includes/libs/MultiHttpClient.php

   1 <?php
   2 /**
   3  * HTTP service client
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  */
  22
  23 use Psr\Log\LoggerAwareInterface;
  24 use Psr\Log\LoggerInterface;
  25 use Psr\Log\NullLogger;
  26
  27 /**
  28  * Class to handle concurrent HTTP requests
  29  *
  30  * HTTP request maps are arrays that use the following format:
  31  *   - method   : GET/HEAD/PUT/POST/DELETE
  32  *   - url      : HTTP/HTTPS URL
  33  *   - query    : <query parameter field/value associative array> (uses RFC 3986)
  34  *   - headers  : <header name/value associative array>
  35  *   - body     : source to get the HTTP request body from;
  36  *                this can simply be a string (always), a resource for
  37  *                PUT requests, and a field/value array for POST request;
  38  *                array bodies are encoded as multipart/form-data and strings
  39  *                use application/x-www-form-urlencoded (headers sent automatically)
  40  *   - stream   : resource to stream the HTTP response body to
  41  *   - proxy    : HTTP proxy to use
  42  *   - flags    : map of boolean flags which supports:
  43  *                  - relayResponseHeaders : write out header via header()
  44  * Request maps can use integer index 0 instead of 'method' and 1 instead of 'url'.
  45  *
  46  * @since 1.23
  47  */
  48 class MultiHttpClient implements LoggerAwareInterface {
  49         /** @var resource */
  50         protected $multiHandle = null; // curl_multi handle
  51         /** @var string|null SSL certificates path  */
  52         protected $caBundlePath;
  53         /** @var int */
  54         protected $connTimeout = 10;
  55         /** @var int */
  56         protected $reqTimeout = 300;
  57         /** @var bool */
  58         protected $usePipelining = false;
  59         /** @var int */
  60         protected $maxConnsPerHost = 50;
  61         /** @var string|null proxy */
  62         protected $proxy;
  63         /** @var string */
  64         protected $userAgent = 'wikimedia/multi-http-client v1.0';
  65         /** @var LoggerInterface */
  66         protected $logger;
  67
  68         /**
  69          * @param array $options
  70          *   - connTimeout     : default connection timeout (seconds)
  71          *   - reqTimeout      : default request timeout (seconds)
  72          *   - proxy           : HTTP proxy to use
  73          *   - usePipelining   : whether to use HTTP pipelining if possible (for all hosts)
  74          *   - maxConnsPerHost : maximum number of concurrent connections (per host)
  75          *   - userAgent       : The User-Agent header value to send
  76          * @throws Exception
  77          */
  78         public function __construct( array $options ) {
  79                 if ( isset( $options['caBundlePath'] ) ) {
  80                         $this->caBundlePath = $options['caBundlePath'];
  81                         if ( !file_exists( $this->caBundlePath ) ) {
  82                                 throw new Exception( "Cannot find CA bundle: " . $this->caBundlePath );
  83                         }
  84                 }
  85                 static $opts = [
  86                         'connTimeout', 'reqTimeout', 'usePipelining', 'maxConnsPerHost',
  87                         'proxy', 'userAgent', 'logger'
  88                 ];
  89                 foreach ( $opts as $key ) {
  90                         if ( isset( $options[$key] ) ) {
  91                                 $this->$key = $options[$key];
  92                         }
  93                 }
  94                 if ( $this->logger === null ) {
  95                         $this->logger = new NullLogger;
  96                 }
  97         }
  98
  99         /**
 100          * Execute an HTTP(S) request
 101          *
 102          * This method returns a response map of:
 103          *   - code    : HTTP response code or 0 if there was a serious cURL error
 104          *   - reason  : HTTP response reason (empty if there was a serious cURL error)
 105          *   - headers : <header name/value associative array>
 106          *   - body    : HTTP response body or resource (if "stream" was set)
 107          *   - error     : Any cURL error string
 108          * The map also stores integer-indexed copies of these values. This lets callers do:
 109          * @code
 110          *              list( $rcode, $rdesc, $rhdrs, $rbody, $rerr ) = $http->run( $req );
 111          * @endcode
 112          * @param array $req HTTP request array
 113          * @param array $opts
 114          *   - connTimeout    : connection timeout per request (seconds)
 115          *   - reqTimeout     : post-connection timeout per request (seconds)
 116          * @return array Response array for request
 117          */
 118         public function run( array $req, array $opts = [] ) {
 119                 return $this->runMulti( [ $req ], $opts )[0]['response'];
 120         }
 121
 122         /**
 123          * Execute a set of HTTP(S) requests concurrently
 124          *
 125          * The maps are returned by this method with the 'response' field set to a map of:
 126          *   - code    : HTTP response code or 0 if there was a serious cURL error
 127          *   - reason  : HTTP response reason (empty if there was a serious cURL error)
 128          *   - headers : <header name/value associative array>
 129          *   - body    : HTTP response body or resource (if "stream" was set)
 130          *   - error   : Any cURL error string
 131          * The map also stores integer-indexed copies of these values. This lets callers do:
 132          * @code
 133          *        list( $rcode, $rdesc, $rhdrs, $rbody, $rerr ) = $req['response'];
 134          * @endcode
 135          * All headers in the 'headers' field are normalized to use lower case names.
 136          * This is true for the request headers and the response headers. Integer-indexed
 137          * method/URL entries will also be changed to use the corresponding string keys.
 138          *
 139          * @param array $reqs Map of HTTP request arrays
 140          * @param array $opts
 141          *   - connTimeout     : connection timeout per request (seconds)
 142          *   - reqTimeout      : post-connection timeout per request (seconds)
 143          *   - usePipelining   : whether to use HTTP pipelining if possible
 144          *   - maxConnsPerHost : maximum number of concurrent connections (per host)
 145          * @return array $reqs With response array populated for each
 146          * @throws Exception
 147          */
 148         public function runMulti( array $reqs, array $opts = [] ) {
 149                 $chm = $this->getCurlMulti();
 150
 151                 // Normalize $reqs and add all of the required cURL handles...
 152                 $handles = [];
 153                 foreach ( $reqs as $index => &$req ) {
 154                         $req['response'] = [
 155                                 'code'     => 0,
 156                                 'reason'   => '',
 157                                 'headers'  => [],
 158                                 'body'     => '',
 159                                 'error'    => ''
 160                         ];
 161                         if ( isset( $req[0] ) ) {
 162                                 $req['method'] = $req[0]; // short-form
 163                                 unset( $req[0] );
 164                         }
 165                         if ( isset( $req[1] ) ) {
 166                                 $req['url'] = $req[1]; // short-form
 167                                 unset( $req[1] );
 168                         }
 169                         if ( !isset( $req['method'] ) ) {
 170                                 throw new Exception( "Request has no 'method' field set." );
 171                         } elseif ( !isset( $req['url'] ) ) {
 172                                 throw new Exception( "Request has no 'url' field set." );
 173                         }
 174                         $this->logger->debug( "{$req['method']}: {$req['url']}" );
 175                         $req['query'] = isset( $req['query'] ) ? $req['query'] : [];
 176                         $headers = []; // normalized headers
 177                         if ( isset( $req['headers'] ) ) {
 178                                 foreach ( $req['headers'] as $name => $value ) {
 179                                         $headers[strtolower( $name )] = $value;
 180                                 }
 181                         }
 182                         $req['headers'] = $headers;
 183                         if ( !isset( $req['body'] ) ) {
 184                                 $req['body'] = '';
 185                                 $req['headers']['content-length'] = 0;
 186                         }
 187                         $req['flags'] = isset( $req['flags'] ) ? $req['flags'] : [];
 188                         $handles[$index] = $this->getCurlHandle( $req, $opts );
 189                         if ( count( $reqs ) > 1 ) {
 190                                 // https://github.com/guzzle/guzzle/issues/349
 191                                 curl_setopt( $handles[$index], CURLOPT_FORBID_REUSE, true );
 192                         }
 193                 }
 194                 unset( $req ); // don't assign over this by accident
 195
 196                 $indexes = array_keys( $reqs );
 197                 if ( isset( $opts['usePipelining'] ) ) {
 198                         curl_multi_setopt( $chm, CURLMOPT_PIPELINING, (int)$opts['usePipelining'] );
 199                 }
 200                 if ( isset( $opts['maxConnsPerHost'] ) ) {
 201                         // Keep these sockets around as they may be needed later in the request
 202                         curl_multi_setopt( $chm, CURLMOPT_MAXCONNECTS, (int)$opts['maxConnsPerHost'] );
 203                 }
 204
 205                 // @TODO: use a per-host rolling handle window (e.g. CURLMOPT_MAX_HOST_CONNECTIONS)
 206                 $batches = array_chunk( $indexes, $this->maxConnsPerHost );
 207                 $infos = [];
 208
 209                 foreach ( $batches as $batch ) {
 210                         // Attach all cURL handles for this batch
 211                         foreach ( $batch as $index ) {
 212                                 curl_multi_add_handle( $chm, $handles[$index] );
 213                         }
 214                         // Execute the cURL handles concurrently...
 215                         $active = null; // handles still being processed
 216                         do {
 217                                 // Do any available work...
 218                                 do {
 219                                         $mrc = curl_multi_exec( $chm, $active );
 220                                         $info = curl_multi_info_read( $chm );
 221                                         if ( $info !== false ) {
 222                                                 $infos[(int)$info['handle']] = $info;
 223                                         }
 224                                 } while ( $mrc == CURLM_CALL_MULTI_PERFORM );
 225                                 // Wait (if possible) for available work...
 226                                 if ( $active > 0 && $mrc == CURLM_OK ) {
 227                                         if ( curl_multi_select( $chm, 10 ) == -1 ) {
 228                                                 // PHP bug 63411; https://curl.haxx.se/libcurl/c/curl_multi_fdset.html
 229                                                 usleep( 5000 ); // 5ms
 230                                         }
 231                                 }
 232                         } while ( $active > 0 && $mrc == CURLM_OK );
 233                 }
 234
 235                 // Remove all of the added cURL handles and check for errors...
 236                 foreach ( $reqs as $index => &$req ) {
 237                         $ch = $handles[$index];
 238                         curl_multi_remove_handle( $chm, $ch );
 239
 240                         if ( isset( $infos[(int)$ch] ) ) {
 241                                 $info = $infos[(int)$ch];
 242                                 $errno = $info['result'];
 243                                 if ( $errno !== 0 ) {
 244                                         $req['response']['error'] = "(curl error: $errno)";
 245                                         if ( function_exists( 'curl_strerror' ) ) {
 246                                                 $req['response']['error'] .= " " . curl_strerror( $errno );
 247                                         }
 248                                         $this->logger->warning( "Error fetching URL \"{$req['url']}\": " .
 249                                                 $req['response']['error'] );
 250                                 }
 251                         } else {
 252                                 $req['response']['error'] = "(curl error: no status set)";
 253                         }
 254
 255                         // For convenience with the list() operator
 256                         $req['response'][0] = $req['response']['code'];
 257                         $req['response'][1] = $req['response']['reason'];
 258                         $req['response'][2] = $req['response']['headers'];
 259                         $req['response'][3] = $req['response']['body'];
 260                         $req['response'][4] = $req['response']['error'];
 261                         curl_close( $ch );
 262                         // Close any string wrapper file handles
 263                         if ( isset( $req['_closeHandle'] ) ) {
 264                                 fclose( $req['_closeHandle'] );
 265                                 unset( $req['_closeHandle'] );
 266                         }
 267                 }
 268                 unset( $req ); // don't assign over this by accident
 269
 270                 // Restore the default settings
 271                 curl_multi_setopt( $chm, CURLMOPT_PIPELINING, (int)$this->usePipelining );
 272                 curl_multi_setopt( $chm, CURLMOPT_MAXCONNECTS, (int)$this->maxConnsPerHost );
 273
 274                 return $reqs;
 275         }
 276
 277         /**
 278          * @param array &$req HTTP request map
 279          * @param array $opts
 280          *   - connTimeout    : default connection timeout
 281          *   - reqTimeout     : default request timeout
 282          * @return resource
 283          * @throws Exception
 284          */
 285         protected function getCurlHandle( array &$req, array $opts = [] ) {
 286                 $ch = curl_init();
 287
 288                 curl_setopt( $ch, CURLOPT_CONNECTTIMEOUT,
 289                         isset( $opts['connTimeout'] ) ? $opts['connTimeout'] : $this->connTimeout );
 290                 curl_setopt( $ch, CURLOPT_PROXY, isset( $req['proxy'] ) ? $req['proxy'] : $this->proxy );
 291                 curl_setopt( $ch, CURLOPT_TIMEOUT,
 292                         isset( $opts['reqTimeout'] ) ? $opts['reqTimeout'] : $this->reqTimeout );
 293                 curl_setopt( $ch, CURLOPT_FOLLOWLOCATION, 1 );
 294                 curl_setopt( $ch, CURLOPT_MAXREDIRS, 4 );
 295                 curl_setopt( $ch, CURLOPT_HEADER, 0 );
 296                 if ( !is_null( $this->caBundlePath ) ) {
 297                         curl_setopt( $ch, CURLOPT_SSL_VERIFYPEER, true );
 298                         curl_setopt( $ch, CURLOPT_CAINFO, $this->caBundlePath );
 299                 }
 300                 curl_setopt( $ch, CURLOPT_RETURNTRANSFER, 1 );
 301
 302                 $url = $req['url'];
 303                 $query = http_build_query( $req['query'], '', '&', PHP_QUERY_RFC3986 );
 304                 if ( $query != '' ) {
 305                         $url .= strpos( $req['url'], '?' ) === false ? "?$query" : "&$query";
 306                 }
 307                 curl_setopt( $ch, CURLOPT_URL, $url );
 308
 309                 curl_setopt( $ch, CURLOPT_CUSTOMREQUEST, $req['method'] );
 310                 if ( $req['method'] === 'HEAD' ) {
 311                         curl_setopt( $ch, CURLOPT_NOBODY, 1 );
 312                 }
 313
 314                 if ( $req['method'] === 'PUT' ) {
 315                         curl_setopt( $ch, CURLOPT_PUT, 1 );
 316                         if ( is_resource( $req['body'] ) ) {
 317                                 curl_setopt( $ch, CURLOPT_INFILE, $req['body'] );
 318                                 if ( isset( $req['headers']['content-length'] ) ) {
 319                                         curl_setopt( $ch, CURLOPT_INFILESIZE, $req['headers']['content-length'] );
 320                                 } elseif ( isset( $req['headers']['transfer-encoding'] ) &&
 321                                         $req['headers']['transfer-encoding'] === 'chunks'
 322                                 ) {
 323                                         curl_setopt( $ch, CURLOPT_UPLOAD, true );
 324                                 } else {
 325                                         throw new Exception( "Missing 'Content-Length' or 'Transfer-Encoding' header." );
 326                                 }
 327                         } elseif ( $req['body'] !== '' ) {
 328                                 $fp = fopen( "php://temp", "wb+" );
 329                                 fwrite( $fp, $req['body'], strlen( $req['body'] ) );
 330                                 rewind( $fp );
 331                                 curl_setopt( $ch, CURLOPT_INFILE, $fp );
 332                                 curl_setopt( $ch, CURLOPT_INFILESIZE, strlen( $req['body'] ) );
 333                                 $req['_closeHandle'] = $fp; // remember to close this later
 334                         } else {
 335                                 curl_setopt( $ch, CURLOPT_INFILESIZE, 0 );
 336                         }
 337                         curl_setopt( $ch, CURLOPT_READFUNCTION,
 338                                 function ( $ch, $fd, $length ) {
 339                                         $data = fread( $fd, $length );
 340                                         $len = strlen( $data );
 341                                         return $data;
 342                                 }
 343                         );
 344                 } elseif ( $req['method'] === 'POST' ) {
 345                         curl_setopt( $ch, CURLOPT_POST, 1 );
 346                         // Don't interpret POST parameters starting with '@' as file uploads, because this
 347                         // makes it impossible to POST plain values starting with '@' (and causes security
 348                         // issues potentially exposing the contents of local files).
 349                         // The PHP manual says this option was introduced in PHP 5.5 defaults to true in PHP 5.6,
 350                         // but we support lower versions, and the option doesn't exist in HHVM 5.6.99.
 351                         if ( defined( 'CURLOPT_SAFE_UPLOAD' ) ) {
 352                                 curl_setopt( $ch, CURLOPT_SAFE_UPLOAD, true );
 353                         } elseif ( is_array( $req['body'] ) ) {
 354                                 // In PHP 5.2 and later, '@' is interpreted as a file upload if POSTFIELDS
 355                                 // is an array, but not if it's a string. So convert $req['body'] to a string
 356                                 // for safety.
 357                                 $req['body'] = http_build_query( $req['body'] );
 358                         }
 359                         curl_setopt( $ch, CURLOPT_POSTFIELDS, $req['body'] );
 360                 } else {
 361                         if ( is_resource( $req['body'] ) || $req['body'] !== '' ) {
 362                                 throw new Exception( "HTTP body specified for a non PUT/POST request." );
 363                         }
 364                         $req['headers']['content-length'] = 0;
 365                 }
 366
 367                 if ( !isset( $req['headers']['user-agent'] ) ) {
 368                         $req['headers']['user-agent'] = $this->userAgent;
 369                 }
 370
 371                 $headers = [];
 372                 foreach ( $req['headers'] as $name => $value ) {
 373                         if ( strpos( $name, ': ' ) ) {
 374                                 throw new Exception( "Headers cannot have ':' in the name." );
 375                         }
 376                         $headers[] = $name . ': ' . trim( $value );
 377                 }
 378                 curl_setopt( $ch, CURLOPT_HTTPHEADER, $headers );
 379
 380                 curl_setopt( $ch, CURLOPT_HEADERFUNCTION,
 381                         function ( $ch, $header ) use ( &$req ) {
 382                                 if ( !empty( $req['flags']['relayResponseHeaders'] ) ) {
 383                                         header( $header );
 384                                 }
 385                                 $length = strlen( $header );
 386                                 $matches = [];
 387                                 if ( preg_match( "/^(HTTP\/1\.[01]) (\d{3}) (.*)/", $header, $matches ) ) {
 388                                         $req['response']['code'] = (int)$matches[2];
 389                                         $req['response']['reason'] = trim( $matches[3] );
 390                                         return $length;
 391                                 }
 392                                 if ( strpos( $header, ":" ) === false ) {
 393                                         return $length;
 394                                 }
 395                                 list( $name, $value ) = explode( ":", $header, 2 );
 396                                 $req['response']['headers'][strtolower( $name )] = trim( $value );
 397                                 return $length;
 398                         }
 399                 );
 400
 401                 if ( isset( $req['stream'] ) ) {
 402                         // Don't just use CURLOPT_FILE as that might give:
 403                         // curl_setopt(): cannot represent a stream of type Output as a STDIO FILE*
 404                         // The callback here handles both normal files and php://temp handles.
 405                         curl_setopt( $ch, CURLOPT_WRITEFUNCTION,
 406                                 function ( $ch, $data ) use ( &$req ) {
 407                                         return fwrite( $req['stream'], $data );
 408                                 }
 409                         );
 410                 } else {
 411                         curl_setopt( $ch, CURLOPT_WRITEFUNCTION,
 412                                 function ( $ch, $data ) use ( &$req ) {
 413                                         $req['response']['body'] .= $data;
 414                                         return strlen( $data );
 415                                 }
 416                         );
 417                 }
 418
 419                 return $ch;
 420         }
 421
 422         /**
 423          * @return resource
 424          */
 425         protected function getCurlMulti() {
 426                 if ( !$this->multiHandle ) {
 427                         $cmh = curl_multi_init();
 428                         curl_multi_setopt( $cmh, CURLMOPT_PIPELINING, (int)$this->usePipelining );
 429                         curl_multi_setopt( $cmh, CURLMOPT_MAXCONNECTS, (int)$this->maxConnsPerHost );
 430                         $this->multiHandle = $cmh;
 431                 }
 432                 return $this->multiHandle;
 433         }
 434
 435         /**
 436          * Register a logger
 437          *
 438          * @param LoggerInterface $logger
 439          */
 440         public function setLogger( LoggerInterface $logger ) {
 441                 $this->logger = $logger;
 442         }
 443
 444         function __destruct() {
 445                 if ( $this->multiHandle ) {
 446                         curl_multi_close( $this->multiHandle );
 447                 }
 448         }
 449 }