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 /**
  24  * Class to handle concurrent HTTP requests
  25  *
  26  * HTTP request maps are arrays that use the following format:
  27  *   - method   : GET/HEAD/PUT/POST/DELETE
  28  *   - url      : HTTP/HTTPS URL
  29  *   - query    : <query parameter field/value associative array> (uses RFC 3986)
  30  *   - headers  : <header name/value associative array>
  31  *   - body     : source to get the HTTP request body from;
  32  *                this can simply be a string (always), a resource for
  33  *                PUT requests, and a field/value array for POST request;
  34  *                array bodies are encoded as multipart/form-data and strings
  35  *                use application/x-www-form-urlencoded (headers sent automatically)
  36  *   - stream   : resource to stream the HTTP response body to
  37  *   - proxy    : HTTP proxy to use
  38  * Request maps can use integer index 0 instead of 'method' and 1 instead of 'url'.
  39  *
  40  * @author Aaron Schulz
  41  * @since 1.23
  42  */
  43 class MultiHttpClient {
  44         /** @var resource */
  45         protected $multiHandle = null; // curl_multi handle
  46         /** @var string|null SSL certificates path  */
  47         protected $caBundlePath;
  48         /** @var integer */
  49         protected $connTimeout = 10;
  50         /** @var integer */
  51         protected $reqTimeout = 300;
  52         /** @var bool */
  53         protected $usePipelining = false;
  54         /** @var integer */
  55         protected $maxConnsPerHost = 50;
  56         /** @var string|null proxy */
  57         protected $proxy;
  58         /** @var string */
  59         protected $userAgent = 'wikimedia/multi-http-client v1.0';
  60
  61         /**
  62          * @param array $options
  63          *   - connTimeout     : default connection timeout (seconds)
  64          *   - reqTimeout      : default request timeout (seconds)
  65          *   - proxy           : HTTP proxy to use
  66          *   - usePipelining   : whether to use HTTP pipelining if possible (for all hosts)
  67          *   - maxConnsPerHost : maximum number of concurrent connections (per host)
  68          *   - userAgent       : The User-Agent header value to send
  69          * @throws Exception
  70          */
  71         public function __construct( array $options ) {
  72                 if ( isset( $options['caBundlePath'] ) ) {
  73                         $this->caBundlePath = $options['caBundlePath'];
  74                         if ( !file_exists( $this->caBundlePath ) ) {
  75                                 throw new Exception( "Cannot find CA bundle: " . $this->caBundlePath );
  76                         }
  77                 }
  78                 static $opts = [
  79                         'connTimeout', 'reqTimeout', 'usePipelining', 'maxConnsPerHost', 'proxy', 'userAgent'
  80                 ];
  81                 foreach ( $opts as $key ) {
  82                         if ( isset( $options[$key] ) ) {
  83                                 $this->$key = $options[$key];
  84                         }
  85                 }
  86         }
  87
  88         /**
  89          * Execute an HTTP(S) request
  90          *
  91          * This method returns a response map of:
  92          *   - code    : HTTP response code or 0 if there was a serious cURL error
  93          *   - reason  : HTTP response reason (empty if there was a serious cURL error)
  94          *   - headers : <header name/value associative array>
  95          *   - body    : HTTP response body or resource (if "stream" was set)
  96          *   - error     : Any cURL error string
  97          * The map also stores integer-indexed copies of these values. This lets callers do:
  98          * @code
  99          *              list( $rcode, $rdesc, $rhdrs, $rbody, $rerr ) = $http->run( $req );
 100          * @endcode
 101          * @param array $req HTTP request array
 102          * @param array $opts
 103          *   - connTimeout    : connection timeout per request (seconds)
 104          *   - reqTimeout     : post-connection timeout per request (seconds)
 105          * @return array Response array for request
 106          */
 107         final public function run( array $req, array $opts = [] ) {
 108                 return $this->runMulti( [ $req ], $opts )[0]['response'];
 109         }
 110
 111         /**
 112          * Execute a set of HTTP(S) requests concurrently
 113          *
 114          * The maps are returned by this method with the 'response' field set to a map of:
 115          *   - code    : HTTP response code or 0 if there was a serious cURL error
 116          *   - reason  : HTTP response reason (empty if there was a serious cURL error)
 117          *   - headers : <header name/value associative array>
 118          *   - body    : HTTP response body or resource (if "stream" was set)
 119          *   - error   : Any cURL error string
 120          * The map also stores integer-indexed copies of these values. This lets callers do:
 121          * @code
 122          *        list( $rcode, $rdesc, $rhdrs, $rbody, $rerr ) = $req['response'];
 123          * @endcode
 124          * All headers in the 'headers' field are normalized to use lower case names.
 125          * This is true for the request headers and the response headers. Integer-indexed
 126          * method/URL entries will also be changed to use the corresponding string keys.
 127          *
 128          * @param array $reqs Map of HTTP request arrays
 129          * @param array $opts
 130          *   - connTimeout     : connection timeout per request (seconds)
 131          *   - reqTimeout      : post-connection timeout per request (seconds)
 132          *   - usePipelining   : whether to use HTTP pipelining if possible
 133          *   - maxConnsPerHost : maximum number of concurrent connections (per host)
 134          * @return array $reqs With response array populated for each
 135          * @throws Exception
 136          */
 137         public function runMulti( array $reqs, array $opts = [] ) {
 138                 $chm = $this->getCurlMulti();
 139
 140                 // Normalize $reqs and add all of the required cURL handles...
 141                 $handles = [];
 142                 foreach ( $reqs as $index => &$req ) {
 143                         $req['response'] = [
 144                                 'code'     => 0,
 145                                 'reason'   => '',
 146                                 'headers'  => [],
 147                                 'body'     => '',
 148                                 'error'    => ''
 149                         ];
 150                         if ( isset( $req[0] ) ) {
 151                                 $req['method'] = $req[0]; // short-form
 152                                 unset( $req[0] );
 153                         }
 154                         if ( isset( $req[1] ) ) {
 155                                 $req['url'] = $req[1]; // short-form
 156                                 unset( $req[1] );
 157                         }
 158                         if ( !isset( $req['method'] ) ) {
 159                                 throw new Exception( "Request has no 'method' field set." );
 160                         } elseif ( !isset( $req['url'] ) ) {
 161                                 throw new Exception( "Request has no 'url' field set." );
 162                         }
 163                         $req['query'] = isset( $req['query'] ) ? $req['query'] : [];
 164                         $headers = []; // normalized headers
 165                         if ( isset( $req['headers'] ) ) {
 166                                 foreach ( $req['headers'] as $name => $value ) {
 167                                         $headers[strtolower( $name )] = $value;
 168                                 }
 169                         }
 170                         $req['headers'] = $headers;
 171                         if ( !isset( $req['body'] ) ) {
 172                                 $req['body'] = '';
 173                                 $req['headers']['content-length'] = 0;
 174                         }
 175                         $handles[$index] = $this->getCurlHandle( $req, $opts );
 176                         if ( count( $reqs ) > 1 ) {
 177                                 // https://github.com/guzzle/guzzle/issues/349
 178                                 curl_setopt( $handles[$index], CURLOPT_FORBID_REUSE, true );
 179                         }
 180                 }
 181                 unset( $req ); // don't assign over this by accident
 182
 183                 $indexes = array_keys( $reqs );
 184                 if ( function_exists( 'curl_multi_setopt' ) ) { // PHP 5.5
 185                         if ( isset( $opts['usePipelining'] ) ) {
 186                                 curl_multi_setopt( $chm, CURLMOPT_PIPELINING, (int)$opts['usePipelining'] );
 187                         }
 188                         if ( isset( $opts['maxConnsPerHost'] ) ) {
 189                                 // Keep these sockets around as they may be needed later in the request
 190                                 curl_multi_setopt( $chm, CURLMOPT_MAXCONNECTS, (int)$opts['maxConnsPerHost'] );
 191                         }
 192                 }
 193
 194                 // @TODO: use a per-host rolling handle window (e.g. CURLMOPT_MAX_HOST_CONNECTIONS)
 195                 $batches = array_chunk( $indexes, $this->maxConnsPerHost );
 196                 $infos = [];
 197
 198                 foreach ( $batches as $batch ) {
 199                         // Attach all cURL handles for this batch
 200                         foreach ( $batch as $index ) {
 201                                 curl_multi_add_handle( $chm, $handles[$index] );
 202                         }
 203                         // Execute the cURL handles concurrently...
 204                         $active = null; // handles still being processed
 205                         do {
 206                                 // Do any available work...
 207                                 do {
 208                                         $mrc = curl_multi_exec( $chm, $active );
 209                                         $info = curl_multi_info_read( $chm );
 210                                         if ( $info !== false ) {
 211                                                 $infos[(int)$info['handle']] = $info;
 212                                         }
 213                                 } while ( $mrc == CURLM_CALL_MULTI_PERFORM );
 214                                 // Wait (if possible) for available work...
 215                                 if ( $active > 0 && $mrc == CURLM_OK ) {
 216                                         if ( curl_multi_select( $chm, 10 ) == -1 ) {
 217                                                 // PHP bug 63411; http://curl.haxx.se/libcurl/c/curl_multi_fdset.html
 218                                                 usleep( 5000 ); // 5ms
 219                                         }
 220                                 }
 221                         } while ( $active > 0 && $mrc == CURLM_OK );
 222                 }
 223
 224                 // Remove all of the added cURL handles and check for errors...
 225                 foreach ( $reqs as $index => &$req ) {
 226                         $ch = $handles[$index];
 227                         curl_multi_remove_handle( $chm, $ch );
 228
 229                         if ( isset( $infos[(int)$ch] ) ) {
 230                                 $info = $infos[(int)$ch];
 231                                 $errno = $info['result'];
 232                                 if ( $errno !== 0 ) {
 233                                         $req['response']['error'] = "(curl error: $errno)";
 234                                         if ( function_exists( 'curl_strerror' ) ) {
 235                                                 $req['response']['error'] .= " " . curl_strerror( $errno );
 236                                         }
 237                                 }
 238                         } else {
 239                                 $req['response']['error'] = "(curl error: no status set)";
 240                         }
 241
 242                         // For convenience with the list() operator
 243                         $req['response'][0] = $req['response']['code'];
 244                         $req['response'][1] = $req['response']['reason'];
 245                         $req['response'][2] = $req['response']['headers'];
 246                         $req['response'][3] = $req['response']['body'];
 247                         $req['response'][4] = $req['response']['error'];
 248                         curl_close( $ch );
 249                         // Close any string wrapper file handles
 250                         if ( isset( $req['_closeHandle'] ) ) {
 251                                 fclose( $req['_closeHandle'] );
 252                                 unset( $req['_closeHandle'] );
 253                         }
 254                 }
 255                 unset( $req ); // don't assign over this by accident
 256
 257                 // Restore the default settings
 258                 if ( function_exists( 'curl_multi_setopt' ) ) { // PHP 5.5
 259                         curl_multi_setopt( $chm, CURLMOPT_PIPELINING, (int)$this->usePipelining );
 260                         curl_multi_setopt( $chm, CURLMOPT_MAXCONNECTS, (int)$this->maxConnsPerHost );
 261                 }
 262
 263                 return $reqs;
 264         }
 265
 266         /**
 267          * @param array $req HTTP request map
 268          * @param array $opts
 269          *   - connTimeout    : default connection timeout
 270          *   - reqTimeout     : default request timeout
 271          * @return resource
 272          * @throws Exception
 273          */
 274         protected function getCurlHandle( array &$req, array $opts = [] ) {
 275                 $ch = curl_init();
 276
 277                 curl_setopt( $ch, CURLOPT_CONNECTTIMEOUT,
 278                         isset( $opts['connTimeout'] ) ? $opts['connTimeout'] : $this->connTimeout );
 279                 curl_setopt( $ch, CURLOPT_PROXY, isset( $req['proxy'] ) ? $req['proxy'] : $this->proxy );
 280                 curl_setopt( $ch, CURLOPT_TIMEOUT,
 281                         isset( $opts['reqTimeout'] ) ? $opts['reqTimeout'] : $this->reqTimeout );
 282                 curl_setopt( $ch, CURLOPT_FOLLOWLOCATION, 1 );
 283                 curl_setopt( $ch, CURLOPT_MAXREDIRS, 4 );
 284                 curl_setopt( $ch, CURLOPT_HEADER, 0 );
 285                 if ( !is_null( $this->caBundlePath ) ) {
 286                         curl_setopt( $ch, CURLOPT_SSL_VERIFYPEER, true );
 287                         curl_setopt( $ch, CURLOPT_CAINFO, $this->caBundlePath );
 288                 }
 289                 curl_setopt( $ch, CURLOPT_RETURNTRANSFER, 1 );
 290
 291                 $url = $req['url'];
 292                 // PHP_QUERY_RFC3986 is PHP 5.4+ only
 293                 $query = str_replace(
 294                         [ '+', '%7E' ],
 295                         [ '%20', '~' ],
 296                         http_build_query( $req['query'], '', '&' )
 297                 );
 298                 if ( $query != '' ) {
 299                         $url .= strpos( $req['url'], '?' ) === false ? "?$query" : "&$query";
 300                 }
 301                 curl_setopt( $ch, CURLOPT_URL, $url );
 302
 303                 curl_setopt( $ch, CURLOPT_CUSTOMREQUEST, $req['method'] );
 304                 if ( $req['method'] === 'HEAD' ) {
 305                         curl_setopt( $ch, CURLOPT_NOBODY, 1 );
 306                 }
 307
 308                 if ( $req['method'] === 'PUT' ) {
 309                         curl_setopt( $ch, CURLOPT_PUT, 1 );
 310                         if ( is_resource( $req['body'] ) ) {
 311                                 curl_setopt( $ch, CURLOPT_INFILE, $req['body'] );
 312                                 if ( isset( $req['headers']['content-length'] ) ) {
 313                                         curl_setopt( $ch, CURLOPT_INFILESIZE, $req['headers']['content-length'] );
 314                                 } elseif ( isset( $req['headers']['transfer-encoding'] ) &&
 315                                         $req['headers']['transfer-encoding'] === 'chunks'
 316                                 ) {
 317                                         curl_setopt( $ch, CURLOPT_UPLOAD, true );
 318                                 } else {
 319                                         throw new Exception( "Missing 'Content-Length' or 'Transfer-Encoding' header." );
 320                                 }
 321                         } elseif ( $req['body'] !== '' ) {
 322                                 $fp = fopen( "php://temp", "wb+" );
 323                                 fwrite( $fp, $req['body'], strlen( $req['body'] ) );
 324                                 rewind( $fp );
 325                                 curl_setopt( $ch, CURLOPT_INFILE, $fp );
 326                                 curl_setopt( $ch, CURLOPT_INFILESIZE, strlen( $req['body'] ) );
 327                                 $req['_closeHandle'] = $fp; // remember to close this later
 328                         } else {
 329                                 curl_setopt( $ch, CURLOPT_INFILESIZE, 0 );
 330                         }
 331                         curl_setopt( $ch, CURLOPT_READFUNCTION,
 332                                 function ( $ch, $fd, $length ) {
 333                                         $data = fread( $fd, $length );
 334                                         $len = strlen( $data );
 335                                         return $data;
 336                                 }
 337                         );
 338                 } elseif ( $req['method'] === 'POST' ) {
 339                         curl_setopt( $ch, CURLOPT_POST, 1 );
 340                         // Don't interpret POST parameters starting with '@' as file uploads, because this
 341                         // makes it impossible to POST plain values starting with '@' (and causes security
 342                         // issues potentially exposing the contents of local files).
 343                         // The PHP manual says this option was introduced in PHP 5.5 defaults to true in PHP 5.6,
 344                         // but we support lower versions, and the option doesn't exist in HHVM 5.6.99.
 345                         if ( defined( 'CURLOPT_SAFE_UPLOAD' ) ) {
 346                                 curl_setopt( $ch, CURLOPT_SAFE_UPLOAD, true );
 347                         } elseif ( is_array( $req['body'] ) ) {
 348                                 // In PHP 5.2 and later, '@' is interpreted as a file upload if POSTFIELDS
 349                                 // is an array, but not if it's a string. So convert $req['body'] to a string
 350                                 // for safety.
 351                                 $req['body'] = wfArrayToCgi( $req['body'] );
 352                         }
 353                         curl_setopt( $ch, CURLOPT_POSTFIELDS, $req['body'] );
 354                 } else {
 355                         if ( is_resource( $req['body'] ) || $req['body'] !== '' ) {
 356                                 throw new Exception( "HTTP body specified for a non PUT/POST request." );
 357                         }
 358                         $req['headers']['content-length'] = 0;
 359                 }
 360
 361                 if ( !isset( $req['headers']['user-agent'] ) ) {
 362                         $req['headers']['user-agent'] = $this->userAgent;
 363                 }
 364
 365                 $headers = [];
 366                 foreach ( $req['headers'] as $name => $value ) {
 367                         if ( strpos( $name, ': ' ) ) {
 368                                 throw new Exception( "Headers cannot have ':' in the name." );
 369                         }
 370                         $headers[] = $name . ': ' . trim( $value );
 371                 }
 372                 curl_setopt( $ch, CURLOPT_HTTPHEADER, $headers );
 373
 374                 curl_setopt( $ch, CURLOPT_HEADERFUNCTION,
 375                         function ( $ch, $header ) use ( &$req ) {
 376                                 $length = strlen( $header );
 377                                 $matches = [];
 378                                 if ( preg_match( "/^(HTTP\/1\.[01]) (\d{3}) (.*)/", $header, $matches ) ) {
 379                                         $req['response']['code'] = (int)$matches[2];
 380                                         $req['response']['reason'] = trim( $matches[3] );
 381                                         return $length;
 382                                 }
 383                                 if ( strpos( $header, ":" ) === false ) {
 384                                         return $length;
 385                                 }
 386                                 list( $name, $value ) = explode( ":", $header, 2 );
 387                                 $req['response']['headers'][strtolower( $name )] = trim( $value );
 388                                 return $length;
 389                         }
 390                 );
 391
 392                 if ( isset( $req['stream'] ) ) {
 393                         // Don't just use CURLOPT_FILE as that might give:
 394                         // curl_setopt(): cannot represent a stream of type Output as a STDIO FILE*
 395                         // The callback here handles both normal files and php://temp handles.
 396                         curl_setopt( $ch, CURLOPT_WRITEFUNCTION,
 397                                 function ( $ch, $data ) use ( &$req ) {
 398                                         return fwrite( $req['stream'], $data );
 399                                 }
 400                         );
 401                 } else {
 402                         curl_setopt( $ch, CURLOPT_WRITEFUNCTION,
 403                                 function ( $ch, $data ) use ( &$req ) {
 404                                         $req['response']['body'] .= $data;
 405                                         return strlen( $data );
 406                                 }
 407                         );
 408                 }
 409
 410                 return $ch;
 411         }
 412
 413         /**
 414          * @return resource
 415          */
 416         protected function getCurlMulti() {
 417                 if ( !$this->multiHandle ) {
 418                         $cmh = curl_multi_init();
 419                         if ( function_exists( 'curl_multi_setopt' ) ) { // PHP 5.5
 420                                 curl_multi_setopt( $cmh, CURLMOPT_PIPELINING, (int)$this->usePipelining );
 421                                 curl_multi_setopt( $cmh, CURLMOPT_MAXCONNECTS, (int)$this->maxConnsPerHost );
 422                         }
 423                         $this->multiHandle = $cmh;
 424                 }
 425                 return $this->multiHandle;
 426         }
 427
 428         function __destruct() {
 429                 if ( $this->multiHandle ) {
 430                         curl_multi_close( $this->multiHandle );
 431                 }
 432         }
 433 }