Merge remote-tracking branch 'origin/next' into fix/url-validator-underscore-hostnames

This commit is contained in:
Andras Bacsai committed 2026-07-07 12:24:32 +02:00
commit b50839d451
505 files changed
+21391 -5998

No files matched your search

+209
View File
@@ -0,0 +1,209 @@
<?php
namespace App\Support;
use Illuminate\Http\UploadedFile;
class DatabaseBackupFileValidator
{
public const ALLOWED_EXTENSIONS = [
'sql',
'sql.gz',
'gz',
'zip',
'tar',
'tar.gz',
'tgz',
'dump',
'bak',
'bson',
'bson.gz',
'archive',
'archive.gz',
'bz2',
'xz',
'dmp',
];
private const DANGEROUS_EXTENSIONS = [
'asp',
'aspx',
'bat',
'bash',
'cgi',
'cmd',
'com',
'exe',
'htm',
'html',
'jar',
'js',
'jsp',
'php',
'php3',
'php4',
'php5',
'phtml',
'pl',
'ps1',
'py',
'rb',
'sh',
];
public static function hasAllowedExtension(string $name): bool
{
return self::extensionFor($name) !== null;
}
public static function isUploadAllowed(UploadedFile $file, int $maxBytes): bool
{
$originalName = $file->getClientOriginalName();
$size = $file->getSize();
if ($size === false || $size > $maxBytes) {
return false;
}
$extension = self::extensionFor($originalName);
if ($extension === null) {
return false;
}
return self::contentMatchesExtension($file->getPathname(), $extension);
}
/**
* Scan a stored backup file (decompressing gzip on the fly) for PostgreSQL
* restore directives that lead to OS command execution.
*/
public static function fileContainsPostgresqlProgramExecution(string $path): bool
{
$contents = self::readPossiblyGzippedText($path);
if ($contents === null) {
return false;
}
return self::containsPostgresqlProgramExecution($contents);
}
public static function containsPostgresqlProgramExecution(string $sql): bool
{
$withoutComments = self::stripSqlComments($sql);
if (preg_match('/^\s*\\\\(?:!|copy\b.*\bprogram\b)/mi', $withoutComments) === 1) {
return true;
}
return preg_match('/\bcopy\b[\s\S]{0,2000}\b(?:from|to)\s+program\b/i', $withoutComments) === 1;
}
private static function extensionFor(string $name): ?string
{
$lower = strtolower($name);
$suffixes = array_map(fn (string $ext) => '.'.$ext, self::ALLOWED_EXTENSIONS);
usort($suffixes, fn (string $a, string $b) => strlen($b) <=> strlen($a));
foreach ($suffixes as $suffix) {
if (! str_ends_with($lower, $suffix)) {
continue;
}
$stem = substr($lower, 0, -strlen($suffix));
if ($stem === '' || str_ends_with($stem, '.')) {
return null;
}
$parts = array_filter(explode('.', $stem));
if (array_intersect($parts, self::DANGEROUS_EXTENSIONS) !== []) {
return null;
}
return ltrim($suffix, '.');
}
return null;
}
private static function contentMatchesExtension(string $path, string $extension): bool
{
$sample = (string) file_get_contents($path, false, null, 0, 4096);
return match ($extension) {
'sql' => self::looksLikeText($sample) && ! self::containsPostgresqlProgramExecution($sample),
'sql.gz', 'gz', 'tar.gz', 'tgz', 'bson.gz', 'archive.gz' => str_starts_with($sample, "\x1f\x8b"),
'zip' => str_starts_with($sample, "PK\x03\x04") || str_starts_with($sample, "PK\x05\x06") || str_starts_with($sample, "PK\x07\x08"),
'tar' => substr($sample, 257, 5) === 'ustar',
'bz2' => str_starts_with($sample, 'BZh'),
'xz' => str_starts_with($sample, "\xfd7zXZ\x00"),
'dump', 'bak', 'archive', 'dmp' => str_starts_with($sample, 'PGDMP')
|| (self::looksLikeText($sample) && ! self::containsPostgresqlProgramExecution($sample)),
'bson' => self::looksLikeBson($path, $sample),
default => false,
};
}
private static function readPossiblyGzippedText(string $path): ?string
{
// Cap the scan so a huge legitimate dump cannot exhaust memory; the
// remote pre-restore scanner inspects the full file as a second layer.
$maxBytes = 50 * 1024 * 1024;
$handle = @fopen($path, 'rb');
if ($handle === false) {
return null;
}
$magic = (string) fread($handle, 2);
fclose($handle);
if ($magic === "\x1f\x8b") {
$gz = @gzopen($path, 'rb');
if ($gz === false) {
return null;
}
$data = '';
while (! gzeof($gz) && strlen($data) < $maxBytes) {
$chunk = gzread($gz, 1024 * 1024);
if ($chunk === false || $chunk === '') {
break;
}
$data .= $chunk;
}
gzclose($gz);
return $data;
}
return (string) file_get_contents($path, false, null, 0, $maxBytes);
}
private static function looksLikeText(string $sample): bool
{
if ($sample === '' || str_contains($sample, "\0")) {
return false;
}
return mb_check_encoding($sample, 'UTF-8') || mb_check_encoding($sample, 'ASCII');
}
private static function looksLikeBson(string $path, string $sample): bool
{
if (strlen($sample) < 5) {
return false;
}
$documentLength = unpack('V', substr($sample, 0, 4))[1] ?? 0;
$fileSize = filesize($path) ?: 0;
return $documentLength >= 5 && $documentLength <= $fileSize;
}
private static function stripSqlComments(string $sql): string
{
$sql = preg_replace('/\/\*[\s\S]*?\*\//', ' ', $sql) ?? $sql;
return preg_replace('/--[^\r\n]*/', ' ', $sql) ?? $sql;
}
}
+185 -4
View File
@@ -94,10 +94,25 @@ class ValidationPatterns
public const DOCKER_NETWORK_PATTERN = '/^[a-zA-Z0-9][a-zA-Z0-9._-]*$/';
/**
* Pattern for Docker-compatible environment variable keys.
* Docker environment entries are KEY=value strings, so keys must be non-empty and cannot contain '=' or NUL.
* Pattern for S3 bucket names.
*
* Bucket names must be 3-63 lowercase characters, start and end with a
* letter or digit, and contain only lowercase letters, digits, dots, and
* hyphens. Additional semantic checks live in isValidS3BucketName().
*/
public const ENVIRONMENT_VARIABLE_KEY_PATTERN = '/\A[^=\x00]+\z/u';
public const S3_BUCKET_NAME_PATTERN = '/\A(?=.{3,63}\z)[a-z0-9][a-z0-9.-]*[a-z0-9]\z/';
/**
* Pattern for Docker-compatible environment variable keys.
* Environment variable keys are later interpolated into shell commands as Docker build args, so only shell-safe identifier characters are allowed.
*/
public const ENVIRONMENT_VARIABLE_KEY_PATTERN = '/\A[A-Za-z_][A-Za-z0-9_.]*\z/u';
/**
* Characters that are valid in some URL positions but unsafe for values
* that are later reused in shell assignment contexts.
*/
public const APPLICATION_DOMAIN_FORBIDDEN_PATTERN = '/[`$;&|<>()\\\\\r\n]/';
/**
* Pattern for SQL-safe unquoted database identifiers (usernames, database names).
@@ -164,7 +179,7 @@ class ValidationPatterns
public static function environmentVariableKeyMessages(string $field = 'key', string $label = 'key'): array
{
return [
"{$field}.regex" => "The {$label} must be a non-empty Docker-compatible environment variable key and cannot contain '=' or NUL characters.",
"{$field}.regex" => "The {$label} must start with a letter or underscore and may only contain letters, numbers, underscores, and dots.",
"{$field}.max" => "The {$label} may not be greater than :max characters.",
];
}
@@ -177,6 +192,22 @@ class ValidationPatterns
return preg_match(self::ENVIRONMENT_VARIABLE_KEY_PATTERN, $value) === 1;
}
/**
* Check if a string is a valid S3 bucket name.
*/
public static function isValidS3BucketName(string $value): bool
{
if (preg_match(self::S3_BUCKET_NAME_PATTERN, $value) !== 1) {
return false;
}
if (str_contains($value, '..') || str_contains($value, '.-') || str_contains($value, '-.')) {
return false;
}
return filter_var($value, FILTER_VALIDATE_IP, FILTER_FLAG_IPV4) === false;
}
/**
* Normalize and validate an environment variable key.
*/
@@ -486,6 +517,156 @@ class ValidationPatterns
return ['nullable', 'string', 'max:'.$maxLength, 'regex:'.self::SHELL_SAFE_COMMAND_PATTERN];
}
/**
* Get validation rules for comma-separated application URL fields.
*/
public static function applicationDomainRules(int $maxLength = 2048): array
{
return [
'nullable',
'string',
'max:'.$maxLength,
function (string $attribute, mixed $value, \Closure $fail): void {
foreach (self::validateApplicationDomains($value) as $error) {
$fail($error);
}
},
];
}
/**
* Validate a comma-separated list of application URLs.
*
* @return array<int, string>
*/
public static function validateApplicationDomains(mixed $value): array
{
if (blank($value)) {
return [];
}
if (! is_string($value)) {
return ['The domains field must be a string.'];
}
$errors = [];
foreach (self::applicationDomainList($value) as $url) {
if (preg_match(self::APPLICATION_DOMAIN_FORBIDDEN_PATTERN, $url) === 1) {
$errors[] = "Invalid URL: {$url}";
continue;
}
if (! isValidDomainUrl($url)) {
$errors[] = "Invalid URL: {$url}";
continue;
}
$scheme = parse_url($url, PHP_URL_SCHEME) ?? '';
if (! in_array(strtolower($scheme), ['http', 'https'], true)) {
$errors[] = "Invalid URL scheme: {$scheme} for URL: {$url}. Only http and https are supported.";
continue;
}
if (blank(parse_url($url, PHP_URL_HOST))) {
$errors[] = "Invalid URL: {$url}";
}
}
return $errors;
}
/**
* Normalize a comma-separated application URL list for storage.
*/
public static function normalizeApplicationDomains(?string $value): ?string
{
$urls = self::applicationDomainList($value);
if ($urls === []) {
return null;
}
return collect($urls)
->map(fn (string $url) => self::normalizeApplicationDomainUrl($url))
->implode(',');
}
/**
* Normalize URL components that are case-insensitive while preserving
* case-sensitive path, query, and fragment components.
*/
private static function normalizeApplicationDomainUrl(string $url): string
{
$components = parse_url($url);
if ($components === false) {
return $url;
}
$normalized = '';
if (isset($components['scheme'])) {
$normalized .= strtolower($components['scheme']).'://';
}
if (isset($components['user'])) {
$normalized .= $components['user'];
if (isset($components['pass'])) {
$normalized .= ':'.$components['pass'];
}
$normalized .= '@';
}
if (isset($components['host'])) {
$normalized .= strtolower($components['host']);
}
if (isset($components['port'])) {
$normalized .= ':'.$components['port'];
}
if (isset($components['path'])) {
$normalized .= $components['path'];
}
if (array_key_exists('query', $components)) {
$normalized .= '?'.$components['query'];
}
if (array_key_exists('fragment', $components)) {
$normalized .= '#'.$components['fragment'];
}
return $normalized;
}
/**
* Split a comma-separated application URL list into trimmed URL strings.
*
* @return array<int, string>
*/
public static function applicationDomainList(?string $value): array
{
if (blank($value)) {
return [];
}
return str($value)
->replaceStart(',', '')
->replaceEnd(',', '')
->trim()
->explode(',')
->map(fn (string $url) => trim($url))
->filter(fn (string $url) => filled($url))
->values()
->all();
}
/**
* Get validation rules for Docker volume name fields
*/