parsers
1 month ago
admin.php
1 month ago
buffers.php
2 weeks ago
core.php
2 weeks ago
engine.php
1 month ago
init.php
9 months ago
mcp.php
1 month ago
parsers.php
1 month ago
rest.php
2 weeks ago
runs.php
1 month ago
support.php
1 month ago
ui.php
1 month ago
core.php
3684 lines
| 1 | <?php |
| 2 | |
| 3 | if ( !class_exists( 'Meow_WPMC_Transient_Exception', false ) ) { |
| 4 | class Meow_WPMC_Transient_Exception extends RuntimeException { |
| 5 | private $retry_after_ms; |
| 6 | |
| 7 | public function __construct( $message, $retry_after_ms = 2000, $previous = null ) { |
| 8 | parent::__construct( $message, 0, $previous ); |
| 9 | $this->retry_after_ms = max( 250, min( 60000, (int) $retry_after_ms ) ); |
| 10 | } |
| 11 | |
| 12 | public function get_retry_after_ms() { |
| 13 | return $this->retry_after_ms; |
| 14 | } |
| 15 | } |
| 16 | } |
| 17 | |
| 18 | class Meow_WPMC_Core { |
| 19 | |
| 20 | // A file that exists but cannot be safely fingerprinted |
| 21 | const FINGERPRINT_UNSAFE = '@unsafe'; |
| 22 | |
| 23 | |
| 24 | public $admin = null; |
| 25 | public $is_rest = false; |
| 26 | public $is_cli = false; |
| 27 | public $is_pro = false; |
| 28 | public $engine = null; |
| 29 | public $runs = null; |
| 30 | public $catch_timeout = true; // This will halt the plugin before reaching the PHP timeout. |
| 31 | 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|woff2|woff|ttf|otf"; |
| 32 | public $current_method = 'media'; |
| 33 | public $servername = null; // meowapps.com (site URL without http/https) |
| 34 | public $site_url = null; // https://meowapps.com |
| 35 | public $upload_path = null; // /www/wp-content/uploads (path to uploads) |
| 36 | public $upload_url = null; // wp-content/uploads (uploads without domain) |
| 37 | private $option_name = 'wpmc_options'; |
| 38 | private $nonce = null; // Nonce for the REST API |
| 39 | |
| 40 | private $regex_file = '/[A-Za-z0-9-_,.\(\)\s]+[.]{1}(MIMETYPES)/'; |
| 41 | |
| 42 | private $refcache = array(); |
| 43 | private $progress_key = 'wpmc_progress'; |
| 44 | private $run_id = 0; |
| 45 | private $run_config = array(); |
| 46 | |
| 47 | private $debug_logs = null; |
| 48 | private $multilingual = false; |
| 49 | private $languages = array(); |
| 50 | private $shortcode_analysis = false; |
| 51 | private $translated_ids_cache = array(); |
| 52 | private $url_id_cache = array(); |
| 53 | private $trash_migration_error = null; |
| 54 | private $parser_query_guard_active = false; |
| 55 | private $parser_query_guard_label = ''; |
| 56 | private $parser_query_guard_warnings = array(); |
| 57 | private $request_start_time = 0; |
| 58 | private $request_time_budget = null; |
| 59 | |
| 60 | public function get_shortcode_analysis() { |
| 61 | return $this->shortcode_analysis; |
| 62 | } |
| 63 | |
| 64 | public function is_debug() { |
| 65 | return $this->debug_logs; |
| 66 | } |
| 67 | |
| 68 | public function __construct() { |
| 69 | $this->request_start_time = isset( $_SERVER['REQUEST_TIME_FLOAT'] ) |
| 70 | ? (float) $_SERVER['REQUEST_TIME_FLOAT'] |
| 71 | : microtime( true ); |
| 72 | add_action( 'plugins_loaded', array( $this, 'plugins_loaded' ) ); |
| 73 | add_action( 'delete_attachment', array( $this, 'delete_attachment_related_data' ), 10, 1 ); |
| 74 | add_action( 'trashed_post', array( $this, 'delete_attachment_related_data' ), 10, 1 ); |
| 75 | } |
| 76 | |
| 77 | function plugins_loaded() { |
| 78 | |
| 79 | |
| 80 | if ( is_admin() ) { |
| 81 | new Meow_WPMC_UI( $this ); |
| 82 | } |
| 83 | |
| 84 | // Admin |
| 85 | $this->admin = new Meow_WPMC_Admin( $this ); |
| 86 | |
| 87 | // Advanced core |
| 88 | if ( class_exists( 'MeowPro_WPMC_Core' ) ) { |
| 89 | new MeowPro_WPMC_Core( $this ); |
| 90 | } |
| 91 | |
| 92 | // Only initialize variables if we are on a relevant screen |
| 93 | $pages = [ 'wpmc_dashboard', 'wpmc_settings' ]; |
| 94 | $page = isset( $_GET["page"] ) ? sanitize_text_field( $_GET["page"] ) : null; |
| 95 | $is_wpmc_screen = in_array( $page, $pages ); |
| 96 | |
| 97 | // Check if this is a REST request specifically for Media Cleaner |
| 98 | $is_wpmc_rest = false; |
| 99 | $is_mcp_rest = false; |
| 100 | if ( MeowKit_WPMC_Helpers::is_rest() ) { |
| 101 | $request_uri = isset( $_SERVER['REQUEST_URI'] ) ? $_SERVER['REQUEST_URI'] : ''; |
| 102 | $is_wpmc_rest = strpos( $request_uri, '/media-cleaner/v1' ) !== false; |
| 103 | // MCP tools are served by AI Engine, on its own route. The scan engine |
| 104 | // still has to be loaded for them to run. |
| 105 | $is_mcp_rest = strpos( $request_uri, '/mcp/v1' ) !== false; |
| 106 | } |
| 107 | |
| 108 | // Variables |
| 109 | $this->site_url = get_site_url(); |
| 110 | $this->multilingual = $this->is_multilingual(); |
| 111 | $this->languages = $this->get_languages(); |
| 112 | $this->current_method = $this->get_option( 'method' ); |
| 113 | $this->regex_file = str_replace( "MIMETYPES", $this->types, $this->regex_file ); |
| 114 | $this->servername = (string) wp_parse_url( $this->site_url, PHP_URL_HOST ); |
| 115 | $uploaddir = wp_upload_dir(); |
| 116 | $this->upload_path = empty( $uploaddir['error'] ) ? untrailingslashit( wp_normalize_path( $uploaddir['basedir'] ) ) : ''; |
| 117 | $this->upload_url = empty( $uploaddir['error'] ) ? untrailingslashit( $uploaddir['baseurl'] ) : ''; |
| 118 | $this->debug_logs = $this->get_option( 'debuglogs' ); |
| 119 | $this->is_rest = $is_wpmc_rest; |
| 120 | $this->is_cli = defined( 'WP_CLI' ) && WP_CLI; |
| 121 | $this->shortcode_analysis = !$this->get_option( 'shortcodes_disabled' ); |
| 122 | |
| 123 | global $wpmc; |
| 124 | $wpmc = $this; |
| 125 | |
| 126 | // MCP tools, served through AI Engine. |
| 127 | if ( class_exists( 'Meow_MWAI_Core' ) || isset( $GLOBALS['mwai'] ) ) { |
| 128 | new Meow_WPMC_MCP( $this ); |
| 129 | } |
| 130 | |
| 131 | $shouldLoad = ( defined( 'WP_CLI' ) && WP_CLI ) || $is_wpmc_screen || $is_wpmc_rest || $is_mcp_rest; |
| 132 | |
| 133 | if ( ! $shouldLoad ) { |
| 134 | return; |
| 135 | } |
| 136 | |
| 137 | $this->runs = new Meow_WPMC_Runs( $this ); |
| 138 | if ( $this->is_rest || $this->is_cli || $is_mcp_rest ) $this->runs->maybe_upgrade(); |
| 139 | |
| 140 | // Language |
| 141 | load_plugin_textdomain( WPMC_DOMAIN, false, basename( WPMC_PATH ) . '/languages' ); |
| 142 | |
| 143 | // Install hooks and engine only if they might be used |
| 144 | if ( is_admin() || $this->is_rest || $this->is_cli || $is_mcp_rest ) { |
| 145 | add_action( 'wpmc_initialize_parsers', array( $this, 'initialize_parsers' ), 10, 0 ); |
| 146 | add_filter( 'wp_unique_filename', array( $this, 'wp_unique_filename' ), 10, 3 ); |
| 147 | $this->engine = new Meow_WPMC_Engine( $this, $this->admin ); |
| 148 | } |
| 149 | |
| 150 | // Only for REST |
| 151 | if ( $this->is_rest ) { |
| 152 | new Meow_WPMC_Rest( $this, $this->admin ); |
| 153 | } |
| 154 | |
| 155 | |
| 156 | |
| 157 | |
| 158 | } |
| 159 | |
| 160 | public function set_run_context( $run_id ) { |
| 161 | if ( !$this->runs ) { |
| 162 | return new WP_Error( 'wpmc_runs_unavailable', __( 'The Media Cleaner run manager is unavailable.', 'media-cleaner' ) ); |
| 163 | } |
| 164 | $run = $this->runs->assert_writable( (int) $run_id ); |
| 165 | if ( is_wp_error( $run ) ) { |
| 166 | return $run; |
| 167 | } |
| 168 | if ( $run->status === 'paused' ) { |
| 169 | $run = $this->runs->resume( (int) $run_id ); |
| 170 | if ( is_wp_error( $run ) ) return $run; |
| 171 | } |
| 172 | $this->run_id = (int) $run->id; |
| 173 | $this->current_method = $run->method; |
| 174 | $config = json_decode( (string) $run->config, true ); |
| 175 | $this->run_config = is_array( $config ) ? $config : array(); |
| 176 | $this->shortcode_analysis = !$this->get_option( 'shortcodes_disabled' ); |
| 177 | return $run; |
| 178 | } |
| 179 | |
| 180 | public function clear_run_context() { |
| 181 | $this->run_id = 0; |
| 182 | $this->run_config = array(); |
| 183 | $this->current_method = $this->get_option( 'method' ); |
| 184 | $this->shortcode_analysis = !$this->get_option( 'shortcodes_disabled' ); |
| 185 | } |
| 186 | |
| 187 | public function safe_do_action( $hook_name, ...$args ) { |
| 188 | global $wp_filter, $wp_actions, $wp_current_filter; |
| 189 | if ( empty( $wp_filter[ $hook_name ] ) || !( $wp_filter[ $hook_name ] instanceof WP_Hook ) ) { |
| 190 | return; |
| 191 | } |
| 192 | $wp_actions[ $hook_name ] = isset( $wp_actions[ $hook_name ] ) ? $wp_actions[ $hook_name ] + 1 : 1; |
| 193 | $wp_current_filter[] = $hook_name; |
| 194 | try { |
| 195 | foreach ( $wp_filter[ $hook_name ]->callbacks as $callbacks ) { |
| 196 | foreach ( $callbacks as $callback ) { |
| 197 | $function = $callback['function']; |
| 198 | $callback_name = $this->callback_name( $function ); |
| 199 | $journal = null; |
| 200 | if ( $hook_name === 'wpmc_scan_once' && $this->run_id > 0 && $this->runs ) { |
| 201 | $journal = $this->runs->get_work( $this->run_id, 'scanOnce', $callback_name ); |
| 202 | if ( $journal && $journal->status === 'complete' ) continue; |
| 203 | if ( !$journal ) { |
| 204 | if ( !$this->runs->enqueue_work( $this->run_id, 'scanOnce', 'parser', $callback_name ) ) { |
| 205 | throw new RuntimeException( sprintf( __( 'Media Cleaner could not journal parser %s.', 'media-cleaner' ), $callback_name ) ); |
| 206 | } |
| 207 | $journal = $this->runs->get_work( $this->run_id, 'scanOnce', $callback_name ); |
| 208 | } |
| 209 | if ( !$journal || !$this->runs->update_work( $journal->id, 'running', $journal->cursor_value ) ) { |
| 210 | throw new RuntimeException( sprintf( __( 'Media Cleaner could not start parser %s safely.', 'media-cleaner' ), $callback_name ) ); |
| 211 | } |
| 212 | } |
| 213 | $accepted_args = max( 0, (int) $callback['accepted_args'] ); |
| 214 | $call_args = $accepted_args === 0 ? array() : array_slice( $args, 0, $accepted_args ); |
| 215 | $started = microtime( true ); |
| 216 | $guard_was_active = $this->parser_query_guard_active; |
| 217 | $previous_guard_label = $this->parser_query_guard_label; |
| 218 | $this->parser_query_guard_active = true; |
| 219 | $this->parser_query_guard_label = $callback_name; |
| 220 | if ( !$guard_was_active ) add_filter( 'query', array( $this, 'guard_parser_query' ), PHP_INT_MAX ); |
| 221 | try { |
| 222 | $this->timeout_check(); |
| 223 | call_user_func_array( $function, $call_args ); |
| 224 | if ( $journal ) { |
| 225 | $this->write_references(); |
| 226 | if ( !$this->runs->update_work( $journal->id, 'complete', $journal->cursor_value ) ) { |
| 227 | throw new RuntimeException( sprintf( __( 'Media Cleaner could not complete parser %s safely.', 'media-cleaner' ), $callback_name ) ); |
| 228 | } |
| 229 | } |
| 230 | } |
| 231 | catch ( Meow_WPMC_Transient_Exception $e ) { |
| 232 | if ( $journal ) $this->runs->update_work( $journal->id, 'pending', $journal->cursor_value, $e ); |
| 233 | throw $e; |
| 234 | } |
| 235 | catch ( Throwable $e ) { |
| 236 | if ( $journal ) $this->runs->update_work( $journal->id, 'failed', $journal->cursor_value, $e ); |
| 237 | throw new RuntimeException( sprintf( __( '%1$s failed in %2$s: %3$s', 'media-cleaner' ), $callback_name, $hook_name, $e->getMessage() ), 0, $e ); |
| 238 | } |
| 239 | finally { |
| 240 | $this->parser_query_guard_active = $guard_was_active; |
| 241 | $this->parser_query_guard_label = $previous_guard_label; |
| 242 | if ( !$guard_was_active ) remove_filter( 'query', array( $this, 'guard_parser_query' ), PHP_INT_MAX ); |
| 243 | } |
| 244 | $elapsed = microtime( true ) - $started; |
| 245 | $memory_limit = $this->parse_ini_bytes( ini_get( 'memory_limit' ) ); |
| 246 | if ( $memory_limit > 0 && memory_get_usage( true ) > $memory_limit * 0.85 ) { |
| 247 | throw new Meow_WPMC_Transient_Exception( sprintf( __( '%1$s reached the safe parser memory budget in %2$s after %3$.1f seconds.', 'media-cleaner' ), $callback_name, $hook_name, $elapsed ), 5000 ); |
| 248 | } |
| 249 | } |
| 250 | } |
| 251 | } |
| 252 | finally { |
| 253 | array_pop( $wp_current_filter ); |
| 254 | } |
| 255 | } |
| 256 | |
| 257 | public function guard_parser_query( $query ) { |
| 258 | if ( !$this->parser_query_guard_active || !is_string( $query ) ) return $query; |
| 259 | $normalized = preg_replace( '/\s+/', ' ', trim( $query ) ); |
| 260 | if ( !preg_match( '/^(SELECT|WITH)\b/i', $normalized ) ) return $query; |
| 261 | if ( preg_match( '/\bLIMIT\s+\d+/i', $normalized ) ) return $query; |
| 262 | if ( preg_match( '/^SELECT\s+(?:DISTINCT\s+)?(?:COUNT|SUM|MIN|MAX|AVG|EXISTS)\s*\(/i', $normalized ) ) return $query; |
| 263 | if ( preg_match( '/\b(?:ID|post_id|meta_id|term_id|term_taxonomy_id|option_id|option_name|user_id)\s*(?:=|IN\s*\()/i', $normalized ) ) return $query; |
| 264 | // Only Media Cleaner's own SQL is a Media Cleaner bug. WordPress and other plugins |
| 265 | // (WPML translates media while our parsers run) issue their own unbounded SELECTs |
| 266 | // during a scan, and blaming the parser for those is noise nobody can act on. |
| 267 | $origin = $this->plugin_query_origin(); |
| 268 | if ( $origin === null ) return $query; |
| 269 | $label = $this->parser_query_guard_label ?: 'A compatibility parser'; |
| 270 | $message = sprintf( |
| 271 | __( '%s attempted an unbounded database query. Its Media Cleaner parser should use pagination.', 'media-cleaner' ), |
| 272 | $label |
| 273 | ); |
| 274 | // The query is rarely written in the parser itself: it is usually a helper it calls. |
| 275 | // Log where it came from, otherwise the warning cannot be acted upon. |
| 276 | $warning_key = $label . '|' . $origin; |
| 277 | if ( empty( $this->parser_query_guard_warnings[ $warning_key ] ) ) { |
| 278 | $this->parser_query_guard_warnings[ $warning_key ] = true; |
| 279 | $this->log( $message . " ({$origin}) " . substr( $normalized, 0, 500 ) ); |
| 280 | } |
| 281 | |
| 282 | // Strict mode is useful for parser development, but is unsafe for normal scans because |
| 283 | // WordPress and third-party APIs can legitimately issue SELECT queries without LIMIT. |
| 284 | if ( apply_filters( 'wpmc_strict_parser_query_guard', false, $normalized, $label ) ) { |
| 285 | throw new RuntimeException( $message ); |
| 286 | } |
| 287 | return $query; |
| 288 | } |
| 289 | |
| 290 | // Returns "file.php:line" when the running query was issued by Media Cleaner itself, |
| 291 | // null when it belongs to WordPress or another plugin. The backtrace is only taken for |
| 292 | // queries that already look unbounded, so it stays out of the normal scan path. |
| 293 | private function plugin_query_origin() { |
| 294 | $issuer = null; |
| 295 | foreach ( debug_backtrace( DEBUG_BACKTRACE_IGNORE_ARGS, 24 ) as $frame ) { |
| 296 | // A frame records where its function was called from, so the outermost $wpdb |
| 297 | // frame ($wpdb->get_var, get_results...) is the one holding the location of |
| 298 | // whoever asked for the query. Everything inside it, and everything above it |
| 299 | // up to this guard, is plumbing. |
| 300 | $class = isset( $frame['class'] ) ? $frame['class'] : ''; |
| 301 | if ( $class !== 'wpdb' && !is_subclass_of( $class, 'wpdb' ) ) continue; |
| 302 | if ( !empty( $frame['file'] ) ) $issuer = $frame; |
| 303 | } |
| 304 | if ( !$issuer ) return null; |
| 305 | $file = wp_normalize_path( $issuer['file'] ); |
| 306 | if ( strpos( $file, trailingslashit( wp_normalize_path( WPMC_PATH ) ) ) !== 0 ) return null; |
| 307 | return basename( $file ) . ':' . ( isset( $issuer['line'] ) ? (int) $issuer['line'] : 0 ); |
| 308 | } |
| 309 | |
| 310 | private function callback_name( $callback ) { |
| 311 | if ( is_string( $callback ) ) return $callback; |
| 312 | if ( is_array( $callback ) && count( $callback ) === 2 ) { |
| 313 | $owner = is_object( $callback[0] ) ? get_class( $callback[0] ) : $callback[0]; |
| 314 | return $owner . '::' . $callback[1]; |
| 315 | } |
| 316 | return $callback instanceof Closure ? 'closure' : 'unknown callback'; |
| 317 | } |
| 318 | |
| 319 | public function get_run_id( $for_write = false ) { |
| 320 | if ( $this->run_id > 0 ) { |
| 321 | return $this->run_id; |
| 322 | } |
| 323 | if ( $for_write ) { |
| 324 | return 0; |
| 325 | } |
| 326 | return $this->runs ? $this->runs->get_active_id() : max( 0, (int) get_option( Meow_WPMC_Runs::ACTIVE_RUN_OPTION, 0 ) ); |
| 327 | } |
| 328 | |
| 329 | public function get_nonce( $force = false ) { |
| 330 | if ( !$force && !is_user_logged_in() ) { |
| 331 | return null; |
| 332 | } |
| 333 | if ( isset( $this->nonce ) ) { |
| 334 | return $this->nonce; |
| 335 | } |
| 336 | |
| 337 | $this->nonce = wp_create_nonce( 'wp_rest' ); |
| 338 | return $this->nonce; |
| 339 | } |
| 340 | |
| 341 | function initialize_parsers() { |
| 342 | include_once( 'parsers.php' ); |
| 343 | new Meow_WPMC_Parsers(); |
| 344 | } |
| 345 | |
| 346 | function deepsleep( $seconds ) { |
| 347 | $start_time = time(); |
| 348 | while( true ) { |
| 349 | if ( ( time() - $start_time ) > $seconds ) { |
| 350 | return false; |
| 351 | } |
| 352 | get_post( array( 'posts_per_page' => 50 ) ); |
| 353 | } |
| 354 | } |
| 355 | |
| 356 | private $start_time; |
| 357 | private $time_elapsed = 0; |
| 358 | private $time_remaining = 0; |
| 359 | private $item_scan_avg_time = 0; |
| 360 | private $wordpress_init_time = 0.5; |
| 361 | private $max_execution_time; |
| 362 | private $items_checked = 0; |
| 363 | private $items_count = 0; |
| 364 | |
| 365 | function get_max_execution_time() { |
| 366 | if ( isset( $this->max_execution_time ) ) |
| 367 | return $this->max_execution_time; |
| 368 | |
| 369 | $this->max_execution_time = (int) ini_get( "max_execution_time" ); |
| 370 | // An unlimited PHP worker can still sit behind a 30 or 60 second proxy. |
| 371 | // Use a conservative fallback so every REST batch remains bounded. |
| 372 | if ( $this->max_execution_time === 0 ) |
| 373 | $this->max_execution_time = 30; |
| 374 | else if ( $this->max_execution_time < 5 ) |
| 375 | $this->max_execution_time = 5; |
| 376 | |
| 377 | return $this->max_execution_time; |
| 378 | } |
| 379 | |
| 380 | public function get_request_time_budget() { |
| 381 | if ( $this->request_time_budget !== null ) return $this->request_time_budget; |
| 382 | $execution_limit = $this->get_max_execution_time(); |
| 383 | $hard_budget = min( 30, $execution_limit ); |
| 384 | $hard_budget = (float) apply_filters( 'wpmc_request_time_budget', $hard_budget, $execution_limit ); |
| 385 | $hard_budget = max( 5, min( $hard_budget, max( 5, $execution_limit ) ) ); |
| 386 | $reserve = max( 2.5, min( 6, $hard_budget * 0.2 ) ); |
| 387 | $this->request_time_budget = max( 2.5, $hard_budget - $reserve ); |
| 388 | return $this->request_time_budget; |
| 389 | } |
| 390 | |
| 391 | public function parse_ini_bytes( $value ) { |
| 392 | $value = trim( (string) $value ); |
| 393 | if ( $value === '' ) { |
| 394 | return 0; |
| 395 | } |
| 396 | if ( $value === '-1' ) { |
| 397 | return -1; |
| 398 | } |
| 399 | $unit = strtolower( substr( $value, -1 ) ); |
| 400 | $number = (float) $value; |
| 401 | switch ( $unit ) { |
| 402 | case 'g': |
| 403 | $number *= 1024; |
| 404 | // Fall through. |
| 405 | case 'm': |
| 406 | $number *= 1024; |
| 407 | // Fall through. |
| 408 | case 'k': |
| 409 | $number *= 1024; |
| 410 | } |
| 411 | return (int) $number; |
| 412 | } |
| 413 | |
| 414 | public function parser_budget_check( $started_at, $label = 'Parser' ) { |
| 415 | $this->timeout_check(); |
| 416 | $budget = max( 3, min( 12, $this->get_request_time_budget() * 0.6 ) ); |
| 417 | if ( !$this->is_cli && microtime( true ) - (float) $started_at > $budget ) { |
| 418 | throw new Meow_WPMC_Transient_Exception( sprintf( __( '%1$s reached its %2$.1f-second scan budget.', 'media-cleaner' ), sanitize_text_field( $label ), $budget ), 3000 ); |
| 419 | } |
| 420 | $memory_limit = $this->parse_ini_bytes( ini_get( 'memory_limit' ) ); |
| 421 | if ( $memory_limit > 0 && memory_get_usage( true ) > $memory_limit * 0.8 ) { |
| 422 | throw new Meow_WPMC_Transient_Exception( sprintf( __( '%s reached the safe parser memory limit.', 'media-cleaner' ), sanitize_text_field( $label ) ), 5000 ); |
| 423 | } |
| 424 | } |
| 425 | |
| 426 | public function run_paged_parser( $label, $fetch_page, $process_page, $page_size = 100, $cursor_resolver = null ) { |
| 427 | if ( !is_callable( $fetch_page ) || !is_callable( $process_page ) ) { |
| 428 | throw new InvalidArgumentException( __( 'A paged Media Cleaner parser requires callable fetch and process functions.', 'media-cleaner' ) ); |
| 429 | } |
| 430 | if ( $cursor_resolver !== null && !is_callable( $cursor_resolver ) ) { |
| 431 | throw new InvalidArgumentException( __( 'A Media Cleaner parser cursor resolver must be callable.', 'media-cleaner' ) ); |
| 432 | } |
| 433 | $label = sanitize_key( $label ); |
| 434 | $page_size = max( 10, min( 250, (int) $page_size ) ); |
| 435 | $work = null; |
| 436 | $offset = 0; |
| 437 | if ( $this->run_id > 0 && $this->runs ) { |
| 438 | $work = $this->runs->get_work( $this->run_id, 'parserPages', $label ); |
| 439 | if ( $work && $work->status === 'complete' ) return true; |
| 440 | if ( !$work ) { |
| 441 | if ( !$this->runs->enqueue_work( $this->run_id, 'parserPages', 'parser', $label ) ) { |
| 442 | throw new RuntimeException( sprintf( __( 'Media Cleaner could not initialize parser page state for %s.', 'media-cleaner' ), $label ) ); |
| 443 | } |
| 444 | $work = $this->runs->get_work( $this->run_id, 'parserPages', $label ); |
| 445 | } |
| 446 | if ( !$work ) throw new RuntimeException( sprintf( __( 'Media Cleaner could not read parser page state for %s.', 'media-cleaner' ), $label ) ); |
| 447 | $offset = max( 0, (int) $work->cursor_value ); |
| 448 | } |
| 449 | |
| 450 | $started = microtime( true ); |
| 451 | do { |
| 452 | $this->timeout_check(); |
| 453 | $rows = call_user_func( $fetch_page, $offset, $page_size ); |
| 454 | global $wpdb; |
| 455 | if ( $wpdb->last_error ) throw new RuntimeException( sprintf( __( '%1$s database error: %2$s', 'media-cleaner' ), $label, $wpdb->last_error ) ); |
| 456 | if ( is_wp_error( $rows ) ) throw new RuntimeException( $rows->get_error_message() ); |
| 457 | if ( !is_array( $rows ) ) $rows = array(); |
| 458 | $count = count( $rows ); |
| 459 | $next_offset = $offset + $count; |
| 460 | if ( $count > 0 && $cursor_resolver ) { |
| 461 | $resolved_cursor = call_user_func( $cursor_resolver, $rows, $offset ); |
| 462 | if ( !is_numeric( $resolved_cursor ) || (int) $resolved_cursor <= $offset ) { |
| 463 | throw new RuntimeException( sprintf( |
| 464 | __( 'Media Cleaner parser %s returned an invalid page cursor.', 'media-cleaner' ), |
| 465 | $label |
| 466 | ) ); |
| 467 | } |
| 468 | $next_offset = (int) $resolved_cursor; |
| 469 | } |
| 470 | call_user_func( $process_page, $rows ); |
| 471 | $this->write_references(); |
| 472 | $offset = $next_offset; |
| 473 | $finished = $count < $page_size; |
| 474 | if ( $work && !$this->runs->update_work( $work->id, $finished ? 'complete' : 'pending', $offset ) ) { |
| 475 | throw new RuntimeException( sprintf( __( 'Media Cleaner could not checkpoint parser %s.', 'media-cleaner' ), $label ) ); |
| 476 | } |
| 477 | if ( !$finished ) $this->parser_budget_check( $started, $label ); |
| 478 | } while ( !$finished ); |
| 479 | return true; |
| 480 | } |
| 481 | |
| 482 | function timeout_check_start( $count ) { |
| 483 | $this->start_time = microtime( true ); |
| 484 | $this->items_count = $count; |
| 485 | $this->items_checked = 0; |
| 486 | $this->item_scan_avg_time = 0; |
| 487 | $this->time_elapsed = 0; |
| 488 | $this->time_remaining = $this->get_request_time_budget() - ( microtime( true ) - $this->request_start_time ); |
| 489 | $this->get_max_execution_time(); |
| 490 | } |
| 491 | |
| 492 | function timeout_get_elapsed() { |
| 493 | return round( $this->time_elapsed, 2 ) . 's'; |
| 494 | } |
| 495 | |
| 496 | function timeout_check() { |
| 497 | $now = microtime( true ); |
| 498 | if ( empty( $this->start_time ) ) $this->start_time = $now; |
| 499 | $this->time_elapsed = $now - $this->start_time; |
| 500 | $this->time_remaining = $this->get_request_time_budget() - ( $now - $this->request_start_time ); |
| 501 | if ( $this->catch_timeout && $this->timeout_should_yield() ) { |
| 502 | error_log("Media Cleaner Timeout! Check the Media Cleaner logs for more info."); |
| 503 | $this->log( "😵 Timeout! Some info for debug:" ); |
| 504 | $this->log( "🍀 Elapsed time: $this->time_elapsed" ); |
| 505 | $this->log( "🍀 WP init time: $this->wordpress_init_time" ); |
| 506 | $this->log( "🍀 Remaining time: $this->time_remaining" ); |
| 507 | $this->log( "🍀 Scan time per item: $this->item_scan_avg_time" ); |
| 508 | $this->log( "🍀 PHP max_execution_time: $this->max_execution_time" ); |
| 509 | throw new Meow_WPMC_Transient_Exception( __( 'Media Cleaner paused this batch before the server execution-time or memory limit.', 'media-cleaner' ), 2000 ); |
| 510 | } |
| 511 | } |
| 512 | |
| 513 | public function timeout_should_yield() { |
| 514 | $now = microtime( true ); |
| 515 | $this->time_remaining = $this->get_request_time_budget() - ( $now - $this->request_start_time ); |
| 516 | $next_item_reserve = max( 1.25, $this->item_scan_avg_time * 1.75 ); |
| 517 | if ( !$this->is_cli && $this->time_remaining <= $next_item_reserve ) return true; |
| 518 | $memory_limit = $this->parse_ini_bytes( ini_get( 'memory_limit' ) ); |
| 519 | return $memory_limit > 0 && memory_get_usage( true ) >= $memory_limit * 0.82; |
| 520 | } |
| 521 | |
| 522 | function delete_attachment_related_data( $post_id ) { |
| 523 | |
| 524 | if ( empty( $post_id ) ) return; |
| 525 | |
| 526 | global $wpdb; |
| 527 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 528 | $run_id = $this->get_run_id(); |
| 529 | |
| 530 | if ( $wpdb->get_var( $wpdb->prepare( "SHOW TABLES LIKE %s", $wpdb->esc_like( $table_name ) ) ) === $table_name ) { |
| 531 | $wpdb->query( $wpdb->prepare( "DELETE FROM $table_name WHERE run_id = %d AND postId = %d", $run_id, $post_id ) ); |
| 532 | } else { |
| 533 | // Table does not exist |
| 534 | } |
| 535 | } |
| 536 | |
| 537 | function timeout_check_additem() { |
| 538 | $this->items_checked++; |
| 539 | $this->time_elapsed = microtime( true ) - $this->start_time; |
| 540 | $this->item_scan_avg_time = ceil( ( $this->time_elapsed / $this->items_checked ) * 10 ) / 10; |
| 541 | } |
| 542 | |
| 543 | // This checks if a new uploaded filename isn't the same one as a currently |
| 544 | // filename in the trash (that would cause issues) |
| 545 | function wp_unique_filename( $filename, $ext, $dir ) { |
| 546 | $fullpath = trailingslashit( $dir ) . $filename; |
| 547 | $relativepath = $this->clean_uploaded_filename( $fullpath ); |
| 548 | $trashfilepath = trailingslashit( $this->get_trashdir() ) . $relativepath; |
| 549 | if ( file_exists( $trashfilepath ) ) { |
| 550 | $path_parts = pathinfo( $fullpath ); |
| 551 | $filename_noext = $path_parts['filename']; |
| 552 | $new_filename = $filename_noext . '-' . date('Ymd-His', time()) . '.' . $path_parts['extension']; |
| 553 | //error_log( 'POTENTIALLY TRASH PATH: ' . $trashfilepath ); |
| 554 | //error_log( 'POTENTIALLY NEW FILE: ' . $new_filename ); |
| 555 | return $new_filename; |
| 556 | } |
| 557 | return $filename; |
| 558 | } |
| 559 | |
| 560 | function array_to_ids_or_urls( $meta, &$ids, &$urls, $recursive = false, $filters = array(), $depth = 0 ) { |
| 561 | if ( $depth > 64 || !is_array( $meta ) ) { |
| 562 | return; |
| 563 | } |
| 564 | foreach ( $meta as $k => $m ) { |
| 565 | |
| 566 | if ( is_numeric( $m ) ) { |
| 567 | |
| 568 | if ( !empty( $filters ) && is_array( $filters ) && !in_array( $k, $filters ) ) { |
| 569 | continue; |
| 570 | } |
| 571 | |
| 572 | // Probably a Media ID |
| 573 | if ( $m > 0 ) |
| 574 | { |
| 575 | array_push( $ids, $m ); |
| 576 | } |
| 577 | } |
| 578 | |
| 579 | else if ( is_array( $m ) ) { |
| 580 | |
| 581 | |
| 582 | if ( $recursive ) { |
| 583 | // If it's an array, we need to go deeper |
| 584 | $this->array_to_ids_or_urls( $m, $ids, $urls, true, $filters, $depth + 1 ); |
| 585 | } |
| 586 | |
| 587 | } |
| 588 | else if ( !empty( $m ) ) { |
| 589 | |
| 590 | if ( !empty( $filters ) && is_array( $filters ) && !in_array( $k, $filters ) ) { |
| 591 | continue; |
| 592 | } |
| 593 | |
| 594 | if ( is_string( $m ) && preg_match( '/^[\d\s,]+$/', $m ) && strpos( $m, ',' ) !== false ) { |
| 595 | // If this is a string that contains only digits, spaces, and commas, and contains at least one comma |
| 596 | // it is probably a list of IDs. So we should explode it to make an array |
| 597 | // Remove any spaces |
| 598 | |
| 599 | $m = str_replace( ' ', '', $m ); |
| 600 | $m = explode( ',', $m ); |
| 601 | |
| 602 | foreach ( $m as $mv ) { |
| 603 | if ( is_numeric( $mv ) && !in_array( (int)$mv, $ids ) ) { |
| 604 | array_push( $ids, (int)$mv ); |
| 605 | } |
| 606 | } |
| 607 | |
| 608 | continue; |
| 609 | } |
| 610 | |
| 611 | // If it's a string, maybe it's a file (with an extension) |
| 612 | if ( preg_match( $this->regex_file, $m ) ) |
| 613 | { |
| 614 | $clean_url = $this->clean_url( $m ); |
| 615 | array_push( $urls, $clean_url ); |
| 616 | } |
| 617 | } |
| 618 | } |
| 619 | } |
| 620 | |
| 621 | function get_favicon() { |
| 622 | // Yoast SEO plugin |
| 623 | $vals = get_option( 'wpseo_titles' ); |
| 624 | if ( !empty( $vals ) && isset( $vals['company_logo'] ) ) { |
| 625 | $url = $vals['company_logo']; |
| 626 | if ( $this->is_url( $url ) ) |
| 627 | return $this->clean_url( $url ); |
| 628 | } |
| 629 | } |
| 630 | |
| 631 | function get_all_shortcodes_attributes( $html, $ids_attr = array(), $urls_attr = array() ) { |
| 632 | // 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 |
| 633 | $urls_values = array(); |
| 634 | $ids_values = array(); |
| 635 | |
| 636 | $pattern = get_shortcode_regex(); |
| 637 | if ( preg_match_all( '/'. $pattern .'/s', $html, $matches ) ) |
| 638 | { |
| 639 | foreach( $matches[0] as $key => $value) { |
| 640 | // $matches[3] return the shortcode attribute as string |
| 641 | // replace space with '&' for parse_str() function |
| 642 | $get = str_replace(" ", "&" , trim( $matches[3][$key] ) ); |
| 643 | $get = str_replace('"', '' , $get ); |
| 644 | parse_str( $get, $sub_output ); |
| 645 | |
| 646 | foreach ( $sub_output as $attr_key => $attr_value ) { |
| 647 | |
| 648 | if ( in_array( $attr_key, $ids_attr ) ) { |
| 649 | if ( is_numeric( $attr_value ) && !in_array( (int)$attr_value, $ids_values ) ) { |
| 650 | array_push( $ids_values, (int)$attr_value ); |
| 651 | } |
| 652 | |
| 653 | // In case of separated by commas |
| 654 | else if ( strpos( $attr_value, ',' ) !== false ) { |
| 655 | $attr_value = str_replace(' ', '', $attr_value ); |
| 656 | $pieces = explode( ',', $attr_value ); |
| 657 | foreach ( $pieces as $pval ) { |
| 658 | if ( is_numeric( $pval ) && !in_array( (int)$pval, $ids_values ) ) { |
| 659 | array_push( $ids_values, (int)$pval ); |
| 660 | } |
| 661 | } |
| 662 | } |
| 663 | } |
| 664 | |
| 665 | else if ( in_array( $attr_key, $urls_attr ) ) { |
| 666 | if ( !empty( trim( $attr_value ) ) && !in_array( trim( $attr_value ), $urls_values ) && !is_numeric( trim( $attr_value ) ) && strpos( trim( $attr_value ), 'http' ) !== false ) { |
| 667 | array_push( $urls_values, trim( $this->clean_url( $attr_value ) ) ); |
| 668 | } |
| 669 | } |
| 670 | } |
| 671 | } |
| 672 | } |
| 673 | |
| 674 | // Remove duplicates |
| 675 | $urls_values = array_unique( $urls_values ); |
| 676 | $ids_values = array_unique( $ids_values ); |
| 677 | |
| 678 | // Return the values |
| 679 | $values = array( |
| 680 | 'urls' => $urls_values, |
| 681 | 'ids' => $ids_values |
| 682 | ); |
| 683 | |
| 684 | return $values; |
| 685 | |
| 686 | } |
| 687 | |
| 688 | |
| 689 | |
| 690 | /** |
| 691 | * Recursively transforms a string with WordPress shortcodes into a |
| 692 | * hierarchical tree structure (an Abstract Syntax Tree). |
| 693 | * |
| 694 | * @param string $content The string containing the shortcodes. |
| 695 | * @return array An array of nodes, where each node can be a shortcode with its |
| 696 | * own 'children' array, or a simple text node. |
| 697 | */ |
| 698 | function nested_shortcodes_to_array(string $content, $depth = 0): array |
| 699 | { |
| 700 | if ( $depth > 32 ) { |
| 701 | return array(); |
| 702 | } |
| 703 | $nodes = []; |
| 704 | $last_pos = 0; |
| 705 | |
| 706 | $pattern = '/\\[' . '(\\[?)' . '([\w-]+)' . '(?![\\w-])' . '(' . '[^\\]\\/]*' . '(?:' . '\\/(?!\\])' . '[^\\]\\/]*' . ')*?' . ')' . '(?:' . '(\\/)' . '\\]' . '|' . '\\]' . '(?:' . '(' . '[^\\[]*+' . '(?:' . '\\[(?!\\/\\2\\])' . '[^\\[]*+' . ')*+' . ')' . '\\[\\/\\2\\]' . ')?' . ')' . '(\\]?)/s'; |
| 707 | |
| 708 | // preg_match_all with PREG_OFFSET_CAPTURE is key to tracking positions. |
| 709 | if (preg_match_all($pattern, $content, $matches, PREG_SET_ORDER | PREG_OFFSET_CAPTURE)) { |
| 710 | foreach ($matches as $match) { |
| 711 | // Get the position and content of the full shortcode match |
| 712 | $match_start_pos = $match[0][1]; |
| 713 | $match_full_string = $match[0][0]; |
| 714 | $match_end_pos = $match_start_pos + strlen($match_full_string); |
| 715 | |
| 716 | // 1. Capture any text that appeared *before* this shortcode |
| 717 | if ($match_start_pos > $last_pos) { |
| 718 | $text_content = substr($content, $last_pos, $match_start_pos - $last_pos); |
| 719 | if (trim($text_content) !== '') { |
| 720 | $nodes[] = [ |
| 721 | 'type' => 'text', |
| 722 | 'content' => $text_content |
| 723 | ]; |
| 724 | } |
| 725 | } |
| 726 | |
| 727 | // 2. Process the shortcode match itself |
| 728 | $tag = $match[2][0]; |
| 729 | $attributes_string = $match[3][0]; |
| 730 | // Use isset since self-closing tags won't have inner content (group 5) |
| 731 | $inner_content = isset($match[5]) ? $match[5][0] : null; |
| 732 | |
| 733 | // Parse attributes from the attribute string |
| 734 | $parsed_attributes = []; |
| 735 | if (preg_match_all('/([\w-]+)\s*=\s*(["\'])([^"\']*?)\2/', $attributes_string, $attr_matches)) { |
| 736 | foreach ($attr_matches[1] as $attr_index => $key) { |
| 737 | $parsed_attributes[$key] = $attr_matches[3][$attr_index]; |
| 738 | } |
| 739 | } |
| 740 | |
| 741 | $shortcode_node = [ |
| 742 | 'type' => 'shortcode', |
| 743 | 'tag' => $tag, |
| 744 | 'attributes' => $parsed_attributes, |
| 745 | ]; |
| 746 | |
| 747 | // 3. This is the recursion! |
| 748 | // If there is inner content, parse it with the same function. |
| 749 | if ($inner_content !== null) { |
| 750 | $children = $this->nested_shortcodes_to_array( $inner_content, $depth + 1 ); |
| 751 | if (!empty($children)) { |
| 752 | $shortcode_node['children'] = $children; |
| 753 | } |
| 754 | } |
| 755 | |
| 756 | $nodes[] = $shortcode_node; |
| 757 | |
| 758 | // Update the last position to the end of the current match |
| 759 | $last_pos = $match_end_pos; |
| 760 | } |
| 761 | } |
| 762 | |
| 763 | // 4. Capture any remaining text after the very last shortcode |
| 764 | if ($last_pos < strlen($content)) { |
| 765 | $text_content = substr($content, $last_pos); |
| 766 | if (trim($text_content) !== '') { |
| 767 | $nodes[] = [ |
| 768 | 'type' => 'text', |
| 769 | 'content' => $text_content |
| 770 | ]; |
| 771 | } |
| 772 | } |
| 773 | |
| 774 | return $nodes; |
| 775 | } |
| 776 | |
| 777 | |
| 778 | |
| 779 | |
| 780 | function get_shortcode_attributes( $shortcode_tag, $post ) { |
| 781 | if ( has_shortcode( $post->post_content, $shortcode_tag ) ) { |
| 782 | $output = array(); |
| 783 | //get shortcode regex pattern wordpress function |
| 784 | $pattern = get_shortcode_regex( [ $shortcode_tag ] ); |
| 785 | if ( preg_match_all( '/'. $pattern .'/s', $post->post_content, $matches ) ) |
| 786 | { |
| 787 | $keys = array(); |
| 788 | $output = array(); |
| 789 | foreach( $matches[0] as $key => $value) { |
| 790 | // $matches[3] return the shortcode attribute as string |
| 791 | // replace space with '&' for parse_str() function |
| 792 | $get = str_replace(" ", "&" , trim( $matches[3][$key] ) ); |
| 793 | $get = str_replace('"', '' , $get ); |
| 794 | parse_str( $get, $sub_output ); |
| 795 | |
| 796 | //get all shortcode attribute keys |
| 797 | $keys = array_unique( array_merge( $keys, array_keys( $sub_output )) ); |
| 798 | $output[] = $sub_output; |
| 799 | } |
| 800 | if ( $keys && $output ) { |
| 801 | // Loop the output array and add the missing shortcode attribute key |
| 802 | foreach ($output as $key => $value) { |
| 803 | // Loop the shortcode attribute key |
| 804 | foreach ($keys as $attr_key) { |
| 805 | $output[$key][$attr_key] = isset( $output[$key] ) && isset( $output[$key] ) ? $output[$key][$attr_key] : NULL; |
| 806 | } |
| 807 | //sort the array key |
| 808 | ksort( $output[$key]); |
| 809 | } |
| 810 | } |
| 811 | } |
| 812 | return $output; |
| 813 | } |
| 814 | else { |
| 815 | return false; |
| 816 | } |
| 817 | } |
| 818 | |
| 819 | // Simply use regex to get URLs from a string return an array of URLs |
| 820 | function get_urls_from_string( $string ) { |
| 821 | if ( $this->analysis_document_too_large( $string ) ) { |
| 822 | return array(); |
| 823 | } |
| 824 | $urls = array(); |
| 825 | // Replace the sanitized urls with the real ones to be sure to get them in the regex |
| 826 | $string = str_replace( '\\', '', $string ); |
| 827 | |
| 828 | $patterns = array( |
| 829 | // Full URLs with protocol |
| 830 | '/(https?:\/\/[^\s\"\'\>\<\?\#]+\.(' . $this->types . '))/i', |
| 831 | |
| 832 | // Relative URLs starting with /wp-content/uploads or /uploads (without protocol) |
| 833 | '/(\/(?:wp-content\/)?uploads\/[^\s\"\'\>\<\?\#]+\.(' . $this->types . '))/i', |
| 834 | ); |
| 835 | |
| 836 | foreach ( $patterns as $pattern ) { |
| 837 | if ( preg_match_all( $pattern, $string, $matches ) ) { |
| 838 | foreach ( $matches[0] as $match ) { |
| 839 | $clean_url = $this->clean_url( $match ); |
| 840 | $urls[] = $clean_url; |
| 841 | } |
| 842 | } |
| 843 | } |
| 844 | |
| 845 | return array_unique( $urls ); |
| 846 | } |
| 847 | |
| 848 | function get_urls_from_html( $html ) { |
| 849 | if ( empty( $html ) ) { |
| 850 | return array(); |
| 851 | } |
| 852 | if ( $this->analysis_document_too_large( $html ) ) { |
| 853 | return array(); |
| 854 | } |
| 855 | |
| 856 | |
| 857 | // Proposal/fix by @copytrans |
| 858 | // Discussion: https://wordpress.org/support/topic/bug-in-core-php/#post-11647775 |
| 859 | // Modified by Jordy again in 2021 for those who don't have MB enabled |
| 860 | if ( function_exists( 'mb_encode_numericentity' ) ) { |
| 861 | $convmap = [0x80, 0xffff, 0, 0xffff]; |
| 862 | $html = mb_encode_numericentity( $html, $convmap, 'UTF-8' ); |
| 863 | } else { |
| 864 | $html = preg_replace_callback( |
| 865 | '/[\x80-\xFF]/', |
| 866 | function( $match ) { |
| 867 | return '&#' . ord( $match[0] ) . ';'; |
| 868 | }, |
| 869 | $html |
| 870 | ); |
| 871 | } |
| 872 | |
| 873 | // Remove any base64 src from the HTML to prevent regex from getting stuck and crashing the site |
| 874 | // Handles both proper (data:image/...) and malformed (image/jpeg;base64,...) base64 |
| 875 | // Also handles HTML-encoded quotes (") and multiline base64 |
| 876 | $html = preg_replace( '/src=["\'](?:data:)?(?:image|video|audio)\/[^"\']+;base64,[^"\']*["\']/', '', $html ); |
| 877 | $html = preg_replace( '/src="(?:data:)?(?:image|video|audio)\/[^&]+;base64,[^&]*"/', '', $html ); |
| 878 | // Catch any remaining base64 data that might cause regex issues (greedy catch-all) |
| 879 | $html = preg_replace( '/;base64,[a-zA-Z0-9+\/=\s]{1000,}/', '', $html ); |
| 880 | |
| 881 | |
| 882 | // Resolve src-set and shortcodes |
| 883 | if ( $this->get_shortcode_analysis() ) { |
| 884 | $html = do_shortcode( $html ); |
| 885 | if ( $this->analysis_document_too_large( $html ) ) { |
| 886 | return array(); |
| 887 | } |
| 888 | } |
| 889 | |
| 890 | // Create the DOM Document |
| 891 | if ( !class_exists("DOMDocument") ) { |
| 892 | error_log( 'Media Cleaner: The DOM extension for PHP is not installed.' ); |
| 893 | throw new Error( 'The DOM extension for PHP is not installed.' ); |
| 894 | } |
| 895 | |
| 896 | |
| 897 | if ( empty( $html ) ) { |
| 898 | return array(); |
| 899 | } |
| 900 | |
| 901 | $previous_libxml_errors = libxml_use_internal_errors( true ); |
| 902 | $dom = new DOMDocument(); |
| 903 | @$dom->loadHTML( $html, LIBXML_NONET | LIBXML_NOWARNING | LIBXML_NOERROR ); |
| 904 | libxml_clear_errors(); |
| 905 | libxml_use_internal_errors( $previous_libxml_errors ); |
| 906 | $results = array(); |
| 907 | |
| 908 | // <meta> tags in <head> area |
| 909 | $metas = $dom->getElementsByTagName( 'meta' ); |
| 910 | foreach ( $metas as $meta ) { |
| 911 | $property = $meta->getAttribute( 'property' ); |
| 912 | if ( $property == 'og:image' || $property == 'og:image:secure_url' || $property == 'twitter:image' ) { |
| 913 | $url = $meta->getAttribute( 'content' ); |
| 914 | if ( $this->is_url( $url ) ) { |
| 915 | $src = $this->clean_url( $url ); |
| 916 | if ( !empty( $src ) ) { |
| 917 | array_push( $results, $src ); |
| 918 | } |
| 919 | } |
| 920 | } |
| 921 | } |
| 922 | |
| 923 | |
| 924 | |
| 925 | // Iframe documents are not fetched while scanning. Fetching page-controlled |
| 926 | // URLs here made analysis depend on networking, allowed recursive documents, |
| 927 | // and could execute expensive remote rendering paths. |
| 928 | |
| 929 | |
| 930 | // Images: src, srcset |
| 931 | $imgs = $dom->getElementsByTagName( 'img' ); |
| 932 | foreach ( $imgs as $img ) { |
| 933 | //error_log($img->getAttribute('src')); |
| 934 | $src = $this->clean_url( $img->getAttribute('src') ); |
| 935 | array_push( $results, $src ); |
| 936 | $srcset = $img->getAttribute('srcset'); |
| 937 | if ( !empty( $srcset ) ) { |
| 938 | $setImgs = explode( ',', trim( $srcset ) ); |
| 939 | foreach ( $setImgs as $setImg ) { |
| 940 | $finalSetImg = explode( ' ', trim( $setImg ) ); |
| 941 | if ( is_array( $finalSetImg ) ) { |
| 942 | array_push( $results, $this->clean_url( $finalSetImg[0] ) ); |
| 943 | } |
| 944 | } |
| 945 | } |
| 946 | } |
| 947 | |
| 948 | // Videos: src, poster, and attached file |
| 949 | $videos = $dom->getElementsByTagName( 'video' ); |
| 950 | foreach ($videos as $video) { |
| 951 | // Get src attribute |
| 952 | $raw_video_src = $video->getAttribute( 'src' ); |
| 953 | $src = $this->clean_url( $raw_video_src ); |
| 954 | if ( !empty( $src ) ) { |
| 955 | $video_id = $this->custom_attachment_url_to_postid( $raw_video_src ); |
| 956 | |
| 957 | $attached_file = get_post_meta( $video_id, '_wp_attached_file', true ); |
| 958 | if ( !empty( $attached_file ) ) { |
| 959 | array_push( $results, $attached_file ); |
| 960 | } |
| 961 | } |
| 962 | |
| 963 | // Get poster attribute |
| 964 | $raw_poster_src = $video->getAttribute( 'poster' ); |
| 965 | $poster = $this->clean_url( $raw_poster_src ); |
| 966 | if ( !empty( $poster ) ) { |
| 967 | $poster_id = $this->custom_attachment_url_to_postid( $raw_poster_src ); |
| 968 | |
| 969 | $attached_file = get_post_meta( $poster_id, '_wp_attached_file', true ); |
| 970 | if ( !empty( $attached_file ) ) { |
| 971 | array_push( $results, $attached_file ); |
| 972 | } |
| 973 | } |
| 974 | |
| 975 | } |
| 976 | |
| 977 | // Audios: src |
| 978 | $audios = $dom->getElementsByTagName( 'audio' ); |
| 979 | foreach ( $audios as $audio ) { |
| 980 | //error_log($audio->getAttribute('src')); |
| 981 | $src = $this->clean_url( $audio->getAttribute('src') ); |
| 982 | array_push( $results, $src ); |
| 983 | } |
| 984 | |
| 985 | // Sources: src |
| 986 | $audios = $dom->getElementsByTagName( 'source' ); |
| 987 | foreach ( $audios as $audio ) { |
| 988 | //error_log($audio->getAttribute('src')); |
| 989 | $src = $this->clean_url( $audio->getAttribute('src') ); |
| 990 | array_push( $results, $src ); |
| 991 | } |
| 992 | |
| 993 | // Links, href |
| 994 | $urls = $dom->getElementsByTagName( 'a' ); |
| 995 | foreach ( $urls as $url ) { |
| 996 | $url_href = $url->getAttribute('href'); // mm change |
| 997 | if ( $this->is_url( $url_href ) ) { // mm change |
| 998 | $src = $this->clean_url( $url_href ); // mm change |
| 999 | if ( !empty( $src ) ) |
| 1000 | array_push( $results, $src ); |
| 1001 | } |
| 1002 | } |
| 1003 | |
| 1004 | // <link> tags in <head> area |
| 1005 | $urls = $dom->getElementsByTagName( 'link' ); |
| 1006 | foreach ( $urls as $url ) { |
| 1007 | $url_href = $url->getAttribute( 'href' ); |
| 1008 | if ( $this->is_url( $url_href ) ) { |
| 1009 | $src = $this->clean_url( $url_href ); |
| 1010 | if ( !empty( $src ) ) { |
| 1011 | array_push( $results, $src ); |
| 1012 | } |
| 1013 | } |
| 1014 | } |
| 1015 | |
| 1016 | |
| 1017 | preg_match_all( "/((https?:\/\/)?[^\\&\#\[\] \"\?]+\.pdf)/", $html, $res ); |
| 1018 | if ( !empty( $res ) && isset( $res[1] ) && count( $res[1] ) > 0 ) { |
| 1019 | foreach ( $res[1] as $url ) { |
| 1020 | if ( $this->is_url( $url ) ) |
| 1021 | array_push( $results, $this->clean_url( $url ) ); |
| 1022 | } |
| 1023 | } |
| 1024 | |
| 1025 | // Background images |
| 1026 | preg_match_all( "/url\(\'?\"?((https?:\/\/)?[^\\&\#\[\] \"\?]+\.(jpe?g|gif|png))\'?\"?/", $html, $res ); |
| 1027 | if ( !empty( $res ) && isset( $res[1] ) && count( $res[1] ) > 0 ) { |
| 1028 | foreach ( $res[1] as $url ) { |
| 1029 | if ( $this->is_url( $url ) ) |
| 1030 | array_push( $results, $this->clean_url( $url ) ); |
| 1031 | } |
| 1032 | } |
| 1033 | |
| 1034 | return $results; |
| 1035 | } |
| 1036 | |
| 1037 | private function analysis_document_too_large( $value ) { |
| 1038 | if ( !is_string( $value ) ) { |
| 1039 | return false; |
| 1040 | } |
| 1041 | $limit = (int) $this->get_option( 'analysis_document_limit' ); |
| 1042 | // A filter can still override the stored setting for advanced setups. |
| 1043 | $limit = (int) apply_filters( 'wpmc_max_analysis_document_bytes', $limit ); |
| 1044 | // -1 disables the limit: documents are analyzed no matter their size. |
| 1045 | if ( $limit < 0 ) { |
| 1046 | return false; |
| 1047 | } |
| 1048 | // Keep a 1 MB floor so a misconfigured tiny limit cannot cripple analysis. |
| 1049 | $limit = max( 1024 * 1024, $limit ); |
| 1050 | if ( strlen( $value ) > $limit ) { |
| 1051 | $this->log( sprintf( |
| 1052 | __( '⚠️ Skipped a content document larger than the %s analysis limit.', 'media-cleaner' ), |
| 1053 | size_format( $limit ) |
| 1054 | ) ); |
| 1055 | return true; |
| 1056 | } |
| 1057 | return false; |
| 1058 | } |
| 1059 | |
| 1060 | /** |
| 1061 | * |
| 1062 | * Get the IDs and URLs from the blocks of a post. |
| 1063 | * |
| 1064 | * @param string $html The HTML content of the post. |
| 1065 | * @param string $prefix The prefix of the blocks to look for. |
| 1066 | * @param array $keys The keys to look for in the blocks. |
| 1067 | * @param array $urls The array to fill with the URLs. |
| 1068 | * @param array $ids The array to fill with the IDs. |
| 1069 | * |
| 1070 | */ |
| 1071 | function get_from_blocks( $html, $prefix, $keys, &$urls, &$ids ) { |
| 1072 | |
| 1073 | $blocks = parse_blocks( $html ); |
| 1074 | |
| 1075 | if ( ! is_array( $blocks ) || ! isset( $blocks[0] ) ) { |
| 1076 | return; |
| 1077 | } |
| 1078 | |
| 1079 | |
| 1080 | foreach ( $blocks as $block ) { |
| 1081 | |
| 1082 | if ( strpos( $block['blockName'], $prefix ) === false ) { |
| 1083 | continue; |
| 1084 | } |
| 1085 | |
| 1086 | $this->array_to_ids_or_urls( $block, $ids, $urls, true, $keys ); |
| 1087 | |
| 1088 | } |
| 1089 | |
| 1090 | |
| 1091 | } |
| 1092 | // Parse a meta, visit all the arrays, look for the attributes, fill $ids and $urls arrays |
| 1093 | // If rawMode is enabled, it will not check if the value is an ID or an URL, it will just returns it in URLs |
| 1094 | function get_from_meta( $meta, $lookFor, &$ids, &$urls, $rawMode = false, $depth = 0 ) { |
| 1095 | if ( $depth > 64 || ( !is_array( $meta ) && !is_object( $meta) ) ) { |
| 1096 | return; |
| 1097 | } |
| 1098 | foreach ( $meta as $key => $value ) { |
| 1099 | if ( is_object( $value ) || is_array( $value ) ) |
| 1100 | $this->get_from_meta( $value, $lookFor, $ids, $urls, $rawMode, $depth + 1 ); |
| 1101 | else if ( in_array( $key, $lookFor ) ) { |
| 1102 | if ( empty( $value ) ) { |
| 1103 | continue; |
| 1104 | } |
| 1105 | else if ( $rawMode ) { |
| 1106 | array_push( $urls, $value ); |
| 1107 | } |
| 1108 | else if ( is_numeric( $value ) ) { |
| 1109 | // It this an ID? |
| 1110 | array_push( $ids, $value ); |
| 1111 | } |
| 1112 | else { |
| 1113 | if ( $this->is_url( $value ) ) { |
| 1114 | // Is this an URL? |
| 1115 | array_push( $urls, $this->clean_url( $value ) ); |
| 1116 | } |
| 1117 | else { |
| 1118 | // Is this an array of IDs, encoded as a string? (like "20,13") |
| 1119 | $pieces = explode( ',', $value ); |
| 1120 | foreach ( $pieces as $pval ) { |
| 1121 | if ( is_numeric( $pval ) ) { |
| 1122 | array_push( $ids, $pval ); |
| 1123 | } |
| 1124 | } |
| 1125 | } |
| 1126 | } |
| 1127 | } |
| 1128 | } |
| 1129 | } |
| 1130 | |
| 1131 | function get_images_from_themes( &$ids, &$urls ) { |
| 1132 | // USE CURRENT THEME AND WP API |
| 1133 | $ch = get_custom_header(); |
| 1134 | if ( !empty( $ch ) && !empty( $ch->url ) ) { |
| 1135 | array_push( $urls, $this->clean_url( $ch->url ) ); |
| 1136 | } |
| 1137 | if ( !empty( $ch ) && !empty( $ch->thumbnail_url ) && $this->is_url( $ch->thumbnail_url ) ) { |
| 1138 | array_push( $urls, $this->clean_url( $ch->thumbnail_url ) ); |
| 1139 | } |
| 1140 | if ( !empty( $ch ) && !empty( $ch->attachment_id ) ) { |
| 1141 | array_push( $ids, $ch->attachment_id ); |
| 1142 | } |
| 1143 | $cl = get_custom_logo(); |
| 1144 | if ( $this->is_url( $cl ) ) { |
| 1145 | $urls = array_merge( $this->get_urls_from_html( $cl ), $urls ); |
| 1146 | } |
| 1147 | $custom_logo = get_theme_mod( 'custom_logo' ); |
| 1148 | if ( !empty( $custom_logo ) && is_numeric( $custom_logo ) ) { |
| 1149 | array_push( $ids, (int)$custom_logo ); |
| 1150 | } |
| 1151 | $si = get_site_icon_url(); |
| 1152 | if ( $this->is_url( $si ) ) { |
| 1153 | array_push( $urls, $this->clean_url( $si ) ); |
| 1154 | } |
| 1155 | $si_id = get_option( 'site_icon' ); |
| 1156 | if ( !empty( $si_id ) && is_numeric( $si_id ) ) { |
| 1157 | array_push( $ids, (int)$si_id ); |
| 1158 | } |
| 1159 | $cd = get_background_image(); |
| 1160 | if ( $this->is_url( $cd ) ) { |
| 1161 | array_push( $urls, $this->clean_url( $cd ) ); |
| 1162 | } |
| 1163 | $photography_hero_image = get_theme_mod( 'photography_hero_image' ); |
| 1164 | if ( !empty( $photography_hero_image ) ) { |
| 1165 | array_push( $ids, $photography_hero_image ); |
| 1166 | } |
| 1167 | $author_profile_picture = get_theme_mod( 'author_profile_picture' ); |
| 1168 | if ( !empty( $author_profile_picture ) ) { |
| 1169 | array_push( $ids, $author_profile_picture ); |
| 1170 | } |
| 1171 | if ( function_exists ( 'get_uploaded_header_images' ) ) { |
| 1172 | $header_images = get_uploaded_header_images(); |
| 1173 | if ( !empty( $header_images ) ) { |
| 1174 | foreach ( $header_images as $hi ) { |
| 1175 | if ( !empty ( $hi['attachment_id'] ) ) { |
| 1176 | array_push( $ids, $hi['attachment_id'] ); |
| 1177 | } |
| 1178 | } |
| 1179 | } |
| 1180 | } |
| 1181 | } |
| 1182 | |
| 1183 | #region LOGS |
| 1184 | |
| 1185 | function log( $data = null, $force = false ) { |
| 1186 | if ( !$this->debug_logs && !$force ) |
| 1187 | return; |
| 1188 | |
| 1189 | $php_logs = $this->get_option( 'php_error_logs' ); |
| 1190 | $log_file_path = $this->get_logs_path(); |
| 1191 | if ( !$log_file_path ) return false; |
| 1192 | if ( file_exists( $log_file_path ) && filesize( $log_file_path ) > 5 * 1024 * 1024 ) { |
| 1193 | if ( file_exists( $log_file_path . '.1' ) ) @unlink( $log_file_path . '.1' ); |
| 1194 | @rename( $log_file_path, $log_file_path . '.1' ); |
| 1195 | } |
| 1196 | |
| 1197 | $fh = @fopen( $log_file_path, 'a' ); |
| 1198 | if ( !$fh ) { return false; } |
| 1199 | $date = date( "Y-m-d H:i:s" ); |
| 1200 | if ( is_null( $data ) ) { |
| 1201 | fwrite( $fh, "\n" ); |
| 1202 | } |
| 1203 | else { |
| 1204 | $message = is_scalar( $data ) ? (string) $data : wp_json_encode( $data ); |
| 1205 | fwrite( $fh, "$date: {$message}\n" ); |
| 1206 | if ( $php_logs ) { |
| 1207 | error_log( "[MEDIA CLEANER] " . $message ); |
| 1208 | } |
| 1209 | } |
| 1210 | fclose( $fh ); |
| 1211 | return true; |
| 1212 | } |
| 1213 | |
| 1214 | //WPMC_PREFIX |
| 1215 | |
| 1216 | function get_logs_path() { |
| 1217 | $private_root = $this->ensure_private_root(); |
| 1218 | if ( is_wp_error( $private_root ) ) return false; |
| 1219 | $log_dir = trailingslashit( $private_root ) . 'logs'; |
| 1220 | if ( !is_dir( $log_dir ) ) { |
| 1221 | wp_mkdir_p( $log_dir ); |
| 1222 | } |
| 1223 | $path = trailingslashit( $log_dir ) . 'media-cleaner.log'; |
| 1224 | $old_path = $this->get_option( 'logs_path' ); |
| 1225 | $legacy_path = wp_normalize_path( WPMC_PATH . '/logs/media-cleaner.log' ); |
| 1226 | $old_normalized = is_string( $old_path ) ? wp_normalize_path( $old_path ) : ''; |
| 1227 | if ( $old_normalized === $legacy_path && $old_normalized !== $path && is_file( $old_normalized ) ) { |
| 1228 | if ( !@rename( $old_normalized, $path ) ) { |
| 1229 | if ( @copy( $old_normalized, $path ) ) @unlink( $old_normalized ); |
| 1230 | } |
| 1231 | } |
| 1232 | if ( !file_exists( $path ) ) { |
| 1233 | @touch( $path ); |
| 1234 | } |
| 1235 | if ( $old_path !== $path ) { |
| 1236 | $options = $this->get_all_options(); |
| 1237 | $options['logs_path'] = $path; |
| 1238 | $this->update_options( $options ); |
| 1239 | } |
| 1240 | return $path; |
| 1241 | } |
| 1242 | |
| 1243 | |
| 1244 | function get_logs() { |
| 1245 | $log_file_path = $this->get_logs_path(); |
| 1246 | |
| 1247 | if ( !$log_file_path || !file_exists( $log_file_path ) ) { |
| 1248 | return __( 'No logs found.', 'media-cleaner' ); |
| 1249 | } |
| 1250 | |
| 1251 | $size = filesize( $log_file_path ); |
| 1252 | $bytes = min( 512 * 1024, max( 0, (int) $size ) ); |
| 1253 | $handle = fopen( $log_file_path, 'rb' ); |
| 1254 | if ( !$handle ) return __( 'No logs found.', 'media-cleaner' ); |
| 1255 | if ( $bytes < $size ) fseek( $handle, -$bytes, SEEK_END ); |
| 1256 | $content = $bytes > 0 ? fread( $handle, $bytes ) : ''; |
| 1257 | fclose( $handle ); |
| 1258 | if ( $bytes < $size ) { |
| 1259 | $first_newline = strpos( $content, "\n" ); |
| 1260 | $content = $first_newline === false ? '' : substr( $content, $first_newline + 1 ); |
| 1261 | } |
| 1262 | $lines = explode( "\n", $content ); |
| 1263 | $lines = array_filter( $lines ); |
| 1264 | $lines = array_slice( array_reverse( $lines ), 0, 2000 ); |
| 1265 | $content = implode( "\n", $lines ); |
| 1266 | return $content; |
| 1267 | } |
| 1268 | |
| 1269 | function clear_logs() { |
| 1270 | $logPath = $this->get_logs_path(); |
| 1271 | if ( $logPath && file_exists( $logPath ) ) { |
| 1272 | unlink( $logPath ); |
| 1273 | } |
| 1274 | |
| 1275 | $options = $this->get_all_options(); |
| 1276 | $options['logs_path'] = null; |
| 1277 | $this->update_options( $options ); |
| 1278 | } |
| 1279 | |
| 1280 | #endregion |
| 1281 | |
| 1282 | /** |
| 1283 | * |
| 1284 | * HELPERS |
| 1285 | * |
| 1286 | */ |
| 1287 | |
| 1288 | private function random_ascii_chars($length = 8) |
| 1289 | { |
| 1290 | $characters = array_merge(range('A', 'Z'), range('a', 'z'), range('0', '9')); |
| 1291 | $characters_length = count($characters); |
| 1292 | $random_string = ''; |
| 1293 | |
| 1294 | for ($i = 0; $i < $length; $i++) { |
| 1295 | $random_string .= $characters[rand(0, $characters_length - 1)]; |
| 1296 | } |
| 1297 | |
| 1298 | return $random_string; |
| 1299 | } |
| 1300 | |
| 1301 | private function get_private_root() { |
| 1302 | $secret = get_option( 'wpmc_trash_secret', null ); |
| 1303 | if ( !$secret ) { |
| 1304 | $candidate = wp_generate_password( 32, false, false ); |
| 1305 | $secret = add_option( 'wpmc_trash_secret', $candidate, '', false ) ? $candidate : get_option( 'wpmc_trash_secret' ); |
| 1306 | } |
| 1307 | if ( !$secret ) $secret = wp_salt( 'auth' ); |
| 1308 | $parent = defined( 'WPMC_TRASH_DIR' ) ? dirname( WPMC_TRASH_DIR ) : WP_CONTENT_DIR; |
| 1309 | $base = trailingslashit( $parent ) . '.media-cleaner-private-' . substr( hash( 'sha256', $secret ), 0, 20 ); |
| 1310 | return untrailingslashit( wp_normalize_path( $base ) ); |
| 1311 | } |
| 1312 | |
| 1313 | private function protect_private_directory( $root ) { |
| 1314 | $guards = array( |
| 1315 | trailingslashit( $root ) . 'index.php' => "<?php\nhttp_response_code( 404 );\nexit;\n", |
| 1316 | trailingslashit( $root ) . '.htaccess' => "Options -Indexes\n<IfModule mod_authz_core.c>Require all denied</IfModule>\n<IfModule !mod_authz_core.c>Deny from all</IfModule>\n", |
| 1317 | trailingslashit( $root ) . 'web.config' => "<?xml version=\"1.0\"?><configuration><system.webServer><security><authorization><remove users=\"*\" roles=\"\" verbs=\"\"/><add accessType=\"Deny\" users=\"*\"/></authorization></security></system.webServer></configuration>", |
| 1318 | ); |
| 1319 | foreach ( $guards as $path => $content ) { |
| 1320 | if ( !file_exists( $path ) && @file_put_contents( $path, $content ) === false ) return false; |
| 1321 | } |
| 1322 | return true; |
| 1323 | } |
| 1324 | |
| 1325 | private function ensure_private_root() { |
| 1326 | $root = $this->get_private_root(); |
| 1327 | if ( is_link( $root ) ) { |
| 1328 | return new WP_Error( 'wpmc_private_storage_unsafe', __( 'Media Cleaner private storage cannot be a symbolic link.', 'media-cleaner' ) ); |
| 1329 | } |
| 1330 | if ( !is_dir( $root ) && !wp_mkdir_p( $root ) ) { |
| 1331 | return new WP_Error( 'wpmc_private_storage_unavailable', __( 'Media Cleaner could not create its private storage directory.', 'media-cleaner' ) ); |
| 1332 | } |
| 1333 | if ( !$this->protect_private_directory( $root ) ) return new WP_Error( 'wpmc_private_storage_unprotected', __( 'Media Cleaner could not protect its private storage directory.', 'media-cleaner' ) ); |
| 1334 | return $root; |
| 1335 | } |
| 1336 | |
| 1337 | function get_trashdir() { |
| 1338 | $trash = defined( 'WPMC_TRASH_DIR' ) ? untrailingslashit( wp_normalize_path( WPMC_TRASH_DIR ) ) : trailingslashit( $this->get_private_root() ) . 'trash'; |
| 1339 | |
| 1340 | $legacy = trailingslashit( $this->upload_path ) . 'wpmc-trash'; |
| 1341 | $trash_is_public = $trash === $this->upload_path || strpos( $trash, trailingslashit( $this->upload_path ) ) === 0; |
| 1342 | if ( is_dir( $legacy ) || is_link( $legacy ) ) { |
| 1343 | if ( $trash_is_public || file_exists( $trash ) || is_link( $legacy ) || !wp_mkdir_p( dirname( $trash ) ) || !@rename( $legacy, $trash ) ) { |
| 1344 | $this->trash_migration_error = new WP_Error( |
| 1345 | 'wpmc_trash_migration_failed', |
| 1346 | __( 'The old public Media Cleaner trash could not be moved to private storage. Cleanup is blocked until that directory can be migrated.', 'media-cleaner' ) |
| 1347 | ); |
| 1348 | } |
| 1349 | } |
| 1350 | return $trash; |
| 1351 | } |
| 1352 | |
| 1353 | private function ensure_trash_directory() { |
| 1354 | $trash = $this->get_trashdir(); |
| 1355 | if ( is_wp_error( $this->trash_migration_error ) ) return $this->trash_migration_error; |
| 1356 | $private_root = $this->ensure_private_root(); |
| 1357 | if ( is_wp_error( $private_root ) ) return $private_root; |
| 1358 | if ( $trash === $this->upload_path || strpos( $trash, trailingslashit( $this->upload_path ) ) === 0 ) { |
| 1359 | return new WP_Error( 'wpmc_public_trash_directory', __( 'Media Cleaner quarantine must be outside the public uploads directory.', 'media-cleaner' ) ); |
| 1360 | } |
| 1361 | if ( !is_dir( $trash ) && !wp_mkdir_p( $trash ) ) { |
| 1362 | return new WP_Error( 'wpmc_trash_unavailable', __( 'Media Cleaner could not create its private trash directory.', 'media-cleaner' ) ); |
| 1363 | } |
| 1364 | $trash_real = realpath( $trash ); |
| 1365 | $upload_real = realpath( $this->upload_path ); |
| 1366 | $trash_real = $trash_real ? untrailingslashit( wp_normalize_path( $trash_real ) ) : ''; |
| 1367 | $upload_real = $upload_real ? untrailingslashit( wp_normalize_path( $upload_real ) ) : ''; |
| 1368 | if ( $upload_real && ( $trash_real === $upload_real || strpos( $trash_real, trailingslashit( $upload_real ) ) === 0 ) ) { |
| 1369 | return new WP_Error( 'wpmc_public_trash_directory', __( 'Media Cleaner quarantine resolves inside the public uploads directory.', 'media-cleaner' ) ); |
| 1370 | } |
| 1371 | if ( is_link( $trash ) || !$this->protect_private_directory( $trash ) ) { |
| 1372 | return new WP_Error( 'wpmc_trash_unprotected', __( 'Media Cleaner quarantine is unsafe or could not be protected.', 'media-cleaner' ) ); |
| 1373 | } |
| 1374 | return wp_normalize_path( $trash ); |
| 1375 | } |
| 1376 | |
| 1377 | public function prepare_private_storage() { |
| 1378 | return $this->ensure_trash_directory(); |
| 1379 | } |
| 1380 | |
| 1381 | public function test_quarantine_roundtrip() { |
| 1382 | $trash = $this->ensure_trash_directory(); |
| 1383 | if ( is_wp_error( $trash ) ) return $trash; |
| 1384 | $source = tempnam( $this->upload_path, '.wpmc-roundtrip-' ); |
| 1385 | if ( !$source || file_put_contents( $source, 'media-cleaner-storage-test' ) === false ) { |
| 1386 | if ( $source && file_exists( $source ) ) @unlink( $source ); |
| 1387 | return new WP_Error( 'wpmc_storage_test_create_failed', __( 'Media Cleaner could not create a storage test file.', 'media-cleaner' ) ); |
| 1388 | } |
| 1389 | $relative = '.wpmc-roundtrip-' . wp_generate_uuid4(); |
| 1390 | $quarantine = $this->resolve_trash_path( $relative ); |
| 1391 | $returned = $source . '.returned'; |
| 1392 | $success = !is_wp_error( $quarantine ) && @rename( $source, $quarantine ) && @rename( $quarantine, $returned ) && @unlink( $returned ); |
| 1393 | foreach ( array( $source, is_wp_error( $quarantine ) ? null : $quarantine, $returned ) as $path ) { |
| 1394 | if ( $path && file_exists( $path ) ) @unlink( $path ); |
| 1395 | } |
| 1396 | return $success ? true : new WP_Error( 'wpmc_storage_roundtrip_failed', __( 'Files cannot be moved safely into and back out of Media Cleaner quarantine. Check mount points and permissions.', 'media-cleaner' ) ); |
| 1397 | } |
| 1398 | |
| 1399 | public function resolve_trash_path( $relative_path, $must_exist = false ) { |
| 1400 | $relative = $this->normalize_upload_relative_path( $relative_path ); |
| 1401 | if ( is_wp_error( $relative ) ) { |
| 1402 | return $relative; |
| 1403 | } |
| 1404 | $root = $this->ensure_trash_directory(); |
| 1405 | if ( is_wp_error( $root ) ) { |
| 1406 | return $root; |
| 1407 | } |
| 1408 | $root_real = realpath( $root ); |
| 1409 | if ( !$root_real ) { |
| 1410 | return new WP_Error( 'wpmc_trash_unavailable', __( 'Media Cleaner trash is unavailable.', 'media-cleaner' ) ); |
| 1411 | } |
| 1412 | $root_real = untrailingslashit( wp_normalize_path( $root_real ) ); |
| 1413 | $candidate = $root_real . ( $relative === '' ? '' : '/' . $relative ); |
| 1414 | if ( is_link( $candidate ) ) { |
| 1415 | return new WP_Error( 'wpmc_trash_path_invalid', __( 'The trash path is unsafe.', 'media-cleaner' ) ); |
| 1416 | } |
| 1417 | if ( $must_exist && !file_exists( $candidate ) ) { |
| 1418 | return new WP_Error( 'wpmc_trash_item_missing', __( 'The requested item no longer exists in Media Cleaner trash.', 'media-cleaner' ) ); |
| 1419 | } |
| 1420 | if ( file_exists( $candidate ) ) { |
| 1421 | $resolved = wp_normalize_path( realpath( $candidate ) ); |
| 1422 | if ( is_link( $candidate ) || ( $resolved !== $root_real && strpos( $resolved, trailingslashit( $root_real ) ) !== 0 ) ) { |
| 1423 | return new WP_Error( 'wpmc_trash_path_invalid', __( 'The trash path is unsafe.', 'media-cleaner' ) ); |
| 1424 | } |
| 1425 | } |
| 1426 | else { |
| 1427 | $ancestor = dirname( $candidate ); |
| 1428 | while ( !file_exists( $ancestor ) && dirname( $ancestor ) !== $ancestor ) { |
| 1429 | $ancestor = dirname( $ancestor ); |
| 1430 | } |
| 1431 | $ancestor_real = realpath( $ancestor ); |
| 1432 | $ancestor_real = $ancestor_real ? untrailingslashit( wp_normalize_path( $ancestor_real ) ) : ''; |
| 1433 | if ( $ancestor_real !== $root_real && strpos( $ancestor_real, trailingslashit( $root_real ) ) !== 0 ) { |
| 1434 | return new WP_Error( 'wpmc_trash_path_invalid', __( 'The trash path is unsafe.', 'media-cleaner' ) ); |
| 1435 | } |
| 1436 | } |
| 1437 | return $candidate; |
| 1438 | } |
| 1439 | |
| 1440 | function get_trashurl() { |
| 1441 | // Trash is intentionally outside the public uploads URL. |
| 1442 | return null; |
| 1443 | } |
| 1444 | |
| 1445 | function clean_ob(){ |
| 1446 | $disabled = $this->get_option( 'output_buffer_cleaning_disabled' ); |
| 1447 | $ob_content = ob_get_contents(); |
| 1448 | if ( is_string( $ob_content ) && trim( $ob_content ) !== '' ) { |
| 1449 | |
| 1450 | if ( $disabled ) { |
| 1451 | $this->log( "🚨 If the server's response was broken, try to let Output Buffer Cleaning enabled." ); |
| 1452 | return; |
| 1453 | } |
| 1454 | |
| 1455 | $this->log( "🧹 The response is broken due to output buffering, it will be cleaned." ); |
| 1456 | $this->log( "📄 Output buffer content: " . $ob_content ); |
| 1457 | |
| 1458 | if ( ob_get_level() > 0 ) { |
| 1459 | ob_end_clean(); |
| 1460 | } |
| 1461 | } |
| 1462 | } |
| 1463 | |
| 1464 | /** |
| 1465 | * |
| 1466 | * I18N RELATED HELPERS |
| 1467 | * |
| 1468 | */ |
| 1469 | |
| 1470 | function is_multilingual() { |
| 1471 | return function_exists( 'icl_get_languages' ); |
| 1472 | } |
| 1473 | |
| 1474 | function get_languages() { |
| 1475 | $results = array(); |
| 1476 | if ( $this->is_multilingual() ) { |
| 1477 | $languages = icl_get_languages(); |
| 1478 | foreach ( $languages as $language ) { |
| 1479 | if ( isset( $language['code'] ) ) { |
| 1480 | array_push( $results, $language['code'] ); |
| 1481 | } |
| 1482 | else if ( isset( $language['language_code'] ) ) { |
| 1483 | array_push( $results, $language['language_code'] ); |
| 1484 | } |
| 1485 | } |
| 1486 | } |
| 1487 | return $results; |
| 1488 | } |
| 1489 | |
| 1490 | function get_translated_media_ids( $mediaId ) { |
| 1491 | if ( isset( $this->translated_ids_cache[ $mediaId ] ) ) { |
| 1492 | return $this->translated_ids_cache[ $mediaId ]; |
| 1493 | } |
| 1494 | $translated_ids = array(); |
| 1495 | foreach ( $this->languages as $language ) { |
| 1496 | $id = apply_filters( 'wpml_object_id', $mediaId, 'attachment', false, $language ); |
| 1497 | if ( !empty( $id ) ) { |
| 1498 | array_push( $translated_ids, $id ); |
| 1499 | } |
| 1500 | } |
| 1501 | $this->translated_ids_cache[ $mediaId ] = $translated_ids; |
| 1502 | return $translated_ids; |
| 1503 | } |
| 1504 | |
| 1505 | /** |
| 1506 | * |
| 1507 | * DELETE / SCANNING / RESET |
| 1508 | * |
| 1509 | */ |
| 1510 | |
| 1511 | function recover_file( $path, &$did_move = null ) { |
| 1512 | $did_move = false; |
| 1513 | $relative = $this->normalize_upload_relative_path( $path ); |
| 1514 | if ( is_wp_error( $relative ) ) return $relative; |
| 1515 | $original_path = $this->resolve_upload_path( $relative ); |
| 1516 | $trash_path = $this->resolve_trash_path( $relative ); |
| 1517 | if ( is_wp_error( $original_path ) ) return $original_path; |
| 1518 | if ( is_wp_error( $trash_path ) ) return $trash_path; |
| 1519 | if ( !file_exists( $trash_path ) ) { |
| 1520 | return file_exists( $original_path ) ? true : new WP_Error( 'wpmc_recovery_source_missing', __( 'The file exists in neither uploads nor Media Cleaner trash.', 'media-cleaner' ) ); |
| 1521 | } |
| 1522 | if ( file_exists( $original_path ) ) { |
| 1523 | return new WP_Error( 'wpmc_recovery_collision', __( 'A file already exists at the recovery destination.', 'media-cleaner' ) ); |
| 1524 | } |
| 1525 | if ( !wp_mkdir_p( dirname( $original_path ) ) || !@rename( $trash_path, $original_path ) ) { |
| 1526 | return new WP_Error( 'wpmc_recovery_move_failed', __( 'The file could not be moved out of Media Cleaner trash.', 'media-cleaner' ) ); |
| 1527 | } |
| 1528 | $did_move = true; |
| 1529 | return true; |
| 1530 | } |
| 1531 | |
| 1532 | function recover( $id, $operation_manifest = array() ) { |
| 1533 | $staged = $this->results_staged_error(); |
| 1534 | if ( $staged ) return $staged; |
| 1535 | global $wpdb; |
| 1536 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 1537 | $issue = $this->get_issue( $id ); |
| 1538 | |
| 1539 | if ( empty( $issue ) ) { |
| 1540 | return new WP_Error( 'wpmc_issue_missing', __( 'The selected Media Cleaner result no longer exists.', 'media-cleaner' ) ); |
| 1541 | } |
| 1542 | if ( empty( $operation_manifest['identity_validated'] ) ) { |
| 1543 | $identity = $this->validate_issue_manifest( $issue ); |
| 1544 | if ( is_wp_error( $identity ) ) return $identity; |
| 1545 | } |
| 1546 | |
| 1547 | // Files |
| 1548 | if ( $issue->type === 0 ) { |
| 1549 | $did_move = false; |
| 1550 | $moved = $this->recover_file( $issue->path, $did_move ); |
| 1551 | if ( is_wp_error( $moved ) ) return $moved; |
| 1552 | $updated = $wpdb->query( $wpdb->prepare( "UPDATE $table_name SET deleted = 0 WHERE id = %d", $id ) ); |
| 1553 | if ( $updated === false ) { |
| 1554 | if ( $did_move ) $this->trash_file( $issue->path ); |
| 1555 | return new WP_Error( 'wpmc_recovery_database_failed', __( 'The file was moved back because its database state could not be updated.', 'media-cleaner' ) ); |
| 1556 | } |
| 1557 | $this->log( "� |
| 1558 | Recovered {$issue->path}." ); |
| 1559 | return true; |
| 1560 | } |
| 1561 | // Media |
| 1562 | else if ( $issue->type === 1 ) { |
| 1563 | |
| 1564 | $paths = $this->get_paths_from_attachment( $issue->postId ); |
| 1565 | $file_manifest = json_decode( (string) $issue->manifest, true ); |
| 1566 | $file_manifest = is_array( $file_manifest ) ? $file_manifest : array(); |
| 1567 | $recovered_paths = array(); |
| 1568 | foreach ( $paths as $path ) { |
| 1569 | if ( array_key_exists( $path, $file_manifest ) && $file_manifest[ $path ] === null ) continue; |
| 1570 | $did_move = false; |
| 1571 | $result = $this->recover_file( $path, $did_move ); |
| 1572 | if ( is_wp_error( $result ) ) { |
| 1573 | foreach ( array_reverse( $recovered_paths ) as $recovered_path ) { |
| 1574 | $this->trash_file( $recovered_path ); |
| 1575 | } |
| 1576 | return $result; |
| 1577 | } |
| 1578 | if ( $did_move ) $recovered_paths[] = $path; |
| 1579 | } |
| 1580 | $previous_post_type = get_post_type( $issue->postId ); |
| 1581 | $post_result = wp_update_post( array( 'ID' => $issue->postId, 'post_type' => 'attachment' ), true ); |
| 1582 | if ( is_wp_error( $post_result ) || !$post_result ) { |
| 1583 | foreach ( array_reverse( $recovered_paths ) as $recovered_path ) { |
| 1584 | $this->trash_file( $recovered_path ); |
| 1585 | } |
| 1586 | return is_wp_error( $post_result ) ? $post_result : new WP_Error( 'wpmc_recovery_post_failed', __( 'The attachment record could not be restored.', 'media-cleaner' ) ); |
| 1587 | } |
| 1588 | $updated = $wpdb->query( $wpdb->prepare( "UPDATE $table_name SET deleted = 0 WHERE id = %d", $id ) ); |
| 1589 | if ( $updated === false ) { |
| 1590 | if ( $previous_post_type && $previous_post_type !== 'attachment' ) { |
| 1591 | wp_update_post( array( 'ID' => $issue->postId, 'post_type' => $previous_post_type ) ); |
| 1592 | } |
| 1593 | foreach ( array_reverse( $recovered_paths ) as $recovered_path ) { |
| 1594 | $this->trash_file( $recovered_path ); |
| 1595 | } |
| 1596 | return new WP_Error( 'wpmc_recovery_database_failed', __( 'The attachment was restored, but Media Cleaner could not update its result record.', 'media-cleaner' ) ); |
| 1597 | } |
| 1598 | $this->log( "� |
| 1599 | Recovered Media #{$issue->postId}." ); |
| 1600 | return true; |
| 1601 | } |
| 1602 | return new WP_Error( 'wpmc_issue_type_invalid', __( 'The selected Media Cleaner result has an unsupported type.', 'media-cleaner' ) ); |
| 1603 | } |
| 1604 | |
| 1605 | function trash_file( $fileIssuePath, &$did_move = null ) { |
| 1606 | $did_move = false; |
| 1607 | $relative = $this->normalize_upload_relative_path( $fileIssuePath ); |
| 1608 | if ( is_wp_error( $relative ) ) return $relative; |
| 1609 | $original_path = $this->resolve_upload_path( $relative ); |
| 1610 | $trash_path = $this->resolve_trash_path( $relative ); |
| 1611 | if ( is_wp_error( $original_path ) ) return $original_path; |
| 1612 | if ( is_wp_error( $trash_path ) ) return $trash_path; |
| 1613 | if ( !file_exists( $original_path ) ) { |
| 1614 | return file_exists( $trash_path ) ? true : new WP_Error( 'wpmc_delete_source_missing', __( 'The file exists in neither uploads nor Media Cleaner trash.', 'media-cleaner' ) ); |
| 1615 | } |
| 1616 | if ( is_dir( $original_path ) || is_link( $original_path ) ) { |
| 1617 | return new WP_Error( 'wpmc_delete_unsafe_path', __( 'Media Cleaner will not move directories or symbolic links.', 'media-cleaner' ) ); |
| 1618 | } |
| 1619 | if ( file_exists( $trash_path ) ) { |
| 1620 | return new WP_Error( 'wpmc_trash_collision', __( 'A different file already exists at the trash destination.', 'media-cleaner' ) ); |
| 1621 | } |
| 1622 | if ( !wp_mkdir_p( dirname( $trash_path ) ) || !@rename( $original_path, $trash_path ) ) { |
| 1623 | return new WP_Error( 'wpmc_trash_move_failed', __( 'The file could not be moved into Media Cleaner trash.', 'media-cleaner' ) ); |
| 1624 | } |
| 1625 | $did_move = true; |
| 1626 | $this->clean_dir( dirname( $original_path ) ); |
| 1627 | return true; |
| 1628 | } |
| 1629 | |
| 1630 | function repair( $id ) { |
| 1631 | $repair = $this->get_repair( $id ); |
| 1632 | if ( empty( $repair ) ) { |
| 1633 | return new WP_Error( 'wpmc_repair_missing', __( 'The selected repair result no longer exists.', 'media-cleaner' ) ); |
| 1634 | } |
| 1635 | $full_path = $this->get_full_upload_path( $repair->path ); |
| 1636 | if ( !$full_path || !is_file( $full_path ) || is_link( $full_path ) ) { |
| 1637 | return new WP_Error( 'wpmc_repair_file_missing', __( 'The file to repair is missing or unsafe.', 'media-cleaner' ) ); |
| 1638 | } |
| 1639 | $relative_path = $this->clean_uploaded_filename( $full_path ); |
| 1640 | global $wpdb; |
| 1641 | $existing_id = (int) $wpdb->get_var( $wpdb->prepare( |
| 1642 | "SELECT post_id FROM $wpdb->postmeta WHERE meta_key = '_wp_attached_file' AND meta_value = %s LIMIT 1", |
| 1643 | $relative_path |
| 1644 | ) ); |
| 1645 | $filetype = wp_check_filetype( basename( $full_path ), null ); |
| 1646 | $wp_upload_dir = wp_upload_dir(); |
| 1647 | $attachment = array( |
| 1648 | 'guid' => trailingslashit( $wp_upload_dir['baseurl'] ) . str_replace( '%2F', '/', rawurlencode( $relative_path ) ), |
| 1649 | 'post_mime_type' => $filetype['type'], |
| 1650 | 'post_title' => preg_replace( '/\.[^.]+$/', '', basename( $full_path ) ), |
| 1651 | 'post_content' => '', |
| 1652 | 'post_status' => 'inherit' |
| 1653 | ); |
| 1654 | |
| 1655 | $attach_id = $existing_id ?: wp_insert_attachment( $attachment, $full_path, 0, true ); |
| 1656 | if ( is_wp_error( $attach_id ) || !$attach_id ) { |
| 1657 | return is_wp_error( $attach_id ) ? $attach_id : new WP_Error( 'wpmc_repair_insert_failed', __( 'WordPress could not create the attachment record.', 'media-cleaner' ) ); |
| 1658 | } |
| 1659 | |
| 1660 | if ( wp_attachment_is_image( $attach_id ) ) { |
| 1661 | require_once( ABSPATH . 'wp-admin/includes/image.php' ); |
| 1662 | $attach_data = wp_generate_attachment_metadata( $attach_id, $full_path ); |
| 1663 | if ( is_wp_error( $attach_data ) || !is_array( $attach_data ) || !wp_update_attachment_metadata( $attach_id, $attach_data ) ) { |
| 1664 | return is_wp_error( $attach_data ) ? $attach_data : new WP_Error( 'wpmc_repair_metadata_failed', __( 'WordPress could not generate attachment metadata. The original file was left untouched.', 'media-cleaner' ) ); |
| 1665 | } |
| 1666 | } |
| 1667 | |
| 1668 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 1669 | $deleted = $wpdb->query( $wpdb->prepare( "DELETE FROM $table_name WHERE run_id = %d AND (id = %d OR parentId = %d)", $this->get_run_id(), $id, $id ) ); |
| 1670 | if ( $deleted === false ) { |
| 1671 | return new WP_Error( 'wpmc_repair_result_update_failed', __( 'The attachment was repaired, but Media Cleaner could not update its result record.', 'media-cleaner' ) ); |
| 1672 | } |
| 1673 | $this->log( "� |
| 1674 | Repaired {$repair->path}." ); |
| 1675 | return true; |
| 1676 | } |
| 1677 | |
| 1678 | function ignore( $id, $ignore ) { |
| 1679 | $staged = $this->results_staged_error(); |
| 1680 | if ( $staged ) return $staged; |
| 1681 | global $wpdb; |
| 1682 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 1683 | $issue = $this->get_issue( $id ); |
| 1684 | |
| 1685 | if ( empty( $issue ) ) { |
| 1686 | $this->log( "🚫 Issue #{$id} does not exist. Cannot ignore this." ); |
| 1687 | return false; |
| 1688 | } |
| 1689 | |
| 1690 | if ( !$ignore ) { |
| 1691 | $updated = $wpdb->query( $wpdb->prepare( "UPDATE $table_name SET ignored = 0 WHERE id = %d", $id ) ); |
| 1692 | } |
| 1693 | else { |
| 1694 | // If it is in trash, recover it |
| 1695 | if ( $issue->deleted ) { |
| 1696 | $recovered = $this->recover( $id ); |
| 1697 | if ( is_wp_error( $recovered ) || $recovered !== true ) { |
| 1698 | return is_wp_error( $recovered ) ? $recovered : new WP_Error( 'wpmc_ignore_recovery_failed', __( 'The item could not be recovered before it was ignored.', 'media-cleaner' ) ); |
| 1699 | } |
| 1700 | } |
| 1701 | $updated = $wpdb->query( $wpdb->prepare( "UPDATE $table_name SET ignored = 1 WHERE id = %d", $id ) ); |
| 1702 | } |
| 1703 | return $updated === false ? new WP_Error( 'wpmc_ignore_database_failed', __( 'Media Cleaner could not update the ignored state.', 'media-cleaner' ) ) : true; |
| 1704 | } |
| 1705 | |
| 1706 | function clean_dir( $dir ) { |
| 1707 | $root = realpath( $this->upload_path ); |
| 1708 | $current = realpath( $dir ); |
| 1709 | if ( !$root || !$current || is_link( $dir ) ) return; |
| 1710 | $root = untrailingslashit( wp_normalize_path( $root ) ); |
| 1711 | $current = untrailingslashit( wp_normalize_path( $current ) ); |
| 1712 | if ( $current === $root || strpos( $current, trailingslashit( $root ) ) !== 0 ) return; |
| 1713 | try { |
| 1714 | $is_empty = !( new FilesystemIterator( $current, FilesystemIterator::SKIP_DOTS ) )->valid(); |
| 1715 | } |
| 1716 | catch ( Throwable $e ) { |
| 1717 | return; |
| 1718 | } |
| 1719 | if ( $is_empty && @rmdir( $current ) ) { |
| 1720 | $this->clean_dir( dirname( $current ) ); |
| 1721 | } |
| 1722 | } |
| 1723 | |
| 1724 | function get_issue( $id ) { |
| 1725 | global $wpdb; |
| 1726 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 1727 | $run_id = $this->get_run_id(); |
| 1728 | $issue = $wpdb->get_row( $wpdb->prepare( "SELECT * FROM $table_name WHERE id = %d AND run_id = %d", $id, $run_id ), OBJECT ); |
| 1729 | if ( empty( $issue ) ) { |
| 1730 | return false; |
| 1731 | } |
| 1732 | $issue->id = (int)$issue->id; |
| 1733 | $issue->postId = (int)$issue->postId; |
| 1734 | $issue->type = (int)$issue->type; |
| 1735 | $issue->deleted = (int)$issue->deleted; |
| 1736 | $issue->ignored = (int)$issue->ignored; |
| 1737 | $issue->path = stripslashes( $issue->path ); |
| 1738 | return $issue; |
| 1739 | } |
| 1740 | |
| 1741 | function get_repair( $id ) { |
| 1742 | global $wpdb; |
| 1743 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 1744 | $run_id = $this->get_run_id(); |
| 1745 | $repair = $wpdb->get_row( $wpdb->prepare( "SELECT |
| 1746 | main.id AS id, |
| 1747 | main.path AS path, |
| 1748 | GROUP_CONCAT(child.id) AS child_ids |
| 1749 | FROM |
| 1750 | $table_name AS main |
| 1751 | LEFT JOIN |
| 1752 | $table_name AS child ON main.id = child.parentId AND child.run_id = main.run_id |
| 1753 | WHERE main.id = %d AND main.run_id = %d |
| 1754 | GROUP BY main.id, main.path", $id, $run_id |
| 1755 | ), OBJECT ); |
| 1756 | if ( empty( $repair ) ) { |
| 1757 | return false; |
| 1758 | } |
| 1759 | |
| 1760 | // If $repair->path is null or empty return false |
| 1761 | if ( empty( $repair->path ) ) { |
| 1762 | $this->log( "🚫 Repair #{$id} does not have a path. Cannot repair this." ); |
| 1763 | return false; |
| 1764 | } |
| 1765 | |
| 1766 | |
| 1767 | $repair->id = (int)$repair->id; |
| 1768 | $regex = "^(.*)(\\s\\(\\+.*)$"; |
| 1769 | $repair->path = preg_replace( '/' . $regex . '/i', '$1', stripslashes( $repair->path ) ); |
| 1770 | $repair->child_ids = $repair->child_ids ? explode( ',', $repair->child_ids ) : []; |
| 1771 | return $repair; |
| 1772 | } |
| 1773 | |
| 1774 | function get_issues_to_repair( $order_by = 'id', $order = 'asc', $search = '', $skip = 0, $limit = 10 ) { |
| 1775 | global $wpdb; |
| 1776 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 1777 | $run_id = $this->get_run_id(); |
| 1778 | |
| 1779 | $search_clause = ''; |
| 1780 | if ( !empty( $search ) ) { |
| 1781 | $search_clause = $wpdb->prepare("AND main.path LIKE %s", ( '%' . $search . '%' )); |
| 1782 | } |
| 1783 | |
| 1784 | $order_clause = 'ORDER BY main.id ASC'; |
| 1785 | if ( $order_by === 'path' ) { |
| 1786 | $order_clause = 'ORDER BY main.path ' . ( $order === 'asc' ? 'ASC' : 'DESC' ); |
| 1787 | } |
| 1788 | else if ( $order_by === 'issue' ) { |
| 1789 | $order_clause = 'ORDER BY main.issue ' . ( $order === 'asc' ? 'ASC' : 'DESC' ); |
| 1790 | } |
| 1791 | else if ( $order_by === 'size' ) { |
| 1792 | $order_clause = 'ORDER BY main.size ' . ( $order === 'asc' ? 'ASC' : 'DESC' ); |
| 1793 | } |
| 1794 | |
| 1795 | $result = $wpdb->get_results( $wpdb->prepare( "SELECT |
| 1796 | main.id AS id, |
| 1797 | main.path AS path, |
| 1798 | GROUP_CONCAT(child.id) AS child_ids, |
| 1799 | GROUP_CONCAT(child.path) AS child_paths, |
| 1800 | main.type AS type, |
| 1801 | main.postId AS postId, |
| 1802 | main.size AS size, |
| 1803 | main.ignored AS ignored, |
| 1804 | main.deleted AS deleted, |
| 1805 | main.issue AS issue |
| 1806 | FROM |
| 1807 | $table_name AS main |
| 1808 | LEFT JOIN |
| 1809 | $table_name AS child ON main.id = child.parentId AND child.run_id = main.run_id |
| 1810 | WHERE |
| 1811 | main.run_id = %d AND main.path IS NOT NULL AND main.parentId IS NULL |
| 1812 | AND main.deleted = 0 AND main.ignored = 0 |
| 1813 | AND main.type = 0 |
| 1814 | $search_clause |
| 1815 | GROUP BY main.id |
| 1816 | $order_clause |
| 1817 | LIMIT %d, %d; |
| 1818 | ", $run_id, $skip, $limit ) ); |
| 1819 | |
| 1820 | return $result; |
| 1821 | } |
| 1822 | |
| 1823 | function get_repair_ids ( $search = '', $cursor = 0, $limit = 100 ) { |
| 1824 | global $wpdb; |
| 1825 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 1826 | $run_id = $this->get_run_id(); |
| 1827 | $cursor = absint( $cursor ); |
| 1828 | $limit = max( 1, min( 100, absint( $limit ) ) ); |
| 1829 | |
| 1830 | $search_clause = ''; |
| 1831 | if ( !empty( $search ) ) { |
| 1832 | $search_clause = $wpdb->prepare( "AND main.path LIKE %s", '%' . $wpdb->esc_like( $search ) . '%' ); |
| 1833 | } |
| 1834 | |
| 1835 | return $wpdb->get_col( $wpdb->prepare( "SELECT main.id |
| 1836 | FROM $table_name AS main |
| 1837 | WHERE |
| 1838 | main.run_id = %d |
| 1839 | AND main.id > %d |
| 1840 | AND main.path IS NOT NULL |
| 1841 | AND main.parentId IS NULL |
| 1842 | AND main.deleted = 0 AND main.ignored = 0 AND main.type = 0 |
| 1843 | $search_clause |
| 1844 | ORDER BY main.id ASC |
| 1845 | LIMIT %d", $run_id, $cursor, $limit ) |
| 1846 | ); |
| 1847 | } |
| 1848 | |
| 1849 | function get_stats_of_issues_to_repair( $search = '' ) { |
| 1850 | global $wpdb; |
| 1851 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 1852 | $run_id = $this->get_run_id(); |
| 1853 | |
| 1854 | $search_clause = ''; |
| 1855 | if ( !empty( $search ) ) { |
| 1856 | $search_clause = $wpdb->prepare("AND main.path LIKE %s", ( '%' . $search . '%' )); |
| 1857 | } |
| 1858 | |
| 1859 | return $wpdb->get_row( $wpdb->prepare( "SELECT |
| 1860 | COUNT(id) AS entries, |
| 1861 | SUM(size) AS size |
| 1862 | FROM ( |
| 1863 | SELECT |
| 1864 | COUNT(DISTINCT main.id) as id, |
| 1865 | main.size as size |
| 1866 | FROM |
| 1867 | $table_name AS main |
| 1868 | LEFT JOIN |
| 1869 | $table_name AS child ON main.id = child.parentId AND child.run_id = main.run_id |
| 1870 | WHERE |
| 1871 | main.run_id = %d AND main.path IS NOT NULL AND main.parentId IS NULL AND main.deleted = 0 AND main.ignored = 0 AND main.type = 0 |
| 1872 | $search_clause |
| 1873 | GROUP BY main.id |
| 1874 | ) t; |
| 1875 | ", $run_id ) ); |
| 1876 | } |
| 1877 | |
| 1878 | function get_count_of_issues_to_repair( $search ) { |
| 1879 | $stats = $this->get_stats_of_issues_to_repair( $search ); |
| 1880 | return $stats->entries; |
| 1881 | } |
| 1882 | |
| 1883 | function delete( $id, $operation_manifest = array() ) { |
| 1884 | global $wpdb; |
| 1885 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 1886 | $issue = $this->get_issue( $id ); |
| 1887 | |
| 1888 | if ( empty( $issue ) ) { |
| 1889 | return new WP_Error( 'wpmc_issue_missing', __( 'The selected Media Cleaner result no longer exists.', 'media-cleaner' ) ); |
| 1890 | } |
| 1891 | if ( empty( $operation_manifest['identity_validated'] ) ) { |
| 1892 | $identity = $this->validate_issue_manifest( $issue ); |
| 1893 | if ( is_wp_error( $identity ) ) return $identity; |
| 1894 | } |
| 1895 | |
| 1896 | $regex = "^(.*)(\\s\\(\\+.*)$"; |
| 1897 | $issue->path = preg_replace( '/' . $regex . '/i', '$1', $issue->path ); // remove " (+ 6 files)" from path |
| 1898 | $skip_trash = $this->get_option( 'skip_trash' ); |
| 1899 | $was_deleted = isset( $operation_manifest['initial_deleted'] ) ? (bool) $operation_manifest['initial_deleted'] : $issue->deleted === 1; |
| 1900 | |
| 1901 | $staged = $this->results_staged_error(); |
| 1902 | if ( $staged ) return $staged; |
| 1903 | |
| 1904 | // Trashing something new is only as safe as the analysis behind it, so it needs |
| 1905 | // results from a finished scan of this version. Emptying the trash does not: |
| 1906 | // that file was already set aside on purpose, and refusing here would strand |
| 1907 | // it. This is the one place deletion is allowed or refused, so the REST API, |
| 1908 | // MCP and WP-CLI all get the same answer. The capability is checked by the |
| 1909 | // callers, since WP-CLI has no current user. |
| 1910 | if ( !$was_deleted && ( !$this->runs || !$this->runs->cleanup_allowed() ) ) { |
| 1911 | return new WP_Error( 'wpmc_cleanup_needs_scan', |
| 1912 | __( 'Media Cleaner needs the results of a completed scan from this version before it can delete anything. Run a scan first. Your trash is untouched and can still be recovered or emptied.', 'media-cleaner' ) ); |
| 1913 | } |
| 1914 | |
| 1915 | if ( $issue->type === 0 ) { |
| 1916 | if ( $was_deleted ) { |
| 1917 | $trash_path = $this->resolve_trash_path( $issue->path ); |
| 1918 | if ( is_wp_error( $trash_path ) ) return $trash_path; |
| 1919 | if ( file_exists( $trash_path ) && ( is_dir( $trash_path ) || is_link( $trash_path ) || !@unlink( $trash_path ) ) ) { |
| 1920 | return new WP_Error( 'wpmc_permanent_delete_failed', __( 'The file could not be permanently removed from Media Cleaner trash.', 'media-cleaner' ) ); |
| 1921 | } |
| 1922 | $deleted = $wpdb->query( $wpdb->prepare( "DELETE FROM $table_name WHERE id = %d", $id ) ); |
| 1923 | return $deleted === false ? new WP_Error( 'wpmc_delete_database_failed', __( 'The file was removed, but its Media Cleaner record could not be deleted.', 'media-cleaner' ) ) : true; |
| 1924 | } |
| 1925 | else if ( $skip_trash ) { |
| 1926 | $original_path = $this->resolve_upload_path( $issue->path ); |
| 1927 | if ( is_wp_error( $original_path ) ) return $original_path; |
| 1928 | if ( file_exists( $original_path ) && ( is_dir( $original_path ) || is_link( $original_path ) || !@unlink( $original_path ) ) ) { |
| 1929 | return new WP_Error( 'wpmc_permanent_delete_failed', __( 'The file could not be permanently deleted.', 'media-cleaner' ) ); |
| 1930 | } |
| 1931 | $this->clean_dir( dirname( $original_path ) ); |
| 1932 | $deleted = $wpdb->query( $wpdb->prepare( "DELETE FROM $table_name WHERE id = %d", $id ) ); |
| 1933 | return $deleted === false ? new WP_Error( 'wpmc_delete_database_failed', __( 'The file was removed, but its Media Cleaner record could not be deleted.', 'media-cleaner' ) ) : true; |
| 1934 | } |
| 1935 | $did_move = false; |
| 1936 | $trashed = $this->trash_file( $issue->path, $did_move ); |
| 1937 | if ( is_wp_error( $trashed ) ) return $trashed; |
| 1938 | $updated = $wpdb->query( $wpdb->prepare( "UPDATE $table_name SET deleted = 1, ignored = 0, time = NOW() WHERE id = %d", $id ) ); |
| 1939 | if ( $updated === false ) { |
| 1940 | if ( $did_move ) $this->recover_file( $issue->path ); |
| 1941 | return new WP_Error( 'wpmc_trash_database_failed', __( 'The file was restored because its Media Cleaner state could not be updated.', 'media-cleaner' ) ); |
| 1942 | } |
| 1943 | return true; |
| 1944 | } |
| 1945 | |
| 1946 | if ( $issue->type === 1 ) { |
| 1947 | if ( $was_deleted || $skip_trash ) { |
| 1948 | if ( $issue->deleted === 1 ) { |
| 1949 | $recovered = $this->recover( $id ); |
| 1950 | if ( is_wp_error( $recovered ) || $recovered !== true ) { |
| 1951 | return is_wp_error( $recovered ) ? $recovered : new WP_Error( 'wpmc_media_recovery_failed', __( 'The attachment could not be restored before permanent deletion.', 'media-cleaner' ) ); |
| 1952 | } |
| 1953 | } |
| 1954 | $deleted_attachment = get_post( $issue->postId ) ? wp_delete_attachment( $issue->postId, true ) : true; |
| 1955 | if ( !$deleted_attachment ) { |
| 1956 | return new WP_Error( 'wpmc_attachment_delete_failed', __( 'WordPress could not permanently delete the attachment.', 'media-cleaner' ) ); |
| 1957 | } |
| 1958 | $deleted = $wpdb->query( $wpdb->prepare( "DELETE FROM $table_name WHERE id = %d", $id ) ); |
| 1959 | return $deleted === false ? new WP_Error( 'wpmc_delete_database_failed', __( 'The attachment was removed, but its Media Cleaner record could not be deleted.', 'media-cleaner' ) ) : true; |
| 1960 | } |
| 1961 | |
| 1962 | $paths = $this->get_paths_from_attachment( $issue->postId ); |
| 1963 | $file_manifest = json_decode( (string) $issue->manifest, true ); |
| 1964 | $file_manifest = is_array( $file_manifest ) ? $file_manifest : array(); |
| 1965 | $trashed_paths = array(); |
| 1966 | foreach ( $paths as $path ) { |
| 1967 | if ( array_key_exists( $path, $file_manifest ) && $file_manifest[ $path ] === null ) continue; |
| 1968 | $did_move = false; |
| 1969 | $result = $this->trash_file( $path, $did_move ); |
| 1970 | if ( is_wp_error( $result ) ) { |
| 1971 | foreach ( array_reverse( $trashed_paths ) as $trashed_path ) { |
| 1972 | $this->recover_file( $trashed_path ); |
| 1973 | } |
| 1974 | return $result; |
| 1975 | } |
| 1976 | if ( $did_move ) $trashed_paths[] = $path; |
| 1977 | } |
| 1978 | $previous_post_type = get_post_type( $issue->postId ); |
| 1979 | $post_result = wp_update_post( array( 'ID' => $issue->postId, 'post_type' => 'wmpc-trash' ), true ); |
| 1980 | if ( is_wp_error( $post_result ) || !$post_result ) { |
| 1981 | foreach ( array_reverse( $trashed_paths ) as $trashed_path ) { |
| 1982 | $this->recover_file( $trashed_path ); |
| 1983 | } |
| 1984 | return is_wp_error( $post_result ) ? $post_result : new WP_Error( 'wpmc_attachment_trash_failed', __( 'The attachment record could not be moved to Media Cleaner trash.', 'media-cleaner' ) ); |
| 1985 | } |
| 1986 | $updated = $wpdb->query( $wpdb->prepare( "UPDATE $table_name SET deleted = 1, ignored = 0, time = NOW() WHERE id = %d", $id ) ); |
| 1987 | if ( $updated === false ) { |
| 1988 | if ( $previous_post_type && $previous_post_type !== 'wmpc-trash' ) { |
| 1989 | wp_update_post( array( 'ID' => $issue->postId, 'post_type' => $previous_post_type ) ); |
| 1990 | } |
| 1991 | foreach ( array_reverse( $trashed_paths ) as $trashed_path ) { |
| 1992 | $this->recover_file( $trashed_path ); |
| 1993 | } |
| 1994 | return new WP_Error( 'wpmc_trash_database_failed', __( 'The attachment was restored because its Media Cleaner state could not be updated.', 'media-cleaner' ) ); |
| 1995 | } |
| 1996 | return true; |
| 1997 | } |
| 1998 | return new WP_Error( 'wpmc_issue_type_invalid', __( 'The selected Media Cleaner result has an unsupported type.', 'media-cleaner' ) ); |
| 1999 | } |
| 2000 | |
| 2001 | function force_trash( $initialize = false, $limit = 100 ) { |
| 2002 | global $wpdb; |
| 2003 | $staged = $this->results_staged_error(); |
| 2004 | if ( $staged ) return $staged; |
| 2005 | $run_id = $this->get_run_id(); |
| 2006 | $table_name = $wpdb->prefix . 'mclean_scan'; |
| 2007 | $tracked_items = (int) $wpdb->get_var( $wpdb->prepare( "SELECT COUNT(*) FROM $table_name WHERE run_id = %d AND deleted = 1", $run_id ) ); |
| 2008 | if ( $tracked_items > 0 ) { |
| 2009 | return new WP_Error( |
| 2010 | 'wpmc_force_trash_has_tracked_items', |
| 2011 | __( 'Retry the tracked trash items individually before removing untracked quarantine files.', 'media-cleaner' ), |
| 2012 | array( 'status' => 409 ) |
| 2013 | ); |
| 2014 | } |
| 2015 | $phase = 'cleanuptrash'; |
| 2016 | $limit = max( 1, min( 100, (int) $limit ) ); |
| 2017 | $trash = $this->ensure_trash_directory(); |
| 2018 | if ( is_wp_error( $trash ) ) return $trash; |
| 2019 | |
| 2020 | if ( $initialize ) { |
| 2021 | if ( !$this->runs->clear_work( $run_id, $phase ) || !$this->runs->enqueue_work( $run_id, $phase, 'directory', '' ) ) { |
| 2022 | return new WP_Error( 'wpmc_trash_queue_failed', __( 'Media Cleaner could not initialize the trash cleanup queue.', 'media-cleaner' ) ); |
| 2023 | } |
| 2024 | } |
| 2025 | if ( $this->runs->failed_work_count( $run_id, $phase ) > 0 ) { |
| 2026 | return new WP_Error( 'wpmc_trash_cleanup_failed', __( 'A trash item could not be removed. Result records were preserved.', 'media-cleaner' ) ); |
| 2027 | } |
| 2028 | |
| 2029 | $work = $this->runs->next_work( $run_id, $phase ); |
| 2030 | if ( !$work ) { |
| 2031 | $deleted = $wpdb->query( $wpdb->prepare( "DELETE FROM $table_name WHERE run_id = %d AND deleted = 1", $run_id ) ); |
| 2032 | if ( $deleted === false ) { |
| 2033 | return new WP_Error( 'wpmc_trash_database_failed', __( 'Trash files were removed, but Media Cleaner could not update its result records.', 'media-cleaner' ) ); |
| 2034 | } |
| 2035 | $this->runs->clear_work( $run_id, $phase ); |
| 2036 | return array( 'finished' => true, 'processed' => 0, 'pending' => 0 ); |
| 2037 | } |
| 2038 | |
| 2039 | $relative = (string) $work->target_key; |
| 2040 | $directory = $this->resolve_trash_path( $relative, true ); |
| 2041 | if ( is_wp_error( $directory ) || !is_dir( $directory ) || is_link( $directory ) ) { |
| 2042 | $error = is_wp_error( $directory ) ? $directory : new WP_Error( 'wpmc_trash_directory_unsafe', __( 'A queued trash directory is missing or unsafe.', 'media-cleaner' ) ); |
| 2043 | $this->runs->update_work( $work->id, 'failed', 0, $error ); |
| 2044 | return $error; |
| 2045 | } |
| 2046 | |
| 2047 | $processed = 0; |
| 2048 | try { |
| 2049 | $iterator = new FilesystemIterator( $directory, FilesystemIterator::SKIP_DOTS ); |
| 2050 | foreach ( $iterator as $entry ) { |
| 2051 | if ( $entry->isLink() ) { |
| 2052 | throw new RuntimeException( __( 'Media Cleaner will not remove a symbolic link from trash.', 'media-cleaner' ) ); |
| 2053 | } |
| 2054 | if ( $entry->isDir() ) { |
| 2055 | $child = ltrim( $relative . '/' . $entry->getFilename(), '/' ); |
| 2056 | if ( !$this->runs->enqueue_work( $run_id, $phase, 'directory', $child ) ) { |
| 2057 | throw new RuntimeException( __( 'A trash subdirectory could not be queued.', 'media-cleaner' ) ); |
| 2058 | } |
| 2059 | if ( !$this->runs->update_work( $work->id, 'waiting', 0 ) ) return new WP_Error( 'wpmc_trash_queue_failed', __( 'Media Cleaner could not checkpoint a trash directory.', 'media-cleaner' ) ); |
| 2060 | return array( 'finished' => false, 'processed' => $processed, 'pending' => $this->runs->pending_work_count( $run_id, $phase ) ); |
| 2061 | } |
| 2062 | if ( !@unlink( $entry->getPathname() ) ) { |
| 2063 | throw new RuntimeException( sprintf( __( 'The trash file %s could not be removed.', 'media-cleaner' ), $entry->getFilename() ) ); |
| 2064 | } |
| 2065 | $processed++; |
| 2066 | if ( $processed >= $limit ) break; |
| 2067 | } |
| 2068 | } |
| 2069 | catch ( Throwable $e ) { |
| 2070 | $error = new WP_Error( 'wpmc_trash_item_delete_failed', $e->getMessage() ); |
| 2071 | $this->runs->update_work( $work->id, 'failed', 0, $error ); |
| 2072 | return $error; |
| 2073 | } |
| 2074 | |
| 2075 | $has_entries = false; |
| 2076 | try { |
| 2077 | $check = new FilesystemIterator( $directory, FilesystemIterator::SKIP_DOTS ); |
| 2078 | $has_entries = $check->valid(); |
| 2079 | } |
| 2080 | catch ( UnexpectedValueException $e ) { |
| 2081 | $has_entries = true; |
| 2082 | } |
| 2083 | if ( $has_entries ) { |
| 2084 | if ( !$this->runs->update_work( $work->id, 'running', 0 ) ) return new WP_Error( 'wpmc_trash_queue_failed', __( 'Media Cleaner could not checkpoint a trash batch.', 'media-cleaner' ) ); |
| 2085 | } |
| 2086 | else { |
| 2087 | if ( $relative !== '' && !@rmdir( $directory ) ) { |
| 2088 | $error = new WP_Error( 'wpmc_trash_directory_delete_failed', __( 'An empty trash directory could not be removed.', 'media-cleaner' ) ); |
| 2089 | $this->runs->update_work( $work->id, 'failed', 0, $error ); |
| 2090 | return $error; |
| 2091 | } |
| 2092 | if ( !$this->runs->update_work( $work->id, 'complete', 0 ) ) return new WP_Error( 'wpmc_trash_queue_failed', __( 'Media Cleaner could not complete a trash directory checkpoint.', 'media-cleaner' ) ); |
| 2093 | $parent = $relative === '' ? '' : dirname( $relative ); |
| 2094 | if ( !$this->runs->wake_work( $run_id, $phase, $parent === '.' ? '' : $parent ) ) return new WP_Error( 'wpmc_trash_queue_failed', __( 'Media Cleaner could not resume the parent trash directory.', 'media-cleaner' ) ); |
| 2095 | } |
| 2096 | |
| 2097 | return array( |
| 2098 | 'finished' => false, |
| 2099 | 'processed' => $processed, |
| 2100 | 'pending' => $this->runs->pending_work_count( $run_id, $phase ), |
| 2101 | ); |
| 2102 | } |
| 2103 | |
| 2104 | // A last resort for trash items that can no longer be emptied the normal way: |
| 2105 | // their file has disappeared, changed, or become unsafe since the scan, so |
| 2106 | // validate_issue_manifest() refuses them and the row stays stuck in the trash |
| 2107 | // with the "run a new scan" error. This removes exactly those broken rows (and |
| 2108 | // any physical file still sitting in quarantine), and leaves healthy, still |
| 2109 | // recoverable trash untouched. It is safe precisely because it only acts on |
| 2110 | // items whose backing file is already gone or unverifiable. Returns the count. |
| 2111 | function force_clean_trash() { |
| 2112 | $staged = $this->results_staged_error(); |
| 2113 | if ( $staged ) return $staged; |
| 2114 | global $wpdb; |
| 2115 | $table_name = $wpdb->prefix . 'mclean_scan'; |
| 2116 | $run_id = $this->get_run_id(); |
| 2117 | $rows = $wpdb->get_results( $wpdb->prepare( "SELECT * FROM $table_name WHERE run_id = %d AND deleted = 1", $run_id ) ); |
| 2118 | if ( $rows === null ) { |
| 2119 | return new WP_Error( 'wpmc_force_clean_trash_failed', __( 'Media Cleaner could not read the trash records.', 'media-cleaner' ) ); |
| 2120 | } |
| 2121 | $removed = 0; |
| 2122 | foreach ( $rows as $issue ) { |
| 2123 | // Only the items that cannot be cleaned normally: a healthy trash item |
| 2124 | // validates fine and is left recoverable. |
| 2125 | if ( !is_wp_error( $this->validate_issue_manifest( $issue ) ) ) continue; |
| 2126 | |
| 2127 | // Best effort: drop any file still physically in quarantine, then the row. |
| 2128 | // Directories and symlinks are never removed, and a missing or unsafe path |
| 2129 | // is skipped rather than fatal — the point is to unstick the record. |
| 2130 | $manifest = json_decode( (string) $issue->manifest, true ); |
| 2131 | $relatives = is_array( $manifest ) ? array_keys( $manifest ) : array(); |
| 2132 | if ( empty( $relatives ) && (int) $issue->type === 0 ) { |
| 2133 | $relatives[] = preg_replace( '/\s\(\+.*$/', '', (string) $issue->path ); |
| 2134 | } |
| 2135 | foreach ( $relatives as $relative ) { |
| 2136 | $trash_path = $this->resolve_trash_path( $relative ); |
| 2137 | if ( is_wp_error( $trash_path ) ) continue; |
| 2138 | if ( file_exists( $trash_path ) && !is_dir( $trash_path ) && !is_link( $trash_path ) ) { |
| 2139 | @unlink( $trash_path ); |
| 2140 | } |
| 2141 | } |
| 2142 | if ( $wpdb->query( $wpdb->prepare( "DELETE FROM $table_name WHERE id = %d", $issue->id ) ) !== false ) { |
| 2143 | $removed++; |
| 2144 | } |
| 2145 | } |
| 2146 | return $removed; |
| 2147 | } |
| 2148 | |
| 2149 | /** |
| 2150 | * |
| 2151 | * SCANNING / RESET |
| 2152 | * |
| 2153 | */ |
| 2154 | |
| 2155 | function add_reference_url( $urlOrUrls, $type, $origin = null, $extra = null ) { |
| 2156 | $urlOrUrls = !is_array( $urlOrUrls ) ? array( $urlOrUrls ) : $urlOrUrls; |
| 2157 | foreach ( $urlOrUrls as $url ) { |
| 2158 | // With files, we need both filename without resolution and filename with resolution, it's important |
| 2159 | // to make sure the original file is not deleted if a size exists for it. |
| 2160 | // With media, all URLs should be without resolution to make sure it matches Media. |
| 2161 | $no_res_url = $this->clean_url_from_resolution( $url ); |
| 2162 | |
| 2163 | $this->add_reference( null, $url, $type, $origin, $extra ); |
| 2164 | $this->add_reference( 0, $no_res_url, $type, $origin, $extra ); |
| 2165 | |
| 2166 | if ( $this->multilingual ) { |
| 2167 | if ( $this->current_method == 'media' ) { |
| 2168 | if ( !array_key_exists( $no_res_url, $this->url_id_cache ) ) { |
| 2169 | $this->url_id_cache[ $no_res_url ] = $this->get_id_from_clean_url( $no_res_url ); |
| 2170 | } |
| 2171 | $id = $this->url_id_cache[ $no_res_url ]; |
| 2172 | if( $id ) $this->add_reference_id( $id, $type, $origin, $extra ); |
| 2173 | } |
| 2174 | } |
| 2175 | } |
| 2176 | } |
| 2177 | |
| 2178 | /** |
| 2179 | * Add an issue to the mclean_scan table. |
| 2180 | * |
| 2181 | * @param string $path The path to the file (relative to uploads). |
| 2182 | * @param string $issue The issue code/type. |
| 2183 | * @param int|null $postId Optional post ID related to the issue. |
| 2184 | */ |
| 2185 | function add_issue( $path, $issue, $postId = null ) { |
| 2186 | global $wpdb; |
| 2187 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 2188 | $clean_path = $this->clean_uploaded_filename( $path ); |
| 2189 | if ( $clean_path === '' ) throw new RuntimeException( __( 'Media Cleaner received an invalid issue path.', 'media-cleaner' ) ); |
| 2190 | $path_hash = hash( 'sha256', $clean_path ); |
| 2191 | $run_id = $this->get_run_id( true ); |
| 2192 | if ( $run_id < 1 ) { |
| 2193 | throw new RuntimeException( __( 'An active scan run is required before adding issues.', 'media-cleaner' ) ); |
| 2194 | } |
| 2195 | $filepath = $this->resolve_upload_path( $clean_path ); |
| 2196 | if ( is_wp_error( $filepath ) ) throw new RuntimeException( $filepath->get_error_message() ); |
| 2197 | $filesize = file_exists( $filepath ) ? filesize( $filepath ) : 0; |
| 2198 | $manifest = $this->build_file_manifest( array( $clean_path ) ); |
| 2199 | if ( is_wp_error( $manifest ) ) throw new RuntimeException( $manifest->get_error_message() ); |
| 2200 | |
| 2201 | // Check if this issue already exists |
| 2202 | $existing = $wpdb->get_var( $wpdb->prepare( |
| 2203 | "SELECT id FROM $table_name WHERE run_id = %d AND path_hash = %s AND path = %s AND issue = %s LIMIT 1", |
| 2204 | $run_id, $path_hash, $clean_path, $issue |
| 2205 | ) ); |
| 2206 | |
| 2207 | if ( $existing ) { |
| 2208 | return; // Issue already exists |
| 2209 | } |
| 2210 | |
| 2211 | // Find potential parent |
| 2212 | $potentialParentPath = $this->clean_url_from_resolution( $clean_path ); |
| 2213 | $parentId = $wpdb->get_var( $wpdb->prepare( "SELECT id FROM $table_name WHERE run_id = %d AND path_hash = %s AND path = %s LIMIT 1", $run_id, hash( 'sha256', $potentialParentPath ), $potentialParentPath ) ); |
| 2214 | $parentId = $parentId ? (int)$parentId : null; |
| 2215 | |
| 2216 | $inserted = $wpdb->insert( $table_name, |
| 2217 | array( |
| 2218 | 'run_id' => $run_id, |
| 2219 | 'time' => current_time('mysql'), |
| 2220 | 'type' => 0, |
| 2221 | 'postId' => $postId, |
| 2222 | 'path' => $clean_path, |
| 2223 | 'path_hash' => $path_hash, |
| 2224 | 'manifest' => wp_json_encode( $manifest ), |
| 2225 | 'size' => $filesize, |
| 2226 | 'issue' => $issue, |
| 2227 | 'parentId' => $parentId |
| 2228 | ) |
| 2229 | ); |
| 2230 | if ( $inserted === false ) throw new RuntimeException( sprintf( __( 'Media Cleaner could not store an issue: %s', 'media-cleaner' ), $wpdb->last_error ) ); |
| 2231 | } |
| 2232 | |
| 2233 | function add_reference_id( $idOrIds, $type, $origin = null, $extra = null ) { |
| 2234 | $idOrIds = !is_array( $idOrIds ) ? array( $idOrIds ) : $idOrIds; |
| 2235 | foreach ( $idOrIds as $id ) { |
| 2236 | $this->add_reference( $id, "", $type, $origin ); |
| 2237 | if ( $this->multilingual ) { |
| 2238 | $translatedIds = $this->get_translated_media_ids( (int)$id ); |
| 2239 | |
| 2240 | // Test for WPML |
| 2241 | // if ( $id === '350') { |
| 2242 | // $translatedIds = $this->get_translated_media_ids( (int)$id ); |
| 2243 | // $count = count($translatedIds); |
| 2244 | // error_log( "${id} => ${count}" ); |
| 2245 | // } |
| 2246 | |
| 2247 | if ( !empty( $translatedIds ) ) { |
| 2248 | foreach ( $translatedIds as $translatedId ) { |
| 2249 | $this->add_reference( $translatedId, "", $type, $origin ); |
| 2250 | } |
| 2251 | } |
| 2252 | } |
| 2253 | } |
| 2254 | } |
| 2255 | |
| 2256 | |
| 2257 | // Returns the reference with the type, origin, related to a Media ID it is referenced |
| 2258 | public function get_reference_for_media_id( $id ) { |
| 2259 | global $wpdb; |
| 2260 | $table_name = $wpdb->prefix . "mclean_refs"; |
| 2261 | $run_id = $this->get_run_id(); |
| 2262 | $refs = $wpdb->get_results( $wpdb->prepare( "SELECT * FROM $table_name WHERE run_id = %d AND mediaId = %d", $run_id, $id ), OBJECT ); |
| 2263 | if ( empty( $refs ) ) { |
| 2264 | return false; |
| 2265 | } |
| 2266 | $ref = $refs[0]; |
| 2267 | $ref->id = (int)$ref->id; |
| 2268 | $ref->mediaId = (int)$ref->mediaId; |
| 2269 | $ref->originType = stripslashes( $ref->originType ); |
| 2270 | $ref->origin = stripslashes( $ref->origin ); |
| 2271 | $ref->parentId = empty( $ref->parentId ) ? null : (int)$ref->parentId; |
| 2272 | return $ref; |
| 2273 | } |
| 2274 | |
| 2275 | // Return the references related to a Post ID |
| 2276 | public function get_references_for_post_id( $id ) { |
| 2277 | global $wpdb; |
| 2278 | $table_name = $wpdb->prefix . "mclean_refs"; |
| 2279 | $run_id = $this->get_run_id(); |
| 2280 | $refs = $wpdb->get_results( $wpdb->prepare( "SELECT * FROM $table_name WHERE run_id = %d AND originType LIKE %s", $run_id, "%[$id]" ), OBJECT ); |
| 2281 | if ( empty( $refs ) ) { |
| 2282 | return []; |
| 2283 | } |
| 2284 | $fresh_refs = array(); |
| 2285 | foreach ( $refs as $ref ) { |
| 2286 | $mediaId = (int)$ref->mediaId > 0 ? (int)$ref->mediaId : null; |
| 2287 | if ( !$mediaId && !empty( $ref->mediaUrl ) ) { |
| 2288 | $mediaId = $this->find_media_id_from_file( $ref->mediaUrl, false ); |
| 2289 | $mediaId = !empty( $mediaId ) ? (int)$mediaId : null; |
| 2290 | } |
| 2291 | if ( !$mediaId ) { |
| 2292 | continue; |
| 2293 | } |
| 2294 | array_push( $fresh_refs, [ |
| 2295 | 'id' => (int)$ref->id, |
| 2296 | 'mediaId' => $mediaId, |
| 2297 | 'mediaUrl' => $ref->mediaUrl, |
| 2298 | 'originType' => $ref->originType, |
| 2299 | 'origin' => $ref->origin, |
| 2300 | 'parentId' => empty( $ref->parentId ) ? null : (int)$ref->parentId, |
| 2301 | ] ); |
| 2302 | } |
| 2303 | return $fresh_refs; |
| 2304 | } |
| 2305 | |
| 2306 | // The references are actually not being added directly in the DB, they are being pushed |
| 2307 | // into a cache ($this->refcache), then written to the database via write_references(). |
| 2308 | private function add_reference( $id, $url, $type, $origin = null, $extra = null ) { |
| 2309 | $type = substr( sanitize_text_field( (string) $type ), 0, 191 ); |
| 2310 | if ( $origin !== null && !is_scalar( $origin ) ) $origin = wp_json_encode( $origin ); |
| 2311 | // The origin is only a label for where the reference was found. If malformed content |
| 2312 | // makes it huge, trim it rather than dropping the reference: the reference marks a file |
| 2313 | // as used, and losing it could let a used file be deleted. |
| 2314 | if ( is_string( $origin ) && strlen( $origin ) > 8192 ) { |
| 2315 | $origin = substr( $origin, 0, 8192 ); |
| 2316 | } |
| 2317 | if ( !empty( $id ) ) { |
| 2318 | $this->queue_reference( array( 'id' => $id, 'url' => null, 'type' => $type, 'origin' => $origin ) ); |
| 2319 | } |
| 2320 | if ( !empty( $url ) ) { |
| 2321 | // A real media path is never this large, so an oversized "URL" is a parser |
| 2322 | // mis-extraction (a data: URI, a concatenated path blob, etc.). It can never match |
| 2323 | // a real file, so skip it instead of aborting the whole scan — the same treatment |
| 2324 | // the http/javascript URLs get just below. This is what stranded Divi users in 7.2. |
| 2325 | if ( is_string( $url ) && strlen( $url ) > 4096 ) { |
| 2326 | return; |
| 2327 | } |
| 2328 | // The URL shouldn't contain http, https, javascript at the beginning (and there are probably many more cases) |
| 2329 | // The URL must be cleaned before being passed as a reference. |
| 2330 | if ( substr( $url, 0, 5 ) === "http:" || substr( $url, 0, 6 ) === "https:" || substr( $url, 0, 11 ) === "javascript:" ) { |
| 2331 | return; |
| 2332 | } |
| 2333 | $this->queue_reference( array( 'id' => null, 'url' => $url, 'type' => $type, 'origin' => $origin ) ); |
| 2334 | } |
| 2335 | } |
| 2336 | |
| 2337 | private function queue_reference( $reference ) { |
| 2338 | $this->refcache[] = $reference; |
| 2339 | if ( count( $this->refcache ) >= 250 ) $this->write_references(); |
| 2340 | } |
| 2341 | |
| 2342 | function insert_references($entries) |
| 2343 | { |
| 2344 | global $wpdb; |
| 2345 | $table = $wpdb->prefix . "mclean_refs"; |
| 2346 | $run_id = $this->get_run_id( true ); |
| 2347 | if ( $run_id < 1 ) { |
| 2348 | throw new RuntimeException( __( 'An active scan run is required before adding references.', 'media-cleaner' ) ); |
| 2349 | } |
| 2350 | |
| 2351 | $refs_buffer = $this->get_option( 'refs_buffer' ); |
| 2352 | if ( empty( $refs_buffer ) || $refs_buffer < 1 ) { |
| 2353 | $refs_buffer = 500; |
| 2354 | } |
| 2355 | |
| 2356 | $values = array(); |
| 2357 | $place_holders = array(); |
| 2358 | $entry_count = 0; |
| 2359 | |
| 2360 | $entries = array_unique( $entries, SORT_REGULAR ); |
| 2361 | |
| 2362 | foreach ( $entries as $value ) { |
| 2363 | $origin = isset( $value['origin'] ) ? $value['origin'] : null; |
| 2364 | if ( !is_null( $value['id'] ) ) { |
| 2365 | // Media Reference |
| 2366 | $hash = md5( $value['id'] . '|' . $value['type'] . '|' . $origin ); |
| 2367 | array_push( $values, $run_id, $value['id'], $value['type'], $origin, $hash ); |
| 2368 | $place_holders[] = "('%d', '%d', NULL, NULL, '%s', '%s', NULL, '%s')"; |
| 2369 | |
| 2370 | if ( $this->debug_logs ) { |
| 2371 | $this->log( "＋ Media #{$value['id']} (as ID)" ); |
| 2372 | } |
| 2373 | $entry_count++; |
| 2374 | } |
| 2375 | else if ( !is_null( $value['url'] ) ) { |
| 2376 | // File Reference |
| 2377 | $parent_id = isset( $value['parentId'] ) ? (int) $value['parentId'] : null; |
| 2378 | $hash = md5( '|' . $value['url'] . '|' . $value['type'] . '|' . $origin . '|' . $parent_id ); |
| 2379 | $url_hash = hash( 'sha256', $value['url'] ); |
| 2380 | if ( $parent_id !== null ) { |
| 2381 | array_push( $values, $run_id, $value['url'], $url_hash, $value['type'], $origin, $parent_id, $hash ); |
| 2382 | $place_holders[] = "('%d', NULL, '%s', '%s', '%s', '%s', '%d', '%s')"; |
| 2383 | if ( $this->debug_logs ) { |
| 2384 | $this->log( "＋ {$value['url']} (as URL) (ParentID: {$value['parentId']})" ); |
| 2385 | } |
| 2386 | } |
| 2387 | else { |
| 2388 | array_push( $values, $run_id, $value['url'], $url_hash, $value['type'], $origin, $hash ); |
| 2389 | $place_holders[] = "('%d', NULL, '%s', '%s', '%s', '%s', NULL, '%s')"; |
| 2390 | if ( $this->debug_logs ) { |
| 2391 | $this->log( "＋ {$value['url']} (as URL)" ); |
| 2392 | } |
| 2393 | } |
| 2394 | $entry_count++; |
| 2395 | } |
| 2396 | |
| 2397 | // Flush to DB when buffer is full |
| 2398 | if ( $entry_count >= $refs_buffer ) { |
| 2399 | $this->log( "Flushing $entry_count references to the database..." ); |
| 2400 | $this->flush_references_to_db( $table, $values, $place_holders ); |
| 2401 | $values = array(); |
| 2402 | $place_holders = array(); |
| 2403 | $entry_count = 0; |
| 2404 | } |
| 2405 | } |
| 2406 | |
| 2407 | // Flush remaining entries |
| 2408 | if ( !empty( $values ) ) { |
| 2409 | $this->log( "Flushing remaining $entry_count references to the database..." ); |
| 2410 | $this->flush_references_to_db( $table, $values, $place_holders ); |
| 2411 | } |
| 2412 | } |
| 2413 | |
| 2414 | function flush_references_to_db( $table, $values, $place_holders ) { |
| 2415 | global $wpdb; |
| 2416 | if ( empty( $values ) ) { |
| 2417 | return; |
| 2418 | } |
| 2419 | $query = "INSERT IGNORE INTO $table (run_id, mediaId, mediaUrl, mediaUrl_hash, originType, origin, parentId, ref_hash) VALUES "; |
| 2420 | $query .= implode( ', ', $place_holders ); |
| 2421 | $prepared = $wpdb->prepare( "$query ", $values ); |
| 2422 | $result = $wpdb->query( $prepared ); |
| 2423 | if ( $result === false ) { |
| 2424 | throw new RuntimeException( sprintf( __( 'Could not store Media Cleaner references: %s', 'media-cleaner' ), $wpdb->last_error ) ); |
| 2425 | } |
| 2426 | } |
| 2427 | |
| 2428 | function reset_progress() { |
| 2429 | // Reset the progress by deleting the transient. |
| 2430 | delete_transient( $this->progress_key ); |
| 2431 | } |
| 2432 | |
| 2433 | function clear_step_progress() { |
| 2434 | // Clear step progress when scanning completes |
| 2435 | delete_transient( $this->progress_key ); |
| 2436 | } |
| 2437 | |
| 2438 | function save_progress( $step, $data = array() ) { |
| 2439 | if ( $this->run_id > 0 && $this->runs ) { |
| 2440 | // Target lists are intentionally not checkpointed. They can be enormous and |
| 2441 | // are rebuilt from deterministic server-side cursors when a scan resumes. |
| 2442 | unset( $data['targets'], $data['doneTargets'] ); |
| 2443 | $saved = $this->runs->checkpoint( $this->run_id, $step, $data ); |
| 2444 | if ( is_wp_error( $saved ) ) throw new RuntimeException( $saved->get_error_message() ); |
| 2445 | if ( $saved !== true ) throw new RuntimeException( __( 'Media Cleaner could not persist the scan checkpoint.', 'media-cleaner' ) ); |
| 2446 | return true; |
| 2447 | } |
| 2448 | // Save progress with step and optional data |
| 2449 | // Data can include type, limit, limitSize, and any other progress information |
| 2450 | $progress = array( |
| 2451 | 'step' => $step, |
| 2452 | 'time' => time(), |
| 2453 | 'data' => $data |
| 2454 | ); |
| 2455 | |
| 2456 | return set_transient( $this->progress_key, $progress, 0 ); |
| 2457 | } |
| 2458 | |
| 2459 | function get_progress() { |
| 2460 | if ( $this->runs ) { |
| 2461 | $run = $this->runs->get_resumable(); |
| 2462 | if ( $run ) { |
| 2463 | $public = $this->runs->to_array( $run ); |
| 2464 | return array( |
| 2465 | 'runId' => $public['id'], |
| 2466 | 'status' => $public['status'], |
| 2467 | 'step' => $public['phase'], |
| 2468 | 'time' => strtotime( $public['updated_at'] . ' UTC' ), |
| 2469 | 'data' => $public['checkpoint'], |
| 2470 | 'errors' => $public['errors'], |
| 2471 | ); |
| 2472 | } |
| 2473 | } |
| 2474 | return get_transient( $this->progress_key ); |
| 2475 | } |
| 2476 | |
| 2477 | function get_step_progress() { |
| 2478 | $options = $this->get_all_options(); |
| 2479 | return isset( $options['step_progress'] ) ? $options['step_progress'] : null; |
| 2480 | } |
| 2481 | |
| 2482 | // The cache containing the references is wrote to the DB. |
| 2483 | function write_references() { |
| 2484 | global $wpdb; |
| 2485 | $table = $wpdb->prefix . "mclean_refs"; |
| 2486 | $run_id = $this->get_run_id( true ); |
| 2487 | |
| 2488 | $potential_parents = array(); |
| 2489 | $potential_children = array(); |
| 2490 | |
| 2491 | foreach ( $this->refcache as $value ) { |
| 2492 | $potentialParentPath = !is_null( $value['url'] ) ? $this->clean_url_from_resolution( $value['url'] ) : null; |
| 2493 | if ( $potentialParentPath === $value['url'] ) { |
| 2494 | $potential_parents[] = $value; |
| 2495 | } |
| 2496 | else { |
| 2497 | $potential_children[] = $value; |
| 2498 | } |
| 2499 | } |
| 2500 | |
| 2501 | $this->insert_references( $potential_parents ); |
| 2502 | |
| 2503 | // Resolve parentId for potential children |
| 2504 | foreach ( $potential_children as &$child ) { |
| 2505 | $potentialParentPath = $this->clean_url_from_resolution( $child['url'] ); |
| 2506 | $parentId = $wpdb->get_var( $wpdb->prepare( "SELECT id FROM $table WHERE run_id = %d AND mediaUrl_hash = %s AND mediaUrl = %s LIMIT 1", $run_id, hash( 'sha256', $potentialParentPath ), $potentialParentPath ) ); |
| 2507 | if ( !empty( $parentId ) ) { |
| 2508 | $child['parentId'] = (int)$parentId; |
| 2509 | } |
| 2510 | } |
| 2511 | |
| 2512 | // Insert potential children with resolved parentIds |
| 2513 | $this->insert_references( $potential_children ); |
| 2514 | $this->refcache = array(); |
| 2515 | } |
| 2516 | |
| 2517 | function check_is_ignore( $file ) { |
| 2518 | global $wpdb; |
| 2519 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 2520 | $run_id = $this->get_run_id(); |
| 2521 | $clean_file = $this->clean_uploaded_filename( $file ); |
| 2522 | $count = $wpdb->get_var( $wpdb->prepare( |
| 2523 | "SELECT COUNT(*) FROM $table_name |
| 2524 | WHERE run_id = %d AND ignored = 1 AND path_hash = %s AND path = %s", |
| 2525 | $run_id, |
| 2526 | hash( 'sha256', $clean_file ), |
| 2527 | $clean_file |
| 2528 | ) ); |
| 2529 | if ( $count > 0 ) { |
| 2530 | $this->log( "🚫 Could not trash $file." ); |
| 2531 | } |
| 2532 | return ($count > 0); |
| 2533 | } |
| 2534 | |
| 2535 | function find_media_id_from_file( $file, $doLog ) { |
| 2536 | global $wpdb; |
| 2537 | $postmeta_table_name = $wpdb->prefix . 'postmeta'; |
| 2538 | $file = $this->clean_uploaded_filename( $file ); |
| 2539 | $sql = $wpdb->prepare( "SELECT post_id |
| 2540 | FROM {$postmeta_table_name} |
| 2541 | WHERE meta_key = '_wp_attached_file' |
| 2542 | AND meta_value = %s |
| 2543 | LIMIT 1", $file |
| 2544 | ); |
| 2545 | $ret = $wpdb->get_var( $sql ); |
| 2546 | if ( $doLog ) { |
| 2547 | if ( empty( $ret ) ) |
| 2548 | $this->log( "🚫 File $file not found as _wp_attached_file (Library)." ); |
| 2549 | else { |
| 2550 | $this->log( "� |
| 2551 | File $file found as Media $ret." ); |
| 2552 | } |
| 2553 | } |
| 2554 | |
| 2555 | return $ret; |
| 2556 | } |
| 2557 | |
| 2558 | function get_thumbnails_urls( $id, $sizes_as_key = false ) { |
| 2559 | $sizes = get_intermediate_image_sizes(); |
| 2560 | // For each size use wp_get_attachment_image_src() to get the URL |
| 2561 | $urls = array(); |
| 2562 | foreach ( $sizes as $size ) { |
| 2563 | $src = wp_get_attachment_image_src( $id, $size ); |
| 2564 | if ( $src ) { |
| 2565 | $urls[$size] = $this->clean_url( $src[0] ); |
| 2566 | } |
| 2567 | } |
| 2568 | |
| 2569 | return $sizes_as_key ? $urls : array_values( $urls ); |
| 2570 | } |
| 2571 | |
| 2572 | function get_thumbnails_urls_from_srcset( $media, $size = 'full' ) { |
| 2573 | |
| 2574 | $id = is_numeric( $media ) ? (int)$media : $this->get_id_from_clean_url( $media, false ); |
| 2575 | |
| 2576 | $image_size = $this->get_attachment_size_by_id( $id, $size ); |
| 2577 | |
| 2578 | $sizes = array_keys( $this->get_image_sizes() ); |
| 2579 | $sizes[] = $image_size; |
| 2580 | |
| 2581 | $urls = array(); |
| 2582 | foreach ( $sizes as $image_size ) { |
| 2583 | $srcset = wp_get_attachment_image_srcset( $id, $image_size ); |
| 2584 | |
| 2585 | // Extract URLs from srcset |
| 2586 | if ( !empty( $srcset ) ) { |
| 2587 | $srcset = explode( ', ', $srcset ); |
| 2588 | foreach ( $srcset as $src ) { |
| 2589 | $parts = explode( ' ', $src ); |
| 2590 | $url = trim( $parts[0] ); |
| 2591 | if ( !empty( $url ) ) { |
| 2592 | $urls[] = $this->clean_url( $url ); |
| 2593 | } |
| 2594 | } |
| 2595 | } |
| 2596 | } |
| 2597 | |
| 2598 | return $urls; |
| 2599 | |
| 2600 | } |
| 2601 | |
| 2602 | function get_attachment_size_by_id( $attachment_id, $default_size = 'full' ) { |
| 2603 | |
| 2604 | if ( ! $attachment_id ) { |
| 2605 | return $default_size; |
| 2606 | } |
| 2607 | |
| 2608 | $url = wp_get_attachment_url( $attachment_id ); |
| 2609 | if ( ! $url ) { |
| 2610 | return $default_size; |
| 2611 | } |
| 2612 | |
| 2613 | $metadata = wp_get_attachment_metadata( $attachment_id ); |
| 2614 | |
| 2615 | if ( ! is_array( $metadata ) ) { |
| 2616 | return $default_size; |
| 2617 | } |
| 2618 | |
| 2619 | $size = $default_size; |
| 2620 | |
| 2621 | if ( isset( $metadata['file'] ) && strpos( $url, $metadata['file'] ) === ( strlen( $url ) - strlen( $metadata['file'] ) ) ) { |
| 2622 | $size = array( $metadata['width'], $metadata['height'] ); |
| 2623 | } elseif ( preg_match( '/-(\d+)x(\d+)\.(jpg|jpeg|gif|png|svg|webp)$/', $url, $match ) ) { |
| 2624 | // Get the image width and height. |
| 2625 | // Example: https://regex101.com/r/7JwGz7/1. |
| 2626 | $size = array( $match[1], $match[2] ); |
| 2627 | } |
| 2628 | |
| 2629 | return $size; |
| 2630 | } |
| 2631 | |
| 2632 | function get_image_sizes() { |
| 2633 | $sizes = array(); |
| 2634 | global $_wp_additional_image_sizes; |
| 2635 | foreach ( get_intermediate_image_sizes() as $s ) { |
| 2636 | $crop = false; |
| 2637 | if ( isset( $_wp_additional_image_sizes[$s] ) ) { |
| 2638 | $width = intval( $_wp_additional_image_sizes[$s]['width'] ); |
| 2639 | $height = intval( $_wp_additional_image_sizes[$s]['height'] ); |
| 2640 | $crop = $_wp_additional_image_sizes[$s]['crop']; |
| 2641 | } else { |
| 2642 | $width = get_option( $s.'_size_w' ); |
| 2643 | $height = get_option( $s.'_size_h' ); |
| 2644 | $crop = get_option( $s.'_crop' ); |
| 2645 | } |
| 2646 | $sizes[$s] = array( 'width' => $width, 'height' => $height, 'crop' => $crop ); |
| 2647 | } |
| 2648 | return $sizes; |
| 2649 | } |
| 2650 | |
| 2651 | /** |
| 2652 | * Get all registered thumbnail sizes formatted for the UI. |
| 2653 | * Returns an array of sizes with name, shortname, width, and height. |
| 2654 | */ |
| 2655 | function get_thumbnail_sizes() { |
| 2656 | $sizes = $this->get_image_sizes(); |
| 2657 | $result = array(); |
| 2658 | foreach ( $sizes as $name => $size ) { |
| 2659 | // Generate a shortname (first 2 letters uppercase) |
| 2660 | $shortname = strtoupper( substr( preg_replace( '/[^a-zA-Z]/', '', $name ), 0, 2 ) ); |
| 2661 | $result[] = array( |
| 2662 | 'name' => $name, |
| 2663 | 'shortname' => $shortname, |
| 2664 | 'width' => $size['width'] ? intval( $size['width'] ) : null, |
| 2665 | 'height' => $size['height'] ? intval( $size['height'] ) : null, |
| 2666 | 'crop' => $size['crop'], |
| 2667 | ); |
| 2668 | } |
| 2669 | return $result; |
| 2670 | } |
| 2671 | |
| 2672 | function clean_url_from_resolution( $url ) { |
| 2673 | if ( !isset( $url ) ) return $url; |
| 2674 | |
| 2675 | $pattern = '/[_-]\d+x\d+(?=\.[a-z]{3,4}$)/'; |
| 2676 | $url = preg_replace( $pattern, '', $url ); |
| 2677 | return $url; |
| 2678 | } |
| 2679 | |
| 2680 | function is_url( $url ) { |
| 2681 | return ( ( |
| 2682 | !empty( $url ) ) && |
| 2683 | is_string( $url ) && |
| 2684 | strlen( $url ) > 4 && ( |
| 2685 | strtolower( substr( $url, 0, 4) ) == 'http' || $url[0] == '/' |
| 2686 | ) |
| 2687 | ); |
| 2688 | } |
| 2689 | |
| 2690 | function get_id_from_clean_url( $clean_url ) { |
| 2691 | $found = false; |
| 2692 | $id = 0; |
| 2693 | |
| 2694 | if( !$found ) { |
| 2695 | $id = $this->find_media_id_from_file( $clean_url, false ); |
| 2696 | if ( $id ) { |
| 2697 | $is_attachment = get_post_type( $id ) === 'attachment'; |
| 2698 | if ( $is_attachment ) { |
| 2699 | $found = true; |
| 2700 | } |
| 2701 | } |
| 2702 | } |
| 2703 | |
| 2704 | if( !$found ) { |
| 2705 | $id = $this->custom_attachment_url_to_postid( $clean_url ); |
| 2706 | if ( $id ) { |
| 2707 | $is_attachment = get_post_type( $id ) === 'attachment'; |
| 2708 | if ( $is_attachment ) { |
| 2709 | $found = true; |
| 2710 | } |
| 2711 | } |
| 2712 | } |
| 2713 | |
| 2714 | if ( !$found ) { |
| 2715 | $id = $this->resolve_from_database( $clean_url ); |
| 2716 | if ( $id ) { |
| 2717 | $is_attachment = get_post_type( $id ) === 'attachment'; |
| 2718 | if ( $is_attachment ) { |
| 2719 | $found = true; |
| 2720 | } |
| 2721 | } |
| 2722 | } |
| 2723 | |
| 2724 | |
| 2725 | return $found ? $id : null; |
| 2726 | } |
| 2727 | |
| 2728 | function resolve_from_database( $url ) { |
| 2729 | global $wpdb; |
| 2730 | $pattern = '/[_-]\d+x\d+(?=\.[a-z]{3,4}$)/'; |
| 2731 | $url = preg_replace( $pattern, '', $url ); |
| 2732 | $url = $this->get_pathinfo_from_image_src( $url ); |
| 2733 | // A guid LIKE '%...%' cannot use an index, so it must never bring back more than the one |
| 2734 | // row this function actually uses. |
| 2735 | $query = $wpdb->prepare( "SELECT ID FROM $wpdb->posts WHERE guid LIKE '%s' LIMIT 1", '%' . $url . '%' ); |
| 2736 | $attachment = $wpdb->get_var( $query ); |
| 2737 | return empty( $attachment ) ? null : $attachment; |
| 2738 | } |
| 2739 | |
| 2740 | function get_pathinfo_from_image_src( $image_src ) { |
| 2741 | $uploads = wp_upload_dir(); |
| 2742 | $uploads_url = trailingslashit( $uploads['baseurl'] ); |
| 2743 | if ( strpos( $image_src, $uploads_url ) === 0 ) |
| 2744 | return ltrim( substr( $image_src, strlen( $uploads_url ) ), '/'); |
| 2745 | else if ( strpos( $image_src, wp_make_link_relative( $uploads_url ) ) === 0 ) |
| 2746 | return ltrim( substr( $image_src, strlen( wp_make_link_relative( $uploads_url ) ) ), '/'); |
| 2747 | $img_info = parse_url( $image_src ); |
| 2748 | return ltrim( $img_info['path'], '/' ); |
| 2749 | } |
| 2750 | |
| 2751 | function clean_url_from_resolution_ref( &$url ) { |
| 2752 | $url = $this->clean_url_from_resolution( $url ); |
| 2753 | } |
| 2754 | |
| 2755 | // From a url to the shortened and cleaned url (for example '2013/02/file.png') |
| 2756 | function clean_url( $url ) { |
| 2757 | if ( !is_string( $url ) || $url === '' || $this->upload_url === '' ) { |
| 2758 | return null; |
| 2759 | } |
| 2760 | $url = html_entity_decode( $url, ENT_QUOTES, 'UTF-8' ); |
| 2761 | $base_path = (string) wp_parse_url( $this->upload_url, PHP_URL_PATH ); |
| 2762 | $url_path = (string) wp_parse_url( $url, PHP_URL_PATH ); |
| 2763 | if ( $url_path === '' || $base_path === '' ) { |
| 2764 | return null; |
| 2765 | } |
| 2766 | // The host is deliberately not compared: content routinely references the |
| 2767 | // uploads directory through www/non-www variants, CDN aliases, or image |
| 2768 | // proxies, and a missed reference is far more dangerous than an extra one. |
| 2769 | // The uploads path is matched anywhere so proxy prefixes also resolve. |
| 2770 | $marker = trailingslashit( $base_path ); |
| 2771 | $position = strpos( $url_path, $marker ); |
| 2772 | if ( $position === false ) { |
| 2773 | return null; |
| 2774 | } |
| 2775 | $relative = substr( $url_path, $position + strlen( $marker ) ); |
| 2776 | $normalized = $this->normalize_upload_relative_path( urldecode( $relative ) ); |
| 2777 | return is_wp_error( $normalized ) ? null : $normalized; |
| 2778 | } |
| 2779 | |
| 2780 | function custom_attachment_url_to_postid( $url ) { |
| 2781 | global $wpdb; |
| 2782 | |
| 2783 | // Remove the query string |
| 2784 | $url = preg_replace('/\?.*/', '', $url); |
| 2785 | |
| 2786 | // Try to find the attachment ID by matching the URL with the guid |
| 2787 | $attachment = $wpdb->get_var( $wpdb->prepare( "SELECT ID FROM $wpdb->posts WHERE guid LIKE %s AND post_type = 'attachment' LIMIT 1;", '%' . $wpdb->esc_like( $url ) ) ); |
| 2788 | |
| 2789 | // If found, return the first attachment ID |
| 2790 | if ( !empty( $attachment ) ) { |
| 2791 | return ( int )$attachment; |
| 2792 | } |
| 2793 | |
| 2794 | // If not found, try to match the URL without the upload directory path |
| 2795 | $upload_dir = wp_upload_dir(); |
| 2796 | $url_relative = str_replace( $upload_dir['baseurl'] . '/', '', $url ); |
| 2797 | |
| 2798 | $attachment = $wpdb->get_var( $wpdb->prepare( "SELECT post_id FROM $wpdb->postmeta WHERE meta_key = '_wp_attached_file' AND meta_value LIKE %s LIMIT 1;", '%' . $wpdb->esc_like( $url_relative ) ) ); |
| 2799 | |
| 2800 | // If found, return the first attachment ID |
| 2801 | if ( !empty( $attachment ) ) { |
| 2802 | return ( int )$attachment; |
| 2803 | } |
| 2804 | |
| 2805 | // If still not found, return 0 |
| 2806 | return 0; |
| 2807 | } |
| 2808 | |
| 2809 | // From a fullpath to the shortened and cleaned path (for example '2013/02/file.png') |
| 2810 | // Original version by Jordy |
| 2811 | // function clean_uploaded_filename( $fullpath ) { |
| 2812 | // $basedir = $this->upload_path; |
| 2813 | // $file = str_replace( $basedir, '', $fullpath ); |
| 2814 | // $file = str_replace( "./", "", $file ); |
| 2815 | // $file = trim( $file, "/" ); |
| 2816 | // return $file; |
| 2817 | // } |
| 2818 | |
| 2819 | // From a fullpath to the shortened and cleaned path (for example '2013/02/file.png') |
| 2820 | // Faster version, more difficult to read, by Mike Meinz |
| 2821 | function clean_uploaded_filename( $fullpath ) { |
| 2822 | if ( !is_string( $fullpath ) || $fullpath === '' ) { |
| 2823 | return ''; |
| 2824 | } |
| 2825 | if ( preg_match( '#^https?://#i', $fullpath ) ) { |
| 2826 | $clean = $this->clean_url( $fullpath ); |
| 2827 | return $clean === null ? '' : $clean; |
| 2828 | } |
| 2829 | $path = wp_normalize_path( rawurldecode( $fullpath ) ); |
| 2830 | $base = untrailingslashit( wp_normalize_path( $this->upload_path ) ); |
| 2831 | if ( $base !== '' && ( $path === $base || strpos( $path, trailingslashit( $base ) ) === 0 ) ) { |
| 2832 | $path = ltrim( substr( $path, strlen( $base ) ), '/' ); |
| 2833 | } |
| 2834 | else if ( substr( $path, 0, 1 ) === '/' ) { |
| 2835 | return ''; |
| 2836 | } |
| 2837 | $normalized = $this->normalize_upload_relative_path( $path ); |
| 2838 | return is_wp_error( $normalized ) ? '' : $normalized; |
| 2839 | } |
| 2840 | |
| 2841 | public function build_file_manifest( $paths ) { |
| 2842 | $manifest = array(); |
| 2843 | foreach ( array_values( array_unique( (array) $paths ) ) as $path ) { |
| 2844 | $relative = $this->normalize_upload_relative_path( $this->clean_uploaded_filename( $path ) ); |
| 2845 | if ( is_wp_error( $relative ) ) return $relative; |
| 2846 | $absolute = $this->resolve_upload_path( $relative ); |
| 2847 | if ( is_wp_error( $absolute ) ) { |
| 2848 | // Instead of propagating an error, mark the file as unsafe and continue. |
| 2849 | // This allows a scan to complete even if some files can't be fingerprinted. |
| 2850 | $code = $absolute->get_error_code(); |
| 2851 | if ( $code === 'wpmc_symlink_rejected' || $code === 'wpmc_path_outside_uploads' ) { |
| 2852 | $manifest[ $relative ] = self::FINGERPRINT_UNSAFE; |
| 2853 | continue; |
| 2854 | } |
| 2855 | return $absolute; |
| 2856 | } |
| 2857 | $manifest[ $relative ] = $this->file_fingerprint( $absolute ); |
| 2858 | } |
| 2859 | return $manifest; |
| 2860 | } |
| 2861 | |
| 2862 | // Returns null for a file that does not exist, a 64-char sha256 fingerprint for a |
| 2863 | // normal file, or the FINGERPRINT_UNSAFE sentinel for a file that exists but cannot be |
| 2864 | // safely fingerprinted. |
| 2865 | public function file_fingerprint( $absolute_path ) { |
| 2866 | if ( !file_exists( $absolute_path ) ) return null; |
| 2867 | if ( !is_file( $absolute_path ) || !is_readable( $absolute_path ) || is_link( $absolute_path ) ) { |
| 2868 | return self::FINGERPRINT_UNSAFE; |
| 2869 | } |
| 2870 | $stat = @lstat( $absolute_path ); |
| 2871 | $handle = @fopen( $absolute_path, 'rb' ); |
| 2872 | if ( !$stat || !$handle ) { |
| 2873 | if ( $handle ) fclose( $handle ); |
| 2874 | return self::FINGERPRINT_UNSAFE; |
| 2875 | } |
| 2876 | $context = hash_init( 'sha256' ); |
| 2877 | hash_update( $context, wp_json_encode( array( |
| 2878 | 'size' => isset( $stat['size'] ) ? (int) $stat['size'] : 0, |
| 2879 | 'mtime' => isset( $stat['mtime'] ) ? (int) $stat['mtime'] : 0, |
| 2880 | 'ino' => isset( $stat['ino'] ) ? (int) $stat['ino'] : 0, |
| 2881 | 'dev' => isset( $stat['dev'] ) ? (int) $stat['dev'] : 0, |
| 2882 | ) ) ); |
| 2883 | $sample_size = 64 * 1024; |
| 2884 | $first = fread( $handle, $sample_size ); |
| 2885 | if ( $first === false ) { |
| 2886 | fclose( $handle ); |
| 2887 | return self::FINGERPRINT_UNSAFE; |
| 2888 | } |
| 2889 | hash_update( $context, $first ); |
| 2890 | if ( (int) $stat['size'] > $sample_size ) { |
| 2891 | fseek( $handle, max( 0, (int) $stat['size'] - $sample_size ), SEEK_SET ); |
| 2892 | $last = fread( $handle, $sample_size ); |
| 2893 | if ( $last === false ) { |
| 2894 | fclose( $handle ); |
| 2895 | return self::FINGERPRINT_UNSAFE; |
| 2896 | } |
| 2897 | hash_update( $context, $last ); |
| 2898 | } |
| 2899 | fclose( $handle ); |
| 2900 | return hash_final( $context ); |
| 2901 | } |
| 2902 | |
| 2903 | public function validate_issue_manifest( $issue ) { |
| 2904 | $manifest = json_decode( (string) $issue->manifest, true ); |
| 2905 | if ( !is_array( $manifest ) ) return true; |
| 2906 | foreach ( $manifest as $relative => $expected ) { |
| 2907 | // We throw an error for the unasfe tag only in the context of cleanup, not in the context of scanning. |
| 2908 | if ( $expected === self::FINGERPRINT_UNSAFE ) { |
| 2909 | return new WP_Error( 'wpmc_file_unsafe_for_cleanup', __( 'This item includes a file that cannot be safely verified (a symbolic link, an unreadable file, or a special file), so Media Cleaner will not touch it. Fix or remove that file, then run a new scan.', 'media-cleaner' ) ); |
| 2910 | } |
| 2911 | $upload = $this->resolve_upload_path( $relative ); |
| 2912 | $trash = $this->resolve_trash_path( $relative ); |
| 2913 | if ( is_wp_error( $upload ) ) return $upload; |
| 2914 | if ( is_wp_error( $trash ) ) return $trash; |
| 2915 | $existing = file_exists( $upload ) ? $upload : ( file_exists( $trash ) ? $trash : null ); |
| 2916 | if ( $expected === null ) { |
| 2917 | if ( $existing ) return new WP_Error( 'wpmc_file_changed_since_scan', __( 'A file appeared after the scan. Run a new scan before cleanup.', 'media-cleaner' ) ); |
| 2918 | continue; |
| 2919 | } |
| 2920 | if ( !$existing ) return new WP_Error( 'wpmc_file_changed_since_scan', __( 'A file disappeared after the scan. Run a new scan before cleanup.', 'media-cleaner' ) ); |
| 2921 | // A real fingerprint that no longer matches — including a file that has since |
| 2922 | // become unsafe and now returns the sentinel — means the file changed. |
| 2923 | $current = $this->file_fingerprint( $existing ); |
| 2924 | if ( !hash_equals( (string) $expected, (string) $current ) ) { |
| 2925 | return new WP_Error( 'wpmc_file_changed_since_scan', __( 'A file changed after the scan. Run a new scan before cleanup.', 'media-cleaner' ) ); |
| 2926 | } |
| 2927 | } |
| 2928 | return true; |
| 2929 | } |
| 2930 | |
| 2931 | public function normalize_upload_relative_path( $path ) { |
| 2932 | if ( !is_string( $path ) || strpos( $path, "\0" ) !== false ) { |
| 2933 | return new WP_Error( 'wpmc_invalid_path', __( 'The storage path is invalid.', 'media-cleaner' ) ); |
| 2934 | } |
| 2935 | $path = wp_normalize_path( rawurldecode( trim( $path ) ) ); |
| 2936 | if ( preg_match( '#^[a-zA-Z]:/#', $path ) || strpos( $path, '://' ) !== false ) { |
| 2937 | return new WP_Error( 'wpmc_absolute_path', __( 'Absolute storage paths are not allowed.', 'media-cleaner' ) ); |
| 2938 | } |
| 2939 | $path = ltrim( $path, '/' ); |
| 2940 | $segments = array(); |
| 2941 | foreach ( explode( '/', $path ) as $segment ) { |
| 2942 | if ( $segment === '' || $segment === '.' ) { |
| 2943 | continue; |
| 2944 | } |
| 2945 | if ( $segment === '..' ) { |
| 2946 | return new WP_Error( 'wpmc_path_traversal', __( 'The storage path attempts to leave the uploads directory.', 'media-cleaner' ) ); |
| 2947 | } |
| 2948 | $segments[] = $segment; |
| 2949 | } |
| 2950 | return implode( '/', $segments ); |
| 2951 | } |
| 2952 | |
| 2953 | public function resolve_upload_path( $relative_path, $must_exist = false ) { |
| 2954 | $relative = $this->normalize_upload_relative_path( $relative_path ); |
| 2955 | if ( is_wp_error( $relative ) ) { |
| 2956 | return $relative; |
| 2957 | } |
| 2958 | $base = realpath( $this->upload_path ); |
| 2959 | if ( !$base ) { |
| 2960 | return new WP_Error( 'wpmc_upload_root_missing', __( 'The uploads directory is unavailable.', 'media-cleaner' ) ); |
| 2961 | } |
| 2962 | $base = untrailingslashit( wp_normalize_path( $base ) ); |
| 2963 | $candidate = $base . ( $relative === '' ? '' : '/' . $relative ); |
| 2964 | if ( is_link( $candidate ) ) { |
| 2965 | return new WP_Error( 'wpmc_symlink_rejected', __( 'Symbolic links are not modified by Media Cleaner.', 'media-cleaner' ) ); |
| 2966 | } |
| 2967 | $existing = realpath( $candidate ); |
| 2968 | if ( $must_exist && !$existing ) { |
| 2969 | return new WP_Error( 'wpmc_path_missing', __( 'The requested file no longer exists.', 'media-cleaner' ) ); |
| 2970 | } |
| 2971 | $resolved = $existing ? wp_normalize_path( $existing ) : $candidate; |
| 2972 | if ( $existing && $resolved !== $base && strpos( $resolved, trailingslashit( $base ) ) !== 0 ) { |
| 2973 | return new WP_Error( 'wpmc_path_outside_uploads', __( 'The requested path is outside the uploads directory.', 'media-cleaner' ) ); |
| 2974 | } |
| 2975 | if ( !$existing ) { |
| 2976 | $ancestor = dirname( $candidate ); |
| 2977 | while ( !file_exists( $ancestor ) && dirname( $ancestor ) !== $ancestor ) { |
| 2978 | $ancestor = dirname( $ancestor ); |
| 2979 | } |
| 2980 | $ancestor_real = realpath( $ancestor ); |
| 2981 | $ancestor_real = $ancestor_real ? untrailingslashit( wp_normalize_path( $ancestor_real ) ) : ''; |
| 2982 | if ( $ancestor_real !== $base && strpos( $ancestor_real, trailingslashit( $base ) ) !== 0 ) { |
| 2983 | return new WP_Error( 'wpmc_path_outside_uploads', __( 'The requested path is outside the uploads directory.', 'media-cleaner' ) ); |
| 2984 | } |
| 2985 | } |
| 2986 | if ( $existing && is_link( $candidate ) ) { |
| 2987 | return new WP_Error( 'wpmc_symlink_rejected', __( 'Symbolic links are not modified by Media Cleaner.', 'media-cleaner' ) ); |
| 2988 | } |
| 2989 | return $candidate; |
| 2990 | } |
| 2991 | |
| 2992 | /** |
| 2993 | * Check if the file or the Media ID is used in the install. |
| 2994 | * That file or ID will be checked against the database of references created by the plugin |
| 2995 | * by the parsers. |
| 2996 | */ |
| 2997 | function reference_exists( $file, $mediaId ) { |
| 2998 | global $wpdb; |
| 2999 | |
| 3000 | $table = $wpdb->prefix . "mclean_refs"; |
| 3001 | $run_id = $this->get_run_id(); |
| 3002 | |
| 3003 | $row = null; |
| 3004 | if ( !empty( $mediaId ) ) { |
| 3005 | $row = $wpdb->get_row( $wpdb->prepare( "SELECT originType FROM $table WHERE run_id = %d AND mediaId = %d LIMIT 1", $run_id, $mediaId ) ); |
| 3006 | if ( !empty( $row ) ) { |
| 3007 | $origin = $row->originType === 'MEDIA LIBRARY' ? 'Media Library' : 'content'; |
| 3008 | $this->log( "� |
| 3009 | Media #{$mediaId} used by {$origin}" ); |
| 3010 | return $row->originType; |
| 3011 | } |
| 3012 | } |
| 3013 | if ( !empty( $file ) ) { |
| 3014 | $row = $wpdb->get_row( $wpdb->prepare( |
| 3015 | "SELECT originType FROM $table WHERE run_id = %d AND mediaUrl_hash = %s AND mediaUrl = %s LIMIT 1", |
| 3016 | $run_id, |
| 3017 | hash( 'sha256', $file ), |
| 3018 | $file |
| 3019 | ) ); |
| 3020 | if ( !empty( $row ) ) { |
| 3021 | $origin = $row->originType === 'MEDIA LIBRARY' ? 'Media Library' : 'content'; |
| 3022 | $this->log( "� |
| 3023 | File {$file} used by {$origin}" ); |
| 3024 | return $row->originType; |
| 3025 | } |
| 3026 | } |
| 3027 | return false; |
| 3028 | } |
| 3029 | |
| 3030 | function get_full_upload_path( $relative_path ) { |
| 3031 | $full_path = $this->resolve_upload_path( $relative_path ); |
| 3032 | return is_wp_error( $full_path ) ? null : $full_path; |
| 3033 | } |
| 3034 | |
| 3035 | function get_paths_from_attachment( $attachmentId ) { |
| 3036 | $paths = array(); |
| 3037 | $fullpath = get_attached_file( $attachmentId ); |
| 3038 | if ( empty( $fullpath ) ) { |
| 3039 | $this->log( 'Could not find attached file for Media ID ' . $attachmentId ); |
| 3040 | return array(); |
| 3041 | } |
| 3042 | $mainfile = $this->clean_uploaded_filename( $fullpath ); |
| 3043 | array_push( $paths, $mainfile ); |
| 3044 | $baseUp = pathinfo( $mainfile ); |
| 3045 | $filespath = trailingslashit( $this->upload_path ) . trailingslashit( $baseUp['dirname'] ); |
| 3046 | $meta = wp_get_attachment_metadata( $attachmentId ); |
| 3047 | if ( isset( $meta['original_image'] ) ) { |
| 3048 | $original_image = $this->clean_uploaded_filename( $filespath . $meta['original_image'] ); |
| 3049 | array_push( $paths, $original_image ); |
| 3050 | } |
| 3051 | $isImage = isset( $meta, $meta['width'], $meta['height'] ); |
| 3052 | $sizes = $this->get_image_sizes(); |
| 3053 | if ( $isImage && isset( $meta['sizes'] ) ) { |
| 3054 | foreach ( $meta['sizes'] as $name => $attr ) { |
| 3055 | if ( isset( $attr['file'] ) ) { |
| 3056 | $file = $this->clean_uploaded_filename( $filespath . $attr['file'] ); |
| 3057 | array_push( $paths, $file ); |
| 3058 | } |
| 3059 | } |
| 3060 | } |
| 3061 | return $paths; |
| 3062 | } |
| 3063 | |
| 3064 | function is_media_ignored( $attachmentId ) { |
| 3065 | global $wpdb; |
| 3066 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 3067 | $run_id = $this->get_run_id(); |
| 3068 | $issue = $wpdb->get_row( $wpdb->prepare( "SELECT * FROM $table_name WHERE run_id = %d AND postId = %d", $run_id, $attachmentId ), OBJECT ); |
| 3069 | //error_log( $attachmentId ); |
| 3070 | //error_log( print_r( $issue, 1 ) ); |
| 3071 | if ( $issue && $issue->ignored ) |
| 3072 | return true; |
| 3073 | return false; |
| 3074 | } |
| 3075 | |
| 3076 | function check_media( $attachmentId, $checkOnly = false ) { |
| 3077 | |
| 3078 | // Is Media ID ignored, consider as used. |
| 3079 | if ( $this->is_media_ignored( $attachmentId ) ) { |
| 3080 | return true; |
| 3081 | } |
| 3082 | |
| 3083 | // Remove everything related to this media from the database. |
| 3084 | if ( !$checkOnly ) { |
| 3085 | $this->delete_attachment_related_data( $attachmentId ); |
| 3086 | } |
| 3087 | |
| 3088 | $size = 0; |
| 3089 | $countfiles = 0; |
| 3090 | $check_content = (bool) $this->get_option( 'content' ); |
| 3091 | $check_broken_media = !$check_content; |
| 3092 | $fullpath = get_attached_file( $attachmentId ); |
| 3093 | $is_broken = apply_filters( 'wpmc_is_file_broken', !file_exists( $fullpath ), $attachmentId ); |
| 3094 | |
| 3095 | // It's a broken-only scan |
| 3096 | if ( $check_broken_media && !$is_broken ) { |
| 3097 | $is_considered_used = apply_filters( 'wpmc_check_media', true, $attachmentId, false ); |
| 3098 | return $is_considered_used; |
| 3099 | } |
| 3100 | |
| 3101 | // Let's analyze the usage of each path (thumbnails included) for this Media ID. |
| 3102 | $issue = 'NO_CONTENT'; |
| 3103 | $paths = $this->get_paths_from_attachment( $attachmentId ); |
| 3104 | foreach ( $paths as $path ) { |
| 3105 | |
| 3106 | // If it's found in the content, we stop the scan right away |
| 3107 | if ( $check_content && $this->reference_exists( $path, $attachmentId ) ) { |
| 3108 | $is_considered_used = apply_filters( 'wpmc_check_media', true, $attachmentId, false ); |
| 3109 | if ( $is_considered_used ) { |
| 3110 | return true; |
| 3111 | } |
| 3112 | } |
| 3113 | |
| 3114 | // Let's count the size of the files for later, in case it's unused |
| 3115 | $filepath = trailingslashit( $this->upload_path ) . $path; |
| 3116 | if ( file_exists( $filepath ) ) |
| 3117 | $size += filesize( $filepath ); |
| 3118 | $countfiles++; |
| 3119 | } |
| 3120 | |
| 3121 | // This Media ID seems not in used (or broken) |
| 3122 | // Let's double-check through the filter (overridable by users) |
| 3123 | $is_considered_used = apply_filters( 'wpmc_check_media', false, $attachmentId, $is_broken ); |
| 3124 | if ( !$is_considered_used ) { |
| 3125 | if ( $is_broken ) { |
| 3126 | $this->log( "🚫 File {$fullpath} does not exist." ); |
| 3127 | $issue = 'ORPHAN_MEDIA'; |
| 3128 | } |
| 3129 | if ( !$checkOnly ) { |
| 3130 | global $wpdb; |
| 3131 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 3132 | $mainfile = $this->clean_uploaded_filename( $fullpath ); |
| 3133 | $display_path = $mainfile !== '' ? $mainfile . ( $countfiles > 0 ? ( " (+ " . $countfiles . " thumbnails)" ) : "" ) : sprintf( 'Media #%d (missing attached file)', $attachmentId ); |
| 3134 | $manifest = $this->build_file_manifest( $paths ); |
| 3135 | if ( is_wp_error( $manifest ) ) throw new RuntimeException( $manifest->get_error_message() ); |
| 3136 | $inserted = $wpdb->insert( $table_name, |
| 3137 | array( |
| 3138 | 'run_id' => $this->get_run_id( true ), |
| 3139 | 'time' => current_time('mysql'), |
| 3140 | 'type' => 1, |
| 3141 | 'size' => $size, |
| 3142 | 'path' => $display_path, |
| 3143 | 'path_hash' => hash( 'sha256', $display_path ), |
| 3144 | 'manifest' => wp_json_encode( $manifest ), |
| 3145 | 'postId' => $attachmentId, |
| 3146 | 'issue' => $issue |
| 3147 | ) |
| 3148 | ); |
| 3149 | if ( $inserted === false ) throw new RuntimeException( sprintf( __( 'Media Cleaner could not store a media issue: %s', 'media-cleaner' ), $wpdb->last_error ) ); |
| 3150 | } |
| 3151 | } |
| 3152 | return $is_considered_used; |
| 3153 | } |
| 3154 | |
| 3155 | // Delete all issues |
| 3156 | function reset_issues( $includingIgnored = false ) { |
| 3157 | global $wpdb; |
| 3158 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 3159 | $run_id = $this->get_run_id( true ); |
| 3160 | if ( $run_id < 1 ) throw new RuntimeException( __( 'A writable scan run is required before resetting issues.', 'media-cleaner' ) ); |
| 3161 | if ( $includingIgnored ) { |
| 3162 | $deleted = $wpdb->query( $wpdb->prepare( "DELETE FROM $table_name WHERE run_id = %d AND deleted = 0", $run_id ) ); |
| 3163 | } |
| 3164 | else { |
| 3165 | $deleted = $wpdb->query( $wpdb->prepare( "DELETE FROM $table_name WHERE run_id = %d AND ignored = 0 AND deleted = 0", $run_id ) ); |
| 3166 | } |
| 3167 | if ( $deleted === false ) throw new RuntimeException( sprintf( __( 'Media Cleaner could not reset staged issues: %s', 'media-cleaner' ), $wpdb->last_error ) ); |
| 3168 | } |
| 3169 | |
| 3170 | function is_image_extension( $ext ) { |
| 3171 | $ext = strtolower( $ext ); |
| 3172 | $valid = apply_filters( 'wpmc_valid_image_extensions', array( 'jpg', 'jpeg', 'png', 'gif', 'bmp', 'tiff', 'ico', 'webp', 'avif' ) ); |
| 3173 | |
| 3174 | return in_array( $ext, $valid ); |
| 3175 | |
| 3176 | } |
| 3177 | |
| 3178 | |
| 3179 | function reset_references() { |
| 3180 | global $wpdb; |
| 3181 | $table_name = $wpdb->prefix . "mclean_refs"; |
| 3182 | $run_id = $this->get_run_id( true ); |
| 3183 | if ( $run_id < 1 ) throw new RuntimeException( __( 'A writable scan run is required before resetting references.', 'media-cleaner' ) ); |
| 3184 | $deleted = $wpdb->query( $wpdb->prepare( "DELETE FROM $table_name WHERE run_id = %d", $run_id ) ); |
| 3185 | if ( $deleted === false ) throw new RuntimeException( sprintf( __( 'Media Cleaner could not reset staged references: %s', 'media-cleaner' ), $wpdb->last_error ) ); |
| 3186 | } |
| 3187 | |
| 3188 | function get_issue_for_postId( $postId ) { |
| 3189 | global $wpdb; |
| 3190 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 3191 | $run_id = $this->get_run_id(); |
| 3192 | $issue = $wpdb->get_row( $wpdb->prepare( "SELECT * FROM $table_name WHERE run_id = %d AND postId = %d", $run_id, $postId ), OBJECT ); |
| 3193 | return $issue; |
| 3194 | } |
| 3195 | |
| 3196 | function echo_issue( $issue ) { |
| 3197 | if ( $issue == 'NO_CONTENT' ) { |
| 3198 | _e( "Not found in content", 'media-cleaner' ); |
| 3199 | } |
| 3200 | else if ( $issue == 'ORPHAN_FILE' ) { |
| 3201 | _e( "Not in Library", 'media-cleaner' ); |
| 3202 | } |
| 3203 | else if ( $issue == 'ORPHAN_RETINA' ) { |
| 3204 | _e( "Orphan Retina", 'media-cleaner' ); |
| 3205 | } |
| 3206 | else if ( $issue == 'ORPHAN_WEBP' ) { |
| 3207 | _e( "Orphan WebP", 'media-cleaner' ); |
| 3208 | } |
| 3209 | else if ( $issue == 'ORPHAN_MEDIA' ) { |
| 3210 | _e( "No attached file", 'media-cleaner' ); |
| 3211 | } |
| 3212 | else { |
| 3213 | echo $issue; |
| 3214 | } |
| 3215 | } |
| 3216 | |
| 3217 | function get_uploads_directory_hierarchy() { |
| 3218 | $uploads_dir = wp_upload_dir(); |
| 3219 | if ( !empty( $uploads_dir['error'] ) ) { |
| 3220 | throw new RuntimeException( $uploads_dir['error'] ); |
| 3221 | } |
| 3222 | $base_dir = wp_normalize_path( $uploads_dir['basedir'] ); |
| 3223 | $root = '/' . wp_basename( $base_dir ); |
| 3224 | $directories = array(); |
| 3225 | |
| 3226 | // Get all subdirectories of the base directory |
| 3227 | $dir_iterator = new RecursiveDirectoryIterator( $base_dir, FilesystemIterator::KEY_AS_PATHNAME | FilesystemIterator::CURRENT_AS_FILEINFO | FilesystemIterator::SKIP_DOTS ); |
| 3228 | $iterator = new RecursiveIteratorIterator( $dir_iterator, RecursiveIteratorIterator::SELF_FIRST ); |
| 3229 | |
| 3230 | $max_directories = (int) apply_filters( 'wpmc_directory_picker_limit', 5000 ); |
| 3231 | foreach ( $iterator as $file ) { |
| 3232 | if ( count( $directories ) >= $max_directories ) { |
| 3233 | break; |
| 3234 | } |
| 3235 | if ( $file->isDir() ) { |
| 3236 | // Normalize path for consistency |
| 3237 | $file_path = wp_normalize_path( $file->getPathname() ); |
| 3238 | // Remove base_dir from path |
| 3239 | $directory = str_replace( $base_dir, '', $file_path ); |
| 3240 | if ( $directory ) { |
| 3241 | $directories[] = $root . $directory; |
| 3242 | } |
| 3243 | } |
| 3244 | } |
| 3245 | |
| 3246 | // Return the hierarchy as a JSON file |
| 3247 | return wp_json_encode( $directories ); |
| 3248 | } |
| 3249 | |
| 3250 | /** |
| 3251 | * |
| 3252 | * Roles & Access Rights |
| 3253 | * |
| 3254 | */ |
| 3255 | public function can_access_settings() { |
| 3256 | return apply_filters( 'wpmc_allow_setup', current_user_can( 'manage_options' ) ); |
| 3257 | } |
| 3258 | |
| 3259 | // Cached per request: the mutators below can be called in batches of 100. |
| 3260 | private $results_staged = null; |
| 3261 | |
| 3262 | public function can_access_features() { |
| 3263 | return apply_filters( 'wpmc_allow_usage', current_user_can( 'manage_options' ) ); |
| 3264 | } |
| 3265 | |
| 3266 | public function can_cleanup() { |
| 3267 | $allowed = current_user_can( 'manage_options' ) && $this->runs && $this->runs->cleanup_allowed(); |
| 3268 | return $allowed && apply_filters( 'wpmc_allow_cleanup', true ); |
| 3269 | } |
| 3270 | |
| 3271 | /** |
| 3272 | * A staged scan copies the ignored and trashed results into itself when it starts, |
| 3273 | * and that copy replaces them when it publishes. Changing them in the meantime |
| 3274 | * would be silently undone, and a trashed row coming back after its file was |
| 3275 | * recovered would let the next Empty Trash erase a file that is in use. |
| 3276 | * |
| 3277 | * So the results hold still while a scan is staged over them. The scan is the |
| 3278 | * user's own, and publishing or cancelling it releases them. |
| 3279 | */ |
| 3280 | private function results_staged_error() { |
| 3281 | if ( $this->results_staged === null ) { |
| 3282 | $this->results_staged = $this->runs && $this->runs->get_resumable() ? true : false; |
| 3283 | } |
| 3284 | if ( !$this->results_staged ) return null; |
| 3285 | return new WP_Error( 'wpmc_scan_staged', |
| 3286 | __( 'A scan is staged over these results, so they cannot be changed yet. Publish it or cancel it first, then try again.', 'media-cleaner' ), |
| 3287 | array( 'status' => 409 ) ); |
| 3288 | } |
| 3289 | |
| 3290 | #region Options |
| 3291 | |
| 3292 | function list_options() { |
| 3293 | return array( |
| 3294 | 'method' => 'media', |
| 3295 | 'content' => true, |
| 3296 | 'filesystem_content' => true, |
| 3297 | 'media_library' => false, |
| 3298 | 'live_content' => false, |
| 3299 | 'debuglogs' => false, |
| 3300 | 'images_only' => false, |
| 3301 | 'attach_is_use' => false, |
| 3302 | 'thumbnails_only' => false, |
| 3303 | 'dirs_filter' => '', |
| 3304 | 'files_filter' => '', |
| 3305 | 'hide_thumbnails' => false, |
| 3306 | 'hide_warning' => false, |
| 3307 | 'skip_trash' => false, |
| 3308 | 'medias_buffer' => 100, |
| 3309 | 'posts_buffer' => 5, |
| 3310 | 'analysis_buffer' => 100, |
| 3311 | 'file_op_buffer' => 20, |
| 3312 | 'uploads_file_buffer' => 500, |
| 3313 | 'delay' => 100, |
| 3314 | 'refs_buffer' => 500, |
| 3315 | 'analysis_document_limit' => 8 * 1024 * 1024, |
| 3316 | 'shortcodes_disabled' => false, |
| 3317 | |
| 3318 | 'output_buffer_cleaning_disabled' => false, |
| 3319 | 'php_error_logs' => false, |
| 3320 | 'posts_per_page' => 10, |
| 3321 | 'clean_uninstall' => false, |
| 3322 | 'repair_mode' => false, |
| 3323 | 'expert_mode' => false, |
| 3324 | 'mcp_support' => false, |
| 3325 | 'logs_path' => null, |
| 3326 | 'thumbnail_force_issues' => [], |
| 3327 | ); |
| 3328 | } |
| 3329 | |
| 3330 | function reset_options() { |
| 3331 | delete_option( $this->option_name ); |
| 3332 | } |
| 3333 | |
| 3334 | function get_option( $option ) { |
| 3335 | if ( $this->run_id > 0 && array_key_exists( $option, $this->run_config ) ) { |
| 3336 | return $this->run_config[ $option ]; |
| 3337 | } |
| 3338 | $options = $this->get_all_options(); |
| 3339 | return $options[$option]; |
| 3340 | } |
| 3341 | |
| 3342 | function get_all_options() { |
| 3343 | $options = get_option( $this->option_name, null ); |
| 3344 | $options = $this->check_options( $options ); |
| 3345 | return $options; |
| 3346 | } |
| 3347 | |
| 3348 | // Let's work on this function if we need it. |
| 3349 | // Right now, it looks like the options are all updated at the same time. |
| 3350 | |
| 3351 | // function update_option( $option, $value ) { |
| 3352 | // if ( !array_key_exists( $name, $options ) ) { |
| 3353 | // return new WP_REST_Response([ 'success' => false, 'message' => 'This option does not exist.' ], 200 ); |
| 3354 | // } |
| 3355 | // $value = is_bool( $params['value'] ) ? ( $params['value'] ? '1' : '' ) : $params['value']; |
| 3356 | // } |
| 3357 | |
| 3358 | function update_options( $options ) { |
| 3359 | $current = get_option( $this->option_name, array() ); |
| 3360 | $current = is_array( $current ) ? $current : array(); |
| 3361 | $options = is_array( $options ) ? $options : array(); |
| 3362 | $clean = array(); |
| 3363 | foreach ( $this->list_options() as $name => $default ) { |
| 3364 | $value = array_key_exists( $name, $options ) ? $options[ $name ] : ( array_key_exists( $name, $current ) ? $current[ $name ] : $default ); |
| 3365 | $clean[ $name ] = $this->sanitize_option_value( $name, $value, $default ); |
| 3366 | } |
| 3367 | update_option( $this->option_name, $clean, false ); |
| 3368 | $options = $this->sanitize_options(); |
| 3369 | return $options; |
| 3370 | } |
| 3371 | |
| 3372 | public function sanitize_scan_config( $config ) { |
| 3373 | $options = $this->get_all_options(); |
| 3374 | $config = is_array( $config ) ? $config : array(); |
| 3375 | $names = array( |
| 3376 | 'content', 'filesystem_content', 'media_library', 'images_only', 'attach_is_use', |
| 3377 | 'thumbnails_only', 'dirs_filter', 'files_filter', 'shortcodes_disabled', |
| 3378 | 'thumbnail_force_issues', 'posts_buffer', 'medias_buffer', 'analysis_buffer', |
| 3379 | 'uploads_file_buffer', 'refs_buffer', 'delay', 'analysis_document_limit', |
| 3380 | ); |
| 3381 | $snapshot = array(); |
| 3382 | foreach ( $names as $name ) { |
| 3383 | $default = array_key_exists( $name, $options ) ? $options[ $name ] : null; |
| 3384 | $value = array_key_exists( $name, $config ) ? $config[ $name ] : $default; |
| 3385 | $snapshot[ $name ] = $this->sanitize_option_value( $name, $value, $default ); |
| 3386 | } |
| 3387 | return $snapshot; |
| 3388 | } |
| 3389 | |
| 3390 | // The accepted range of every numeric option. Stored values and the values proposed by |
| 3391 | // Meow_WPMC_Buffers are clamped through this same table, so "too big" is defined once. |
| 3392 | public function option_ranges() { |
| 3393 | return array( |
| 3394 | 'medias_buffer' => array( 1, 500 ), |
| 3395 | 'posts_buffer' => array( 1, 100 ), |
| 3396 | 'analysis_buffer' => array( 1, 500 ), |
| 3397 | 'file_op_buffer' => array( 1, 100 ), |
| 3398 | 'uploads_file_buffer' => array( 10, 1000 ), |
| 3399 | 'delay' => array( 0, 10000 ), |
| 3400 | 'refs_buffer' => array( 10, 1000 ), |
| 3401 | 'posts_per_page' => array( 5, 100 ), |
| 3402 | ); |
| 3403 | } |
| 3404 | |
| 3405 | private function sanitize_option_value( $name, $value, $default ) { |
| 3406 | $boolean_options = array( |
| 3407 | 'content', 'filesystem_content', 'media_library', 'live_content', 'debuglogs', |
| 3408 | 'images_only', 'attach_is_use', 'thumbnails_only', 'hide_thumbnails', 'hide_warning', |
| 3409 | 'skip_trash', 'shortcodes_disabled', 'output_buffer_cleaning_disabled', |
| 3410 | 'php_error_logs', 'clean_uninstall', 'repair_mode', 'expert_mode', |
| 3411 | 'mcp_support', |
| 3412 | ); |
| 3413 | if ( in_array( $name, $boolean_options, true ) ) { |
| 3414 | return rest_sanitize_boolean( $value ); |
| 3415 | } |
| 3416 | |
| 3417 | $ranges = $this->option_ranges(); |
| 3418 | if ( isset( $ranges[ $name ] ) ) { |
| 3419 | $number = is_numeric( $value ) ? (int) $value : (int) $default; |
| 3420 | return max( $ranges[ $name ][0], min( $ranges[ $name ][1], $number ) ); |
| 3421 | } |
| 3422 | |
| 3423 | if ( $name === 'analysis_document_limit' ) { |
| 3424 | $number = is_numeric( $value ) ? (int) $value : (int) $default; |
| 3425 | // -1 is the sentinel for "no limit"; any other value is floored at 1 MB. |
| 3426 | return $number === -1 ? -1 : max( 1024 * 1024, $number ); |
| 3427 | } |
| 3428 | |
| 3429 | if ( $name === 'method' ) { |
| 3430 | $value = sanitize_key( $value ); |
| 3431 | return in_array( $value, array( 'media', 'files', 'duplicates', 'optimize_thumbnails' ), true ) ? $value : 'media'; |
| 3432 | } |
| 3433 | if ( in_array( $name, array( 'dirs_filter', 'files_filter' ), true ) ) { |
| 3434 | $value = is_string( $value ) ? trim( $value ) : ''; |
| 3435 | return $value !== '' && @preg_match( $value, '' ) === false ? '' : $value; |
| 3436 | } |
| 3437 | if ( $name === 'thumbnail_force_issues' ) { |
| 3438 | return is_array( $value ) ? array_values( array_unique( array_map( 'sanitize_key', $value ) ) ) : array(); |
| 3439 | } |
| 3440 | if ( $name === 'logs_path' ) { |
| 3441 | return is_string( $value ) ? sanitize_text_field( $value ) : null; |
| 3442 | } |
| 3443 | return is_scalar( $value ) || is_array( $value ) ? $value : $default; |
| 3444 | } |
| 3445 | |
| 3446 | // Upgrade from the old way of storing options to the new way. |
| 3447 | function check_options( $options = [] ) { |
| 3448 | $plugin_options = $this->list_options(); |
| 3449 | $options = empty( $options ) ? [] : $options; |
| 3450 | $clean_options = array_intersect_key( $options, $plugin_options ); |
| 3451 | $hasChanges = count( $clean_options ) !== count( $options ); |
| 3452 | $options = $clean_options; |
| 3453 | foreach ( $plugin_options as $option => $default ) { |
| 3454 | // The option already exists |
| 3455 | if ( isset( $options[$option] ) ) { |
| 3456 | continue; |
| 3457 | } |
| 3458 | // The option does not exist, so we need to add it. |
| 3459 | // Let's use the old value if any, or the default value. |
| 3460 | $options[$option] = get_option( 'wpmc_' . $option, $default ); |
| 3461 | delete_option( 'wpmc_' . $option ); |
| 3462 | $hasChanges = true; |
| 3463 | } |
| 3464 | if ( $hasChanges ) { |
| 3465 | update_option( $this->option_name , $options ); |
| 3466 | } |
| 3467 | |
| 3468 | // Runtime information is not persisted with settings. In particular, scan |
| 3469 | // checkpoints must never be copied into the options row. |
| 3470 | $options['thumbnail_sizes'] = $this->get_thumbnail_sizes(); |
| 3471 | global $mwai; |
| 3472 | $options['mwai_has_mcp'] = !empty( $mwai ) && method_exists( $mwai, 'hasMCP' ) && $mwai->hasMCP(); |
| 3473 | |
| 3474 | return $options; |
| 3475 | } |
| 3476 | |
| 3477 | // Validate and keep the options clean and logical. |
| 3478 | function sanitize_options() { |
| 3479 | $options = get_option( $this->option_name, array() ); |
| 3480 | $options = is_array( $options ) ? $options : array(); |
| 3481 | $clean = array(); |
| 3482 | foreach ( $this->list_options() as $name => $default ) { |
| 3483 | $value = array_key_exists( $name, $options ) ? $options[ $name ] : $default; |
| 3484 | $clean[ $name ] = $this->sanitize_option_value( $name, $value, $default ); |
| 3485 | } |
| 3486 | if ( $clean !== $options ) { |
| 3487 | update_option( $this->option_name, $clean, false ); |
| 3488 | } |
| 3489 | return $this->check_options( $clean ); |
| 3490 | } |
| 3491 | |
| 3492 | #endregion |
| 3493 | } |
| 3494 | |
| 3495 | // Check the DB. If does not exist, let's create it. |
| 3496 | function wpmc_check_database() { |
| 3497 | wpmc_create_database(); |
| 3498 | } |
| 3499 | |
| 3500 | function wpmc_create_database() { |
| 3501 | global $wpdb; |
| 3502 | $table_name = $wpdb->prefix . "mclean_scan"; |
| 3503 | $charset_collate = $wpdb->get_charset_collate(); |
| 3504 | $sql = "CREATE TABLE $table_name ( |
| 3505 | id BIGINT(20) NOT NULL AUTO_INCREMENT, |
| 3506 | run_id BIGINT(20) UNSIGNED NOT NULL DEFAULT 0, |
| 3507 | time DATETIME NULL, |
| 3508 | type TINYINT(1) NOT NULL, |
| 3509 | postId BIGINT(20) NULL, |
| 3510 | path TEXT NULL, |
| 3511 | path_hash CHAR(64) NULL, |
| 3512 | manifest LONGTEXT NULL, |
| 3513 | size BIGINT(20) UNSIGNED NULL, |
| 3514 | ignored TINYINT(1) NOT NULL DEFAULT 0, |
| 3515 | deleted TINYINT(1) NOT NULL DEFAULT 0, |
| 3516 | issue VARCHAR(191) NOT NULL, |
| 3517 | parentId BIGINT(20) NULL, |
| 3518 | PRIMARY KEY (id), |
| 3519 | KEY run_state_index (run_id, deleted, ignored, id), |
| 3520 | KEY run_post_index (run_id, postId), |
| 3521 | KEY run_path_index (run_id, path_hash), |
| 3522 | KEY run_parent_index (run_id, parentId) |
| 3523 | ) " . $charset_collate . ";" ; |
| 3524 | require_once( ABSPATH . 'wp-admin/includes/upgrade.php' ); |
| 3525 | dbDelta( $sql ); |
| 3526 | |
| 3527 | $table_name = $wpdb->prefix . "mclean_refs"; |
| 3528 | $charset_collate = $wpdb->get_charset_collate(); |
| 3529 | // This key doesn't work on too many installs because of the 'Specified key was too long' issue |
| 3530 | // KEY mediaLookUp (mediaId, mediaUrl) |
| 3531 | $sql = "CREATE TABLE $table_name ( |
| 3532 | id BIGINT(20) NOT NULL AUTO_INCREMENT, |
| 3533 | run_id BIGINT(20) UNSIGNED NOT NULL DEFAULT 0, |
| 3534 | mediaId BIGINT(20) NULL, |
| 3535 | mediaUrl TEXT NULL, |
| 3536 | mediaUrl_hash CHAR(64) NULL, |
| 3537 | originType VARCHAR(191) NOT NULL, |
| 3538 | origin TEXT NULL, |
| 3539 | parentId BIGINT(20) NULL, |
| 3540 | ref_hash VARCHAR(32) NULL, |
| 3541 | PRIMARY KEY (id), |
| 3542 | KEY run_media_index (run_id, mediaId), |
| 3543 | KEY run_url_index (run_id, mediaUrl_hash), |
| 3544 | KEY run_origin_index (run_id, originType), |
| 3545 | UNIQUE KEY run_ref_hash_unique (run_id, ref_hash) |
| 3546 | ) " . $charset_collate . ";"; |
| 3547 | require_once( ABSPATH . 'wp-admin/includes/upgrade.php' ); |
| 3548 | dbDelta( $sql ); |
| 3549 | |
| 3550 | $runs_table = $wpdb->prefix . 'mclean_runs'; |
| 3551 | $sql = "CREATE TABLE $runs_table ( |
| 3552 | id BIGINT(20) UNSIGNED NOT NULL AUTO_INCREMENT, |
| 3553 | owner_id BIGINT(20) UNSIGNED NOT NULL DEFAULT 0, |
| 3554 | method VARCHAR(32) NOT NULL, |
| 3555 | status VARCHAR(24) NOT NULL, |
| 3556 | phase VARCHAR(64) NOT NULL, |
| 3557 | config LONGTEXT NULL, |
| 3558 | checkpoint LONGTEXT NULL, |
| 3559 | counters LONGTEXT NULL, |
| 3560 | errors LONGTEXT NULL, |
| 3561 | error_count INT(10) UNSIGNED NOT NULL DEFAULT 0, |
| 3562 | created_at DATETIME NOT NULL, |
| 3563 | updated_at DATETIME NOT NULL, |
| 3564 | heartbeat_at DATETIME NOT NULL, |
| 3565 | finished_at DATETIME NULL, |
| 3566 | published_at DATETIME NULL, |
| 3567 | PRIMARY KEY (id), |
| 3568 | KEY status_heartbeat_index (status, heartbeat_at), |
| 3569 | KEY owner_status_index (owner_id, status) |
| 3570 | ) " . $charset_collate . ";"; |
| 3571 | dbDelta( $sql ); |
| 3572 | |
| 3573 | $work_table = $wpdb->prefix . 'mclean_work'; |
| 3574 | $sql = "CREATE TABLE $work_table ( |
| 3575 | id BIGINT(20) UNSIGNED NOT NULL AUTO_INCREMENT, |
| 3576 | run_id BIGINT(20) UNSIGNED NOT NULL, |
| 3577 | phase VARCHAR(64) NOT NULL, |
| 3578 | target_type VARCHAR(32) NOT NULL, |
| 3579 | target_key TEXT NOT NULL, |
| 3580 | target_hash CHAR(64) NOT NULL, |
| 3581 | cursor_value BIGINT(20) UNSIGNED NOT NULL DEFAULT 0, |
| 3582 | status VARCHAR(24) NOT NULL DEFAULT 'pending', |
| 3583 | attempts SMALLINT(5) UNSIGNED NOT NULL DEFAULT 0, |
| 3584 | last_error TEXT NULL, |
| 3585 | snapshot_token CHAR(64) NULL, |
| 3586 | updated_at DATETIME NOT NULL, |
| 3587 | PRIMARY KEY (id), |
| 3588 | UNIQUE KEY run_target_unique (run_id, phase, target_hash), |
| 3589 | KEY lease_index (run_id, phase, status, id) |
| 3590 | ) " . $charset_collate . ";"; |
| 3591 | dbDelta( $sql ); |
| 3592 | |
| 3593 | $operations_table = $wpdb->prefix . 'mclean_operations'; |
| 3594 | $sql = "CREATE TABLE $operations_table ( |
| 3595 | id BIGINT(20) UNSIGNED NOT NULL AUTO_INCREMENT, |
| 3596 | run_id BIGINT(20) UNSIGNED NOT NULL, |
| 3597 | issue_id BIGINT(20) UNSIGNED NOT NULL, |
| 3598 | operation VARCHAR(24) NOT NULL, |
| 3599 | state VARCHAR(32) NOT NULL, |
| 3600 | request_key VARCHAR(64) NOT NULL, |
| 3601 | manifest LONGTEXT NULL, |
| 3602 | error_code VARCHAR(64) NULL, |
| 3603 | error_message TEXT NULL, |
| 3604 | created_at DATETIME NOT NULL, |
| 3605 | updated_at DATETIME NOT NULL, |
| 3606 | PRIMARY KEY (id), |
| 3607 | UNIQUE KEY request_issue_unique (request_key, issue_id, operation), |
| 3608 | KEY run_state_index (run_id, state, id) |
| 3609 | ) " . $charset_collate . ";"; |
| 3610 | dbDelta( $sql ); |
| 3611 | |
| 3612 | $duplicates_table = $wpdb->prefix . 'mclean_duplicates'; |
| 3613 | $sql = "CREATE TABLE $duplicates_table ( |
| 3614 | id BIGINT(20) UNSIGNED NOT NULL AUTO_INCREMENT, |
| 3615 | run_id BIGINT(20) UNSIGNED NOT NULL, |
| 3616 | media_id BIGINT(20) UNSIGNED NOT NULL, |
| 3617 | path TEXT NOT NULL, |
| 3618 | path_hash CHAR(64) NOT NULL, |
| 3619 | size BIGINT(20) UNSIGNED NOT NULL DEFAULT 0, |
| 3620 | content_hash CHAR(64) NULL, |
| 3621 | is_canonical TINYINT(1) NOT NULL DEFAULT 0, |
| 3622 | PRIMARY KEY (id), |
| 3623 | UNIQUE KEY run_media_unique (run_id, media_id), |
| 3624 | KEY run_size_index (run_id, size), |
| 3625 | KEY run_content_index (run_id, content_hash) |
| 3626 | ) " . $charset_collate . ";"; |
| 3627 | dbDelta( $sql ); |
| 3628 | } |
| 3629 | |
| 3630 | function wpmc_remove_database() { |
| 3631 | global $wpdb; |
| 3632 | $table_name1 = $wpdb->prefix . "mclean_scan"; |
| 3633 | $table_name2 = $wpdb->prefix . "mclean_refs"; |
| 3634 | $table_name3 = $wpdb->prefix . "wpmcleaner"; |
| 3635 | $table_name4 = $wpdb->prefix . "mclean_runs"; |
| 3636 | $table_name5 = $wpdb->prefix . "mclean_work"; |
| 3637 | $table_name6 = $wpdb->prefix . "mclean_operations"; |
| 3638 | $table_name7 = $wpdb->prefix . "mclean_duplicates"; |
| 3639 | $sql = "DROP TABLE IF EXISTS $table_name1, $table_name2, $table_name3, $table_name4, $table_name5, $table_name6, $table_name7;"; |
| 3640 | $wpdb->query( $sql ); |
| 3641 | } |
| 3642 | |
| 3643 | #region Install / Uninstall |
| 3644 | |
| 3645 | /* |
| 3646 | INSTALL / UNINSTALL |
| 3647 | */ |
| 3648 | |
| 3649 | function wpmc_install() { |
| 3650 | $previous_schema = (int) get_option( Meow_WPMC_Runs::SCHEMA_OPTION, 0 ); |
| 3651 | wpmc_create_database(); |
| 3652 | $runs = new Meow_WPMC_Runs( null ); |
| 3653 | if ( $runs->tables_exist( true ) ) { |
| 3654 | if ( $previous_schema < 2 ) { |
| 3655 | delete_transient( 'wpmc_progress' ); |
| 3656 | } |
| 3657 | // Schema 2 to 4 force-disabled shortcode analysis, which caused false |
| 3658 | // positives for rendered shortcodes (Foo Gallery and others). Undo it once. |
| 3659 | if ( $previous_schema >= 2 && $previous_schema < 5 ) { |
| 3660 | $options = get_option( 'wpmc_options', array() ); |
| 3661 | if ( is_array( $options ) && !empty( $options['shortcodes_disabled'] ) ) { |
| 3662 | $options['shortcodes_disabled'] = false; |
| 3663 | update_option( 'wpmc_options', $options, false ); |
| 3664 | } |
| 3665 | } |
| 3666 | update_option( Meow_WPMC_Runs::SCHEMA_OPTION, Meow_WPMC_Runs::SCHEMA_VERSION, false ); |
| 3667 | } |
| 3668 | } |
| 3669 | |
| 3670 | function wpmc_reset () { |
| 3671 | wpmc_remove_database(); |
| 3672 | delete_option( Meow_WPMC_Runs::ACTIVE_RUN_OPTION ); |
| 3673 | delete_option( Meow_WPMC_Runs::LOCK_OPTION ); |
| 3674 | delete_option( Meow_WPMC_Runs::SCHEMA_OPTION ); |
| 3675 | delete_option( Meow_WPMC_Runs::SCHEMA_LOCK_OPTION ); |
| 3676 | wpmc_create_database(); |
| 3677 | $runs = new Meow_WPMC_Runs( null ); |
| 3678 | if ( $runs->tables_exist( true ) ) { |
| 3679 | update_option( Meow_WPMC_Runs::SCHEMA_OPTION, Meow_WPMC_Runs::SCHEMA_VERSION, false ); |
| 3680 | } |
| 3681 | } |
| 3682 | |
| 3683 | #endregion |
| 3684 |