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