added memchached compatibility to background downloads.
[lhc/web/wiklou.git] / 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
15 public static function request( $method, $url, $opts = array() ){
16 $opts['method'] = ( strtoupper( $method ) == 'GET' || strtoupper( $method ) == 'POST' ) ? strtoupper( $method ) : null;
17 $req = new HttpRequest( $url, $opts );
18 $status = $req->doRequest();
19 if( $status->isOK() ){
20 return $status->value;
21 } else {
22 wfDebug( 'http error: ' . $status->getWikiText() );
23 return false;
24 }
25 }
26
27 /**
28 * Simple wrapper for Http::request( 'GET' )
29 */
30 public static function get( $url, $timeout = false ) {
31 $opts = array();
32 if( $timeout )
33 $opts['timeout'] = $timeout;
34 return Http::request( 'GET', $url, $opts );
35 }
36
37 /**
38 * Simple wrapper for Http::request( 'POST' )
39 */
40 public static function post( $url, $opts = array() ) {
41 return Http::request( 'POST', $url, $opts );
42 }
43
44 public static function doDownload( $url, $target_file_path, $dl_mode = self::SYNC_DOWNLOAD, $redirectCount = 0 ){
45 global $wgPhpCli, $wgMaxUploadSize, $wgMaxRedirects;
46 // do a quick check to HEAD to insure the file size is not > $wgMaxUploadSize
47 $headRequest = new HttpRequest( $url, array( 'headers_only' => true ) );
48 $headResponse = $headRequest->doRequest();
49 if( !$headResponse->isOK() ){
50 return $headResponse;
51 }
52 $head = $headResponse->value;
53
54 // check for redirects:
55 if( isset( $head['Location'] ) && strrpos( $head[0], '302' ) !== false ){
56 if( $redirectCount < $wgMaxRedirects ){
57 if( UploadFromUrl::isValidURI( $head['Location'] ) ){
58 return self::doDownload( $head['Location'], $target_file_path, $dl_mode, $redirectCount++ );
59 } else {
60 return Status::newFatal( 'upload-proto-error' );
61 }
62 } else {
63 return Status::newFatal( 'upload-too-many-redirects' );
64 }
65 }
66 // we did not get a 200 ok response:
67 if( strrpos( $head[0], '200 OK' ) === false ){
68 return Status::newFatal( 'upload-http-error', htmlspecialchars( $head[0] ) );
69 }
70
71 $content_length = ( isset( $head['Content-Length'] ) ) ? $head['Content-Length'] : null;
72 if( $content_length ){
73 if( $content_length > $wgMaxUploadSize ){
74 return Status::newFatal( 'requested file length ' . $content_length . ' is greater than $wgMaxUploadSize: ' . $wgMaxUploadSize );
75 }
76 }
77
78 // check if we can find phpCliPath (for doing a background shell request to php to do the download:
79 if( $wgPhpCli && wfShellExecEnabled() && $dl_mode == self::ASYNC_DOWNLOAD ){
80 wfDebug( __METHOD__ . "\ASYNC_DOWNLOAD\n" );
81 //setup session and shell call:
82 return self::initBackgroundDownload( $url, $target_file_path, $content_length );
83 } else {
84 wfDebug( __METHOD__ . "\nSYNC_DOWNLOAD\n" );
85 // SYNC_DOWNLOAD download as much as we can in the time we have to execute
86 $opts['method'] = 'GET';
87 $opts['target_file_path'] = $target_file_path;
88 $req = new HttpRequest( $url, $opts );
89 return $req->doRequest();
90 }
91 }
92
93 /**
94 * a non blocking request (generally an exit point in the application)
95 * should write to a file location and give updates
96 *
97 */
98 private static function initBackgroundDownload( $url, $target_file_path, $content_length = null ){
99 global $wgMaxUploadSize, $IP, $wgPhpCli, $wgServer;
100 $status = Status::newGood();
101
102 // generate a session id with all the details for the download (pid, target_file_path )
103 $upload_session_key = self::getUploadSessionKey();
104 $session_id = session_id();
105
106 // store the url and target path:
107 $_SESSION['wsDownload'][$upload_session_key]['url'] = $url;
108 $_SESSION['wsDownload'][$upload_session_key]['target_file_path'] = $target_file_path;
109 // since we request from the cmd line we lose the original host name pass in the session:
110 $_SESSION['wsDownload'][$upload_session_key]['orgServer'] = $wgServer;
111
112 if( $content_length )
113 $_SESSION['wsDownload'][$upload_session_key]['content_length'] = $content_length;
114
115 // set initial loaded bytes:
116 $_SESSION['wsDownload'][$upload_session_key]['loaded'] = 0;
117
118 // run the background download request:
119 $cmd = $wgPhpCli . ' ' . $IP . "/maintenance/http_session_download.php --sid {$session_id} --usk {$upload_session_key}";
120 $pid = wfShellBackgroundExec( $cmd );
121 // the pid is not of much use since we won't be visiting this same apache any-time soon.
122 if( !$pid )
123 return Status::newFatal( 'could not run background shell exec' );
124
125 // update the status value with the $upload_session_key (for the user to check on the status of the upload)
126 $status->value = $upload_session_key;
127
128 // return good status
129 return $status;
130 }
131
132 static function getUploadSessionKey(){
133 $key = mt_rand( 0, 0x7fffffff );
134 $_SESSION['wsUploadData'][$key] = array();
135 return $key;
136 }
137
138 /**
139 * used to run a session based download. Is initiated via the shell.
140 *
141 * @param $session_id String: the session id to grab download details from
142 * @param $upload_session_key String: the key of the given upload session
143 * (a given client could have started a few http uploads at once)
144 */
145 public static function doSessionIdDownload( $session_id, $upload_session_key ){
146 global $wgUser, $wgEnableWriteAPI, $wgAsyncHTTPTimeout, $wgServer,
147 $wgSessionsInMemcached, $wgSessionHandler, $wgSessionStarted;
148 wfDebug( __METHOD__ . "\n\n doSessionIdDownload :\n\n" );
149 // set session to the provided key:
150 session_id( $session_id );
151 //fire up mediaWiki session system:
152 wfSetupSession();
153
154 // start the session
155 if( session_start() === false ){
156 wfDebug( __METHOD__ . ' could not start session' );
157 }
158 // get all the vars we need from session_id
159 if( !isset( $_SESSION[ 'wsDownload' ][$upload_session_key] ) ){
160 wfDebug( __METHOD__ . ' Error:could not find upload session');
161 exit();
162 }
163 // setup the global user from the session key we just inherited
164 $wgUser = User::newFromSession();
165
166 // grab the session data to setup the request:
167 $sd =& $_SESSION['wsDownload'][$upload_session_key];
168
169 // update the wgServer var ( since cmd line thinks we are localhost when we are really orgServer)
170 if( isset( $sd['orgServer'] ) && $sd['orgServer'] ){
171 $wgServer = $sd['orgServer'];
172 }
173 // close down the session so we can other http queries can get session updates: (if not $wgSessionsInMemcached)
174 if( !$wgSessionsInMemcached )
175 session_write_close();
176
177 $req = new HttpRequest( $sd['url'], array(
178 'target_file_path' => $sd['target_file_path'],
179 'upload_session_key'=> $upload_session_key,
180 'timeout' => $wgAsyncHTTPTimeout,
181 'do_close_session_update' => true
182 ) );
183 // run the actual request .. (this can take some time)
184 wfDebug( __METHOD__ . 'do Session Download :: ' . $sd['url'] . ' tf: ' . $sd['target_file_path'] . "\n\n");
185 $status = $req->doRequest();
186 //wfDebug("done with req status is: ". $status->isOK(). ' '.$status->getWikiText(). "\n");
187
188 // start up the session again:
189 if( session_start() === false ){
190 wfDebug( __METHOD__ . ' ERROR:: Could not start session');
191 }
192 // grab the updated session data pointer
193 $sd =& $_SESSION['wsDownload'][$upload_session_key];
194 // if error update status:
195 if( !$status->isOK() ){
196 $sd['apiUploadResult'] = ApiFormatJson::getJsonEncode(
197 array( 'error' => $status->getWikiText() )
198 );
199 }
200 // if status okay process upload using fauxReq to api:
201 if( $status->isOK() ){
202 // setup the FauxRequest
203 $fauxReqData = $sd['mParams'];
204
205 // Fix boolean parameters
206 foreach( $fauxReqData as $k => $v ) {
207 if( $v === false )
208 unset( $fauxReqData[$k] );
209 }
210
211 $fauxReqData['action'] = 'upload';
212 $fauxReqData['format'] = 'json';
213 $fauxReqData['internalhttpsession'] = $upload_session_key;
214 // evil but no other clean way about it:
215 $faxReq = new FauxRequest( $fauxReqData, true );
216 $processor = new ApiMain( $faxReq, $wgEnableWriteAPI );
217
218 //init the mUpload var for the $processor
219 $processor->execute();
220 $processor->getResult()->cleanUpUTF8();
221 $printer = $processor->createPrinterByName( 'json' );
222 $printer->initPrinter( false );
223 ob_start();
224 $printer->execute();
225 $apiUploadResult = ob_get_clean();
226
227 //wfDebug( __METHOD__ . "\n\n got api result:: $apiUploadResult \n" );
228 // the status updates runner will grab the result form the session:
229 $sd['apiUploadResult'] = $apiUploadResult;
230 }
231 // close the session:
232 session_write_close();
233 }
234
235 /**
236 * Check if the URL can be served by localhost
237 * @param $url string Full url to check
238 * @return bool
239 */
240 public static function isLocalURL( $url ) {
241 global $wgCommandLineMode, $wgConf;
242 if ( $wgCommandLineMode ) {
243 return false;
244 }
245
246 // Extract host part
247 $matches = array();
248 if ( preg_match( '!^http://([\w.-]+)[/:].*$!', $url, $matches ) ) {
249 $host = $matches[1];
250 // Split up dotwise
251 $domainParts = explode( '.', $host );
252 // Check if this domain or any superdomain is listed in $wgConf as a local virtual host
253 $domainParts = array_reverse( $domainParts );
254 for ( $i = 0; $i < count( $domainParts ); $i++ ) {
255 $domainPart = $domainParts[$i];
256 if ( $i == 0 ) {
257 $domain = $domainPart;
258 } else {
259 $domain = $domainPart . '.' . $domain;
260 }
261 if ( $wgConf->isLocalVHost( $domain ) ) {
262 return true;
263 }
264 }
265 }
266 return false;
267 }
268
269 /**
270 * Return a standard user-agent we can use for external requests.
271 */
272 public static function userAgent() {
273 global $wgVersion;
274 return "MediaWiki/$wgVersion";
275 }
276 }
277
278 class HttpRequest {
279 var $target_file_path;
280 var $upload_session_key;
281
282 function __construct( $url, $opt ){
283 global $wgSyncHTTPTimeout;
284 // double check that it's a valid url:
285 $this->url = $url;
286
287 // set the timeout to default sync timeout (unless the timeout option is provided)
288 $this->timeout = ( isset( $opt['timeout'] ) ) ? $opt['timeout'] : $wgSyncHTTPTimeout;
289 $this->method = ( isset( $opt['method'] ) ) ? $opt['method'] : 'GET';
290 $this->target_file_path = ( isset( $opt['target_file_path'] ) ) ? $opt['target_file_path'] : false;
291 $this->upload_session_key = ( isset( $opt['upload_session_key'] ) ) ? $opt['upload_session_key'] : false;
292 $this->headers_only = ( isset( $opt['headers_only'] ) ) ? $opt['headers_only'] : false;
293 $this->do_close_session_update = isset( $opt['do_close_session_update'] );
294 $this->postData = isset( $opt['postdata'] ) ? $opt['postdata'] : '';
295 }
296
297 /**
298 * Get the contents of a file by HTTP
299 * @param $url string Full URL to act on
300 * @param $Opt associative array Optional array of options:
301 * 'method' => 'GET', 'POST' etc.
302 * 'target_file_path' => if curl should output to a target file
303 * 'adapter' => 'curl', 'soket'
304 */
305 public function doRequest() {
306 # Make sure we have a valid url
307 if( !UploadFromUrl::isValidURI( $this->url ) )
308 return Status::newFatal('bad-url');
309
310 # Use curl if available
311 if ( function_exists( 'curl_init' ) ) {
312 return $this->doCurlReq();
313 } else {
314 return $this->doPhpReq();
315 }
316 }
317
318 private function doCurlReq(){
319 global $wgHTTPProxy, $wgTitle;
320
321 $status = Status::newGood();
322 $c = curl_init( $this->url );
323
324 // proxy setup:
325 if ( Http::isLocalURL( $this->url ) ) {
326 curl_setopt( $c, CURLOPT_PROXY, 'localhost:80' );
327 } else if ( $wgHTTPProxy ) {
328 curl_setopt( $c, CURLOPT_PROXY, $wgHTTPProxy );
329 }
330
331 curl_setopt( $c, CURLOPT_TIMEOUT, $this->timeout );
332 curl_setopt( $c, CURLOPT_USERAGENT, Http::userAgent() );
333
334 if ( $this->headers_only ) {
335 curl_setopt( $c, CURLOPT_NOBODY, true );
336 curl_setopt( $c, CURLOPT_HEADER, true );
337 } elseif ( $this->method == 'POST' ) {
338 curl_setopt( $c, CURLOPT_POST, true );
339 curl_setopt( $c, CURLOPT_POSTFIELDS, $this->postData );
340 // Suppress 'Expect: 100-continue' header, as some servers
341 // will reject it with a 417 and Curl won't auto retry
342 // with HTTP 1.0 fallback
343 curl_setopt( $c, CURLOPT_HTTPHEADER, array( 'Expect:' ) );
344 } else {
345 curl_setopt( $c, CURLOPT_CUSTOMREQUEST, $this->method );
346 }
347
348 # Set the referer to $wgTitle, even in command-line mode
349 # This is useful for interwiki transclusion, where the foreign
350 # server wants to know what the referring page is.
351 # $_SERVER['REQUEST_URI'] gives a less reliable indication of the
352 # referring page.
353 if ( is_object( $wgTitle ) ) {
354 curl_setopt( $c, CURLOPT_REFERER, $wgTitle->getFullURL() );
355 }
356
357 // set the write back function (if we are writing to a file)
358 if( $this->target_file_path ){
359 $cwrite = new simpleFileWriter( $this->target_file_path,
360 $this->upload_session_key,
361 $this->do_close_session_update
362 );
363 if( !$cwrite->status->isOK() ){
364 wfDebug( __METHOD__ . "ERROR in setting up simpleFileWriter\n" );
365 $status = $cwrite->status;
366 return $status;
367 }
368 curl_setopt( $c, CURLOPT_WRITEFUNCTION, array( $cwrite, 'callbackWriteBody' ) );
369 }
370
371 // start output grabber:
372 if( !$this->target_file_path )
373 ob_start();
374
375 //run the actual curl_exec:
376 try {
377 if ( false === curl_exec( $c ) ) {
378 $error_txt ='Error sending request: #' . curl_errno( $c ) .' '. curl_error( $c );
379 wfDebug( __METHOD__ . $error_txt . "\n" );
380 $status = Status::newFatal( $error_txt );
381 }
382 } catch ( Exception $e ) {
383 // do something with curl exec error?
384 }
385 // if direct request output the results to the stats value:
386 if( !$this->target_file_path && $status->isOK() ){
387 $status->value = ob_get_contents();
388 ob_end_clean();
389 }
390 // if we wrote to a target file close up or return error
391 if( $this->target_file_path ){
392 $cwrite->close();
393 if( !$cwrite->status->isOK() ){
394 return $cwrite->status;
395 }
396 }
397
398 if ( $this->headers_only ) {
399 $headers = explode( "\n", $status->value );
400 $headerArray = array();
401 foreach ( $headers as $header ) {
402 if ( !strlen( trim( $header ) ) )
403 continue;
404 $headerParts = explode( ':', $header, 2 );
405 if ( count( $headerParts ) == 1 ) {
406 $headerArray[] = trim( $header );
407 } else {
408 list( $key, $val ) = $headerParts;
409 $headerArray[trim( $key )] = trim( $val );
410 }
411 }
412 $status->value = $headerArray;
413 } else {
414 # Don't return the text of error messages, return false on error
415 $retcode = curl_getinfo( $c, CURLINFO_HTTP_CODE );
416 if ( $retcode != 200 ) {
417 wfDebug( __METHOD__ . ": HTTP return code $retcode\n" );
418 $status = Status::newFatal( "HTTP return code $retcode\n" );
419 }
420 # Don't return truncated output
421 $errno = curl_errno( $c );
422 if ( $errno != CURLE_OK ) {
423 $errstr = curl_error( $c );
424 wfDebug( __METHOD__ . ": CURL error code $errno: $errstr\n" );
425 $status = Status::newFatal( " CURL error code $errno: $errstr\n" );
426 }
427 }
428
429 curl_close( $c );
430
431 // return the result obj
432 return $status;
433 }
434
435 public function doPhpReq(){
436 global $wgTitle, $wgHTTPProxy;
437 # Check for php.ini allow_url_fopen
438 if( !ini_get( 'allow_url_fopen' ) ){
439 return Status::newFatal( 'allow_url_fopen needs to be enabled for http copy to work' );
440 }
441
442 // start with good status:
443 $status = Status::newGood();
444
445 if ( $this->headers_only ) {
446 $status->value = get_headers( $this->url, 1 );
447 return $status;
448 }
449
450 // setup the headers
451 $headers = array( "User-Agent: " . Http::userAgent() );
452 if ( is_object( $wgTitle ) ) {
453 $headers[] = "Referer: ". $wgTitle->getFullURL();
454 }
455
456 if( strcasecmp( $this->method, 'post' ) == 0 ) {
457 // Required for HTTP 1.0 POSTs
458 $headers[] = "Content-Length: 0";
459 }
460 $fcontext = stream_context_create ( array(
461 'http' => array(
462 'method' => $this->method,
463 'header' => implode( "\r\n", $headers ),
464 'timeout' => $this->timeout )
465 )
466 );
467 $fh = fopen( $this->url, "r", false, $fcontext);
468
469 // set the write back function (if we are writing to a file)
470 if( $this->target_file_path ){
471 $cwrite = new simpleFileWriter( $this->target_file_path, $this->upload_session_key, $this->do_close_session_update );
472 if( !$cwrite->status->isOK() ){
473 wfDebug( __METHOD__ . "ERROR in setting up simpleFileWriter\n" );
474 $status = $cwrite->status;
475 return $status;
476 }
477
478 // read $fh into the simpleFileWriter (grab in 64K chunks since its likely a ~large~ media file)
479 while ( !feof( $fh ) ) {
480 $contents = fread( $fh, 65536 );
481 $cwrite->callbackWriteBody( $fh, $contents );
482 }
483 $cwrite->close();
484 // check for simpleFileWriter error:
485 if( !$cwrite->status->isOK() ){
486 return $cwrite->status;
487 }
488 } else {
489 // read $fh into status->value
490 $status->value = @stream_get_contents( $fh );
491 }
492 //close the url file wrapper
493 fclose( $fh );
494
495 // check for "false"
496 if( $status->value === false ){
497 $status->error( 'file_get_contents-failed' );
498 }
499 return $status;
500 }
501
502 }
503
504 /**
505 * a simpleFileWriter with session id updates
506 */
507 class simpleFileWriter {
508 var $target_file_path;
509 var $status = null;
510 var $session_id = null;
511 var $session_update_interval = 0; // how often to update the session while downloading
512
513 function simpleFileWriter( $target_file_path, $upload_session_key, $do_close_session_update = false ){
514 $this->target_file_path = $target_file_path;
515 $this->upload_session_key = $upload_session_key;
516 $this->status = Status::newGood();
517 $this->do_close_session_update = $do_close_session_update;
518 // open the file:
519 $this->fp = fopen( $this->target_file_path, 'w' );
520 if( $this->fp === false ){
521 $this->status = Status::newFatal( 'HTTP::could-not-open-file-for-writing' );
522 }
523 // true start time
524 $this->prevTime = time();
525 }
526
527 public function callbackWriteBody( $ch, $data_packet ){
528 global $wgMaxUploadSize, $wgLang;
529
530 // write out the content
531 if( fwrite( $this->fp, $data_packet ) === false ){
532 wfDebug( __METHOD__ ." ::could-not-write-to-file\n" );
533 $this->status = Status::newFatal( 'HTTP::could-not-write-to-file' );
534 return 0;
535 }
536
537 // check file size:
538 clearstatcache();
539 $this->current_fsize = filesize( $this->target_file_path );
540
541 if( $this->current_fsize > $wgMaxUploadSize ){
542 wfDebug( __METHOD__ . " ::http download too large\n" );
543 $this->status = Status::newFatal( 'HTTP::file-has-grown-beyond-upload-limit-killing: downloaded more than ' .
544 $wgLang->formatSize( $wgMaxUploadSize ) . ' ' );
545 return 0;
546 }
547 // if more than session_update_interval second have passed update_session_progress
548 if( $this->do_close_session_update && $this->upload_session_key &&
549 ( ( time() - $this->prevTime ) > $this->session_update_interval ) ) {
550 $this->prevTime = time();
551 $session_status = $this->update_session_progress();
552 if( !$session_status->isOK() ){
553 $this->status = $session_status;
554 wfDebug( __METHOD__ . ' update session failed or was canceled');
555 return 0;
556 }
557 }
558 return strlen( $data_packet );
559 }
560
561 public function update_session_progress(){
562 global $wgSessionsInMemcached;
563 $status = Status::newGood();
564 // start the session (if necessary)
565 if( !$wgSessionsInMemcached ){
566 if( @session_start() === false){
567 wfDebug( __METHOD__ . ' could not start session' );
568 exit( 0 );
569 }
570 }
571 $sd =& $_SESSION['wsDownload'][ $this->upload_session_key ];
572 // check if the user canceled the request:
573 if( isset( $sd['user_cancel'] ) && $sd['user_cancel'] == true ){
574 //@@todo kill the download
575 return Status::newFatal( 'user-canceled-request' );
576 }
577 // update the progress bytes download so far:
578 $sd['loaded'] = $this->current_fsize;
579
580 // close down the session so we can other http queries can get session updates:
581 if( !$wgSessionsInMemcached )
582 session_write_close();
583
584 return $status;
585 }
586
587 public function close(){
588 // do a final session update:
589 if( $this->do_close_session_update ){
590 $this->update_session_progress();
591 }
592 // close up the file handle:
593 if( false === fclose( $this->fp ) ){
594 $this->status = Status::newFatal( 'HTTP::could-not-close-file' );
595 }
596 }
597
598 }