includes/HttpFunctions.php

   1 <?php
   2 /**
   3  * HTTP handling class
   4  * @defgroup HTTP HTTP
   5  * @file
   6  * @ingroup HTTP
   7  */
   8
   9 class Http {
  10         const SYNC_DOWNLOAD = 1;  // syncronys upload (in a single request)
  11         const ASYNC_DOWNLOAD = 2; // asynchronous upload we should spawn out another process and monitor progress if possible)
  12
  13         var $body = '';
  14         public static function request($method, $url, $opts = Array() ){
  15                 $opts['method'] = ( strtoupper( $method ) == 'GET' || strtoupper( $method ) == 'POST' ) ? strtoupper( $method ) : null;
  16                 $req = new HttpRequest( $url, $opts );
  17                 $status = $req->doRequest();
  18                 if( $status->isOK() ){
  19                         return $status->value;
  20                 } else {
  21                         wfDebug( 'http error: ' . $status->getWikiText() );
  22                         return false;
  23                 }
  24         }
  25         /**
  26          * Simple wrapper for Http::request( 'GET' )
  27          */
  28         public static function get( $url, $timeout = false ) {
  29                 $opts = Array();
  30                 if( $timeout )
  31                         $opts['timeout'] = $timeout;
  32                 return Http::request( 'GET', $url, $opts );
  33         }
  34
  35         /**
  36          * Simple wrapper for Http::request( 'POST' )
  37          */
  38         public static function post( $url, $opts = array() ) {
  39                 return Http::request( 'POST', $url, $opts );
  40         }
  41
  42         public static function doDownload( $url, $target_file_path , $dl_mode = self::SYNC_DOWNLOAD , $redirectCount = 0 ){
  43                 global $wgPhpCli, $wgMaxUploadSize, $wgMaxRedirects;
  44                 // do a quick check to HEAD to insure the file size is not > $wgMaxUploadSize
  45                 $headRequest = new HttpRequest( $url, array( 'headers_only' => true ) );
  46                 $headResponse = $headRequest->doRequest();
  47                 if( !$headResponse->isOK() ){
  48                         return $headResponse;
  49                 }
  50                 $head = $headResponse->value;
  51
  52                 // check for redirects:
  53                 if( isset( $head['Location'] ) && strrpos( $head[0], '302' ) !== false ){
  54                         if( $redirectCount < $wgMaxRedirects ){
  55                                 if( UploadFromUrl::isValidURI( $head['Location'] ) ){
  56                                         return self::doDownload( $head['Location'], $target_file_path , $dl_mode, $redirectCount++ );
  57                                 } else {
  58                                         return Status::newFatal( 'upload-proto-error' );
  59                                 }
  60                         } else {
  61                                 return Status::newFatal( 'upload-too-many-redirects' );
  62                         }
  63                 }
  64                 // we did not get a 200 ok response:
  65                 if( strrpos( $head[0], '200 OK' ) === false ){
  66                         return Status::newFatal( 'upload-http-error', htmlspecialchars( $head[0] ) );
  67                 }
  68
  69                 $content_length = ( isset( $head['Content-Length'] ) ) ? $head['Content-Length'] : null;
  70                 if( $content_length ){
  71                         if( $content_length > $wgMaxUploadSize ){
  72                                 return Status::newFatal( 'requested file length ' . $content_length . ' is greater than $wgMaxUploadSize: ' . $wgMaxUploadSize );
  73                         }
  74                 }
  75
  76                 // check if we can find phpCliPath (for doing a background shell request to php to do the download:
  77                 if( $wgPhpCli && wfShellExecEnabled() && $dl_mode == self::ASYNC_DOWNLOAD ){
  78                         wfDebug( __METHOD__ . "\ASYNC_DOWNLOAD\n" );
  79                         // setup session and shell call:
  80                         return self::initBackgroundDownload( $url, $target_file_path, $content_length );
  81                 } else if( $dl_mode == self::SYNC_DOWNLOAD ){
  82                         wfDebug( __METHOD__ . "\nSYNC_DOWNLOAD\n" );
  83                         // SYNC_DOWNLOAD download as much as we can in the time we have to execute
  84                         $opts['method'] = 'GET';
  85                         $opts['target_file_path'] = $target_file_path;
  86                         $req = new HttpRequest( $url, $opts );
  87                         return $req->doRequest();
  88                 }
  89         }
  90
  91         /**
  92          * a non blocking request (generally an exit point in the application)
  93          * should write to a file location and give updates
  94          *
  95          */
  96         private function initBackgroundDownload( $url, $target_file_path, $content_length = null ){
  97                 global $wgMaxUploadSize, $IP, $wgPhpCli, $wgServer;
  98                 $status = Status::newGood();
  99
 100                 // generate a session id with all the details for the download (pid, target_file_path )
 101                 $upload_session_key = self::getUploadSessionKey();
 102                 $session_id = session_id();
 103
 104                 // store the url and target path:
 105                 $_SESSION['wsDownload'][$upload_session_key]['url'] = $url;
 106                 $_SESSION['wsDownload'][$upload_session_key]['target_file_path'] = $target_file_path;
 107                 //since we request from the cmd line we lose the original host name pass in the session:
 108                 $_SESSION['wsDownload'][$upload_session_key]['orgServer'] = $wgServer;
 109
 110                 if( $content_length )
 111                         $_SESSION['wsDownload'][$upload_session_key]['content_length'] = $content_length;
 112
 113                 // set initial loaded bytes:
 114                 $_SESSION['wsDownload'][$upload_session_key]['loaded'] = 0;
 115
 116                 // run the background download request:
 117                 $cmd = $wgPhpCli . ' ' . $IP . "/maintenance/http_session_download.php --sid {$session_id} --usk {$upload_session_key}";
 118                 $pid = wfShellBackgroundExec( $cmd, $retval );
 119                 // the pid is not of much use since we won't be visiting this same apache any-time soon.
 120                 if( !$pid )
 121                         return Status::newFatal( 'could not run background shell exec' );
 122
 123                 // update the status value with the $upload_session_key (for the user to check on the status of the upload)
 124                 $status->value = $upload_session_key;
 125
 126                 // return good status
 127                 return $status;
 128         }
 129
 130         function getUploadSessionKey(){
 131                 $key = mt_rand( 0, 0x7fffffff );
 132                 $_SESSION['wsUploadData'][$key] = array();
 133                 return $key;
 134         }
 135
 136         /**
 137          * used to run a session based download. Is initiated via the shell.
 138          *
 139          * @param $session_id  String: the session id to grab download details from
 140          * @param $upload_session_key String: the key of the given upload session
 141          *  (a given client could have started a few http uploads at once)
 142          */
 143         public static function doSessionIdDownload( $session_id, $upload_session_key ){
 144                 global $wgUser, $wgEnableWriteAPI, $wgAsyncHTTPTimeout, $wgServer;
 145                 wfDebug( __METHOD__ . "\n\ndoSessionIdDownload:\n\n" );
 146                 // set session to the provided key:
 147                 session_id( $session_id );
 148                 // start the session
 149                 if( session_start() === false ){
 150                         wfDebug( __METHOD__ . ' could not start session' );
 151                 }
 152                 //get all the vars we need from session_id
 153                 if(!isset($_SESSION[ 'wsDownload' ][$upload_session_key])){
 154                         wfDebug(  __METHOD__ .' Error:could not find upload session');
 155                         exit();
 156                 }
 157                 // setup the global user from the session key we just inherited
 158                 $wgUser = User::newFromSession();
 159
 160                 // grab the session data to setup the request:
 161                 $sd =& $_SESSION['wsDownload'][$upload_session_key];
 162
 163                 //update the wgServer var ( since cmd line thinks we are localhost when we are really orgServer)
 164                 if(isset($sd['orgServer']) && $sd['orgServer']){
 165                         $wgServer = $sd['orgServer'];
 166                 }
 167                 // close down the session so we can other http queries can get session updates:
 168                 session_write_close();
 169
 170                 $req = new HttpRequest( $sd['url'], array(
 171                         'target_file_path'  => $sd['target_file_path'],
 172                         'upload_session_key'=> $upload_session_key,
 173                         'timeout'           => $wgAsyncHTTPTimeout
 174                 ) );
 175                 // run the actual request .. (this can take some time)
 176                 wfDebug( __METHOD__ . "do Request: " . $sd['url'] . ' tf: ' . $sd['target_file_path'] );
 177                 $status = $req->doRequest();
 178                 //wfDebug("done with req status is: ". $status->isOK(). ' '.$status->getWikiText(). "\n");
 179
 180                 // start up the session again:
 181                 if( session_start() === false ){
 182                         wfDebug( __METHOD__ . ' ERROR:: Could not start session');
 183                 }
 184                 // grab the updated session data pointer
 185                 $sd =& $_SESSION['wsDownload'][$upload_session_key];
 186                 // if error update status:
 187                 if( !$status->isOK() ){
 188                         $sd['apiUploadResult'] = ApiFormatJson::getJsonEncode(
 189                                 array( 'error' => $status->getWikiText() )
 190                         );
 191                 }
 192                 // if status okay process upload using fauxReq to api:
 193                 if( $status->isOK() ){
 194                         // setup the FauxRequest
 195                         $fauxReqData = $sd['mParams'];
 196                         $fauxReqData['action'] = 'upload';
 197                         $fauxReqData['format'] = 'json';
 198                         $fauxReqData['internalhttpsession'] = $upload_session_key;
 199
 200                         // evil but no other clean way about it:
 201                         $faxReq = new FauxRequest( $fauxReqData, true );
 202                         $processor = new ApiMain( $faxReq, $wgEnableWriteAPI );
 203
 204                         //init the mUpload var for the $processor
 205                         $processor->execute();
 206                         $processor->getResult()->cleanUpUTF8();
 207                         $printer = $processor->createPrinterByName( 'json' );
 208                         $printer->initPrinter( false );
 209                         ob_start();
 210                         $printer->execute();
 211                         $apiUploadResult = ob_get_clean();
 212
 213                         wfDebug( __METHOD__ . "\n\n got api result:: $apiUploadResult \n" );
 214                         // the status updates runner will grab the result form the session:
 215                         $sd['apiUploadResult'] = $apiUploadResult;
 216                 }
 217                 // close the session:
 218                 session_write_close();
 219         }
 220
 221         /**
 222          * Check if the URL can be served by localhost
 223          * @param $url string Full url to check
 224          * @return bool
 225          */
 226         public static function isLocalURL( $url ) {
 227                 global $wgCommandLineMode, $wgConf;
 228                 if ( $wgCommandLineMode ) {
 229                         return false;
 230                 }
 231
 232                 // Extract host part
 233                 $matches = array();
 234                 if ( preg_match( '!^http://([\w.-]+)[/:].*$!', $url, $matches ) ) {
 235                         $host = $matches[1];
 236                         // Split up dotwise
 237                         $domainParts = explode( '.', $host );
 238                         // Check if this domain or any superdomain is listed in $wgConf as a local virtual host
 239                         $domainParts = array_reverse( $domainParts );
 240                         for ( $i = 0; $i < count( $domainParts ); $i++ ) {
 241                                 $domainPart = $domainParts[$i];
 242                                 if ( $i == 0 ) {
 243                                         $domain = $domainPart;
 244                                 } else {
 245                                         $domain = $domainPart . '.' . $domain;
 246                                 }
 247                                 if ( $wgConf->isLocalVHost( $domain ) ) {
 248                                         return true;
 249                                 }
 250                         }
 251                 }
 252                 return false;
 253         }
 254
 255         /**
 256          * Return a standard user-agent we can use for external requests.
 257          */
 258         public static function userAgent() {
 259                 global $wgVersion;
 260                 return "MediaWiki/$wgVersion";
 261         }
 262 }
 263 class HttpRequest {
 264         var $target_file_path;
 265         var $upload_session_key;
 266
 267         function __construct( $url, $opt ){
 268                 global $wgSyncHTTPTimeout;
 269                 //double check its a valid url:
 270                 $this->url = $url;
 271
 272                 // set the timeout to default sync timeout (unless the timeout option is provided)
 273                 $this->timeout = ( isset( $opt['timeout'] ) ) ? $opt['timeout'] : $wgSyncHTTPTimeout;
 274                 $this->method = ( isset( $opt['method'] ) ) ? $opt['method'] : 'GET';
 275                 $this->target_file_path = ( isset( $opt['target_file_path'] ) ) ? $opt['target_file_path'] : false;
 276                 $this->upload_session_key = ( isset( $opt['upload_session_key'] ) ) ? $opt['upload_session_key'] : false;
 277                 $this->headers_only = ( isset( $opt['headers_only'] ) ) ? $opt['headers_only'] : false;
 278         }
 279
 280         /**
 281          * Get the contents of a file by HTTP
 282          * @param $url string Full URL to act on
 283          * @param $Opt associative array Optional array of options:
 284          *     'method'           => 'GET', 'POST' etc.
 285          *     'target_file_path' => if curl should output to a target file
 286          *     'adapter'          => 'curl', 'soket'
 287          */
 288         public function doRequest() {
 289
 290                 #make sure we have a valid url
 291                 if( !UploadFromUrl::isValidURI( $this->url ) )
 292                         return Status::newFatal('bad-url');
 293
 294                 # Use curl if available
 295                 if ( function_exists( 'curl_init' ) ) {
 296                         return $this->doCurlReq();
 297                 } else {
 298                         return $this->doPhpReq();
 299                 }
 300         }
 301
 302         private function doCurlReq(){
 303                 global $wgHTTPProxy, $wgTitle;
 304
 305                 $status = Status::newGood();
 306                 $c = curl_init( $this->url );
 307
 308                 // proxy setup:
 309                 if ( Http::isLocalURL( $this->url ) ) {
 310                         curl_setopt( $c, CURLOPT_PROXY, 'localhost:80' );
 311                 } else if ( $wgHTTPProxy ) {
 312                         curl_setopt( $c, CURLOPT_PROXY, $wgHTTPProxy );
 313                 }
 314
 315                 curl_setopt( $c, CURLOPT_TIMEOUT, $this->timeout );
 316                 curl_setopt( $c, CURLOPT_USERAGENT, Http::userAgent() );
 317
 318                 if ( $this->headers_only ) {
 319                         curl_setopt( $c, CURLOPT_NOBODY, true );
 320                         curl_setopt( $c, CURLOPT_HEADER, true );
 321                 } elseif ( $this->method == 'POST' ) {
 322                         curl_setopt( $c, CURLOPT_POST, true );
 323                         curl_setopt( $c, CURLOPT_POSTFIELDS, '' );
 324                 } else {
 325                         curl_setopt( $c, CURLOPT_CUSTOMREQUEST, $this->method );
 326                 }
 327
 328                 # Set the referer to $wgTitle, even in command-line mode
 329                 # This is useful for interwiki transclusion, where the foreign
 330                 # server wants to know what the referring page is.
 331                 # $_SERVER['REQUEST_URI'] gives a less reliable indication of the
 332                 # referring page.
 333                 if ( is_object( $wgTitle ) ) {
 334                         curl_setopt( $c, CURLOPT_REFERER, $wgTitle->getFullURL() );
 335                 }
 336
 337                 // set the write back function (if we are writing to a file)
 338                 if( $this->target_file_path ){
 339                         $cwrite = new simpleFileWriter( $this->target_file_path, $this->upload_session_key );
 340                         if( !$cwrite->status->isOK() ){
 341                                 wfDebug( __METHOD__ . "ERROR in setting up simpleFileWriter\n" );
 342                                 $status = $cwrite->status;
 343                                 return $status;
 344                         }
 345                         curl_setopt( $c, CURLOPT_WRITEFUNCTION, array( $cwrite, 'callbackWriteBody' ) );
 346                 }
 347
 348                 // start output grabber:
 349                 if( !$this->target_file_path )
 350                         ob_start();
 351
 352                 //run the actual curl_exec:
 353                 try {
 354                         if ( false === curl_exec( $c ) ) {
 355                                 $error_txt ='Error sending request: #' . curl_errno( $c ) .' '. curl_error( $c );
 356                                 wfDebug( __METHOD__ . $error_txt . "\n" );
 357                                 $status = Status::newFatal( $error_txt );
 358                         }
 359                 } catch ( Exception $e ) {
 360                         // do something with curl exec error?
 361                 }
 362                 // if direct request output the results to the stats value:
 363                 if( !$this->target_file_path && $status->isOK() ){
 364                         $status->value = ob_get_contents();
 365                         ob_end_clean();
 366                 }
 367                 // if we wrote to a target file close up or return error
 368                 if( $this->target_file_path ){
 369                         $cwrite->close();
 370                         if( !$cwrite->status->isOK() ){
 371                                 return $cwrite->status;
 372                         }
 373                 }
 374
 375                 if ( $this->headers_only ) {
 376                         $headers = explode( "\n", $status->value );
 377                         $headerArray = array();
 378                         foreach ( $headers as $header ) {
 379                                 if ( !strlen( trim( $header ) ) )
 380                                         continue;
 381                                 $headerParts = explode( ':', $header, 2 );
 382                                 if ( count( $headerParts ) == 1 ) {
 383                                         $headerArray[] = trim( $header );
 384                                 } else {
 385                                         list( $key, $val ) = $headerParts;
 386                                         $headerArray[trim( $key )] = trim( $val );
 387                                 }
 388                         }
 389                         $status->value = $headerArray;
 390                 } else {
 391                         # Don't return the text of error messages, return false on error
 392                         $retcode = curl_getinfo( $c, CURLINFO_HTTP_CODE );
 393                         if ( $retcode != 200 ) {
 394                                 wfDebug( __METHOD__ . ": HTTP return code $retcode\n" );
 395                                 $status = Status::newFatal( "HTTP return code $retcode\n" );
 396                         }
 397                         # Don't return truncated output
 398                         $errno = curl_errno( $c );
 399                         if ( $errno != CURLE_OK ) {
 400                                 $errstr = curl_error( $c );
 401                                 wfDebug( __METHOD__ . ": CURL error code $errno: $errstr\n" );
 402                                 $status = Status::newFatal( " CURL error code $errno: $errstr\n" );
 403                         }
 404                 }
 405
 406                 curl_close( $c );
 407
 408                 // return the result obj
 409                 return $status;
 410         }
 411
 412         public function doPhpReq(){
 413                 global $wgTitle, $wgHTTPProxy;
 414
 415                 #check for php.ini allow_url_fopen
 416                 if( !ini_get( 'allow_url_fopen' ) ){
 417                         return Status::newFatal( 'allow_url_fopen needs to be enabled for http copy to work' );
 418                 }
 419
 420                 //start with good status:
 421                 $status = Status::newGood();
 422
 423                 if ( $this->headers_only ) {
 424                         $status->value = get_headers( $this->url, 1 );
 425                         return $status;
 426                 }
 427
 428                 //setup the headers
 429                 $headers = array( "User-Agent: " . Http :: userAgent() );
 430                 if ( is_object( $wgTitle ) ) {
 431                         $headers[] = "Referer: ". $wgTitle->getFullURL();
 432                 }
 433
 434                 if( strcasecmp( $this->method, 'post' ) == 0 ) {
 435                         // Required for HTTP 1.0 POSTs
 436                         $headers[] = "Content-Length: 0";
 437                 }
 438                 $fcontext = stream_context_create ( array(
 439                         'http' => array(
 440                                 'method' => $this->method,
 441                                 'header' => implode( "\r\n", $headers ),
 442                                 'timeout' => $this->timeout )
 443                         )
 444                 );
 445
 446                 $fh = fopen( $this->url, "r", false, $fcontext);
 447
 448                 // set the write back function (if we are writing to a file)
 449                 if( $this->target_file_path ){
 450                         $cwrite = new simpleFileWriter( $this->target_file_path, $this->upload_session_key );
 451                         if( !$cwrite->status->isOK() ){
 452                                 wfDebug( __METHOD__ . "ERROR in setting up simpleFileWriter\n" );
 453                                 $status = $cwrite->status;
 454                                 return $status;
 455                         }
 456                         //read $fh into the simpleFileWriter (grab in 64K chunks since its likely a media file)
 457                         while ( !feof( $fh )) {
 458                                  $contents = fread($fh, 65536);
 459                                  $cwrite->callbackWriteBody($fh, $contents );
 460                         }
 461
 462                         $cwrite->close();
 463                         //check for simpleFileWriter error:
 464                         if( !$cwrite->status->isOK() ){
 465                                 return $cwrite->status;
 466                         }
 467                 } else {
 468                         //read $fh into status->value
 469                         $status->value = @stream_get_contents( $fh );
 470                 }
 471                 //close the url file wrapper
 472                 fclose( $fh );
 473
 474                 //check for "false"
 475                 if( $status->value === false ){
 476                         $status->error( 'file_get_contents-failed' );
 477                 }
 478                 return $status;
 479         }
 480
 481 }
 482
 483 /**
 484  * a simpleFileWriter with session id updates
 485  */
 486 class simpleFileWriter {
 487         var $target_file_path;
 488         var $status = null;
 489         var $session_id = null;
 490         var $session_update_interval = 0; // how often to update the session while downloading
 491
 492         function simpleFileWriter( $target_file_path, $upload_session_key ){
 493                 $this->target_file_path = $target_file_path;
 494                 $this->upload_session_key = $upload_session_key;
 495                 $this->status = Status::newGood();
 496                 // open the file:
 497                 $this->fp = fopen( $this->target_file_path, 'w' );
 498                 if( $this->fp === false ){
 499                         $this->status = Status::newFatal( 'HTTP::could-not-open-file-for-writing' );
 500                 }
 501                 // true start time
 502                 $this->prevTime = time();
 503         }
 504
 505         public function callbackWriteBody($ch, $data_packet){
 506                 global $wgMaxUploadSize;
 507
 508                 // write out the content
 509                 if( fwrite( $this->fp, $data_packet ) === false ){
 510                         wfDebug( __METHOD__ ." ::could-not-write-to-file\n" );
 511                         $this->status = Status::newFatal( 'HTTP::could-not-write-to-file' );
 512                         return 0;
 513                 }
 514
 515                 // check file size:
 516                 clearstatcache();
 517                 $this->current_fsize = filesize( $this->target_file_path );
 518
 519                 if( $this->current_fsize > $wgMaxUploadSize ){
 520                         wfDebug( __METHOD__ . " ::http download too large\n" );
 521                         $this->status = Status::newFatal( 'HTTP::file-has-grown-beyond-upload-limit-killing: downloaded more than ' .
 522                                 Language::formatSize( $wgMaxUploadSize ) . ' ' );
 523                         return 0;
 524                 }
 525
 526                 // if more than session_update_interval second have passed update_session_progress
 527                 if( $this->upload_session_key && ( ( time() - $this->prevTime ) > $this->session_update_interval ) ) {
 528                         $this->prevTime = time();
 529                         $session_status = $this->update_session_progress();
 530                         if( !$session_status->isOK() ){
 531                                 $this->status = $session_status;
 532                                 wfDebug( __METHOD__ . ' update session failed or was canceled');
 533                                 return 0;
 534                         }
 535                 }
 536                 return strlen( $data_packet );
 537         }
 538
 539         public function update_session_progress(){
 540                 $status = Status::newGood();
 541                 // start the session
 542                 if( session_start() === false){
 543                         wfDebug( __METHOD__ . ' could not start session' );
 544                         exit( 0 );
 545                 }
 546                 $sd =& $_SESSION['wsDownload'][$this->upload_session_key];
 547                 // check if the user canceled the request:
 548                 if( isset( $sd['user_cancel'] ) && $sd['user_cancel'] == true ){
 549                         // kill the download
 550                         return Status::newFatal( 'user-canceled-request' );
 551                 }
 552                 // update the progress bytes download so far:
 553                 $sd['loaded'] = $this->current_fsize;
 554                 wfDebug( __METHOD__ . ': set session loaded amount to: ' . $sd['loaded'] . "\n");
 555                 // close down the session so we can other http queries can get session updates:
 556                 session_write_close();
 557                 return $status;
 558         }
 559
 560         public function close(){
 561                 // do a final session update:
 562                 $this->update_session_progress();
 563                 // close up the file handle:
 564                 if( false === fclose( $this->fp ) ){
 565                         $this->status = Status::newFatal( 'HTTP::could-not-close-file' );
 566                 }
 567         }
 568
 569 }