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_INC . '/inc/search.php'); 19 20require_once(DOKU_PLUGIN . 'siteexport/inc/functions.php'); 21require_once(DOKU_PLUGIN . 'siteexport/inc/httpproxy.php'); 22require_once(DOKU_PLUGIN . 'siteexport/inc/filewriter.php'); 23require_once(DOKU_PLUGIN . 'siteexport/inc/toc.php'); 24require_once(DOKU_PLUGIN . 'siteexport/inc/javahelp.php'); 25 26use dokuwiki\File\PageResolver; 27use dokuwiki\File\MediaResolver; 28 29class action_plugin_siteexport_ajax extends DokuWiki_Action_Plugin 30{ 31 /** 32 * New internal variables for better structure 33 */ 34 private $filewriter = null; 35 public $functions = null; 36 37 // List of files that have already been checked 38 private $fileChecked = array(); 39 40 // Namespace of the page to export 41 private $namespace = ''; 42 43 /** 44 * Register Plugin in DW 45 **/ 46 public function register(Doku_Event_Handler $controller) { 47 $controller->register_hook('AJAX_CALL_UNKNOWN', 'BEFORE', $this, 'ajax_siteexport_provider'); 48 $controller->register_hook('ACTION_ACT_PREPROCESS', 'BEFORE', $this, 'siteexport_action'); 49 } 50 51 /** 52 * AJAX Provider - check what is going to be done 53 * @param $event 54 * @param $args 55 */ 56 public function ajax_siteexport_provider(Doku_Event &$event, $args) { 57 58 // If this is not a siteexport call, ignore it. 59 if (!strstr($event->data, '__siteexport')) 60 { 61 return; 62 } 63 64 $this->__init_functions(true); 65 66 switch ($event->data) { 67 case '__siteexport_getsitelist': $this->ajax_siteexport_getsitelist($event); break; 68 case '__siteexport_addsite': $this->ajax_siteexport_addsite($event); break; 69 case '__siteexport_generateurl': $this->ajax_siteexport_generateurl($event); break; 70 case '__siteexport_aggregate': $this->ajax_siteexport_aggregate($event); break; 71 } 72 } 73 74 /** 75 * Export from a URL - action 76 * @param $event 77 */ 78 public function siteexport_action( Doku_Event &$event ) { 79 global $ID; 80 81 // Check if the 'do' was siteexport 82 $keys = is_array($event->data) ? array_keys($event->data) : null; 83 $command = is_array($keys) ? array_shift($keys) : $event->data; 84 if ( $command != 'siteexport' ) { return false; } 85 $event->data = act_clean($event->data); 86 87 if ( headers_sent() ) { 88 msg("The siteexport function has to be called prior to any header output.", -1); 89 } 90 91 $this->__init_functions(); 92 93 $this->functions->debug->message("========================================", null, 1); 94 $this->functions->debug->message("Starting export from URL call", null, 1); 95 $this->functions->debug->message("----------------------------------------", null, 1); 96 97 $event->preventDefault(); 98 $event->stopPropagation(); 99 100 // Fake security Token if none given 101 if (empty($_REQUEST['sectok'])) { 102 $_REQUEST['sectok'] = $this->functions->getSecurityToken(); 103 } 104 105 // The timer will be used to do redirects if needed to prevent timeouts 106 $starttimer = time(); 107 $timerdiff = $this->getConf('max_execution_time'); 108 109 $data = $this->__get_siteexport_list_and_init_tocs($ID, !empty($_REQUEST['startcounter'])); 110 111 if ($data === false) { 112 header("HTTP/1.0 401 Unauthorized"); 113 print 'Unauthorized'; 114 exit; 115 } 116 117 $counter = 0; 118 119 if ( count($data) == 0 && !$this->functions->settings->hasValidCacheFile ) { 120 exit( "No Data to export" ); 121 } 122 123 foreach ( $data as $site ) { 124 125 if ( intval($site['exists']) == 1 || !isset($site['exists']) ) { 126 127 // Skip over the amount of urls that have been exported already 128 if ( empty($_REQUEST['startcounter']) || $counter >= intval($_REQUEST['startcounter']) ) { 129 $status = $this->__siteexport_add_site($site['id']); 130 131 if ( $status === false ) { 132 $this->functions->debug->message("----------------------------------------", null, 1); 133 $this->functions->debug->message("Errors during export from URL call", null, 1); 134 $this->functions->debug->message("========================================", null, 1); 135 print $this->functions->debug->runtimeErrors; 136 exit(0); // We need to stop 137 } 138 } 139 } 140 141 $counter++; 142 if (time()-$starttimer >= $timerdiff) { 143 $this->functions->debug->message("Will Redirect", null, 1); 144 $this->handleRuntimeErrorOutput(); 145 $this->functions->startRedirctProcess($counter); 146 } 147 } 148 149 $this->functions->debug->message("----------------------------------------", null, 1); 150 $this->functions->debug->message("Finishing export from URL call", null, 1); 151 $this->functions->debug->message("========================================", null, 1); 152 153 $this->cleanCacheFiles(); 154 155 $URL = ml($this->functions->settings->origZipFile, array('cache' => 'nocache', 'siteexport' => $this->functions->settings->pattern, 'sectok' => $this->functions->getSecurityToken()), true, '&'); 156 $this->functions->debug->message("Redirecting to final file", $URL, 2); 157 158 $this->handleRuntimeErrorOutput(); 159 send_redirect($URL); 160 exit(0); // Should not be reached, but anyways 161 } 162 163 private function handleRuntimeErrorOutput() 164 { 165 if (!empty($this->functions->debug->runtimeErrors)) 166 { 167 $this->filewriter->__moveDataToZip($this->functions->debug->runtimeErrors, '_runtime_error/' . time() . '.html'); 168 } 169 } 170 171 public function __init_functions($isAJAX = false) 172 { 173 global $conf; 174 175 $conf['useslash'] = 1; 176 177 $this->functions = new siteexport_functions(true, $isAJAX); 178 $this->filewriter = new siteexport_zipfilewriter($this->functions); 179 180 // Check for PDF Capabilities 181 if ($this->filewriter->canDoPDF()) { 182 $this->functions->settings->fileType = 'pdf'; 183 } 184 } 185 186 /** 187 * Prepares the generated URL for direct download access 188 * Also gives back the parameters for this URL 189 * @param $event init event of the ajax request 190 */ 191 private function ajax_siteexport_prepareURL_and_POSTData(Doku_Event &$event) { 192 193 $event->preventDefault(); 194 $event->stopPropagation(); 195 196 // Retrieve Information for download URL 197 $this->functions->debug->message("Prepared URL and POST from Request:", $_REQUEST, 2); 198 $url = $this->functions->prepare_POSTData($_REQUEST); 199 $combined = $this->functions->urlToPathAndParams($url); 200 list($path, $query) = explode('?', $combined, 2); 201 $return = array($url, $combined, $path, $query); 202 203 $this->functions->debug->message("Prepared URL and POST data:", $return, 2); 204 return $return; 205 } 206 207 /** 208 * generate direct access URL 209 **/ 210 private function ajax_siteexport_generateurl(Doku_Event &$event) { 211 212 global $INPUT; 213 214 list($url, $combined, $path, $POSTData) = $this->ajax_siteexport_prepareURL_and_POSTData($event); 215 216 // WGET Redirects - this is an option for wget only. 217 // 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 218 // Therefore we assume it takes about 5s for each page - that gives the freedom to have anough time for redirect. 219 $maxRedirectNumber = ceil((count($this->__get_siteexport_list($INPUT->str('ns'), true))*5)/$this->getConf('max_execution_time')); 220 $maxRedirect = $maxRedirectNumber > 0 ? '--max-redirect=' . ($maxRedirectNumber+3) . ' ' : ''; 221 $maxRedirs = $maxRedirectNumber > 0 ? '--max-redirs ' . ($maxRedirectNumber+3) . ' ' : ''; 222 223 $this->functions->debug->message("Generating Direct Download URL", $url, 2); 224 225 // If there was a Runtime Exception 226 if (!$this->functions->debug->firstRE()) { 227 $this->functions->debug->message("There have been errors while generating the download URLs.", null, 4); 228 return; 229 } 230 231 $zipFile = explode(":", ($this->getConf('zipfilename'))); 232 $zipFile = array_pop($zipFile); 233 234 echo $url; 235 echo "\n"; 236 echo 'wget ' . $maxRedirect . '--output-document=' . $zipFile . ' --post-data="' . $POSTData . '" ' . wl(cleanID($path), null, true) . ' --http-user=USER --http-passwd=PASSWD'; 237 echo "\n"; 238 echo 'curl -L ' . $maxRedirs . '-o ' . $zipFile . ' -d "' . $POSTData . '" ' . wl(cleanID($path), null, true) . ' --anyauth --user USER:PASSWD'; 239 echo "\n"; 240 241 $this->functions->debug->message("Checking for Cron parameters: ", $combined, 1); 242 243 return; 244 } 245 246 /** 247 * Get List of sites to be exported for AJAX (wrapper) 248 **/ 249 private function ajax_siteexport_getsitelist(Doku_Event &$event) { 250 251 global $INPUT; 252 253 $event->preventDefault(); 254 $event->stopPropagation(); 255 256 $data = $this->__get_siteexport_list_and_init_tocs($INPUT->str('ns')); 257 258 // Important for reconaisance of the session 259 260 if ($data === false) 261 { 262 $this->functions->debug->runtimeException("No data generated. List of Files is 'false'."); 263 return; 264 } 265 266 if (empty($data) && !$this->functions->settings->hasValidCacheFile) 267 { 268 $this->functions->debug->runtimeException("Generated list is empty."); 269 return; 270 } 271 272 // If there was a Runtime Exception 273 if (!$this->functions->debug->firstRE()) 274 { 275 $this->functions->debug->message("There have been errors while generating site list.", null, 4); 276 return; 277 } 278 279 echo "{$this->functions->settings->pattern}\n"; 280 echo $this->functions->downloadURL() . "\n"; 281 foreach ($data as $line) { 282 echo $line['id'] . "\n"; 283 } 284 285 return; 286 } 287 288 private function ajax_siteexport_aggregate(Doku_Event &$event) { 289 290 // Quick preparations for one page only 291 if ($this->filewriter->hasValidCacheFile($_REQUEST)) { 292 $this->functions->debug->message("Had a valid cache file and will use it.", null, 2); 293 print $this->functions->downloadURL(); 294 295 $event->preventDefault(); 296 $event->stopPropagation(); 297 } else { 298 // Then go for it! 299 $this->functions->debug->message("Will create a new cache thing.", null, 2); 300 $this->ajax_siteexport_addsite($event); 301 } 302 303 } 304 305 /** 306 * Add a page to the package (for AJAX calls - Wrapper) 307 **/ 308 private function ajax_siteexport_addsite(Doku_Event &$event) { 309 310 global $INPUT; 311 312 $event->preventDefault(); 313 $event->stopPropagation(); 314 315 $this->functions->debug->message("========================================", null, 1); 316 $this->functions->debug->message("Starting export from AJAX call", null, 1); 317 $this->functions->debug->message("----------------------------------------", null, 1); 318 319 $status = $this->__siteexport_add_site($INPUT->str('site')); 320 if ( $status === false ) { 321 $this->functions->debug->message("----------------------------------------", null, 1); 322 $this->functions->debug->message("Errors during export from AJAX call", null, 1); 323 $this->functions->debug->message("========================================", null, 1); 324 return; 325 } 326 327 $this->functions->debug->message("----------------------------------------", null, 1); 328 $this->functions->debug->message("Finishing export from AJAX call", null, 1); 329 $this->functions->debug->message("========================================", null, 1); 330 331 // Print the download zip-File 332 $this->cleanCacheFiles(); 333 334 // If there was a Runtime Exception 335 if (!$this->functions->debug->firstRE()) { 336 $this->functions->debug->message("There have been errors during the export.", null, 4); 337 return; 338 } 339 340 print $this->functions->downloadURL(); 341 return; 342 } 343 344 /** 345 * Fetch the list of pages to be exported 346 **/ 347 private function __get_siteexport_list($NS, $overrideCache = false) { 348 global $conf, $INPUT; 349 350 $PAGE = ""; 351 $NS = $this->namespace = $this->functions->getNamespaceFromID($NS, $PAGE); 352 $this->functions->debug->message("ROOT Namespace to export from: '{$NS}' / {$this->namespace}", null, 2); 353 354 $depth = $this->getConf('depth'); 355 $query = ''; 356 $doSearch = 'search_allpages'; 357 358 switch ($INPUT->int('depthType')) { 359 case 0: 360 $PAGE = $this->functions->cleanID(str_replace(":", "/", $NS . ':' . $PAGE)); 361 $NS = (new PageResolver($NS))->resolveId($PAGE); 362 $exists = page_exists($NS); 363 364 if ($exists) { 365 $data = array(array('id' => $PAGE)); 366 367 $this->functions->debug->message("Checking for Cache, depthType:0", null, 2); 368 if (!$overrideCache && $this->filewriter->hasValidCacheFile($_REQUEST, $data)) 369 { 370 return array(); 371 } 372 373 $this->functions->debug->message("Exporting the following sites: ", $data, 2); 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, $INPUT; 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 = $INPUT->int('exportbody', 0, true) == 1 ? $INPUT->str('renderer', $conf['renderer_xhtml'], true) : ''; 489 490 if ($do == 'pdf' && $this->filewriter->canDoPDF()) 491 { 492 $do = 'export_siteexport_pdf'; 493 $INPUT->set('origRenderer', $INPUT->str('renderer', $conf['renderer_xhtml'], true)); 494 } else if ($INPUT->str('renderer') == 'dw2pdf') { 495 $do = 'pdf'; 496 } 497 498 $do = ($do == $conf['renderer_xhtml'] && $INPUT->int('exportbody', 0, true) != 1) ? '' : 'export_' . $do; 499 500 if ($do != 'export_' && !empty($do) ) 501 { 502 $request['do'] = $do; 503 } 504 505 // set Template 506 if (!empty($INPUT->str('template'))) { 507 $request['template'] = $INPUT->str('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 // Match CSS url(...) but avoid plain text like "API-URL (foo)" 654 $PATTERNCSS = '(?<![A-Za-z0-9_-])(url\s*?)\(([^\)]*)\)'; 655 $DATA = preg_replace_callback("/$PATTERNCSS/i", $CALLBACK, $DATA); 656 657 $PATTERNSRCSET = '(srcset)="([^"]*)"'; 658 $CALLBACK = array($this, '__fetchAndReplaceSrcset'); 659 $DATA = preg_replace_callback("/$PATTERNSRCSET/i", $CALLBACK, $DATA); 660 } 661 662 /** 663 * Support for 'srcset' image atributes 664 */ 665 private function __fetchAndReplaceSrcset($DATA) { 666 667 $this->functions->debug->message("SRCSET", $DATA, 2); 668 if ( strtolower($DATA[1]) != 'srcset' ) { return $DATA[0]; } 669 670 $SRCSETS = array_map( 'trim', explode( ",", $DATA[2] ) ); 671 $NEWSRCSETS = array(); 672 foreach( $SRCSETS as $SRCSET ) { 673 674 // tehre should be no other unencoded spaces in here 675 list($url, $size) = explode( " ", $SRCSET, 2 ); 676 677 $this->functions->debug->message("SRCSET: URL before: '$url'", null, 2); 678 $url = $this->__fetchAndReplaceLink(array( $url, '', $url )); 679 $url = substr( $url, 2, -1 ); 680 $this->functions->debug->message("SRCSET: URL after: '$url'", null, 2); 681 682 $NEWSRCSETS[]=$url . ' ' . $size; 683 } 684 685 return 'srcset="' . implode( ',', $NEWSRCSETS ) . '"'; 686 } 687 688 /** 689 * Deep Fetch and replace of links inside the texts matched by __getInternalLinks 690 **/ 691 private function __fetchAndReplaceLink($DATA) { 692 global $conf, $CURRENT_ID, $CURRENT_PARENT; 693 694 $noDeepReplace = true; 695 $newAdditionalParameters = $this->functions->settings->additionalParameters; 696 $newDepth = $this->functions->settings->depth; 697 $hadBase = false; 698 699 // Clean data[2], remote ' and " 700 $DATA[2] = preg_replace("/^\s*?['\"]?(.*?)['\"]?\s*?$/", '\1', trim($DATA[2])); 701 702 $this->functions->debug->message("Starting Link Replacement", array('data' => $DATA, 'additional Params' => $newAdditionalParameters, 'newDepth' => $newDepth, 'currentID' => $CURRENT_ID, 'currentParent' => $CURRENT_PARENT), 2); 703 704 // STEP 1: check for well known links that can be returned 705 if ( $this->__fetchAndReplaceWellKnownLinks( $DATA ) ) { 706 return $this->__rebuildLink($DATA, ""); 707 } 708 709 // 2014-07-21: Origdata before anything else - or it will be missing some things. 710 $ORIGDATA2 = $DATA; 711 // $ORIGDATA2 = $DATA[2]; // 08/10/2010 - this line required a $this->functions->wl which may mess up with the base URL 712 $this->functions->debug->message("OrigDATA is:", $ORIGDATA2, 1); 713 714 // strip all things out 715 // changed Data 716 $PARAMS = @parse_url($DATA[2], PHP_URL_QUERY); 717 $ANCHOR = @parse_url($DATA[2], PHP_URL_FRAGMENT); 718 $DATA[2] = @parse_url($DATA[2], PHP_URL_PATH); 719 720 // 2014-05-12 - fix problem with URLs starting with a ./ or ../ ... they seem to need the current IDs root 721 if (preg_match("#^\.\.?/#", $DATA[2])) { 722 $DATA[2] = getNS($CURRENT_ID) . ':' . $DATA[2]; 723 } 724 725 // 2010-08-25 - fix problem with relative movement in links ( "test/../test2" ) 726 // 2014-06-30 - what? to what will this end relatively? 727 $tmpData2 = ''; 728 while ($tmpData2 != $DATA[2]) { 729 $tmpData2 = $DATA[2]; 730 $DATA[2] = preg_replace("#/(?!\.\.)[^\/]*?/\.\./#", '/', $DATA[2]); 731 } 732 733 $temp = preg_replace("%^" . preg_quote(DOKU_BASE, '%') . "%", "", $DATA[2]); 734 if ($temp != $DATA[2]) { 735 $DATA[2] = $temp; 736 $hadBase = true; // 2010-08-23 Check if there has been a rewrite here that will have to be considered later on 737 } 738 739 $this->functions->debug->message("URL before rewriting option for others than 1", array($DATA, $PARAMS, $hadBase), 1); 740 741 742 // Handle rewrites other than 1 - just for non-lib-files 743 $this->__fetchAndReplaceLinkHandleRewrite( $DATA, $PARAMS ); 744 745 $this->functions->debug->message("URL before rewriting option", array($DATA, $PARAMS), 2); 746 747 // Generate ID 748 $DATA[2] = str_replace('/', ':', $DATA[2]); 749 750 // If Data was empty this must be the same file!; 751 if (empty($DATA[2])) { 752 $DATA[2] = $CURRENT_ID; 753 } 754 755 $ID = $DATA[2]; 756 $MEDIAMATCHER = "#(_media(/|:)|media=|_detail(/|:)|_export(/|:)|do=export_)#i"; // 2010-10-23 added "(/|:)" for the ID may not contain slashes anymore 757 $ISMEDIA = preg_match($MEDIAMATCHER, $DATA[2]) === 1; 758 if ($ISMEDIA !== false && $conf['userewrite'] == 1) { 759 $ID = preg_replace("#^_(detail|media)(/|:)#", "", $ID); 760 } 761 762 $ID = $this->functions->cleanID($DATA[2], null, $ISMEDIA); 763 // $ID = $this->functions->cleanID($DATA[2], null, strstr($DATA[2], 'media') ); // Export anpassung nun weiter unten 764 765 // $IDexists = page_exists($ID); // 08/10/2010 - Not needed. This will be done in the next block. 766 // $this->functions->debug->message("Current ID: '$ID' exists: '" . ($IDexists ? 'true' : 'false') . "' (will be set to 'false' anyway)", null, 1); 767 768 $IDifIDnotExists = $ID; // 08/10/2010 - Save ID - with possible upper cases to preserve them 769 $IDexists = false; 770 771 $this->functions->debug->message("Resolving ID: '$ID'", null, 2); 772 if ($ISMEDIA !== false) { 773 $ID = (new MediaResolver('root'))->resolveId($ID); 774 $IDexists = media_exists($ID); 775 $this->functions->debug->message("Current mediaID to filename: '" . mediaFN($ID) . "'", null, 2); 776 } else { 777 $ID = (new PageResolver('root'))->resolveId($ID); 778 $IDexists = page_exists($ID); 779 $this->functions->debug->message("Current ID to filename: '" . wikiFN($ID) . "'", null, 2); 780 } 781 782 $this->functions->debug->message("Current ID after resolvement: '$ID' the ID does exist: '" . ($IDexists ? 'true' : 'false') . "'", null, 2); 783 // $ORIGDATA2 = @parse_url($this->functions->wl($ORIGDATA2, null, true)); // What was the next 2 line for? It did mess up with links from {{jdoc>}} 784 // $this->functions->debug->message("OrigData ID after parse:", $ORIGDATA2, 1); // 08/10/2010 - The lines are obsolete when the $ORIGDATA2 = $DATA. $ORIGDATA is only for fallback 785 786 // 08/10/2010 - If the ID does not exist, we may have a problem here with upper cases - they will all be lower by now! 787 if (!$IDexists) { 788 $ID = $IDifIDnotExists; // there may have been presevered Upper cases. We will need them! 789 } 790 791 // $this->functions->cleanID($DATA[2], null, strstr($DATA[2], 'media') || strstr($DATA[2], 'export') ); 792 if (substr($ID, -1) == ':' || empty($ID)) $ID .= $conf['start']; 793 794 // Generate Download URL 795 // $PARAMS = trim(str_replace('&', '&', $PARAMS)); 796 $PARAMS = trim($PARAMS); 797 $this->functions->removeWikiVariables($PARAMS, false, true); 798 799 $url = $this->functions->wl($ID, null, true, null, null, true, $hadBase) . (!empty($ANCHOR) ? '#' . $ANCHOR : '') . (!empty($PARAMS) ? '?' . $PARAMS : ''); 800 $this->functions->debug->message("URL from ID: '$url'", null, 2); 801 802 // Parse URI PATH and add "html" 803 $uri = @parse_url($url); 804 $DATA[2] = $uri['path']; 805 806 $this->functions->debug->message("DATA after parsing.", $DATA, 2); 807 808 // Second Rewrite for UseRewrite = 2 809 if ($conf['userewrite'] == 2 && preg_match("%((/lib/exe/(fetch|detail|indexer)|feed|doku)\.php)/?(.*?)$%", $DATA[2], $matches)) { 810 811 // The actual file in lib 812 $DATA[2] = $matches[1]; 813 $PARAMS .= '&' . (in_array($matches[3], array('fetch', 'detail')) ? 'media' : 'id') . '=' . cleanID(str_replace('/', ':', $matches[4])); 814 815 $this->functions->debug->message("DATA after second rewrite with UseRewrite = 2", array($DATA, $matches, $PARAMS), 1); 816 } 817 818 $DATA['ANCHOR'] = $ANCHOR; 819 $DATA['PARAMS'] = $PARAMS; 820 $elements = explode('/', $DATA[2]); 821 822 // Main Switch to check the link 823 $result = $this->__fetchAndReplaceLinkMainSwitch( $elements, $DATA, $url, $newAdditionalParameters, $PARAMS, $noDeepReplace, $fileName, $newDepth, $ID, $IDexists ); 824 if ( $result !== null ) { 825 return $result; 826 } 827 828 $this->functions->debug->message("DATA after SWITCH CASE decision", array($DATA, $noDeepReplace, $fileName, $newDepth), 1); 829 830 if ($this->filewriter->canDoPDF()) { 831 $this->functions->addAdditionalParametersToURL($url, $newAdditionalParameters); 832 $DATA[2] = $url; 833 unset($DATA['PARAMS']); 834 $rebuilt = $this->__rebuildLink($DATA, ''); 835 // __rebuildLink returns a full href|src|action="…" fragment; log the inner URL for readability. 836 if (preg_match('/^\\w+\\s*=\\s*"([^"]*)"/', $rebuilt, $m)) { 837 $this->functions->debug->message("Creating PDF: fetching '{$m[1]}' (replacement {$rebuilt})", null, 2); 838 } else { 839 $this->functions->debug->message("Creating PDF with replacement '$rebuilt'", null, 2); 840 } 841 842 return $rebuilt; 843 } 844 845 // Finalize 846 return $this->__fetchAndReplaceLinkFinish( $DATA, $url, $noDeepReplace, $newAdditionalParameters, $ORIGDATA2, $newDepth, $IDexists, $fileName ); 847 } 848 849 private function __fetchAndReplaceLinkMainSwitch( &$elements, &$DATA, &$url, &$newAdditionalParameters, &$PARAMS, &$noDeepReplace, &$fileName, &$newDepth, &$ID, &$IDexists ) { 850 switch (array_pop($elements)) { 851 // CSS Extra Handling with extra rewrites 852 case 'css.php' : // $DATA[2] .= ( !$this->functions->settings->addParams || empty($PARAMS) ? '' : '.' . $this->functions->cleanID(preg_replace("/(=|\?|&)/", ".", $PARAMS))) . '.css'; 853 $DATA[2] .= '.' . $this->functions->cleanID(preg_replace("/(=|\?|&)/", ".", $PARAMS)) . '.css'; // allways put parameters behind 854 // No paramters needed since they are rewritten. 855 $DATA['PARAMS'] = ""; 856 $noDeepReplace = false; 857 $fileName = $this->functions->getSiteName($ID, true); 858 859 // NewDepth has to be relative to the css file itself ... 860 $newDepth = './' . str_repeat('../', count(explode('/', $fileName))-1); // it is an ID at this point. 861 $newAdditionalParameters['do'] = 'siteexport'; 862 863 $this->functions->debug->message("This is CSS file", array($DATA, $noDeepReplace, $fileName, $newDepth, $newAdditionalParameters), 2); 864 865 break; 866 case 'jquery.php' : 867 case 'js.php' : // $DATA[2] .= ( !$this->functions->settings->addParams || empty($PARAMS) ? '' : '.' . $this->functions->cleanID(preg_replace("/(=|\?|&)/", ".", $PARAMS))) . '.js'; 868 $DATA[2] .= '.t.' . $this->functions->cleanID($_REQUEST['template']) . '.js'; // allways put parameters behind 869 // set Template 870 if (!empty($_REQUEST['template'])) { 871 $url .= (strstr($url, '?') ? '&' : '?') . 'template=' . $_REQUEST['template']; 872 } 873 // No paramters needed since they are rewritten. 874 $DATA['PARAMS'] = ""; 875 $newAdditionalParameters['do'] = 'siteexport'; 876 877 $this->functions->debug->message("This is JS file", array($DATA, $url, $newAdditionalParameters), 2); 878 879 break; 880 // Detail Handling with extra Rewrites if Paramaters are available - otherwise this is just the fetch 881 case 'indexer.php' : 882 $this->functions->debug->message("Skipping indexer", null, 2); 883 return ""; 884 case 'detail.php' : 885 $noDeepReplace = false; 886 887 $this->__getParamsAndDataRewritten($DATA, $PARAMS, 'media'); 888 $ID = $this->functions->cleanID(str_replace('/', ':', $DATA[2]), null, strstr($DATA[2], 'media')); 889 $fileName = $this->functions->getSiteName($ID, true); // 2010-09-03 - rewrite with override enabled 890 891 $newDepth = str_repeat('../', count(explode('/', $fileName))-1); 892 $this->__rebuildDataForNormalFiles($DATA, $PARAMS); 893 $DATA[2] .= '.detail.html'; 894 895 $this->functions->debug->message("This is detail.php file with addParams", array($DATA, $ID, $fileName, $newDepth, $newAdditionalParameters), 2); 896 break; 897 case 'doku.php' : 898 899 $noDeepReplace = false; 900 $this->__getParamsAndDataRewritten($DATA, $PARAMS, 'id'); 901 $ID = $this->functions->cleanID($DATA[2], null, strstr($DATA[2], 'id')); 902 903 $this->functions->debug->message("Current ID to filename (doku.php): '" . wikiFN($ID) . "'", null, 2); 904 905 $fileName = $this->functions->getSiteName($ID); // 2010-09-03 - rewrite with override enabled 906 907 $newDepth = str_repeat('../', count(explode('/', $fileName))-1); 908 $this->__rebuildDataForNormalFiles($DATA, $PARAMS); 909 $DATA2Name = explode('/', $fileName); 910 $DATA[2] .= '.' . array_pop($DATA2Name); 911 912 $this->functions->debug->message("This is doku.php file with addParams", array($DATA, $ID, $fileName, $newDepth, $newAdditionalParameters), 2); 913 return $this->__rebuildLink($DATA); 914 915 // Fetch Handling for media - rewriting everything 916 case 'fetch.php': 917 $this->__getParamsAndDataRewritten($DATA, $PARAMS, 'media'); 918 919 $DATA[2] = str_replace('/', ':', $DATA[2]); 920 $ID = $this->functions->cleanID($DATA[2], null, strstr($DATA[2], 'media')); 921 $ID = (new MediaResolver('root'))->resolveId($ID); 922 $IDexists = media_exists($ID); 923 924 $DATA[2] = $this->functions->wl($ID, null, null, null, $IDexists, true); 925 $this->__rebuildDataForNormalFiles($DATA, $PARAMS); 926 927 $DATA['PARAMS'] = ""; 928 $newAdditionalParameters = array(); 929 930 $this->functions->debug->message("This is fetch.php file", array($DATA, $ID, $PARAMS), 2); 931 break; 932 933 // default Handling for Pages 934 case 'feed.php': 935 return ""; // Ignore. Has no sense to export. 936 default: 937 if (preg_match("%" . preg_quote(DOKU_BASE, '%') . "_detail/%", $DATA[2])) { 938 939 // GET ID Param from origdata2 940 preg_match("#id=(.*?)(&|\")#i", $DATA[0], $backlinkID); 941 $this->__rebuildDataForNormalFiles($DATA, $PARAMS); 942 943 $fileIDPart = isset($backlinkID[1]) && !empty($backlinkID[1]) ? $this->functions->cleanID(urldecode($backlinkID[1])) : 'detail'; 944 945 $ID = preg_replace("#^_detail(/|:)#", "", $ID); 946 $DATA[2] .= ':' . $fileIDPart . '.' . $this->functions->settings->fileType; // add namespace and subpage for back button and add filetype 947 948 $noDeepReplace = false; 949 $fileName = $this->functions->shortenName($DATA[2]); 950 $newDepth = str_repeat('../', count(explode('/', $fileName))-1); 951 $url .= (strstr($url, '?') ? '&' : '?') . 'id=' . $fileIDPart; // add id-part to URL for backlinks 952 953 $DATA['PARAMS'] = ""; 954 955 $this->functions->debug->message("This is something with '_detail' file", array($DATA, $backlinkID, $newDepth, $url, $ID), 2); 956 } else if (preg_match("%" . preg_quote(DOKU_BASE, '%') . "_export/(.*?)/%", $DATA[2], $fileType)) { 957 958 // Fixes multiple codeblocks in one file 959 $this->__rebuildDataForNormalFiles($DATA, $PARAMS); 960 961 // add the Params no matter what they are. This is export. We don't mess with other files 962 // adding the "/" fixes the usage of multiple codeblocks in the same namespace 963 $DATA[2] .= (empty($PARAMS) ? '' : '/' . $PARAMS) . '.' . $fileType[1]; 964 965 $DATA['PARAMS'] = ""; 966 $this->functions->debug->message("This is something with '_export' file", $DATA, 2); 967 968 } else if ($IDexists) { // 08/10/2010 - was page_exists($ID) - but this should do as well. 969 // If this is a page ... skip it! 970 $DATA[2] .= (!$this->functions->settings->addParams || empty($PARAMS) ? '' : '.' . $this->functions->cleanID(preg_replace("/(=|\?|&)/", ".", $PARAMS))) . '.' . $this->functions->settings->fileType; 971 972 $DATA[2] = $this->functions->shortenName($DATA[2]); 973 974 // If Parameters are to be included in the filename - they must not be added twice 975 if ($this->functions->settings->addParams) $DATA['PARAMS'] = ""; 976 977 $this->functions->debug->message("This page really exists", $DATA, 1); 978 979 return $this->__rebuildLink($DATA, null, $ID); 980 } else { 981 $this->__rebuildDataForNormalFiles($DATA, $PARAMS, true); 982 $newAdditionalParameters = null; // 2014-06-27 - when using the "normal" files way we will not need any additional stuff. 983 // This would make problems with e.g. ditaa plugin 984 } 985 986 unset($newAdditionalParameters['diPlu']); 987 } 988 989 return null; 990 } 991 992 private function __fetchAndReplaceLinkFinish( $DATA, $url, $noDeepReplace, $newAdditionalParameters, $ORIGDATA2, $newDepth, $IDexists, $fileName ) { 993 global $conf, $CURRENT_ID, $CURRENT_PARENT; 994 995 // Create Name to save the file at 996 $DATA[2] = str_replace(':', '_', $DATA[2]); 997 $DATA[2] = $this->functions->shortenName($DATA[2]); 998 999 1000 // File already loaded? 1001 // 2010-10-23 - changes in_array from DATA[2] to $url - to check real URLs, the DATA[2] file will be checked with fileExistsInZip 1002 if (in_array($url, array_keys($this->fileChecked))) { 1003 $DATA[2] = $this->fileChecked[$url]; 1004 $this->functions->debug->message("File has been checked before.", array($DATA, $url), 2); 1005 return $this->__rebuildLink($DATA); 1006 } 1007 1008 // 2010-09-03 - second check if the file is in the ZIP already. 1009 if ($this->filewriter->fileExistsInZip($DATA[2])) { 1010 $this->functions->debug->message("File with DATA exists in ZIP.", $DATA, 3); 1011 return $this->__rebuildLink($DATA); 1012 } 1013 1014 // 2010-10-23 - What if this is a fetch.php? than we produced an error. 1015 // $this->fileChecked[] = $DATA[2]; 1016 1017 // get tempFile and save it 1018 $origDepth = $this->functions->settings->depth; 1019 $this->functions->settings->depth = $newDepth; 1020 1021 $tmpID = $CURRENT_ID; 1022 $tmpParent = $CURRENT_PARENT; 1023 1024 $CURRENT_PARENT = $fileName; 1025 $this->functions->debug->message("Going to get the file", array($url, $noDeepReplace, $newAdditionalParameters), 2); 1026 $tmpFile = $this->__getHTTPFile($url, $noDeepReplace, $newAdditionalParameters); 1027 $this->functions->debug->message("The getHTTPFile result is still empty", $tmpFile === false ? 'YES' : 'NO', 2); 1028 1029 $CURRENT_PARENT = $tmpParent; 1030 $CURRENT_ID = $tmpID; 1031 $this->functions->settings->depth = $origDepth; // 2010-09-03 - Reset depth at the very end 1032 1033 if ($tmpFile === false) { 1034 // Keep an potentially extra link intact 1035 1036 $this->functions->debug->message("The fetched file '$url' is 'false'", null, 3); 1037 if ($IDexists === false) { 1038 $this->functions->debug->message("The file does not exist, fallback to ORIGDATA", $ORIGDATA2, 2); 1039 $DATA[2] = $this->functions->shortenName($ORIGDATA2[2]); // get Origdata Path 1040 } 1041 1042 $this->fileChecked[$url] = $DATA[2]; // 2010-09-03 - One URL to one FileName 1043 $link = $this->__rebuildLink($DATA); 1044 $this->functions->debug->message("Final Link after empty file from '$url'", null, 2); 1045 1046 return $link; 1047 } elseif ($tmpFile === true) { 1048 // Non-200 ignored; keep original link intact. 1049 return is_array($ORIGDATA2) && isset($ORIGDATA2[0]) ? $ORIGDATA2[0] : "file_not_found_and_ignored.html"; 1050 } 1051 1052 $this->functions->debug->message("The fetched file looks good.", $tmpFile, 2); 1053 $dirname = dirname($DATA[2]); 1054 1055 // If a Filename was given that does not comply to the original name, us this one! 1056 // 2014-02-28 But only if we are on PDF Mode. Does this produce any other Problems? 1057 if ( $this->filewriter->canDoPDF() && !empty($tmpFile[1]) && !strstr($DATA[2], $tmpFile[1]) ) { 1058 $DATA[2] = $dirname . '/' . $tmpFile[1]; 1059 $this->functions->debug->message("Changed filename.", $DATA[2], 2); 1060 } 1061 1062 // Custom extension if not set already - 2014-07-02 1063 if ( !empty($tmpFile[2]) && !preg_match("#\.{$tmpFile[2]}$#", $DATA[2]) ) { 1064 $DATA[2] = preg_match("#(\.[^\.]+)$#", $DATA[2]) ? preg_replace("#(\.[^\.]+)$#", '.' . $tmpFile[2], $DATA[2]) : $DATA[2] . '.' . $tmpFile[2]; 1065 $this->functions->debug->message("Added extension provided from Server.", $DATA[2], 2); 1066 } 1067 1068 // Add to zip 1069 $this->fileChecked[$url] = $DATA[2]; // 2010-09-03 - One URL to one FileName 1070 1071 $this->filewriter->__addFileToZip($tmpFile[0], $DATA[2]); 1072 if (@unlink($tmpFile[0]) === false) { 1073 $this->functions->debug->message("Could not delete temporary file.", $tmpFile[0], 2); 1074 } 1075 1076 $newURL = $this->__rebuildLink($DATA); 1077 $this->functions->debug->message("Returning final Link to document: '$newURL'", null, 2); 1078 1079 return $newURL; 1080 } 1081 1082 private function __fetchAndReplaceWellKnownLinks( $DATA ) { 1083 // $DATA[2] = urldecode($DATA[2]); // Leads to problems because it does not re-encode the url 1084 // External and mailto links 1085 if (preg_match("%^(https?://|mailto:|javascript:|data:)%", $DATA[2])) { 1086 $this->functions->debug->message("Don't like http, mailto, data or javascript links here", null, 1); 1087 return true; 1088 } 1089 //if ( preg_match("%^(https?://|mailto:|" . DOKU_BASE . "/_export/)%", $DATA[2]) ) { return $this->__rebuildLink($DATA, ""); } 1090 // External media - this is deep down in the link, so we have to grep it out 1091 if (preg_match("%media=(https?://.*?$)%", $DATA[2], $matches)) { 1092 $DATA[2] = $matches[1]; 1093 $this->functions->debug->message("This is an HTTP like somewhere else", $DATA, 1); 1094 return true; 1095 } 1096 // reference only links won't have to be rewritten 1097 if (preg_match("%^#.*?$%", $DATA[2])) { 1098 $this->functions->debug->message("This is a refercence only", null, 1); 1099 return true; 1100 } 1101 1102 return false; 1103 } 1104 1105 // Handle rewrites other than 1 - just for non-lib-files 1106 private function __fetchAndReplaceLinkHandleRewrite( &$DATA, &$PARAMS ) { 1107 global $conf; 1108 if ( !preg_match('$^(' . DOKU_BASE . ')?lib/$', $DATA[2]) ) { 1109 $this->functions->debug->message("Did not match '$^(" . DOKU_BASE . ")?lib/$' userewrite == {$conf['userewrite']}", null, 2); 1110 if ( $conf['userewrite'] == 2 ) { 1111 $DATA[2] = $this->__getInternalRewriteURL($DATA[2]); 1112 } elseif ( $conf['userewrite'] == 0 ) { 1113 $this->__getParamsAndDataRewritten($DATA, $PARAMS); 1114 } 1115 } else { 1116 $this->functions->debug->message("This file must be inside lib ...", null, 2); 1117 } 1118 } 1119 1120 /** 1121 * build the new link to be put in place for the donwloaded site 1122 **/ 1123 private function __rebuildLink($DATA, $DEPTH = null, $existingPageID = null) { 1124 global $CURRENT_ID, $CURRENT_PARENT; 1125 1126 // depth is set, skip this one 1127 if (is_null($DEPTH)) $DEPTH = $this->functions->settings->depth; 1128 $DATA[2] .= (!empty($DATA['PARAMS']) && $this->functions->settings->addParams ? '?' . $DATA['PARAMS'] : '') . (!empty($DATA['ANCHOR']) ? '#' . $DATA['ANCHOR'] : ''); 1129 1130 $intermediateURL = $DEPTH . $DATA[2]; 1131 1132 // 2012-06-15 originally has an absolute path ... we might need a relative one if not in our namespace 1133 if (empty($_REQUEST['absolutePath']) && preg_match("#^(\.\./)+#", $intermediateURL)) { 1134 1135 $this->functions->debug->message("OK, this is not to be absolute: ", array($intermediateURL, $CURRENT_PARENT), 1); 1136 $intermediateURL = $this->functions->getRelativeURL($intermediateURL, $CURRENT_PARENT, $existingPageID); 1137 } 1138 1139 $newURL = $DATA[1] == 'url' ? $DATA[1] . '(' . $intermediateURL . ')' : $DATA[1] . '="' . $intermediateURL . '"'; 1140 $this->functions->debug->message("Re-created URL: '$newURL'", $DEPTH, 2); 1141 1142 return $newURL; 1143 } 1144 1145 1146 /** 1147 * remove an old zip file 1148 **/ 1149 private function __removeOldZip($FILENAMEID = null, $checkForMore = true, $reauthenticated = false) { 1150 global $INFO; 1151 global $conf; 1152 1153 $returnValue = true; 1154 1155 if (empty($FILENAMEID)) { 1156 $FILENAMEID = $this->functions->settings->origZipFile; 1157 } 1158 1159 if (!file_exists(mediaFN($FILENAMEID))) { 1160 $returnValue = true; 1161 } else { 1162 1163 require_once(DOKU_INC . 'inc/media.php'); 1164 if (!media_delete($FILENAMEID, $INFO['perm'])) { 1165 1166 if (!$reauthenticated) { 1167 $this->functions->authenticate(); 1168 return $this->__removeOldZip($FILENAMEID, $checkForMore, true); 1169 } 1170 1171 $returnValue = false; 1172 } 1173 } 1174 1175 if ($checkForMore) { 1176 // Try to remove more files. 1177 $ns = getNS($FILENAMEID); 1178 $fn = $this->functions->getSpecialExportFileName(noNS($FILENAMEID), '.+'); 1179 1180 $data = array(); 1181 search($data, $conf['mediadir'], 'search_media', array('pattern' => "/$fn$/i"), $ns); 1182 1183 if (count($data) > 0) { 1184 1185 // 30 Minuten Cache Zeit 1186 $cache = $this->functions->settings->cachetime; 1187 foreach ($data as $media) { 1188 1189 //decide if has to be deleted needed: 1190 if ($media['mtime'] < time()-$cache) { 1191 $this->__removeOldZip($media['id'], false, $reauthenticated); 1192 } 1193 } 1194 } 1195 1196 } 1197 1198 return $returnValue; 1199 } 1200 1201 /** 1202 * if confrewrite is set to internal rewrite, use this function - taken from a DW renderer 1203 **/ 1204 private function __getInternalRewriteURL($url) { 1205 global $conf; 1206 1207 //construct page id from request URI 1208 if ($conf['userewrite'] != 2) { return $url; } 1209 1210 //get the script URL 1211 if ($conf['basedir']) { 1212 $relpath = ''; 1213 $script = $conf['basedir'] . $relpath . basename($_SERVER['SCRIPT_FILENAME']); 1214 } elseif ($_SERVER['DOCUMENT_ROOT'] && $_SERVER['SCRIPT_FILENAME']) { 1215 $script = preg_replace('/^' . preg_quote($_SERVER['DOCUMENT_ROOT'], '/') . '/', '', 1216 $_SERVER['SCRIPT_FILENAME']); 1217 $script = '/' . $script; 1218 } else { 1219 $script = $_SERVER['SCRIPT_NAME']; 1220 } 1221 1222 //clean script and request (fixes a windows problem) 1223 $script = preg_replace('/\/\/+/', '/', $script); 1224 $request = preg_replace('/\/\/+/', '/', $url); 1225 1226 //remove script URL and Querystring to gain the id 1227 $id = $request; 1228 if (preg_match('/^' . preg_quote($script, '/') . '(.*)/', $request, $match)) { 1229 $id = preg_replace('/\?.*/', '', $match[1]); 1230 } 1231 $id = urldecode($id); 1232 //strip leading slashes 1233 $id = preg_replace('!^/+!', '', $id); 1234 1235 return $id; 1236 } 1237 1238 /** 1239 * rewrite parameter calls 1240 **/ 1241 private function __getParamsAndDataRewritten(&$DATA, &$PARAMS, $IDKEY = 'id') { 1242 1243 if (empty($PARAMS)) 1244 return array(); 1245 1246 $PARRAY = explode('&', str_replace('&', '&', $PARAMS)); 1247 $PARAMS = array(); 1248 1249 foreach ($PARRAY as $item) { 1250 list($key, $value) = array_pad( explode('=', $item, 2), 2, null ); 1251 if (empty($key) || empty($value)) 1252 continue; 1253 1254 if (strtolower(trim($key)) == $IDKEY) { 1255 $DATA[2] = preg_replace("%^" . preg_quote(DOKU_BASE, '%') . "%", "", str_replace(':', '/', $value)); 1256 continue; 1257 } 1258 1259 $PARAMS[] = "$key=$value"; 1260 } 1261 1262 sort($PARAMS); 1263 1264 $PARAMS = implode('&', $PARAMS); 1265 } 1266 1267 /** 1268 * rewrite detail.php calls 1269 **/ 1270 private function __rebuildDataForNormalFiles(&$DATA, &$PARAMS, $addHash = false) { 1271 $PARTS = explode('.', $DATA[2]); 1272 $EXT = ''; 1273 if (count($PARTS) > 1) { 1274 $EXT = '.' . array_pop($PARTS); 1275 } 1276 1277 $internalParams = $PARAMS = preg_replace("/(=|\?|&)/", ".", $PARAMS); 1278 1279 // Do not map *.php → export fileType for real PHP endpoints (doku.php, lib/exe/*, …). 1280 // Otherwise PDF export turns form actions into e.g. /doku.pdf and breaks crawled HTML. 1281 $trimPath = ltrim($DATA[2], '/'); 1282 $skipPhpRemap = (strpos($trimPath, 'lib/exe/') === 0) 1283 || (bool) preg_match('#(^|/)(doku|index)\.php$#', $trimPath); 1284 1285 // add anyways - if on overridde 1286 if (!$this->functions->settings->addParams && !empty($PARAMS) && $addHash) { 1287 $internalParams = md5($PARAMS); 1288 } else if (!$this->functions->settings->addParams) { 1289 $internalParams = null; 1290 } 1291 1292 $tailExt = ($EXT == '.php' && !$skipPhpRemap) ? '.' . $this->functions->settings->fileType : $EXT; 1293 $DATA[2] = implode('.', $PARTS) . (empty($internalParams) ? '' : '.' . $this->functions->cleanID($internalParams)) . $tailExt; 1294 $DATA[2] = preg_replace("/\.+/", ".", $DATA[2]); 1295 $this->functions->debug->message("Rebuilding Data for normal file.", $DATA[2], 1); 1296 } 1297 1298 /* 1299 * Clean JS and CSS cache files 1300 */ 1301 private function cleanCacheFiles() { 1302 1303 $_SERVER['HTTP_HOST'] = preg_replace("/:?\d+$/", '', $_SERVER['HTTP_HOST']); 1304 $cache = getCacheName('scripts' . $_SERVER['HTTP_HOST'] . '-siteexport-js-' . $_SERVER['SERVER_PORT'], '.js'); 1305 $this->unlinkIfExists($cache); 1306 1307 $tpl = trim(preg_replace('/[^\w-]+/', '', $_REQUEST['template'])); 1308 if ($tpl) 1309 { 1310 $tplinc = DOKU_INC . 'lib/tpl/' . $tpl . '/'; 1311 } else { 1312 $tplinc = DOKU_TPLINC; 1313 } 1314 1315 // The generated script depends on some dynamic options 1316 $cache = getCacheName('styles' . $_SERVER['HTTP_HOST'] . '-siteexport-js-' . $_SERVER['SERVER_PORT'] . DOKU_BASE . $tplinc , '.css'); 1317 $this->unlinkIfExists($cache); 1318 } 1319 1320 /** 1321 * Clear Cache 1322 */ 1323 private function unlinkIfExists($cache) { 1324 if (file_exists($cache) && @unlink($cache) === false) { 1325 $this->functions->debug->message('Could not remove file', $cache, 1 ); 1326 } 1327 1328 if (function_exists('gzopen') && @unlink("{$cache}.gz") === false ) { 1329 $this->functions->debug->message('Could not remove file', $cache . '.gz', 1 ); 1330 } 1331 } 1332} 1333