1<?php 2/** 3 * Site Export Plugin 4 * 5 * @license GPL 2 (http://www.gnu.org/licenses/gpl.html) 6 * @author i-net software <tools@inetsoftware.de> 7 * @author Gerry Weissbach <gweissbach@inetsoftware.de> 8 */ 9 10// must be run within Dokuwiki 11if(!defined('DOKU_INC')) define('DOKU_INC',realpath(dirname(__FILE__).'/../../../../').'/'); 12if(!defined('DOKU_PLUGIN')) { 13 // Just for sanity 14 require_once(DOKU_INC.'inc/plugin.php'); 15 define('DOKU_PLUGIN',DOKU_INC.'lib/plugins/'); 16} 17 18require_once(DOKU_PLUGIN.'action.php'); 19require_once(DOKU_INC.'/inc/search.php'); 20 21require_once(DOKU_PLUGIN.'siteexport/inc/functions.php'); 22require_once(DOKU_PLUGIN.'siteexport/inc/httpproxy.php'); 23require_once(DOKU_PLUGIN.'siteexport/inc/filewriter.php'); 24require_once(DOKU_PLUGIN.'siteexport/inc/toc.php'); 25require_once(DOKU_PLUGIN.'siteexport/inc/javahelp.php'); 26 27class action_plugin_siteexport_ajax extends DokuWiki_Action_Plugin 28{ 29 /** 30 * New internal variables for better structure 31 */ 32 private $filewriter = null; 33 public $functions = null; 34 35 // List of files that have already been checked 36 private $fileChecked = array(); 37 38 // Namespace of the page to export 39 private $namespace = ''; 40 41 /** 42 * for backward compatability 43 * @see inc/DokuWiki_Plugin#getInfo() 44 */ 45 function getInfo(){ 46 if ( method_exists(parent, 'getInfo')) { 47 $info = parent::getInfo(); 48 } 49 return is_array($info) ? $info : confToHash(dirname(__FILE__).'/../plugin.info.txt'); 50 } 51 52 /** 53 * Register Plugin in DW 54 **/ 55 function register(&$controller) { 56 $controller->register_hook('AJAX_CALL_UNKNOWN', 'BEFORE', $this, 'ajax_siteexport_provider'); 57 $controller->register_hook('ACTION_ACT_PREPROCESS', 'BEFORE', $this, 'siteexport_action'); 58 } 59 60 /** 61 * AJAX Provider - check what is going to be done 62 * @param $event 63 * @param $args 64 */ 65 function ajax_siteexport_provider(&$event, $args) { 66 67 // If this is not a siteexport call, ignore it. 68 if ( !strstr($event->data, '__siteexport' ) ) 69 { 70 return; 71 } 72 73 $this->__init_functions(); 74 75 switch( $event->data ) { 76 case '__siteexport_getsitelist': $this->ajax_siteexport_getsitelist( $event ); break; 77 case '__siteexport_addsite': $this->ajax_siteexport_addsite( $event ); break; 78 case '__siteexport_generateurl': $this->ajax_siteexport_generateurl( $event ); break; 79 } 80 } 81 82 /** 83 * Export from a URL - action 84 * @param $event 85 */ 86 function siteexport_action( &$event ) { 87 global $ID; 88 89 // Check if the 'do' was siteexport 90 if ( $event->data != 'siteexport' ) { return false; } 91 if ( headers_sent() ) { 92 msg("The siteexport function has to be called prior to any header output.", -1); 93 } 94 95 $this->__init_functions(); 96 97 $this->functions->debug->message("========================================", null, 1); 98 $this->functions->debug->message("Starting export from URL call", null, 1); 99 100 $event->preventDefault(); 101 $event->stopPropagation(); 102 103 // Fake security Token if none given 104 if ( empty( $_REQUEST['sectok'] ) ) { 105 $_REQUEST['sectok'] = getSecurityToken(); 106 } 107 108 // The timer will be used to do redirects if needed to prevent timeouts 109 $starttimer = time(); 110 $timerdiff = $this->getConf('max_execution_time'); 111 112 $data = $this->__get_siteexport_list_and_init_tocs($ID, !empty($_REQUEST['startcounter'])); 113 114 if ( $data === false ) { 115 header("HTTP/1.0 401 Unauthorized"); 116 print 'Unauthorized'; 117 exit; 118 } 119 120 $counter = 0; 121 122 if ( count($data) == 0 && !$this->functions->settings->hasValidCacheFile ) { 123 exit(); 124 } 125 126 foreach ( $data as $site ) { 127 128 if ( intval($site['exists']) == 1 || !isset($site['exists']) ) { 129 130 // Skip over the amount of urls that have been exported already 131 if ( empty($_REQUEST['startcounter']) || $counter >= intval($_REQUEST['startcounter']) ) { 132 $status = $this->__siteexport_add_site($site['id']); 133 } 134 } 135 136 $counter ++; 137 if ( time() - $starttimer >= $timerdiff ) { 138 $this->functions->debug->message("Will Redirect", null, 1); 139 $this->handleRuntimeErrorOutput(); 140 $this->functions->startRedirctProcess($counter); 141 } 142 } 143 144 $this->functions->debug->message("Finishing export from URL call", null, 1); 145 $this->functions->debug->message("========================================", null, 1); 146 147 $this->cleanCacheFiles(); 148 149 $URL = ml($this->functions->settings->origZipFile, array('cache' => 'nocache', 'siteexport' => $this->functions->settings->pattern, 'sectok' => getSecurityToken()), true, '&'); 150 $this->functions->debug->message("Redirecting to final file", $URL, 2); 151 152 $this->handleRuntimeErrorOutput(); 153 send_redirect($URL); 154 exit(0); // Should not be reached, but anyways 155 } 156 157 private function handleRuntimeErrorOutput() 158 { 159 if ( !empty($this->functions->debug->runtimeErrors) ) 160 { 161 $this->filewriter->__moveDataToZip($this->functions->debug->runtimeErrors, '_runtime_error/' . time() . '.html'); 162 } 163 } 164 165 public function __init_functions() 166 { 167 $this->functions = new siteexport_functions(); 168 $this->functions->debug->isAJAX = true; 169 $this->filewriter = new siteexport_zipfilewriter($this->functions); 170 171 // Check for PDF Capabilities 172 if ( $this->filewriter->canDoPDF() ) { 173 $this->functions->settings->fileType = 'pdf'; 174 } 175 } 176 177 /** 178 * Prepares the generated URL for direct download access 179 * Also gives back the parameters for this URL 180 * @param $event init event of the ajax request 181 */ 182 function ajax_siteexport_prepareURL_and_POSTData( &$event ) { 183 184 $event->preventDefault(); 185 $event->stopPropagation(); 186 187 // Retrieve Information for download URL 188 $url = $this->functions->prepare_POSTData($_REQUEST); 189 $combined = $this->functions->urlToPathAndParams($url); 190 list($path, $query) = explode('?', $combined, 2); 191 $return = array($url, $combined, $path, $query); 192 193 $this->functions->debug->message("Prepared URL and POST data:", $return, 2); 194 return $return; 195 } 196 197 /** 198 * Executes a Cron Job Action 199 * @param $event 200 */ 201 function ajax_siteexport_cronaction( &$event ) 202 { 203 $cronOverwriteExisting = intval($_REQUEST['cronOverwriteExisting']) == 1; 204 list($url, $combined) = $this->ajax_siteexport_prepareURL_and_POSTData($event); 205 206 if ( !$function =& plugin_load('cron', 'siteexport' ) ) 207 { 208 $this->functions->debug->message("Tried to do an action with siteexport/cron, but the cron plugin is missing.", null, 4); 209 } 210 211 $status = null; 212 switch( $event->data ) { 213 case '__siteexport_savecron': $status = $function->saveCronDataWithParameters($combined, $cronOverwriteExisting); break; 214 case '__siteexport_deletecron': $status = $function->deleteCronDataWithParameters($combined); break; 215 } 216 217 if ( !empty($status) ) 218 { 219 $this->functions->debug->message("Tried to do an action with siteexport/cron, but failed.", $status, 4); 220 } 221 } 222 223 /** 224 * generate direct access URL 225 **/ 226 function ajax_siteexport_generateurl( &$event ) { 227 228 list($url, $combined, $path, $POSTData) = $this->ajax_siteexport_prepareURL_and_POSTData($event); 229 230 // WGET Redirects - this is an option for wget only. 231 // Calculate the maximum redirects that we want to allow. A Problem is that we don't know how long it will take to fetch one page 232 // Therefore we assume it takes about 5s for each page - that gives the freedom to have anough time for redirect. 233 $maxRedirectNumber = ceil( ( count($this->__get_siteexport_list($NS, true)) * 5) / $this->getConf('max_execution_time') ); 234 $maxRedirect = $maxRedirectNumber > 0 ? '--max-redirect=' . ($maxRedirectNumber+3) . ' ' : ''; 235 $maxRedirs = $maxRedirectNumber > 0 ? '--max-redirs ' . ($maxRedirectNumber+3) . ' ' : ''; 236 237 $this->functions->debug->message("Generating Direct Download URL", $url, 2); 238 239 // If there was a Runtime Exception 240 if ( !$this->functions->debug->firstRE() ) { 241 $this->functions->debug->message("There have been errors while generating the download URLs.", null, 4); 242 return; 243 } 244 245 echo $url; 246 echo "\n"; 247 echo 'wget ' . $maxRedirect . '--output-document=' . array_pop(explode(":", ($this->getConf('zipfilename')))) . ' --post-data="' . $POSTData . '" ' . wl(cleanID($path), null, true) . ' --http-user=USER --http-passwd=PASSWD'; 248 echo "\n"; 249 echo 'curl -L ' . $maxRedirs . '-o ' . array_pop(explode(":", ($this->getConf('zipfilename')))) . ' -d "' . $POSTData . '" ' . wl(cleanID($path), null, true) . ' --anyauth --user USER:PASSWD'; 250 echo "\n"; 251 252 $this->functions->debug->message("Checking for Cron parameters: ", $combined, 1); 253 if ( !$functions =& plugin_load('cron', 'siteexport' ) || 254 !$functions->hasCronJobForParameters($combined) ) { 255 echo "false"; 256 } else 257 { 258 echo "true"; 259 } 260 261 return; 262 } 263 264 /** 265 * Get List of sites to be exported for AJAX (wrapper) 266 **/ 267 function ajax_siteexport_getsitelist( &$event ) { 268 269 $event->preventDefault(); 270 $event->stopPropagation(); 271 272 $data = $this->__get_siteexport_list_and_init_tocs($_REQUEST['ns']); 273 274 // Important for reconaisance of the session 275 276 if ( $data === false ) 277 { 278 $this->functions->debug->runtimeException("No data generated. List of Files is 'false'."); 279 return; 280 } 281 282 if ( empty($data) && !$this->functions->settings->hasValidCacheFile ) 283 { 284 $this->functions->debug->runtimeException("Generated list is empty."); 285 return; 286 } 287 288 // If there was a Runtime Exception 289 if ( !$this->functions->debug->firstRE() ) 290 { 291 $this->functions->debug->message("There have been errors while generating site list.", null, 4); 292 return; 293 } 294 295 echo "{$this->functions->settings->pattern}\n"; 296 echo $this->functions->downloadURL() . "\n"; 297 foreach($data as $line ){ 298 echo $line['id'] . "\n"; 299 } 300 301 return; 302 } 303 304 /** 305 * Add a page to the package (for AJAX calls - Wrapper) 306 **/ 307 function ajax_siteexport_addsite( &$event ) { 308 309 $event->preventDefault(); 310 $event->stopPropagation(); 311 312 $this->functions->debug->message("========================================", null, 1); 313 $this->functions->debug->message("Starting export from AJAX call", null, 1); 314 315 $status = $this->__siteexport_add_site($_REQUEST['site']); 316 if ( $status === false ) { return; } 317 318 $this->functions->debug->message("Finishing export from AJAX call", null, 1); 319 $this->functions->debug->message("========================================", null, 1); 320 321 // Print the download zip-File 322 $this->cleanCacheFiles(); 323 324 // If there was a Runtime Exception 325 if ( !$this->functions->debug->firstRE() ) { 326 $this->functions->debug->message("There have been errors during the export.", null, 4); 327 return; 328 } 329 330 print $this->functions->downloadURL(); 331 return; 332 } 333 334 /** 335 * Fetch the list of pages to be exported 336 **/ 337 function __get_siteexport_list($NS, $overrideCache=false) { 338 global $conf; 339 340 $NS = $this->namespace = $this->functions->getNamespaceFromID($NS, $PAGE); 341 342 $depth = $this->getConf('depth'); 343 $query = ''; 344 $doSearch = 'search_allpages'; 345 346 switch( intval($_REQUEST['depthType']) ) { 347 case 0: 348 $query = $this->functions->cleanID(str_replace(":", "/", $NS.':'.$PAGE)); 349 resolve_pageid($NS, $PAGE, $exists); 350 351 if ( $exists ) { 352 $data = array( array( 'id' => $PAGE) ); 353 354 $this->functions->debug->message("Checking for Cache", null, 2); 355 if ( !$overrideCache && $this->filewriter->hasValidCacheFile($_REQUEST, $data) ) 356 { 357 return array(); 358 } 359 360 return $data; 361 } 362 case 1: $depth = 0; 363 break; 364 case 2: $depth = intval($_REQUEST['depth']); 365 break; 366 } 367 368 $opts = array( 'depth' => $depth, 'skipacl' => $this->getConf('skipacl'), 'query' => $query); 369 $data = array(); 370 require_once (DOKU_INC.'inc/search.php'); 371 372 // Check, which TOC to take 373 if ( !$this->functions->settings->useTOCFile ) { 374 search($data, $conf['datadir'], $doSearch, $opts, $this->namespace); 375 } else { 376 $this->functions->debug->message("Using TOC for data", null, 2); 377 378 $doSearch = 'search_pagename'; 379 380 // Create Data of the TOC File should be used instead 381 $opts['query'] = 'toc.txt'; 382 383 $RAWdata = array(); 384 search($RAWdata, $conf['datadir'], $doSearch, $opts, $this->namespace); 385 386 // There may be more than one toc and all of them have to be merged. 387 $data = array(); 388 foreach( $RAWdata as $entry ) 389 { 390 $tmpData = p_get_metadata($entry['id'], 'sitetoc siteexportTOC', true); 391 392 if ( is_array($tmpData) ) 393 { 394 $data = array_merge($data, $tmpData); 395 } 396 } 397 } 398 399 $this->functions->debug->message("Checking for Cache", null, 2); 400 if ( !$overrideCache && $this->filewriter->hasValidCacheFile($_REQUEST, $data) ) 401 { 402 return array(); 403 } 404 405 $this->functions->debug->message("Exporting the following sites: ", $data, 2); 406 return $data; 407 } 408 409 function __get_siteexport_list_and_init_tocs($NS, $isRedirected=false ) { 410 411 // Clean up if not redirected 412 if ( !$isRedirected && !$this->__removeOldZip() ) { 413 $this->functions->debug->runtimeException("Can't remove old files."); 414 return false; 415 } 416 417 $data = $this->__get_siteexport_list($NS, $isRedirected); 418 if ( $isRedirected || empty($data) ) 419 { 420 // if we have been redirected, simply return the data 421 return $data; 422 } 423 424 // Create Eclipse Documentation Pages - TOC.xml, Context.xml 425 if ( !empty($_REQUEST['absolutePath']) ) $this->namespace = ""; 426// $this->__removeOldZip( $this->functions->settings->eclipseZipFile ); 427 428 if ( !empty($_REQUEST['eclipseDocZip']) ) 429 { 430 $toc = new siteexport_toc($this->functions); 431 $this->functions->debug->message("Generating eclipseDocZip", null, 2); 432 $this->filewriter->__moveDataToZip($toc->__getTOCXML($data), 'toc.xml'); 433 $this->filewriter->__moveDataToZip($toc->__getContextXML($data), 'context.xml'); 434 } else if ( !empty($_REQUEST['JavaHelpDocZip']) ) 435 { 436 $toc = new siteexport_javahelp($this->functions, $this->filewriter); 437 $toc->createTOCFiles($data); 438 439/* $toc = new siteexport_toc($this->functions); 440 list($tocData, $mapData) = $toc->__getJavaHelpTOCXML($data); 441 $this->functions->debug->message("Generating JavaHelpDocZip", null, 2); 442 $this->filewriter->__moveDataToZip($tocData, 'toc.xml'); 443 $this->filewriter->__moveDataToZip($mapData, 'map.xml'); 444*/ } 445 446 return $data; 447 } 448 449 /** 450 * Add page with ID to the package 451 **/ 452 function __siteexport_add_site( $ID ) { 453 global $conf, $currentID; 454 455 // Which is the current ID? 456 $currentID = $ID; 457 458 $this->functions->debug->message("========================================", null, 2); 459 $this->functions->debug->message("Adding Site: '$ID'", null, 2); 460 461 $request = $this->functions->settings->additionalParameters; 462 unset($request['diPlu']); // This will not be needed for the first request. 463 unset($request['diInv']); // This will not be needed for the first request. 464 465 // say, what to export and Build URL 466 // http://documentation:81/helpdesk/de/hds/getting-started?depthType=0&do=siteexport&ens=helpdesk%3Ade%3Ahds%3Agetting-started&pdfExport=1&renderer=siteexport_siteexportpdf&template=helpdesk 467 468 $do = (intval($_REQUEST['exportbody']) == 1 ? (empty($_REQUEST['renderer']) ? $conf['renderer_xhtml'] : $_REQUEST['renderer'] ) : '' ); 469 470 if ($do == 'pdf' && $this->filewriter->canDoPDF() ) 471 { 472 $do = 'export_siteexport_pdf'; 473 $_REQUEST['origRenderer'] = (empty($_REQUEST['renderer']) ? $conf['renderer_xhtml'] : $_REQUEST['renderer'] ); 474 } 475 476 $do = ($do == $conf['renderer_xhtml'] && intval($_REQUEST['exportbody']) != 1) ? '' : 'export_' . $do; 477 478 if ( $do != 'export_' && !empty($do) ) 479 { 480 $request['do'] = $do; 481 } 482 483 // set Template 484 if ( !empty( $_REQUEST['template'] ) ) { 485 $request['template'] = $_REQUEST['template']; 486 } 487 488 $this->functions->debug->message("REQUEST for add_site:", $request, 2); 489 490 $ID = $this->functions->cleanID($ID); 491 $url = $this->functions->wl($ID, $request, true, '&'); 492 493 // Parse URI PATH and add "html" 494 $fileName = $this->functions->getSiteName($ID, true); 495 496 $this->fileChecked[$url] = $fileName; // 2010-09-03 - One URL to one FileName 497 $this->functions->settings->depth = str_repeat('../', count(explode('/', $fileName))-1); 498 499 // fetch URL and save it in temp file 500 $tmpFile = $this->__getHTTPFile($url); 501 if ( $tmpFile === false ) { 502 return $this->functions->debug->message("Creating temporary download file failed for '$url'. See log for more information."); 503 return $this->functions->debug->runtimeException("Creating temporary download file failed for '$url'. See log for more information."); 504 } 505 506 // If a Filename was given that does not comply to the original name, use this one! 507 if ( !empty($tmpFile[1]) && !strstr($fileName, $tmpFile[1]) ) { 508 509 $dParts = explode('/', $fileName); 510 array_pop($dParts); 511 $dParts[] = $tmpFile[1]; 512 513 $fileName = implode('/', $dParts); 514 $this->fileChecked[$url] = $fileName; 515 } 516 517 // Add to zip 518 $status = $this->filewriter->__addFileToZip($tmpFile[0], $fileName); 519 @unlink($tmpFile[0]); 520 521 return $status; 522 } 523 524 /** 525 * Download the file via HTTP URL + recurse if this is not an image 526 * The file will be saved as temporary file. The filename is the result. 527 **/ 528 function __getHTTPFile($URL, $RECURSE=false, $newAdditionalParameters=null) { 529 global $conf; 530 531 $EXCLUDE = str_replace('/', '\/', ($this->getConf('exclude'))); 532 533 if ( !empty($EXCLUDE) && preg_match("/(".preg_quote($EXCLUDE,"/").")/i", $URL) ) { return false; } 534 535 require_once( DOKU_INC . 'inc/HTTPClient.php'); 536 537 $http = new HTTPProxy($this->functions->debug); 538 $http->max_bodysize = $conf['fetchsize']; 539 // $http->user = $_SERVER['PHP_AUTH_USER']; // Must not be set, or the files will be authenticated and have the edit thingies 540 // $http->pass = $_SERVER['PHP_AUTH_PW']; // Must not be set, or the files will be authenticated and have the edit thingies 541 542 // Add additional Params 543 $this->functions->addAdditionalParametersToURL($URL, $newAdditionalParameters); 544 545 $this->functions->debug->message("Fetching URL: '$URL'", null, 2); 546 $getData = $http->get($URL); 547 548 if( $getData === false ) { 549 $this->functions->debug->message("Sending request failed with error, HTTP status was '{$http->status}'.", $URL, 4); 550 return false; 551 } 552 553 if( empty($getData) ) { 554 $this->functions->debug->message("No data fetched.", null , 4); 555 return false; 556 } 557 558 $tmpFile = tempnam($this->functions->settings->tmpDir , 'siteexport__'); 559 $this->functions->debug->message("Temporary filename", $tmpFile, 1); 560 561 $fp = fopen( $tmpFile, "w"); 562 if(!$fp) { 563 $this->functions->debug->message("Can't open temporary File '$tmpFile'.", null , 4); 564 return false; 565 } 566 567 if ( !$RECURSE ) { 568 // Parse URI PATH and add "html" 569 $this->functions->debug->message("========================================", null, 1); 570 $this->functions->debug->message("Starting to recurse file '$URL'", null , 1); 571 $this->functions->debug->message("========================================", null, 1); 572 $this->__getInternalLinks($getData); 573 $this->functions->debug->message("========================================", null, 1); 574 $this->functions->debug->message("Finished to recurse file '$URL'", null , 1); 575 $this->functions->debug->message("========================================", null, 1); 576 } 577 578 fwrite($fp,$getData); 579 fclose($fp); 580 581 return array($tmpFile, preg_replace("/.*?filename=\"?(.*?)\"?;?$/", "$1", $http->resp_headers['content-disposition'])); 582 } 583 584 /** 585 * Find internal links in the currently downloaded file. This also matches inside CSS files 586 **/ 587 function __getInternalLinks(&$DATA) { 588 589 $PATTERN = '(href|src|action)="([^"]*)"'; 590 $CALLBACK = array($this, '__fetchAndReplaceLink'); 591 $DATA = preg_replace_callback("/$PATTERN/i", $CALLBACK, $DATA); 592 593 $PATTERNCSS = '(url\s*?)\([^\)]*\)'; 594 $DATA = preg_replace_callback("/$PATTERNCSS/i", $CALLBACK, $DATA); 595 } 596 597 /** 598 * Deep Fetch and replace of links inside the texts matched by __getInternalLinks 599 **/ 600 function __fetchAndReplaceLink($DATA) { 601 global $conf, $currentID; 602 603 $noDeepReplace = true; 604 $newAdditionalParameters = $this->functions->settings->additionalParameters; 605 $newDepth = $this->functions->settings->depth; 606 $hadBase = false; 607 608 $this->functions->debug->message("Starting Link Replacement", $DATA, 2); 609 610 // $DATA[2] = urldecode($DATA[2]); // Leads to problems because it does not re-encode the url 611 // External and mailto links 612 if ( preg_match("%^(https?://|mailto:|javascript:|data:)%", $DATA[2]) ) { 613 $this->functions->debug->message("Don't like http, mailto, data or javascript links here", null, 1); 614 return $this->__rebuildLink($DATA, ""); 615 } 616 //if ( preg_match("%^(https?://|mailto:|" . DOKU_BASE . "/_export/)%", $DATA[2]) ) { return $this->__rebuildLink($DATA, ""); } 617 // External media - this is deep down in the link, so we have to grep it out 618 if ( preg_match("%media=(https?://.*?$)%", $DATA[2], $matches) ) { 619 $DATA[2] = $matches[1]; 620 $this->functions->debug->message("This is an HTTP like somewhere else", $DATA, 1); 621 return $this->__rebuildLink($DATA, ""); 622 } 623 // reference only links won't have to be rewritten 624 if ( preg_match("%^#.*?$%", $DATA[2]) ) { 625 $this->functions->debug->message("This is a refercence only", null, 1); 626 return $this->__rebuildLink($DATA, ""); 627 } 628 629 // strip all things out 630 // changed Data 631 $PARAMS = @parse_url($DATA[2], PHP_URL_QUERY); 632 $ANCHOR = @parse_url($DATA[2], PHP_URL_FRAGMENT); 633 $DATA[2] = @parse_url($DATA[2], PHP_URL_PATH); 634 635 // 2010-08-25 - fix problem with relative movement in links ( "test/../test2" ) 636 $tmpData2 = ''; 637 while( $tmpData2 != $DATA[2] ) { 638 $tmpData2 = $DATA[2]; 639 $DATA[2] = preg_replace("#/(?!\.\.)[^\/]*?/\.\./#", '/', $DATA[2]); 640 } 641 642 $temp = preg_replace("%^" . DOKU_BASE . "%", "", $DATA[2]); 643 if ( $temp != $DATA[2] ) { 644 $DATA[2] = $temp; 645 $hadBase = true; // 2010-08-23 Check if there has been a rewrite here that will have to be considered later on 646 } 647 648 $this->functions->debug->message("URL before rewriting option for others than 1", array($DATA, $PARAMS, $hadBase), 1); 649 650 // Handle rewrites other than 1 651 if ( !preg_match('$^/?lib/$', $DATA[2]) ) { 652 if ( $conf['userewrite'] == 2 ) { 653 $DATA[2] = $this->__getInternalRewriteURL($DATA[2]); 654 } elseif ( $conf['userewrite'] == 0 ) { 655 $this->__getParamsAndDataRewritten($DATA, $PARAMS); 656 } 657 } 658 659 $this->functions->debug->message("URL before rewriting option", array($DATA, $PARAMS), 2); 660 661 $ORIGDATA2 = $DATA; 662 // $ORIGDATA2 = $DATA[2]; // 08/10/2010 - this line required a $this->functions->wl which may mess up with the base URL 663 $this->functions->debug->message("OrigDATA is:", $ORIGDATA2, 1); 664 665 // Generate ID 666 $DATA[2] = str_replace('/', ':', $DATA[2]); 667 668 // If Data was empty this must be the same file!; 669 if ( empty( $DATA[2] ) ) { 670 $DATA[2] = $currentID; 671 } 672 673 $ID = $DATA[2]; 674 $MEDIAMATCHER = "#(_media(/|:)|media=|_detail(/|:)|_export(/|:)|do=export_)#i"; // 2010-10-23 added "(/|:)" for the ID may not contain slashes anymore 675 $ID = $this->functions->cleanID($DATA[2], null, preg_match($MEDIAMATCHER, $DATA[2]) ); 676 // $ID = $this->functions->cleanID($DATA[2], null, strstr($DATA[2], 'media') ); // Export anpassung nun weiter unten 677 678 // $IDexists = page_exists($ID); // 08/10/2010 - Not needed. This will be done in the next block. 679 // $this->functions->debug->message("Current ID: '$ID' exists: '" . ($IDexists ? 'true' : 'false') . "' (will be set to 'false' anyway)", null, 1); 680 681 $IDifIDnotExists = $ID; // 08/10/2010 - Save ID - with possible upper cases to preserve them 682 $IDexists = false; 683 684 $this->functions->debug->message("Resolving ID: '$ID'", null, 2); 685 if ( preg_match($MEDIAMATCHER, $DATA[2]) ) { 686 resolve_mediaid(null, $ID, $IDexists); 687 688 $this->functions->debug->message("Current mediaID to filename: '" . mediaFN($ID) . "'", null, 2); 689 } else { 690 resolve_pageid(null, $ID, $IDexists); 691 $this->functions->debug->message("Current ID to filename: '" . wikiFN($ID) . "'", null, 2); 692 } 693 694 $this->functions->debug->message("Current ID after resolvement: '$ID' the ID does exist: '" . ($IDexists ? 'true' : 'false') . "'", null, 2); 695 // $ORIGDATA2 = @parse_url($this->functions->wl($ORIGDATA2, null, true)); // What was the next 2 line for? It did mess up with links from {{jdoc>}} 696 // $this->functions->debug->message("OrigData ID after parse:", $ORIGDATA2, 1); // 08/10/2010 - The lines are obsolete when the $ORIGDATA2 = $DATA. $ORIGDATA is only for fallback 697 698 // 08/10/2010 - If the ID does not exist, we may have a problem here with upper cases - they will all be lower by now! 699 if ( !$IDexists ) { 700 $ID = $IDifIDnotExists; // there may have been presevered Upper cases. We will need them! 701 } 702 703 // $this->functions->cleanID($DATA[2], null, strstr($DATA[2], 'media') || strstr($DATA[2], 'export') ); 704 if ( substr($ID, -1) == ':' || empty($ID) ) $ID .= $conf['start']; 705 706 // Generate Download URL 707 // $PARAMS = trim(str_replace('&', '&', $PARAMS)); 708 $PARAMS = trim($PARAMS); 709 $this->functions->removeWikiVariables($PARAMS, false, true); 710 711 $url = $this->functions->wl($ID, null, true, null, null, true, $hadBase) . ( !empty( $ANCHOR) ? '#' . $ANCHOR : '' ) . ( !empty( $PARAMS) ? '?' . $PARAMS : '' ); 712 $this->functions->debug->message("URL from ID: '$url'", null, 2); 713 714 // Parse URI PATH and add "html" 715 $uri = @parse_url($url); 716 $DATA[2] = $uri['path']; 717 $DATA['ANCHOR'] = $ANCHOR; 718 $DATA['PARAMS'] = $PARAMS; 719 720 $this->functions->debug->message("DATA after parsing.", $DATA, 2); 721 722 // Second Rewrite for UseRewrite = 2 723 if ( $conf['userewrite'] == 2 ) { 724 $DATA[2] = preg_replace( '$/lib/.*?fetch\.php$', '', $DATA[2]); 725 $DATA[2] = preg_replace( '%(/lib/.*?detail\.php.*$)%', '\1' . '.' . $this->functions->settings->fileType, $DATA[2]); 726 727 if ( preg_match( '%/(lib/.*?detail|doku)\.php%', $DATA[2])) { 728 $noDeepReplace = false; 729 $fileName = $this->functions->getSiteName($ID); 730 $newDepth = str_repeat('../', count(explode('/', $fileName))-1); 731 } 732 733 $this->functions->debug->message("DATA after second rewrite with UseRewrite = 2", array($DATA, $noDeepReplace, $fileName, $newDepth), 1); 734 } 735 736 switch ( array_pop(explode('/', $DATA[2])) ) { 737 // CSS Extra Handling with extra rewrites 738 case 'css.php' : // $DATA[2] .= ( !$this->functions->settings->addParams || empty($PARAMS) ? '' : '.' . $this->functions->cleanID(preg_replace("/(=|\?|&)/", ".", $PARAMS))) . '.css'; 739 $DATA[2] .= '.' . $this->functions->cleanID(preg_replace("/(=|\?|&)/", ".", $PARAMS)) . '.css'; // allways put parameters behind 740 // No paramters needed since they are rewritten. 741 $DATA['PARAMS'] = ""; 742 $noDeepReplace = false; 743 $fileName = $this->functions->getSiteName($ID); 744 $newDepth = str_repeat('../', count(explode('/', $fileName))-1); 745 $newAdditionalParameters['do'] = 'siteexport'; 746 747 $this->functions->debug->message("This is CSS file", array($DATA, $noDeepReplace, $fileName, $newDepth, $newAdditionalParameters), 2); 748 749 break; 750 case 'js.php' : // $DATA[2] .= ( !$this->functions->settings->addParams || empty($PARAMS) ? '' : '.' . $this->functions->cleanID(preg_replace("/(=|\?|&)/", ".", $PARAMS))) . '.js'; 751 $DATA[2] .= '.t.' . $this->functions->cleanID($_REQUEST['template']) . '.js'; // allways put parameters behind 752 // set Template 753 if ( !empty( $_REQUEST['template'] ) ) { 754 $url .= ( strstr($url, '?') ? '&' : '?' ) . 'template=' . $_REQUEST['template']; 755 } 756 // No paramters needed since they are rewritten. 757 $DATA['PARAMS'] = ""; 758 $newAdditionalParameters['do'] = 'siteexport'; 759 760 $this->functions->debug->message("This is JS file", array($DATA, $url, $fileName, $newAdditionalParameters), 2); 761 762 break; 763 // Detail Handling with extra Rewrites if Paramaters are available - otherwise this is just the fetch 764 case 'indexer.php' : 765 $this->functions->debug->message("Skipping indexer", null, 2); 766 return ""; 767 break; 768 case 'detail.php' : 769 $fileName = $this->functions->getSiteName($ID, true); // 2010-09-03 - rewrite with override enabled 770 case 'doku.php' : 771 if ( $this->functions->settings->addParams ) { 772 $noDeepReplace = false; 773 774 if ( empty($fileName) ) { 775 $fileName = $this->functions->getSiteName($ID); // 2010-09-03 - rewrite with override enabled 776 } 777 778 $newDepth = str_repeat('../', count(explode('/', $fileName))-1); 779 $this->__rebuildDataForNormalFiles($DATA, $PARAMS); 780 781 $this->functions->debug->message("This is doku.php or detail.php file with addParams", array($DATA, $fileName, $newDepth, $newAdditionalParameters), 2); 782 break; 783 } 784 785 $url = str_replace('detail.php', 'fetch.php', $url); 786 $this->functions->debug->message("This is doku.php or detail.php file '$url'", null, 2); 787 // Fetch Handling for media - rewriting everything 788 case 'fetch.php': 789 $this->__getParamsAndDataRewritten($DATA, $PARAMS, 'media'); 790 791 $DATA[2] = str_replace('/', ':', $DATA[2]); 792 $ID = $this->functions->cleanID($DATA[2], null, strstr($DATA[2], 'media')); 793 794 $urlM = ml($ID, null, true); 795 $uriM = @parse_url($urlM); 796 $DATA[2] = $uriM['path'] . ( !empty( $ANCHOR) ? '#' . $ANCHOR : '' ) . ( !empty( $PARAMS) ? '?' . $PARAMS : '' ); 797 798 $DATA['PARAMS'] = ""; 799 $newAdditionalParameters = array(); 800 801 $this->functions->debug->message("This is fetch.php file", array($DATA, $ID), 2); 802 break; 803 804 // default Handling for Pages 805 default : 806 if ( preg_match("%" . DOKU_BASE . "_detail/%", $DATA[2]) ) { 807 808 // GET ID Param from origdata2 809 preg_match("#id=(.*?)(&|\")#i", $DATA[0], $backlinkID); 810 $this->__rebuildDataForNormalFiles($DATA, $PARAMS); 811 812 $fileIDPart = isset($backlinkID[1]) && !empty($backlinkID[1]) ? $this->functions->cleanID(urldecode($backlinkID[1])) : 'detail'; 813 814 $DATA[2] .= '/' . $fileIDPart . '.' . $this->functions->settings->fileType; // add namespace and subpage for back button and add filetype 815 816 $noDeepReplace = false; 817 $fileName = $this->functions->shortenName($DATA[2]); 818 $newDepth = str_repeat('../', count(explode('/', $fileName))-1); 819 $url .= ( strstr($url, '?') ? '&' : '?' ) . 'id=' . $fileIDPart; // add id-part to URL for backlinks 820 821 $DATA['PARAMS'] = ""; 822 823 $this->functions->debug->message("This is something with '_detail' file", array($DATA, $backlinkID, $newDepth, $url), 2); 824 } else if ( preg_match("%" . DOKU_BASE . "_export/(.*?)/%", $DATA[2], $fileType) ) { 825 826 // Fixes multiple codeblocks in one file 827 $this->__rebuildDataForNormalFiles($DATA, $PARAMS); 828 829 // add the Params no matter what they are. This is export. We don't mess with other files 830 // adding the "/" fixes the usage of multiple codeblocks in the same namespace 831 $DATA[2] .= (empty( $PARAMS ) ? '' : '/' . $PARAMS) . '.'. $fileType[1]; 832 833 $DATA['PARAMS'] = ""; 834 $this->functions->debug->message("This is something with '_export' file", $DATA, 2); 835 836 } else if ( $IDexists ) { // 08/10/2010 - was page_exists($ID) - but this should do as well. 837 // If this is a page ... skip it! 838 $DATA[2] .= ( !$this->functions->settings->addParams || empty($PARAMS) ? '' : '.' . $this->functions->cleanID(preg_replace("/(=|\?|&)/", ".", $PARAMS))) . '.' . $this->functions->settings->fileType; 839 840 // 2012-06-15 originally has an absolute path ... we might need a relative one if not in our namespace 841 $this->functions->debug->message("OK, this is to be absolute: " . (empty($_REQUEST['absolutePath'])?'false':'true'), null, 1); 842 if ( empty($_REQUEST['absolutePath']) ) 843 { 844 $DATA[2] = $this->functions->getRelativeURL($DATA[2], $currentID); 845 } 846 847 $DATA[2] = $this->functions->shortenName($DATA[2]); 848 849 // If Parameters are to be included in the filename - they must not be added twice 850 if ( $this->functions->settings->addParams ) $DATA['PARAMS'] = ""; 851 852 $this->functions->debug->message("This page really exists", $DATA, 1); 853 854 return $this->__rebuildLink($DATA); 855 } else { 856 $this->__rebuildDataForNormalFiles($DATA, $PARAMS); 857 } 858 859 unset($newAdditionalParameters['diPlu']); 860 } 861 862 863 $this->functions->debug->message("DATA after SWITCH CASE decision", array($DATA, $noDeepReplace, $fileName, $newDepth), 1); 864 865 if ( $this->filewriter->canDoPDF() ) { 866 $this->functions->addAdditionalParametersToURL($url, $newAdditionalParameters); 867 $DATA[2] = $url; 868 unset($DATA['PARAMS']); 869 $url = $this->__rebuildLink($DATA, ''); 870 871 $this->functions->debug->message("Creating PDF with URL '$url'", null, 2); 872 873 return $url; 874 } 875 876 // Create Name to save the file at 877 $DATA[2] = str_replace(':', '_', $DATA[2]); 878 $DATA[2] = $this->functions->shortenName($DATA[2]); 879 880 881 // File already loaded? 882 // 2010-10-23 - changes in_array from DATA[2] to $url - to check real URLs, the DATA[2] file will be checked with fileExistsInZip 883 if ( in_array($url, array_keys($this->fileChecked)) ) { 884 $DATA[2] = $this->fileChecked[$url]; 885 $this->functions->debug->message("File has been checked before.", array($DATA, $url), 2); 886 return $this->__rebuildLink($DATA); 887 } 888 889 // 2010-09-03 - second check if the file is in the ZIP already. 890 if ( $this->filewriter->fileExistsInZip($DATA[2]) ) { 891 $this->functions->debug->message("File with DATA exists in ZIP.", $DATA, 3); 892 return $this->__rebuildLink($DATA); 893 } 894 895 // 2010-10-23 - What if this is a fetch.php? than we produced an error. 896 // $this->fileChecked[] = $DATA[2]; 897 $this->fileChecked[$url] = $DATA[2]; // 2010-09-03 - One URL to one FileName 898 899 // get tempFile and save it 900 $origDepth = $this->functions->settings->depth; 901 $this->functions->settings->depth = $newDepth; 902 903 $tmpID = $currentID; 904 $tmpFile === false; 905 906 $this->functions->debug->message("Going to get the file", array($url, $noDeepReplace, $newAdditionalParameters), 2); 907 $tmpFile = $this->__getHTTPFile($url, $noDeepReplace, $newAdditionalParameters); 908 $this->functions->debug->message("This is the getHTTPFile result", $tmpFile, 2); 909 910 $currentID = $tmpID; 911 $this->functions->settings->depth = $origDepth; // 2010-09-03 - Reset depth at the very end 912 913 if ( $tmpFile === false ) { 914 // Keep an potentially extra link intact 915 916 $this->functions->debug->message("The fetched file '$url' is 'false'", null, 3); 917 if ( $IDexists === false ) { 918 $this->functions->debug->message("The file does not exist, fallback to ORIGDATA", $ORIGDATA2, 2); 919 $DATA[2] = $this->functions->shortenName($ORIGDATA2[2]); // get Origdata Path 920 } 921 922 $this->fileChecked[$url] = $DATA[2]; // 2010-09-03 - One URL to one FileName 923 $link = $this->__rebuildLink($DATA); 924 $this->functions->debug->message("Final Link after empty file from '$url'", null, 2); 925 926 return $link; 927 } 928 929 $this->functions->debug->message("The fetched file looks good.", $tmpFile, 1); 930 931 // If a Filename was given that does not comply to the original name, us this one! 932 if ( !empty($tmpFile[1]) && !strstr($DATA[2], $tmpFile[1]) ) { 933 934 $dParts = explode('/', $DATA[2]); 935 array_pop($dParts); 936 $dParts[] = $tmpFile[1]; 937 938 $DATA[2] = implode('/', $dParts); 939 } 940 941 // Add to zip 942 $this->fileChecked[$url] = $DATA[2]; // 2010-09-03 - One URL to one FileName 943 944 $status = $this->filewriter->__addFileToZip($tmpFile[0], $DATA[2]); 945 @unlink($tmpFile[0]); 946 947 $newURL = $this->__rebuildLink($DATA); 948 $this->functions->debug->message("Returning final Link to document: '$newURL'", null, 2); 949 950 return $newURL; 951 } 952 953 /** 954 * build the new link to be put in place for the donwloaded site 955 **/ 956 function __rebuildLink($DATA, $DEPTH = null) { 957 958 // depth is set, skip this one 959 if ( is_null( $DEPTH ) ) $DEPTH = $this->functions->settings->depth; 960 $DATA[2] .= ( !empty( $DATA['PARAMS']) ? '?' . $DATA['PARAMS'] : '' ) . ( !empty( $DATA['ANCHOR'] ) ? '#' . $DATA['ANCHOR'] : '' ); 961 962 $newURL = $DATA[1] == 'url' ? $DATA[1] . '(' . $DEPTH . $DATA[2] . ')' : $DATA[1] . '="' . $DEPTH . $DATA[2] . '"'; 963 $this->functions->debug->message("Re-created URL: '$newURL'", null, 2); 964 965 return $newURL; 966 } 967 968 969 /** 970 * remove an old zip file 971 **/ 972 function __removeOldZip( $FILENAMEID=null, $checkForMore=true ) { 973 global $INFO; 974 global $conf; 975 976 $returnValue = true; 977 978 if ( empty($FILENAMEID) ) { 979 $FILENAMEID = $this->functions->settings->origZipFile; 980 } 981 982 if ( !file_exists(mediaFN($FILENAMEID)) ) { 983 $returnValue = true; 984 } else { 985 986 if ( !$this->functions->settings->isCLI ) 987 { 988 $INFO = pageinfo(); 989 if ( $INFO['perm'] < AUTH_DELETE && !$this->functions->settings->isAuthed ) { 990 list ( $USER, $PASS) = $this->functions->basic_authentication(); 991 auth_login($USER, $PASS); 992 $this->functions->settings->isAuthed = true; 993 } 994 } 995 996 require_once( DOKU_INC . 'inc/media.php'); 997 if ( !media_delete($FILENAMEID, $INFO['perm']) ) { 998 $returnValue = false; 999 } 1000 } 1001 1002 if ( $checkForMore ) { 1003 // Try to remove more files. 1004 $ns = getNS($FILENAMEID); 1005 $fn = $this->functions->getSpecialExportFileName(noNS($FILENAMEID), '.+'); 1006 1007 $data = array(); 1008 search($data, $conf['mediadir'], 'search_media', array('pattern' => "/$fn$/i"), $ns); 1009 1010 if ( count($data > 0) ) { 1011 1012 // 30 Minuten Cache Zeit 1013 $cache = $this->getConf('cachetime'); 1014 foreach ( $data as $media ) { 1015 1016 //decide if has to be deleted needed: 1017 if( $media['mtime'] < time()-$cache) { 1018 $this->__removeOldZip($media['id'], false); 1019 } 1020 } 1021 } 1022 1023 } 1024 1025 return $returnValue; 1026 } 1027 1028 /** 1029 * if confrewrite is set to internal rewrite, use this function - taken from a DW renderer 1030 **/ 1031 function __getInternalRewriteURL($url) { 1032 global $conf; 1033 1034 //construct page id from request URI 1035 if( $conf['userewrite'] != 2) { return $url; } 1036 1037 //get the script URL 1038 if($conf['basedir']) { 1039 $relpath = ''; 1040 $script = $conf['basedir'].$relpath.basename($_SERVER['SCRIPT_FILENAME']); 1041 } elseif($_SERVER['DOCUMENT_ROOT'] && $_SERVER['SCRIPT_FILENAME']){ 1042 $script = preg_replace ('/^'.preg_quote($_SERVER['DOCUMENT_ROOT'],'/').'/','', 1043 $_SERVER['SCRIPT_FILENAME']); 1044 $script = '/'.$script; 1045 }else{ 1046 $script = $_SERVER['SCRIPT_NAME']; 1047 } 1048 1049 //clean script and request (fixes a windows problem) 1050 $script = preg_replace('/\/\/+/','/',$script); 1051 $request = preg_replace('/\/\/+/','/',$url); 1052 1053 //remove script URL and Querystring to gain the id 1054 if(preg_match('/^'.preg_quote($script,'/').'(.*)/',$request, $match)){ 1055 $id = preg_replace ('/\?.*/','',$match[1]); 1056 } 1057 $id = urldecode($id); 1058 //strip leading slashes 1059 $id = preg_replace('!^/+!','',$id); 1060 1061 return $id; 1062 } 1063 1064 /** 1065 * rewrite parameter calls 1066 **/ 1067 function __getParamsAndDataRewritten(&$DATA, &$PARAMS, $IDKEY='id') { 1068 1069 $PARRAY = explode('&', str_replace('&', '&', $PARAMS) ); 1070 $PARAMS = ""; 1071 1072 foreach ( $PARRAY as $item ) { 1073 list($key, $value) = explode('=', $item, 2); 1074 if ( empty($key) || empty($value) ) 1075 continue; 1076 1077 if ( strtolower(trim($key)) == $IDKEY ) { 1078 $DATA[2] = preg_replace("%^" . DOKU_BASE . "%", "", $value); 1079 continue; 1080 } 1081 1082 if ( !empty( $PARAMS) ) { 1083 $PARAMS .= '&'; 1084 } 1085 1086 $PARAMS .= "$key=$value"; 1087 } 1088 } 1089 1090 /** 1091 * rewrite detail.php calls 1092 **/ 1093 function __rebuildDataForNormalFiles(&$DATA, &$PARAMS) { 1094 $PARTS = explode('.', $DATA[2]); 1095 if ( count($PARTS) > 1 ) { 1096 $EXT = '.' . array_pop($PARTS); 1097 } 1098 1099 $PARAMS = preg_replace("/(=|\?|&)/", ".", $PARAMS); 1100 $DATA[2] = implode('.', $PARTS) . ( !$this->functions->settings->addParams || empty($PARAMS) ? '' : '.' . $this->functions->cleanID($PARAMS)) . ( $EXT == '.php' ? '.' . $this->functions->settings->fileType : $EXT ); 1101 $DATA[2] = preg_replace("/\.+/", ".", $DATA[2]); 1102 } 1103 1104 1105 1106 1107 /* 1108 * Clean JS and CSS cache files 1109 */ 1110 function cleanCacheFiles() { 1111 1112 $_SERVER['HTTP_HOST'] = preg_replace("/:?\d+$/", '', $_SERVER['HTTP_HOST']); 1113 $cache = getCacheName('scripts'.$_SERVER['HTTP_HOST'].'-siteexport-js-'.$_SERVER['SERVER_PORT'],'.js'); 1114 $this->unlinkIfExists($cache); 1115 1116 $tpl = trim(preg_replace('/[^\w-]+/','',$_REQUEST['template'])); 1117 if($tpl) 1118 { 1119 $tplinc = DOKU_INC.'lib/tpl/'.$tpl.'/'; 1120 $tpldir = DOKU_BASE.'lib/tpl/'.$tpl.'/'; 1121 } else { 1122 $tplinc = DOKU_TPLINC; 1123 $tpldir = DOKU_TPL; 1124 } 1125 1126 // The generated script depends on some dynamic options 1127 $cache = getCacheName('styles'.$_SERVER['HTTP_HOST'].'-siteexport-js-'.$_SERVER['SERVER_PORT'].DOKU_BASE.$tplinc.$style,'.css'); 1128 $this->unlinkIfExists($cache); 1129 } 1130 1131 function unlinkIfExists($cache) { 1132 if ( file_exists($cache) ) { 1133 @unlink($cache); 1134 if(function_exists('gzopen')) @unlink("$cache.gz"); 1135 } 1136 } 1137 1138 // Private unset function 1139 private function clear(&$variable) 1140 { 1141 if ( isset($variable) ) 1142 { 1143 unset($variable); 1144 } 1145 } 1146}