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