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