includes/libs/MultiHttpClient.php

   1 <?php
   2 /**
   3  * HTTP service client
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  */
  22
  23 use Psr\Log\LoggerAwareInterface;
  24 use Psr\Log\LoggerInterface;
  25 use Psr\Log\NullLogger;
  26
  27 /**
  28  * Class to handle concurrent HTTP requests
  29  *
  30  * HTTP request maps are arrays that use the following format:
  31  *   - method   : GET/HEAD/PUT/POST/DELETE
  32  *   - url      : HTTP/HTTPS URL
  33  *   - query    : <query parameter field/value associative array> (uses RFC 3986)
  34  *   - headers  : <header name/value associative array>
  35  *   - body     : source to get the HTTP request body from;
  36  *                this can simply be a string (always), a resource for
  37  *                PUT requests, and a field/value array for POST request;
  38  *                array bodies are encoded as multipart/form-data and strings
  39  *                use application/x-www-form-urlencoded (headers sent automatically)
  40  *   - stream   : resource to stream the HTTP response body to
  41  *   - proxy    : HTTP proxy to use
  42  *   - flags    : map of boolean flags which supports:
  43  *                  - relayResponseHeaders : write out header via header()
  44  * Request maps can use integer index 0 instead of 'method' and 1 instead of 'url'.
  45  *
  46  * @since 1.23
  47  */
  48 class MultiHttpClient implements LoggerAwareInterface {
  49         /** @var resource */
  50         protected $multiHandle = null; // curl_multi handle
  51         /** @var string|null SSL certificates path */
  52         protected $caBundlePath;
  53         /** @var float */
  54         protected $connTimeout = 10;
  55         /** @var float */
  56         protected $reqTimeout = 300;
  57         /** @var bool */
  58         protected $usePipelining = false;
  59         /** @var int */
  60         protected $maxConnsPerHost = 50;
  61         /** @var string|null proxy */
  62         protected $proxy;
  63         /** @var string */
  64         protected $userAgent = 'wikimedia/multi-http-client v1.0';
  65         /** @var LoggerInterface */
  66         protected $logger;
  67
  68         // In PHP 7 due to https://bugs.php.net/bug.php?id=76480 the request/connect
  69         // timeouts are periodically polled instead of being accurately respected.
  70         // The select timeout is set to the minimum timeout multiplied by this factor.
  71         const TIMEOUT_ACCURACY_FACTOR = 0.1;
  72
  73         /**
  74          * @param array $options
  75          *   - connTimeout     : default connection timeout (seconds)
  76          *   - reqTimeout      : default request timeout (seconds)
  77          *   - proxy           : HTTP proxy to use
  78          *   - usePipelining   : whether to use HTTP pipelining if possible (for all hosts)
  79          *   - maxConnsPerHost : maximum number of concurrent connections (per host)
  80          *   - userAgent       : The User-Agent header value to send
  81          * @throws Exception
  82          */
  83         public function __construct( array $options ) {
  84                 if ( isset( $options['caBundlePath'] ) ) {
  85                         $this->caBundlePath = $options['caBundlePath'];
  86                         if ( !file_exists( $this->caBundlePath ) ) {
  87                                 throw new Exception( "Cannot find CA bundle: " . $this->caBundlePath );
  88                         }
  89                 }
  90                 static $opts = [
  91                         'connTimeout', 'reqTimeout', 'usePipelining', 'maxConnsPerHost',
  92                         'proxy', 'userAgent', 'logger'
  93                 ];
  94                 foreach ( $opts as $key ) {
  95                         if ( isset( $options[$key] ) ) {
  96                                 $this->$key = $options[$key];
  97                         }
  98                 }
  99                 if ( $this->logger === null ) {
 100                         $this->logger = new NullLogger;
 101                 }
 102         }
 103
 104         /**
 105          * Execute an HTTP(S) request
 106          *
 107          * This method returns a response map of:
 108          *   - code    : HTTP response code or 0 if there was a serious cURL error
 109          *   - reason  : HTTP response reason (empty if there was a serious cURL error)
 110          *   - headers : <header name/value associative array>
 111          *   - body    : HTTP response body or resource (if "stream" was set)
 112          *   - error     : Any cURL error string
 113          * The map also stores integer-indexed copies of these values. This lets callers do:
 114          * @code
 115          *              list( $rcode, $rdesc, $rhdrs, $rbody, $rerr ) = $http->run( $req );
 116          * @endcode
 117          * @param array $req HTTP request array
 118          * @param array $opts
 119          *   - connTimeout    : connection timeout per request (seconds)
 120          *   - reqTimeout     : post-connection timeout per request (seconds)
 121          * @return array Response array for request
 122          */
 123         public function run( array $req, array $opts = [] ) {
 124                 return $this->runMulti( [ $req ], $opts )[0]['response'];
 125         }
 126
 127         /**
 128          * Execute a set of HTTP(S) requests concurrently
 129          *
 130          * The maps are returned by this method with the 'response' field set to a map of:
 131          *   - code    : HTTP response code or 0 if there was a serious cURL error
 132          *   - reason  : HTTP response reason (empty if there was a serious cURL error)
 133          *   - headers : <header name/value associative array>
 134          *   - body    : HTTP response body or resource (if "stream" was set)
 135          *   - error   : Any cURL error string
 136          * The map also stores integer-indexed copies of these values. This lets callers do:
 137          * @code
 138          *        list( $rcode, $rdesc, $rhdrs, $rbody, $rerr ) = $req['response'];
 139          * @endcode
 140          * All headers in the 'headers' field are normalized to use lower case names.
 141          * This is true for the request headers and the response headers. Integer-indexed
 142          * method/URL entries will also be changed to use the corresponding string keys.
 143          *
 144          * @param array $reqs Map of HTTP request arrays
 145          * @param array $opts
 146          *   - connTimeout     : connection timeout per request (seconds)
 147          *   - reqTimeout      : post-connection timeout per request (seconds)
 148          *   - usePipelining   : whether to use HTTP pipelining if possible
 149          *   - maxConnsPerHost : maximum number of concurrent connections (per host)
 150          * @return array $reqs With response array populated for each
 151          * @throws Exception
 152          */
 153         public function runMulti( array $reqs, array $opts = [] ) {
 154                 $chm = $this->getCurlMulti();
 155
 156                 $selectTimeout = $this->getSelectTimeout( $opts );
 157
 158                 // Normalize $reqs and add all of the required cURL handles...
 159                 $handles = [];
 160                 foreach ( $reqs as $index => &$req ) {
 161                         $req['response'] = [
 162                                 'code'     => 0,
 163                                 'reason'   => '',
 164                                 'headers'  => [],
 165                                 'body'     => '',
 166                                 'error'    => ''
 167                         ];
 168                         if ( isset( $req[0] ) ) {
 169                                 $req['method'] = $req[0]; // short-form
 170                                 unset( $req[0] );
 171                         }
 172                         if ( isset( $req[1] ) ) {
 173                                 $req['url'] = $req[1]; // short-form
 174                                 unset( $req[1] );
 175                         }
 176                         if ( !isset( $req['method'] ) ) {
 177                                 throw new Exception( "Request has no 'method' field set." );
 178                         } elseif ( !isset( $req['url'] ) ) {
 179                                 throw new Exception( "Request has no 'url' field set." );
 180                         }
 181                         $this->logger->debug( "{$req['method']}: {$req['url']}" );
 182                         $req['query'] = $req['query'] ?? [];
 183                         $headers = []; // normalized headers
 184                         if ( isset( $req['headers'] ) ) {
 185                                 foreach ( $req['headers'] as $name => $value ) {
 186                                         $headers[strtolower( $name )] = $value;
 187                                 }
 188                         }
 189                         $req['headers'] = $headers;
 190                         if ( !isset( $req['body'] ) ) {
 191                                 $req['body'] = '';
 192                                 $req['headers']['content-length'] = 0;
 193                         }
 194                         $req['flags'] = $req['flags'] ?? [];
 195                         $handles[$index] = $this->getCurlHandle( $req, $opts );
 196                         if ( count( $reqs ) > 1 ) {
 197                                 // https://github.com/guzzle/guzzle/issues/349
 198                                 curl_setopt( $handles[$index], CURLOPT_FORBID_REUSE, true );
 199                         }
 200                 }
 201                 unset( $req ); // don't assign over this by accident
 202
 203                 $indexes = array_keys( $reqs );
 204                 if ( isset( $opts['usePipelining'] ) ) {
 205                         curl_multi_setopt( $chm, CURLMOPT_PIPELINING, (int)$opts['usePipelining'] );
 206                 }
 207                 if ( isset( $opts['maxConnsPerHost'] ) ) {
 208                         // Keep these sockets around as they may be needed later in the request
 209                         curl_multi_setopt( $chm, CURLMOPT_MAXCONNECTS, (int)$opts['maxConnsPerHost'] );
 210                 }
 211
 212                 // @TODO: use a per-host rolling handle window (e.g. CURLMOPT_MAX_HOST_CONNECTIONS)
 213                 $batches = array_chunk( $indexes, $this->maxConnsPerHost );
 214                 $infos = [];
 215
 216                 foreach ( $batches as $batch ) {
 217                         // Attach all cURL handles for this batch
 218                         foreach ( $batch as $index ) {
 219                                 curl_multi_add_handle( $chm, $handles[$index] );
 220                         }
 221                         // Execute the cURL handles concurrently...
 222                         $active = null; // handles still being processed
 223                         do {
 224                                 // Do any available work...
 225                                 do {
 226                                         $mrc = curl_multi_exec( $chm, $active );
 227                                         $info = curl_multi_info_read( $chm );
 228                                         if ( $info !== false ) {
 229                                                 $infos[(int)$info['handle']] = $info;
 230                                         }
 231                                 } while ( $mrc == CURLM_CALL_MULTI_PERFORM );
 232                                 // Wait (if possible) for available work...
 233                                 if ( $active > 0 && $mrc == CURLM_OK ) {
 234                                         if ( curl_multi_select( $chm, $selectTimeout ) == -1 ) {
 235                                                 // PHP bug 63411; https://curl.haxx.se/libcurl/c/curl_multi_fdset.html
 236                                                 usleep( 5000 ); // 5ms
 237                                         }
 238                                 }
 239                         } while ( $active > 0 && $mrc == CURLM_OK );
 240                 }
 241
 242                 // Remove all of the added cURL handles and check for errors...
 243                 foreach ( $reqs as $index => &$req ) {
 244                         $ch = $handles[$index];
 245                         curl_multi_remove_handle( $chm, $ch );
 246
 247                         if ( isset( $infos[(int)$ch] ) ) {
 248                                 $info = $infos[(int)$ch];
 249                                 $errno = $info['result'];
 250                                 if ( $errno !== 0 ) {
 251                                         $req['response']['error'] = "(curl error: $errno)";
 252                                         if ( function_exists( 'curl_strerror' ) ) {
 253                                                 $req['response']['error'] .= " " . curl_strerror( $errno );
 254                                         }
 255                                         $this->logger->warning( "Error fetching URL \"{$req['url']}\": " .
 256                                                 $req['response']['error'] );
 257                                 }
 258                         } else {
 259                                 $req['response']['error'] = "(curl error: no status set)";
 260                         }
 261
 262                         // For convenience with the list() operator
 263                         $req['response'][0] = $req['response']['code'];
 264                         $req['response'][1] = $req['response']['reason'];
 265                         $req['response'][2] = $req['response']['headers'];
 266                         $req['response'][3] = $req['response']['body'];
 267                         $req['response'][4] = $req['response']['error'];
 268                         curl_close( $ch );
 269                         // Close any string wrapper file handles
 270                         if ( isset( $req['_closeHandle'] ) ) {
 271                                 fclose( $req['_closeHandle'] );
 272                                 unset( $req['_closeHandle'] );
 273                         }
 274                 }
 275                 unset( $req ); // don't assign over this by accident
 276
 277                 // Restore the default settings
 278                 curl_multi_setopt( $chm, CURLMOPT_PIPELINING, (int)$this->usePipelining );
 279                 curl_multi_setopt( $chm, CURLMOPT_MAXCONNECTS, (int)$this->maxConnsPerHost );
 280
 281                 return $reqs;
 282         }
 283
 284         /**
 285          * @param array &$req HTTP request map
 286          * @param array $opts
 287          *   - connTimeout    : default connection timeout
 288          *   - reqTimeout     : default request timeout
 289          * @return resource
 290          * @throws Exception
 291          */
 292         protected function getCurlHandle( array &$req, array $opts = [] ) {
 293                 $ch = curl_init();
 294
 295                 curl_setopt( $ch, CURLOPT_CONNECTTIMEOUT_MS,
 296                         ( $opts['connTimeout'] ?? $this->connTimeout ) * 1000 );
 297                 curl_setopt( $ch, CURLOPT_PROXY, $req['proxy'] ?? $this->proxy );
 298                 curl_setopt( $ch, CURLOPT_TIMEOUT_MS,
 299                         ( $opts['reqTimeout'] ?? $this->reqTimeout ) * 1000 );
 300                 curl_setopt( $ch, CURLOPT_FOLLOWLOCATION, 1 );
 301                 curl_setopt( $ch, CURLOPT_MAXREDIRS, 4 );
 302                 curl_setopt( $ch, CURLOPT_HEADER, 0 );
 303                 if ( !is_null( $this->caBundlePath ) ) {
 304                         curl_setopt( $ch, CURLOPT_SSL_VERIFYPEER, true );
 305                         curl_setopt( $ch, CURLOPT_CAINFO, $this->caBundlePath );
 306                 }
 307                 curl_setopt( $ch, CURLOPT_RETURNTRANSFER, 1 );
 308
 309                 $url = $req['url'];
 310                 $query = http_build_query( $req['query'], '', '&', PHP_QUERY_RFC3986 );
 311                 if ( $query != '' ) {
 312                         $url .= strpos( $req['url'], '?' ) === false ? "?$query" : "&$query";
 313                 }
 314                 curl_setopt( $ch, CURLOPT_URL, $url );
 315
 316                 curl_setopt( $ch, CURLOPT_CUSTOMREQUEST, $req['method'] );
 317                 if ( $req['method'] === 'HEAD' ) {
 318                         curl_setopt( $ch, CURLOPT_NOBODY, 1 );
 319                 }
 320
 321                 if ( $req['method'] === 'PUT' ) {
 322                         curl_setopt( $ch, CURLOPT_PUT, 1 );
 323                         if ( is_resource( $req['body'] ) ) {
 324                                 curl_setopt( $ch, CURLOPT_INFILE, $req['body'] );
 325                                 if ( isset( $req['headers']['content-length'] ) ) {
 326                                         curl_setopt( $ch, CURLOPT_INFILESIZE, $req['headers']['content-length'] );
 327                                 } elseif ( isset( $req['headers']['transfer-encoding'] ) &&
 328                                         $req['headers']['transfer-encoding'] === 'chunks'
 329                                 ) {
 330                                         curl_setopt( $ch, CURLOPT_UPLOAD, true );
 331                                 } else {
 332                                         throw new Exception( "Missing 'Content-Length' or 'Transfer-Encoding' header." );
 333                                 }
 334                         } elseif ( $req['body'] !== '' ) {
 335                                 $fp = fopen( "php://temp", "wb+" );
 336                                 fwrite( $fp, $req['body'], strlen( $req['body'] ) );
 337                                 rewind( $fp );
 338                                 curl_setopt( $ch, CURLOPT_INFILE, $fp );
 339                                 curl_setopt( $ch, CURLOPT_INFILESIZE, strlen( $req['body'] ) );
 340                                 $req['_closeHandle'] = $fp; // remember to close this later
 341                         } else {
 342                                 curl_setopt( $ch, CURLOPT_INFILESIZE, 0 );
 343                         }
 344                         curl_setopt( $ch, CURLOPT_READFUNCTION,
 345                                 function ( $ch, $fd, $length ) {
 346                                         $data = fread( $fd, $length );
 347                                         $len = strlen( $data );
 348                                         return $data;
 349                                 }
 350                         );
 351                 } elseif ( $req['method'] === 'POST' ) {
 352                         curl_setopt( $ch, CURLOPT_POST, 1 );
 353                         // Don't interpret POST parameters starting with '@' as file uploads, because this
 354                         // makes it impossible to POST plain values starting with '@' (and causes security
 355                         // issues potentially exposing the contents of local files).
 356                         curl_setopt( $ch, CURLOPT_SAFE_UPLOAD, true );
 357                         curl_setopt( $ch, CURLOPT_POSTFIELDS, $req['body'] );
 358                 } else {
 359                         if ( is_resource( $req['body'] ) || $req['body'] !== '' ) {
 360                                 throw new Exception( "HTTP body specified for a non PUT/POST request." );
 361                         }
 362                         $req['headers']['content-length'] = 0;
 363                 }
 364
 365                 if ( !isset( $req['headers']['user-agent'] ) ) {
 366                         $req['headers']['user-agent'] = $this->userAgent;
 367                 }
 368
 369                 $headers = [];
 370                 foreach ( $req['headers'] as $name => $value ) {
 371                         if ( strpos( $name, ': ' ) ) {
 372                                 throw new Exception( "Headers cannot have ':' in the name." );
 373                         }
 374                         $headers[] = $name . ': ' . trim( $value );
 375                 }
 376                 curl_setopt( $ch, CURLOPT_HTTPHEADER, $headers );
 377
 378                 curl_setopt( $ch, CURLOPT_HEADERFUNCTION,
 379                         function ( $ch, $header ) use ( &$req ) {
 380                                 if ( !empty( $req['flags']['relayResponseHeaders'] ) ) {
 381                                         header( $header );
 382                                 }
 383                                 $length = strlen( $header );
 384                                 $matches = [];
 385                                 if ( preg_match( "/^(HTTP\/1\.[01]) (\d{3}) (.*)/", $header, $matches ) ) {
 386                                         $req['response']['code'] = (int)$matches[2];
 387                                         $req['response']['reason'] = trim( $matches[3] );
 388                                         return $length;
 389                                 }
 390                                 if ( strpos( $header, ":" ) === false ) {
 391                                         return $length;
 392                                 }
 393                                 list( $name, $value ) = explode( ":", $header, 2 );
 394                                 $req['response']['headers'][strtolower( $name )] = trim( $value );
 395                                 return $length;
 396                         }
 397                 );
 398
 399                 if ( isset( $req['stream'] ) ) {
 400                         // Don't just use CURLOPT_FILE as that might give:
 401                         // curl_setopt(): cannot represent a stream of type Output as a STDIO FILE*
 402                         // The callback here handles both normal files and php://temp handles.
 403                         curl_setopt( $ch, CURLOPT_WRITEFUNCTION,
 404                                 function ( $ch, $data ) use ( &$req ) {
 405                                         return fwrite( $req['stream'], $data );
 406                                 }
 407                         );
 408                 } else {
 409                         curl_setopt( $ch, CURLOPT_WRITEFUNCTION,
 410                                 function ( $ch, $data ) use ( &$req ) {
 411                                         $req['response']['body'] .= $data;
 412                                         return strlen( $data );
 413                                 }
 414                         );
 415                 }
 416
 417                 return $ch;
 418         }
 419
 420         /**
 421          * Get a suitable select timeout for the given options.
 422          *
 423          * @param array $opts
 424          * @return float
 425          */
 426         private function getSelectTimeout( $opts ) {
 427                 $connTimeout = $opts['connTimeout'] ?? $this->connTimeout;
 428                 $reqTimeout = $opts['reqTimeout'] ?? $this->reqTimeout;
 429                 $timeouts = array_filter( [ $connTimeout, $reqTimeout ] );
 430                 if ( count( $timeouts ) === 0 ) {
 431                         return 1;
 432                 }
 433
 434                 $selectTimeout = min( $timeouts ) * self::TIMEOUT_ACCURACY_FACTOR;
 435                 // Minimum 10us for sanity
 436                 if ( $selectTimeout < 10e-6 ) {
 437                         $selectTimeout = 10e-6;
 438                 }
 439                 return $selectTimeout;
 440         }
 441
 442         /**
 443          * @return resource
 444          * @throws Exception
 445          */
 446         protected function getCurlMulti() {
 447                 if ( !$this->multiHandle ) {
 448                         if ( !function_exists( 'curl_multi_init' ) ) {
 449                                 throw new Exception( "PHP cURL extension missing. " .
 450                                                                          "Check https://www.mediawiki.org/wiki/Manual:CURL" );
 451                         }
 452                         $cmh = curl_multi_init();
 453                         curl_multi_setopt( $cmh, CURLMOPT_PIPELINING, (int)$this->usePipelining );
 454                         curl_multi_setopt( $cmh, CURLMOPT_MAXCONNECTS, (int)$this->maxConnsPerHost );
 455                         $this->multiHandle = $cmh;
 456                 }
 457                 return $this->multiHandle;
 458         }
 459
 460         /**
 461          * Register a logger
 462          *
 463          * @param LoggerInterface $logger
 464          */
 465         public function setLogger( LoggerInterface $logger ) {
 466                 $this->logger = $logger;
 467         }
 468
 469         function __destruct() {
 470                 if ( $this->multiHandle ) {
 471                         curl_multi_close( $this->multiHandle );
 472                 }
 473         }
 474 }