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;
  28         /** @var integer */
  29         protected $reqTimeout;
  30
  31         /**
  32          * @param array $options
  33          */
  34         public function __construct( array $options ) {
  35                 if ( isset( $options['caBundlePath'] ) ) {
  36                         $this->caBundlePath = $options['caBundlePath'];
  37                         if ( !file_exists( $this->caBundlePath ) ) {
  38                                 throw new Exception( "Cannot find CA bundle: " . $this->caBundlePath );
  39                         }
  40                 }
  41                 static $defaults = array( 'connTimeout' => 10, 'reqTimeout' => 300 );
  42                 foreach ( $defaults as $key => $default ) {
  43                         $this->$key = isset( $options[$key] ) ? $options[$key] : $default;
  44                 }
  45         }
  46
  47         /**
  48          * Execute an HTTP(S) request
  49          *
  50          * This method returns a response map of:
  51          *   - code    : HTTP response code or 0 if there was a serious cURL error
  52          *   - reason  : HTTP response reason (empty if there was a serious cURL error)
  53          *   - headers : <header name/value associative array>
  54          *   - body    : HTTP response body or resource (if "stream" was set)
  55          *   - err     : Any cURL error string
  56          * The map also stores integer-indexed copies of these values. This lets callers do:
  57          *      <code>
  58          *              list( $rcode, $rdesc, $rhdrs, $rbody, $rerr ) = $req;
  59          *  </code>
  60          * @param array $req HTTP request array
  61          * @return array Response array for request
  62          */
  63         public function run( array $req ) {
  64                 $req = $this->runMulti( array( $req ) );
  65                 return $req[0]['response'];
  66         }
  67
  68         /**
  69          * Execute a set of HTTP(S) request concurrently
  70          *
  71          * The maps are returned by this method with the 'response' field set to a map of:
  72          *   - code    : HTTP response code or 0 if there was a serious cURL error
  73          *   - reason  : HTTP response reason (empty if there was a serious cURL error)
  74          *   - headers : <header name/value associative array>
  75          *   - body    : HTTP response body or resource (if "stream" was set)
  76          *   - err     : Any cURL error string
  77          * The map also stores integer-indexed copies of these values. This lets callers do:
  78          *      <code>
  79          *              list( $rcode, $rdesc, $rhdrs, $rbody, $rerr ) = $req;
  80          *  </code>
  81          * All headers in the 'headers' field are normalized to use lower case names.
  82          * This is true for the request headers and the response headers.
  83          *
  84          * @param array $req Map of HTTP request arrays
  85          * @return array $reqs With response array populated for each
  86          */
  87         public function runMulti( array $reqs ) {
  88                 $multiHandle = $this->getCurlMulti();
  89
  90                 // Normalize $reqs and add all of the required cURL handles...
  91                 $handles = array();
  92                 foreach ( $reqs as $index => &$req ) {
  93                         $req['response'] = array(
  94                                 'code'     => 0,
  95                                 'reason'   => '',
  96                                 'headers'  => array(),
  97                                 'body'     => '',
  98                                 'error'    => ''
  99                         );
 100                         if ( !isset( $req['method'] ) ) {
 101                                 throw new Exception( "Request has no 'method' field set." );
 102                         } elseif ( !isset( $req['url'] ) ) {
 103                                 throw new Exception( "Request has no 'url' field set." );
 104                         }
 105                         $req['query'] = isset( $req['query'] ) ? $req['query'] : array();
 106                         $headers = array(); // normalized headers
 107                         if ( isset( $req['headers'] ) ) {
 108                                 foreach ( $req['headers'] as $name => $value ) {
 109                                         $headers[strtolower( $name )] = $value;
 110                                 }
 111                         }
 112                         $req['headers'] = $headers;
 113                         if ( !isset( $req['body'] ) ) {
 114                                 $req['body'] = '';
 115                                 $req['headers']['content-length'] = 0;
 116                         }
 117                         $handles[$index] = $this->getCurlHandle( $req );
 118                         if ( count( $reqs ) > 1 ) {
 119                                 // https://github.com/guzzle/guzzle/issues/349
 120                                 curl_setopt( $handles[$index], CURLOPT_FORBID_REUSE, true );
 121                         }
 122                         curl_multi_add_handle( $multiHandle, $handles[$index] );
 123                 }
 124
 125                 // Execute the cURL handles concurrently...
 126                 $active = null; // handles still being processed
 127                 do {
 128                         // Do any available work...
 129                         do {
 130                                 $mrc = curl_multi_exec( $multiHandle, $active );
 131                         } while ( $mrc == CURLM_CALL_MULTI_PERFORM );
 132                         // Wait (if possible) for available work...
 133                         if ( $active > 0 && $mrc == CURLM_OK ) {
 134                                 if ( curl_multi_select( $multiHandle, 10 ) == -1 ) {
 135                                         // PHP bug 63411; http://curl.haxx.se/libcurl/c/curl_multi_fdset.html
 136                                         usleep( 5000 ); // 5ms
 137                                 }
 138                         }
 139                 } while ( $active > 0 && $mrc == CURLM_OK );
 140
 141                 // Remove all of the added cURL handles and check for errors...
 142                 foreach ( $reqs as $index => &$req ) {
 143                         $ch = $handles[$index];
 144                         curl_multi_remove_handle( $multiHandle, $ch );
 145                         if ( curl_errno( $ch ) !== 0 ) {
 146                                 $req['error'] = "(curl error: " . curl_errno( $ch ) . ") " . curl_error( $ch );
 147                         }
 148                         // For convenience with the list() operator
 149                         $req['response'][0] = $req['response']['code'];
 150                         $req['response'][1] = $req['response']['reason'];
 151                         $req['response'][2] = $req['response']['headers'];
 152                         $req['response'][3] = $req['response']['body'];
 153                         $req['response'][4] = $req['response']['error'];
 154                         curl_close( $ch );
 155                         // Close any string wrapper file handles
 156                         if ( isset( $req['_closeHandle'] ) ) {
 157                                 fclose( $req['_closeHandle'] );
 158                                 unset( $req['_closeHandle'] );
 159                         }
 160                 }
 161
 162                 return $reqs;
 163         }
 164
 165         /**
 166          * @param array $req HTTP request map
 167          * @return resource
 168          */
 169         protected function getCurlHandle( array &$req ) {
 170                 $ch = curl_init();
 171
 172                 curl_setopt( $ch, CURLOPT_CONNECTTIMEOUT, $this->connTimeout );
 173                 curl_setopt( $ch, CURLOPT_TIMEOUT, $this->reqTimeout );
 174                 curl_setopt( $ch, CURLOPT_FOLLOWLOCATION, 1 );
 175                 curl_setopt( $ch, CURLOPT_MAXREDIRS, 4 );
 176                 curl_setopt( $ch, CURLOPT_HEADER, 0 );
 177                 if ( !is_null( $this->caBundlePath ) ) {
 178                         curl_setopt( $ch, CURLOPT_SSL_VERIFYPEER, true );
 179                         curl_setopt( $ch, CURLOPT_CAINFO, $this->caBundlePath );
 180                 }
 181                 curl_setopt( $ch, CURLOPT_RETURNTRANSFER, 1 );
 182
 183                 $url = $req['url'];
 184                 // PHP_QUERY_RFC3986 is PHP 5.4+ only
 185                 $query = str_replace(
 186                         array( '+', '%7E' ),
 187                         array( '%20', '~' ),
 188                         http_build_query( $req['query'], '', '&' )
 189                 );
 190                 if ( $query != '' ) {
 191                         $url .= strpos( $req['url'], '?' ) === false ? "?$query" : "&$query";
 192                 }
 193                 curl_setopt( $ch, CURLOPT_URL, $url );
 194
 195                 curl_setopt( $ch, CURLOPT_CUSTOMREQUEST, $req['method'] );
 196                 if ( $req['method'] === 'HEAD' ) {
 197                         curl_setopt( $ch, CURLOPT_NOBODY, 1 );
 198                 }
 199
 200                 if ( $req['method'] === 'PUT' ) {
 201                         curl_setopt( $ch, CURLOPT_PUT, 1 );
 202                         if ( is_resource( $req['body'] ) ) {
 203                                 curl_setopt( $ch, CURLOPT_INFILE, $req['body'] );
 204                                 if ( isset( $req['headers']['content-length'] ) ) {
 205                                         curl_setopt( $ch, CURLOPT_INFILESIZE, $req['headers']['content-length'] );
 206                                 } elseif ( isset( $req['headers']['transfer-encoding'] ) &&
 207                                         $req['headers']['transfer-encoding'] === 'chunks'
 208                                 ) {
 209                                         curl_setopt( $ch, CURLOPT_UPLOAD, true );
 210                                 } else {
 211                                         throw new Exception( "Missing 'Content-Length' or 'Transfer-Encoding' header." );
 212                                 }
 213                         } elseif ( $req['body'] !== '' ) {
 214                                 $fp = fopen( "php://temp", "wb+" );
 215                                 fwrite( $fp, $req['body'], strlen( $req['body'] ) );
 216                                 rewind( $fp );
 217                                 curl_setopt( $ch, CURLOPT_INFILE, $fp );
 218                                 curl_setopt( $ch, CURLOPT_INFILESIZE, strlen( $req['body'] ) );
 219                                 $req['_closeHandle'] = $fp; // remember to close this later
 220                         } else {
 221                                 curl_setopt( $ch, CURLOPT_INFILESIZE, 0 );
 222                         }
 223                         curl_setopt( $ch, CURLOPT_READFUNCTION,
 224                                 function ( $ch, $fd, $length ) {
 225                                         $data = fread( $fd, $length );
 226                                         $len = strlen( $data );
 227                                         return $data;
 228                                 }
 229                         );
 230                 } elseif ( $req['method'] === 'POST' ) {
 231                         curl_setopt( $ch, CURLOPT_POST, 1 );
 232                         curl_setopt( $ch, CURLOPT_POSTFIELDS, $req['body'] );
 233                 } else {
 234                         if ( is_resource( $req['body'] ) || $req['body'] !== '' ) {
 235                                 throw new Exception( "HTTP body specified for a non PUT/POST request." );
 236                         }
 237                         $req['headers']['content-length'] = 0;
 238                 }
 239
 240                 $headers = array();
 241                 foreach ( $req['headers'] as $name => $value ) {
 242                         if ( strpos( $name, ': ' ) ) {
 243                                 throw new Exception( "Headers cannot have ':' in the name." );
 244                         }
 245                         $headers[] = $name . ': ' . trim( $value );
 246                 }
 247                 curl_setopt( $ch, CURLOPT_HTTPHEADER, $headers );
 248
 249                 curl_setopt( $ch, CURLOPT_HEADERFUNCTION,
 250                         function ( $ch, $header ) use ( &$req ) {
 251                                 $length = strlen( $header );
 252                                 $matches = array();
 253                                 if ( preg_match( "/^(HTTP\/1\.[01]) (\d{3}) (.*)/", $header, $matches ) ) {
 254                                         $req['response']['code'] = (int)$matches[2];
 255                                         $req['response']['reason'] = trim( $matches[3] );
 256                                         return $length;
 257                                 }
 258                                 if ( strpos( $header, ":" ) === false ) {
 259                                         return $length;
 260                                 }
 261                                 list( $name, $value ) = explode( ":", $header, 2 );
 262                                 $req['response']['headers'][strtolower( $name )] = trim( $value );
 263                                 return $length;
 264                         }
 265                 );
 266
 267                 if ( isset( $req['stream'] ) ) {
 268                         // Don't just use CURLOPT_FILE as that might give:
 269                         // curl_setopt(): cannot represent a stream of type Output as a STDIO FILE*
 270                         // The callback here handles both normal files and php://temp handles.
 271                         curl_setopt( $ch, CURLOPT_WRITEFUNCTION,
 272                                 function ( $ch, $data ) use ( &$req ) {
 273                                         return fwrite( $req['stream'], $data );
 274                                 }
 275                         );
 276                 } else {
 277                         curl_setopt( $ch, CURLOPT_WRITEFUNCTION,
 278                                 function ( $ch, $data ) use ( &$req ) {
 279                                         $req['response']['body'] .= $data;
 280                                         return strlen( $data );
 281                                 }
 282                         );
 283                 }
 284
 285                 return $ch;
 286         }
 287
 288         /**
 289          * @return resource
 290          */
 291         protected function getCurlMulti() {
 292                 if ( !$this->multiHandle ) {
 293                         $this->multiHandle = curl_multi_init();
 294                 }
 295                 return $this->multiHandle;
 296         }
 297
 298         function __destruct() {
 299                 if ( $this->multiHandle ) {
 300                         curl_multi_close( $this->multiHandle );
 301                 }
 302         }
 303 }