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