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