class-file-tree-producer.php
3 weeks ago
class-hmac-client.php
3 weeks ago
class-hmac-server.php
3 weeks ago
class-http-server.php
3 weeks ago
class-mysql-dump-producer.php
3 weeks ago
class-pdo-polyfill.php
3 weeks ago
class-sqlite-driver-pdo.php
3 weeks ago
class-staged-artifacts.php
3 weeks ago
class-staged-endpoints.php
3 weeks ago
class-staged-push-stream-protocol.php
3 weeks ago
class-wpdb-driver-pdo.php
3 weeks ago
export.php
3 weeks ago
utils.php
3 weeks ago
class-file-tree-producer.php
668 lines
| 1 | <?php |
| 2 | /** |
| 3 | * Streams a provided list of filesystem paths in sorted order with |
| 4 | * cursor-based resumption. The caller must pass the same paths array |
| 5 | * on each request when resuming from a cursor. Callers are responsible |
| 6 | * for encoding cursors for transport (e.g. base64 for HTTP headers). |
| 7 | */ |
| 8 | class FileTreeProducer |
| 9 | { |
| 10 | const DEFAULT_CHUNK_SIZE = 5 * 1024 * 1024; |
| 11 | |
| 12 | const PHASE_STREAMING = "streaming"; |
| 13 | const PHASE_FINISHED = "finished"; |
| 14 | |
| 15 | /** @var array */ |
| 16 | private $directories; |
| 17 | /** @var int */ |
| 18 | private $chunk_size; |
| 19 | /** @var bool */ |
| 20 | private $index_only; |
| 21 | /** @var string|null */ |
| 22 | private $filesystem_root; |
| 23 | |
| 24 | /** @var string */ |
| 25 | private $phase; |
| 26 | /** @var array|null */ |
| 27 | private $current_chunk = null; |
| 28 | |
| 29 | /** Explicit list of paths to stream, sorted on first use. */ |
| 30 | /** @var array */ |
| 31 | private $paths; |
| 32 | /** @var bool */ |
| 33 | private $paths_sorted = false; |
| 34 | /** @var bool */ |
| 35 | private $paths_positioned = false; |
| 36 | /** Ephemeral index into $paths; NOT stored in cursor. */ |
| 37 | /** @var int */ |
| 38 | private $paths_position = 0; |
| 39 | |
| 40 | /** State for the file currently being streamed in chunks. */ |
| 41 | private $streaming_file_handle = null; |
| 42 | |
| 43 | public function __destruct() |
| 44 | { |
| 45 | if ($this->streaming_file_handle !== null) { |
| 46 | fclose($this->streaming_file_handle); |
| 47 | $this->streaming_file_handle = null; |
| 48 | } |
| 49 | } |
| 50 | /** @var int */ |
| 51 | private $streaming_file_offset = 0; |
| 52 | /** @var array|null */ |
| 53 | private $current_file_meta = null; |
| 54 | |
| 55 | /** Tracks the last emitted path for cursor generation. */ |
| 56 | /** @var string|null */ |
| 57 | private $last_emitted_path = null; |
| 58 | /** @var int|null */ |
| 59 | private $last_emitted_ctime = null; |
| 60 | |
| 61 | /** |
| 62 | * @param string|array $directories Root directories to scan. |
| 63 | * @param array $options { |
| 64 | * @type int $chunk_size Bytes per file chunk (default 5MB). |
| 65 | * @type bool $index_only Emit index entries instead of file contents. |
| 66 | * @type string $cursor JSON cursor string for resumption. |
| 67 | * @type array $paths Paths to stream (required). |
| 68 | * } |
| 69 | */ |
| 70 | public function __construct($directories, array $options = []) |
| 71 | { |
| 72 | $this->directories = $this->normalize_directories($directories); |
| 73 | $this->chunk_size = $options["chunk_size"] ?? self::DEFAULT_CHUNK_SIZE; |
| 74 | $this->index_only = $options["index_only"] ?? false; |
| 75 | |
| 76 | if (!isset($options["paths"]) || !is_array($options["paths"])) { |
| 77 | throw new InvalidArgumentException( |
| 78 | "The 'paths' option is required and must be an array" |
| 79 | ); |
| 80 | } |
| 81 | $this->paths = $options["paths"]; |
| 82 | |
| 83 | if (isset($options["cursor"])) { |
| 84 | $this->initialize_from_cursor($options["cursor"]); |
| 85 | } else { |
| 86 | $this->initialize_new(); |
| 87 | } |
| 88 | } |
| 89 | |
| 90 | /** |
| 91 | * Sets up a fresh traversal from the beginning. |
| 92 | */ |
| 93 | private function initialize_new(): void |
| 94 | { |
| 95 | $this->phase = self::PHASE_STREAMING; |
| 96 | $dirs = $this->directories; |
| 97 | sort($dirs, SORT_STRING); |
| 98 | $this->filesystem_root = $dirs[0] ?? "/"; |
| 99 | |
| 100 | $this->current_chunk = null; |
| 101 | $this->streaming_file_handle = null; |
| 102 | $this->streaming_file_offset = 0; |
| 103 | $this->current_file_meta = null; |
| 104 | $this->last_emitted_path = null; |
| 105 | $this->last_emitted_ctime = null; |
| 106 | $this->paths_sorted = false; |
| 107 | $this->paths_positioned = false; |
| 108 | $this->paths_position = 0; |
| 109 | } |
| 110 | |
| 111 | /** |
| 112 | * Restores producer state from a JSON cursor string. |
| 113 | * |
| 114 | * Cursor format is minimal: (path, ctime, byte_offset) |
| 115 | * - path: the file/dir/symlink we were processing or just finished |
| 116 | * - ctime: the ctime of the file when we started (for change detection) |
| 117 | * - bytes: byte offset within the file (0 if finished or non-file) |
| 118 | * |
| 119 | * On resume, position within the paths array is determined by binary |
| 120 | * search based on the path. This ensures correctness even when the |
| 121 | * paths array changes between requests. |
| 122 | */ |
| 123 | private function initialize_from_cursor(string $cursor_json): void |
| 124 | { |
| 125 | $cursor = json_decode($cursor_json, true); |
| 126 | if ($cursor === null && json_last_error() !== JSON_ERROR_NONE) { |
| 127 | throw new InvalidArgumentException( |
| 128 | "Invalid cursor format: " . json_last_error_msg() |
| 129 | ); |
| 130 | } |
| 131 | |
| 132 | $this->phase = $cursor["phase"] ?? self::PHASE_STREAMING; |
| 133 | $this->filesystem_root = isset($cursor["root"]) |
| 134 | ? base64_decode($cursor["root"]) |
| 135 | : ($this->directories[0] ?? "/"); |
| 136 | $this->current_chunk = null; |
| 137 | $this->streaming_file_handle = null; |
| 138 | $this->paths_sorted = false; |
| 139 | $this->paths_positioned = false; |
| 140 | $this->paths_position = 0; |
| 141 | |
| 142 | if ($this->phase !== self::PHASE_STREAMING) { |
| 143 | $this->phase = self::PHASE_FINISHED; |
| 144 | return; |
| 145 | } |
| 146 | |
| 147 | $path = isset($cursor["path"]) ? base64_decode($cursor["path"]) : null; |
| 148 | $ctime = $cursor["ctime"] ?? null; |
| 149 | $byte_offset = $cursor["bytes"] ?? 0; |
| 150 | |
| 151 | $this->last_emitted_path = null; |
| 152 | $this->last_emitted_ctime = null; |
| 153 | |
| 154 | if ($path !== null && $byte_offset > 0) { |
| 155 | // Resuming mid-file. |
| 156 | clearstatcache(true, $path); |
| 157 | $size = @filesize($path); |
| 158 | if ($size === false) { |
| 159 | // File disappeared; treat as completed. |
| 160 | $this->current_file_meta = null; |
| 161 | $this->streaming_file_offset = 0; |
| 162 | $this->last_emitted_path = $path; |
| 163 | } else { |
| 164 | $this->current_file_meta = [ |
| 165 | "path" => $path, |
| 166 | "ctime" => $ctime, |
| 167 | "size" => $size, |
| 168 | ]; |
| 169 | $this->streaming_file_offset = $byte_offset; |
| 170 | $this->last_emitted_path = $path; |
| 171 | } |
| 172 | } else { |
| 173 | $this->current_file_meta = null; |
| 174 | $this->streaming_file_offset = 0; |
| 175 | $this->last_emitted_path = $path; |
| 176 | } |
| 177 | // Position within paths array will be resolved by binary search |
| 178 | // when get_next_path_entry() is first called |
| 179 | } |
| 180 | |
| 181 | /** @param string|array $directories */ |
| 182 | private function normalize_directories($directories): array |
| 183 | { |
| 184 | if (is_string($directories)) { |
| 185 | return [rtrim($directories, "/")]; |
| 186 | } |
| 187 | return array_map(function ($d) { |
| 188 | return rtrim($d, "/"); |
| 189 | }, $directories); |
| 190 | } |
| 191 | |
| 192 | /** |
| 193 | * Advances to the next chunk. Returns false when finished. |
| 194 | */ |
| 195 | public function next_chunk(): bool |
| 196 | { |
| 197 | if ($this->phase === self::PHASE_FINISHED) { |
| 198 | return false; |
| 199 | } |
| 200 | |
| 201 | $this->stream_step(); |
| 202 | return $this->phase !== self::PHASE_FINISHED; |
| 203 | } |
| 204 | |
| 205 | /** |
| 206 | * Produces the next chunk: file data, index entry, directory, or symlink. |
| 207 | */ |
| 208 | private function stream_step(): void |
| 209 | { |
| 210 | if ($this->current_file_meta !== null) { |
| 211 | $this->stream_file_chunk($this->current_file_meta); |
| 212 | return; |
| 213 | } |
| 214 | |
| 215 | while (true) { |
| 216 | // get_next_server_file() may set current_chunk directly for |
| 217 | // symlinks and directories, so clear it before each iteration. |
| 218 | $this->current_chunk = null; |
| 219 | |
| 220 | $server_file = $this->get_next_server_file(); |
| 221 | |
| 222 | if ($this->current_chunk !== null) { |
| 223 | return; |
| 224 | } |
| 225 | |
| 226 | if ($server_file === null) { |
| 227 | $this->phase = self::PHASE_FINISHED; |
| 228 | $this->current_chunk = null; |
| 229 | return; |
| 230 | } |
| 231 | |
| 232 | if ($this->index_only) { |
| 233 | $this->emit_index_chunk($server_file); |
| 234 | return; |
| 235 | } |
| 236 | |
| 237 | $this->stream_file_chunk($server_file); |
| 238 | return; |
| 239 | } |
| 240 | } |
| 241 | |
| 242 | /** |
| 243 | * Emits an index entry chunk without streaming file contents. |
| 244 | */ |
| 245 | private function emit_index_chunk(array $file): void |
| 246 | { |
| 247 | $this->current_chunk = [ |
| 248 | "type" => "index", |
| 249 | "path" => $file["path"], |
| 250 | "ctime" => $file["ctime"], |
| 251 | "size" => $file["size"], |
| 252 | ]; |
| 253 | $this->last_emitted_path = $file["path"]; |
| 254 | $this->last_emitted_ctime = $file["ctime"]; |
| 255 | $this->current_file_meta = null; |
| 256 | } |
| 257 | |
| 258 | /** |
| 259 | * Returns the next file entry, or sets current_chunk for non-file entries |
| 260 | * (symlinks, directories, missing paths) and returns null. |
| 261 | */ |
| 262 | private function get_next_server_file(): ?array |
| 263 | { |
| 264 | return $this->get_next_path_entry(); |
| 265 | } |
| 266 | |
| 267 | /** |
| 268 | * Returns the next entry from the paths array. |
| 269 | * |
| 270 | * Paths are sorted on first access. Position is determined by binary |
| 271 | * search based on last_emitted_path, not by a stored index. This ensures |
| 272 | * correctness even when the paths array changes between requests. |
| 273 | */ |
| 274 | private function get_next_path_entry(): ?array |
| 275 | { |
| 276 | if (!$this->paths_sorted) { |
| 277 | sort($this->paths, SORT_STRING); |
| 278 | $this->paths_sorted = true; |
| 279 | } |
| 280 | |
| 281 | if (!$this->paths_positioned) { |
| 282 | if ($this->last_emitted_path !== null) { |
| 283 | $this->paths_position = $this->binary_search_next( |
| 284 | $this->paths, |
| 285 | $this->last_emitted_path |
| 286 | ); |
| 287 | } else { |
| 288 | $this->paths_position = 0; |
| 289 | } |
| 290 | $this->paths_positioned = true; |
| 291 | } |
| 292 | |
| 293 | while ($this->paths_position < count($this->paths)) { |
| 294 | $path = $this->paths[$this->paths_position]; |
| 295 | $this->paths_position++; |
| 296 | |
| 297 | $resolved_path = $this->resolve_path($path); |
| 298 | if ($resolved_path === null) { |
| 299 | // Path doesn't exist or isn't accessible, emit as missing |
| 300 | $this->last_emitted_path = $path; |
| 301 | $this->last_emitted_ctime = null; |
| 302 | $this->current_chunk = [ |
| 303 | "type" => "missing", |
| 304 | "path" => $path, |
| 305 | ]; |
| 306 | return null; |
| 307 | } |
| 308 | |
| 309 | $info = $this->lstat_path($resolved_path); |
| 310 | if ($info === null) { |
| 311 | continue; |
| 312 | } |
| 313 | |
| 314 | if ($info["type"] === "link") { |
| 315 | $target = @readlink($resolved_path); |
| 316 | $this->last_emitted_path = $resolved_path; |
| 317 | $this->last_emitted_ctime = $info["ctime"]; |
| 318 | $this->current_chunk = [ |
| 319 | "type" => "symlink", |
| 320 | "path" => $resolved_path, |
| 321 | "target" => $target !== false ? $target : "", |
| 322 | "ctime" => $info["ctime"] ?? 0, |
| 323 | ]; |
| 324 | return null; |
| 325 | } |
| 326 | |
| 327 | if ($info["type"] === "dir") { |
| 328 | $this->last_emitted_path = $resolved_path; |
| 329 | $this->last_emitted_ctime = $info["ctime"] ?? null; |
| 330 | $this->current_chunk = [ |
| 331 | "type" => "directory", |
| 332 | "path" => $resolved_path, |
| 333 | "ctime" => $info["ctime"] ?? 0, |
| 334 | ]; |
| 335 | return null; |
| 336 | } |
| 337 | |
| 338 | if ($info["type"] === "file") { |
| 339 | $ctime = $info["ctime"]; |
| 340 | $size = $info["size"]; |
| 341 | if ($ctime === null || $size === null) { |
| 342 | continue; |
| 343 | } |
| 344 | $this->current_file_meta = [ |
| 345 | "path" => $resolved_path, |
| 346 | "ctime" => $ctime, |
| 347 | "size" => $size, |
| 348 | ]; |
| 349 | $this->streaming_file_offset = 0; |
| 350 | return $this->current_file_meta; |
| 351 | } |
| 352 | } |
| 353 | |
| 354 | return null; |
| 355 | } |
| 356 | |
| 357 | /** |
| 358 | * Resolves a path that might be relative to one of the root directories. |
| 359 | * |
| 360 | * Uses both file_exists() and is_link() because file_exists() follows |
| 361 | * symlinks and returns false for broken symlinks, but the symlink |
| 362 | * itself is still a valid filesystem entry we want to stream. |
| 363 | */ |
| 364 | private function resolve_path(string $path): ?string |
| 365 | { |
| 366 | if ($path === "") { |
| 367 | return null; |
| 368 | } |
| 369 | |
| 370 | clearstatcache(true, $path); |
| 371 | if ($path[0] === "/" && (file_exists($path) || is_link($path))) { |
| 372 | return $path; |
| 373 | } |
| 374 | |
| 375 | foreach ($this->directories as $dir) { |
| 376 | $candidate = $dir . "/" . ltrim($path, "/"); |
| 377 | clearstatcache(true, $candidate); |
| 378 | if (file_exists($candidate) || is_link($candidate)) { |
| 379 | return $candidate; |
| 380 | } |
| 381 | } |
| 382 | |
| 383 | if ($path[0] === "/") { |
| 384 | return null; |
| 385 | } |
| 386 | |
| 387 | return null; |
| 388 | } |
| 389 | |
| 390 | /** |
| 391 | * Reads the next chunk from the current file and performs post-read |
| 392 | * change detection via ctime comparison. |
| 393 | */ |
| 394 | private function stream_file_chunk(array $file): void |
| 395 | { |
| 396 | if ($this->streaming_file_handle === null) { |
| 397 | clearstatcache(true, $file["path"]); |
| 398 | $pre_stat = @lstat($file["path"]); |
| 399 | if ($pre_stat === false || (($pre_stat["mode"] & 0170000) !== 0100000)) { |
| 400 | $this->streaming_file_handle = null; |
| 401 | $this->current_file_meta = null; |
| 402 | $this->current_chunk = [ |
| 403 | "type" => "error", |
| 404 | "error_type" => $pre_stat === false ? "file_missing" : "file_changed", |
| 405 | "path" => $file["path"], |
| 406 | "message" => $pre_stat === false |
| 407 | ? "File disappeared before read" |
| 408 | : "Path is no longer a regular file", |
| 409 | ]; |
| 410 | $this->last_emitted_path = $file["path"]; |
| 411 | $this->last_emitted_ctime = $file["ctime"]; |
| 412 | return; |
| 413 | } |
| 414 | |
| 415 | $this->streaming_file_handle = @fopen($file["path"], "r"); |
| 416 | if (!$this->streaming_file_handle) { |
| 417 | $this->streaming_file_handle = null; |
| 418 | $this->current_file_meta = null; |
| 419 | $this->current_chunk = [ |
| 420 | "type" => "error", |
| 421 | "error_type" => "file_open", |
| 422 | "path" => $file["path"], |
| 423 | "message" => "Failed to open file", |
| 424 | ]; |
| 425 | $this->last_emitted_path = $file["path"]; |
| 426 | $this->last_emitted_ctime = $file["ctime"]; |
| 427 | return; |
| 428 | } |
| 429 | if ($this->streaming_file_offset > 0) { |
| 430 | $seek_result = fseek( |
| 431 | $this->streaming_file_handle, |
| 432 | $this->streaming_file_offset |
| 433 | ); |
| 434 | if ($seek_result === -1) { |
| 435 | fclose($this->streaming_file_handle); |
| 436 | $this->streaming_file_handle = null; |
| 437 | $this->current_file_meta = null; |
| 438 | $this->current_chunk = [ |
| 439 | "type" => "error", |
| 440 | "error_type" => "file_seek", |
| 441 | "path" => $file["path"], |
| 442 | "message" => "Failed to seek to offset {$this->streaming_file_offset}", |
| 443 | ]; |
| 444 | $this->last_emitted_path = $file["path"]; |
| 445 | $this->last_emitted_ctime = $file["ctime"]; |
| 446 | $this->streaming_file_offset = 0; |
| 447 | return; |
| 448 | } |
| 449 | } |
| 450 | } |
| 451 | |
| 452 | $data = fread($this->streaming_file_handle, $this->chunk_size); |
| 453 | if (false === $data || ("" === $data && $file["size"] !== 0)) { |
| 454 | fclose($this->streaming_file_handle); |
| 455 | $this->streaming_file_handle = null; |
| 456 | $this->streaming_file_offset = 0; |
| 457 | $this->last_emitted_path = $file["path"]; |
| 458 | $this->last_emitted_ctime = $file["ctime"]; |
| 459 | $this->current_file_meta = null; |
| 460 | $this->current_chunk = [ |
| 461 | "type" => "error", |
| 462 | "error_type" => "file_read", |
| 463 | "path" => $file["path"], |
| 464 | "message" => "Failed to read file", |
| 465 | ]; |
| 466 | return; |
| 467 | } |
| 468 | |
| 469 | $offset = $this->streaming_file_offset; |
| 470 | $this->streaming_file_offset += strlen($data); |
| 471 | |
| 472 | $is_first = $offset === 0; |
| 473 | $is_last = feof($this->streaming_file_handle); |
| 474 | |
| 475 | $changed = false; |
| 476 | $change_ctime = null; |
| 477 | $change_size = null; |
| 478 | $error_type = "file_changed"; |
| 479 | |
| 480 | // Detect whether the file changed while we were reading it. |
| 481 | clearstatcache(true, $file["path"]); |
| 482 | $stat = @stat($file["path"]); |
| 483 | if ($stat === false) { |
| 484 | $changed = true; |
| 485 | $error_type = "file_missing"; |
| 486 | } else { |
| 487 | $now_ctime = $stat["ctime"]; |
| 488 | if ($now_ctime !== $file["ctime"]) { |
| 489 | $changed = true; |
| 490 | $change_ctime = $now_ctime; |
| 491 | } |
| 492 | } |
| 493 | |
| 494 | if ($changed) { |
| 495 | fclose($this->streaming_file_handle); |
| 496 | $this->streaming_file_handle = null; |
| 497 | $this->streaming_file_offset = 0; |
| 498 | $this->last_emitted_path = $file["path"]; |
| 499 | $this->last_emitted_ctime = $file["ctime"]; |
| 500 | $this->current_file_meta = null; |
| 501 | $this->current_chunk = [ |
| 502 | "type" => "error", |
| 503 | "error_type" => $error_type, |
| 504 | "path" => $file["path"], |
| 505 | "message" => |
| 506 | $error_type === "file_missing" |
| 507 | ? "File disappeared during stream" |
| 508 | : "File changed during stream", |
| 509 | "expected_ctime" => $file["ctime"], |
| 510 | "actual_ctime" => $change_ctime, |
| 511 | ]; |
| 512 | return; |
| 513 | } |
| 514 | |
| 515 | $this->current_chunk = [ |
| 516 | "type" => "file", |
| 517 | "path" => $file["path"], |
| 518 | "data" => $data, |
| 519 | "size" => $file["size"], |
| 520 | "ctime" => $file["ctime"], |
| 521 | "offset" => $offset, |
| 522 | "is_first_chunk" => $is_first, |
| 523 | "is_last_chunk" => $is_last, |
| 524 | "file_changed" => $changed, |
| 525 | "change_ctime" => $change_ctime, |
| 526 | "change_size" => $change_size, |
| 527 | ]; |
| 528 | |
| 529 | if ($is_last) { |
| 530 | fclose($this->streaming_file_handle); |
| 531 | $this->streaming_file_handle = null; |
| 532 | $this->streaming_file_offset = 0; |
| 533 | $this->last_emitted_path = $file["path"]; |
| 534 | $this->last_emitted_ctime = $file["ctime"]; |
| 535 | $this->current_file_meta = null; |
| 536 | } |
| 537 | } |
| 538 | |
| 539 | /** |
| 540 | * Returns the chunk produced by the last call to next_chunk(). |
| 541 | */ |
| 542 | public function get_current_chunk(): ?array |
| 543 | { |
| 544 | return $this->current_chunk; |
| 545 | } |
| 546 | |
| 547 | /** |
| 548 | * Serializes state into a JSON cursor string. |
| 549 | * |
| 550 | * Cursor format is minimal: (path, ctime, byte_offset) |
| 551 | * - path: last emitted path, or current file being streamed |
| 552 | * - ctime: ctime of file when we started reading (for change detection) |
| 553 | * - bytes: byte offset within the current file (0 if not mid-file) |
| 554 | * |
| 555 | * No traversal stack or list indices are stored. On resume, position is |
| 556 | * determined by binary search based on the path. This ensures correctness |
| 557 | * even when the filesystem changes between requests. |
| 558 | */ |
| 559 | public function get_reentrancy_cursor(): string |
| 560 | { |
| 561 | if ($this->phase === self::PHASE_FINISHED) { |
| 562 | return json_encode([ |
| 563 | "phase" => self::PHASE_FINISHED, |
| 564 | "root" => base64_encode($this->filesystem_root), |
| 565 | ]); |
| 566 | } |
| 567 | |
| 568 | $cursor = [ |
| 569 | "phase" => $this->phase, |
| 570 | "root" => base64_encode($this->filesystem_root), |
| 571 | ]; |
| 572 | |
| 573 | if ($this->current_file_meta !== null) { |
| 574 | $cursor["path"] = base64_encode($this->current_file_meta["path"]); |
| 575 | $cursor["ctime"] = $this->current_file_meta["ctime"]; |
| 576 | $cursor["bytes"] = $this->streaming_file_offset; |
| 577 | } else if ($this->last_emitted_path !== null) { |
| 578 | $cursor["path"] = base64_encode($this->last_emitted_path); |
| 579 | $cursor["ctime"] = $this->last_emitted_ctime; |
| 580 | $cursor["bytes"] = 0; |
| 581 | } |
| 582 | return json_encode($cursor); |
| 583 | } |
| 584 | |
| 585 | /** |
| 586 | * Returns progress metadata for logging and UI updates. |
| 587 | */ |
| 588 | public function get_progress(): array |
| 589 | { |
| 590 | $progress = [ |
| 591 | "phase" => $this->phase, |
| 592 | ]; |
| 593 | |
| 594 | if ($this->phase === self::PHASE_STREAMING) { |
| 595 | if ($this->last_emitted_path !== null) { |
| 596 | $progress["last_path"] = base64_encode($this->last_emitted_path); |
| 597 | } |
| 598 | if ($this->current_file_meta) { |
| 599 | $file = $this->current_file_meta; |
| 600 | $progress["current_file"] = [ |
| 601 | "path" => base64_encode($file["path"]), |
| 602 | "size" => $file["size"], |
| 603 | "bytes_read" => $this->streaming_file_offset, |
| 604 | ]; |
| 605 | } |
| 606 | } |
| 607 | |
| 608 | return $progress; |
| 609 | } |
| 610 | |
| 611 | /** |
| 612 | * Returns the filesystem root path. |
| 613 | */ |
| 614 | public function get_filesystem_root(): ?string |
| 615 | { |
| 616 | return $this->filesystem_root; |
| 617 | } |
| 618 | |
| 619 | /** |
| 620 | * Returns the index of the first entry strictly greater than $last |
| 621 | * in a sorted array, using binary search. |
| 622 | */ |
| 623 | private function binary_search_next(array $entries, string $last): int |
| 624 | { |
| 625 | $low = 0; |
| 626 | $high = count($entries); |
| 627 | while ($low < $high) { |
| 628 | $mid = intdiv($low + $high, 2); |
| 629 | if (strcmp($entries[$mid], $last) <= 0) { |
| 630 | $low = $mid + 1; |
| 631 | } else { |
| 632 | $high = $mid; |
| 633 | } |
| 634 | } |
| 635 | return $low; |
| 636 | } |
| 637 | |
| 638 | /** |
| 639 | * Classifies a path as file, dir, link, or other via a single lstat() call. |
| 640 | * |
| 641 | * @return array|null {type: 'file'|'dir'|'link'|'other', ctime: ?int, size: ?int} |
| 642 | */ |
| 643 | private function lstat_path(string $path): ?array |
| 644 | { |
| 645 | clearstatcache(true, $path); |
| 646 | $stat = @lstat($path); |
| 647 | if ($stat === false) { |
| 648 | return null; |
| 649 | } |
| 650 | |
| 651 | $mode = $stat["mode"] & 0170000; |
| 652 | $type = "other"; |
| 653 | if ($mode === 0120000) { |
| 654 | $type = "link"; |
| 655 | } elseif ($mode === 0040000) { |
| 656 | $type = "dir"; |
| 657 | } elseif ($mode === 0100000) { |
| 658 | $type = "file"; |
| 659 | } |
| 660 | |
| 661 | return [ |
| 662 | "type" => $type, |
| 663 | "ctime" => isset($stat["ctime"]) ? (int) $stat["ctime"] : null, |
| 664 | "size" => isset($stat["size"]) ? (int) $stat["size"] : null, |
| 665 | ]; |
| 666 | } |
| 667 | } |
| 668 |