← All changes
|
vendor/wp-php-toolkit/reprint-server/src/class-file-index-processor.php
+1149
-0
16.2-beta
→
16.3
View file →
| @@ -1,0 +1,1149 @@ | ||
| 1 | +<?php | |
| 2 | + | |
| 3 | +namespace WordPress\Reprint\Server; | |
| 4 | + | |
| 5 | +require_once __DIR__ . '/utils.php'; | |
| 6 | + | |
| 7 | +use InvalidArgumentException; | |
| 8 | +use LogicException; | |
| 9 | + | |
| 10 | +// phpcs:disable WordPress.Security.EscapeOutput.ExceptionNotEscaped -- Traversal failures become API or CLI values, never HTML output. | |
| 11 | + | |
| 12 | +/** | |
| 13 | + * Walks filesystem paths for the file index one resumable step at a time. | |
| 14 | + * | |
| 15 | + * The processor owns directory traversal and path inspection. Callers own | |
| 16 | + * what happens to the resulting index entries: the HTTP endpoint batches and | |
| 17 | + * encodes them, while local push indexing will write them to its JSONL file. | |
| 18 | + * | |
| 19 | + * A cursor contains the active directory stack and the last settled name in | |
| 20 | + * each directory. Reopening from that cursor rescans only the active | |
| 21 | + * directories and continues after those names. Directory names remain sorted | |
| 22 | + * exactly as they were in endpoint_file_index() before this extraction. | |
| 23 | + * | |
| 24 | + * One step either produces zero or more index entries for one filesystem path, | |
| 25 | + * skips one path, reports one directory failure, or finishes one directory. | |
| 26 | + * Following a symlink may produce additional intermediate-symlink entries in | |
| 27 | + * the same step because they share the path's cursor boundary. | |
| 28 | + * | |
| 29 | + * Directory names are still read with scandir(). This deliberately preserves | |
| 30 | + * the endpoint's established ordering and cursor behavior. It also means one | |
| 31 | + * unusually wide directory is held in memory; changing that requires a | |
| 32 | + * separate traversal design rather than hiding it inside this extraction. | |
| 33 | + * | |
| 34 | + * @phpstan-type FileIndexRoot ( | |
| 35 | + * array{requested_path:string,resolved_path:string,type:'directory'|'file'|'symlink'} | |
| 36 | + * | array{requested_path:string,resolved_path:null,type:'missing'} | |
| 37 | + * ) | |
| 38 | + */ | |
| 39 | +final class FileIndexProcessor { | |
| 40 | + | |
| 41 | + const STATUS_INDEXED = "indexed"; | |
| 42 | + const STATUS_SKIPPED = "skipped"; | |
| 43 | + const STATUS_PATH_UNAVAILABLE = "path_unavailable"; | |
| 44 | + const STATUS_DIRECTORY_COMPLETE = "directory_complete"; | |
| 45 | + const STATUS_DIRECTORY_ERROR = "directory_error"; | |
| 46 | + | |
| 47 | + const STAT_TYPE_MASK = 0170000; | |
| 48 | + const STAT_TYPE_LINK = 0120000; | |
| 49 | + const STAT_TYPE_FILE = 0100000; | |
| 50 | + const STAT_TYPE_DIR = 0040000; | |
| 51 | + | |
| 52 | + /** @var array[] Configured file-index roots, in requested order. */ | |
| 53 | + private $roots; | |
| 54 | + | |
| 55 | + /** @var string[] Canonical directories selected by the request. */ | |
| 56 | + private $configured_directories; | |
| 57 | + | |
| 58 | + /** @var bool Whether directory symlinks may lead outside the allowed directories. */ | |
| 59 | + private $follow_symlinks; | |
| 60 | + | |
| 61 | + /** @var bool Whether generated caches and development files are included. */ | |
| 62 | + private $include_caches; | |
| 63 | + | |
| 64 | + /** @var string Canonical Reprint storage path omitted from the index, or an empty string. */ | |
| 65 | + private $storage_path; | |
| 66 | + | |
| 67 | + /** @var array[] Active directory stack, from scheduled roots to the current directory. */ | |
| 68 | + private $directory_stack; | |
| 69 | + | |
| 70 | + /** @var string Directory reported as X-Index-Dir for this traversal. */ | |
| 71 | + private $index_directory; | |
| 72 | + | |
| 73 | + /** @var array[] Intermediate symlinks emitted before a new traversal begins. */ | |
| 74 | + private $initial_index_entries; | |
| 75 | + | |
| 76 | + /** @var string[] Requested named roots still to index, one per step. */ | |
| 77 | + private $pending_named_roots = []; | |
| 78 | + | |
| 79 | + /** @var string[]|null Sorted names in the current directory. */ | |
| 80 | + private $current_directory_names = null; | |
| 81 | + | |
| 82 | + /** @var int Position of the next name in $current_directory_names. */ | |
| 83 | + private $current_directory_position = 0; | |
| 84 | + | |
| 85 | + /** @var string|null Current directory used for endpoint exception reporting. */ | |
| 86 | + private $current_directory = null; | |
| 87 | + | |
| 88 | + /** @var string|null Result of the most recent step. */ | |
| 89 | + private $step_status = null; | |
| 90 | + | |
| 91 | + /** @var array[] Entries produced by the most recent indexed step. */ | |
| 92 | + private $index_entries = []; | |
| 93 | + | |
| 94 | + /** @var array|null Directory failure produced by the most recent step. */ | |
| 95 | + private $directory_error = null; | |
| 96 | + | |
| 97 | + /** @var bool Whether close() has been called. */ | |
| 98 | + private $closed = false; | |
| 99 | + | |
| 100 | + /** | |
| 101 | + * Starts a traversal at the requested root and schedules the other roots. | |
| 102 | + * | |
| 103 | + * @param FileIndexRoot[] $roots Structured roots scheduled for this index. | |
| 104 | + * @param FileIndexRoot $start_root Root scheduled first. It may be an | |
| 105 | + * external directory reached by a followed link. | |
| 106 | + * @param bool $follow_symlinks Whether directory symlinks may lead outside the allowed directories. | |
| 107 | + * @param bool $include_caches Whether generated caches and development files are included. | |
| 108 | + * @param string $storage_path Reprint storage path omitted from the index, or an empty string. | |
| 109 | + * @return self New file-index processor. | |
| 110 | + */ | |
| 111 | + public static function start( | |
| 112 | + array $roots, | |
| 113 | + array $start_root, | |
| 114 | + bool $follow_symlinks, | |
| 115 | + bool $include_caches, | |
| 116 | + string $storage_path | |
| 117 | + ): self { | |
| 118 | + $roots = self::validate_roots($roots); | |
| 119 | + $start_root = self::validate_root($start_root); | |
| 120 | + $start_root_is_configured = false; | |
| 121 | + foreach ($roots as $root) { | |
| 122 | + if ($root["requested_path"] === $start_root["requested_path"]) { | |
| 123 | + $start_root_is_configured = true; | |
| 124 | + break; | |
| 125 | + } | |
| 126 | + } | |
| 127 | + if (!$start_root_is_configured && $start_root["type"] !== "directory") { | |
| 128 | + throw new InvalidArgumentException( | |
| 129 | + "File-index start root must be a configured root or a directory: {$start_root["requested_path"]}" | |
| 130 | + ); | |
| 131 | + } | |
| 132 | + | |
| 133 | + $configured_directories = self::resolved_directory_roots($roots, $follow_symlinks); | |
| 134 | + | |
| 135 | + // Visit the requested directory first, followed by every other root in | |
| 136 | + // stable byte order. Stable ordering makes a cursor independent of the | |
| 137 | + // order in which configuration discovered the additional roots. | |
| 138 | + $ordered_roots = [$start_root]; | |
| 139 | + $extra_roots = []; | |
| 140 | + foreach ($roots as $root) { | |
| 141 | + if ($root["requested_path"] !== $start_root["requested_path"]) { | |
| 142 | + $extra_roots[] = $root; | |
| 143 | + } | |
| 144 | + } | |
| 145 | + usort($extra_roots, static function (array $left, array $right): int { | |
| 146 | + return strcmp($left["requested_path"], $right["requested_path"]); | |
| 147 | + }); | |
| 148 | + $ordered_roots = array_merge($ordered_roots, $extra_roots); | |
| 149 | + | |
| 150 | + // A selected directory symlink has two responsibilities: emit its | |
| 151 | + // requested link entry and traverse its resolved target. Keep the | |
| 152 | + // two work lists separate so each follows its own coordinate. | |
| 153 | + $traversal_directories = self::resolved_directory_roots($ordered_roots, $follow_symlinks); | |
| 154 | + $pending_named_roots = []; | |
| 155 | + foreach ($ordered_roots as $root) { | |
| 156 | + if ($root["type"] !== "directory") { | |
| 157 | + $pending_named_roots[] = $root["requested_path"]; | |
| 158 | + } | |
| 159 | + } | |
| 160 | + | |
| 161 | + // The last stack element is visited next, so reverse the desired order | |
| 162 | + // while constructing the depth-first traversal stack. | |
| 163 | + $directory_stack = []; | |
| 164 | + for ($i = count($traversal_directories) - 1; $i >= 0; $i--) { | |
| 165 | + $directory_stack[] = [ | |
| 166 | + "dir" => $traversal_directories[$i], | |
| 167 | + "after" => null, | |
| 168 | + ]; | |
| 169 | + } | |
| 170 | + | |
| 171 | + // Keep parent-link discovery as the first traversal event. This | |
| 172 | + // preserves the endpoint's established ordering: any link entries | |
| 173 | + // found here must precede ordinary directory entries. | |
| 174 | + $initial_index_entries = []; | |
| 175 | + if ($follow_symlinks) { | |
| 176 | + foreach ($ordered_roots as $root) { | |
| 177 | + if ($root["type"] === "directory") { | |
| 178 | + $initial_index_entries = array_merge( | |
| 179 | + $initial_index_entries, | |
| 180 | + self::find_parent_symlinks($root["requested_path"]) | |
| 181 | + ); | |
| 182 | + } | |
| 183 | + } | |
| 184 | + } | |
| 185 | + | |
| 186 | + // X-Index-Dir names a directory, so a named path reports its parent. | |
| 187 | + $reported_index_directory = $start_root["type"] === "directory" | |
| 188 | + ? $start_root["resolved_path"] | |
| 189 | + : dirname($start_root["requested_path"]); | |
| 190 | + | |
| 191 | + return new self( | |
| 192 | + $roots, | |
| 193 | + $configured_directories, | |
| 194 | + $follow_symlinks, | |
| 195 | + $include_caches, | |
| 196 | + $storage_path, | |
| 197 | + $directory_stack, | |
| 198 | + $reported_index_directory, | |
| 199 | + $initial_index_entries, | |
| 200 | + $pending_named_roots | |
| 201 | + ); | |
| 202 | + } | |
| 203 | + | |
| 204 | + /** | |
| 205 | + * Resumes traversal from a cursor returned by get_cursor(). | |
| 206 | + * | |
| 207 | + * @param FileIndexRoot[] $roots Structured roots scheduled for this index. | |
| 208 | + * @param string $cursor_json JSON cursor returned by the preceding request. | |
| 209 | + * @param bool $follow_symlinks Whether directory symlinks may lead outside the allowed directories. | |
| 210 | + * @param bool $include_caches Whether generated caches and development files are included. | |
| 211 | + * @param string $storage_path Reprint storage path omitted from the index, or an empty string. | |
| 212 | + * @return self Resumed file-index processor. | |
| 213 | + */ | |
| 214 | + public static function resume( | |
| 215 | + array $roots, | |
| 216 | + string $cursor_json, | |
| 217 | + bool $follow_symlinks, | |
| 218 | + bool $include_caches, | |
| 219 | + string $storage_path | |
| 220 | + ): self { | |
| 221 | + $roots = self::validate_roots($roots); | |
| 222 | + $configured_directories = self::resolved_directory_roots($roots, $follow_symlinks); | |
| 223 | + | |
| 224 | + // A cursor is caller-held continuation state. Reject malformed JSON or | |
| 225 | + // a missing stack before any filesystem work begins. | |
| 226 | + $cursor = json_decode($cursor_json, true); | |
| 227 | + if (!is_array($cursor)) { | |
| 228 | + throw new InvalidArgumentException("Invalid index cursor format"); | |
| 229 | + } | |
| 230 | + if (!isset($cursor["stack"]) || !is_array($cursor["stack"])) { | |
| 231 | + throw new InvalidArgumentException("Index cursor missing stack"); | |
| 232 | + } | |
| 233 | + | |
| 234 | + // Paths are base64 text because filesystem names are arbitrary bytes | |
| 235 | + // while JSON strings must be valid UTF-8. Decode each frame back into | |
| 236 | + // the in-memory stack used by traversal. | |
| 237 | + $directory_stack = []; | |
| 238 | + foreach ($cursor["stack"] as $frame) { | |
| 239 | + if (!is_array($frame)) { | |
| 240 | + throw new InvalidArgumentException("Invalid index cursor frame"); | |
| 241 | + } | |
| 242 | + $encoded_directory = isset($frame["dir"]) ? $frame["dir"] : null; | |
| 243 | + if (!is_string($encoded_directory) || $encoded_directory === "") { | |
| 244 | + throw new InvalidArgumentException("Index cursor frame missing dir"); | |
| 245 | + } | |
| 246 | + $directory = base64_decode($encoded_directory, true); | |
| 247 | + if ($directory === false || $directory === "") { | |
| 248 | + throw new InvalidArgumentException("Index cursor frame has invalid dir encoding"); | |
| 249 | + } | |
| 250 | + | |
| 251 | + $encoded_after = array_key_exists("after", $frame) ? $frame["after"] : null; | |
| 252 | + if ($encoded_after !== null && !is_string($encoded_after)) { | |
| 253 | + throw new InvalidArgumentException("Index cursor frame invalid after"); | |
| 254 | + } | |
| 255 | + $after = null; | |
| 256 | + if ($encoded_after !== null) { | |
| 257 | + $after = base64_decode($encoded_after, true); | |
| 258 | + if ($after === false) { | |
| 259 | + throw new InvalidArgumentException("Index cursor frame has invalid after encoding"); | |
| 260 | + } | |
| 261 | + } | |
| 262 | + | |
| 263 | + $directory_stack[] = [ | |
| 264 | + "dir" => $directory, | |
| 265 | + "after" => $after, | |
| 266 | + ]; | |
| 267 | + } | |
| 268 | + | |
| 269 | + // Absent from cursors written before this field existed. | |
| 270 | + $pending_named_roots = []; | |
| 271 | + $encoded_path_roots = isset($cursor["paths"]) ? $cursor["paths"] : []; | |
| 272 | + if (!is_array($encoded_path_roots)) { | |
| 273 | + throw new InvalidArgumentException("Index cursor paths must be an array"); | |
| 274 | + } | |
| 275 | + foreach ($encoded_path_roots as $encoded_path_root) { | |
| 276 | + if (!is_string($encoded_path_root) || $encoded_path_root === "") { | |
| 277 | + throw new InvalidArgumentException("Index cursor path entry must be a non-empty string"); | |
| 278 | + } | |
| 279 | + $path_root = base64_decode($encoded_path_root, true); | |
| 280 | + if ($path_root === false || $path_root === "") { | |
| 281 | + throw new InvalidArgumentException("Index cursor path entry has invalid encoding"); | |
| 282 | + } | |
| 283 | + $pending_named_roots[] = $path_root; | |
| 284 | + } | |
| 285 | + | |
| 286 | + // During continuation, the active directory is the best description | |
| 287 | + // of what this request is indexing. A completed cursor has no active | |
| 288 | + // directory, so it falls back to the first configured root. | |
| 289 | + $index_directory = !empty($directory_stack) | |
| 290 | + ? $directory_stack[count($directory_stack) - 1]["dir"] | |
| 291 | + : ( isset($configured_directories[0]) ? $configured_directories[0] : "/" ); | |
| 292 | + | |
| 293 | + return new self( | |
| 294 | + $roots, | |
| 295 | + $configured_directories, | |
| 296 | + $follow_symlinks, | |
| 297 | + $include_caches, | |
| 298 | + $storage_path, | |
| 299 | + $directory_stack, | |
| 300 | + $index_directory, | |
| 301 | + [], | |
| 302 | + $pending_named_roots | |
| 303 | + ); | |
| 304 | + } | |
| 305 | + | |
| 306 | + /** | |
| 307 | + * Performs one traversal step. | |
| 308 | + * | |
| 309 | + * @return bool Whether a current step is available. False means traversal is complete. | |
| 310 | + */ | |
| 311 | + public function next_index_step(): bool | |
| 312 | + { | |
| 313 | + // A closed processor has discarded its retained directory names and | |
| 314 | + // cannot safely take another step. | |
| 315 | + if ($this->closed) { | |
| 316 | + throw new LogicException("Cannot take a file-index step after close()."); | |
| 317 | + } | |
| 318 | + | |
| 319 | + // Step accessors describe only the current call. Clear the preceding | |
| 320 | + // result before deciding which traversal event comes next. | |
| 321 | + $this->step_status = null; | |
| 322 | + $this->index_entries = []; | |
| 323 | + $this->directory_error = null; | |
| 324 | + | |
| 325 | + // Emit the parent links discovered during start() before descendants. | |
| 326 | + // They share one cursor boundary because traversal has not begun yet. | |
| 327 | + if (!empty($this->initial_index_entries)) { | |
| 328 | + $this->step_status = self::STATUS_INDEXED; | |
| 329 | + $this->index_entries = $this->initial_index_entries; | |
| 330 | + $this->initial_index_entries = []; | |
| 331 | + return true; | |
| 332 | + } | |
| 333 | + | |
| 334 | + // Index one selected named root before walking directories. This keeps | |
| 335 | + // each step bounded and makes its cursor boundary unambiguous. | |
| 336 | + if (!empty($this->pending_named_roots)) { | |
| 337 | + $this->index_next_named_root(); | |
| 338 | + return true; | |
| 339 | + } | |
| 340 | + | |
| 341 | + // Load the directory at the top of the stack only when no sorted name | |
| 342 | + // list is retained. A directory failure is itself a step; an empty | |
| 343 | + // stack means traversal has no further event. | |
| 344 | + if ($this->current_directory_names === null) { | |
| 345 | + if (!$this->open_current_directory()) { | |
| 346 | + return $this->step_status !== null; | |
| 347 | + } | |
| 348 | + } | |
| 349 | + | |
| 350 | + // Finishing a directory is observable so callers may stop at this | |
| 351 | + // exact cursor before the processor returns to its parent directory. | |
| 352 | + if ($this->current_directory_position >= count($this->current_directory_names)) { | |
| 353 | + array_pop($this->directory_stack); | |
| 354 | + $this->forget_current_directory_names(); | |
| 355 | + $this->step_status = self::STATUS_DIRECTORY_COMPLETE; | |
| 356 | + return true; | |
| 357 | + } | |
| 358 | + | |
| 359 | + // Select exactly one name for this step. Move the cursor first so every | |
| 360 | + // later outcome, including omission or disappearance, settles the name. | |
| 361 | + $frame_index = count($this->directory_stack) - 1; | |
| 362 | + $entry_name = $this->current_directory_names[$this->current_directory_position]; | |
| 363 | + ++$this->current_directory_position; | |
| 364 | + // Set "after" before any skip or stat call. The cursor must move past | |
| 365 | + // a cache path or a path that disappears between scandir() and lstat(), | |
| 366 | + // or every resumed request would inspect that same name again. | |
| 367 | + $this->directory_stack[$frame_index]["after"] = $entry_name; | |
| 368 | + $path = wp_join_unix_paths($this->current_directory, $entry_name); | |
| 369 | + | |
| 370 | + // Apply omissions before lstat() and before a directory can enter the | |
| 371 | + // stack. Omitted subtrees therefore cost no extra filesystem calls. | |
| 372 | + if (!$this->include_caches && self::path_is_default_skipped($path)) { | |
| 373 | + $this->step_status = self::STATUS_SKIPPED; | |
| 374 | + return true; | |
| 375 | + } | |
| 376 | + if ( | |
| 377 | + $this->storage_path !== "" | |
| 378 | + && \WordPress\Reprint\Server\path_is_same_as_or_descendant_of($path, $this->storage_path) | |
| 379 | + ) { | |
| 380 | + $this->step_status = self::STATUS_SKIPPED; | |
| 381 | + return true; | |
| 382 | + } | |
| 383 | + | |
| 384 | + // A name returned by scandir() may disappear before inspection. Its | |
| 385 | + // cursor is already settled, so continuation moves to the next name. | |
| 386 | + clearstatcache(true, $path); | |
| 387 | + $stat = @lstat($path); | |
| 388 | + if ($stat === false) { | |
| 389 | + $this->step_status = self::STATUS_PATH_UNAVAILABLE; | |
| 390 | + return true; | |
| 391 | + } | |
| 392 | + | |
| 393 | + $inspected_path = self::index_entries_for_path($path, $stat, $this->follow_symlinks); | |
| 394 | + $this->index_entries = $inspected_path["entries"]; | |
| 395 | + $type = $inspected_path["type"]; | |
| 396 | + $this->step_status = self::STATUS_INDEXED; | |
| 397 | + | |
| 398 | + // Depth-first traversal enters a new directory before returning to the | |
| 399 | + // remaining names in its parent. An exact scheduled root is already | |
| 400 | + // on the stack, and traversing an ancestor would expose paths outside | |
| 401 | + // the requested tree before entering that root again. | |
| 402 | + if ($type === "dir") { | |
| 403 | + $canonical_directory = realpath($path); | |
| 404 | + if ( | |
| 405 | + $canonical_directory === false | |
| 406 | + || !\WordPress\Reprint\Server\path_is_same_as_or_descendant_of($this->configured_directories, $canonical_directory) | |
| 407 | + ) { | |
| 408 | + $this->directory_stack[] = [ | |
| 409 | + "dir" => $path, | |
| 410 | + "after" => null, | |
| 411 | + ]; | |
| 412 | + $this->forget_current_directory_names(); | |
| 413 | + } | |
| 414 | + } | |
| 415 | + | |
| 416 | + return true; | |
| 417 | + } | |
| 418 | + | |
| 419 | + /** | |
| 420 | + * Returns what the most recent step did. | |
| 421 | + * | |
| 422 | + * @return string|null One STATUS_* value, or null before the first step and after completion. | |
| 423 | + */ | |
| 424 | + public function get_step_status() | |
| 425 | + { | |
| 426 | + return $this->step_status; | |
| 427 | + } | |
| 428 | + | |
| 429 | + /** | |
| 430 | + * Returns entries produced by the most recent indexed step. | |
| 431 | + * | |
| 432 | + * @return array[] File-index entries, normally containing exactly one entry. | |
| 433 | + */ | |
| 434 | + public function get_index_entries(): array | |
| 435 | + { | |
| 436 | + return $this->index_entries; | |
| 437 | + } | |
| 438 | + | |
| 439 | + /** | |
| 440 | + * Returns the directory failure produced by the most recent step. | |
| 441 | + * | |
| 442 | + * @return array|null { | |
| 443 | + * Directory failure, or null when the current step did not report one. | |
| 444 | + * | |
| 445 | + * @type string $error_type Protocol error type. | |
| 446 | + * @type string $path Filesystem path that could not be traversed. | |
| 447 | + * @type string $message Human-readable explanation. | |
| 448 | + * } | |
| 449 | + */ | |
| 450 | + public function get_directory_error() | |
| 451 | + { | |
| 452 | + return $this->directory_error; | |
| 453 | + } | |
| 454 | + | |
| 455 | + /** | |
| 456 | + * Returns a JSON-safe cursor for the next traversal step. | |
| 457 | + * | |
| 458 | + * @return array { | |
| 459 | + * File-index cursor. | |
| 460 | + * | |
| 461 | + * @type array[] $stack Active directories with base64-encoded path names. | |
| 462 | + * @type string[] $paths Base64-encoded named paths not yet inspected. | |
| 463 | + * } | |
| 464 | + */ | |
| 465 | + public function get_cursor(): array | |
| 466 | + { | |
| 467 | + $encoded_stack = []; | |
| 468 | + foreach ($this->directory_stack as $frame) { | |
| 469 | + $encoded_stack[] = [ | |
| 470 | + "dir" => base64_encode($frame["dir"]), | |
| 471 | + "after" => $frame["after"] !== null ? base64_encode($frame["after"]) : null, | |
| 472 | + ]; | |
| 473 | + } | |
| 474 | + $encoded_path_roots = []; | |
| 475 | + foreach ($this->pending_named_roots as $path_root) { | |
| 476 | + $encoded_path_roots[] = base64_encode($path_root); | |
| 477 | + } | |
| 478 | + return ["stack" => $encoded_stack, "paths" => $encoded_path_roots]; | |
| 479 | + } | |
| 480 | + | |
| 481 | + /** | |
| 482 | + * Returns the directory reported by the endpoint as X-Index-Dir. | |
| 483 | + * | |
| 484 | + * @return string Index directory for this traversal. | |
| 485 | + */ | |
| 486 | + public function get_index_directory(): string | |
| 487 | + { | |
| 488 | + return $this->index_directory; | |
| 489 | + } | |
| 490 | + | |
| 491 | + /** | |
| 492 | + * Returns the directory active during the most recent step. | |
| 493 | + * | |
| 494 | + * @return string|null Current directory, or null before traversal begins. | |
| 495 | + */ | |
| 496 | + public function get_current_directory() | |
| 497 | + { | |
| 498 | + return $this->current_directory; | |
| 499 | + } | |
| 500 | + | |
| 501 | + /** | |
| 502 | + * Releases in-memory directory data without performing another step. | |
| 503 | + */ | |
| 504 | + public function close(): void | |
| 505 | + { | |
| 506 | + if ($this->closed) { | |
| 507 | + return; | |
| 508 | + } | |
| 509 | + $this->closed = true; | |
| 510 | + $this->initial_index_entries = []; | |
| 511 | + $this->forget_current_directory_names(); | |
| 512 | + } | |
| 513 | + | |
| 514 | + /** | |
| 515 | + * Reports whether a path belongs to the established default skip set. | |
| 516 | + * | |
| 517 | + * @param string $path Filesystem path to classify. | |
| 518 | + * @return bool Whether the path should be omitted unless caches are included. | |
| 519 | + */ | |
| 520 | + public static function path_is_default_skipped(string $path): bool | |
| 521 | + { | |
| 522 | + // Sentinel slashes make component matches independent of whether the | |
| 523 | + // component appears at the beginning, middle, or end of the path. | |
| 524 | + $path_with_boundaries = "/" . trim($path, "/") . "/"; | |
| 525 | + | |
| 526 | + // These generated directories are limited to wp-content. A directory | |
| 527 | + // named cache elsewhere may contain user files and remains included. | |
| 528 | + // wpcomsh-cache is wp.com Atomic's filesystem cache shadow; wflogs is | |
| 529 | + // Wordfence request and scan data which can grow to gigabytes. | |
| 530 | + static $cache_directories = [ | |
| 531 | + "/wp-content/cache/", | |
| 532 | + "/wp-content/upgrade/", | |
| 533 | + "/wp-content/wpcomsh-cache/", | |
| 534 | + "/wp-content/wflogs/", | |
| 535 | + ]; | |
| 536 | + foreach ($cache_directories as $directory) { | |
| 537 | + if (strpos($path_with_boundaries, $directory) !== false) { | |
| 538 | + return true; | |
| 539 | + } | |
| 540 | + } | |
| 541 | + | |
| 542 | + // Version-control metadata and local development dependencies match | |
| 543 | + // complete path components. Similar names such as cache-control or | |
| 544 | + // node_modules-backup remain included. | |
| 545 | + static $skipped_components = [ | |
| 546 | + ".git", ".svn", ".hg", ".bzr", | |
| 547 | + "node_modules", | |
| 548 | + ".idea", ".vscode", | |
| 549 | + ".cache", ".npm", ".yarn", ".pnpm-store", | |
| 550 | + ]; | |
| 551 | + foreach ($skipped_components as $component) { | |
| 552 | + if (strpos($path_with_boundaries, "/" . $component . "/") !== false) { | |
| 553 | + return true; | |
| 554 | + } | |
| 555 | + } | |
| 556 | + | |
| 557 | + // Operating-system metadata matches only the basename. | |
| 558 | + $basename = basename($path); | |
| 559 | + static $skipped_basenames = [ | |
| 560 | + ".DS_Store", "._.DS_Store", | |
| 561 | + "Thumbs.db", "desktop.ini", "ehthumbs.db", | |
| 562 | + ]; | |
| 563 | + if (in_array($basename, $skipped_basenames, true)) { | |
| 564 | + return true; | |
| 565 | + } | |
| 566 | + // Editor and merge scratch files: Emacs locks and autosaves, trailing | |
| 567 | + // tildes, Vim swaps, backups, and conflict leftovers. | |
| 568 | + if ($basename !== "" && $basename[0] === "." && isset($basename[1]) && $basename[1] === "#") { | |
| 569 | + return true; | |
| 570 | + } | |
| 571 | + if (strlen($basename) >= 3 && $basename[0] === "#" && substr($basename, -1) === "#") { | |
| 572 | + return true; | |
| 573 | + } | |
| 574 | + if (preg_match("/(?:~|\\.(?:swp|swo|swn|bak|orig|rej))$/", $basename) === 1) { | |
| 575 | + return true; | |
| 576 | + } | |
| 577 | + | |
| 578 | + return false; | |
| 579 | + } | |
| 580 | + | |
| 581 | + /** | |
| 582 | + * Initializes common traversal state. | |
| 583 | + * | |
| 584 | + * @param array[] $roots Structured file-index roots. | |
| 585 | + * @param string[] $configured_directories Canonical directories selected by the request. | |
| 586 | + * @param bool $follow_symlinks Whether directory symlinks may leave the allowed directories. | |
| 587 | + * @param bool $include_caches Whether generated caches and development files are included. | |
| 588 | + * @param string $storage_path Reprint storage path omitted from the index, or an empty string. | |
| 589 | + * @param array[] $directory_stack Active directory stack. | |
| 590 | + * @param string $index_directory Directory reported by the endpoint. | |
| 591 | + * @param array[] $initial_index_entries Intermediate symlinks emitted before traversal. | |
| 592 | + * @param string[] $pending_named_roots Requested named roots still to index, one per step. | |
| 593 | + */ | |
| 594 | + private function __construct( | |
| 595 | + array $roots, | |
| 596 | + array $configured_directories, | |
| 597 | + bool $follow_symlinks, | |
| 598 | + bool $include_caches, | |
| 599 | + string $storage_path, | |
| 600 | + array $directory_stack, | |
| 601 | + string $index_directory, | |
| 602 | + array $initial_index_entries, | |
| 603 | + array $pending_named_roots = [] | |
| 604 | + ) { | |
| 605 | + $this->roots = $roots; | |
| 606 | + $this->configured_directories = $configured_directories; | |
| 607 | + $this->follow_symlinks = $follow_symlinks; | |
| 608 | + $this->include_caches = $include_caches; | |
| 609 | + $this->storage_path = self::canonical_storage_path($storage_path); | |
| 610 | + $this->directory_stack = $directory_stack; | |
| 611 | + $this->index_directory = $index_directory; | |
| 612 | + $this->initial_index_entries = $initial_index_entries; | |
| 613 | + $this->pending_named_roots = $pending_named_roots; | |
| 614 | + } | |
| 615 | + | |
| 616 | + /** | |
| 617 | + * Opens and positions the directory at the top of the traversal stack. | |
| 618 | + * | |
| 619 | + * @return bool Whether directory entries are ready for the current step. | |
| 620 | + */ | |
| 621 | + private function open_current_directory(): bool | |
| 622 | + { | |
| 623 | + // An empty stack is normal completion, not a directory failure. | |
| 624 | + if (empty($this->directory_stack)) { | |
| 625 | + return false; | |
| 626 | + } | |
| 627 | + | |
| 628 | + // The top frame names the next directory and the last name settled in | |
| 629 | + // it. Keep the directory available for any failure reported this step. | |
| 630 | + $frame_index = count($this->directory_stack) - 1; | |
| 631 | + $frame = $this->directory_stack[$frame_index]; | |
| 632 | + $this->current_directory = $frame["dir"]; | |
| 633 | + | |
| 634 | + // A directory may disappear while it waits on the stack. Remove that | |
| 635 | + // frame so a later call continues with its parent or the next root. | |
| 636 | + clearstatcache(true, $this->current_directory); | |
| 637 | + $canonical_directory = realpath($this->current_directory); | |
| 638 | + if ($canonical_directory === false || !is_dir($canonical_directory)) { | |
| 639 | + array_pop($this->directory_stack); | |
| 640 | + $this->directory_error = [ | |
| 641 | + "error_type" => "dir_open", | |
| 642 | + "path" => $this->current_directory, | |
| 643 | + "message" => "Directory does not exist or is not accessible", | |
| 644 | + ]; | |
| 645 | + $this->step_status = self::STATUS_DIRECTORY_ERROR; | |
| 646 | + return false; | |
| 647 | + } | |
| 648 | + | |
| 649 | + // When following links is disabled, every canonical directory must | |
| 650 | + // remain inside a configured root. Reject one that crosses that | |
| 651 | + // boundary, then continue with the remaining stack. | |
| 652 | + if ( | |
| 653 | + !$this->follow_symlinks | |
| 654 | + && !\WordPress\Reprint\Server\path_is_same_as_or_descendant_of($canonical_directory, $this->configured_directories) | |
| 655 | + ) { | |
| 656 | + array_pop($this->directory_stack); | |
| 657 | + $this->directory_error = [ | |
| 658 | + "error_type" => "dir_outside_root", | |
| 659 | + "path" => $canonical_directory, | |
| 660 | + "message" => "Directory is outside allowed roots", | |
| 661 | + ]; | |
| 662 | + $this->step_status = self::STATUS_DIRECTORY_ERROR; | |
| 663 | + return false; | |
| 664 | + } | |
| 665 | + | |
| 666 | + // Canonical paths keep split roots and followed symlinks in one | |
| 667 | + // namespace, matching the endpoint's previous traversal. | |
| 668 | + $this->directory_stack[$frame_index]["dir"] = $canonical_directory; | |
| 669 | + $this->current_directory = $canonical_directory; | |
| 670 | + | |
| 671 | + // scandir() supplies the stable byte order on which cursor resumption | |
| 672 | + // depends. Failure settles this directory rather than retrying it on | |
| 673 | + // every subsequent request. | |
| 674 | + clearstatcache(true, $canonical_directory); | |
| 675 | + $directory_names = @scandir($canonical_directory, SCANDIR_SORT_ASCENDING); | |
| 676 | + if ($directory_names === false) { | |
| 677 | + array_pop($this->directory_stack); | |
| 678 | + $this->directory_error = [ | |
| 679 | + "error_type" => "dir_open", | |
| 680 | + "path" => $canonical_directory, | |
| 681 | + "message" => "Failed to open directory", | |
| 682 | + ]; | |
| 683 | + $this->step_status = self::STATUS_DIRECTORY_ERROR; | |
| 684 | + return false; | |
| 685 | + } | |
| 686 | + | |
| 687 | + // Tests may change the scanned names to exercise traversal boundaries. | |
| 688 | + // Production traversal has no hook and uses scandir() results directly. | |
| 689 | + if (getenv("SITE_EXPORT_TEST_MODE") && function_exists("_e2e_call_hook")) { | |
| 690 | + $hook_arguments = [$canonical_directory, &$directory_names]; | |
| 691 | + _e2e_call_hook("test_hook_during_dir_scan", $hook_arguments); | |
| 692 | + } | |
| 693 | + | |
| 694 | + // Remove the two navigation names, then seek past the last settled name. | |
| 695 | + // Binary search keeps continuation cheap for unusually wide directories. | |
| 696 | + $this->current_directory_names = []; | |
| 697 | + foreach ($directory_names as $directory_name) { | |
| 698 | + if ($directory_name !== "." && $directory_name !== "..") { | |
| 699 | + $this->current_directory_names[] = $directory_name; | |
| 700 | + } | |
| 701 | + } | |
| 702 | + $this->current_directory_position = 0; | |
| 703 | + $after = isset($frame["after"]) ? $frame["after"] : null; | |
| 704 | + if ($after !== null && $after !== "") { | |
| 705 | + $this->current_directory_position = self::position_after_name( | |
| 706 | + $this->current_directory_names, | |
| 707 | + $after | |
| 708 | + ); | |
| 709 | + } | |
| 710 | + | |
| 711 | + return true; | |
| 712 | + } | |
| 713 | + | |
| 714 | + /** | |
| 715 | + * Drops the cached names for the current directory. | |
| 716 | + */ | |
| 717 | + private function forget_current_directory_names(): void | |
| 718 | + { | |
| 719 | + $this->current_directory_names = null; | |
| 720 | + $this->current_directory_position = 0; | |
| 721 | + } | |
| 722 | + | |
| 723 | + /** | |
| 724 | + * Returns a canonical storage path when the configured path exists. | |
| 725 | + * | |
| 726 | + * @param string $storage_path Configured Reprint storage path. | |
| 727 | + * @return string Canonical or normalized storage path, or an empty string. | |
| 728 | + */ | |
| 729 | + private static function canonical_storage_path(string $storage_path): string | |
| 730 | + { | |
| 731 | + // Reprint storage may live inside the document root on hosts that can | |
| 732 | + // write nowhere else. It must never enter an index or a push could | |
| 733 | + // copy or delete its own work while using it. Traversal canonicalizes | |
| 734 | + // directories with realpath(), so the comparison uses the same form. | |
| 735 | + // rtrim() also prevents a trailing slash from missing an exact match. | |
| 736 | + $storage_path = rtrim($storage_path, "/"); | |
| 737 | + if ($storage_path === "") { | |
| 738 | + return ""; | |
| 739 | + } | |
| 740 | + $canonical_storage_path = realpath($storage_path); | |
| 741 | + return $canonical_storage_path !== false ? $canonical_storage_path : $storage_path; | |
| 742 | + } | |
| 743 | + | |
| 744 | + /** | |
| 745 | + * Finds the first sorted directory name after a cursor name. | |
| 746 | + * | |
| 747 | + * @param string[] $directory_names Sorted directory names. | |
| 748 | + * @param string $after_name Last settled directory name. | |
| 749 | + * @return int Position of the next directory name. | |
| 750 | + */ | |
| 751 | + private static function position_after_name(array $directory_names, string $after_name): int | |
| 752 | + { | |
| 753 | + $low = 0; | |
| 754 | + $high = count($directory_names); | |
| 755 | + while ($low < $high) { | |
| 756 | + $middle = (int) ( ( $low + $high ) / 2 ); | |
| 757 | + if (strcmp($directory_names[$middle], $after_name) <= 0) { | |
| 758 | + $low = $middle + 1; | |
| 759 | + } else { | |
| 760 | + $high = $middle; | |
| 761 | + } | |
| 762 | + } | |
| 763 | + return $low; | |
| 764 | + } | |
| 765 | + | |
| 766 | + /** | |
| 767 | + * Indexes one requested named root using traversal's exclusions. | |
| 768 | + */ | |
| 769 | + private function index_next_named_root(): void | |
| 770 | + { | |
| 771 | + // Settle the cursor first so a skipped or vanished path is not retried. | |
| 772 | + $requested_path = array_shift($this->pending_named_roots); | |
| 773 | + $root = $this->find_root($requested_path); | |
| 774 | + if ($root === null) { | |
| 775 | + throw new InvalidArgumentException("Index cursor names a root absent from this request: {$requested_path}"); | |
| 776 | + } | |
| 777 | + | |
| 778 | + if ($root["type"] === "missing") { | |
| 779 | + $this->step_status = self::STATUS_PATH_UNAVAILABLE; | |
| 780 | + return; | |
| 781 | + } | |
| 782 | + | |
| 783 | + $path_root = $root["requested_path"]; | |
| 784 | + | |
| 785 | + if (!$this->include_caches && self::path_is_default_skipped($path_root)) { | |
| 786 | + $this->step_status = self::STATUS_SKIPPED; | |
| 787 | + return; | |
| 788 | + } | |
| 789 | + if ( | |
| 790 | + $this->storage_path !== "" | |
| 791 | + && \WordPress\Reprint\Server\path_is_same_as_or_descendant_of($path_root, $this->storage_path) | |
| 792 | + ) { | |
| 793 | + $this->step_status = self::STATUS_SKIPPED; | |
| 794 | + return; | |
| 795 | + } | |
| 796 | + | |
| 797 | + clearstatcache(true, $path_root); | |
| 798 | + $stat = @lstat($path_root); | |
| 799 | + if ($stat === false) { | |
| 800 | + $this->step_status = self::STATUS_PATH_UNAVAILABLE; | |
| 801 | + return; | |
| 802 | + } | |
| 803 | + | |
| 804 | + $entries = []; | |
| 805 | + if ($this->follow_symlinks) { | |
| 806 | + // Record links in the requested parent path. The inspected root may | |
| 807 | + // add links from its own symlink target, so keep both entry sets. | |
| 808 | + $entries = self::find_parent_symlinks(dirname($path_root)); | |
| 809 | + } | |
| 810 | + $inspected_path = self::index_entries_for_path($path_root, $stat, $this->follow_symlinks); | |
| 811 | + $resolved_target_was_indexed = $root["resolved_path"] !== null | |
| 812 | + && $this->resolved_target_was_indexed($root); | |
| 813 | + | |
| 814 | + // A selected symlink always remains at its requested path. When | |
| 815 | + // followed, its target content is emitted in the resolved-path namespace | |
| 816 | + // that normal traversal already uses. Two aliases may therefore share | |
| 817 | + // one target entry while both link entries remain present. | |
| 818 | + if (!( $root["type"] === "file" && $resolved_target_was_indexed )) { | |
| 819 | + $entries = array_merge($entries, $inspected_path["entries"]); | |
| 820 | + } | |
| 821 | + if ( | |
| 822 | + $this->follow_symlinks | |
| 823 | + && $root["type"] === "symlink" | |
| 824 | + && $root["resolved_path"] !== null | |
| 825 | + && !is_dir($root["resolved_path"]) | |
| 826 | + && !$resolved_target_was_indexed | |
| 827 | + ) { | |
| 828 | + clearstatcache(true, $root["resolved_path"]); | |
| 829 | + $target_stat = @lstat($root["resolved_path"]); | |
| 830 | + if (is_array($target_stat)) { | |
| 831 | + $target = self::index_entries_for_path($root["resolved_path"], $target_stat, false); | |
| 832 | + $entries = array_merge($entries, $target["entries"]); | |
| 833 | + } | |
| 834 | + } | |
| 835 | + if ( | |
| 836 | + $root["type"] === "file" | |
| 837 | + && $root["resolved_path"] !== null | |
| 838 | + && $root["resolved_path"] !== $root["requested_path"] | |
| 839 | + && !$resolved_target_was_indexed | |
| 840 | + ) { | |
| 841 | + // A regular root reached through no link normally has identical | |
| 842 | + // coordinates. Keep this branch for records supplied by callers | |
| 843 | + // which already normalized a resolved file root. | |
| 844 | + $entries = array_merge( | |
| 845 | + $entries, | |
| 846 | + self::index_entries_for_path($root["resolved_path"], $stat, false)["entries"] | |
| 847 | + ); | |
| 848 | + } | |
| 849 | + $this->index_entries = $entries; | |
| 850 | + $this->step_status = self::STATUS_INDEXED; | |
| 851 | + } | |
| 852 | + | |
| 853 | + /** Finds the current structured root by its requested path. */ | |
| 854 | + private function find_root(string $requested_path): ?array | |
| 855 | + { | |
| 856 | + foreach ($this->roots as $root) { | |
| 857 | + if ($root["requested_path"] === $requested_path) { | |
| 858 | + return $root; | |
| 859 | + } | |
| 860 | + } | |
| 861 | + return null; | |
| 862 | + } | |
| 863 | + | |
| 864 | + /** Whether an earlier named root already emitted this resolved target. */ | |
| 865 | + private function resolved_target_was_indexed(array $root): bool | |
| 866 | + { | |
| 867 | + foreach ($this->roots as $candidate) { | |
| 868 | + if ( | |
| 869 | + $candidate["requested_path"] !== $root["requested_path"] | |
| 870 | + && | |
| 871 | + $candidate["resolved_path"] === $root["resolved_path"] | |
| 872 | + && !in_array($candidate["requested_path"], $this->pending_named_roots, true) | |
| 873 | + ) { | |
| 874 | + return true; | |
| 875 | + } | |
| 876 | + } | |
| 877 | + return false; | |
| 878 | + } | |
| 879 | + | |
| 880 | + /** | |
| 881 | + * Validates root records produced by the endpoint resolver or local callers. | |
| 882 | + * | |
| 883 | + * @param array[] $roots File-index roots. | |
| 884 | + * @return FileIndexRoot[] | |
| 885 | + */ | |
| 886 | + private static function validate_roots(array $roots): array | |
| 887 | + { | |
| 888 | + $validated_roots = []; | |
| 889 | + foreach ($roots as $root) { | |
| 890 | + $validated_roots[] = self::validate_root($root); | |
| 891 | + } | |
| 892 | + return $validated_roots; | |
| 893 | + } | |
| 894 | + | |
| 895 | + /** | |
| 896 | + * @param mixed $root Candidate file-index root. | |
| 897 | + * @return FileIndexRoot Validated file-index root. | |
| 898 | + */ | |
| 899 | + private static function validate_root($root): array | |
| 900 | + { | |
| 901 | + if (!is_array($root) || !isset($root["requested_path"], $root["type"])) { | |
| 902 | + throw new InvalidArgumentException("File-index roots must contain requested_path and type"); | |
| 903 | + } | |
| 904 | + if (!is_string($root["requested_path"]) || !is_string($root["type"])) { | |
| 905 | + throw new InvalidArgumentException("File-index root fields have invalid types"); | |
| 906 | + } | |
| 907 | + $requested_path = $root["requested_path"]; | |
| 908 | + if ( | |
| 909 | + $requested_path === "" | |
| 910 | + || \WordPress\Reprint\Server\normalize_path($requested_path) !== $requested_path | |
| 911 | + ) { | |
| 912 | + throw new InvalidArgumentException("File-index root requested_path must be normalized"); | |
| 913 | + } | |
| 914 | + $resolved_path = $root["resolved_path"] ?? null; | |
| 915 | + if ($resolved_path !== null && !is_string($resolved_path)) { | |
| 916 | + throw new InvalidArgumentException("File-index root resolved_path has invalid type"); | |
| 917 | + } | |
| 918 | + if (!in_array($root["type"], ["directory", "file", "symlink", "missing"], true)) { | |
| 919 | + throw new InvalidArgumentException("File-index root type is invalid: {$root["type"]}"); | |
| 920 | + } | |
| 921 | + if ($root["type"] === "missing") { | |
| 922 | + if ($resolved_path !== null) { | |
| 923 | + throw new InvalidArgumentException( | |
| 924 | + "Missing file-index root has a resolved_path: {$requested_path}" | |
| 925 | + ); | |
| 926 | + } | |
| 927 | + } elseif ($resolved_path === null || $resolved_path === "") { | |
| 928 | + throw new InvalidArgumentException("File-index root missing resolved_path: {$requested_path}"); | |
| 929 | + } | |
| 930 | + return [ | |
| 931 | + "requested_path" => $requested_path, | |
| 932 | + "resolved_path" => $resolved_path, | |
| 933 | + "type" => $root["type"], | |
| 934 | + ]; | |
| 935 | + } | |
| 936 | + | |
| 937 | + /** | |
| 938 | + * Returns resolved directory roots, including followed directory links. | |
| 939 | + * | |
| 940 | + * @param FileIndexRoot[] $roots Structured roots. Each has requested_path, | |
| 941 | + * resolved_path, and type keys; type is directory, | |
| 942 | + * file, symlink, or missing. | |
| 943 | + * @return string[] Resolved directory paths. | |
| 944 | + */ | |
| 945 | + private static function resolved_directory_roots(array $roots, bool $follow_symlinks): array | |
| 946 | + { | |
| 947 | + $directories = []; | |
| 948 | + foreach ($roots as $root) { | |
| 949 | + if ( | |
| 950 | + $root["type"] === "directory" | |
| 951 | + || ( $follow_symlinks && $root["type"] === "symlink" && $root["resolved_path"] !== null && is_dir($root["resolved_path"]) ) | |
| 952 | + ) { | |
| 953 | + if (!in_array($root["resolved_path"], $directories, true)) { | |
| 954 | + $directories[] = $root["resolved_path"]; | |
| 955 | + } | |
| 956 | + } | |
| 957 | + } | |
| 958 | + return $directories; | |
| 959 | + } | |
| 960 | + | |
| 961 | + /** | |
| 962 | + * Builds the index entries describing one inspected path. | |
| 963 | + * | |
| 964 | + * @param string $path Absolute path already confirmed by lstat(). | |
| 965 | + * @param array $stat lstat() result for the path. | |
| 966 | + * @param bool $follow_symlinks Whether directory links may reveal intermediate links. | |
| 967 | + * @return array { | |
| 968 | + * @type array[] $entries Intermediate links, then the path's own entry. | |
| 969 | + * @type string $type One of file, link, dir, or other. | |
| 970 | + * } | |
| 971 | + */ | |
| 972 | + private static function index_entries_for_path( | |
| 973 | + string $path, | |
| 974 | + array $stat, | |
| 975 | + bool $follow_symlinks | |
| 976 | + ): array { | |
| 977 | + $mode = $stat["mode"] & self::STAT_TYPE_MASK; | |
| 978 | + $type = "file"; | |
| 979 | + $link_target = null; | |
| 980 | + $intermediate_symlinks = []; | |
| 981 | + if ($mode === self::STAT_TYPE_LINK) { | |
| 982 | + $type = "link"; | |
| 983 | + $resolved_symlink = self::resolve_symlink_target($path); | |
| 984 | + $link_target = $resolved_symlink["target"]; | |
| 985 | + if ($follow_symlinks) { | |
| 986 | + $intermediate_symlinks = $resolved_symlink["intermediates"]; | |
| 987 | + } | |
| 988 | + } elseif ($mode === self::STAT_TYPE_DIR) { | |
| 989 | + $type = "dir"; | |
| 990 | + } elseif ($mode !== self::STAT_TYPE_FILE) { | |
| 991 | + $type = "other"; | |
| 992 | + } | |
| 993 | + | |
| 994 | + // Directory size does not describe its descendants, so it is zeroed. | |
| 995 | + $item = [ | |
| 996 | + "path" => $path, | |
| 997 | + "ctime" => (int) ( isset($stat["ctime"]) ? $stat["ctime"] : 0 ), | |
| 998 | + "size" => $type === "file" || $type === "link" ? (int) ( isset($stat["size"]) ? $stat["size"] : 0 ) : 0, | |
| 999 | + "type" => $type, | |
| 1000 | + ]; | |
| 1001 | + if ($link_target !== null) { | |
| 1002 | + $item["target"] = $link_target; | |
| 1003 | + } | |
| 1004 | + if ($type === "dir") { | |
| 1005 | + // Actual empty directory, not a directory with all its children | |
| 1006 | + // excluded from the synchronization | |
| 1007 | + $directory_handle = @opendir($path); | |
| 1008 | + if ($directory_handle !== false) { | |
| 1009 | + $item["empty"] = true; | |
| 1010 | + while (true) { | |
| 1011 | + $directory_entry = readdir($directory_handle); | |
| 1012 | + if ($directory_entry === false) { | |
| 1013 | + break; | |
| 1014 | + } | |
| 1015 | + if ($directory_entry !== "." && $directory_entry !== "..") { | |
| 1016 | + $item["empty"] = false; | |
| 1017 | + break; | |
| 1018 | + } | |
| 1019 | + } | |
| 1020 | + closedir($directory_handle); | |
| 1021 | + } | |
| 1022 | + // If opendir() fails, leave "empty" absent. Pull reports the | |
| 1023 | + // directory error and push does not plan deletions from it. | |
| 1024 | + } | |
| 1025 | + | |
| 1026 | + // Intermediate links and the inspected path share one step because a | |
| 1027 | + // cursor cannot stop between them without losing one of the entries. | |
| 1028 | + $entries = $intermediate_symlinks; | |
| 1029 | + // Descendants imply non-empty parents: /a/file already implies /a. | |
| 1030 | + // Emit a directory only when it is empty or uninspectable, when no | |
| 1031 | + // descendant can establish that it exists. | |
| 1032 | + if ( | |
| 1033 | + $type !== "dir" | |
| 1034 | + || !isset($item["empty"]) | |
| 1035 | + || $item["empty"] | |
| 1036 | + ) { | |
| 1037 | + $entries[] = $item; | |
| 1038 | + } | |
| 1039 | + | |
| 1040 | + return ["entries" => $entries, "type" => $type]; | |
| 1041 | + } | |
| 1042 | + | |
| 1043 | + /** | |
| 1044 | + * Resolves a directory symlink and finds symlinks in its unresolved path. | |
| 1045 | + * | |
| 1046 | + * Managed WordPress hosts often chain symlinks: /srv may point to /, | |
| 1047 | + * /srv/wordpress may point to /wordpress, and readlink() may return a | |
| 1048 | + * relative path containing more symlinks. realpath() gives the final | |
| 1049 | + * canonical directory used for further indexing. File symlinks do not get | |
| 1050 | + * a canonical target because the pull client does not traverse them. | |
| 1051 | + * | |
| 1052 | + * realpath() skips the intermediate links, so the unresolved readlink() | |
| 1053 | + * path is also walked. Those intermediate entries let pull recreate the | |
| 1054 | + * complete path rather than only its final target. | |
| 1055 | + * | |
| 1056 | + * @param string $path Absolute path to the symlink. | |
| 1057 | + * @return array { | |
| 1058 | + * Symlink details for file indexing. | |
| 1059 | + * | |
| 1060 | + * @type string|null $target Canonical directory target, or null. | |
| 1061 | + * @type array[] $intermediates Symlinks encountered before that target. | |
| 1062 | + * } | |
| 1063 | + */ | |
| 1064 | + private static function resolve_symlink_target(string $path): array | |
| 1065 | + { | |
| 1066 | + // Only links ending at a directory need a canonical target because only | |
| 1067 | + // directories can add more traversal work. Broken, self-referential, | |
| 1068 | + // and file links remain ordinary link entries without a target. | |
| 1069 | + clearstatcache(true, $path); | |
| 1070 | + $resolved_target = @realpath($path); | |
| 1071 | + if ( | |
| 1072 | + $resolved_target === false | |
| 1073 | + || $resolved_target === $path | |
| 1074 | + || !is_dir($resolved_target) | |
| 1075 | + ) { | |
| 1076 | + return ["target" => null, "intermediates" => []]; | |
| 1077 | + } | |
| 1078 | + | |
| 1079 | + // realpath() jumps directly to the final directory. Walk the unresolved | |
| 1080 | + // target as well so links along that path are included in the index. | |
| 1081 | + $intermediates = []; | |
| 1082 | + $raw_target = @readlink($path); | |
| 1083 | + if ($raw_target !== false && $raw_target !== "") { | |
| 1084 | + if ($raw_target[0] !== "/") { | |
| 1085 | + $raw_target = wp_join_unix_paths(dirname($path), $raw_target); | |
| 1086 | + } | |
| 1087 | + // Resolve only textual dot segments. realpath() would skip the | |
| 1088 | + // intermediate links that this walk must inspect. | |
| 1089 | + $absolute_raw_target = \WordPress\Reprint\Server\normalize_path($raw_target); | |
| 1090 | + if ( | |
| 1091 | + $absolute_raw_target !== "" | |
| 1092 | + && $absolute_raw_target[0] === "/" | |
| 1093 | + && $absolute_raw_target !== $resolved_target | |
| 1094 | + ) { | |
| 1095 | + $intermediates = self::find_parent_symlinks($absolute_raw_target); | |
| 1096 | + } | |
| 1097 | + } | |
| 1098 | + | |
| 1099 | + return ["target" => $resolved_target, "intermediates" => $intermediates]; | |
| 1100 | + } | |
| 1101 | + | |
| 1102 | + /** | |
| 1103 | + * Returns symlinks found while walking the parents of an absolute path. | |
| 1104 | + * | |
| 1105 | + * For `/srv/wordpress/wp-content/plugins`, this inspects `/srv`, then | |
| 1106 | + * `/srv/wordpress`, and so on. After finding a link, traversal continues | |
| 1107 | + * from its canonical path. The walk is intentionally not recursive into | |
| 1108 | + * the final target; pull decides whether another directory needs indexing. | |
| 1109 | + * | |
| 1110 | + * @param string $absolute_path Absolute filesystem path. | |
| 1111 | + * @return array[] Intermediate symlink index entries. | |
| 1112 | + */ | |
| 1113 | + private static function find_parent_symlinks(string $absolute_path): array | |
| 1114 | + { | |
| 1115 | + $entries = []; | |
| 1116 | + $parts = explode("/", $absolute_path); | |
| 1117 | + $current = ""; | |
| 1118 | + | |
| 1119 | + // Keep the requested spelling while inspecting each parent. PHP follows | |
| 1120 | + // a parent link when checking the next component, so changing $current | |
| 1121 | + // to realpath() would turn later emitted links into resolved paths. | |
| 1122 | + foreach ($parts as $part) { | |
| 1123 | + if ($part === "") { | |
| 1124 | + $current = "/"; | |
| 1125 | + continue; | |
| 1126 | + } | |
| 1127 | + $current = wp_join_unix_paths($current, $part); | |
| 1128 | + if (!@is_link($current)) { | |
| 1129 | + continue; | |
| 1130 | + } | |
| 1131 | + | |
| 1132 | + // Preserve the link spelling returned by readlink(); pull needs it | |
| 1133 | + // to reconstruct the same link rather than only its final directory. | |
| 1134 | + $target = @readlink($current); | |
| 1135 | + if ($target !== false && $target !== "") { | |
| 1136 | + $stat = @lstat($current); | |
| 1137 | + $entries[] = [ | |
| 1138 | + "path" => $current, | |
| 1139 | + "ctime" => (int) ( is_array($stat) && isset($stat["ctime"]) ? $stat["ctime"] : 0 ), | |
| 1140 | + "size" => 0, | |
| 1141 | + "type" => "link", | |
| 1142 | + "target" => $target, | |
| 1143 | + "intermediate" => true, | |
| 1144 | + ]; | |
| 1145 | + } | |
| 1146 | + } | |
| 1147 | + return $entries; | |
| 1148 | + } | |
| 1149 | +} | |