed to delete/trash the file." ); |
| 1366 | } |
| 1367 | |
| 1368 | if ( $issue->type === 1 ) { |
| 1369 | |
| 1370 | // Trash Media definitely by recovering it (to be like a normal Media) and remove it through the |
| 1371 | // standard WordPress workflow |
| 1372 | if ( $issue->deleted === 1 || $skip_trash ) { |
| 1373 | if ( $issue->deleted === 1 ) { |
| 1374 | $this->recover( $id ); |
| 1375 | } |
| 1376 | wp_update_post( array( 'ID' => $issue->postId, 'post_type' => 'attachment' ) ); |
| 1377 | wp_delete_attachment( $issue->postId, true ); |
| 1378 | $wpdb->query( $wpdb->prepare( "DELETE FROM $table_name WHERE id = %d", $id ) ); |
| 1379 | return true; |
| 1380 | } |
| 1381 | else { |
| 1382 | // Move Media to trash |
| 1383 | // Let's copy the images to the trash so that it can be recovered. |
| 1384 | $paths = $this->get_paths_from_attachment( $issue->postId ); |
| 1385 | foreach ( $paths as $path ) { |
| 1386 | if ( !$this->trash_file( $path ) ) { |
| 1387 | $this->log( "🚫 Could not trash $path." ); |
| 1388 | error_log( "Media Cleaner: Could not trash $path." ); |
| 1389 | return false; |
| 1390 | } |
| 1391 | } |
| 1392 | wp_update_post( array( 'ID' => $issue->postId, 'post_type' => 'wmpc-trash' ) ); |
| 1393 | $wpdb->query( $wpdb->prepare( "UPDATE $table_name SET deleted = 1, ignored = 0, time = NOW() WHERE id = %d", $id ) ); |
| 1394 | return true; |
| 1395 | } |
| 1396 | } |
| 1397 | return false; |
| 1398 | } |
| 1399 | |
| 1400 | function delete_directory_recurcively( $dir ) { |
| 1401 | if ( !is_dir( $dir ) ) { |
| 1402 | return; |
| 1403 | } |
| 1404 | $files = array_diff( scandir( $dir ), array( '.', '..' ) ); |
| 1405 | foreach ( $files as $file ) { |
| 1406 | if ( is_dir( "$dir/$file" ) ) { |
| 1407 | $this->delete_directory_recurcively( "$dir/$file" ); |
| 1408 | } |
| 1409 | else { |
| 1410 | unlink( "$dir/$file" ); |
| 1411 | } |
| 1412 | } |
| 1413 | rmdir( $dir ); |
| 1414 | } |
| 1415 | |
| 1416 | function force_trash() { |
| 1417 | |
| 1418 | $res = [ |
| 1419 | 'message' => 'The trash folder has been emptied.', |
| 1420 | 'success' => true |
| 1421 | ]; |
| 1422 | |
| 1423 | // Delete all the files in the trash folder. |
| 1424 | $trashDirPath = trailingslashit( $this->get_trashdir() ); |
| 1425 | if ( file_exists( $trashDirPath ) && is_dir( $trashDirPath ) ) { |
| 1426 | $this->delete_directory_recurcively( $trashDirPath, true ); |
| 1427 | } |
| 1428 | |
| 1429 | // Clean the Database: DELETE FROM wp_mclean_scan WHERE deleted = 1 |
| 1430 | global $wpdb; |
| 1431 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 1432 | $wpdb->query( $wpdb->prepare( "DELETE FROM $table_name WHERE deleted = 1" ) ); |
| 1433 | |
| 1434 | |
| 1435 | return $res; |
| 1436 | } |
| 1437 | |
| 1438 | /** |
| 1439 | * |
| 1440 | * SCANNING / RESET |
| 1441 | * |
| 1442 | */ |
| 1443 | |
| 1444 | function add_reference_url( $urlOrUrls, $type, $origin = null, $extra = null ) { |
| 1445 | $urlOrUrls = !is_array( $urlOrUrls ) ? array( $urlOrUrls ) : $urlOrUrls; |
| 1446 | foreach ( $urlOrUrls as $url ) { |
| 1447 | // With files, we need both filename without resolution and filename with resolution, it's important |
| 1448 | // to make sure the original file is not deleted if a size exists for it. |
| 1449 | // With media, all URLs should be without resolution to make sure it matches Media. |
| 1450 | if ( $this->current_method == 'files' ) { |
| 1451 | $this->add_reference( null, $url, $type, $origin, $extra ); |
| 1452 | $this->add_reference( 0, $this->clean_url_from_resolution( $url ), $type, $origin, $extra ); |
| 1453 | } |
| 1454 | else { |
| 1455 | // 2021/11/08: I added this, the problem is that sometimes users create image filenames with the resolution |
| 1456 | // in it, even though it is the original. |
| 1457 | $this->add_reference( null, $url, $type, $origin, $extra ); |
| 1458 | |
| 1459 | $this->add_reference( 0, $this->clean_url_from_resolution( $url ), $type, $origin, $extra ); |
| 1460 | } |
| 1461 | } |
| 1462 | } |
| 1463 | |
| 1464 | function add_reference_id( $idOrIds, $type, $origin = null, $extra = null ) { |
| 1465 | $idOrIds = !is_array( $idOrIds ) ? array( $idOrIds ) : $idOrIds; |
| 1466 | foreach ( $idOrIds as $id ) { |
| 1467 | $this->add_reference( $id, "", $type, $origin ); |
| 1468 | if ( $this->multilingual ) { |
| 1469 | $translatedIds = $this->get_translated_media_ids( (int)$id ); |
| 1470 | |
| 1471 | // Test for WPML |
| 1472 | // if ( $id === '350') { |
| 1473 | // $translatedIds = $this->get_translated_media_ids( (int)$id ); |
| 1474 | // $count = count($translatedIds); |
| 1475 | // error_log( "${id} => ${count}" ); |
| 1476 | // } |
| 1477 | |
| 1478 | if ( !empty( $translatedIds ) ) { |
| 1479 | foreach ( $translatedIds as $translatedId ) { |
| 1480 | $this->add_reference( $translatedId, "", $type, $origin ); |
| 1481 | } |
| 1482 | } |
| 1483 | } |
| 1484 | } |
| 1485 | } |
| 1486 | |
| 1487 | |
| 1488 | // Returns the reference with the type, origin, related to a Media ID it is referenced |
| 1489 | public function get_reference_for_media_id( $id ) { |
| 1490 | global $wpdb; |
| 1491 | $table_name = $wpdb->prefix . "mclean_refs"; |
| 1492 | $refs = $wpdb->get_results( $wpdb->prepare( "SELECT * FROM $table_name WHERE mediaId = %d", $id ), OBJECT ); |
| 1493 | if ( empty( $refs ) ) { |
| 1494 | return false; |
| 1495 | } |
| 1496 | $ref = $refs[0]; |
| 1497 | $ref->id = (int)$ref->id; |
| 1498 | $ref->mediaId = (int)$ref->mediaId; |
| 1499 | $ref->originType = (int)$ref->originType; |
| 1500 | $ref->origin = stripslashes( $ref->origin ); |
| 1501 | $ref->parentId = empty( $ref->parentId ) ? null : (int)$ref->parentId; |
| 1502 | return $ref; |
| 1503 | } |
| 1504 | |
| 1505 | // Return the references related to a Post ID |
| 1506 | public function get_references_for_post_id( $id ) { |
| 1507 | global $wpdb; |
| 1508 | $table_name = $wpdb->prefix . "mclean_refs"; |
| 1509 | $refs = $wpdb->get_results( $wpdb->prepare( "SELECT * FROM $table_name WHERE originType LIKE %s", "%[$id]" ), OBJECT ); |
| 1510 | if ( empty( $refs ) ) { |
| 1511 | return []; |
| 1512 | } |
| 1513 | $fresh_refs = array(); |
| 1514 | foreach ( $refs as $ref ) { |
| 1515 | $mediaId = (int)$ref->mediaId > 0 ? (int)$ref->mediaId : null; |
| 1516 | if ( !$mediaId && !empty( $ref->mediaUrl ) ) { |
| 1517 | $mediaId = $this->find_media_id_from_file( $ref->mediaUrl, false ); |
| 1518 | $mediaId = !empty( $mediaId ) ? (int)$mediaId : null; |
| 1519 | } |
| 1520 | if ( !$mediaId ) { |
| 1521 | continue; |
| 1522 | } |
| 1523 | array_push( $fresh_refs, [ |
| 1524 | 'id' => (int)$ref->id, |
| 1525 | 'mediaId' => $mediaId, |
| 1526 | 'mediaUrl' => $ref->mediaUrl, |
| 1527 | 'originType' => $ref->originType, |
| 1528 | 'parentId' => empty( $ref->parentId ) ? null : (int)$ref->parentId, |
| 1529 | ] ); |
| 1530 | } |
| 1531 | return $fresh_refs; |
| 1532 | } |
| 1533 | |
| 1534 | // The references are actually not being added directly in the DB, they are being pushed |
| 1535 | // into a cache ($this->refcache). |
| 1536 | private function add_reference( $id, $url, $type, $origin = null, $extra = null ) { |
| 1537 | |
| 1538 | $force_no_cache = $extra && isset( $extra['force_no_cache'] ) ? $extra['force_no_cache'] : false; |
| 1539 | $force_cache = $extra && isset( $extra['force_cache'] ) ? $extra['force_cache'] : false; |
| 1540 | if ( $force_no_cache ) { |
| 1541 | $this->use_cached_references = false; |
| 1542 | } |
| 1543 | |
| 1544 | if ( $force_cache ) { |
| 1545 | $this->use_cached_references = true; |
| 1546 | } |
| 1547 | |
| 1548 | if ( !empty( $origin ) ) { |
| 1549 | $type = $type . " [$origin]"; |
| 1550 | } |
| 1551 | |
| 1552 | if ( !empty( $id ) ) { |
| 1553 | |
| 1554 | if( $this->use_cached_references ) { |
| 1555 | |
| 1556 | $added = $this->add_cached_id( $id ); |
| 1557 | if ( $added ) { |
| 1558 | array_push( $this->refcache, array( 'id' => $id, 'url' => null, 'type' => $type, 'origin' => $origin ) ); |
| 1559 | } |
| 1560 | |
| 1561 | |
| 1562 | } |
| 1563 | |
| 1564 | if( !$this->use_cached_references ) { |
| 1565 | array_push( $this->refcache, array( 'id' => $id, 'url' => null, 'type' => $type, 'origin' => $origin ) ); |
| 1566 | } |
| 1567 | |
| 1568 | } |
| 1569 | if ( !empty( $url ) ) { |
| 1570 | // The URL shouldn't contain http, https, javascript at the beginning (and there are probably many more cases) |
| 1571 | // The URL must be cleaned before being passed as a reference. |
| 1572 | if ( substr( $url, 0, 5 ) === "http:" || substr( $url, 0, 6 ) === "https:" || substr( $url, 0, 11 ) === "javascript:" ) { |
| 1573 | return; |
| 1574 | } |
| 1575 | |
| 1576 | if( $this->use_cached_references ) { |
| 1577 | |
| 1578 | $added = $this->add_cached_url( $url ); |
| 1579 | if ( $added ) { |
| 1580 | array_push( $this->refcache, array( 'id' => null, 'url' => $url, 'type' => $type, 'origin' => $origin ) ); |
| 1581 | } |
| 1582 | |
| 1583 | } |
| 1584 | |
| 1585 | if( !$this->use_cached_references ) { |
| 1586 | array_push( $this->refcache, array( 'id' => null, 'url' => $url, 'type' => $type, 'origin' => $origin ) ); |
| 1587 | } |
| 1588 | |
| 1589 | } |
| 1590 | |
| 1591 | } |
| 1592 | |
| 1593 | private function get_cached_ids() { |
| 1594 | global $wpdb; |
| 1595 | $table_name = $wpdb->prefix . "mclean_cache"; |
| 1596 | $cached_ids = $wpdb->get_col( $wpdb->prepare( |
| 1597 | "SELECT cache_value FROM $table_name WHERE cache_key = %s AND cache_type = %s", |
| 1598 | $this->cached_ids_key, |
| 1599 | 'id' |
| 1600 | ) ); |
| 1601 | return $cached_ids ? $cached_ids : array(); |
| 1602 | } |
| 1603 | |
| 1604 | private function get_cached_urls() { |
| 1605 | global $wpdb; |
| 1606 | $table_name = $wpdb->prefix . "mclean_cache"; |
| 1607 | $cached_urls = $wpdb->get_col( $wpdb->prepare( |
| 1608 | "SELECT cache_value FROM $table_name WHERE cache_key = %s AND cache_type = %s", |
| 1609 | $this->cached_urls_key, |
| 1610 | 'url' |
| 1611 | ) ); |
| 1612 | return $cached_urls ? $cached_urls : array(); |
| 1613 | } |
| 1614 | |
| 1615 | private function add_cached_id($id) { |
| 1616 | global $wpdb; |
| 1617 | $table_name = $wpdb->prefix . "mclean_cache"; |
| 1618 | |
| 1619 | // Try to insert, ignore if duplicate (UNIQUE KEY will prevent duplicates) |
| 1620 | $result = $wpdb->query( $wpdb->prepare( |
| 1621 | "INSERT IGNORE INTO $table_name (cache_key, cache_value, cache_type) VALUES (%s, %s, %s)", |
| 1622 | $this->cached_ids_key, |
| 1623 | $id, |
| 1624 | 'id' |
| 1625 | ) ); |
| 1626 | |
| 1627 | // Return true if a row was inserted |
| 1628 | return $result > 0; |
| 1629 | } |
| 1630 | |
| 1631 | private function add_cached_url($url) { |
| 1632 | global $wpdb; |
| 1633 | $table_name = $wpdb->prefix . "mclean_cache"; |
| 1634 | |
| 1635 | // Try to insert, ignore if duplicate (UNIQUE KEY will prevent duplicates) |
| 1636 | $result = $wpdb->query( $wpdb->prepare( |
| 1637 | "INSERT IGNORE INTO $table_name (cache_key, cache_value, cache_type) VALUES (%s, %s, %s)", |
| 1638 | $this->cached_urls_key, |
| 1639 | $url, |
| 1640 | 'url' |
| 1641 | ) ); |
| 1642 | |
| 1643 | // Return true if a row was inserted |
| 1644 | return $result > 0; |
| 1645 | } |
| 1646 | |
| 1647 | function reset_cached_references() { |
| 1648 | global $wpdb; |
| 1649 | $table_name = $wpdb->prefix . "mclean_cache"; |
| 1650 | |
| 1651 | // Delete all cached references from the cache table |
| 1652 | $wpdb->query( "TRUNCATE TABLE $table_name" ); |
| 1653 | |
| 1654 | $this->cached_ids_cli = array(); |
| 1655 | $this->cached_urls_cli = array(); |
| 1656 | } |
| 1657 | |
| 1658 | function insert_references($entries) |
| 1659 | { |
| 1660 | global $wpdb; |
| 1661 | $table = $wpdb->prefix . "mclean_refs"; |
| 1662 | $values = array(); |
| 1663 | $place_holders = array(); |
| 1664 | $query = "INSERT INTO $table (mediaId, mediaUrl, originType, parentId) VALUES "; |
| 1665 | |
| 1666 | foreach ( $entries as $value ) { |
| 1667 | if ( !is_null($value['id'] ) ) { |
| 1668 | // Media Reference |
| 1669 | array_push( $values, $value['id'], $value['type'] ); |
| 1670 | $place_holders[] = "('%d', NULL, '%s', NULL)"; |
| 1671 | |
| 1672 | if ($this->debug_logs) { |
| 1673 | $this->log("＋ Media #{$value['id']} (as ID)"); |
| 1674 | } |
| 1675 | } |
| 1676 | else if ( !is_null($value['url'] ) ) { |
| 1677 | // File Reference |
| 1678 | array_push( $values, $value['url'], $value['type'] ); |
| 1679 | if ( isset( $value['parentId'] ) ) { |
| 1680 | array_push( $values, $value['parentId'] ); |
| 1681 | $place_holders[] = "(NULL, '%s', '%s', '%d')"; |
| 1682 | if ( $this->debug_logs ) { |
| 1683 | $this->log( "＋ {$value['url']} (as URL) (ParentID: {$value['parentId']})" ); |
| 1684 | } |
| 1685 | } else { |
| 1686 | $place_holders[] = "(NULL, '%s', '%s', NULL)"; |
| 1687 | if ( $this->debug_logs ) { |
| 1688 | $this->log("＋ {$value['url']} (as URL)"); |
| 1689 | } |
| 1690 | } |
| 1691 | } |
| 1692 | } |
| 1693 | |
| 1694 | if ( !empty( $values ) ) { |
| 1695 | $query .= implode( ', ', $place_holders ); |
| 1696 | $prepared = $wpdb->prepare( "$query ", $values ); |
| 1697 | $wpdb->query( $prepared ); |
| 1698 | } |
| 1699 | } |
| 1700 | |
| 1701 | function reset_progress() { |
| 1702 | // Reset the progress by deleting the transient. |
| 1703 | delete_transient( $this->progress_key ); |
| 1704 | } |
| 1705 | |
| 1706 | function clear_step_progress() { |
| 1707 | // Clear step progress when scanning completes |
| 1708 | delete_transient( $this->progress_key ); |
| 1709 | } |
| 1710 | |
| 1711 | function save_progress( $step, $data = array() ) { |
| 1712 | // Save progress with step and optional data |
| 1713 | // Data can include type, limit, limitSize, and any other progress information |
| 1714 | $progress = array( |
| 1715 | 'step' => $step, |
| 1716 | 'time' => time(), |
| 1717 | 'data' => $data |
| 1718 | ); |
| 1719 | |
| 1720 | set_transient( $this->progress_key, $progress, 0 ); |
| 1721 | } |
| 1722 | |
| 1723 | function get_progress() { |
| 1724 | return get_transient( $this->progress_key ); |
| 1725 | } |
| 1726 | |
| 1727 | function get_step_progress() { |
| 1728 | $options = $this->get_all_options(); |
| 1729 | return isset( $options['step_progress'] ) ? $options['step_progress'] : null; |
| 1730 | } |
| 1731 | |
| 1732 | // The cache containing the references is wrote to the DB. |
| 1733 | function write_references() { |
| 1734 | global $wpdb; |
| 1735 | $table = $wpdb->prefix . "mclean_refs"; |
| 1736 | |
| 1737 | $potential_parents = array(); |
| 1738 | $potential_children = array(); |
| 1739 | |
| 1740 | foreach ( $this->refcache as $value ) { |
| 1741 | $potentialParentPath = !is_null( $value['url'] ) ? $this->clean_url_from_resolution( $value['url'] ) : null; |
| 1742 | if ( $potentialParentPath === $value['url'] ) { |
| 1743 | $potential_parents[] = $value; |
| 1744 | } |
| 1745 | else { |
| 1746 | $potential_children[] = $value; |
| 1747 | } |
| 1748 | } |
| 1749 | |
| 1750 | $this->insert_references( $potential_parents ); |
| 1751 | |
| 1752 | // Resolve parentId for potential children |
| 1753 | foreach ( $potential_children as &$child ) { |
| 1754 | $potentialParentPath = $this->clean_url_from_resolution( $child['url'] ); |
| 1755 | $parentId = $wpdb->get_var( $wpdb->prepare( "SELECT id FROM $table WHERE mediaUrl = %s", $potentialParentPath ) ); |
| 1756 | if ( !empty( $parentId ) ) { |
| 1757 | $child['parentId'] = (int)$parentId; |
| 1758 | } |
| 1759 | } |
| 1760 | |
| 1761 | // Insert potential children with resolved parentIds |
| 1762 | $this->insert_references( $potential_children ); |
| 1763 | $this->refcache = array(); |
| 1764 | } |
| 1765 | |
| 1766 | function check_is_ignore( $file ) { |
| 1767 | global $wpdb; |
| 1768 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 1769 | $count = $wpdb->get_var( "SELECT COUNT(*) |
| 1770 | FROM $table_name |
| 1771 | WHERE ignored = 1 |
| 1772 | AND path LIKE '%". esc_sql( $wpdb->esc_like( $file ) ) . "%'" ); |
| 1773 | if ( $count > 0 ) { |
| 1774 | $this->log( "🚫 Could not trash $file." ); |
| 1775 | } |
| 1776 | return ($count > 0); |
| 1777 | } |
| 1778 | |
| 1779 | function find_media_id_from_file( $file, $doLog ) { |
| 1780 | global $wpdb; |
| 1781 | $postmeta_table_name = $wpdb->prefix . 'postmeta'; |
| 1782 | $file = $this->clean_uploaded_filename( $file ); |
| 1783 | $sql = $wpdb->prepare( "SELECT post_id |
| 1784 | FROM {$postmeta_table_name} |
| 1785 | WHERE meta_key = '_wp_attached_file' |
| 1786 | AND meta_value = %s", $file |
| 1787 | ); |
| 1788 | $ret = $wpdb->get_var( $sql ); |
| 1789 | if ( $doLog ) { |
| 1790 | if ( empty( $ret ) ) |
| 1791 | $this->log( "🚫 File $file not found as _wp_attached_file (Library)." ); |
| 1792 | else { |
| 1793 | $this->log( "� |
| 1794 | File $file found as Media $ret." ); |
| 1795 | } |
| 1796 | } |
| 1797 | |
| 1798 | return $ret; |
| 1799 | } |
| 1800 | |
| 1801 | function get_thumbnails_urls( $id, $sizes_as_key = false ) { |
| 1802 | $sizes = get_intermediate_image_sizes(); |
| 1803 | // For each size use wp_get_attachment_image_src() to get the URL |
| 1804 | $urls = array(); |
| 1805 | foreach ( $sizes as $size ) { |
| 1806 | $src = wp_get_attachment_image_src( $id, $size ); |
| 1807 | if ( $src ) { |
| 1808 | $urls[$size] = $this->clean_url( $src[0] ); |
| 1809 | } |
| 1810 | } |
| 1811 | |
| 1812 | return $sizes_as_key ? $urls : array_values( $urls ); |
| 1813 | } |
| 1814 | |
| 1815 | |
| 1816 | function get_thumbnails_urls_from_srcset( $id, $size = 'full' ) { |
| 1817 | |
| 1818 | $image_size = $this->get_attachment_size_by_id( $id, $size ); |
| 1819 | |
| 1820 | $sizes = array_keys( $this->get_image_sizes() ); |
| 1821 | $sizes[] = $image_size; |
| 1822 | |
| 1823 | $urls = array(); |
| 1824 | foreach ( $sizes as $image_size ) { |
| 1825 | $srcset = wp_get_attachment_image_srcset( $id, $image_size ); |
| 1826 | |
| 1827 | // Extract URLs from srcset |
| 1828 | if ( !empty( $srcset ) ) { |
| 1829 | $srcset = explode( ', ', $srcset ); |
| 1830 | foreach ( $srcset as $src ) { |
| 1831 | $parts = explode( ' ', $src ); |
| 1832 | $url = trim( $parts[0] ); |
| 1833 | if ( !empty( $url ) ) { |
| 1834 | $urls[] = $this->clean_url( $url ); |
| 1835 | } |
| 1836 | } |
| 1837 | } |
| 1838 | } |
| 1839 | |
| 1840 | return $urls; |
| 1841 | |
| 1842 | } |
| 1843 | |
| 1844 | function get_attachment_size_by_id( $attachment_id, $default_size = 'full' ) { |
| 1845 | |
| 1846 | if ( ! $attachment_id ) { |
| 1847 | return $default_size; |
| 1848 | } |
| 1849 | |
| 1850 | $url = wp_get_attachment_url( $attachment_id ); |
| 1851 | if ( ! $url ) { |
| 1852 | return $default_size; |
| 1853 | } |
| 1854 | |
| 1855 | $metadata = wp_get_attachment_metadata( $attachment_id ); |
| 1856 | |
| 1857 | if ( ! is_array( $metadata ) ) { |
| 1858 | return $default_size; |
| 1859 | } |
| 1860 | |
| 1861 | $size = $default_size; |
| 1862 | |
| 1863 | if ( isset( $metadata['file'] ) && strpos( $url, $metadata['file'] ) === ( strlen( $url ) - strlen( $metadata['file'] ) ) ) { |
| 1864 | $size = array( $metadata['width'], $metadata['height'] ); |
| 1865 | } elseif ( preg_match( '/-(\d+)x(\d+)\.(jpg|jpeg|gif|png|svg|webp)$/', $url, $match ) ) { |
| 1866 | // Get the image width and height. |
| 1867 | // Example: https://regex101.com/r/7JwGz7/1. |
| 1868 | $size = array( $match[1], $match[2] ); |
| 1869 | } |
| 1870 | |
| 1871 | return $size; |
| 1872 | } |
| 1873 | |
| 1874 | function get_image_sizes() { |
| 1875 | $sizes = array(); |
| 1876 | global $_wp_additional_image_sizes; |
| 1877 | foreach ( get_intermediate_image_sizes() as $s ) { |
| 1878 | $crop = false; |
| 1879 | if ( isset( $_wp_additional_image_sizes[$s] ) ) { |
| 1880 | $width = intval( $_wp_additional_image_sizes[$s]['width'] ); |
| 1881 | $height = intval( $_wp_additional_image_sizes[$s]['height'] ); |
| 1882 | $crop = $_wp_additional_image_sizes[$s]['crop']; |
| 1883 | } else { |
| 1884 | $width = get_option( $s.'_size_w' ); |
| 1885 | $height = get_option( $s.'_size_h' ); |
| 1886 | $crop = get_option( $s.'_crop' ); |
| 1887 | } |
| 1888 | $sizes[$s] = array( 'width' => $width, 'height' => $height, 'crop' => $crop ); |
| 1889 | } |
| 1890 | return $sizes; |
| 1891 | } |
| 1892 | |
| 1893 | function clean_url_from_resolution( $url ) { |
| 1894 | if ( !isset( $url ) ) return $url; |
| 1895 | |
| 1896 | $pattern = '/[_-]\d+x\d+(?=\.[a-z]{3,4}$)/'; |
| 1897 | $url = preg_replace( $pattern, '', $url ); |
| 1898 | return $url; |
| 1899 | } |
| 1900 | |
| 1901 | function is_url( $url ) { |
| 1902 | return ( ( |
| 1903 | !empty( $url ) ) && |
| 1904 | is_string( $url ) && |
| 1905 | strlen( $url ) > 4 && ( |
| 1906 | strtolower( substr( $url, 0, 4) ) == 'http' || $url[0] == '/' |
| 1907 | ) |
| 1908 | ); |
| 1909 | } |
| 1910 | |
| 1911 | function clean_url_from_resolution_ref( &$url ) { |
| 1912 | $url = $this->clean_url_from_resolution( $url ); |
| 1913 | } |
| 1914 | |
| 1915 | // From a url to the shortened and cleaned url (for example '2013/02/file.png') |
| 1916 | function clean_url( $url ) { |
| 1917 | // if ( is_array( $url ) ) { |
| 1918 | // error_log( print_r( $url, 1 ) ); |
| 1919 | // } |
| 1920 | $dirIndex = strpos( $url, $this->upload_url ); |
| 1921 | if ( empty( $url ) || $dirIndex === false ) { |
| 1922 | $finalUrl = null; |
| 1923 | } |
| 1924 | else { |
| 1925 | $finalUrl = urldecode( substr( $url, 1 + strlen( $this->upload_url ) + $dirIndex ) ); |
| 1926 | } |
| 1927 | return $finalUrl; |
| 1928 | } |
| 1929 | |
| 1930 | function custom_attachment_url_to_postid( $url ) { |
| 1931 | global $wpdb; |
| 1932 | |
| 1933 | // Remove the query string |
| 1934 | $url = preg_replace('/\?.*/', '', $url); |
| 1935 | |
| 1936 | // Try to find the attachment ID by matching the URL with the guid |
| 1937 | $attachment = $wpdb->get_col( $wpdb->prepare( "SELECT ID FROM $wpdb->posts WHERE guid LIKE %s AND post_type = 'attachment';", '%' . $wpdb->esc_like( $url ) ) ); |
| 1938 | |
| 1939 | // If found, return the first attachment ID |
| 1940 | if ( !empty( $attachment ) ) { |
| 1941 | return ( int )$attachment[0]; |
| 1942 | } |
| 1943 | |
| 1944 | // If not found, try to match the URL without the upload directory path |
| 1945 | $upload_dir = wp_upload_dir(); |
| 1946 | $url_relative = str_replace( $upload_dir['baseurl'] . '/', '', $url ); |
| 1947 | |
| 1948 | $attachment = $wpdb->get_col( $wpdb->prepare( "SELECT post_id FROM $wpdb->postmeta WHERE meta_key = '_wp_attached_file' AND meta_value LIKE %s;", '%' . $wpdb->esc_like( $url_relative ) ) ); |
| 1949 | |
| 1950 | // If found, return the first attachment ID |
| 1951 | if ( !empty( $attachment ) ) { |
| 1952 | return ( int )$attachment[0]; |
| 1953 | } |
| 1954 | |
| 1955 | // If still not found, return 0 |
| 1956 | return 0; |
| 1957 | } |
| 1958 | |
| 1959 | // From a fullpath to the shortened and cleaned path (for example '2013/02/file.png') |
| 1960 | // Original version by Jordy |
| 1961 | // function clean_uploaded_filename( $fullpath ) { |
| 1962 | // $basedir = $this->upload_path; |
| 1963 | // $file = str_replace( $basedir, '', $fullpath ); |
| 1964 | // $file = str_replace( "./", "", $file ); |
| 1965 | // $file = trim( $file, "/" ); |
| 1966 | // return $file; |
| 1967 | // } |
| 1968 | |
| 1969 | // From a fullpath to the shortened and cleaned path (for example '2013/02/file.png') |
| 1970 | // Faster version, more difficult to read, by Mike Meinz |
| 1971 | function clean_uploaded_filename( $fullpath ) { |
| 1972 | $dirIndex = strpos( $fullpath, $this->upload_url ); |
| 1973 | if ( $dirIndex == false ) { |
| 1974 | $file = $fullpath; |
| 1975 | } |
| 1976 | else { |
| 1977 | // Remove first part of the path leaving yyyy/mm/filename.ext |
| 1978 | $file = substr( $fullpath, 1 + strlen( $this->upload_url ) + $dirIndex ); |
| 1979 | } |
| 1980 | if ( substr( $file, 0, 2 ) == './' ) { |
| 1981 | $file = substr( $file, 2 ); |
| 1982 | } |
| 1983 | if ( substr( $file, 0, 1 ) == '/' ) { |
| 1984 | $file = substr( $file, 1 ); |
| 1985 | } |
| 1986 | return $file; |
| 1987 | } |
| 1988 | |
| 1989 | /** |
| 1990 | * Check if the file or the Media ID is used in the install. |
| 1991 | * That file or ID will be checked against the database of references created by the plugin |
| 1992 | * by the parsers. |
| 1993 | */ |
| 1994 | function reference_exists( $file, $mediaId ) { |
| 1995 | global $wpdb; |
| 1996 | |
| 1997 | $table = $wpdb->prefix . "mclean_refs"; |
| 1998 | |
| 1999 | $row = null; |
| 2000 | if ( !empty( $mediaId ) ) { |
| 2001 | $row = $wpdb->get_row( $wpdb->prepare( "SELECT originType FROM $table WHERE mediaId = %d", $mediaId ) ); |
| 2002 | if ( !empty( $row ) ) { |
| 2003 | $origin = $row->originType === 'MEDIA LIBRARY' ? 'Media Library' : 'content'; |
| 2004 | $this->log( "� |
| 2005 | Media #{$mediaId} used by {$origin}" ); |
| 2006 | return $row->originType; |
| 2007 | } |
| 2008 | } |
| 2009 | if ( !empty( $file ) ) { |
| 2010 | $row = $wpdb->get_row( $wpdb->prepare( "SELECT originType FROM $table WHERE mediaUrl = %s", $file ) ); |
| 2011 | if ( !empty( $row ) ) { |
| 2012 | $origin = $row->originType === 'MEDIA LIBRARY' ? 'Media Library' : 'content'; |
| 2013 | $this->log( "� |
| 2014 | File {$file} used by {$origin}" ); |
| 2015 | return $row->originType; |
| 2016 | } |
| 2017 | } |
| 2018 | return false; |
| 2019 | } |
| 2020 | |
| 2021 | function get_full_upload_path( $relative_path ) { |
| 2022 | $wp_upload_dir = wp_upload_dir(); |
| 2023 | $full_path = trailingslashit( $wp_upload_dir['basedir'] ) . $relative_path; |
| 2024 | return $full_path; |
| 2025 | } |
| 2026 | |
| 2027 | function get_paths_from_attachment( $attachmentId ) { |
| 2028 | $paths = array(); |
| 2029 | $fullpath = get_attached_file( $attachmentId ); |
| 2030 | if ( empty( $fullpath ) ) { |
| 2031 | $this->log( 'Could not find attached file for Media ID ' . $attachmentId ); |
| 2032 | return array(); |
| 2033 | } |
| 2034 | $mainfile = $this->clean_uploaded_filename( $fullpath ); |
| 2035 | array_push( $paths, $mainfile ); |
| 2036 | $baseUp = pathinfo( $mainfile ); |
| 2037 | $filespath = trailingslashit( $this->upload_path ) . trailingslashit( $baseUp['dirname'] ); |
| 2038 | $meta = wp_get_attachment_metadata( $attachmentId ); |
| 2039 | if ( isset( $meta['original_image'] ) ) { |
| 2040 | $original_image = $this->clean_uploaded_filename( $filespath . $meta['original_image'] ); |
| 2041 | array_push( $paths, $original_image ); |
| 2042 | } |
| 2043 | $isImage = isset( $meta, $meta['width'], $meta['height'] ); |
| 2044 | $sizes = $this->get_image_sizes(); |
| 2045 | if ( $isImage && isset( $meta['sizes'] ) ) { |
| 2046 | foreach ( $meta['sizes'] as $name => $attr ) { |
| 2047 | if ( isset( $attr['file'] ) ) { |
| 2048 | $file = $this->clean_uploaded_filename( $filespath . $attr['file'] ); |
| 2049 | array_push( $paths, $file ); |
| 2050 | } |
| 2051 | } |
| 2052 | } |
| 2053 | return $paths; |
| 2054 | } |
| 2055 | |
| 2056 | function is_media_ignored( $attachmentId ) { |
| 2057 | global $wpdb; |
| 2058 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 2059 | $issue = $wpdb->get_row( $wpdb->prepare( "SELECT * FROM $table_name WHERE postId = %d", $attachmentId ), OBJECT ); |
| 2060 | //error_log( $attachmentId ); |
| 2061 | //error_log( print_r( $issue, 1 ) ); |
| 2062 | if ( $issue && $issue->ignored ) |
| 2063 | return true; |
| 2064 | return false; |
| 2065 | } |
| 2066 | |
| 2067 | function check_media( $attachmentId, $checkOnly = false ) { |
| 2068 | |
| 2069 | // Is Media ID ignored, consider as used. |
| 2070 | if ( $this->is_media_ignored( $attachmentId ) ) { |
| 2071 | return true; |
| 2072 | } |
| 2073 | |
| 2074 | // Remove everything related to this media from the database. |
| 2075 | if ( !$checkOnly ) { |
| 2076 | $this->delete_attachment_related_data( $attachmentId ); |
| 2077 | } |
| 2078 | |
| 2079 | $size = 0; |
| 2080 | $countfiles = 0; |
| 2081 | $check_broken_media = !$this->check_content; |
| 2082 | $fullpath = get_attached_file( $attachmentId ); |
| 2083 | $is_broken = apply_filters( 'wpmc_is_file_broken', !file_exists( $fullpath ), $attachmentId ); |
| 2084 | |
| 2085 | // It's a broken-only scan |
| 2086 | if ( $check_broken_media && !$is_broken ) { |
| 2087 | $is_considered_used = apply_filters( 'wpmc_check_media', true, $attachmentId, false ); |
| 2088 | return $is_considered_used; |
| 2089 | } |
| 2090 | |
| 2091 | // Let's analyze the usage of each path (thumbnails included) for this Media ID. |
| 2092 | $issue = 'NO_CONTENT'; |
| 2093 | $paths = $this->get_paths_from_attachment( $attachmentId ); |
| 2094 | foreach ( $paths as $path ) { |
| 2095 | |
| 2096 | // If it's found in the content, we stop the scan right away |
| 2097 | if ( $this->check_content && $this->reference_exists( $path, $attachmentId ) ) { |
| 2098 | $is_considered_used = apply_filters( 'wpmc_check_media', true, $attachmentId, false ); |
| 2099 | if ( $is_considered_used ) { |
| 2100 | return true; |
| 2101 | } |
| 2102 | } |
| 2103 | |
| 2104 | // Let's count the size of the files for later, in case it's unused |
| 2105 | $filepath = trailingslashit( $this->upload_path ) . $path; |
| 2106 | if ( file_exists( $filepath ) ) |
| 2107 | $size += filesize( $filepath ); |
| 2108 | $countfiles++; |
| 2109 | } |
| 2110 | |
| 2111 | // This Media ID seems not in used (or broken) |
| 2112 | // Let's double-check through the filter (overridable by users) |
| 2113 | $is_considered_used = apply_filters( 'wpmc_check_media', false, $attachmentId, $is_broken ); |
| 2114 | if ( !$is_considered_used ) { |
| 2115 | if ( $is_broken ) { |
| 2116 | $this->log( "🚫 File {$fullpath} does not exist." ); |
| 2117 | $issue = 'ORPHAN_MEDIA'; |
| 2118 | } |
| 2119 | if ( !$checkOnly ) { |
| 2120 | global $wpdb; |
| 2121 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 2122 | $mainfile = $this->clean_uploaded_filename( $fullpath ); |
| 2123 | $wpdb->insert( $table_name, |
| 2124 | array( |
| 2125 | 'time' => current_time('mysql'), |
| 2126 | 'type' => 1, |
| 2127 | 'size' => $size, |
| 2128 | 'path' => $mainfile . ( $countfiles > 0 ? ( " (+ " . $countfiles . " files)" ) : "" ), |
| 2129 | 'postId' => $attachmentId, |
| 2130 | 'issue' => $issue |
| 2131 | ) |
| 2132 | ); |
| 2133 | } |
| 2134 | } |
| 2135 | return $is_considered_used; |
| 2136 | } |
| 2137 | |
| 2138 | // Delete all issues |
| 2139 | function reset_issues( $includingIgnored = false ) { |
| 2140 | global $wpdb; |
| 2141 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 2142 | if ( $includingIgnored ) { |
| 2143 | $wpdb->query( "DELETE FROM $table_name WHERE deleted = 0" ); |
| 2144 | } |
| 2145 | else { |
| 2146 | $wpdb->query( "DELETE FROM $table_name WHERE ignored = 0 AND deleted = 0" ); |
| 2147 | } |
| 2148 | if ( file_exists( WPMC_PATH . '/logs/media-cleaner.log' ) ) { |
| 2149 | file_put_contents( WPMC_PATH . '/logs/media-cleaner.log', '' ); |
| 2150 | } |
| 2151 | } |
| 2152 | |
| 2153 | function is_image_extension( $ext ) { |
| 2154 | $ext = strtolower( $ext ); |
| 2155 | $valid = apply_filters( 'wpmc_valid_image_extensions', array( 'jpg', 'jpeg', 'png', 'gif', 'bmp', 'tiff', 'ico', 'webp', 'avif' ) ); |
| 2156 | |
| 2157 | return in_array( $ext, $valid ); |
| 2158 | |
| 2159 | } |
| 2160 | |
| 2161 | |
| 2162 | function reset_references() { |
| 2163 | global $wpdb; |
| 2164 | $table_name = $wpdb->prefix . "mclean_refs"; |
| 2165 | $wpdb->query("TRUNCATE $table_name"); |
| 2166 | $this->reset_cached_references(); |
| 2167 | } |
| 2168 | |
| 2169 | function get_issue_for_postId( $postId ) { |
| 2170 | global $wpdb; |
| 2171 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 2172 | $issue = $wpdb->get_row( $wpdb->prepare( "SELECT * FROM $table_name WHERE postId = %d", $postId ), OBJECT ); |
| 2173 | return $issue; |
| 2174 | } |
| 2175 | |
| 2176 | function echo_issue( $issue ) { |
| 2177 | if ( $issue == 'NO_CONTENT' ) { |
| 2178 | _e( "Not found in content", 'media-cleaner' ); |
| 2179 | } |
| 2180 | else if ( $issue == 'ORPHAN_FILE' ) { |
| 2181 | _e( "Not in Library", 'media-cleaner' ); |
| 2182 | } |
| 2183 | else if ( $issue == 'ORPHAN_RETINA' ) { |
| 2184 | _e( "Orphan Retina", 'media-cleaner' ); |
| 2185 | } |
| 2186 | else if ( $issue == 'ORPHAN_WEBP' ) { |
| 2187 | _e( "Orphan WebP", 'media-cleaner' ); |
| 2188 | } |
| 2189 | else if ( $issue == 'ORPHAN_MEDIA' ) { |
| 2190 | _e( "No attached file", 'media-cleaner' ); |
| 2191 | } |
| 2192 | else { |
| 2193 | echo $issue; |
| 2194 | } |
| 2195 | } |
| 2196 | |
| 2197 | function get_uploads_directory_hierarchy() { |
| 2198 | $uploads_dir = wp_upload_dir(); |
| 2199 | $base_dir = wp_normalize_path( $uploads_dir['basedir'] ); |
| 2200 | $root = '/' . wp_basename( $base_dir ); |
| 2201 | $directories = array(); |
| 2202 | |
| 2203 | // Get all subdirectories of the base directory |
| 2204 | $dir_iterator = new RecursiveDirectoryIterator( $base_dir, FilesystemIterator::KEY_AS_PATHNAME | FilesystemIterator::CURRENT_AS_FILEINFO | FilesystemIterator::SKIP_DOTS ); |
| 2205 | $iterator = new RecursiveIteratorIterator( $dir_iterator, RecursiveIteratorIterator::SELF_FIRST ); |
| 2206 | |
| 2207 | foreach ( $iterator as $file ) { |
| 2208 | if ( $file->isDir() ) { |
| 2209 | // Normalize path for consistency |
| 2210 | $file_path = wp_normalize_path( $file->getPathname() ); |
| 2211 | // Remove base_dir from path |
| 2212 | $directory = str_replace( $base_dir, '', $file_path ); |
| 2213 | if ( $directory ) { |
| 2214 | $directories[] = $root . $directory; |
| 2215 | } |
| 2216 | } |
| 2217 | } |
| 2218 | |
| 2219 | // Return the hierarchy as a JSON file |
| 2220 | return json_encode( $directories ); |
| 2221 | } |
| 2222 | |
| 2223 | /** |
| 2224 | * |
| 2225 | * Roles & Access Rights |
| 2226 | * |
| 2227 | */ |
| 2228 | public function can_access_settings() { |
| 2229 | return apply_filters( 'wpmc_allow_setup', current_user_can( 'manage_options' ) ); |
| 2230 | } |
| 2231 | |
| 2232 | public function can_access_features() { |
| 2233 | return apply_filters( 'wpmc_allow_usage', current_user_can( 'administrator' ) ); |
| 2234 | } |
| 2235 | |
| 2236 | #region Options |
| 2237 | |
| 2238 | function list_options() { |
| 2239 | return array( |
| 2240 | 'method' => 'media', |
| 2241 | 'content' => true, |
| 2242 | 'filesystem_content' => false, |
| 2243 | 'media_library' => true, |
| 2244 | 'live_content' => false, |
| 2245 | 'debuglogs' => false, |
| 2246 | 'images_only' => false, |
| 2247 | 'attach_is_use' => false, |
| 2248 | 'thumbnails_only' => false, |
| 2249 | 'dirs_filter' => '', |
| 2250 | 'files_filter' => '', |
| 2251 | 'hide_thumbnails' => false, |
| 2252 | 'hide_warning' => false, |
| 2253 | 'skip_trash' => false, |
| 2254 | 'medias_buffer' => 100, |
| 2255 | 'posts_buffer' => 5, |
| 2256 | 'analysis_buffer' => 100, |
| 2257 | 'file_op_buffer' => 20, |
| 2258 | 'delay' => 100, |
| 2259 | 'shortcodes_disabled' => false, |
| 2260 | 'use_cached_references' => true, |
| 2261 | 'output_buffer_cleaning_disabled' => false, |
| 2262 | 'php_error_logs' => false, |
| 2263 | 'posts_per_page' => 10, |
| 2264 | 'clean_uninstall' => false, |
| 2265 | 'repair_mode' => false, |
| 2266 | 'expert_mode' => false, |
| 2267 | 'logs_path' => null, |
| 2268 | ); |
| 2269 | } |
| 2270 | |
| 2271 | function reset_options() { |
| 2272 | delete_option( $this->option_name ); |
| 2273 | } |
| 2274 | |
| 2275 | function get_option( $option ) { |
| 2276 | $options = $this->get_all_options(); |
| 2277 | return $options[$option]; |
| 2278 | } |
| 2279 | |
| 2280 | function get_all_options() { |
| 2281 | $options = get_option( $this->option_name, null ); |
| 2282 | $options = $this->check_options( $options ); |
| 2283 | return $options; |
| 2284 | } |
| 2285 | |
| 2286 | // Let's work on this function if we need it. |
| 2287 | // Right now, it looks like the options are all updated at the same time. |
| 2288 | |
| 2289 | // function update_option( $option, $value ) { |
| 2290 | // if ( !array_key_exists( $name, $options ) ) { |
| 2291 | // return new WP_REST_Response([ 'success' => false, 'message' => 'This option does not exist.' ], 200 ); |
| 2292 | // } |
| 2293 | // $value = is_bool( $params['value'] ) ? ( $params['value'] ? '1' : '' ) : $params['value']; |
| 2294 | // } |
| 2295 | |
| 2296 | function update_options( $options ) { |
| 2297 | if ( !update_option( $this->option_name, $options, false ) ) { |
| 2298 | return false; |
| 2299 | } |
| 2300 | $options = $this->sanitize_options(); |
| 2301 | return $options; |
| 2302 | } |
| 2303 | |
| 2304 | // Upgrade from the old way of storing options to the new way. |
| 2305 | function check_options( $options = [] ) { |
| 2306 | $plugin_options = $this->list_options(); |
| 2307 | $options = empty( $options ) ? [] : $options; |
| 2308 | $hasChanges = false; |
| 2309 | foreach ( $plugin_options as $option => $default ) { |
| 2310 | // The option already exists |
| 2311 | if ( isset( $options[$option] ) ) { |
| 2312 | continue; |
| 2313 | } |
| 2314 | // The option does not exist, so we need to add it. |
| 2315 | // Let's use the old value if any, or the default value. |
| 2316 | $options[$option] = get_option( 'wpmc_' . $option, $default ); |
| 2317 | delete_option( 'wpmc_' . $option ); |
| 2318 | $hasChanges = true; |
| 2319 | } |
| 2320 | if ( $hasChanges ) { |
| 2321 | update_option( $this->option_name , $options ); |
| 2322 | } |
| 2323 | |
| 2324 | // Dynamically added options |
| 2325 | //TODO: we should have a rest route to fetch this instead of using the options directly. This is temporary. |
| 2326 | $options['scan_progress'] = get_transient( $this->progress_key ); |
| 2327 | |
| 2328 | return $options; |
| 2329 | } |
| 2330 | |
| 2331 | // Validate and keep the options clean and logical. |
| 2332 | function sanitize_options() { |
| 2333 | $options = $this->get_all_options(); |
| 2334 | $medias = $options['medias_buffer']; |
| 2335 | $posts = $options['posts_buffer']; |
| 2336 | $analysis = $options['analysis_buffer']; |
| 2337 | $fileOp = $options['file_op_buffer']; |
| 2338 | $delay = $options['delay']; |
| 2339 | $hasChanges = false; |
| 2340 | if ( $medias === '' ) { |
| 2341 | $options['medias_buffer'] = 100; |
| 2342 | $hasChanges = true; |
| 2343 | } |
| 2344 | if ( $posts === '' ) { |
| 2345 | $options['posts_buffer'] = 5; |
| 2346 | $hasChanges = true; |
| 2347 | } |
| 2348 | if ( $analysis === '' ) { |
| 2349 | $options['analysis_buffer'] = 100; |
| 2350 | $hasChanges = true; |
| 2351 | } |
| 2352 | if ( $fileOp === '' ) { |
| 2353 | $options['file_op_buffer'] = 20; |
| 2354 | $hasChanges = true; |
| 2355 | } |
| 2356 | if ( $delay === '' ) { |
| 2357 | $options['delay'] = 100; |
| 2358 | $hasChanges = true; |
| 2359 | } |
| 2360 | if ( $hasChanges ) { |
| 2361 | update_option( $this->option_name, $options, false ); |
| 2362 | } |
| 2363 | return $options; |
| 2364 | } |
| 2365 | |
| 2366 | #endregion |
| 2367 | } |
| 2368 | |
| 2369 | // Check the DB. If does not exist, let's create it. |
| 2370 | function wpmc_check_database() { |
| 2371 | wpmc_create_database(); |
| 2372 | } |
| 2373 | |
| 2374 | function wpmc_create_database() { |
| 2375 | global $wpdb; |
| 2376 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 2377 | $charset_collate = $wpdb->get_charset_collate(); |
| 2378 | $sql = "CREATE TABLE $table_name ( |
| 2379 | id BIGINT(20) NOT NULL AUTO_INCREMENT, |
| 2380 | time DATETIME DEFAULT '0000-00-00 00:00:00' NOT NULL, |
| 2381 | type TINYINT(1) NOT NULL, |
| 2382 | postId BIGINT(20) NULL, |
| 2383 | path TINYTEXT NULL, |
| 2384 | size INT(9) NULL, |
| 2385 | ignored TINYINT(1) NOT NULL DEFAULT 0, |
| 2386 | deleted TINYINT(1) NOT NULL DEFAULT 0, |
| 2387 | issue TINYTEXT NOT NULL, |
| 2388 | parentId BIGINT(20) NULL, |
| 2389 | PRIMARY KEY (id), |
| 2390 | KEY PostIdIndex (postId), |
| 2391 | KEY IgnoredIndex (ignored) |
| 2392 | ) " . $charset_collate . ";" ; |
| 2393 | require_once( ABSPATH . 'wp-admin/includes/upgrade.php' ); |
| 2394 | dbDelta( $sql ); |
| 2395 | |
| 2396 | $table_name = $wpdb->prefix . "mclean_refs"; |
| 2397 | $charset_collate = $wpdb->get_charset_collate(); |
| 2398 | // This key doesn't work on too many installs because of the 'Specified key was too long' issue |
| 2399 | // KEY mediaLookUp (mediaId, mediaUrl) |
| 2400 | $sql = "CREATE TABLE $table_name ( |
| 2401 | id BIGINT(20) NOT NULL AUTO_INCREMENT, |
| 2402 | mediaId BIGINT(20) NULL, |
| 2403 | mediaUrl TINYTEXT NULL, |
| 2404 | originType TINYTEXT NOT NULL, |
| 2405 | parentId BIGINT(20) NULL, |
| 2406 | PRIMARY KEY (id), |
| 2407 | KEY mediaId_index (mediaId) |
| 2408 | ) " . $charset_collate . ";"; |
| 2409 | require_once( ABSPATH . 'wp-admin/includes/upgrade.php' ); |
| 2410 | dbDelta( $sql ); |
| 2411 | |
| 2412 | // Create cache table for cached IDs and URLs |
| 2413 | $table_name = $wpdb->prefix . "mclean_cache"; |
| 2414 | $sql = "CREATE TABLE $table_name ( |
| 2415 | id BIGINT(20) NOT NULL AUTO_INCREMENT, |
| 2416 | cache_key VARCHAR(50) NOT NULL, |
| 2417 | cache_value VARCHAR(255) NOT NULL, |
| 2418 | cache_type VARCHAR(20) NOT NULL, |
| 2419 | PRIMARY KEY (id), |
| 2420 | UNIQUE KEY cache_lookup (cache_key, cache_value, cache_type) |
| 2421 | ) " . $charset_collate . ";"; |
| 2422 | dbDelta( $sql ); |
| 2423 | } |
| 2424 | |
| 2425 | function wpmc_remove_database() { |
| 2426 | global $wpdb; |
| 2427 | $table_name1 = $wpdb->prefix . "mclean_scan"; |
| 2428 | $table_name2 = $wpdb->prefix . "mclean_refs"; |
| 2429 | $table_name3 = $wpdb->prefix . "wpmcleaner"; |
| 2430 | $table_name4 = $wpdb->prefix . "mclean_cache"; |
| 2431 | $sql = "DROP TABLE IF EXISTS $table_name1, $table_name2, $table_name3, $table_name4;"; |
| 2432 | $wpdb->query( $sql ); |
| 2433 | } |
| 2434 | |
| 2435 | #region Install / Uninstall |
| 2436 | |
| 2437 | /* |
| 2438 | INSTALL / UNINSTALL |
| 2439 | */ |
| 2440 | |
| 2441 | function wpmc_init( $mainfile ) { |
| 2442 | //register_activation_hook( $mainfile, 'wpmc_install' ); |
| 2443 | //register_deactivation_hook( $mainfile, 'wpmc_uninstall' ); |
| 2444 | register_uninstall_hook( $mainfile, 'wpmc_uninstall' ); |
| 2445 | } |
| 2446 | |
| 2447 | function wpmc_install() { |
| 2448 | wpmc_create_database(); |
| 2449 | } |
| 2450 | |
| 2451 | function wpmc_reset () { |
| 2452 | wpmc_remove_database(); |
| 2453 | wpmc_create_database(); |
| 2454 | } |
| 2455 | |
| 2456 | function wpmc_remove_options() { |
| 2457 | global $wpdb; |
| 2458 | $options = $wpdb->get_results( "SELECT option_name FROM $wpdb->options WHERE option_name LIKE 'wpmc_%'" ); |
| 2459 | foreach( $options as $option ) { |
| 2460 | delete_option( $option->option_name ); |
| 2461 | } |
| 2462 | } |
| 2463 | |
| 2464 | function wpmc_uninstall () { |
| 2465 | $options = get_option( 'wpmc_options', [] ); |
| 2466 | $cleanUninstall = $options['clean_uninstall']; |
| 2467 | if ($cleanUninstall) { |
| 2468 | wpmc_remove_options(); |
| 2469 | wpmc_remove_database(); |
| 2470 | } |
| 2471 | } |
| 2472 | |
| 2473 | #endregion |