includes/libs/MultiHttpClient.php

   1 <?php
   2
   3 /**
   4  * Class to handle concurrent HTTP requests
   5  *
   6  * HTTP request maps use the following format:
   7  *   - method   : GET/HEAD/PUT/POST/DELETE
   8  *   - url      : HTTP/HTTPS URL
   9  *   - query    : <query parameter field/value associative array> (uses RFC 3986)
  10  *   - headers  : <header name/value associative array>
  11  *   - body     : source to get the HTTP request body from;
  12  *                this can simply be a string (always), a resource for
  13  *                PUT requests, and a field/value array for POST request;
  14  *                array bodies are encoded as multipart/form-data and strings
  15  *                use application/x-www-form-urlencoded (headers sent automatically)
  16  *   - stream   : resource to stream the HTTP response body to
  17  *
  18  * @author Aaron Schulz
  19  * @since 1.23
  20  */
  21 class MultiHttpClient {
  22         /** @var resource */
  23         protected $multiHandle = null; // curl_multi handle
  24         /** @var string|null SSL certificates path  */
  25         protected $caBundlePath;
  26         /** @var integer */
  27         protected $connTimeout = 10;
  28         /** @var integer */
  29         protected $reqTimeout = 300;
  30         /** @var bool */
  31         protected $usePipelining = false;
  32         /** @var integer */
  33         protected $maxConnsPerHost = 50;
  34
  35         /**
  36          * @param array $options
  37          *   - connTimeout     : default connection timeout
  38          *   - reqTimeout      : default request timeout
  39          *   - usePipelining   : whether to use HTTP pipelining if possible (for all hosts)
  40          *   - maxConnsPerHost : maximum number of concurrent connections (per host)
  41          */
  42         public function __construct( array $options ) {
  43                 if ( isset( $options['caBundlePath'] ) ) {
  44                         $this->caBundlePath = $options['caBundlePath'];
  45                         if ( !file_exists( $this->caBundlePath ) ) {
  46                                 throw new Exception( "Cannot find CA bundle: " . $this->caBundlePath );
  47                         }
  48                 }
  49                 static $opts = array( 'connTimeout', 'reqTimeout', 'usePipelining', 'maxConnsPerHost' );
  50                 foreach ( $opts as $key ) {
  51                         if ( isset( $options[$key] ) ) {
  52                                 $this->$key = $options[$key];
  53                         }
  54                 }
  55         }
  56
  57         /**
  58          * Execute an HTTP(S) request
  59          *
  60          * This method returns a response map of:
  61          *   - code    : HTTP response code or 0 if there was a serious cURL error
  62          *   - reason  : HTTP response reason (empty if there was a serious cURL error)
  63          *   - headers : <header name/value associative array>
  64          *   - body    : HTTP response body or resource (if "stream" was set)
  65          *   - err     : Any cURL error string
  66          * The map also stores integer-indexed copies of these values. This lets callers do:
  67          *      <code>
  68          *              list( $rcode, $rdesc, $rhdrs, $rbody, $rerr ) = $req;
  69          *  </code>
  70          * @param array $req HTTP request array
  71          * @param array $opts
  72          *   - connTimeout    : connection timeout per request
  73          *   - reqTimeout     : post-connection timeout per request
  74          * @return array Response array for request
  75          */
  76         public function run( array $req, array $opts = array() ) {
  77                 $req = $this->runMulti( array( $req ), $opts );
  78                 return $req[0]['response'];
  79         }
  80
  81         /**
  82          * Execute a set of HTTP(S) request concurrently
  83          *
  84          * The maps are returned by this method with the 'response' field set to a map of:
  85          *   - code    : HTTP response code or 0 if there was a serious cURL error
  86          *   - reason  : HTTP response reason (empty if there was a serious cURL error)
  87          *   - headers : <header name/value associative array>
  88          *   - body    : HTTP response body or resource (if "stream" was set)
  89          *   - err     : Any cURL error string
  90          * The map also stores integer-indexed copies of these values. This lets callers do:
  91          *      <code>
  92          *              list( $rcode, $rdesc, $rhdrs, $rbody, $rerr ) = $req;
  93          *  </code>
  94          * All headers in the 'headers' field are normalized to use lower case names.
  95          * This is true for the request headers and the response headers.
  96          *
  97          * @param array $req Map of HTTP request arrays
  98          * @param array $opts
  99          *   - connTimeout     : connection timeout per request
 100          *   - reqTimeout      : post-connection timeout per request
 101          *   - usePipelining   : whether to use HTTP pipelining if possible
 102          *   - maxConnsPerHost : maximum number of concurrent connections (per host)
 103          * @return array $reqs With response array populated for each
 104          */
 105         public function runMulti( array $reqs, array $opts = array() ) {
 106                 $chm = $this->getCurlMulti();
 107
 108                 // Normalize $reqs and add all of the required cURL handles...
 109                 $handles = array();
 110                 foreach ( $reqs as $index => &$req ) {
 111                         $req['response'] = array(
 112                                 'code'     => 0,
 113                                 'reason'   => '',
 114                                 'headers'  => array(),
 115                                 'body'     => '',
 116                                 'error'    => ''
 117                         );
 118                         if ( !isset( $req['method'] ) ) {
 119                                 throw new Exception( "Request has no 'method' field set." );
 120                         } elseif ( !isset( $req['url'] ) ) {
 121                                 throw new Exception( "Request has no 'url' field set." );
 122                         }
 123                         $req['query'] = isset( $req['query'] ) ? $req['query'] : array();
 124                         $headers = array(); // normalized headers
 125                         if ( isset( $req['headers'] ) ) {
 126                                 foreach ( $req['headers'] as $name => $value ) {
 127                                         $headers[strtolower( $name )] = $value;
 128                                 }
 129                         }
 130                         $req['headers'] = $headers;
 131                         if ( !isset( $req['body'] ) ) {
 132                                 $req['body'] = '';
 133                                 $req['headers']['content-length'] = 0;
 134                         }
 135                         $handles[$index] = $this->getCurlHandle( $req, $opts );
 136                         if ( count( $reqs ) > 1 ) {
 137                                 // https://github.com/guzzle/guzzle/issues/349
 138                                 curl_setopt( $handles[$index], CURLOPT_FORBID_REUSE, true );
 139                         }
 140                 }
 141
 142                 $indexes = array_keys( $reqs );
 143                 if ( function_exists( 'curl_multi_setopt' ) ) { // PHP 5.5
 144                         if ( isset( $opts['usePipelining'] ) ) {
 145                                 curl_multi_setopt( $chm, CURLMOPT_PIPELINING, (int)$opts['usePipelining'] );
 146                         }
 147                         if ( isset( $opts['maxConnsPerHost'] ) ) {
 148                                 // Keep these sockets around as they may be needed later in the request
 149                                 curl_multi_setopt( $chm, CURLMOPT_MAXCONNECTS, (int)$opts['maxConnsPerHost'] );
 150                         }
 151                 }
 152
 153                 // @TODO: use a per-host rolling handle window (e.g. CURLMOPT_MAX_HOST_CONNECTIONS)
 154                 $batches = array_chunk( $indexes, $this->maxConnsPerHost );
 155
 156                 foreach ( $batches as $batch ) {
 157                         // Attach all cURL handles for this batch
 158                         foreach ( $batch as $index ) {
 159                                 curl_multi_add_handle( $chm, $handles[$index] );
 160                         }
 161                         // Execute the cURL handles concurrently...
 162                         $active = null; // handles still being processed
 163                         do {
 164                                 // Do any available work...
 165                                 do {
 166                                         $mrc = curl_multi_exec( $chm, $active );
 167                                 } while ( $mrc == CURLM_CALL_MULTI_PERFORM );
 168                                 // Wait (if possible) for available work...
 169                                 if ( $active > 0 && $mrc == CURLM_OK ) {
 170                                         if ( curl_multi_select( $chm, 10 ) == -1 ) {
 171                                                 // PHP bug 63411; http://curl.haxx.se/libcurl/c/curl_multi_fdset.html
 172                                                 usleep( 5000 ); // 5ms
 173                                         }
 174                                 }
 175                         } while ( $active > 0 && $mrc == CURLM_OK );
 176                 }
 177
 178                 // Remove all of the added cURL handles and check for errors...
 179                 foreach ( $reqs as $index => &$req ) {
 180                         $ch = $handles[$index];
 181                         curl_multi_remove_handle( $chm, $ch );
 182                         if ( curl_errno( $ch ) !== 0 ) {
 183                                 $req['error'] = "(curl error: " . curl_errno( $ch ) . ") " . curl_error( $ch );
 184                         }
 185                         // For convenience with the list() operator
 186                         $req['response'][0] = $req['response']['code'];
 187                         $req['response'][1] = $req['response']['reason'];
 188                         $req['response'][2] = $req['response']['headers'];
 189                         $req['response'][3] = $req['response']['body'];
 190                         $req['response'][4] = $req['response']['error'];
 191                         curl_close( $ch );
 192                         // Close any string wrapper file handles
 193                         if ( isset( $req['_closeHandle'] ) ) {
 194                                 fclose( $req['_closeHandle'] );
 195                                 unset( $req['_closeHandle'] );
 196                         }
 197                 }
 198
 199                 // Restore the default settings
 200                 if ( function_exists( 'curl_multi_setopt' ) ) { // PHP 5.5
 201                         curl_multi_setopt( $chm, CURLMOPT_PIPELINING, (int)$this->usePipelining );
 202                         curl_multi_setopt( $chm, CURLMOPT_MAXCONNECTS, (int)$this->maxConnsPerHost );
 203                 }
 204
 205                 return $reqs;
 206         }
 207
 208         /**
 209          * @param array $req HTTP request map
 210          * @param array $opts
 211          *   - connTimeout    : default connection timeout
 212          *   - reqTimeout     : default request timeout
 213          * @return resource
 214          */
 215         protected function getCurlHandle( array &$req, array $opts = array() ) {
 216                 $ch = curl_init();
 217
 218                 curl_setopt( $ch, CURLOPT_CONNECTTIMEOUT,
 219                         isset( $opts['connTimeout'] ) ? $opts['connTimeout'] : $this->connTimeout );
 220                 curl_setopt( $ch, CURLOPT_TIMEOUT,
 221                         isset( $opts['reqTimeout'] ) ? $opts['reqTimeout'] : $this->reqTimeout );
 222                 curl_setopt( $ch, CURLOPT_FOLLOWLOCATION, 1 );
 223                 curl_setopt( $ch, CURLOPT_MAXREDIRS, 4 );
 224                 curl_setopt( $ch, CURLOPT_HEADER, 0 );
 225                 if ( !is_null( $this->caBundlePath ) ) {
 226                         curl_setopt( $ch, CURLOPT_SSL_VERIFYPEER, true );
 227                         curl_setopt( $ch, CURLOPT_CAINFO, $this->caBundlePath );
 228                 }
 229                 curl_setopt( $ch, CURLOPT_RETURNTRANSFER, 1 );
 230
 231                 $url = $req['url'];
 232                 // PHP_QUERY_RFC3986 is PHP 5.4+ only
 233                 $query = str_replace(
 234                         array( '+', '%7E' ),
 235                         array( '%20', '~' ),
 236                         http_build_query( $req['query'], '', '&' )
 237                 );
 238                 if ( $query != '' ) {
 239                         $url .= strpos( $req['url'], '?' ) === false ? "?$query" : "&$query";
 240                 }
 241                 curl_setopt( $ch, CURLOPT_URL, $url );
 242
 243                 curl_setopt( $ch, CURLOPT_CUSTOMREQUEST, $req['method'] );
 244                 if ( $req['method'] === 'HEAD' ) {
 245                         curl_setopt( $ch, CURLOPT_NOBODY, 1 );
 246                 }
 247
 248                 if ( $req['method'] === 'PUT' ) {
 249                         curl_setopt( $ch, CURLOPT_PUT, 1 );
 250                         if ( is_resource( $req['body'] ) ) {
 251                                 curl_setopt( $ch, CURLOPT_INFILE, $req['body'] );
 252                                 if ( isset( $req['headers']['content-length'] ) ) {
 253                                         curl_setopt( $ch, CURLOPT_INFILESIZE, $req['headers']['content-length'] );
 254                                 } elseif ( isset( $req['headers']['transfer-encoding'] ) &&
 255                                         $req['headers']['transfer-encoding'] === 'chunks'
 256                                 ) {
 257                                         curl_setopt( $ch, CURLOPT_UPLOAD, true );
 258                                 } else {
 259                                         throw new Exception( "Missing 'Content-Length' or 'Transfer-Encoding' header." );
 260                                 }
 261                         } elseif ( $req['body'] !== '' ) {
 262                                 $fp = fopen( "php://temp", "wb+" );
 263                                 fwrite( $fp, $req['body'], strlen( $req['body'] ) );
 264                                 rewind( $fp );
 265                                 curl_setopt( $ch, CURLOPT_INFILE, $fp );
 266                                 curl_setopt( $ch, CURLOPT_INFILESIZE, strlen( $req['body'] ) );
 267                                 $req['_closeHandle'] = $fp; // remember to close this later
 268                         } else {
 269                                 curl_setopt( $ch, CURLOPT_INFILESIZE, 0 );
 270                         }
 271                         curl_setopt( $ch, CURLOPT_READFUNCTION,
 272                                 function ( $ch, $fd, $length ) {
 273                                         $data = fread( $fd, $length );
 274                                         $len = strlen( $data );
 275                                         return $data;
 276                                 }
 277                         );
 278                 } elseif ( $req['method'] === 'POST' ) {
 279                         curl_setopt( $ch, CURLOPT_POST, 1 );
 280                         curl_setopt( $ch, CURLOPT_POSTFIELDS, $req['body'] );
 281                 } else {
 282                         if ( is_resource( $req['body'] ) || $req['body'] !== '' ) {
 283                                 throw new Exception( "HTTP body specified for a non PUT/POST request." );
 284                         }
 285                         $req['headers']['content-length'] = 0;
 286                 }
 287
 288                 $headers = array();
 289                 foreach ( $req['headers'] as $name => $value ) {
 290                         if ( strpos( $name, ': ' ) ) {
 291                                 throw new Exception( "Headers cannot have ':' in the name." );
 292                         }
 293                         $headers[] = $name . ': ' . trim( $value );
 294                 }
 295                 curl_setopt( $ch, CURLOPT_HTTPHEADER, $headers );
 296
 297                 curl_setopt( $ch, CURLOPT_HEADERFUNCTION,
 298                         function ( $ch, $header ) use ( &$req ) {
 299                                 $length = strlen( $header );
 300                                 $matches = array();
 301                                 if ( preg_match( "/^(HTTP\/1\.[01]) (\d{3}) (.*)/", $header, $matches ) ) {
 302                                         $req['response']['code'] = (int)$matches[2];
 303                                         $req['response']['reason'] = trim( $matches[3] );
 304                                         return $length;
 305                                 }
 306                                 if ( strpos( $header, ":" ) === false ) {
 307                                         return $length;
 308                                 }
 309                                 list( $name, $value ) = explode( ":", $header, 2 );
 310                                 $req['response']['headers'][strtolower( $name )] = trim( $value );
 311                                 return $length;
 312                         }
 313                 );
 314
 315                 if ( isset( $req['stream'] ) ) {
 316                         // Don't just use CURLOPT_FILE as that might give:
 317                         // curl_setopt(): cannot represent a stream of type Output as a STDIO FILE*
 318                         // The callback here handles both normal files and php://temp handles.
 319                         curl_setopt( $ch, CURLOPT_WRITEFUNCTION,
 320                                 function ( $ch, $data ) use ( &$req ) {
 321                                         return fwrite( $req['stream'], $data );
 322                                 }
 323                         );
 324                 } else {
 325                         curl_setopt( $ch, CURLOPT_WRITEFUNCTION,
 326                                 function ( $ch, $data ) use ( &$req ) {
 327                                         $req['response']['body'] .= $data;
 328                                         return strlen( $data );
 329                                 }
 330                         );
 331                 }
 332
 333                 return $ch;
 334         }
 335
 336         /**
 337          * @return resource
 338          */
 339         protected function getCurlMulti() {
 340                 if ( !$this->multiHandle ) {
 341                         $cmh = curl_multi_init();
 342                         if ( function_exists( 'curl_multi_setopt' ) ) { // PHP 5.5
 343                                 curl_multi_setopt( $cmh, CURLMOPT_PIPELINING, (int)$this->usePipelining );
 344                                 curl_multi_setopt( $cmh, CURLMOPT_MAXCONNECTS, (int)$this->maxConnsPerHost );
 345                         }
 346                         $this->multiHandle = $cmh;
 347                 }
 348                 return $this->multiHandle;
 349         }
 350
 351         function __destruct() {
 352                 if ( $this->multiHandle ) {
 353                         curl_multi_close( $this->multiHandle );
 354                 }
 355         }
 356 }