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