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 } else { 528 $this->functions->debug->message( "Tempfile is:", $tmpFile ); 529 } 530 531 $dirname = dirname($fileName); 532 // If a Filename was given that does not comply to the original name, use this one! 533 if ( $this->filewriter->canDoPDF() ) { 534 535 $this->functions->debug->message("Will replace old filename '{$fileName}' with {$ID}", null, 1); 536 $extension = explode('.', $fileName); 537 $extension = array_pop($extension); 538 539 // 2014-04-29 added cleanID to ensure that links are generated consistently when using [[this>...]] or another local, relativ linking 540 $fileName = $dirname . '/' . $this->functions->cleanID($this->functions->getSiteTitle($ID)) . '.' . $extension; 541 } else if ( !empty($tmpFile[1]) /*&& !strstr($DATA[2], $tmpFile[1])*/ ) { // 2017-11-30: $DATA is never defined 542 543 $this->functions->debug->message("Will replace old filename '{$fileName}' with {$dirname}/{$tmpFile[1]}", null, 1); 544 $fileName = $dirname . '/' . $tmpFile[1]; 545 } 546 547 // Add to zip 548 $this->fileChecked[$url] = $fileName; 549 $status = $this->filewriter->__addFileToZip($tmpFile[0], $fileName); 550 if (@unlink($tmpFile[0]) === false) { 551 $this->functions->debug->message("Could not remove temporary file", $tmpFile[0], 1); 552 } 553 554 return $status; 555 } 556 557 /** 558 * Download the file via HTTP URL + recurse if this is not an image 559 * The file will be saved as temporary file. The filename is the result. 560 **/ 561 private function __getHTTPFile($URL, $RECURSE=false, $newAdditionalParameters=null) { 562 global $conf; 563 564 $EXCLUDE = $this->functions->settings->excludePattern; 565 if ( !empty($EXCLUDE) ) { 566 $PATTERN = "/(" . implode('|', explode(' ', preg_quote($EXCLUDE, '/'))) . ")/i"; 567 568 $this->functions->debug->message("Checking for exclude: ", array( 569 "pattern" => $PATTERN, 570 "file" => $URL, 571 "matches" => preg_match($PATTERN, $URL) ? 'match' : 'no match' 572 ), 2); 573 574 if ( preg_match($PATTERN, $URL) ) { return false; } 575 } 576 577 $http = new HTTPProxy($this->functions); 578 $http->max_bodysize = $conf['fetchsize']; 579 580 // Add additional Params 581 $this->functions->addAdditionalParametersToURL($URL, $newAdditionalParameters); 582 583 $this->functions->debug->message("Fetching URL: '$URL'", null, 2); 584 $getData = $http->get($URL, true); // true == sloopy, get 304 body as well. 585 586 if( $getData === false ) { // || ($http->status != 200 && !$this->functions->settings->ignoreNon200) ) { 587 588 if ( $http->status != 200 && !$this->functions->settings->ignoreNon200 ) { 589 $this->functions->debug->message("Sending request failed with error, HTTP status was '{$http->status}'.", $URL, 4); 590 return false; 591 } 592 } 593 594 if( empty($getData) ) { 595 $this->functions->debug->message("No data fetched", $URL, 4); 596 return false; 597 } 598 599 $this->functions->debug->message("Headers received", $http->resp_headers, 2); 600 601 if ( !$RECURSE ) { 602 // Parse URI PATH and add "html" 603 $this->functions->debug->message("========================================", null, 1); 604 $this->functions->debug->message("Starting to recurse file '$URL'", null , 1); 605 $this->functions->debug->message("----------------------------------------", null, 1); 606 $this->__getInternalLinks($getData); 607 $this->functions->debug->message("----------------------------------------", null, 1); 608 $this->functions->debug->message("Finished to recurse file '$URL'", null , 1); 609 $this->functions->debug->message("========================================", null, 1); 610 } 611 612 $tmpFile = tempnam($this->functions->settings->tmpDir , 'siteexport__') ?: $this->functions->settings->tmpDir . "siteexport__"; 613 $this->functions->debug->message("Temporary filename", $tmpFile, 1); 614 615 $fp = fopen( $tmpFile, "w"); 616 if(!$fp) { 617 $this->functions->debug->message("Can't open temporary File '$tmpFile'.", null , 4); 618 return false; 619 } 620 621 fwrite($fp,$getData); 622 fclose($fp); 623 624 // plain/text; ... 625 $extension = explode(';', $http->resp_headers['content-type'], 2); 626 $extension = array_shift($extension); 627 $extension = explode('/', $extension, 2); 628 if ( $extension[0] == 'image' && preg_match("/^[a-zA-Z0-9]{3,}$/", $extension[1]) ) { 629 $extension = strtolower($extension[1]); 630 $this->functions->debug->message("Found new image extension:", $extension, 2); 631 } else { 632 unset($extension); 633 } 634 635 return array($tmpFile, preg_replace("/.*?filename=\"?(.*?)\"?;?$/", "$1", $http->resp_headers['content-disposition'] ?? ''), $extension ?? ''); 636 } 637 638 /** 639 * Find internal links in the currently downloaded file. This also matches inside CSS files 640 **/ 641 private function __getInternalLinks(&$DATA) { 642 643 $PATTERN = '(href|src|action)="([^"]*)"'; 644 if (!$this->functions->settings->exportLinkedPages) { 645 // no links or forms 646 $PATTERN = '((?<!<a )href|src|action)="([^"]*)"'; 647 } 648 649 $CALLBACK = array($this, '__fetchAndReplaceLink'); 650 $DATA = preg_replace_callback("/$PATTERN/i", $CALLBACK, $DATA); 651 652 $PATTERNCSS = '(url\s*?)\(([^\)]*)\)'; 653 $DATA = preg_replace_callback("/$PATTERNCSS/i", $CALLBACK, $DATA); 654 655 $PATTERNSRCSET = '(srcset)="([^"]*)"'; 656 $CALLBACK = array($this, '__fetchAndReplaceSrcset'); 657 $DATA = preg_replace_callback("/$PATTERNSRCSET/i", $CALLBACK, $DATA); 658 } 659 660 /** 661 * Support for 'srcset' image atributes 662 */ 663 private function __fetchAndReplaceSrcset($DATA) { 664 665 $this->functions->debug->message("SRCSET", $DATA, 2); 666 if ( strtolower($DATA[1]) != 'srcset' ) { return $DATA[0]; } 667 668 $SRCSETS = array_map( 'trim', explode( ",", $DATA[2] ) ); 669 $NEWSRCSETS = array(); 670 foreach( $SRCSETS as $SRCSET ) { 671 672 // tehre should be no other unencoded spaces in here 673 list($url, $size) = explode( " ", $SRCSET, 2 ); 674 675 $this->functions->debug->message("SRCSET: URL before: '$url'", null, 2); 676 $url = $this->__fetchAndReplaceLink(array( $url, '', $url )); 677 $url = substr( $url, 2, -1 ); 678 $this->functions->debug->message("SRCSET: URL after: '$url'", null, 2); 679 680 $NEWSRCSETS[]=$url . ' ' . $size; 681 } 682 683 return 'srcset="' . implode( ',', $NEWSRCSETS ) . '"'; 684 } 685 686 /** 687 * Deep Fetch and replace of links inside the texts matched by __getInternalLinks 688 **/ 689 private function __fetchAndReplaceLink($DATA) { 690 global $conf, $CURRENT_ID, $CURRENT_PARENT; 691 692 $noDeepReplace = true; 693 $newAdditionalParameters = $this->functions->settings->additionalParameters; 694 $newDepth = $this->functions->settings->depth; 695 $hadBase = false; 696 697 // Clean data[2], remote ' and " 698 $DATA[2] = preg_replace("/^\s*?['\"]?(.*?)['\"]?\s*?$/", '\1', trim($DATA[2])); 699 700 $this->functions->debug->message("Starting Link Replacement", array('data' => $DATA, 'additional Params' => $newAdditionalParameters, 'newDepth' => $newDepth, 'currentID' => $CURRENT_ID, 'currentParent' => $CURRENT_PARENT), 2); 701 702 // STEP 1: check for well known links that can be returned 703 if ( $this->__fetchAndReplaceWellKnownLinks( $DATA ) ) { 704 return $this->__rebuildLink($DATA, ""); 705 } 706 707 // 2014-07-21: Origdata before anything else - or it will be missing some things. 708 $ORIGDATA2 = $DATA; 709 // $ORIGDATA2 = $DATA[2]; // 08/10/2010 - this line required a $this->functions->wl which may mess up with the base URL 710 $this->functions->debug->message("OrigDATA is:", $ORIGDATA2, 1); 711 712 // strip all things out 713 // changed Data 714 $PARAMS = @parse_url($DATA[2], PHP_URL_QUERY); 715 $ANCHOR = @parse_url($DATA[2], PHP_URL_FRAGMENT); 716 $DATA[2] = @parse_url($DATA[2], PHP_URL_PATH); 717 718 // 2014-05-12 - fix problem with URLs starting with a ./ or ../ ... they seem to need the current IDs root 719 if (preg_match("#^\.\.?/#", $DATA[2])) { 720 $DATA[2] = getNS($CURRENT_ID) . ':' . $DATA[2]; 721 } 722 723 // 2010-08-25 - fix problem with relative movement in links ( "test/../test2" ) 724 // 2014-06-30 - what? to what will this end relatively? 725 $tmpData2 = ''; 726 while ($tmpData2 != $DATA[2]) { 727 $tmpData2 = $DATA[2]; 728 $DATA[2] = preg_replace("#/(?!\.\.)[^\/]*?/\.\./#", '/', $DATA[2]); 729 } 730 731 $temp = preg_replace("%^" . preg_quote(DOKU_BASE, '%') . "%", "", $DATA[2]); 732 if ($temp != $DATA[2]) { 733 $DATA[2] = $temp; 734 $hadBase = true; // 2010-08-23 Check if there has been a rewrite here that will have to be considered later on 735 } 736 737 $this->functions->debug->message("URL before rewriting option for others than 1", array($DATA, $PARAMS, $hadBase), 1); 738 739 740 // Handle rewrites other than 1 - just for non-lib-files 741 $this->__fetchAndReplaceLinkHandleRewrite( $DATA, $PARAMS ); 742 743 $this->functions->debug->message("URL before rewriting option", array($DATA, $PARAMS), 2); 744 745 // Generate ID 746 $DATA[2] = str_replace('/', ':', $DATA[2]); 747 748 // If Data was empty this must be the same file!; 749 if (empty($DATA[2])) { 750 $DATA[2] = $CURRENT_ID; 751 } 752 753 $ID = $DATA[2]; 754 $MEDIAMATCHER = "#(_media(/|:)|media=|_detail(/|:)|_export(/|:)|do=export_)#i"; // 2010-10-23 added "(/|:)" for the ID may not contain slashes anymore 755 $ISMEDIA = preg_match($MEDIAMATCHER, $DATA[2]) === 1; 756 if ($ISMEDIA !== false && $conf['userewrite'] == 1) { 757 $ID = preg_replace("#^_(detail|media)(/|:)#", "", $ID); 758 } 759 760 $ID = $this->functions->cleanID($DATA[2], null, $ISMEDIA); 761 // $ID = $this->functions->cleanID($DATA[2], null, strstr($DATA[2], 'media') ); // Export anpassung nun weiter unten 762 763 // $IDexists = page_exists($ID); // 08/10/2010 - Not needed. This will be done in the next block. 764 // $this->functions->debug->message("Current ID: '$ID' exists: '" . ($IDexists ? 'true' : 'false') . "' (will be set to 'false' anyway)", null, 1); 765 766 $IDifIDnotExists = $ID; // 08/10/2010 - Save ID - with possible upper cases to preserve them 767 $IDexists = false; 768 769 $this->functions->debug->message("Resolving ID: '$ID'", null, 2); 770 if ($ISMEDIA !== false) { 771 $ID = (new MediaResolver('root'))->resolveId($ID); 772 $IDexists = media_exists($ID); 773 $this->functions->debug->message("Current mediaID to filename: '" . mediaFN($ID) . "'", null, 2); 774 } else { 775 $ID = (new PageResolver('root'))->resolveId($ID); 776 $IDexists = page_exists($ID); 777 $this->functions->debug->message("Current ID to filename: '" . wikiFN($ID) . "'", null, 2); 778 } 779 780 $this->functions->debug->message("Current ID after resolvement: '$ID' the ID does exist: '" . ($IDexists ? 'true' : 'false') . "'", null, 2); 781 // $ORIGDATA2 = @parse_url($this->functions->wl($ORIGDATA2, null, true)); // What was the next 2 line for? It did mess up with links from {{jdoc>}} 782 // $this->functions->debug->message("OrigData ID after parse:", $ORIGDATA2, 1); // 08/10/2010 - The lines are obsolete when the $ORIGDATA2 = $DATA. $ORIGDATA is only for fallback 783 784 // 08/10/2010 - If the ID does not exist, we may have a problem here with upper cases - they will all be lower by now! 785 if (!$IDexists) { 786 $ID = $IDifIDnotExists; // there may have been presevered Upper cases. We will need them! 787 } 788 789 // $this->functions->cleanID($DATA[2], null, strstr($DATA[2], 'media') || strstr($DATA[2], 'export') ); 790 if (substr($ID, -1) == ':' || empty($ID)) $ID .= $conf['start']; 791 792 // Generate Download URL 793 // $PARAMS = trim(str_replace('&', '&', $PARAMS)); 794 $PARAMS = trim($PARAMS); 795 $this->functions->removeWikiVariables($PARAMS, false, true); 796 797 $url = $this->functions->wl($ID, null, true, null, null, true, $hadBase) . (!empty($ANCHOR) ? '#' . $ANCHOR : '') . (!empty($PARAMS) ? '?' . $PARAMS : ''); 798 $this->functions->debug->message("URL from ID: '$url'", null, 2); 799 800 // Parse URI PATH and add "html" 801 $uri = @parse_url($url); 802 $DATA[2] = $uri['path']; 803 804 $this->functions->debug->message("DATA after parsing.", $DATA, 2); 805 806 // Second Rewrite for UseRewrite = 2 807 if ($conf['userewrite'] == 2 && preg_match("%((/lib/exe/(fetch|detail|indexer)|feed|doku)\.php)/?(.*?)$%", $DATA[2], $matches)) { 808 809 // The actual file in lib 810 $DATA[2] = $matches[1]; 811 $PARAMS .= '&' . (in_array($matches[3], array('fetch', 'detail')) ? 'media' : 'id') . '=' . cleanID(str_replace('/', ':', $matches[4])); 812 813 $this->functions->debug->message("DATA after second rewrite with UseRewrite = 2", array($DATA, $matches, $PARAMS), 1); 814 } 815 816 $DATA['ANCHOR'] = $ANCHOR; 817 $DATA['PARAMS'] = $PARAMS; 818 $elements = explode('/', $DATA[2]); 819 820 // Main Switch to check the link 821 $result = $this->__fetchAndReplaceLinkMainSwitch( $elements, $DATA, $url, $newAdditionalParameters, $PARAMS, $noDeepReplace, $fileName, $newDepth, $ID, $IDexists ); 822 if ( $result !== null ) { 823 return $result; 824 } 825 826 $this->functions->debug->message("DATA after SWITCH CASE decision", array($DATA, $noDeepReplace, $fileName, $newDepth), 1); 827 828 if ($this->filewriter->canDoPDF()) { 829 $this->functions->addAdditionalParametersToURL($url, $newAdditionalParameters); 830 $DATA[2] = $url; 831 unset($DATA['PARAMS']); 832 $url = $this->__rebuildLink($DATA, ''); 833 834 $this->functions->debug->message("Creating PDF with URL '$url'", null, 2); 835 836 return $url; 837 } 838 839 // Finalize 840 return $this->__fetchAndReplaceLinkFinish( $DATA, $url, $noDeepReplace, $newAdditionalParameters, $ORIGDATA2, $newDepth, $IDexists, $fileName ); 841 } 842 843 private function __fetchAndReplaceLinkMainSwitch( &$elements, &$DATA, &$url, &$newAdditionalParameters, &$PARAMS, &$noDeepReplace, &$fileName, &$newDepth, &$ID, &$IDexists ) { 844 switch (array_pop($elements)) { 845 // CSS Extra Handling with extra rewrites 846 case 'css.php' : // $DATA[2] .= ( !$this->functions->settings->addParams || empty($PARAMS) ? '' : '.' . $this->functions->cleanID(preg_replace("/(=|\?|&)/", ".", $PARAMS))) . '.css'; 847 $DATA[2] .= '.' . $this->functions->cleanID(preg_replace("/(=|\?|&)/", ".", $PARAMS)) . '.css'; // allways put parameters behind 848 // No paramters needed since they are rewritten. 849 $DATA['PARAMS'] = ""; 850 $noDeepReplace = false; 851 $fileName = $this->functions->getSiteName($ID, true); 852 853 // NewDepth has to be relative to the css file itself ... 854 $newDepth = './' . str_repeat('../', count(explode('/', $fileName))-1); // it is an ID at this point. 855 $newAdditionalParameters['do'] = 'siteexport'; 856 857 $this->functions->debug->message("This is CSS file", array($DATA, $noDeepReplace, $fileName, $newDepth, $newAdditionalParameters), 2); 858 859 break; 860 case 'jquery.php' : 861 case 'js.php' : // $DATA[2] .= ( !$this->functions->settings->addParams || empty($PARAMS) ? '' : '.' . $this->functions->cleanID(preg_replace("/(=|\?|&)/", ".", $PARAMS))) . '.js'; 862 $DATA[2] .= '.t.' . $this->functions->cleanID($_REQUEST['template']) . '.js'; // allways put parameters behind 863 // set Template 864 if (!empty($_REQUEST['template'])) { 865 $url .= (strstr($url, '?') ? '&' : '?') . 'template=' . $_REQUEST['template']; 866 } 867 // No paramters needed since they are rewritten. 868 $DATA['PARAMS'] = ""; 869 $newAdditionalParameters['do'] = 'siteexport'; 870 871 $this->functions->debug->message("This is JS file", array($DATA, $url, $newAdditionalParameters), 2); 872 873 break; 874 // Detail Handling with extra Rewrites if Paramaters are available - otherwise this is just the fetch 875 case 'indexer.php' : 876 $this->functions->debug->message("Skipping indexer", null, 2); 877 return ""; 878 case 'detail.php' : 879 $noDeepReplace = false; 880 881 $this->__getParamsAndDataRewritten($DATA, $PARAMS, 'media'); 882 $ID = $this->functions->cleanID(str_replace('/', ':', $DATA[2]), null, strstr($DATA[2], 'media')); 883 $fileName = $this->functions->getSiteName($ID, true); // 2010-09-03 - rewrite with override enabled 884 885 $newDepth = str_repeat('../', count(explode('/', $fileName))-1); 886 $this->__rebuildDataForNormalFiles($DATA, $PARAMS); 887 $DATA[2] .= '.detail.html'; 888 889 $this->functions->debug->message("This is detail.php file with addParams", array($DATA, $ID, $fileName, $newDepth, $newAdditionalParameters), 2); 890 break; 891 case 'doku.php' : 892 893 $noDeepReplace = false; 894 $this->__getParamsAndDataRewritten($DATA, $PARAMS, 'id'); 895 $ID = $this->functions->cleanID($DATA[2], null, strstr($DATA[2], 'id')); 896 897 $this->functions->debug->message("Current ID to filename (doku.php): '" . wikiFN($ID) . "'", null, 2); 898 899 $fileName = $this->functions->getSiteName($ID); // 2010-09-03 - rewrite with override enabled 900 901 $newDepth = str_repeat('../', count(explode('/', $fileName))-1); 902 $this->__rebuildDataForNormalFiles($DATA, $PARAMS); 903 $DATA2Name = explode('/', $fileName); 904 $DATA[2] .= '.' . array_pop($DATA2Name); 905 906 $this->functions->debug->message("This is doku.php file with addParams", array($DATA, $ID, $fileName, $newDepth, $newAdditionalParameters), 2); 907 return $this->__rebuildLink($DATA); 908 909 // Fetch Handling for media - rewriting everything 910 case 'fetch.php': 911 $this->__getParamsAndDataRewritten($DATA, $PARAMS, 'media'); 912 913 $DATA[2] = str_replace('/', ':', $DATA[2]); 914 $ID = $this->functions->cleanID($DATA[2], null, strstr($DATA[2], 'media')); 915 $ID = (new MediaResolver('root'))->resolveId($ID); 916 $IDexists = media_exists($ID); 917 918 $DATA[2] = $this->functions->wl($ID, null, null, null, $IDexists, true); 919 $this->__rebuildDataForNormalFiles($DATA, $PARAMS); 920 921 $DATA['PARAMS'] = ""; 922 $newAdditionalParameters = array(); 923 924 $this->functions->debug->message("This is fetch.php file", array($DATA, $ID, $PARAMS), 2); 925 break; 926 927 // default Handling for Pages 928 case 'feed.php': 929 return ""; // Ignore. Has no sense to export. 930 default: 931 if (preg_match("%" . preg_quote(DOKU_BASE, '%') . "_detail/%", $DATA[2])) { 932 933 // GET ID Param from origdata2 934 preg_match("#id=(.*?)(&|\")#i", $DATA[0], $backlinkID); 935 $this->__rebuildDataForNormalFiles($DATA, $PARAMS); 936 937 $fileIDPart = isset($backlinkID[1]) && !empty($backlinkID[1]) ? $this->functions->cleanID(urldecode($backlinkID[1])) : 'detail'; 938 939 $ID = preg_replace("#^_detail(/|:)#", "", $ID); 940 $DATA[2] .= ':' . $fileIDPart . '.' . $this->functions->settings->fileType; // add namespace and subpage for back button and add filetype 941 942 $noDeepReplace = false; 943 $fileName = $this->functions->shortenName($DATA[2]); 944 $newDepth = str_repeat('../', count(explode('/', $fileName))-1); 945 $url .= (strstr($url, '?') ? '&' : '?') . 'id=' . $fileIDPart; // add id-part to URL for backlinks 946 947 $DATA['PARAMS'] = ""; 948 949 $this->functions->debug->message("This is something with '_detail' file", array($DATA, $backlinkID, $newDepth, $url, $ID), 2); 950 } else if (preg_match("%" . preg_quote(DOKU_BASE, '%') . "_export/(.*?)/%", $DATA[2], $fileType)) { 951 952 // Fixes multiple codeblocks in one file 953 $this->__rebuildDataForNormalFiles($DATA, $PARAMS); 954 955 // add the Params no matter what they are. This is export. We don't mess with other files 956 // adding the "/" fixes the usage of multiple codeblocks in the same namespace 957 $DATA[2] .= (empty($PARAMS) ? '' : '/' . $PARAMS) . '.' . $fileType[1]; 958 959 $DATA['PARAMS'] = ""; 960 $this->functions->debug->message("This is something with '_export' file", $DATA, 2); 961 962 } else if ($IDexists) { // 08/10/2010 - was page_exists($ID) - but this should do as well. 963 // If this is a page ... skip it! 964 $DATA[2] .= (!$this->functions->settings->addParams || empty($PARAMS) ? '' : '.' . $this->functions->cleanID(preg_replace("/(=|\?|&)/", ".", $PARAMS))) . '.' . $this->functions->settings->fileType; 965 966 $DATA[2] = $this->functions->shortenName($DATA[2]); 967 968 // If Parameters are to be included in the filename - they must not be added twice 969 if ($this->functions->settings->addParams) $DATA['PARAMS'] = ""; 970 971 $this->functions->debug->message("This page really exists", $DATA, 1); 972 973 return $this->__rebuildLink($DATA, null, $ID); 974 } else { 975 $this->__rebuildDataForNormalFiles($DATA, $PARAMS, true); 976 $newAdditionalParameters = null; // 2014-06-27 - when using the "normal" files way we will not need any additional stuff. 977 // This would make problems with e.g. ditaa plugin 978 } 979 980 unset($newAdditionalParameters['diPlu']); 981 } 982 983 return null; 984 } 985 986 private function __fetchAndReplaceLinkFinish( $DATA, $url, $noDeepReplace, $newAdditionalParameters, $ORIGDATA2, $newDepth, $IDexists, $fileName ) { 987 global $conf, $CURRENT_ID, $CURRENT_PARENT; 988 989 // Create Name to save the file at 990 $DATA[2] = str_replace(':', '_', $DATA[2]); 991 $DATA[2] = $this->functions->shortenName($DATA[2]); 992 993 994 // File already loaded? 995 // 2010-10-23 - changes in_array from DATA[2] to $url - to check real URLs, the DATA[2] file will be checked with fileExistsInZip 996 if (in_array($url, array_keys($this->fileChecked))) { 997 $DATA[2] = $this->fileChecked[$url]; 998 $this->functions->debug->message("File has been checked before.", array($DATA, $url), 2); 999 return $this->__rebuildLink($DATA); 1000 } 1001 1002 // 2010-09-03 - second check if the file is in the ZIP already. 1003 if ($this->filewriter->fileExistsInZip($DATA[2])) { 1004 $this->functions->debug->message("File with DATA exists in ZIP.", $DATA, 3); 1005 return $this->__rebuildLink($DATA); 1006 } 1007 1008 // 2010-10-23 - What if this is a fetch.php? than we produced an error. 1009 // $this->fileChecked[] = $DATA[2]; 1010 1011 // get tempFile and save it 1012 $origDepth = $this->functions->settings->depth; 1013 $this->functions->settings->depth = $newDepth; 1014 1015 $tmpID = $CURRENT_ID; 1016 $tmpParent = $CURRENT_PARENT; 1017 1018 $CURRENT_PARENT = $fileName; 1019 $this->functions->debug->message("Going to get the file", array($url, $noDeepReplace, $newAdditionalParameters), 2); 1020 $tmpFile = $this->__getHTTPFile($url, $noDeepReplace, $newAdditionalParameters); 1021 $this->functions->debug->message("The getHTTPFile result is still empty", $tmpFile === false ? 'YES' : 'NO', 2); 1022 1023 $CURRENT_PARENT = $tmpParent; 1024 $CURRENT_ID = $tmpID; 1025 $this->functions->settings->depth = $origDepth; // 2010-09-03 - Reset depth at the very end 1026 1027 if ($tmpFile === false) { 1028 // Keep an potentially extra link intact 1029 1030 $this->functions->debug->message("The fetched file '$url' is 'false'", null, 3); 1031 if ($IDexists === false) { 1032 $this->functions->debug->message("The file does not exist, fallback to ORIGDATA", $ORIGDATA2, 2); 1033 $DATA[2] = $this->functions->shortenName($ORIGDATA2[2]); // get Origdata Path 1034 } 1035 1036 $this->fileChecked[$url] = $DATA[2]; // 2010-09-03 - One URL to one FileName 1037 $link = $this->__rebuildLink($DATA); 1038 $this->functions->debug->message("Final Link after empty file from '$url'", null, 2); 1039 1040 return $link; 1041 } 1042 1043 $this->functions->debug->message("The fetched file looks good.", $tmpFile, 2); 1044 $dirname = dirname($DATA[2]); 1045 1046 // If a Filename was given that does not comply to the original name, us this one! 1047 // 2014-02-28 But only if we are on PDF Mode. Does this produce any other Problems? 1048 if ( $this->filewriter->canDoPDF() && !empty($tmpFile[1]) && !strstr($DATA[2], $tmpFile[1]) ) { 1049 $DATA[2] = $dirname . '/' . $tmpFile[1]; 1050 $this->functions->debug->message("Changed filename.", $DATA[2], 2); 1051 } 1052 1053 // Custom extension if not set already - 2014-07-02 1054 if ( !empty($tmpFile[2]) && !preg_match("#\.{$tmpFile[2]}$#", $DATA[2]) ) { 1055 $DATA[2] = preg_match("#(\.[^\.]+)$#", $DATA[2]) ? preg_replace("#(\.[^\.]+)$#", '.' . $tmpFile[2], $DATA[2]) : $DATA[2] . '.' . $tmpFile[2]; 1056 $this->functions->debug->message("Added extension provided from Server.", $DATA[2], 2); 1057 } 1058 1059 // Add to zip 1060 $this->fileChecked[$url] = $DATA[2]; // 2010-09-03 - One URL to one FileName 1061 1062 $this->filewriter->__addFileToZip($tmpFile[0], $DATA[2]); 1063 if (@unlink($tmpFile[0]) === false) { 1064 $this->functions->debug->message("Could not delete temporary file.", $tmpFile[0], 2); 1065 } 1066 1067 $newURL = $this->__rebuildLink($DATA); 1068 $this->functions->debug->message("Returning final Link to document: '$newURL'", null, 2); 1069 1070 return $newURL; 1071 } 1072 1073 private function __fetchAndReplaceWellKnownLinks( $DATA ) { 1074 // $DATA[2] = urldecode($DATA[2]); // Leads to problems because it does not re-encode the url 1075 // External and mailto links 1076 if (preg_match("%^(https?://|mailto:|javascript:|data:)%", $DATA[2])) { 1077 $this->functions->debug->message("Don't like http, mailto, data or javascript links here", null, 1); 1078 return true; 1079 } 1080 //if ( preg_match("%^(https?://|mailto:|" . DOKU_BASE . "/_export/)%", $DATA[2]) ) { return $this->__rebuildLink($DATA, ""); } 1081 // External media - this is deep down in the link, so we have to grep it out 1082 if (preg_match("%media=(https?://.*?$)%", $DATA[2], $matches)) { 1083 $DATA[2] = $matches[1]; 1084 $this->functions->debug->message("This is an HTTP like somewhere else", $DATA, 1); 1085 return true; 1086 } 1087 // reference only links won't have to be rewritten 1088 if (preg_match("%^#.*?$%", $DATA[2])) { 1089 $this->functions->debug->message("This is a refercence only", null, 1); 1090 return true; 1091 } 1092 1093 return false; 1094 } 1095 1096 // Handle rewrites other than 1 - just for non-lib-files 1097 private function __fetchAndReplaceLinkHandleRewrite( &$DATA, &$PARAMS ) { 1098 global $conf; 1099 if ( !preg_match('$^(' . DOKU_BASE . ')?lib/$', $DATA[2]) ) { 1100 $this->functions->debug->message("Did not match '$^(" . DOKU_BASE . ")?lib/$' userewrite == {$conf['userewrite']}", null, 2); 1101 if ( $conf['userewrite'] == 2 ) { 1102 $DATA[2] = $this->__getInternalRewriteURL($DATA[2]); 1103 } elseif ( $conf['userewrite'] == 0 ) { 1104 $this->__getParamsAndDataRewritten($DATA, $PARAMS); 1105 } 1106 } else { 1107 $this->functions->debug->message("This file must be inside lib ...", null, 2); 1108 } 1109 } 1110 1111 /** 1112 * build the new link to be put in place for the donwloaded site 1113 **/ 1114 private function __rebuildLink($DATA, $DEPTH = null, $existingPageID = null) { 1115 global $CURRENT_ID, $CURRENT_PARENT; 1116 1117 // depth is set, skip this one 1118 if (is_null($DEPTH)) $DEPTH = $this->functions->settings->depth; 1119 $DATA[2] .= (!empty($DATA['PARAMS']) && $this->functions->settings->addParams ? '?' . $DATA['PARAMS'] : '') . (!empty($DATA['ANCHOR']) ? '#' . $DATA['ANCHOR'] : ''); 1120 1121 $intermediateURL = $DEPTH . $DATA[2]; 1122 1123 // 2012-06-15 originally has an absolute path ... we might need a relative one if not in our namespace 1124 if (empty($_REQUEST['absolutePath']) && preg_match("#^(\.\./)+#", $intermediateURL)) { 1125 1126 $this->functions->debug->message("OK, this is not to be absolute: ", array($intermediateURL, $CURRENT_PARENT), 1); 1127 $intermediateURL = $this->functions->getRelativeURL($intermediateURL, $CURRENT_PARENT, $existingPageID); 1128 } 1129 1130 $newURL = $DATA[1] == 'url' ? $DATA[1] . '(' . $intermediateURL . ')' : $DATA[1] . '="' . $intermediateURL . '"'; 1131 $this->functions->debug->message("Re-created URL: '$newURL'", $DEPTH, 2); 1132 1133 return $newURL; 1134 } 1135 1136 1137 /** 1138 * remove an old zip file 1139 **/ 1140 private function __removeOldZip($FILENAMEID = null, $checkForMore = true, $reauthenticated = false) { 1141 global $INFO; 1142 global $conf; 1143 1144 $returnValue = true; 1145 1146 if (empty($FILENAMEID)) { 1147 $FILENAMEID = $this->functions->settings->origZipFile; 1148 } 1149 1150 if (!file_exists(mediaFN($FILENAMEID))) { 1151 $returnValue = true; 1152 } else { 1153 1154 require_once(DOKU_INC . 'inc/media.php'); 1155 if (!media_delete($FILENAMEID, $INFO['perm'])) { 1156 1157 if (!$reauthenticated) { 1158 $this->functions->authenticate(); 1159 return $this->__removeOldZip($FILENAMEID, $checkForMore, true); 1160 } 1161 1162 $returnValue = false; 1163 } 1164 } 1165 1166 if ($checkForMore) { 1167 // Try to remove more files. 1168 $ns = getNS($FILENAMEID); 1169 $fn = $this->functions->getSpecialExportFileName(noNS($FILENAMEID), '.+'); 1170 1171 $data = array(); 1172 search($data, $conf['mediadir'], 'search_media', array('pattern' => "/$fn$/i"), $ns); 1173 1174 if (count($data) > 0) { 1175 1176 // 30 Minuten Cache Zeit 1177 $cache = $this->functions->settings->cachetime; 1178 foreach ($data as $media) { 1179 1180 //decide if has to be deleted needed: 1181 if ($media['mtime'] < time()-$cache) { 1182 $this->__removeOldZip($media['id'], false, $reauthenticated); 1183 } 1184 } 1185 } 1186 1187 } 1188 1189 return $returnValue; 1190 } 1191 1192 /** 1193 * if confrewrite is set to internal rewrite, use this function - taken from a DW renderer 1194 **/ 1195 private function __getInternalRewriteURL($url) { 1196 global $conf; 1197 1198 //construct page id from request URI 1199 if ($conf['userewrite'] != 2) { return $url; } 1200 1201 //get the script URL 1202 if ($conf['basedir']) { 1203 $relpath = ''; 1204 $script = $conf['basedir'] . $relpath . basename($_SERVER['SCRIPT_FILENAME']); 1205 } elseif ($_SERVER['DOCUMENT_ROOT'] && $_SERVER['SCRIPT_FILENAME']) { 1206 $script = preg_replace('/^' . preg_quote($_SERVER['DOCUMENT_ROOT'], '/') . '/', '', 1207 $_SERVER['SCRIPT_FILENAME']); 1208 $script = '/' . $script; 1209 } else { 1210 $script = $_SERVER['SCRIPT_NAME']; 1211 } 1212 1213 //clean script and request (fixes a windows problem) 1214 $script = preg_replace('/\/\/+/', '/', $script); 1215 $request = preg_replace('/\/\/+/', '/', $url); 1216 1217 //remove script URL and Querystring to gain the id 1218 $id = $request; 1219 if (preg_match('/^' . preg_quote($script, '/') . '(.*)/', $request, $match)) { 1220 $id = preg_replace('/\?.*/', '', $match[1]); 1221 } 1222 $id = urldecode($id); 1223 //strip leading slashes 1224 $id = preg_replace('!^/+!', '', $id); 1225 1226 return $id; 1227 } 1228 1229 /** 1230 * rewrite parameter calls 1231 **/ 1232 private function __getParamsAndDataRewritten(&$DATA, &$PARAMS, $IDKEY = 'id') { 1233 1234 $PARRAY = explode('&', str_replace('&', '&', $PARAMS)); 1235 $PARAMS = array(); 1236 1237 foreach ($PARRAY as $item) { 1238 list($key, $value) = explode('=', $item, 2); 1239 if (empty($key) || empty($value)) 1240 continue; 1241 1242 if (strtolower(trim($key)) == $IDKEY) { 1243 $DATA[2] = preg_replace("%^" . preg_quote(DOKU_BASE, '%') . "%", "", str_replace(':', '/', $value)); 1244 continue; 1245 } 1246 1247 $PARAMS[] = "$key=$value"; 1248 } 1249 1250 sort($PARAMS); 1251 1252 $PARAMS = implode('&', $PARAMS); 1253 } 1254 1255 /** 1256 * rewrite detail.php calls 1257 **/ 1258 private function __rebuildDataForNormalFiles(&$DATA, &$PARAMS, $addHash = false) { 1259 $PARTS = explode('.', $DATA[2]); 1260 $EXT = ''; 1261 if (count($PARTS) > 1) { 1262 $EXT = '.' . array_pop($PARTS); 1263 } 1264 1265 $internalParams = $PARAMS = preg_replace("/(=|\?|&)/", ".", $PARAMS); 1266 1267 // add anyways - if on overridde 1268 if (!$this->functions->settings->addParams && !empty($PARAMS) && $addHash) { 1269 $internalParams = md5($PARAMS); 1270 } else if (!$this->functions->settings->addParams) { 1271 $internalParams = null; 1272 } 1273 1274 $DATA[2] = implode('.', $PARTS) . (empty($internalParams) ? '' : '.' . $this->functions->cleanID($internalParams)) . ($EXT == '.php' ? '.' . $this->functions->settings->fileType : $EXT); 1275 $DATA[2] = preg_replace("/\.+/", ".", $DATA[2]); 1276 $this->functions->debug->message("Rebuilding Data for normal file.", $DATA[2], 1); 1277 } 1278 1279 /* 1280 * Clean JS and CSS cache files 1281 */ 1282 private function cleanCacheFiles() { 1283 1284 $_SERVER['HTTP_HOST'] = preg_replace("/:?\d+$/", '', $_SERVER['HTTP_HOST']); 1285 $cache = getCacheName('scripts' . $_SERVER['HTTP_HOST'] . '-siteexport-js-' . $_SERVER['SERVER_PORT'], '.js'); 1286 $this->unlinkIfExists($cache); 1287 1288 $tpl = trim(preg_replace('/[^\w-]+/', '', $_REQUEST['template'])); 1289 if ($tpl) 1290 { 1291 $tplinc = DOKU_INC . 'lib/tpl/' . $tpl . '/'; 1292 } else { 1293 $tplinc = DOKU_TPLINC; 1294 } 1295 1296 // The generated script depends on some dynamic options 1297 $cache = getCacheName('styles' . $_SERVER['HTTP_HOST'] . '-siteexport-js-' . $_SERVER['SERVER_PORT'] . DOKU_BASE . $tplinc , '.css'); 1298 $this->unlinkIfExists($cache); 1299 } 1300 1301 /** 1302 * Clear Cache 1303 */ 1304 private function unlinkIfExists($cache) { 1305 if (file_exists($cache) && @unlink($cache) === false) { 1306 $this->functions->debug->message('Could not remove file', $cache, 1 ); 1307 } 1308 1309 if (function_exists('gzopen') && @unlink("{$cache}.gz") === false ) { 1310 $this->functions->debug->message('Could not remove file', $cache . '.gz', 1 ); 1311 } 1312 } 1313} 1314