API
6 days ago
Access
1 month ago
Application
1 month ago
Archive
1 month ago
ArchiveProcessor
6 days ago
Archiver
2 years ago
AssetManager
1 month ago
Auth
1 month ago
Category
1 month ago
Changes
3 months ago
CliMulti
1 month ago
Columns
6 days ago
Concurrency
1 month ago
Config
1 month ago
Container
3 months ago
CronArchive
1 month ago
DataAccess
6 days ago
DataFiles
2 years ago
DataTable
1 month ago
Db
6 days ago
DeviceDetector
1 year ago
Email
2 years ago
Exception
6 months ago
Http
6 days ago
Intl
5 months ago
Log
2 years ago
Mail
1 year ago
Measurable
8 months ago
Menu
1 month ago
Metrics
1 month ago
Notification
8 months ago
Period
1 month ago
Plugin
6 days ago
Policy
3 months ago
ProfessionalServices
1 year ago
Report
1 year ago
ReportRenderer
6 days ago
Request
1 month ago
Scheduler
1 month ago
Segment
1 month ago
Session
6 days ago
Settings
6 days ago
Tracker
6 days ago
Translation
3 months ago
Twig
1 year ago
UpdateCheck
5 months ago
Updater
1 month ago
Updates
2 months ago
Validators
1 year ago
View
3 months ago
ViewDataTable
6 days ago
Visualization
6 days ago
Widget
1 month ago
.htaccess
2 years ago
Access.php
1 month ago
Archive.php
3 months ago
ArchiveProcessor.php
6 days ago
AssetManager.php
1 month ago
Auth.php
8 months ago
AuthResult.php
8 months ago
BaseFactory.php
2 years ago
Cache.php
1 month ago
CacheId.php
6 months ago
CliMulti.php
1 month ago
Common.php
1 month ago
Config.php
1 month ago
Console.php
5 months ago
Context.php
2 years ago
Cookie.php
1 month ago
CronArchive.php
1 month ago
DI.php
5 months ago
DataArray.php
1 month ago
DataTable.php
6 days ago
Date.php
3 months ago
Db.php
3 months ago
DbHelper.php
1 month ago
Development.php
1 month ago
ErrorHandler.php
8 months ago
EventDispatcher.php
3 months ago
ExceptionHandler.php
6 months ago
FileIntegrity.php
1 month ago
Filechecks.php
1 year ago
Filesystem.php
6 days ago
FrontController.php
1 month ago
Http.php
6 days ago
IP.php
1 year ago
Log.php
5 months ago
LogDeleter.php
1 month ago
Mail.php
1 year ago
Metrics.php
3 months ago
NoAccessException.php
2 years ago
Nonce.php
8 months ago
Notification.php
3 months ago
NumberFormatter.php
1 month ago
Option.php
1 month ago
Period.php
1 month ago
Piwik.php
1 month ago
Plugin.php
3 months ago
Process.php
3 months ago
Profiler.php
1 month ago
ProxyHeaders.php
6 months ago
ProxyHttp.php
1 month ago
QuickForm2.php
5 months ago
RankingQuery.php
3 months ago
ReportRenderer.php
6 days ago
Request.php
3 months ago
Segment.php
6 days ago
Sequence.php
1 month ago
Session.php
2 months ago
SettingsPiwik.php
1 month ago
SettingsServer.php
1 month ago
Singleton.php
2 years ago
Site.php
3 months ago
SiteContentDetector.php
6 days ago
SupportedBrowser.php
2 years ago
TCPDF.php
1 year ago
Theme.php
1 month ago
Timer.php
1 month ago
Tracker.php
1 month ago
Twig.php
6 days ago
Unzip.php
1 year ago
UpdateCheck.php
3 months ago
Updater.php
1 month ago
UpdaterErrorException.php
2 years ago
Updates.php
1 month ago
Url.php
1 month ago
UrlHelper.php
3 months ago
Version.php
6 days ago
View.php
1 month ago
bootstrap.php
1 year ago
dispatch.php
2 years ago
testMinimumPhpVersion.php
8 months ago
Http.php
1049 lines
| 1 | <?php |
| 2 | |
| 3 | /** |
| 4 | * Matomo - free/libre analytics platform |
| 5 | * |
| 6 | * @link https://matomo.org |
| 7 | * @license https://www.gnu.org/licenses/gpl-3.0.html GPL v3 or later |
| 8 | */ |
| 9 | namespace Piwik; |
| 10 | |
| 11 | use Composer\CaBundle\CaBundle; |
| 12 | use Exception; |
| 13 | use Piwik\Config\GeneralConfig; |
| 14 | use Piwik\Container\StaticContainer; |
| 15 | use Piwik\Http\EgressBlockedException; |
| 16 | use Piwik\Http\EgressHostValidator; |
| 17 | /** |
| 18 | * Contains HTTP client related helper methods that can retrieve content from remote servers |
| 19 | * and optionally save to a local file. |
| 20 | * |
| 21 | * Used to check for the latest Piwik version and download updates. |
| 22 | * |
| 23 | */ |
| 24 | class Http |
| 25 | { |
| 26 | /** |
| 27 | * Returns the "best" available transport method for {@link sendHttpRequest()} calls. |
| 28 | * |
| 29 | * @return string|null Either curl, fopen, socket or null if no method is supported. |
| 30 | * @api |
| 31 | */ |
| 32 | public static function getTransportMethod() |
| 33 | { |
| 34 | $method = 'curl'; |
| 35 | if (!self::isCurlEnabled()) { |
| 36 | $method = 'fopen'; |
| 37 | if (@ini_get('allow_url_fopen') != '1') { |
| 38 | $method = 'socket'; |
| 39 | if (!self::isSocketEnabled()) { |
| 40 | return null; |
| 41 | } |
| 42 | } |
| 43 | } |
| 44 | return $method; |
| 45 | } |
| 46 | /** |
| 47 | * @return bool |
| 48 | */ |
| 49 | protected static function isSocketEnabled() |
| 50 | { |
| 51 | return function_exists('fsockopen'); |
| 52 | } |
| 53 | /** |
| 54 | * @return bool |
| 55 | */ |
| 56 | protected static function isCurlEnabled() |
| 57 | { |
| 58 | return function_exists('curl_init') && function_exists('curl_exec'); |
| 59 | } |
| 60 | /** |
| 61 | * Sends an HTTP request using best available transport method. |
| 62 | * |
| 63 | * @param string $aUrl The target URL. |
| 64 | * @param int $timeout The number of seconds to wait before aborting the HTTP request. |
| 65 | * @param string|null $userAgent The user agent to use. |
| 66 | * @param string|null $destinationPath If supplied, the HTTP response will be saved to the file specified by |
| 67 | * this path. |
| 68 | * @param int|null $followDepth Internal redirect count. Should always pass `null` for this parameter. |
| 69 | * @param bool|string $acceptLanguage The value to use for the `'Accept-Language'` HTTP request header. |
| 70 | * @param array|bool $byteRange For `Range:` header. Should be two element array of bytes, eg, `array(0, 1024)` |
| 71 | * Doesn't work w/ `fopen` transport method. |
| 72 | * @param bool $getExtendedInfo If true returns the status code, headers & response, if false just the response. |
| 73 | * @param string $httpMethod The HTTP method to use. Defaults to `'GET'`. |
| 74 | * @param string $httpUsername HTTP Auth username |
| 75 | * @param string $httpPassword HTTP Auth password |
| 76 | * @param bool $checkHostIsAllowed whether we should check if the target host is allowed or not. This should only |
| 77 | * be set to false when using a hardcoded URL. |
| 78 | * @param bool $validateEgressIp when true, serves the request over the SSRF-safe path: the resolved host must be a |
| 79 | * public IP (or covered by `[General] allowed_private_egress_ranges`), every redirect |
| 80 | * hop is re-validated and the connection pinned to it. Use this whenever the URL comes |
| 81 | * from untrusted input (e.g. a site's own configured URL). |
| 82 | * Requires curl, bypasses any configured or environment proxy, retains the method and |
| 83 | * body across same-origin hops, drops credentials, caller headers and the body on an |
| 84 | * origin change, and does not follow redirects when downloading to a file. |
| 85 | * A refused target or unmet precondition throws {@see EgressBlockedException}. |
| 86 | * |
| 87 | * @return string|array|bool If `$destinationPath` is not specified the HTTP response is returned on success. `false` |
| 88 | * is returned on failure. |
| 89 | * If `$getExtendedInfo` is `true` and `$destinationPath` is not specified an array with |
| 90 | * the following information is returned on success: |
| 91 | * |
| 92 | * - **status**: the HTTP status code |
| 93 | * - **headers**: the HTTP headers |
| 94 | * - **data**: the HTTP response data |
| 95 | * |
| 96 | * `false` is still returned on failure. |
| 97 | * @throws Exception if the response cannot be saved to `$destinationPath`, if the HTTP response cannot be sent, |
| 98 | * if there are more than 5 redirects or if the request times out. |
| 99 | * @phpstan-return ($destinationPath is null ? ($getExtendedInfo is true ? array{status: ?int, headers?: ?array, data?: ?string} : string|false) : bool) |
| 100 | * @api |
| 101 | */ |
| 102 | public static function sendHttpRequest($aUrl, $timeout, $userAgent = null, $destinationPath = null, $followDepth = 0, $acceptLanguage = \false, $byteRange = \false, $getExtendedInfo = \false, $httpMethod = 'GET', $httpUsername = null, $httpPassword = null, $checkHostIsAllowed = \true, $validateEgressIp = \false) |
| 103 | { |
| 104 | // create output file |
| 105 | $file = self::ensureDestinationDirectoryExists($destinationPath); |
| 106 | $transport = self::getTransportMethod(); |
| 107 | if ($validateEgressIp) { |
| 108 | // The SSRF-safe path only pins and re-validates reliably over curl, so fail |
| 109 | // closed rather than silently degrade to an unprotected transport. |
| 110 | if (!self::isCurlEnabled()) { |
| 111 | throw new EgressBlockedException('SSRF-safe HTTP requests require the curl PHP extension.'); |
| 112 | } |
| 113 | $transport = 'curl'; |
| 114 | } |
| 115 | $acceptLanguage = $acceptLanguage ? 'Accept-Language: ' . $acceptLanguage : ''; |
| 116 | return self::sendHttpRequestBy($transport, $aUrl, $timeout, $userAgent, $destinationPath, $file, $followDepth ?? 0, $acceptLanguage, $acceptInvalidSslCertificate = \false, $byteRange, $getExtendedInfo, $httpMethod, $httpUsername, $httpPassword, null, [], null, $checkHostIsAllowed, $validateEgressIp); |
| 117 | } |
| 118 | /** |
| 119 | * @param string|null $destinationPath |
| 120 | * @return resource|null |
| 121 | * @throws Exception |
| 122 | */ |
| 123 | public static function ensureDestinationDirectoryExists($destinationPath) |
| 124 | { |
| 125 | if ($destinationPath) { |
| 126 | \Piwik\Filesystem::mkdir(dirname($destinationPath)); |
| 127 | if (($file = @fopen($destinationPath, 'wb')) === \false || !is_resource($file)) { |
| 128 | throw new Exception('Error while creating the file: ' . $destinationPath); |
| 129 | } |
| 130 | return $file; |
| 131 | } |
| 132 | return null; |
| 133 | } |
| 134 | /** |
| 135 | * Throws when the host matches any `http.blocklist.hosts` wildcard rule. |
| 136 | */ |
| 137 | private static function assertHostNotBlocked(?string $host) : void |
| 138 | { |
| 139 | if (empty($host)) { |
| 140 | return; |
| 141 | } |
| 142 | $disallowedHosts = StaticContainer::get('http.blocklist.hosts'); |
| 143 | foreach ($disallowedHosts as $disallowedHost) { |
| 144 | if (preg_match(self::convertWildcardToPattern($disallowedHost), $host) === 1) { |
| 145 | throw new Exception(sprintf('Hostname %s is in list of disallowed hosts', $host)); |
| 146 | } |
| 147 | } |
| 148 | } |
| 149 | private static function convertWildcardToPattern(string $wildcardHost) : string |
| 150 | { |
| 151 | $flexibleStart = $flexibleEnd = \false; |
| 152 | if (strpos($wildcardHost, '*.') === 0) { |
| 153 | $flexibleStart = \true; |
| 154 | $wildcardHost = substr($wildcardHost, 2); |
| 155 | } |
| 156 | if (\Piwik\Common::stringEndsWith($wildcardHost, '.*')) { |
| 157 | $flexibleEnd = \true; |
| 158 | $wildcardHost = substr($wildcardHost, 0, -2); |
| 159 | } |
| 160 | $pattern = preg_quote($wildcardHost); |
| 161 | if ($flexibleStart) { |
| 162 | $pattern = '.*\\.' . $pattern; |
| 163 | } |
| 164 | if ($flexibleEnd) { |
| 165 | $pattern .= '\\..*'; |
| 166 | } |
| 167 | return '/^' . $pattern . '$/i'; |
| 168 | } |
| 169 | /** |
| 170 | * Sends an HTTP request using the specified transport method. |
| 171 | * |
| 172 | * @param string|null $method |
| 173 | * @param string $aUrl |
| 174 | * @param int $timeout in seconds |
| 175 | * @param string|null $userAgent |
| 176 | * @param string|null $destinationPath |
| 177 | * @param resource|null $file |
| 178 | * @param int $followDepth |
| 179 | * @param string|false $acceptLanguage Accept-language header |
| 180 | * @param bool $acceptInvalidSslCertificate Only used with $method == 'curl'. If set to true (NOT recommended!) the SSL certificate will not be checked |
| 181 | * @param array|false $byteRange For Range: header. Should be two element array of bytes, eg, array(0, 1024) |
| 182 | * Doesn't work w/ fopen method. |
| 183 | * @param bool $getExtendedInfo True to return status code, headers & response, false if just response. |
| 184 | * @param string $httpMethod The HTTP method to use. Defaults to `'GET'`. |
| 185 | * @param string|null $httpUsername HTTP Auth username |
| 186 | * @param string|null $httpPassword HTTP Auth password |
| 187 | * @param array|string|null $requestBody If $httpMethod is 'POST' this may accept an array of variables or a string that needs to be posted |
| 188 | * @param array $additionalHeaders List of additional headers to set for the request |
| 189 | * @param bool|null $forcePost If true, forces POST redirects to remain POST requests (curl only). Ignored on the |
| 190 | * `$validateEgressIp` path, where the method and body are retained on same-origin |
| 191 | * redirects only and cross-origin redirects are downgraded to GET without a body. |
| 192 | * @param bool $checkHostIsAllowed whether we should check if the target host is allowed or not. This should only |
| 193 | * be set to false when using a hardcoded URL. |
| 194 | * @param bool $validateEgressIp when true, the request is served over the SSRF-safe path: public-IP validation, |
| 195 | * manual per-hop redirect re-validation and connection pinning. See |
| 196 | * {@see sendHttpRequest()} for the full contract. |
| 197 | * |
| 198 | * @return ($destinationPath is null ? ($getExtendedInfo is true ? array{status: ?int, headers?: ?array, data?: ?string} : string|false) : bool) |
| 199 | * @throws Exception |
| 200 | */ |
| 201 | public static function sendHttpRequestBy($method, $aUrl, $timeout, $userAgent = null, $destinationPath = null, $file = null, $followDepth = 0, $acceptLanguage = \false, $acceptInvalidSslCertificate = \false, $byteRange = \false, $getExtendedInfo = \false, $httpMethod = 'GET', $httpUsername = null, $httpPassword = null, $requestBody = null, $additionalHeaders = array(), $forcePost = null, $checkHostIsAllowed = \true, $validateEgressIp = \false) |
| 202 | { |
| 203 | if ($followDepth > 5) { |
| 204 | throw new Exception('Too many redirects (' . $followDepth . ')'); |
| 205 | } |
| 206 | $aUrl = preg_replace('/[\\x00-\\x1F\\x7F]/', '', trim($aUrl)); |
| 207 | $parsedUrl = @parse_url($aUrl); |
| 208 | if (empty($parsedUrl['scheme'])) { |
| 209 | throw new Exception('Missing scheme in given url'); |
| 210 | } |
| 211 | $allowedProtocols = GeneralConfig::getConfigValue('allowed_outgoing_protocols'); |
| 212 | $isAllowed = \false; |
| 213 | foreach (explode(',', $allowedProtocols) as $protocol) { |
| 214 | if (strtolower($parsedUrl['scheme']) === strtolower(trim($protocol))) { |
| 215 | $isAllowed = \true; |
| 216 | break; |
| 217 | } |
| 218 | } |
| 219 | if (!$isAllowed) { |
| 220 | throw new Exception(sprintf('Protocol %s not in list of allowed protocols: %s', $parsedUrl['scheme'], $allowedProtocols)); |
| 221 | } |
| 222 | if ($checkHostIsAllowed) { |
| 223 | self::assertHostNotBlocked($parsedUrl['host'] ?? null); |
| 224 | } |
| 225 | // SSRF-safe path: only curl can pin the validated address |
| 226 | // we handle redirects manually below, and refuse any other transport |
| 227 | // or a forward proxy rather than fetch unsafely. |
| 228 | $pinnedResolveEntry = null; |
| 229 | if ($validateEgressIp) { |
| 230 | if ($method !== 'curl') { |
| 231 | throw new EgressBlockedException('SSRF-safe HTTP requests require the curl transport.'); |
| 232 | } |
| 233 | if (!self::isCurlEnabled()) { |
| 234 | throw new EgressBlockedException('SSRF-safe HTTP requests require the curl PHP extension.'); |
| 235 | } |
| 236 | // Restrict to http(s): other schemes have different default ports |
| 237 | $scheme = strtolower((string) $parsedUrl['scheme']); |
| 238 | if ($scheme !== 'http' && $scheme !== 'https') { |
| 239 | throw new EgressBlockedException('SSRF-safe HTTP requests only support the http and https schemes.'); |
| 240 | } |
| 241 | [$configuredProxyHost] = self::getProxyConfiguration($aUrl); |
| 242 | if (!empty($configuredProxyHost)) { |
| 243 | throw new EgressBlockedException('SSRF-safe HTTP requests cannot be routed through a configured proxy.'); |
| 244 | } |
| 245 | $effectivePort = isset($parsedUrl['port']) ? (int) $parsedUrl['port'] : ($scheme === 'https' ? 443 : 80); |
| 246 | // Resolved via DI so tests can substitute a validator that accepts the local fixture server. |
| 247 | [$canonicalHost, $pinnedIp] = StaticContainer::get(EgressHostValidator::class)->resolveTarget((string) ($parsedUrl['host'] ?? '')); |
| 248 | // Rewrite the URL to the canonical host when it differs (IDN folding, casing, a trailing dot) |
| 249 | if ($canonicalHost !== trim((string) ($parsedUrl['host'] ?? ''), '[]')) { |
| 250 | $aUrl = self::replaceUrlHost($parsedUrl, $canonicalHost); |
| 251 | // Re-check the blocklist against the host curl will actually connect to. The check |
| 252 | // above ran on the raw host, so canonicalisation (a trailing dot, IDN folding) could |
| 253 | // otherwise slip a blocked host like "s3.amazonaws.com." past the wildcard rules. |
| 254 | if ($checkHostIsAllowed) { |
| 255 | self::assertHostNotBlocked($canonicalHost); |
| 256 | } |
| 257 | } |
| 258 | // For a DNS host, pin the name to the validated IP so curl cannot re-resolve to |
| 259 | // a different address. An IP literal (canonicalHost === pinnedIp) needs no pin. |
| 260 | // @todo PHP 8.1 min: strpos($pinnedIp, ':') !== false can become str_contains(). |
| 261 | if ($canonicalHost !== $pinnedIp) { |
| 262 | $pinnedAddress = strpos($pinnedIp, ':') !== \false ? '[' . $pinnedIp . ']' : $pinnedIp; |
| 263 | $pinnedResolveEntry = $canonicalHost . ':' . $effectivePort . ':' . $pinnedAddress; |
| 264 | } |
| 265 | } |
| 266 | // When sending an insecure request, but https is forced, and we would care about valid certificates, log a warning |
| 267 | // Note: accepting invalid ssl certificates should only be used when requesting data from a configured website |
| 268 | if ($parsedUrl['scheme'] === 'http' && \Piwik\SettingsPiwik::isHttpsForced() && $acceptInvalidSslCertificate === \false) { |
| 269 | \Piwik\Log::warning('Matomo is configured to force HTTPS, but is sending an insecure request to ' . $aUrl); |
| 270 | } |
| 271 | $contentLength = 0; |
| 272 | $fileLength = 0; |
| 273 | if (!empty($requestBody) && is_array($requestBody)) { |
| 274 | $requestBodyQuery = self::buildQuery($requestBody); |
| 275 | } else { |
| 276 | $requestBodyQuery = $requestBody; |
| 277 | } |
| 278 | if (empty($userAgent)) { |
| 279 | $userAgent = self::getUserAgent(); |
| 280 | } |
| 281 | $via = 'Via: ' . (isset($_SERVER['HTTP_VIA']) && !empty($_SERVER['HTTP_VIA']) ? $_SERVER['HTTP_VIA'] . ', ' : '') . \Piwik\Version::VERSION . ' ' . ($userAgent ? " ({$userAgent})" : ''); |
| 282 | // range header |
| 283 | $rangeBytes = ''; |
| 284 | $rangeHeader = ''; |
| 285 | if (!empty($byteRange)) { |
| 286 | $rangeBytes = $byteRange[0] . '-' . $byteRange[1]; |
| 287 | $rangeHeader = 'Range: bytes=' . $rangeBytes . "\r\n"; |
| 288 | } |
| 289 | [$proxyHost, $proxyPort, $proxyUser, $proxyPassword] = self::getProxyConfiguration($aUrl); |
| 290 | /** @var int|null $status */ |
| 291 | $status = null; |
| 292 | /** @var array<string, string> $headers */ |
| 293 | $headers = array(); |
| 294 | /** @var string|null $response */ |
| 295 | $response = null; |
| 296 | $httpAuthIsUsed = !empty($httpUsername) || !empty($httpPassword); |
| 297 | $httpAuth = ''; |
| 298 | if ($httpAuthIsUsed) { |
| 299 | $httpAuth = 'Authorization: Basic ' . base64_encode($httpUsername . ':' . $httpPassword) . "\r\n"; |
| 300 | } |
| 301 | $httpEventParams = array('httpMethod' => $httpMethod, 'body' => $requestBody, 'userAgent' => $userAgent, 'timeout' => $timeout, 'headers' => array_map('trim', array_filter(array_merge([$rangeHeader, $via, $httpAuth, $acceptLanguage], $additionalHeaders))), 'verifySsl' => !$acceptInvalidSslCertificate, 'destinationPath' => $destinationPath, 'validateEgressIp' => $validateEgressIp); |
| 302 | /** |
| 303 | * Triggered to send an HTTP request. Allows plugins to resolve the HTTP request themselves or to find out |
| 304 | * when an HTTP request is triggered to log this information for example to a monitoring tool. |
| 305 | * |
| 306 | * @param string $url The URL that needs to be requested |
| 307 | * @param array $params HTTP params like |
| 308 | * - 'httpMethod' (eg GET, POST, ...), |
| 309 | * - 'body' the request body if the HTTP method needs to be posted |
| 310 | * - 'userAgent' |
| 311 | * - 'timeout' After how many seconds a request should time out |
| 312 | * - 'headers' An array of header strings like array('Accept-Language: en', '...') |
| 313 | * - 'verifySsl' A boolean whether SSL certificate should be verified |
| 314 | * - 'destinationPath' If set, the response of the HTTP request should be saved to this file |
| 315 | * - 'validateEgressIp' Whether the caller asked for SSRF-safe semantics. A listener |
| 316 | * resolving the request itself must honour them or leave it unhandled |
| 317 | * @param string &$response A plugin listening to this event should assign the HTTP response it received to this variable, for example "{value: true}" |
| 318 | * @param int &$status A plugin listening to this event should assign the HTTP status code it received to this variable, for example "200" |
| 319 | * @param array &$headers A plugin listening to this event should assign the HTTP headers it received to this variable, eg array('Content-Length' => '5') |
| 320 | */ |
| 321 | \Piwik\Piwik::postEvent('Http.sendHttpRequest', array($aUrl, $httpEventParams, &$response, &$status, &$headers)); |
| 322 | if ($response !== null || $status !== null || !empty($headers)) { |
| 323 | // was handled by event above... |
| 324 | /** |
| 325 | * described below |
| 326 | * @ignore |
| 327 | */ |
| 328 | \Piwik\Piwik::postEvent('Http.sendHttpRequest.end', array($aUrl, $httpEventParams, &$response, &$status, &$headers)); |
| 329 | if ($destinationPath && file_exists($destinationPath)) { |
| 330 | return \true; |
| 331 | } |
| 332 | if ($getExtendedInfo) { |
| 333 | return array('status' => $status, 'headers' => $headers, 'data' => $response); |
| 334 | } else { |
| 335 | return trim($response); |
| 336 | } |
| 337 | } |
| 338 | if ($method == 'socket') { |
| 339 | if (!self::isSocketEnabled()) { |
| 340 | // can be triggered in tests |
| 341 | throw new Exception("HTTP socket support is not enabled (php function fsockopen is not available) "); |
| 342 | } |
| 343 | // initialization |
| 344 | $url = @parse_url($aUrl); |
| 345 | if ($url === \false || !isset($url['scheme'])) { |
| 346 | throw new Exception('Malformed URL: ' . $aUrl); |
| 347 | } |
| 348 | if ($url['scheme'] != 'http' && $url['scheme'] != 'https') { |
| 349 | throw new Exception('Invalid protocol/scheme: ' . $url['scheme']); |
| 350 | } |
| 351 | $host = $url['host']; |
| 352 | $port = isset($url['port']) ? $url['port'] : ('https' == $url['scheme'] ? 443 : 80); |
| 353 | $path = isset($url['path']) ? $url['path'] : '/'; |
| 354 | if (isset($url['query'])) { |
| 355 | $path .= '?' . $url['query']; |
| 356 | } |
| 357 | $errno = null; |
| 358 | $errstr = null; |
| 359 | if (!empty($proxyHost) && !empty($proxyPort) || !empty($byteRange)) { |
| 360 | $httpVer = '1.1'; |
| 361 | } else { |
| 362 | $httpVer = '1.0'; |
| 363 | } |
| 364 | $proxyAuth = null; |
| 365 | if (!empty($proxyHost) && !empty($proxyPort)) { |
| 366 | $connectHost = $proxyHost; |
| 367 | $connectPort = $proxyPort; |
| 368 | if (!empty($proxyUser) && !empty($proxyPassword)) { |
| 369 | $proxyAuth = 'Proxy-Authorization: Basic ' . base64_encode("{$proxyUser}:{$proxyPassword}") . "\r\n"; |
| 370 | } |
| 371 | $requestHeader = "{$httpMethod} {$aUrl} HTTP/{$httpVer}\r\n"; |
| 372 | } else { |
| 373 | $connectHost = $host; |
| 374 | $connectPort = $port; |
| 375 | $requestHeader = "{$httpMethod} {$path} HTTP/{$httpVer}\r\n"; |
| 376 | if ('https' == $url['scheme']) { |
| 377 | $connectHost = 'tls://' . $connectHost; |
| 378 | } |
| 379 | } |
| 380 | // connection attempt |
| 381 | if (($fsock = @fsockopen($connectHost, $connectPort, $errno, $errstr, $timeout)) === \false || !is_resource($fsock)) { |
| 382 | if (is_resource($file)) { |
| 383 | @fclose($file); |
| 384 | } |
| 385 | throw new Exception("Error while connecting to: {$host}. Please try again later. {$errstr}"); |
| 386 | } |
| 387 | // send HTTP request header |
| 388 | $requestHeader .= "Host: {$host}" . ($port != 80 && ('https' == $url['scheme'] && $port != 443) ? ':' . $port : '') . "\r\n" . ($httpAuth ? $httpAuth : '') . ($proxyAuth ? $proxyAuth : '') . 'User-Agent: ' . $userAgent . "\r\n" . ($acceptLanguage ? $acceptLanguage . "\r\n" : '') . $via . "\r\n" . $rangeHeader . (!empty($additionalHeaders) ? implode("\r\n", $additionalHeaders) . "\r\n" : '') . "Connection: close\r\n"; |
| 389 | fwrite($fsock, $requestHeader); |
| 390 | if (strtolower($httpMethod) === 'post' && !empty($requestBodyQuery)) { |
| 391 | fwrite($fsock, self::buildHeadersForPost($requestBodyQuery)); |
| 392 | fwrite($fsock, "\r\n"); |
| 393 | fwrite($fsock, $requestBodyQuery); |
| 394 | } else { |
| 395 | fwrite($fsock, "\r\n"); |
| 396 | } |
| 397 | $streamMetaData = array('timed_out' => \false); |
| 398 | @stream_set_blocking($fsock, \true); |
| 399 | if (function_exists('stream_set_timeout')) { |
| 400 | @stream_set_timeout($fsock, $timeout); |
| 401 | } elseif (function_exists('socket_set_timeout')) { |
| 402 | @socket_set_timeout($fsock, $timeout); |
| 403 | } |
| 404 | // process header |
| 405 | $status = null; |
| 406 | while (!feof($fsock)) { |
| 407 | $line = fgets($fsock, 4096); |
| 408 | $streamMetaData = @stream_get_meta_data($fsock); |
| 409 | if ($streamMetaData['timed_out']) { |
| 410 | if (is_resource($file)) { |
| 411 | @fclose($file); |
| 412 | } |
| 413 | @fclose($fsock); |
| 414 | throw new Exception('Timed out waiting for server response'); |
| 415 | } |
| 416 | // a blank line marks the end of the server response header |
| 417 | if (rtrim($line, "\r\n") == '') { |
| 418 | break; |
| 419 | } |
| 420 | // parse first line of server response header |
| 421 | if (!$status) { |
| 422 | // expect first line to be HTTP response status line, e.g., HTTP/1.1 200 OK |
| 423 | if (!preg_match('~^HTTP/(\\d\\.\\d)\\s+(\\d+)(\\s*.*)?~', $line, $m)) { |
| 424 | if (is_resource($file)) { |
| 425 | @fclose($file); |
| 426 | } |
| 427 | @fclose($fsock); |
| 428 | throw new Exception('Expected server response code. Got ' . rtrim($line, "\r\n")); |
| 429 | } |
| 430 | $status = (int) $m[2]; |
| 431 | // Informational 1xx or Client Error 4xx |
| 432 | if ($status < 200 || $status >= 400) { |
| 433 | if (is_resource($file)) { |
| 434 | @fclose($file); |
| 435 | } |
| 436 | @fclose($fsock); |
| 437 | if (!$getExtendedInfo) { |
| 438 | return \false; |
| 439 | } else { |
| 440 | return array('status' => $status); |
| 441 | } |
| 442 | } |
| 443 | continue; |
| 444 | } |
| 445 | // handle redirect |
| 446 | if (preg_match('/^Location:\\s*(.+)/', rtrim($line, "\r\n"), $m)) { |
| 447 | if (is_resource($file)) { |
| 448 | @fclose($file); |
| 449 | } |
| 450 | @fclose($fsock); |
| 451 | // Successful 2xx vs Redirect 3xx |
| 452 | if ($status < 300) { |
| 453 | throw new Exception('Unexpected redirect to Location: ' . rtrim($line) . ' for status code ' . $status); |
| 454 | } |
| 455 | return self::sendHttpRequestBy($method, trim($m[1]), $timeout, $userAgent, $destinationPath, $file, $followDepth + 1, $acceptLanguage, $acceptInvalidSslCertificate = \false, $byteRange, $getExtendedInfo, $httpMethod, $httpUsername, $httpPassword, $requestBodyQuery, $additionalHeaders); |
| 456 | } |
| 457 | // save expected content length for later verification |
| 458 | if (preg_match('/^Content-Length:\\s*(\\d+)/', $line, $m)) { |
| 459 | $contentLength = (int) $m[1]; |
| 460 | } |
| 461 | self::parseHeaderLine($headers, $line); |
| 462 | } |
| 463 | if (feof($fsock) && $httpMethod != 'HEAD') { |
| 464 | throw new Exception('Unexpected end of transmission'); |
| 465 | } |
| 466 | // process content/body |
| 467 | $response = ''; |
| 468 | while (!feof($fsock)) { |
| 469 | $line = fread($fsock, 8192); |
| 470 | $streamMetaData = @stream_get_meta_data($fsock); |
| 471 | if ($streamMetaData['timed_out']) { |
| 472 | if (is_resource($file)) { |
| 473 | @fclose($file); |
| 474 | } |
| 475 | @fclose($fsock); |
| 476 | throw new Exception('Timed out waiting for server response'); |
| 477 | } |
| 478 | $fileLength += strlen($line); |
| 479 | if (is_resource($file)) { |
| 480 | // save to file |
| 481 | fwrite($file, $line); |
| 482 | } else { |
| 483 | // concatenate to response string |
| 484 | $response .= $line; |
| 485 | } |
| 486 | } |
| 487 | // determine success or failure |
| 488 | @fclose(@$fsock); |
| 489 | } elseif ($method == 'fopen') { |
| 490 | $response = \false; |
| 491 | // we make sure the request takes less than a few seconds to fail |
| 492 | // we create a stream_context (works in php >= 5.2.1) |
| 493 | // we also set the socket_timeout (for php < 5.2.1) |
| 494 | $default_socket_timeout = @ini_get('default_socket_timeout'); |
| 495 | @ini_set('default_socket_timeout', (string) $timeout); |
| 496 | $ctx = null; |
| 497 | if (function_exists('stream_context_create')) { |
| 498 | $stream_options = array('http' => array( |
| 499 | 'header' => 'User-Agent: ' . $userAgent . "\r\n" . ($httpAuth ? $httpAuth : '') . ($acceptLanguage ? $acceptLanguage . "\r\n" : '') . $via . "\r\n" . (!empty($additionalHeaders) ? implode("\r\n", $additionalHeaders) . "\r\n" : '') . $rangeHeader, |
| 500 | 'max_redirects' => 5, |
| 501 | // PHP 5.1.0 |
| 502 | 'timeout' => $timeout, |
| 503 | )); |
| 504 | if (!empty($proxyHost) && !empty($proxyPort)) { |
| 505 | $stream_options['http']['proxy'] = 'tcp://' . $proxyHost . ':' . $proxyPort; |
| 506 | $stream_options['http']['request_fulluri'] = \true; |
| 507 | // required by squid proxy |
| 508 | if (!empty($proxyUser) && !empty($proxyPassword)) { |
| 509 | $stream_options['http']['header'] .= 'Proxy-Authorization: Basic ' . base64_encode("{$proxyUser}:{$proxyPassword}") . "\r\n"; |
| 510 | } |
| 511 | } |
| 512 | if (strtolower($httpMethod) === 'post' && !empty($requestBodyQuery)) { |
| 513 | $postHeader = self::buildHeadersForPost($requestBodyQuery); |
| 514 | $postHeader .= "\r\n"; |
| 515 | $stream_options['http']['method'] = 'POST'; |
| 516 | $stream_options['http']['header'] .= $postHeader; |
| 517 | $stream_options['http']['content'] = $requestBodyQuery; |
| 518 | } |
| 519 | $ctx = stream_context_create($stream_options); |
| 520 | } |
| 521 | // save to file |
| 522 | if (is_resource($file)) { |
| 523 | if (!($handle = fopen($aUrl, 'rb', \false, $ctx))) { |
| 524 | throw new Exception("Unable to open {$aUrl}"); |
| 525 | } |
| 526 | while (!feof($handle)) { |
| 527 | $response = fread($handle, 8192); |
| 528 | $fileLength += strlen($response); |
| 529 | fwrite($file, $response); |
| 530 | } |
| 531 | fclose($handle); |
| 532 | if (function_exists('http_get_last_response_headers')) { |
| 533 | $http_response_header = http_get_last_response_headers(); |
| 534 | } |
| 535 | } else { |
| 536 | $response = @file_get_contents($aUrl, \false, $ctx); |
| 537 | if (function_exists('http_get_last_response_headers')) { |
| 538 | $http_response_header = http_get_last_response_headers(); |
| 539 | } |
| 540 | // try to get http status code from response headers |
| 541 | if (!empty($http_response_header) && preg_match('~^HTTP/(\\d\\.\\d)\\s+(\\d+)(\\s*.*)?~', implode("\n", $http_response_header), $m)) { |
| 542 | $status = (int) $m[2]; |
| 543 | } |
| 544 | if (!$status && $response === \false) { |
| 545 | $error = \Piwik\ErrorHandler::getLastError(); |
| 546 | throw new \Exception($error); |
| 547 | } |
| 548 | $fileLength = strlen($response); |
| 549 | } |
| 550 | foreach ($http_response_header as $line) { |
| 551 | self::parseHeaderLine($headers, $line); |
| 552 | } |
| 553 | // restore the socket_timeout value |
| 554 | if (!empty($default_socket_timeout)) { |
| 555 | @ini_set('default_socket_timeout', $default_socket_timeout); |
| 556 | } |
| 557 | } elseif ($method == 'curl') { |
| 558 | if (!self::isCurlEnabled()) { |
| 559 | // can be triggered in tests |
| 560 | throw new Exception("CURL is not enabled in php.ini, but is being used."); |
| 561 | } |
| 562 | $ch = @curl_init(); |
| 563 | if (!empty($proxyHost) && !empty($proxyPort)) { |
| 564 | @curl_setopt($ch, \CURLOPT_PROXY, $proxyHost . ':' . $proxyPort); |
| 565 | if (!empty($proxyUser) && !empty($proxyPassword)) { |
| 566 | // PROXYAUTH defaults to BASIC |
| 567 | @curl_setopt($ch, \CURLOPT_PROXYUSERPWD, $proxyUser . ':' . $proxyPassword); |
| 568 | } |
| 569 | } |
| 570 | $curl_options = array( |
| 571 | // curl options (sorted oldest to newest) |
| 572 | \CURLOPT_URL => $aUrl, |
| 573 | \CURLOPT_USERAGENT => $userAgent, |
| 574 | \CURLOPT_HTTPHEADER => array_merge(array($via, $acceptLanguage), $additionalHeaders), |
| 575 | // only get header info if not saving directly to file |
| 576 | \CURLOPT_HEADER => is_resource($file) ? \false : \true, |
| 577 | \CURLOPT_CONNECTTIMEOUT => $timeout, |
| 578 | \CURLOPT_TIMEOUT => $timeout, |
| 579 | ); |
| 580 | if ($rangeBytes) { |
| 581 | curl_setopt($ch, \CURLOPT_RANGE, $rangeBytes); |
| 582 | } else { |
| 583 | // see https://github.com/matomo-org/matomo/pull/17009 for more info |
| 584 | // NOTE: we only do this when CURLOPT_RANGE is not being used, because when using both the |
| 585 | // response is empty. |
| 586 | $curl_options[\CURLOPT_ENCODING] = ""; |
| 587 | } |
| 588 | // Case core:archive command is triggering archiving on https:// and the certificate is not valid |
| 589 | if ($acceptInvalidSslCertificate) { |
| 590 | $curl_options += array(\CURLOPT_SSL_VERIFYHOST => \false, \CURLOPT_SSL_VERIFYPEER => \false); |
| 591 | } |
| 592 | @curl_setopt($ch, \CURLOPT_CUSTOMREQUEST, $httpMethod); |
| 593 | if ($httpMethod == 'HEAD') { |
| 594 | @curl_setopt($ch, \CURLOPT_NOBODY, \true); |
| 595 | } |
| 596 | if (in_array(strtolower($httpMethod), ['post', 'put']) && !empty($requestBodyQuery)) { |
| 597 | curl_setopt($ch, \CURLOPT_POST, 1); |
| 598 | curl_setopt($ch, \CURLOPT_POSTFIELDS, $requestBodyQuery); |
| 599 | } |
| 600 | if (!empty($httpUsername) && !empty($httpPassword)) { |
| 601 | $curl_options += array(\CURLOPT_USERPWD => $httpUsername . ':' . $httpPassword); |
| 602 | } |
| 603 | @curl_setopt_array($ch, $curl_options); |
| 604 | self::configCurlCertificate($ch); |
| 605 | if ($validateEgressIp) { |
| 606 | // Follow redirects manually so every hop is re-validated, and pin |
| 607 | // the connection to the address we validated to close the DNS-rebinding window. |
| 608 | // CURLOPT_RESOLVE keeps the original hostname for SNI and certificate checks. |
| 609 | // @todo when switching to a new HTTP library: this transport-specific |
| 610 | // pinning/redirect wiring should probably be re-implemented against it (e.g. Guzzle's |
| 611 | // curl.options + redirect middleware). EgressHostValidator is reusable as-is. |
| 612 | @curl_setopt($ch, \CURLOPT_FOLLOWLOCATION, \false); |
| 613 | // Disable any environment proxy (http_proxy etc.) so it cannot re-resolve the host and bypass the pin. |
| 614 | @curl_setopt($ch, \CURLOPT_PROXY, ''); |
| 615 | if ($pinnedResolveEntry !== null) { |
| 616 | @curl_setopt($ch, \CURLOPT_RESOLVE, array($pinnedResolveEntry)); |
| 617 | } |
| 618 | } |
| 619 | /* |
| 620 | * as of php 5.2.0, CURLOPT_FOLLOWLOCATION can't be set if |
| 621 | * in safe_mode or open_basedir is set |
| 622 | */ |
| 623 | if (!$validateEgressIp && (string) ini_get('safe_mode') == '' && ini_get('open_basedir') == '') { |
| 624 | $protocols = 0; |
| 625 | foreach (explode(',', $allowedProtocols) as $protocol) { |
| 626 | if (defined('CURLPROTO_' . strtoupper(trim($protocol)))) { |
| 627 | $protocols |= constant('CURLPROTO_' . strtoupper(trim($protocol))); |
| 628 | } |
| 629 | } |
| 630 | $curl_options = array( |
| 631 | // curl options (sorted oldest to newest) |
| 632 | \CURLOPT_FOLLOWLOCATION => \true, |
| 633 | \CURLOPT_REDIR_PROTOCOLS => $protocols, |
| 634 | \CURLOPT_MAXREDIRS => 5, |
| 635 | ); |
| 636 | if ($forcePost) { |
| 637 | $curl_options[\CURLOPT_POSTREDIR] = \CURL_REDIR_POST_ALL; |
| 638 | } |
| 639 | @curl_setopt_array($ch, $curl_options); |
| 640 | } |
| 641 | if (is_resource($file)) { |
| 642 | // write output directly to file |
| 643 | @curl_setopt($ch, \CURLOPT_FILE, $file); |
| 644 | } else { |
| 645 | // internal to ext/curl |
| 646 | @curl_setopt($ch, \CURLOPT_RETURNTRANSFER, \true); |
| 647 | } |
| 648 | ob_start(); |
| 649 | $response = @curl_exec($ch); |
| 650 | ob_end_clean(); |
| 651 | if ($response === \true) { |
| 652 | $response = ''; |
| 653 | } elseif ($response === \false) { |
| 654 | $errstr = curl_error($ch); |
| 655 | if ($errstr != '') { |
| 656 | throw new Exception('curl_exec: ' . $errstr . '. Hostname requested was: ' . \Piwik\UrlHelper::getHostFromUrl($aUrl)); |
| 657 | } |
| 658 | $response = ''; |
| 659 | } else { |
| 660 | $header = ''; |
| 661 | // redirects are included in the output html, so we look for the last line that starts w/ HTTP/... |
| 662 | // to split the response |
| 663 | while (substr($response, 0, 5) == "HTTP/") { |
| 664 | $split = explode("\r\n\r\n", $response, 2); |
| 665 | if (count($split) == 2) { |
| 666 | [$header, $response] = $split; |
| 667 | } else { |
| 668 | $response = ''; |
| 669 | $header = reset($split); |
| 670 | } |
| 671 | } |
| 672 | foreach (explode("\r\n", $header) as $line) { |
| 673 | self::parseHeaderLine($headers, $line); |
| 674 | } |
| 675 | } |
| 676 | $contentLength = @curl_getinfo($ch, \CURLINFO_CONTENT_LENGTH_DOWNLOAD); |
| 677 | $fileLength = is_resource($file) ? @curl_getinfo($ch, \CURLINFO_SIZE_DOWNLOAD) : strlen($response); |
| 678 | $status = @curl_getinfo($ch, \CURLINFO_HTTP_CODE); |
| 679 | $elapsed = (float) @curl_getinfo($ch, \CURLINFO_TOTAL_TIME); |
| 680 | $curlRedirectUrl = (string) @curl_getinfo($ch, \CURLINFO_REDIRECT_URL); |
| 681 | @curl_close($ch); |
| 682 | unset($ch); |
| 683 | // SSRF-safe path follows redirects manually so each hop is re-validated and re-pinned. |
| 684 | if ($validateEgressIp && $status >= 300 && $status < 400 && $status !== 304) { |
| 685 | if (is_resource($file)) { |
| 686 | // CURLOPT_HEADER is off for file downloads, so the Location cannot be re-validated: fail closed. |
| 687 | @fclose($file); |
| 688 | if ($destinationPath) { |
| 689 | @unlink($destinationPath); |
| 690 | } |
| 691 | throw new EgressBlockedException('SSRF-safe HTTP requests cannot follow redirects when downloading to a file.'); |
| 692 | } |
| 693 | // Read from curl, not from $headers: the splitter above keeps the last "HTTP/" block |
| 694 | // it finds, so a response body can forge one. |
| 695 | $redirectUrl = $curlRedirectUrl; |
| 696 | if ($redirectUrl !== '') { |
| 697 | // Cross-origin: drop credentials, caller headers and the body, and downgrade to GET. |
| 698 | // $acceptInvalidSslCertificate stays, since it guards no secret past this point and |
| 699 | // dropping it would fail the common http -> https hop for self-signed sites. |
| 700 | if (!self::urlsSameOrigin($aUrl, $redirectUrl)) { |
| 701 | $httpUsername = null; |
| 702 | $httpPassword = null; |
| 703 | $additionalHeaders = array(); |
| 704 | $requestBody = null; |
| 705 | $httpMethod = 'GET'; |
| 706 | $forcePost = null; |
| 707 | } |
| 708 | // Shrink the timeout by what this hop already spent so the whole redirect chain |
| 709 | // stays within the caller's original budget instead of granting it to every hop. |
| 710 | $remainingTimeout = max(1, (int) floor($timeout - $elapsed)); |
| 711 | return self::sendHttpRequestBy($method, $redirectUrl, $remainingTimeout, $userAgent, $destinationPath, $file, $followDepth + 1, $acceptLanguage, $acceptInvalidSslCertificate, $byteRange, $getExtendedInfo, $httpMethod, $httpUsername, $httpPassword, $requestBody, $additionalHeaders, $forcePost, $checkHostIsAllowed, $validateEgressIp); |
| 712 | } |
| 713 | } |
| 714 | } else { |
| 715 | throw new Exception('Invalid request method: ' . $method); |
| 716 | } |
| 717 | if (is_resource($file)) { |
| 718 | fflush($file); |
| 719 | @fclose($file); |
| 720 | $fileSize = filesize($destinationPath); |
| 721 | if ($contentLength > 0 && $fileSize != $contentLength) { |
| 722 | throw new Exception('File size error: ' . $destinationPath . '; expected ' . $contentLength . ' bytes; received ' . $fileLength . ' bytes; saved ' . $fileSize . ' bytes to file'); |
| 723 | } |
| 724 | return \true; |
| 725 | } |
| 726 | /** |
| 727 | * Triggered when an HTTP request finished. A plugin can for example listen to this and alter the response, |
| 728 | * status code, or finish a timer in case the plugin is measuring how long it took to execute the request |
| 729 | * |
| 730 | * @param string $url The URL that needs to be requested |
| 731 | * @param array $params HTTP params like |
| 732 | * - 'httpMethod' (eg GET, POST, ...), |
| 733 | * - 'body' the request body if the HTTP method needs to be posted |
| 734 | * - 'userAgent' |
| 735 | * - 'timeout' After how many seconds a request should time out |
| 736 | * - 'headers' An array of header strings like array('Accept-Language: en', '...') |
| 737 | * - 'verifySsl' A boolean whether SSL certificate should be verified |
| 738 | * - 'destinationPath' If set, the response of the HTTP request should be saved to this file |
| 739 | * - 'validateEgressIp' Whether the caller asked for SSRF-safe semantics. A listener |
| 740 | * resolving the request itself must honour them or leave it unhandled |
| 741 | * @param string &$response The response of the HTTP request, for example "{value: true}" |
| 742 | * @param int &$status The returned HTTP status code, for example "200" |
| 743 | * @param array &$headers The returned headers, eg array('Content-Length' => '5') |
| 744 | */ |
| 745 | \Piwik\Piwik::postEvent('Http.sendHttpRequest.end', array($aUrl, $httpEventParams, &$response, &$status, &$headers)); |
| 746 | if (!$getExtendedInfo) { |
| 747 | return trim($response); |
| 748 | } else { |
| 749 | return array('status' => $status, 'headers' => $headers, 'data' => $response); |
| 750 | } |
| 751 | } |
| 752 | public static function buildQuery($params) |
| 753 | { |
| 754 | return http_build_query($params, '', '&'); |
| 755 | } |
| 756 | private static function buildHeadersForPost(string $requestBody) : string |
| 757 | { |
| 758 | $postHeader = "Content-Type: application/x-www-form-urlencoded\r\n"; |
| 759 | $postHeader .= "Content-Length: " . strlen($requestBody) . "\r\n"; |
| 760 | return $postHeader; |
| 761 | } |
| 762 | /** |
| 763 | * Downloads the next chunk of a specific file. The next chunk's byte range |
| 764 | * is determined by the existing file's size and the expected file size, which |
| 765 | * is stored in the option table before starting a download. The expected |
| 766 | * file size is obtained through a `HEAD` HTTP request. |
| 767 | * |
| 768 | * _Note: this function uses the **Range** HTTP header to accomplish downloading in |
| 769 | * parts. Not every server supports this header._ |
| 770 | * |
| 771 | * The proper use of this function is to call it once per request. The browser |
| 772 | * should continue to send requests to Piwik which will in turn call this method |
| 773 | * until the file has completely downloaded. In this way, the user can be informed |
| 774 | * of a download's progress. |
| 775 | * |
| 776 | * **Example Usage** |
| 777 | * |
| 778 | * ``` |
| 779 | * // browser JavaScript |
| 780 | * var downloadFile = function (isStart) { |
| 781 | * var ajax = new ajaxHelper(); |
| 782 | * ajax.addParams({ |
| 783 | * module: 'MyPlugin', |
| 784 | * action: 'myAction', |
| 785 | * isStart: isStart ? 1 : 0 |
| 786 | * }, 'post'); |
| 787 | * ajax.setCallback(function (response) { |
| 788 | * var progress = response.progress |
| 789 | * // ...update progress... |
| 790 | * |
| 791 | * downloadFile(false); |
| 792 | * }); |
| 793 | * ajax.send(); |
| 794 | * } |
| 795 | * |
| 796 | * downloadFile(true); |
| 797 | * ``` |
| 798 | * |
| 799 | * ``` |
| 800 | * // PHP controller action |
| 801 | * public function myAction() |
| 802 | * { |
| 803 | * $outputPath = PIWIK_INCLUDE_PATH . '/tmp/averybigfile.zip'; |
| 804 | * $isStart = Common::getRequestVar('isStart', 1, 'int'); |
| 805 | * Http::downloadChunk("https://bigfiles.com/averybigfile.zip", $outputPath, $isStart == 1); |
| 806 | * } |
| 807 | * ``` |
| 808 | * |
| 809 | * @param string $url The url to download from. |
| 810 | * @param string $outputPath The path to the file to save/append to. |
| 811 | * @param bool $isContinuation `true` if this is the continuation of a download, |
| 812 | * or if we're starting a fresh one. |
| 813 | * @throws Exception if the file already exists and we're starting a new download, |
| 814 | * if we're trying to continue a download that never started |
| 815 | * @return array |
| 816 | * @api |
| 817 | */ |
| 818 | public static function downloadChunk($url, $outputPath, $isContinuation) |
| 819 | { |
| 820 | // make sure file doesn't already exist if we're starting a new download |
| 821 | if (!$isContinuation && file_exists($outputPath)) { |
| 822 | throw new Exception(\Piwik\Piwik::translate('General_DownloadFail_FileExists', "'" . $outputPath . "'") . ' ' . \Piwik\Piwik::translate('General_DownloadPleaseRemoveExisting')); |
| 823 | } |
| 824 | // if we're starting a download, get the expected file size & save as an option |
| 825 | $downloadOption = $outputPath . '_expectedDownloadSize'; |
| 826 | if (!$isContinuation) { |
| 827 | $expectedFileSizeResult = \Piwik\Http::sendHttpRequest($url, $timeout = 300, $userAgent = null, $destinationPath = null, $followDepth = 0, $acceptLanguage = \false, $byteRange = \false, $getExtendedInfo = \true, $httpMethod = 'HEAD'); |
| 828 | $expectedFileSize = 0; |
| 829 | if (isset($expectedFileSizeResult['headers']['Content-Length'])) { |
| 830 | $expectedFileSize = (int) $expectedFileSizeResult['headers']['Content-Length']; |
| 831 | } |
| 832 | if ($expectedFileSize == 0) { |
| 833 | \Piwik\Log::info("HEAD request for '%s' failed, got following: %s", $url, print_r($expectedFileSizeResult, \true)); |
| 834 | throw new Exception(\Piwik\Piwik::translate('General_DownloadFail_HttpRequestFail')); |
| 835 | } |
| 836 | \Piwik\Option::set($downloadOption, (string) $expectedFileSize); |
| 837 | } else { |
| 838 | $expectedFileSize = \Piwik\Option::get($downloadOption); |
| 839 | if ($expectedFileSize === \false) { |
| 840 | // sanity check |
| 841 | throw new Exception("Trying to continue a download that never started?! That's not supposed to happen..."); |
| 842 | } |
| 843 | $expectedFileSize = (int) $expectedFileSize; |
| 844 | } |
| 845 | // if existing file is already big enough, then fail so we don't accidentally overwrite |
| 846 | // existing DB |
| 847 | $existingSize = file_exists($outputPath) ? filesize($outputPath) : 0; |
| 848 | if ($existingSize >= $expectedFileSize) { |
| 849 | throw new Exception(\Piwik\Piwik::translate('General_DownloadFail_FileExistsContinue', "'" . $outputPath . "'") . ' ' . \Piwik\Piwik::translate('General_DownloadPleaseRemoveExisting')); |
| 850 | } |
| 851 | // download a chunk of the file |
| 852 | $result = \Piwik\Http::sendHttpRequest($url, $timeout = 300, $userAgent = null, $destinationPath = null, $followDepth = 0, $acceptLanguage = \false, $byteRange = array($existingSize, min($existingSize + 1024 * 1024 - 1, $expectedFileSize)), $getExtendedInfo = \true); |
| 853 | if ($result['status'] < 200 || $result['status'] > 299) { |
| 854 | $result['data'] = self::truncateStr($result['data'], 1024); |
| 855 | \Piwik\Log::info("Failed to download range '%s-%s' of file from url '%s'. Got result: %s", $byteRange[0], $byteRange[1], $url, print_r($result, \true)); |
| 856 | throw new Exception(\Piwik\Piwik::translate('General_DownloadFail_HttpRequestFail')); |
| 857 | } |
| 858 | // write chunk to file |
| 859 | $f = fopen($outputPath, 'ab'); |
| 860 | fwrite($f, $result['data']); |
| 861 | fclose($f); |
| 862 | clearstatcache($clear_realpath_cache = \true, $outputPath); |
| 863 | return array('current_size' => filesize($outputPath), 'expected_file_size' => $expectedFileSize); |
| 864 | } |
| 865 | /** |
| 866 | * Will configure CURL handle $ch |
| 867 | * to use local list of Certificate Authorities, |
| 868 | */ |
| 869 | public static function configCurlCertificate(&$ch) |
| 870 | { |
| 871 | $cacertPath = GeneralConfig::getConfigValue('custom_cacert_pem'); |
| 872 | if (empty($cacertPath)) { |
| 873 | $cacertPath = CaBundle::getBundledCaBundlePath(); |
| 874 | } |
| 875 | @curl_setopt($ch, \CURLOPT_CAINFO, $cacertPath); |
| 876 | } |
| 877 | public static function getUserAgent() |
| 878 | { |
| 879 | return !empty($_SERVER['HTTP_USER_AGENT']) ? $_SERVER['HTTP_USER_AGENT'] : 'Matomo/' . \Piwik\Version::VERSION; |
| 880 | } |
| 881 | public static function getClientHintsFromServerVariables() : array |
| 882 | { |
| 883 | $clientHints = []; |
| 884 | foreach ($_SERVER as $key => $value) { |
| 885 | if (0 === strpos(strtolower($key), strtolower('HTTP_SEC_CH_UA')) || 'X_HTTP_REQUESTED_WITH' === strtoupper($key)) { |
| 886 | $clientHints[$key] = $value; |
| 887 | } |
| 888 | } |
| 889 | ksort($clientHints); |
| 890 | return $clientHints; |
| 891 | } |
| 892 | /** |
| 893 | * Fetches a file located at `$url` and saves it to `$destinationPath`. |
| 894 | * |
| 895 | * @param string $url The URL of the file to download. |
| 896 | * @param string $destinationPath The path to download the file to. |
| 897 | * @param int $tries (deprecated) |
| 898 | * @param int $timeout The amount of seconds to wait before aborting the HTTP request. |
| 899 | * @return string|bool |
| 900 | * @throws Exception if the response cannot be saved to `$destinationPath`, if the HTTP response cannot be sent, |
| 901 | * if there are more than 5 redirects or if the request times out. |
| 902 | * @phpstan-return ($destinationPath is null ? false|string : bool) |
| 903 | * @api |
| 904 | */ |
| 905 | public static function fetchRemoteFile($url, $destinationPath = null, $tries = 0, $timeout = 10) |
| 906 | { |
| 907 | @ignore_user_abort(\true); |
| 908 | \Piwik\SettingsServer::setMaxExecutionTime(0); |
| 909 | return self::sendHttpRequest($url, $timeout, 'Update', $destinationPath); |
| 910 | } |
| 911 | /** |
| 912 | * Utility function, parses an HTTP header line into key/value & sets header |
| 913 | * array with them. |
| 914 | * |
| 915 | * @param array $headers |
| 916 | * @param string $line |
| 917 | */ |
| 918 | private static function parseHeaderLine(&$headers, $line) : void |
| 919 | { |
| 920 | $parts = explode(':', $line, 2); |
| 921 | if (count($parts) == 1) { |
| 922 | return; |
| 923 | } |
| 924 | [$name, $value] = $parts; |
| 925 | $name = trim($name); |
| 926 | $headers[$name] = trim($value); |
| 927 | /** |
| 928 | * With HTTP/2 Cloudflare is passing headers in lowercase (e.g. 'content-type' instead of 'Content-Type') |
| 929 | * which breaks any code which uses the header data. |
| 930 | */ |
| 931 | $camelName = ucwords($name, '-'); |
| 932 | if ($camelName !== $name) { |
| 933 | $headers[$camelName] = trim($value); |
| 934 | } |
| 935 | } |
| 936 | /** |
| 937 | * Whether two URLs share the same origin (scheme, host and effective port). Fails closed: |
| 938 | * a parse failure or missing component counts as a different origin. |
| 939 | */ |
| 940 | private static function urlsSameOrigin(string $urlA, string $urlB) : bool |
| 941 | { |
| 942 | $a = @parse_url($urlA); |
| 943 | $b = @parse_url($urlB); |
| 944 | if (!is_array($a) || !is_array($b) || !isset($a['host'], $b['host'], $a['scheme'], $b['scheme'])) { |
| 945 | return \false; |
| 946 | } |
| 947 | $schemeA = strtolower($a['scheme']); |
| 948 | $schemeB = strtolower($b['scheme']); |
| 949 | if ($schemeA !== $schemeB || strcasecmp($a['host'], $b['host']) !== 0) { |
| 950 | return \false; |
| 951 | } |
| 952 | $defaultPort = $schemeA === 'https' ? 443 : 80; |
| 953 | $portA = isset($a['port']) ? (int) $a['port'] : $defaultPort; |
| 954 | $portB = isset($b['port']) ? (int) $b['port'] : $defaultPort; |
| 955 | return $portA === $portB; |
| 956 | } |
| 957 | /** |
| 958 | * Rebuilds a URL from its parse_url() parts with a replacement host, preserving every other |
| 959 | * component. Used on the SSRF-safe path, so the connected URL carries the pinned canonical host. |
| 960 | * |
| 961 | * @param array<string, mixed> $parts |
| 962 | */ |
| 963 | private static function replaceUrlHost(array $parts, string $newHost) : string |
| 964 | { |
| 965 | // @todo PHP 8.1 min (Matomo 6): strpos($newHost, ':') !== false can become str_contains(). |
| 966 | $scheme = isset($parts['scheme']) ? $parts['scheme'] . '://' : ''; |
| 967 | $user = (string) ($parts['user'] ?? ''); |
| 968 | $pass = isset($parts['pass']) ? ':' . $parts['pass'] : ''; |
| 969 | $auth = $user !== '' ? $user . $pass . '@' : ''; |
| 970 | $hostPart = strpos($newHost, ':') !== \false ? '[' . $newHost . ']' : $newHost; |
| 971 | $port = isset($parts['port']) ? ':' . $parts['port'] : ''; |
| 972 | $path = (string) ($parts['path'] ?? ''); |
| 973 | $query = isset($parts['query']) ? '?' . $parts['query'] : ''; |
| 974 | $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : ''; |
| 975 | return $scheme . $auth . $hostPart . $port . $path . $query . $fragment; |
| 976 | } |
| 977 | /** |
| 978 | * Utility function that truncates a string to an arbitrary limit. |
| 979 | * |
| 980 | * @param string $str The string to truncate. |
| 981 | * @param int $limit The maximum length of the truncated string. |
| 982 | * @return string |
| 983 | */ |
| 984 | private static function truncateStr($str, $limit) |
| 985 | { |
| 986 | if (strlen($str) > $limit) { |
| 987 | return substr($str, 0, $limit) . '...'; |
| 988 | } |
| 989 | return $str; |
| 990 | } |
| 991 | /** |
| 992 | * Returns the If-Modified-Since HTTP header if it can be found. If it cannot be |
| 993 | * found, an empty string is returned. |
| 994 | * |
| 995 | * @return string |
| 996 | */ |
| 997 | public static function getModifiedSinceHeader() |
| 998 | { |
| 999 | $modifiedSince = ''; |
| 1000 | if (isset($_SERVER['HTTP_IF_MODIFIED_SINCE'])) { |
| 1001 | $modifiedSince = $_SERVER['HTTP_IF_MODIFIED_SINCE']; |
| 1002 | // strip any trailing data appended to header |
| 1003 | if (\false !== ($semicolonPos = strpos($modifiedSince, ';'))) { |
| 1004 | $modifiedSince = substr($modifiedSince, 0, $semicolonPos); |
| 1005 | } |
| 1006 | } |
| 1007 | return $modifiedSince; |
| 1008 | } |
| 1009 | /** |
| 1010 | * Returns Proxy to use for connecting via HTTP to given URL |
| 1011 | * |
| 1012 | * @param string $url |
| 1013 | * @return array{0: string|null, 1: string|null, 2: string|null, 3: string|null} |
| 1014 | */ |
| 1015 | private static function getProxyConfiguration($url) : array |
| 1016 | { |
| 1017 | $hostname = \Piwik\UrlHelper::getHostFromUrl($url); |
| 1018 | if (\Piwik\Url::isLocalHost($hostname)) { |
| 1019 | return [null, null, null, null]; |
| 1020 | } |
| 1021 | // proxy configuration |
| 1022 | $proxyHost = \Piwik\Config::getInstance()->proxy['host']; |
| 1023 | $proxyPort = \Piwik\Config::getInstance()->proxy['port']; |
| 1024 | $proxyUser = \Piwik\Config::getInstance()->proxy['username']; |
| 1025 | $proxyPassword = \Piwik\Config::getInstance()->proxy['password']; |
| 1026 | $proxyExclude = \Piwik\Config::getInstance()->proxy['exclude']; |
| 1027 | if (!empty($proxyExclude)) { |
| 1028 | $excludes = explode(',', $proxyExclude); |
| 1029 | $excludes = array_map('trim', $excludes); |
| 1030 | $excludes = array_filter($excludes); |
| 1031 | if (in_array($hostname, $excludes)) { |
| 1032 | return [null, null, null, null]; |
| 1033 | } |
| 1034 | } |
| 1035 | return array($proxyHost, $proxyPort, $proxyUser, $proxyPassword); |
| 1036 | } |
| 1037 | /** |
| 1038 | * Checks if HTTPS is available |
| 1039 | * |
| 1040 | * @return bool |
| 1041 | */ |
| 1042 | public static function isUpdatingOverHttps() |
| 1043 | { |
| 1044 | $openSslEnabled = extension_loaded('openssl'); |
| 1045 | $usingMethodSupportingHttps = \Piwik\Http::getTransportMethod() !== 'socket'; |
| 1046 | return $openSslEnabled && $usingMethodSupportingHttps; |
| 1047 | } |
| 1048 | } |
| 1049 |