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', /** @scrutinizer ignore-type */ 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 27use dokuwiki\File\PageResolver; 28use dokuwiki\File\MediaResolver; 29 30class action_plugin_siteexport_ajax extends DokuWiki_Action_Plugin 31{ 32 /** 33 * New internal variables for better structure 34 */ 35 private $filewriter = null; 36 public $functions = null; 37 38 // List of files that have already been checked 39 private $fileChecked = array(); 40 41 // Namespace of the page to export 42 private $namespace = ''; 43 44 /** 45 * Register Plugin in DW 46 **/ 47 public function register(Doku_Event_Handler $controller) { 48 $controller->register_hook('AJAX_CALL_UNKNOWN', 'BEFORE', $this, 'ajax_siteexport_provider'); 49 $controller->register_hook('ACTION_ACT_PREPROCESS', 'BEFORE', $this, 'siteexport_action'); 50 } 51 52 /** 53 * AJAX Provider - check what is going to be done 54 * @param $event 55 * @param $args 56 */ 57 public function ajax_siteexport_provider(Doku_Event &$event, $args) { 58 59 // If this is not a siteexport call, ignore it. 60 if (!strstr($event->data, '__siteexport')) 61 { 62 return; 63 } 64 65 $this->__init_functions(true); 66 67 switch ($event->data) { 68 case '__siteexport_getsitelist': $this->ajax_siteexport_getsitelist($event); break; 69 case '__siteexport_addsite': $this->ajax_siteexport_addsite($event); break; 70 case '__siteexport_generateurl': $this->ajax_siteexport_generateurl($event); break; 71 case '__siteexport_aggregate': $this->ajax_siteexport_aggregate($event); break; 72 } 73 } 74 75 /** 76 * Export from a URL - action 77 * @param $event 78 */ 79 public function siteexport_action( Doku_Event &$event ) { 80 global $ID; 81 82 // Check if the 'do' was siteexport 83 $keys = is_array($event->data) ? array_keys($event->data) : null; 84 $command = is_array($keys) ? array_shift($keys) : $event->data; 85 if ( $command != 'siteexport' ) { return false; } 86 $event->data = act_clean($event->data); 87 88 if ( headers_sent() ) { 89 msg("The siteexport function has to be called prior to any header output.", -1); 90 } 91 92 $this->__init_functions(); 93 94 $this->functions->debug->message("========================================", null, 1); 95 $this->functions->debug->message("Starting export from URL call", null, 1); 96 $this->functions->debug->message("----------------------------------------", null, 1); 97 98 $event->preventDefault(); 99 $event->stopPropagation(); 100 101 // Fake security Token if none given 102 if (empty($_REQUEST['sectok'])) { 103 $_REQUEST['sectok'] = $this->functions->getSecurityToken(); 104 } 105 106 // The timer will be used to do redirects if needed to prevent timeouts 107 $starttimer = time(); 108 $timerdiff = $this->getConf('max_execution_time'); 109 110 $data = $this->__get_siteexport_list_and_init_tocs($ID, !empty($_REQUEST['startcounter'])); 111 112 if ($data === false) { 113 header("HTTP/1.0 401 Unauthorized"); 114 print 'Unauthorized'; 115 exit; 116 } 117 118 $counter = 0; 119 120 if ( count($data) == 0 && !$this->functions->settings->hasValidCacheFile ) { 121 exit( "No Data to export" ); 122 } 123 124 foreach ( $data as $site ) { 125 126 if ( intval($site['exists']) == 1 || !isset($site['exists']) ) { 127 128 // Skip over the amount of urls that have been exported already 129 if ( empty($_REQUEST['startcounter']) || $counter >= intval($_REQUEST['startcounter']) ) { 130 $status = $this->__siteexport_add_site($site['id']); 131 132 if ( $status === false ) { 133 $this->functions->debug->message("----------------------------------------", null, 1); 134 $this->functions->debug->message("Errors during export from URL call", null, 1); 135 $this->functions->debug->message("========================================", null, 1); 136 print $this->functions->debug->runtimeErrors; 137 exit(0); // We need to stop 138 } 139 } 140 } 141 142 $counter++; 143 if (time()-$starttimer >= $timerdiff) { 144 $this->functions->debug->message("Will Redirect", null, 1); 145 $this->handleRuntimeErrorOutput(); 146 $this->functions->startRedirctProcess($counter); 147 } 148 } 149 150 $this->functions->debug->message("----------------------------------------", null, 1); 151 $this->functions->debug->message("Finishing export from URL call", null, 1); 152 $this->functions->debug->message("========================================", null, 1); 153 154 $this->cleanCacheFiles(); 155 156 $URL = ml($this->functions->settings->origZipFile, array('cache' => 'nocache', 'siteexport' => $this->functions->settings->pattern, 'sectok' => $this->functions->getSecurityToken()), true, '&'); 157 $this->functions->debug->message("Redirecting to final file", $URL, 2); 158 159 $this->handleRuntimeErrorOutput(); 160 send_redirect($URL); 161 exit(0); // Should not be reached, but anyways 162 } 163 164 private function handleRuntimeErrorOutput() 165 { 166 if (!empty($this->functions->debug->runtimeErrors)) 167 { 168 $this->filewriter->__moveDataToZip($this->functions->debug->runtimeErrors, '_runtime_error/' . time() . '.html'); 169 } 170 } 171 172 public function __init_functions($isAJAX = false) 173 { 174 global $conf; 175 176 $conf['useslash'] = 1; 177 178 $this->functions = new siteexport_functions(true, $isAJAX); 179 $this->filewriter = new siteexport_zipfilewriter($this->functions); 180 181 // Check for PDF Capabilities 182 if ($this->filewriter->canDoPDF()) { 183 $this->functions->settings->fileType = 'pdf'; 184 } 185 } 186 187 /** 188 * Prepares the generated URL for direct download access 189 * Also gives back the parameters for this URL 190 * @param $event init event of the ajax request 191 */ 192 private function ajax_siteexport_prepareURL_and_POSTData(Doku_Event &$event) { 193 194 $event->preventDefault(); 195 $event->stopPropagation(); 196 197 // Retrieve Information for download URL 198 $this->functions->debug->message("Prepared URL and POST from Request:", $_REQUEST, 2); 199 $url = $this->functions->prepare_POSTData($_REQUEST); 200 $combined = $this->functions->urlToPathAndParams($url); 201 list($path, $query) = explode('?', $combined, 2); 202 $return = array($url, $combined, $path, $query); 203 204 $this->functions->debug->message("Prepared URL and POST data:", $return, 2); 205 return $return; 206 } 207 208 /** 209 * generate direct access URL 210 **/ 211 private function ajax_siteexport_generateurl(Doku_Event &$event) { 212 213 global $INPUT; 214 215 list($url, $combined, $path, $POSTData) = $this->ajax_siteexport_prepareURL_and_POSTData($event); 216 217 // WGET Redirects - this is an option for wget only. 218 // 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 219 // Therefore we assume it takes about 5s for each page - that gives the freedom to have anough time for redirect. 220 $maxRedirectNumber = ceil((count($this->__get_siteexport_list($INPUT->str('ns'), true))*5)/$this->getConf('max_execution_time')); 221 $maxRedirect = $maxRedirectNumber > 0 ? '--max-redirect=' . ($maxRedirectNumber+3) . ' ' : ''; 222 $maxRedirs = $maxRedirectNumber > 0 ? '--max-redirs ' . ($maxRedirectNumber+3) . ' ' : ''; 223 224 $this->functions->debug->message("Generating Direct Download URL", $url, 2); 225 226 // If there was a Runtime Exception 227 if (!$this->functions->debug->firstRE()) { 228 $this->functions->debug->message("There have been errors while generating the download URLs.", null, 4); 229 return; 230 } 231 232 $zipFile = explode(":", ($this->getConf('zipfilename'))); 233 $zipFile = array_pop($zipFile); 234 235 echo $url; 236 echo "\n"; 237 echo 'wget ' . $maxRedirect . '--output-document=' . $zipFile . ' --post-data="' . $POSTData . '" ' . wl(cleanID($path), null, true) . ' --http-user=USER --http-passwd=PASSWD'; 238 echo "\n"; 239 echo 'curl -L ' . $maxRedirs . '-o ' . $zipFile . ' -d "' . $POSTData . '" ' . wl(cleanID($path), null, true) . ' --anyauth --user USER:PASSWD'; 240 echo "\n"; 241 242 $this->functions->debug->message("Checking for Cron parameters: ", $combined, 1); 243 244 return; 245 } 246 247 /** 248 * Get List of sites to be exported for AJAX (wrapper) 249 **/ 250 private function ajax_siteexport_getsitelist(Doku_Event &$event) { 251 252 global $INPUT; 253 254 $event->preventDefault(); 255 $event->stopPropagation(); 256 257 $data = $this->__get_siteexport_list_and_init_tocs($INPUT->str('ns')); 258 259 // Important for reconaisance of the session 260 261 if ($data === false) 262 { 263 $this->functions->debug->runtimeException("No data generated. List of Files is 'false'."); 264 return; 265 } 266 267 if (empty($data) && !$this->functions->settings->hasValidCacheFile) 268 { 269 $this->functions->debug->runtimeException("Generated list is empty."); 270 return; 271 } 272 273 // If there was a Runtime Exception 274 if (!$this->functions->debug->firstRE()) 275 { 276 $this->functions->debug->message("There have been errors while generating site list.", null, 4); 277 return; 278 } 279 280 echo "{$this->functions->settings->pattern}\n"; 281 echo $this->functions->downloadURL() . "\n"; 282 foreach ($data as $line) { 283 echo $line['id'] . "\n"; 284 } 285 286 return; 287 } 288 289 private function ajax_siteexport_aggregate(Doku_Event &$event) { 290 291 // Quick preparations for one page only 292 if ($this->filewriter->hasValidCacheFile($_REQUEST)) { 293 $this->functions->debug->message("Had a valid cache file and will use it.", null, 2); 294 print $this->functions->downloadURL(); 295 296 $event->preventDefault(); 297 $event->stopPropagation(); 298 } else { 299 // Then go for it! 300 $this->functions->debug->message("Will create a new cache thing.", null, 2); 301 $this->ajax_siteexport_addsite($event); 302 } 303 304 } 305 306 /** 307 * Add a page to the package (for AJAX calls - Wrapper) 308 **/ 309 private function ajax_siteexport_addsite(Doku_Event &$event) { 310 311 global $INPUT; 312 313 $event->preventDefault(); 314 $event->stopPropagation(); 315 316 $this->functions->debug->message("========================================", null, 1); 317 $this->functions->debug->message("Starting export from AJAX call", null, 1); 318 $this->functions->debug->message("----------------------------------------", null, 1); 319 320 $status = $this->__siteexport_add_site($INPUT->str('site')); 321 if ( $status === false ) { 322 $this->functions->debug->message("----------------------------------------", null, 1); 323 $this->functions->debug->message("Errors during export from AJAX call", null, 1); 324 $this->functions->debug->message("========================================", null, 1); 325 return; 326 } 327 328 $this->functions->debug->message("----------------------------------------", null, 1); 329 $this->functions->debug->message("Finishing export from AJAX call", null, 1); 330 $this->functions->debug->message("========================================", null, 1); 331 332 // Print the download zip-File 333 $this->cleanCacheFiles(); 334 335 // If there was a Runtime Exception 336 if (!$this->functions->debug->firstRE()) { 337 $this->functions->debug->message("There have been errors during the export.", null, 4); 338 return; 339 } 340 341 print $this->functions->downloadURL(); 342 return; 343 } 344 345 /** 346 * Fetch the list of pages to be exported 347 **/ 348 private function __get_siteexport_list($NS, $overrideCache = false) { 349 global $conf, $INPUT; 350 351 $PAGE = ""; 352 $NS = $this->namespace = $this->functions->getNamespaceFromID($NS, $PAGE); 353 $this->functions->debug->message("ROOT Namespace to export from: '{$NS}' / {$this->namespace}", null, 1); 354 355 $depth = $this->getConf('depth'); 356 $query = ''; 357 $doSearch = 'search_allpages'; 358 359 switch ($INPUT->int('depthType')) { 360 case 0: 361 $query = $this->functions->cleanID(str_replace(":", "/", $NS . ':' . $PAGE)); 362 $NS = (new PageResolver($NS))->resolveId($PAGE); 363 $exists = page_exists($NS); 364 365 if ($exists) { 366 $data = array(array('id' => $PAGE)); 367 368 $this->functions->debug->message("Checking for Cache, depthType:0", null, 2); 369 if (!$overrideCache && $this->filewriter->hasValidCacheFile($_REQUEST, $data)) 370 { 371 return array(); 372 } 373 374 return $data; 375 } else { 376 // Does not exist, try next case 377 } 378 case 1: $depth = 0; 379 break; 380 case 2: $depth = $INPUT->int('depth'); 381 break; 382 } 383 384 $opts = array('depth' => $depth, 'skipacl' => $this->getConf('skipacl'), 'query' => $query); 385 $this->functions->debug->message("Options", $opts, 2); 386 387 $data = array(); 388 require_once (DOKU_INC . 'inc/search.php'); 389 390 // Check, which TOC to take 391 if (!$this->functions->settings->useTOCFile) { 392 search($data, $conf['datadir'], $doSearch, $opts, $this->namespace); 393 } else { 394 $this->functions->debug->message("Using TOC for data", null, 2); 395 396 $doSearch = 'search_pagename'; 397 398 // Create Data of the TOC File should be used instead 399 $opts['query'] = 'toc.txt'; 400 401 $RAWdata = array(); 402 search($RAWdata, $conf['datadir'], $doSearch, $opts, $this->namespace); 403 404 // There may be more than one toc and all of them have to be merged. 405 $data = array(); 406 foreach ($RAWdata as $entry) 407 { 408 $tmpData = p_get_metadata($entry['id'], 'sitetoc siteexportTOC'); 409 410 if (is_array($tmpData)) 411 { 412 $data = array_merge($data, $tmpData); 413 } 414 } 415 } 416 417 $this->functions->debug->message("Checking for Cache after lookup of pages", null, 2); 418 if (!$overrideCache && $this->filewriter->hasValidCacheFile($_REQUEST, $data)) 419 { 420 return array(); 421 } 422 423 $this->functions->debug->message("Exporting the following sites: ", $data, 2); 424 return $data; 425 } 426 427 private function __get_siteexport_list_and_init_tocs($NS, $isRedirected = false) { 428 429 // Clean up if not redirected 430 if (!$isRedirected && !$this->__removeOldZip()) { 431 $this->functions->debug->runtimeException("Can't remove old files."); 432 return false; 433 } 434 435 $data = $this->__get_siteexport_list($NS, $isRedirected); 436 if ($isRedirected || empty($data)) 437 { 438 // if we have been redirected, simply return the data 439 $this->functions->debug->message("List is empty I guess. Used NS: '{$NS}' ", null, 1); 440 return $data; 441 } 442 443 // Create Eclipse Documentation Pages - TOC.xml, Context.xml 444 if (!empty($_REQUEST['absolutePath'])) $this->namespace = ""; 445// $this->__removeOldZip( $this->functions->settings->eclipseZipFile ); 446 447 if (!empty($_REQUEST['eclipseDocZip'])) 448 { 449 $toc = new siteexport_toc($this->functions, $NS); 450 $this->functions->debug->message("Generating eclipseDocZip", null, 2); 451 $this->filewriter->__moveDataToZip($toc->__getTOCXML($data), 'toc.xml'); 452 $this->filewriter->__moveDataToZip($toc->__getContextXML($data), 'context.xml'); 453 } else if (!empty($_REQUEST['JavaHelpDocZip'])) 454 { 455 $toc = new siteexport_javahelp($this->functions, $this->filewriter, $NS); 456 $toc->createTOCFiles($data); 457 458/* $toc = new siteexport_toc($this->functions); 459 list($tocData, $mapData) = $toc->__getJavaHelpTOCXML($data); 460 $this->functions->debug->message("Generating JavaHelpDocZip", null, 2); 461 $this->filewriter->__moveDataToZip($tocData, 'toc.xml'); 462 $this->filewriter->__moveDataToZip($mapData, 'map.xml'); 463*/ } 464 465 return $data; 466 } 467 468 /** 469 * Add page with ID to the package 470 **/ 471 private function __siteexport_add_site($ID) { 472 global $conf, $CURRENT_ID, $CURRENT_PARENT; 473 474 // Which is the current ID? 475 $CURRENT_ID = $ID; 476 477 $this->functions->debug->message("========================================", null, 2); 478 $this->functions->debug->message("Adding Site: '$ID'", null, 2); 479 $this->functions->debug->message("----------------------------------------", $_REQUEST, 2); 480 481 $request = $this->functions->settings->additionalParameters; 482 unset($request['diPlu']); // This will not be needed for the first request. 483 unset($request['diInv']); // This will not be needed for the first request. 484 485 // say, what to export and Build URL 486 // 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 487 488 $do = (intval($_REQUEST['exportbody'] ?? 0) == 1 ? (empty($_REQUEST['renderer'] ?? '') ? $conf['renderer_xhtml'] : $_REQUEST['renderer']) : ''); 489 490 if ($do == 'pdf' && $this->filewriter->canDoPDF()) 491 { 492 $do = 'export_siteexport_pdf'; 493 $_REQUEST['origRenderer'] = (empty($_REQUEST['renderer']) ? $conf['renderer_xhtml'] : $_REQUEST['renderer']); 494 } else if ($_REQUEST['renderer'] ?? '' == 'dw2pdf') { 495 $do = 'pdf'; 496 } 497 498 $do = ($do == $conf['renderer_xhtml'] && intval($_REQUEST['exportbody']) != 1) ? '' : 'export_' . $do; 499 500 if ($do != 'export_' && !empty($do)) 501 { 502 $request['do'] = $do; 503 } 504 505 // set Template 506 if (!empty($_REQUEST['template'])) { 507 $request['template'] = $_REQUEST['template']; 508 } 509 510 $this->functions->debug->message("REQUEST for add_site:", $request, 2); 511 512 $ID = $this->functions->cleanID($ID); 513 $url = $this->functions->wl($ID, $request, true, '&'); 514 515 // Parse URI PATH and add "html" 516 $CURRENT_PARENT = $fileName = $this->functions->getSiteName($ID, true); 517 $this->functions->debug->message("Filename could be:", $fileName, 2); 518 519 $this->fileChecked[$url] = $fileName; // 2010-09-03 - One URL to one FileName 520 $this->functions->settings->depth = str_repeat('../', count(explode('/', $fileName))-1); 521 522 // fetch URL and save it in temp file 523 $tmpFile = $this->__getHTTPFile($url); 524 if ( $tmpFile === false ) { 525 $this->functions->debug->runtimeException("Creating temporary download file failed for '$url'. See log for more information."); 526 return false; 527 } 528 529 $dirname = dirname($fileName); 530 // If a Filename was given that does not comply to the original name, use this one! 531 if ( $this->filewriter->canDoPDF() ) { 532 533 $this->functions->debug->message("Will replace old filename '{$fileName}' with {$ID}", null, 1); 534 $extension = explode('.', $fileName); 535 $extension = array_pop($extension); 536 537 // 2014-04-29 added cleanID to ensure that links are generated consistently when using [[this>...]] or another local, relativ linking 538 $fileName = $dirname . '/' . $this->functions->cleanID($this->functions->getSiteTitle($ID)) . '.' . $extension; 539 } else if ( !empty($tmpFile[1]) /*&& !strstr($DATA[2], $tmpFile[1])*/ ) { // 2017-11-30: $DATA is never defined 540 541 $this->functions->debug->message("Will replace old filename '{$fileName}' with {$dirname}/{$tmpFile[1]}", null, 1); 542 $fileName = $dirname . '/' . $tmpFile[1]; 543 } 544 545 // Add to zip 546 $this->fileChecked[$url] = $fileName; 547 $status = $this->filewriter->__addFileToZip($tmpFile[0], $fileName); 548 if (@unlink($tmpFile[0]) === false) { 549 $this->functions->debug->message("Could not remove temporary file", $tmpFile[0], 1); 550 } 551 552 return $status; 553 } 554 555 /** 556 * Download the file via HTTP URL + recurse if this is not an image 557 * The file will be saved as temporary file. The filename is the result. 558 **/ 559 private function __getHTTPFile($URL, $RECURSE=false, $newAdditionalParameters=null) { 560 global $conf; 561 562 $EXCLUDE = $this->functions->settings->excludePattern; 563 if ( !empty($EXCLUDE) ) { 564 $PATTERN = "/(" . implode('|', explode(' ', preg_quote($EXCLUDE, '/'))) . ")/i"; 565 566 $this->functions->debug->message("Checking for exclude: ", array( 567 "pattern" => $PATTERN, 568 "file" => $URL, 569 "matches" => preg_match($PATTERN, $URL) ? 'match' : 'no match' 570 ), 2); 571 572 if ( preg_match($PATTERN, $URL) ) { return false; } 573 } 574 575 $http = new HTTPProxy($this->functions); 576 $http->max_bodysize = $conf['fetchsize']; 577 578 // Add additional Params 579 $this->functions->addAdditionalParametersToURL($URL, $newAdditionalParameters); 580 581 $this->functions->debug->message("Fetching URL: '$URL'", null, 2); 582 $getData = $http->get($URL, true); // true == sloopy, get 304 body as well. 583 584 if( $getData === false ) { // || ($http->status != 200 && !$this->functions->settings->ignoreNon200) ) { 585 586 if ( $http->status != 200 && $this->functions->settings->ignoreNon200 ) { 587 $this->functions->debug->message("HTTP status was '{$http->status}' - but I was told to ignore it by the settings.", $URL, 3); 588 return true; 589 } 590 591 $this->functions->debug->message("Sending request failed with error, HTTP status was '{$http->status}'.", $URL, 4); 592 return false; 593 } 594 595 if( empty($getData) ) { 596 $this->functions->debug->message("No data fetched", $URL, 4); 597 return false; 598 } 599 600 $this->functions->debug->message("Headers received", $http->resp_headers, 2); 601 602 if ( !$RECURSE ) { 603 // Parse URI PATH and add "html" 604 $this->functions->debug->message("========================================", null, 1); 605 $this->functions->debug->message("Starting to recurse file '$URL'", null , 1); 606 $this->functions->debug->message("----------------------------------------", null, 1); 607 $this->__getInternalLinks($getData); 608 $this->functions->debug->message("----------------------------------------", null, 1); 609 $this->functions->debug->message("Finished to recurse file '$URL'", null , 1); 610 $this->functions->debug->message("========================================", null, 1); 611 } 612 613 $tmpFile = tempnam($this->functions->settings->tmpDir , 'siteexport__') ?: $this->functions->settings->tmpDir . "siteexport__"; 614 $this->functions->debug->message("Temporary filename", $tmpFile, 1); 615 616 $fp = fopen( $tmpFile, "w"); 617 if(!$fp) { 618 $this->functions->debug->message("Can't open temporary File '$tmpFile'.", null , 4); 619 return false; 620 } 621 622 fwrite($fp,$getData); 623 fclose($fp); 624 625 // plain/text; ... 626 $extension = explode(';', $http->resp_headers['content-type'], 2); 627 $extension = array_shift($extension); 628 $extension = explode('/', $extension, 2); 629 if ( $extension[0] == 'image' && preg_match("/^[a-zA-Z0-9]{3,}$/", $extension[1]) ) { 630 $extension = strtolower($extension[1]); 631 $this->functions->debug->message("Found new image extension:", $extension, 2); 632 } else { 633 unset($extension); 634 } 635 636 return array($tmpFile, preg_replace("/.*?filename=\"?(.*?)\"?;?$/", "$1", $http->resp_headers['content-disposition'] ?? ''), $extension ?? ''); 637 } 638 639 /** 640 * Find internal links in the currently downloaded file. This also matches inside CSS files 641 **/ 642 private function __getInternalLinks(&$DATA) { 643 644 $PATTERN = '(href|src|action)="([^"]*)"'; 645 if (!$this->functions->settings->exportLinkedPages) { 646 // no links or forms 647 $PATTERN = '((?<!<a )href|src|action)="([^"]*)"'; 648 } 649 650 $CALLBACK = array($this, '__fetchAndReplaceLink'); 651 $DATA = preg_replace_callback("/$PATTERN/i", $CALLBACK, $DATA); 652 653 $PATTERNCSS = '(url\s*?)\(([^\)]*)\)'; 654 $DATA = preg_replace_callback("/$PATTERNCSS/i", $CALLBACK, $DATA); 655 656 $PATTERNSRCSET = '(srcset)="([^"]*)"'; 657 $CALLBACK = array($this, '__fetchAndReplaceSrcset'); 658 $DATA = preg_replace_callback("/$PATTERNSRCSET/i", $CALLBACK, $DATA); 659 } 660 661 /** 662 * Support for 'srcset' image atributes 663 */ 664 private function __fetchAndReplaceSrcset($DATA) { 665 666 $this->functions->debug->message("SRCSET", $DATA, 2); 667 if ( strtolower($DATA[1]) != 'srcset' ) { return $DATA[0]; } 668 669 $SRCSETS = array_map( 'trim', explode( ",", $DATA[2] ) ); 670 $NEWSRCSETS = array(); 671 foreach( $SRCSETS as $SRCSET ) { 672 673 // tehre should be no other unencoded spaces in here 674 list($url, $size) = explode( " ", $SRCSET, 2 ); 675 676 $this->functions->debug->message("SRCSET: URL before: '$url'", null, 2); 677 $url = $this->__fetchAndReplaceLink(array( $url, '', $url )); 678 $url = substr( $url, 2, -1 ); 679 $this->functions->debug->message("SRCSET: URL after: '$url'", null, 2); 680 681 $NEWSRCSETS[]=$url . ' ' . $size; 682 } 683 684 return 'srcset="' . implode( ',', $NEWSRCSETS ) . '"'; 685 } 686 687 /** 688 * Deep Fetch and replace of links inside the texts matched by __getInternalLinks 689 **/ 690 private function __fetchAndReplaceLink($DATA) { 691 global $conf, $CURRENT_ID, $CURRENT_PARENT; 692 693 $noDeepReplace = true; 694 $newAdditionalParameters = $this->functions->settings->additionalParameters; 695 $newDepth = $this->functions->settings->depth; 696 $hadBase = false; 697 698 // Clean data[2], remote ' and " 699 $DATA[2] = preg_replace("/^\s*?['\"]?(.*?)['\"]?\s*?$/", '\1', trim($DATA[2])); 700 701 $this->functions->debug->message("Starting Link Replacement", array('data' => $DATA, 'additional Params' => $newAdditionalParameters, 'newDepth' => $newDepth, 'currentID' => $CURRENT_ID, 'currentParent' => $CURRENT_PARENT), 2); 702 703 // STEP 1: check for well known links that can be returned 704 if ( $this->__fetchAndReplaceWellKnownLinks( $DATA ) ) { 705 return $this->__rebuildLink($DATA, ""); 706 } 707 708 // 2014-07-21: Origdata before anything else - or it will be missing some things. 709 $ORIGDATA2 = $DATA; 710 // $ORIGDATA2 = $DATA[2]; // 08/10/2010 - this line required a $this->functions->wl which may mess up with the base URL 711 $this->functions->debug->message("OrigDATA is:", $ORIGDATA2, 1); 712 713 // strip all things out 714 // changed Data 715 $PARAMS = @parse_url($DATA[2], PHP_URL_QUERY); 716 $ANCHOR = @parse_url($DATA[2], PHP_URL_FRAGMENT); 717 $DATA[2] = @parse_url($DATA[2], PHP_URL_PATH); 718 719 // 2014-05-12 - fix problem with URLs starting with a ./ or ../ ... they seem to need the current IDs root 720 if (preg_match("#^\.\.?/#", $DATA[2])) { 721 $DATA[2] = getNS($CURRENT_ID) . ':' . $DATA[2]; 722 } 723 724 // 2010-08-25 - fix problem with relative movement in links ( "test/../test2" ) 725 // 2014-06-30 - what? to what will this end relatively? 726 $tmpData2 = ''; 727 while ($tmpData2 != $DATA[2]) { 728 $tmpData2 = $DATA[2]; 729 $DATA[2] = preg_replace("#/(?!\.\.)[^\/]*?/\.\./#", '/', $DATA[2]); 730 } 731 732 $temp = preg_replace("%^" . preg_quote(DOKU_BASE, '%') . "%", "", $DATA[2]); 733 if ($temp != $DATA[2]) { 734 $DATA[2] = $temp; 735 $hadBase = true; // 2010-08-23 Check if there has been a rewrite here that will have to be considered later on 736 } 737 738 $this->functions->debug->message("URL before rewriting option for others than 1", array($DATA, $PARAMS, $hadBase), 1); 739 740 741 // Handle rewrites other than 1 - just for non-lib-files 742 $this->__fetchAndReplaceLinkHandleRewrite( $DATA, $PARAMS ); 743 744 $this->functions->debug->message("URL before rewriting option", array($DATA, $PARAMS), 2); 745 746 // Generate ID 747 $DATA[2] = str_replace('/', ':', $DATA[2]); 748 749 // If Data was empty this must be the same file!; 750 if (empty($DATA[2])) { 751 $DATA[2] = $CURRENT_ID; 752 } 753 754 $ID = $DATA[2]; 755 $MEDIAMATCHER = "#(_media(/|:)|media=|_detail(/|:)|_export(/|:)|do=export_)#i"; // 2010-10-23 added "(/|:)" for the ID may not contain slashes anymore 756 $ISMEDIA = preg_match($MEDIAMATCHER, $DATA[2]) === 1; 757 if ($ISMEDIA !== false && $conf['userewrite'] == 1) { 758 $ID = preg_replace("#^_(detail|media)(/|:)#", "", $ID); 759 } 760 761 $ID = $this->functions->cleanID($DATA[2], null, $ISMEDIA); 762 // $ID = $this->functions->cleanID($DATA[2], null, strstr($DATA[2], 'media') ); // Export anpassung nun weiter unten 763 764 // $IDexists = page_exists($ID); // 08/10/2010 - Not needed. This will be done in the next block. 765 // $this->functions->debug->message("Current ID: '$ID' exists: '" . ($IDexists ? 'true' : 'false') . "' (will be set to 'false' anyway)", null, 1); 766 767 $IDifIDnotExists = $ID; // 08/10/2010 - Save ID - with possible upper cases to preserve them 768 $IDexists = false; 769 770 $this->functions->debug->message("Resolving ID: '$ID'", null, 2); 771 if ($ISMEDIA !== false) { 772 $ID = (new MediaResolver('root'))->resolveId($ID); 773 $IDexists = media_exists($ID); 774 $this->functions->debug->message("Current mediaID to filename: '" . mediaFN($ID) . "'", null, 2); 775 } else { 776 $ID = (new PageResolver('root'))->resolveId($ID); 777 $IDexists = page_exists($ID); 778 $this->functions->debug->message("Current ID to filename: '" . wikiFN($ID) . "'", null, 2); 779 } 780 781 $this->functions->debug->message("Current ID after resolvement: '$ID' the ID does exist: '" . ($IDexists ? 'true' : 'false') . "'", null, 2); 782 // $ORIGDATA2 = @parse_url($this->functions->wl($ORIGDATA2, null, true)); // What was the next 2 line for? It did mess up with links from {{jdoc>}} 783 // $this->functions->debug->message("OrigData ID after parse:", $ORIGDATA2, 1); // 08/10/2010 - The lines are obsolete when the $ORIGDATA2 = $DATA. $ORIGDATA is only for fallback 784 785 // 08/10/2010 - If the ID does not exist, we may have a problem here with upper cases - they will all be lower by now! 786 if (!$IDexists) { 787 $ID = $IDifIDnotExists; // there may have been presevered Upper cases. We will need them! 788 } 789 790 // $this->functions->cleanID($DATA[2], null, strstr($DATA[2], 'media') || strstr($DATA[2], 'export') ); 791 if (substr($ID, -1) == ':' || empty($ID)) $ID .= $conf['start']; 792 793 // Generate Download URL 794 // $PARAMS = trim(str_replace('&', '&', $PARAMS)); 795 $PARAMS = trim($PARAMS); 796 $this->functions->removeWikiVariables($PARAMS, false, true); 797 798 $url = $this->functions->wl($ID, null, true, null, null, true, $hadBase) . (!empty($ANCHOR) ? '#' . $ANCHOR : '') . (!empty($PARAMS) ? '?' . $PARAMS : ''); 799 $this->functions->debug->message("URL from ID: '$url'", null, 2); 800 801 // Parse URI PATH and add "html" 802 $uri = @parse_url($url); 803 $DATA[2] = $uri['path']; 804 805 $this->functions->debug->message("DATA after parsing.", $DATA, 2); 806 807 // Second Rewrite for UseRewrite = 2 808 if ($conf['userewrite'] == 2 && preg_match("%((/lib/exe/(fetch|detail|indexer)|feed|doku)\.php)/?(.*?)$%", $DATA[2], $matches)) { 809 810 // The actual file in lib 811 $DATA[2] = $matches[1]; 812 $PARAMS .= '&' . (in_array($matches[3], array('fetch', 'detail')) ? 'media' : 'id') . '=' . cleanID(str_replace('/', ':', $matches[4])); 813 814 $this->functions->debug->message("DATA after second rewrite with UseRewrite = 2", array($DATA, $matches, $PARAMS), 1); 815 } 816 817 $DATA['ANCHOR'] = $ANCHOR; 818 $DATA['PARAMS'] = $PARAMS; 819 $elements = explode('/', $DATA[2]); 820 821 // Main Switch to check the link 822 $result = $this->__fetchAndReplaceLinkMainSwitch( $elements, $DATA, $url, $newAdditionalParameters, $PARAMS, $noDeepReplace, $fileName, $newDepth, $ID, $IDexists ); 823 if ( $result !== null ) { 824 return $result; 825 } 826 827 $this->functions->debug->message("DATA after SWITCH CASE decision", array($DATA, $noDeepReplace, $fileName, $newDepth), 1); 828 829 if ($this->filewriter->canDoPDF()) { 830 $this->functions->addAdditionalParametersToURL($url, $newAdditionalParameters); 831 $DATA[2] = $url; 832 unset($DATA['PARAMS']); 833 $url = $this->__rebuildLink($DATA, ''); 834 835 $this->functions->debug->message("Creating PDF with URL '$url'", null, 2); 836 837 return $url; 838 } 839 840 // Finalize 841 return $this->__fetchAndReplaceLinkFinish( $DATA, $url, $noDeepReplace, $newAdditionalParameters, $ORIGDATA2, $newDepth, $IDexists, $fileName ); 842 } 843 844 private function __fetchAndReplaceLinkMainSwitch( &$elements, &$DATA, &$url, &$newAdditionalParameters, &$PARAMS, &$noDeepReplace, &$fileName, &$newDepth, &$ID, &$IDexists ) { 845 switch (array_pop($elements)) { 846 // CSS Extra Handling with extra rewrites 847 case 'css.php' : // $DATA[2] .= ( !$this->functions->settings->addParams || empty($PARAMS) ? '' : '.' . $this->functions->cleanID(preg_replace("/(=|\?|&)/", ".", $PARAMS))) . '.css'; 848 $DATA[2] .= '.' . $this->functions->cleanID(preg_replace("/(=|\?|&)/", ".", $PARAMS)) . '.css'; // allways put parameters behind 849 // No paramters needed since they are rewritten. 850 $DATA['PARAMS'] = ""; 851 $noDeepReplace = false; 852 $fileName = $this->functions->getSiteName($ID, true); 853 854 // NewDepth has to be relative to the css file itself ... 855 $newDepth = './' . str_repeat('../', count(explode('/', $fileName))-1); // it is an ID at this point. 856 $newAdditionalParameters['do'] = 'siteexport'; 857 858 $this->functions->debug->message("This is CSS file", array($DATA, $noDeepReplace, $fileName, $newDepth, $newAdditionalParameters), 2); 859 860 break; 861 case 'jquery.php' : 862 case 'js.php' : // $DATA[2] .= ( !$this->functions->settings->addParams || empty($PARAMS) ? '' : '.' . $this->functions->cleanID(preg_replace("/(=|\?|&)/", ".", $PARAMS))) . '.js'; 863 $DATA[2] .= '.t.' . $this->functions->cleanID($_REQUEST['template']) . '.js'; // allways put parameters behind 864 // set Template 865 if (!empty($_REQUEST['template'])) { 866 $url .= (strstr($url, '?') ? '&' : '?') . 'template=' . $_REQUEST['template']; 867 } 868 // No paramters needed since they are rewritten. 869 $DATA['PARAMS'] = ""; 870 $newAdditionalParameters['do'] = 'siteexport'; 871 872 $this->functions->debug->message("This is JS file", array($DATA, $url, $newAdditionalParameters), 2); 873 874 break; 875 // Detail Handling with extra Rewrites if Paramaters are available - otherwise this is just the fetch 876 case 'indexer.php' : 877 $this->functions->debug->message("Skipping indexer", null, 2); 878 return ""; 879 case 'detail.php' : 880 $noDeepReplace = false; 881 882 $this->__getParamsAndDataRewritten($DATA, $PARAMS, 'media'); 883 $ID = $this->functions->cleanID(str_replace('/', ':', $DATA[2]), null, strstr($DATA[2], 'media')); 884 $fileName = $this->functions->getSiteName($ID, true); // 2010-09-03 - rewrite with override enabled 885 886 $newDepth = str_repeat('../', count(explode('/', $fileName))-1); 887 $this->__rebuildDataForNormalFiles($DATA, $PARAMS); 888 $DATA[2] .= '.detail.html'; 889 890 $this->functions->debug->message("This is detail.php file with addParams", array($DATA, $ID, $fileName, $newDepth, $newAdditionalParameters), 2); 891 break; 892 case 'doku.php' : 893 894 $noDeepReplace = false; 895 $this->__getParamsAndDataRewritten($DATA, $PARAMS, 'id'); 896 $ID = $this->functions->cleanID($DATA[2], null, strstr($DATA[2], 'id')); 897 898 $this->functions->debug->message("Current ID to filename (doku.php): '" . wikiFN($ID) . "'", null, 2); 899 900 $fileName = $this->functions->getSiteName($ID); // 2010-09-03 - rewrite with override enabled 901 902 $newDepth = str_repeat('../', count(explode('/', $fileName))-1); 903 $this->__rebuildDataForNormalFiles($DATA, $PARAMS); 904 $DATA2Name = explode('/', $fileName); 905 $DATA[2] .= '.' . array_pop($DATA2Name); 906 907 $this->functions->debug->message("This is doku.php file with addParams", array($DATA, $ID, $fileName, $newDepth, $newAdditionalParameters), 2); 908 return $this->__rebuildLink($DATA); 909 910 // Fetch Handling for media - rewriting everything 911 case 'fetch.php': 912 $this->__getParamsAndDataRewritten($DATA, $PARAMS, 'media'); 913 914 $DATA[2] = str_replace('/', ':', $DATA[2]); 915 $ID = $this->functions->cleanID($DATA[2], null, strstr($DATA[2], 'media')); 916 $ID = (new MediaResolver('root'))->resolveId($ID); 917 $IDexists = media_exists($ID); 918 919 $DATA[2] = $this->functions->wl($ID, null, null, null, $IDexists, true); 920 $this->__rebuildDataForNormalFiles($DATA, $PARAMS); 921 922 $DATA['PARAMS'] = ""; 923 $newAdditionalParameters = array(); 924 925 $this->functions->debug->message("This is fetch.php file", array($DATA, $ID, $PARAMS), 2); 926 break; 927 928 // default Handling for Pages 929 case 'feed.php': 930 return ""; // Ignore. Has no sense to export. 931 default: 932 if (preg_match("%" . preg_quote(DOKU_BASE, '%') . "_detail/%", $DATA[2])) { 933 934 // GET ID Param from origdata2 935 preg_match("#id=(.*?)(&|\")#i", $DATA[0], $backlinkID); 936 $this->__rebuildDataForNormalFiles($DATA, $PARAMS); 937 938 $fileIDPart = isset($backlinkID[1]) && !empty($backlinkID[1]) ? $this->functions->cleanID(urldecode($backlinkID[1])) : 'detail'; 939 940 $ID = preg_replace("#^_detail(/|:)#", "", $ID); 941 $DATA[2] .= ':' . $fileIDPart . '.' . $this->functions->settings->fileType; // add namespace and subpage for back button and add filetype 942 943 $noDeepReplace = false; 944 $fileName = $this->functions->shortenName($DATA[2]); 945 $newDepth = str_repeat('../', count(explode('/', $fileName))-1); 946 $url .= (strstr($url, '?') ? '&' : '?') . 'id=' . $fileIDPart; // add id-part to URL for backlinks 947 948 $DATA['PARAMS'] = ""; 949 950 $this->functions->debug->message("This is something with '_detail' file", array($DATA, $backlinkID, $newDepth, $url, $ID), 2); 951 } else if (preg_match("%" . preg_quote(DOKU_BASE, '%') . "_export/(.*?)/%", $DATA[2], $fileType)) { 952 953 // Fixes multiple codeblocks in one file 954 $this->__rebuildDataForNormalFiles($DATA, $PARAMS); 955 956 // add the Params no matter what they are. This is export. We don't mess with other files 957 // adding the "/" fixes the usage of multiple codeblocks in the same namespace 958 $DATA[2] .= (empty($PARAMS) ? '' : '/' . $PARAMS) . '.' . $fileType[1]; 959 960 $DATA['PARAMS'] = ""; 961 $this->functions->debug->message("This is something with '_export' file", $DATA, 2); 962 963 } else if ($IDexists) { // 08/10/2010 - was page_exists($ID) - but this should do as well. 964 // If this is a page ... skip it! 965 $DATA[2] .= (!$this->functions->settings->addParams || empty($PARAMS) ? '' : '.' . $this->functions->cleanID(preg_replace("/(=|\?|&)/", ".", $PARAMS))) . '.' . $this->functions->settings->fileType; 966 967 $DATA[2] = $this->functions->shortenName($DATA[2]); 968 969 // If Parameters are to be included in the filename - they must not be added twice 970 if ($this->functions->settings->addParams) $DATA['PARAMS'] = ""; 971 972 $this->functions->debug->message("This page really exists", $DATA, 1); 973 974 return $this->__rebuildLink($DATA, null, $ID); 975 } else { 976 $this->__rebuildDataForNormalFiles($DATA, $PARAMS, true); 977 $newAdditionalParameters = null; // 2014-06-27 - when using the "normal" files way we will not need any additional stuff. 978 // This would make problems with e.g. ditaa plugin 979 } 980 981 unset($newAdditionalParameters['diPlu']); 982 } 983 984 return null; 985 } 986 987 private function __fetchAndReplaceLinkFinish( $DATA, $url, $noDeepReplace, $newAdditionalParameters, $ORIGDATA2, $newDepth, $IDexists, $fileName ) { 988 global $conf, $CURRENT_ID, $CURRENT_PARENT; 989 990 // Create Name to save the file at 991 $DATA[2] = str_replace(':', '_', $DATA[2]); 992 $DATA[2] = $this->functions->shortenName($DATA[2]); 993 994 995 // File already loaded? 996 // 2010-10-23 - changes in_array from DATA[2] to $url - to check real URLs, the DATA[2] file will be checked with fileExistsInZip 997 if (in_array($url, array_keys($this->fileChecked))) { 998 $DATA[2] = $this->fileChecked[$url]; 999 $this->functions->debug->message("File has been checked before.", array($DATA, $url), 2); 1000 return $this->__rebuildLink($DATA); 1001 } 1002 1003 // 2010-09-03 - second check if the file is in the ZIP already. 1004 if ($this->filewriter->fileExistsInZip($DATA[2])) { 1005 $this->functions->debug->message("File with DATA exists in ZIP.", $DATA, 3); 1006 return $this->__rebuildLink($DATA); 1007 } 1008 1009 // 2010-10-23 - What if this is a fetch.php? than we produced an error. 1010 // $this->fileChecked[] = $DATA[2]; 1011 1012 // get tempFile and save it 1013 $origDepth = $this->functions->settings->depth; 1014 $this->functions->settings->depth = $newDepth; 1015 1016 $tmpID = $CURRENT_ID; 1017 $tmpParent = $CURRENT_PARENT; 1018 1019 $CURRENT_PARENT = $fileName; 1020 $this->functions->debug->message("Going to get the file", array($url, $noDeepReplace, $newAdditionalParameters), 2); 1021 $tmpFile = $this->__getHTTPFile($url, $noDeepReplace, $newAdditionalParameters); 1022 $this->functions->debug->message("The getHTTPFile result is still empty", $tmpFile === false ? 'YES' : 'NO', 2); 1023 1024 $CURRENT_PARENT = $tmpParent; 1025 $CURRENT_ID = $tmpID; 1026 $this->functions->settings->depth = $origDepth; // 2010-09-03 - Reset depth at the very end 1027 1028 if ($tmpFile === false) { 1029 // Keep an potentially extra link intact 1030 1031 $this->functions->debug->message("The fetched file '$url' is 'false'", null, 3); 1032 if ($IDexists === false) { 1033 $this->functions->debug->message("The file does not exist, fallback to ORIGDATA", $ORIGDATA2, 2); 1034 $DATA[2] = $this->functions->shortenName($ORIGDATA2[2]); // get Origdata Path 1035 } 1036 1037 $this->fileChecked[$url] = $DATA[2]; // 2010-09-03 - One URL to one FileName 1038 $link = $this->__rebuildLink($DATA); 1039 $this->functions->debug->message("Final Link after empty file from '$url'", null, 2); 1040 1041 return $link; 1042 } 1043 1044 $this->functions->debug->message("The fetched file looks good.", $tmpFile, 2); 1045 $dirname = dirname($DATA[2]); 1046 1047 // If a Filename was given that does not comply to the original name, us this one! 1048 // 2014-02-28 But only if we are on PDF Mode. Does this produce any other Problems? 1049 if ( $this->filewriter->canDoPDF() && !empty($tmpFile[1]) && !strstr($DATA[2], $tmpFile[1]) ) { 1050 $DATA[2] = $dirname . '/' . $tmpFile[1]; 1051 $this->functions->debug->message("Changed filename.", $DATA[2], 2); 1052 } 1053 1054 // Custom extension if not set already - 2014-07-02 1055 if ( !empty($tmpFile[2]) && !preg_match("#\.{$tmpFile[2]}$#", $DATA[2]) ) { 1056 $DATA[2] = preg_match("#(\.[^\.]+)$#", $DATA[2]) ? preg_replace("#(\.[^\.]+)$#", '.' . $tmpFile[2], $DATA[2]) : $DATA[2] . '.' . $tmpFile[2]; 1057 $this->functions->debug->message("Added extension provided from Server.", $DATA[2], 2); 1058 } 1059 1060 // Add to zip 1061 $this->fileChecked[$url] = $DATA[2]; // 2010-09-03 - One URL to one FileName 1062 1063 $this->filewriter->__addFileToZip($tmpFile[0], $DATA[2]); 1064 if (@unlink($tmpFile[0]) === false) { 1065 $this->functions->debug->message("Could not delete temporary file.", $tmpFile[0], 2); 1066 } 1067 1068 $newURL = $this->__rebuildLink($DATA); 1069 $this->functions->debug->message("Returning final Link to document: '$newURL'", null, 2); 1070 1071 return $newURL; 1072 } 1073 1074 private function __fetchAndReplaceWellKnownLinks( $DATA ) { 1075 // $DATA[2] = urldecode($DATA[2]); // Leads to problems because it does not re-encode the url 1076 // External and mailto links 1077 if (preg_match("%^(https?://|mailto:|javascript:|data:)%", $DATA[2])) { 1078 $this->functions->debug->message("Don't like http, mailto, data or javascript links here", null, 1); 1079 return true; 1080 } 1081 //if ( preg_match("%^(https?://|mailto:|" . DOKU_BASE . "/_export/)%", $DATA[2]) ) { return $this->__rebuildLink($DATA, ""); } 1082 // External media - this is deep down in the link, so we have to grep it out 1083 if (preg_match("%media=(https?://.*?$)%", $DATA[2], $matches)) { 1084 $DATA[2] = $matches[1]; 1085 $this->functions->debug->message("This is an HTTP like somewhere else", $DATA, 1); 1086 return true; 1087 } 1088 // reference only links won't have to be rewritten 1089 if (preg_match("%^#.*?$%", $DATA[2])) { 1090 $this->functions->debug->message("This is a refercence only", null, 1); 1091 return true; 1092 } 1093 1094 return false; 1095 } 1096 1097 // Handle rewrites other than 1 - just for non-lib-files 1098 private function __fetchAndReplaceLinkHandleRewrite( &$DATA, &$PARAMS ) { 1099 global $conf; 1100 if ( !preg_match('$^(' . DOKU_BASE . ')?lib/$', $DATA[2]) ) { 1101 $this->functions->debug->message("Did not match '$^(" . DOKU_BASE . ")?lib/$' userewrite == {$conf['userewrite']}", null, 2); 1102 if ( $conf['userewrite'] == 2 ) { 1103 $DATA[2] = $this->__getInternalRewriteURL($DATA[2]); 1104 } elseif ( $conf['userewrite'] == 0 ) { 1105 $this->__getParamsAndDataRewritten($DATA, $PARAMS); 1106 } 1107 } else { 1108 $this->functions->debug->message("This file must be inside lib ...", null, 2); 1109 } 1110 } 1111 1112 /** 1113 * build the new link to be put in place for the donwloaded site 1114 **/ 1115 private function __rebuildLink($DATA, $DEPTH = null, $existingPageID = null) { 1116 global $CURRENT_ID, $CURRENT_PARENT; 1117 1118 // depth is set, skip this one 1119 if (is_null($DEPTH)) $DEPTH = $this->functions->settings->depth; 1120 $DATA[2] .= (!empty($DATA['PARAMS']) && $this->functions->settings->addParams ? '?' . $DATA['PARAMS'] : '') . (!empty($DATA['ANCHOR']) ? '#' . $DATA['ANCHOR'] : ''); 1121 1122 $intermediateURL = $DEPTH . $DATA[2]; 1123 1124 // 2012-06-15 originally has an absolute path ... we might need a relative one if not in our namespace 1125 if (empty($_REQUEST['absolutePath']) && preg_match("#^(\.\./)+#", $intermediateURL)) { 1126 1127 $this->functions->debug->message("OK, this is not to be absolute: ", array($intermediateURL, $CURRENT_PARENT), 1); 1128 $intermediateURL = $this->functions->getRelativeURL($intermediateURL, $CURRENT_PARENT, $existingPageID); 1129 } 1130 1131 $newURL = $DATA[1] == 'url' ? $DATA[1] . '(' . $intermediateURL . ')' : $DATA[1] . '="' . $intermediateURL . '"'; 1132 $this->functions->debug->message("Re-created URL: '$newURL'", $DEPTH, 2); 1133 1134 return $newURL; 1135 } 1136 1137 1138 /** 1139 * remove an old zip file 1140 **/ 1141 private function __removeOldZip($FILENAMEID = null, $checkForMore = true, $reauthenticated = false) { 1142 global $INFO; 1143 global $conf; 1144 1145 $returnValue = true; 1146 1147 if (empty($FILENAMEID)) { 1148 $FILENAMEID = $this->functions->settings->origZipFile; 1149 } 1150 1151 if (!file_exists(mediaFN($FILENAMEID))) { 1152 $returnValue = true; 1153 } else { 1154 1155 require_once(DOKU_INC . 'inc/media.php'); 1156 if (!media_delete($FILENAMEID, $INFO['perm'])) { 1157 1158 if (!$reauthenticated) { 1159 $this->functions->authenticate(); 1160 return $this->__removeOldZip($FILENAMEID, $checkForMore, true); 1161 } 1162 1163 $returnValue = false; 1164 } 1165 } 1166 1167 if ($checkForMore) { 1168 // Try to remove more files. 1169 $ns = getNS($FILENAMEID); 1170 $fn = $this->functions->getSpecialExportFileName(noNS($FILENAMEID), '.+'); 1171 1172 $data = array(); 1173 search($data, $conf['mediadir'], 'search_media', array('pattern' => "/$fn$/i"), $ns); 1174 1175 if (count($data) > 0) { 1176 1177 // 30 Minuten Cache Zeit 1178 $cache = $this->functions->settings->cachetime; 1179 foreach ($data as $media) { 1180 1181 //decide if has to be deleted needed: 1182 if ($media['mtime'] < time()-$cache) { 1183 $this->__removeOldZip($media['id'], false, $reauthenticated); 1184 } 1185 } 1186 } 1187 1188 } 1189 1190 return $returnValue; 1191 } 1192 1193 /** 1194 * if confrewrite is set to internal rewrite, use this function - taken from a DW renderer 1195 **/ 1196 private function __getInternalRewriteURL($url) { 1197 global $conf; 1198 1199 //construct page id from request URI 1200 if ($conf['userewrite'] != 2) { return $url; } 1201 1202 //get the script URL 1203 if ($conf['basedir']) { 1204 $relpath = ''; 1205 $script = $conf['basedir'] . $relpath . basename($_SERVER['SCRIPT_FILENAME']); 1206 } elseif ($_SERVER['DOCUMENT_ROOT'] && $_SERVER['SCRIPT_FILENAME']) { 1207 $script = preg_replace('/^' . preg_quote($_SERVER['DOCUMENT_ROOT'], '/') . '/', '', 1208 $_SERVER['SCRIPT_FILENAME']); 1209 $script = '/' . $script; 1210 } else { 1211 $script = $_SERVER['SCRIPT_NAME']; 1212 } 1213 1214 //clean script and request (fixes a windows problem) 1215 $script = preg_replace('/\/\/+/', '/', $script); 1216 $request = preg_replace('/\/\/+/', '/', $url); 1217 1218 //remove script URL and Querystring to gain the id 1219 $id = $request; 1220 if (preg_match('/^' . preg_quote($script, '/') . '(.*)/', $request, $match)) { 1221 $id = preg_replace('/\?.*/', '', $match[1]); 1222 } 1223 $id = urldecode($id); 1224 //strip leading slashes 1225 $id = preg_replace('!^/+!', '', $id); 1226 1227 return $id; 1228 } 1229 1230 /** 1231 * rewrite parameter calls 1232 **/ 1233 private function __getParamsAndDataRewritten(&$DATA, &$PARAMS, $IDKEY = 'id') { 1234 1235 $PARRAY = explode('&', str_replace('&', '&', $PARAMS)); 1236 $PARAMS = array(); 1237 1238 foreach ($PARRAY as $item) { 1239 list($key, $value) = explode('=', $item, 2); 1240 if (empty($key) || empty($value)) 1241 continue; 1242 1243 if (strtolower(trim($key)) == $IDKEY) { 1244 $DATA[2] = preg_replace("%^" . preg_quote(DOKU_BASE, '%') . "%", "", str_replace(':', '/', $value)); 1245 continue; 1246 } 1247 1248 $PARAMS[] = "$key=$value"; 1249 } 1250 1251 sort($PARAMS); 1252 1253 $PARAMS = implode('&', $PARAMS); 1254 } 1255 1256 /** 1257 * rewrite detail.php calls 1258 **/ 1259 private function __rebuildDataForNormalFiles(&$DATA, &$PARAMS, $addHash = false) { 1260 $PARTS = explode('.', $DATA[2]); 1261 $EXT = ''; 1262 if (count($PARTS) > 1) { 1263 $EXT = '.' . array_pop($PARTS); 1264 } 1265 1266 $internalParams = $PARAMS = preg_replace("/(=|\?|&)/", ".", $PARAMS); 1267 1268 // add anyways - if on overridde 1269 if (!$this->functions->settings->addParams && !empty($PARAMS) && $addHash) { 1270 $internalParams = md5($PARAMS); 1271 } else if (!$this->functions->settings->addParams) { 1272 $internalParams = null; 1273 } 1274 1275 $DATA[2] = implode('.', $PARTS) . (empty($internalParams) ? '' : '.' . $this->functions->cleanID($internalParams)) . ($EXT == '.php' ? '.' . $this->functions->settings->fileType : $EXT); 1276 $DATA[2] = preg_replace("/\.+/", ".", $DATA[2]); 1277 $this->functions->debug->message("Rebuilding Data for normal file.", $DATA[2], 1); 1278 } 1279 1280 /* 1281 * Clean JS and CSS cache files 1282 */ 1283 private function cleanCacheFiles() { 1284 1285 $_SERVER['HTTP_HOST'] = preg_replace("/:?\d+$/", '', $_SERVER['HTTP_HOST']); 1286 $cache = getCacheName('scripts' . $_SERVER['HTTP_HOST'] . '-siteexport-js-' . $_SERVER['SERVER_PORT'], '.js'); 1287 $this->unlinkIfExists($cache); 1288 1289 $tpl = trim(preg_replace('/[^\w-]+/', '', $_REQUEST['template'])); 1290 if ($tpl) 1291 { 1292 $tplinc = DOKU_INC . 'lib/tpl/' . $tpl . '/'; 1293 } else { 1294 $tplinc = DOKU_TPLINC; 1295 } 1296 1297 // The generated script depends on some dynamic options 1298 $cache = getCacheName('styles' . $_SERVER['HTTP_HOST'] . '-siteexport-js-' . $_SERVER['SERVER_PORT'] . DOKU_BASE . $tplinc , '.css'); 1299 $this->unlinkIfExists($cache); 1300 } 1301 1302 /** 1303 * Clear Cache 1304 */ 1305 private function unlinkIfExists($cache) { 1306 if (file_exists($cache) && @unlink($cache) === false) { 1307 $this->functions->debug->message('Could not remove file', $cache, 1 ); 1308 } 1309 1310 if (function_exists('gzopen') && @unlink("{$cache}.gz") === false ) { 1311 $this->functions->debug->message('Could not remove file', $cache . '.gz', 1 ); 1312 } 1313 } 1314} 1315