| 1 |
<?php |
| 2 |
/** |
| 3 |
* Shared utility functions used by both export.php and import.php. |
| 4 |
* |
| 5 |
* These helpers live in a namespace so they don't collide with global |
| 6 |
* functions of the same name declared by third-party plugins or |
| 7 |
* WordPress drop-ins. Generic names like parse_size() or normalize_path() |
| 8 |
* are guaranteed to clash sooner or later if they sit in the global |
| 9 |
* namespace, and more than one plugin on a WordPress.com site loads this |
| 10 |
* file. |
| 11 |
* |
| 12 |
* Consumers require this file when they need its helpers. Keep this file |
| 13 |
* limited to guarded function declarations: do not add I/O, hooks, mutable |
| 14 |
* global state, or eager class definitions here. |
| 15 |
* |
| 16 |
* The two str_* polyfills stay global on purpose: they backfill functions |
| 17 |
* unavailable before PHP 8.0, so callers reach them via the global namespace |
| 18 |
* without a use-statement. |
| 19 |
*/ |
| 20 |
|
| 21 |
// Polyfill for PHP versions before 8.0, which lack str_starts_with(). |
| 22 |
namespace { |
| 23 |
if (!function_exists('str_starts_with')) { |
| 24 |
function str_starts_with(string $haystack, string $needle): bool { |
| 25 |
return $needle === '' || strncmp($haystack, $needle, strlen($needle)) === 0; |
| 26 |
} |
| 27 |
} |
| 28 |
|
| 29 |
// Polyfill for PHP versions before 8.0, which lack str_contains(). |
| 30 |
if (!function_exists('str_contains')) { |
| 31 |
function str_contains(string $haystack, string $needle): bool { |
| 32 |
return $needle === '' || strpos($haystack, $needle) !== false; |
| 33 |
} |
| 34 |
} |
| 35 |
} |
| 36 |
|
| 37 |
namespace WordPress\Reprint\Server { |
| 38 |
|
| 39 |
use InvalidArgumentException; |
| 40 |
use RuntimeException; |
| 41 |
|
| 42 |
// Every declaration below carries its own function_exists() guard, and the |
| 43 |
// guards are deliberately per-function rather than one block-wide check. |
| 44 |
// |
| 45 |
// Two plugins on the same site can each ship a copy of this package — wpcomsh |
| 46 |
// and Jetpack both do on WordPress.com — so one of them declares these |
| 47 |
// functions first and the other must not redeclare them. A single guard keyed |
| 48 |
// on one sentinel function would work only while both copies declare exactly |
| 49 |
// the same set: the moment one ships a helper the other lacks, the second copy |
| 50 |
// skips the whole block and the first call to that helper is a fatal. |
| 51 |
// Per-function guards degrade to "whoever loaded first wins the functions it |
| 52 |
// has, this copy supplies the rest" instead. |
| 53 |
// |
| 54 |
// The guards earn their keep when two copies are loaded: a monorepo checkout |
| 55 |
// that loads both packages/reprint-server/src/utils.php and the vendor/ mirror |
| 56 |
// of it, or a site still carrying reprint-exporter v0.1.47 |
| 57 |
// under its old package name. That older copy declares its helpers in |
| 58 |
// WordPress\Reprint\Exporter, a namespace nothing here uses any more, so it |
| 59 |
// cannot reach these names at all — but the guards cost nothing and they |
| 60 |
// document the hazard. |
| 61 |
// |
| 62 |
// Function bodies stay unindented inside their guards, matching how the |
| 63 |
// bracketed namespace blocks in this file are written. |
| 64 |
|
| 65 |
if (!function_exists(__NAMESPACE__ . '\\generate_random_bytes')) { |
| 66 |
/** |
| 67 |
* Returns cryptographically secure random bytes on every supported PHP version. |
| 68 |
* |
| 69 |
* @param int $length Number of bytes to return. |
| 70 |
* @return string Random bytes. |
| 71 |
* @throws RuntimeException When the runtime has no secure random-byte source. |
| 72 |
*/ |
| 73 |
function generate_random_bytes(int $length): string |
| 74 |
{ |
| 75 |
if (function_exists('random_bytes')) { |
| 76 |
return random_bytes($length); |
| 77 |
} |
| 78 |
|
| 79 |
if (function_exists('openssl_random_pseudo_bytes')) { |
| 80 |
$strong = false; |
| 81 |
$bytes = openssl_random_pseudo_bytes($length, $strong); |
| 82 |
if ($bytes !== false && $strong && strlen($bytes) === $length) { |
| 83 |
return $bytes; |
| 84 |
} |
| 85 |
} |
| 86 |
|
| 87 |
throw new RuntimeException('The PHP runtime has no cryptographically secure random-byte source.'); |
| 88 |
} |
| 89 |
} |
| 90 |
|
| 91 |
if (!function_exists(__NAMESPACE__ . '\\integer_divide')) { |
| 92 |
/** |
| 93 |
* Divides two integers and rounds the result toward zero. |
| 94 |
* |
| 95 |
* @param int $dividend Number to divide. |
| 96 |
* @param int $divisor Number to divide by. |
| 97 |
* @return int Integer quotient. |
| 98 |
*/ |
| 99 |
function integer_divide(int $dividend, int $divisor): int |
| 100 |
{ |
| 101 |
if (function_exists('intdiv')) { |
| 102 |
return intdiv($dividend, $divisor); |
| 103 |
} |
| 104 |
|
| 105 |
return intval($dividend / $divisor); |
| 106 |
} |
| 107 |
} |
| 108 |
|
| 109 |
if (!function_exists(__NAMESPACE__ . '\\build_pdo_dsn')) { |
| 110 |
/** |
| 111 |
* Builds a PDO DSN string from a WordPress DB_HOST value. |
| 112 |
* |
| 113 |
* WordPress's DB_HOST supports several non-standard formats that shared |
| 114 |
* hosts commonly use: |
| 115 |
* - "localhost" → standard hostname |
| 116 |
* - "db.host.com:3307" → hostname with port |
| 117 |
* - "localhost:/path/sock" → hostname with Unix socket |
| 118 |
* - "/path/to/mysql.sock" → bare Unix socket path |
| 119 |
* - "::1" → IPv6 address |
| 120 |
* - "[::1]" → bracketed IPv6 |
| 121 |
* - "[::1]:3306" → bracketed IPv6 with port |
| 122 |
* - "[::1]:/path/to/socket" → bracketed IPv6 with Unix socket |
| 123 |
* |
| 124 |
* PDO needs these broken out into separate DSN parameters (host, port, |
| 125 |
* unix_socket), so we parse the value the same way WordPress core does. |
| 126 |
* |
| 127 |
* @param string $db_host Raw DB_HOST value. |
| 128 |
* @param string $db_name Database name. |
| 129 |
* @return string PDO DSN string. |
| 130 |
*/ |
| 131 |
function build_pdo_dsn(string $db_host, string $db_name): string |
| 132 |
{ |
| 133 |
$socket = ''; |
| 134 |
$host = $db_host; |
| 135 |
$port = ''; |
| 136 |
|
| 137 |
if (str_starts_with($db_host, '/') && file_exists($db_host)) { |
| 138 |
// Bare socket path: "/var/run/mysqld/mysqld.sock" |
| 139 |
$socket = $db_host; |
| 140 |
$host = ''; |
| 141 |
} elseif ( |
| 142 |
str_starts_with($db_host, '[') && |
| 143 |
($bracket_end = strpos($db_host, ']')) !== false |
| 144 |
) { |
| 145 |
// Bracketed IPv6: "[::1]", "[::1]:3306", "[::1]:/path/to/socket" |
| 146 |
$host = substr($db_host, 1, $bracket_end - 1); |
| 147 |
$after = substr($db_host, $bracket_end + 1); |
| 148 |
$candidate_socket = str_starts_with($after, ':/') ? substr($after, 1) : ''; |
| 149 |
if ($candidate_socket !== '' && file_exists($candidate_socket)) { |
| 150 |
$socket = $candidate_socket; |
| 151 |
} elseif (str_starts_with($after, ':')) { |
| 152 |
$port = substr($after, 1); |
| 153 |
} |
| 154 |
} elseif (($socket_pos = strpos($db_host, ':/')) !== false) { |
| 155 |
// "host:/path/to/socket" — check before general colon split |
| 156 |
// to avoid misinterpreting IPv6 addresses as host:port |
| 157 |
$candidate_socket = substr($db_host, $socket_pos + 1); |
| 158 |
if (file_exists($candidate_socket)) { |
| 159 |
$host = substr($db_host, 0, $socket_pos); |
| 160 |
$socket = $candidate_socket; |
| 161 |
} elseif (substr_count($db_host, ':') === 1) { |
| 162 |
// Single colon but not a socket — treat as host:port |
| 163 |
[$host, $port] = explode(':', $db_host, 2); |
| 164 |
} |
| 165 |
} elseif ( |
| 166 |
str_contains($db_host, ':') && |
| 167 |
substr_count($db_host, ':') === 1 |
| 168 |
) { |
| 169 |
// Exactly one colon: "host:port" — not IPv6 |
| 170 |
[$host, $port] = explode(':', $db_host, 2); |
| 171 |
} |
| 172 |
// Otherwise (multiple colons, no socket marker): bare IPv6 like "::1" |
| 173 |
// — $host stays as the full value. |
| 174 |
|
| 175 |
if ($socket !== '') { |
| 176 |
return "mysql:unix_socket={$socket};dbname={$db_name};charset=utf8mb4"; |
| 177 |
} |
| 178 |
|
| 179 |
$dsn = "mysql:host={$host}"; |
| 180 |
if ($port !== '') { |
| 181 |
$dsn .= ";port={$port}"; |
| 182 |
} |
| 183 |
$dsn .= ";dbname={$db_name};charset=utf8mb4"; |
| 184 |
return $dsn; |
| 185 |
} |
| 186 |
} |
| 187 |
|
| 188 |
if (!function_exists(__NAMESPACE__ . '\\parse_size')) { |
| 189 |
/** |
| 190 |
* Parse a human-readable size string (e.g. "16M", "1G", "512K") into bytes. |
| 191 |
* Accepts plain integers as well. |
| 192 |
*/ |
| 193 |
function parse_size(string $value): int |
| 194 |
{ |
| 195 |
$value = trim($value); |
| 196 |
if (!preg_match('/^(\d+(?:\.\d+)?)\s*([KMGkmg])?[Bb]?$/', $value, $m)) { |
| 197 |
throw new InvalidArgumentException( |
| 198 |
"Invalid size value: '{$value}'. Use a number optionally followed by K, M, or G (e.g. 64M)." |
| 199 |
); |
| 200 |
} |
| 201 |
$num = (float) $m[1]; |
| 202 |
$suffix = strtoupper($m[2] ?? ""); |
| 203 |
switch ($suffix) { |
| 204 |
case "K": |
| 205 |
return (int) ($num * 1024); |
| 206 |
case "M": |
| 207 |
return (int) ($num * 1024 * 1024); |
| 208 |
case "G": |
| 209 |
return (int) ($num * 1024 * 1024 * 1024); |
| 210 |
default: |
| 211 |
return (int) $num; |
| 212 |
} |
| 213 |
} |
| 214 |
} |
| 215 |
|
| 216 |
if (!function_exists(__NAMESPACE__ . '\\json_encode_or_throw')) { |
| 217 |
/** |
| 218 |
* Throws on json_encode failure instead of returning false. |
| 219 |
* |
| 220 |
* Do NOT use inside error/shutdown handlers — those need hardcoded fallback strings. |
| 221 |
*/ |
| 222 |
function json_encode_or_throw($value, int $flags = 0): string |
| 223 |
{ |
| 224 |
$json = json_encode($value, $flags); |
| 225 |
if ($json === false) { |
| 226 |
throw new RuntimeException("json_encode failed: " . json_last_error_msg()); |
| 227 |
} |
| 228 |
return $json; |
| 229 |
} |
| 230 |
} |
| 231 |
|
| 232 |
if (!function_exists(__NAMESPACE__ . '\\normalize_path')) { |
| 233 |
/** |
| 234 |
* Resolve ".." and "." segments in a path without touching the filesystem. |
| 235 |
* |
| 236 |
* Unlike realpath(), this works on paths that don't exist yet. |
| 237 |
*/ |
| 238 |
function normalize_path(string $path): string |
| 239 |
{ |
| 240 |
$parts = explode("/", $path); |
| 241 |
$resolved = []; |
| 242 |
foreach ($parts as $part) { |
| 243 |
if ($part === "" || $part === ".") { |
| 244 |
continue; |
| 245 |
} |
| 246 |
if ($part === "..") { |
| 247 |
array_pop($resolved); |
| 248 |
} else { |
| 249 |
$resolved[] = $part; |
| 250 |
} |
| 251 |
} |
| 252 |
return "/" . implode("/", $resolved); |
| 253 |
} |
| 254 |
} |
| 255 |
|
| 256 |
if (!function_exists(__NAMESPACE__ . '\\trim_right_slash')) { |
| 257 |
/** |
| 258 |
* Removes trailing slashes without changing the filesystem root into an empty path. |
| 259 |
* |
| 260 |
* Unlike rtrim($path, '/'), this returns `/` for both the filesystem root and |
| 261 |
* an empty input. It only changes the lexical spelling; it does not validate |
| 262 |
* the path or resolve dot segments and symlinks. |
| 263 |
* |
| 264 |
* Examples: |
| 265 |
* |
| 266 |
* trim_right_slash('/srv/site///'); // '/srv/site' |
| 267 |
* trim_right_slash('/'); // '/' |
| 268 |
* trim_right_slash(''); // '/' |
| 269 |
* |
| 270 |
* @param string $path Path whose trailing slashes to remove. |
| 271 |
* @return string A path without trailing slashes, or `/` for the filesystem root. |
| 272 |
*/ |
| 273 |
function trim_right_slash(string $path): string |
| 274 |
{ |
| 275 |
return rtrim($path, '/') ?: '/'; |
| 276 |
} |
| 277 |
} |
| 278 |
|
| 279 |
if (!function_exists(__NAMESPACE__ . '\\realpath_with_missing_tail')) { |
| 280 |
/** |
| 281 |
* Canonicalizes an absolute path through the nearest ancestor realpath() can resolve. |
| 282 |
* |
| 283 |
* The final components need not exist. The function resolves a real ancestor, |
| 284 |
* then appends the missing components without creating them. A broken symlink |
| 285 |
* cannot be resolved safely, so its normalized lexical spelling is retained. |
| 286 |
* |
| 287 |
* Examples: |
| 288 |
* |
| 289 |
* realpath_with_missing_tail('/srv/site'); |
| 290 |
* // '/srv/site' when /srv/site exists |
| 291 |
* |
| 292 |
* realpath_with_missing_tail('/srv/site/state/push'); |
| 293 |
* // '/srv/site/state/push' when /srv/site exists but state/push do not |
| 294 |
* |
| 295 |
* realpath_with_missing_tail('/links/site/state'); |
| 296 |
* // '/srv/site/state' when /links/site is a symlink to /srv/site |
| 297 |
* |
| 298 |
* This does not create, remove, or otherwise modify filesystem entries. |
| 299 |
* |
| 300 |
* @throws InvalidArgumentException When $absolute_path is not absolute. |
| 301 |
*/ |
| 302 |
function realpath_with_missing_tail(string $absolute_path): string |
| 303 |
{ |
| 304 |
if ($absolute_path === '' || $absolute_path[0] !== '/') { |
| 305 |
throw new InvalidArgumentException('Path must be absolute: ' . $absolute_path); |
| 306 |
} |
| 307 |
|
| 308 |
$normalized_path = normalize_path($absolute_path); |
| 309 |
$missing_components = []; |
| 310 |
$existing_ancestor = $normalized_path; |
| 311 |
$canonical_existing_ancestor = realpath($existing_ancestor); |
| 312 |
|
| 313 |
while ($canonical_existing_ancestor === false) { |
| 314 |
// Keep a broken symlink lexical: resolving past it would change what a |
| 315 |
// future replacement of that link means. |
| 316 |
if (is_link($existing_ancestor)) { |
| 317 |
return $normalized_path; |
| 318 |
} |
| 319 |
|
| 320 |
$parent = dirname($existing_ancestor); |
| 321 |
if ($parent === $existing_ancestor) { |
| 322 |
return $normalized_path; |
| 323 |
} |
| 324 |
array_unshift($missing_components, basename($existing_ancestor)); |
| 325 |
$existing_ancestor = $parent; |
| 326 |
$canonical_existing_ancestor = realpath($existing_ancestor); |
| 327 |
} |
| 328 |
|
| 329 |
if ($missing_components === []) { |
| 330 |
return normalize_path($canonical_existing_ancestor); |
| 331 |
} |
| 332 |
|
| 333 |
return normalize_path( |
| 334 |
$canonical_existing_ancestor . '/' . implode('/', $missing_components) |
| 335 |
); |
| 336 |
} |
| 337 |
} |
| 338 |
|
| 339 |
if (!function_exists(__NAMESPACE__ . '\\normalize_excluded_paths')) { |
| 340 |
/** |
| 341 |
* Normalizes document-root-relative excluded paths. |
| 342 |
* |
| 343 |
* Rejects non-string, empty, absolute, NUL-containing, backslash-containing, |
| 344 |
* and empty/dot/parent-component paths, then sorts and deduplicates them. |
| 345 |
* |
| 346 |
* @param string[] $excluded_paths Paths which a push must not change. |
| 347 |
* @phpstan-param array<mixed> $excluded_paths |
| 348 |
* @return list<string> Validated excluded paths in bytewise order. |
| 349 |
*/ |
| 350 |
function normalize_excluded_paths(array $excluded_paths): array |
| 351 |
{ |
| 352 |
// phpcs:disable WordPress.Security.EscapeOutput.ExceptionNotEscaped -- These validation exceptions are never rendered, and arbitrary path bytes are represented as base64. |
| 353 |
$normalized_excluded_paths = []; |
| 354 |
foreach ($excluded_paths as $path) { |
| 355 |
if (!is_string($path)) { |
| 356 |
throw new InvalidArgumentException('Each excluded path must be a string; observed ' . gettype($path) . '.'); |
| 357 |
} |
| 358 |
if ($path !== '' && $path[0] === '/') { |
| 359 |
throw new InvalidArgumentException('Excluded path must be document-root-relative: ' . base64_encode($path) . '.'); |
| 360 |
} |
| 361 |
assert_valid_relative_path($path, 'Excluded path'); |
| 362 |
$normalized_excluded_paths[] = $path; |
| 363 |
} |
| 364 |
sort($normalized_excluded_paths, SORT_STRING); |
| 365 |
$normalized_excluded_paths = array_values(array_unique($normalized_excluded_paths)); |
| 366 |
if (count($normalized_excluded_paths) > 100) { |
| 367 |
throw new InvalidArgumentException( |
| 368 |
'Push supports at most 100 excluded paths; received ' |
| 369 |
. count($normalized_excluded_paths) |
| 370 |
. ' after normalization.' |
| 371 |
); |
| 372 |
} |
| 373 |
return $normalized_excluded_paths; |
| 374 |
} |
| 375 |
} |
| 376 |
|
| 377 |
if (!function_exists(__NAMESPACE__ . '\\assert_valid_relative_path')) { |
| 378 |
/** |
| 379 |
* Validates a document-root-relative path carried as raw bytes. |
| 380 |
* |
| 381 |
* A valid path has one or more slash-delimited components. It cannot be |
| 382 |
* absolute, use Windows separators, include a NUL byte, or contain empty, |
| 383 |
* current-directory, or parent-directory components. It deliberately does |
| 384 |
* not trim whitespace: spaces and other non-reserved bytes are valid file |
| 385 |
* name bytes. |
| 386 |
* |
| 387 |
* Examples: |
| 388 |
* |
| 389 |
* assert_valid_relative_path('wp-content/plugins', 'Excluded path'); |
| 390 |
* assert_valid_relative_path('index.php', 'Document-root-relative path'); |
| 391 |
* |
| 392 |
* @param string $path Raw path bytes to validate. |
| 393 |
* @param string $label Human-readable name at the start of validation errors. |
| 394 |
* @throws InvalidArgumentException When the path has a reserved form. |
| 395 |
*/ |
| 396 |
function assert_valid_relative_path(string $path, string $label): void |
| 397 |
{ |
| 398 |
if ($path === '') { |
| 399 |
throw new InvalidArgumentException("{$label} must not be empty."); |
| 400 |
} |
| 401 |
if ($path[0] === '/') { |
| 402 |
throw new InvalidArgumentException("{$label} must not be absolute: " . base64_encode($path) . '.'); |
| 403 |
} |
| 404 |
if (strpos($path, "\0") !== false) { |
| 405 |
throw new InvalidArgumentException("{$label} must not contain a NUL byte: " . base64_encode($path) . '.'); |
| 406 |
} |
| 407 |
if (strpos($path, '\\') !== false) { |
| 408 |
throw new InvalidArgumentException("{$label} must not contain a backslash: " . base64_encode($path) . '.'); |
| 409 |
} |
| 410 |
foreach (explode('/', $path) as $component) { |
| 411 |
if ($component === '') { |
| 412 |
throw new InvalidArgumentException("{$label} must not contain an empty component: " . base64_encode($path) . '.'); |
| 413 |
} |
| 414 |
if ($component === '.') { |
| 415 |
throw new InvalidArgumentException("{$label} must not contain a dot component: " . base64_encode($path) . '.'); |
| 416 |
} |
| 417 |
if ($component === '..') { |
| 418 |
throw new InvalidArgumentException("{$label} must not contain a parent component: " . base64_encode($path) . '.'); |
| 419 |
} |
| 420 |
} |
| 421 |
} |
| 422 |
} |
| 423 |
|
| 424 |
// phpcs:enable WordPress.Security.EscapeOutput.ExceptionNotEscaped |
| 425 |
|
| 426 |
if (!function_exists(__NAMESPACE__ . '\\path_is_same_as_or_descendant_of')) { |
| 427 |
/** |
| 428 |
* Indicates whether a candidate path is the same as or a descendant of an |
| 429 |
* ancestor. |
| 430 |
* |
| 431 |
* Either argument may be a list. The result is true when any candidate-and- |
| 432 |
* ancestor pair matches. The filesystem root matches every absolute path and |
| 433 |
* cannot use the normal ancestor-plus-slash prefix because that would produce |
| 434 |
* `//`. |
| 435 |
* |
| 436 |
* Examples: |
| 437 |
* |
| 438 |
* path_is_same_as_or_descendant_of('/srv/site', '/srv/site'); // true |
| 439 |
* path_is_same_as_or_descendant_of('/srv/site/wp-content', '/srv/site'); // true |
| 440 |
* path_is_same_as_or_descendant_of('/srv/site-old', '/srv/site'); // false |
| 441 |
* path_is_same_as_or_descendant_of('/', '/'); // true |
| 442 |
* |
| 443 |
* @param string|list<string> $path Candidate path or paths. |
| 444 |
* @param string|list<string> $ancestor Ancestor path or paths. |
| 445 |
* @return bool Whether a candidate is the same as or a descendant of an |
| 446 |
* ancestor. |
| 447 |
* @throws InvalidArgumentException If either scalar value is not a string. |
| 448 |
*/ |
| 449 |
function path_is_same_as_or_descendant_of($path, $ancestor): bool |
| 450 |
{ |
| 451 |
if (is_array($path)) { |
| 452 |
foreach ($path as $candidate_path) { |
| 453 |
if (path_is_same_as_or_descendant_of($candidate_path, $ancestor)) { |
| 454 |
return true; |
| 455 |
} |
| 456 |
} |
| 457 |
return false; |
| 458 |
} |
| 459 |
if (is_array($ancestor)) { |
| 460 |
foreach ($ancestor as $candidate_ancestor) { |
| 461 |
if (path_is_same_as_or_descendant_of($path, $candidate_ancestor)) { |
| 462 |
return true; |
| 463 |
} |
| 464 |
} |
| 465 |
return false; |
| 466 |
} |
| 467 |
if (!is_string($path) || !is_string($ancestor)) { |
| 468 |
throw new InvalidArgumentException('Path containment expects strings or lists of strings.'); |
| 469 |
} |
| 470 |
if ($ancestor === "/") { |
| 471 |
return str_starts_with($path, "/"); |
| 472 |
} |
| 473 |
return $path === $ancestor || str_starts_with($path, $ancestor . "/"); |
| 474 |
} |
| 475 |
} |
| 476 |
|
| 477 |
if (!function_exists(__NAMESPACE__ . '\\path_is_descendant_of')) { |
| 478 |
/** |
| 479 |
* Indicates whether a candidate path is a descendant of an ancestor. |
| 480 |
* |
| 481 |
* Either argument may be a list. The result is true when any candidate-and- |
| 482 |
* ancestor pair has a component-boundary match below the ancestor. Unlike |
| 483 |
* path_is_same_as_or_descendant_of(), equal paths do not match. The |
| 484 |
* filesystem root contains every absolute descendant, but not itself. |
| 485 |
* |
| 486 |
* Examples: |
| 487 |
* |
| 488 |
* path_is_descendant_of('/srv/site/wp-content', '/srv/site'); // true |
| 489 |
* path_is_descendant_of('/srv/site', '/srv/site'); // false |
| 490 |
* path_is_descendant_of('/srv/site-old', '/srv/site'); // false |
| 491 |
* path_is_descendant_of('/wp-content', '/'); // true |
| 492 |
* path_is_descendant_of('/', '/'); // false |
| 493 |
* |
| 494 |
* @param string|list<string> $path Candidate path or paths. |
| 495 |
* @param string|list<string> $ancestor Ancestor path or paths. |
| 496 |
* @return bool Whether a candidate is a descendant of an ancestor. |
| 497 |
* @throws InvalidArgumentException If either scalar value is not a string. |
| 498 |
*/ |
| 499 |
function path_is_descendant_of($path, $ancestor): bool |
| 500 |
{ |
| 501 |
if (is_array($path)) { |
| 502 |
foreach ($path as $candidate_path) { |
| 503 |
if (path_is_descendant_of($candidate_path, $ancestor)) { |
| 504 |
return true; |
| 505 |
} |
| 506 |
} |
| 507 |
return false; |
| 508 |
} |
| 509 |
if (is_array($ancestor)) { |
| 510 |
foreach ($ancestor as $candidate_ancestor) { |
| 511 |
if (path_is_descendant_of($path, $candidate_ancestor)) { |
| 512 |
return true; |
| 513 |
} |
| 514 |
} |
| 515 |
return false; |
| 516 |
} |
| 517 |
if (!path_is_same_as_or_descendant_of($path, $ancestor)) { |
| 518 |
return false; |
| 519 |
} |
| 520 |
return $path !== $ancestor; |
| 521 |
} |
| 522 |
} |
| 523 |
|
| 524 |
if (!function_exists(__NAMESPACE__ . '\\path_remainder_under')) { |
| 525 |
/** |
| 526 |
* Returns the remainder of $path underneath $prefix. |
| 527 |
* |
| 528 |
* An exact match returns an empty string. A descendant returns the remainder |
| 529 |
* beginning with "/". A path outside $prefix returns null. |
| 530 |
*/ |
| 531 |
function path_remainder_under(string $path, string $prefix): ?string |
| 532 |
{ |
| 533 |
$path = rtrim($path, "/"); |
| 534 |
$prefix = rtrim($prefix, "/"); |
| 535 |
|
| 536 |
if ($path === $prefix) { |
| 537 |
return ""; |
| 538 |
} |
| 539 |
|
| 540 |
if (str_starts_with($path, $prefix . "/")) { |
| 541 |
return substr($path, strlen($prefix)); |
| 542 |
} |
| 543 |
|
| 544 |
return null; |
| 545 |
} |
| 546 |
} |
| 547 |
|
| 548 |
if (!function_exists(__NAMESPACE__ . '\\relative_path_under')) { |
| 549 |
/** |
| 550 |
* Returns a path relative to a slash-delimited root, or null when it is not |
| 551 |
* equal to or below that root. |
| 552 |
* |
| 553 |
* Use this when a caller needs a path for a root-relative field. It performs |
| 554 |
* the component-boundary test and removes the separating slash in one step, |
| 555 |
* rather than letting a byte-offset slice treat `/srv/site-old` as below |
| 556 |
* `/srv/site`. |
| 557 |
* |
| 558 |
* Examples: |
| 559 |
* |
| 560 |
* relative_path_under('/srv/site/wp-content', '/srv/site'); // 'wp-content' |
| 561 |
* relative_path_under('/srv/site', '/srv/site'); // '' |
| 562 |
* relative_path_under('/srv/site-old', '/srv/site'); // null |
| 563 |
* relative_path_under('/wp-content', '/'); // 'wp-content' |
| 564 |
* relative_path_under('wp-content/plugins', ''); // 'wp-content/plugins' |
| 565 |
* |
| 566 |
* Trailing slashes do not change the result. This is a lexical operation: it |
| 567 |
* does not resolve dot segments or symlinks, and it also accepts relative |
| 568 |
* slash-delimited paths. An empty root contains every relative path, but no |
| 569 |
* absolute path. |
| 570 |
* |
| 571 |
* @param string $path Candidate path to make relative. |
| 572 |
* @param string $root Root that must contain the candidate path. |
| 573 |
* @return string|null A path without a leading slash, an empty string for an |
| 574 |
* exact match, or null when the path is outside the root. |
| 575 |
*/ |
| 576 |
function relative_path_under(string $path, string $root): ?string |
| 577 |
{ |
| 578 |
if ($root === "") { |
| 579 |
return str_starts_with($path, "/") ? null : rtrim($path, "/"); |
| 580 |
} |
| 581 |
$remainder = path_remainder_under($path, $root); |
| 582 |
return $remainder === null ? null : ltrim($remainder, "/"); |
| 583 |
} |
| 584 |
} |
| 585 |
|
| 586 |
if (!function_exists(__NAMESPACE__ . '\\assert_valid_path')) { |
| 587 |
/** |
| 588 |
* Validates that a path is a non-empty absolute string without NUL bytes |
| 589 |
* or dot-segments (. or ..). |
| 590 |
* |
| 591 |
* Useful anywhere untrusted or remote paths need to be checked before |
| 592 |
* use — both the exporter (directory config) and the importer (remote |
| 593 |
* paths from the server) share this validation. |
| 594 |
* |
| 595 |
* @param string $path The path to validate. |
| 596 |
* @param string $label Human-readable label for error messages (e.g. "directory", "remote path"). |
| 597 |
* @throws InvalidArgumentException When the path fails any check. |
| 598 |
*/ |
| 599 |
function assert_valid_path(string $path, string $label = "path"): void |
| 600 |
{ |
| 601 |
$path = trim($path); |
| 602 |
if ($path === "") { |
| 603 |
throw new InvalidArgumentException("{$label} must be a non-empty string"); |
| 604 |
} |
| 605 |
if ($path[0] !== "/") { |
| 606 |
throw new InvalidArgumentException("{$label} must be an absolute path: {$path}"); |
| 607 |
} |
| 608 |
if (strpos($path, "\0") !== false) { |
| 609 |
throw new InvalidArgumentException("{$label} must not contain NUL bytes"); |
| 610 |
} |
| 611 |
foreach (explode("/", $path) as $segment) { |
| 612 |
if ($segment === "." || $segment === "..") { |
| 613 |
throw new InvalidArgumentException( |
| 614 |
"{$label} must not contain dot-segments (. or ..): {$path}" |
| 615 |
); |
| 616 |
} |
| 617 |
} |
| 618 |
} |
| 619 |
} |
| 620 |
|
| 621 |
// --------------------------------------------------------------------------- |
| 622 |
// Vendored from wp-php-toolkit/filesystem. |
| 623 |
// |
| 624 |
// This is a copy of WordPress\Filesystem\wp_join_unix_paths(), kept in sync by |
| 625 |
// hand. Do not "fix" it by importing the original: reprint-server must require |
| 626 |
// nothing but PHP. |
| 627 |
// |
| 628 |
// Consumers vendor this package into Composer autoloaders that are not scoped |
| 629 |
// to one plugin — Jetpack's is the one that bites. It folds every installed |
| 630 |
// package's psr-4, classmap and files entries into site-global manifests that |
| 631 |
// arbitrate class and function names, by version, across every plugin on the |
| 632 |
// site. Requiring wp-php-toolkit/filesystem would publish WordPress\Filesystem |
| 633 |
// site-wide, where it would be arbitrated against the copy WordPress Importer |
| 634 |
// already ships through data-liberation. Two copies of one namespace in one |
| 635 |
// version-arbitrated manifest is what produced Automattic/jetpack#51027. |
| 636 |
// |
| 637 |
// WordPress core's path_join() is not a substitute. It takes two arguments |
| 638 |
// rather than being variadic, does not collapse duplicate slashes, and returns |
| 639 |
// the second argument alone when that is absolute, discarding the base. Its |
| 640 |
// path_is_absolute() check also calls realpath() plus a stream-wrapper lookup, |
| 641 |
// and class-file-index-processor.php calls this once per directory entry in |
| 642 |
// the file walk. |
| 643 |
// --------------------------------------------------------------------------- |
| 644 |
if (!function_exists(__NAMESPACE__ . '\\wp_join_unix_paths')) { |
| 645 |
/** |
| 646 |
* Joins path segments into one Unix path, collapsing duplicate slashes. |
| 647 |
* |
| 648 |
* Empty segments are skipped. A leading slash on the first non-empty segment |
| 649 |
* is preserved. Trailing slashes are left as the caller wrote them. |
| 650 |
* |
| 651 |
* Examples: |
| 652 |
* |
| 653 |
* wp_join_unix_paths('/srv/site', 'wp-content'); // '/srv/site/wp-content' |
| 654 |
* wp_join_unix_paths('/srv/site/', '/uploads'); // '/srv/site/uploads' |
| 655 |
* wp_join_unix_paths('', 'wp-content', ''); // 'wp-content' |
| 656 |
* |
| 657 |
* @param string ...$path_segments Segments to join. |
| 658 |
* @return string The joined path. |
| 659 |
*/ |
| 660 |
function wp_join_unix_paths(...$path_segments) |
| 661 |
{ |
| 662 |
$input_starts_with_slash = null; |
| 663 |
|
| 664 |
$paths = []; |
| 665 |
foreach ($path_segments as $path_segment) { |
| 666 |
if ($path_segment !== '') { |
| 667 |
$paths[] = $path_segment; |
| 668 |
if ($input_starts_with_slash === null) { |
| 669 |
$input_starts_with_slash = strncmp($path_segment, '/', strlen('/')) === 0; |
| 670 |
} |
| 671 |
} |
| 672 |
} |
| 673 |
$path = implode('/', $paths); |
| 674 |
|
| 675 |
$result = preg_replace('#/+#', '/', $path); |
| 676 |
if ($input_starts_with_slash && strncmp($result, '/', strlen('/')) !== 0) { |
| 677 |
$result = '/' . $result; |
| 678 |
} |
| 679 |
|
| 680 |
return $result; |
| 681 |
} |
| 682 |
} |
| 683 |
|
| 684 |
} |
| 685 |
|
