1c3437056SNickeau<?php 2c3437056SNickeau 3c3437056SNickeau 4c3437056SNickeauuse ComboStrap\DatabasePageRow; 504fd306cSNickeauuse ComboStrap\ExceptionBadArgument; 604fd306cSNickeauuse ComboStrap\ExceptionBadSyntax; 704fd306cSNickeauuse ComboStrap\ExceptionCompile; 804fd306cSNickeauuse ComboStrap\ExceptionSqliteNotAvailable; 904fd306cSNickeauuse ComboStrap\ExecutionContext; 1004fd306cSNickeauuse ComboStrap\FileSystems; 11c3437056SNickeauuse ComboStrap\HttpResponse; 1204fd306cSNickeauuse ComboStrap\HttpResponseStatus; 13c3437056SNickeauuse ComboStrap\Identity; 14c3437056SNickeauuse ComboStrap\LogUtility; 1504fd306cSNickeauuse ComboStrap\MarkupPath; 1604fd306cSNickeauuse ComboStrap\Meta\Field\AliasType; 17c3437056SNickeauuse ComboStrap\Mime; 18c3437056SNickeauuse ComboStrap\PageId; 19c3437056SNickeauuse ComboStrap\PageRules; 20c3437056SNickeauuse ComboStrap\PageUrlPath; 2182a60d03SNickeauuse ComboStrap\PageUrlType; 2204fd306cSNickeauuse ComboStrap\RouterBestEndPage; 23c3437056SNickeauuse ComboStrap\Site; 2404fd306cSNickeauuse ComboStrap\SiteConfig; 25c3437056SNickeauuse ComboStrap\Sqlite; 2604fd306cSNickeauuse ComboStrap\Web\Url; 2754743e42Sgerardnicouse ComboStrap\Web\UrlRewrite; 2804fd306cSNickeauuse ComboStrap\WikiPath; 29c3437056SNickeau 3004fd306cSNickeaurequire_once(__DIR__ . '/../vendor/autoload.php'); 31c3437056SNickeau 32c3437056SNickeau/** 33c3437056SNickeau * Class action_plugin_combo_url 34c3437056SNickeau * 35c3437056SNickeau * The actual URL manager 36c3437056SNickeau * 37c3437056SNickeau * 38c3437056SNickeau */ 39c3437056SNickeauclass action_plugin_combo_router extends DokuWiki_Action_Plugin 40c3437056SNickeau{ 41c3437056SNickeau 42c3437056SNickeau /** 43c3437056SNickeau * @deprecated 44c3437056SNickeau */ 45c3437056SNickeau const URL_MANAGER_ENABLE_CONF = "enableUrlManager"; 46c3437056SNickeau const ROUTER_ENABLE_CONF = "enableRouter"; 47c3437056SNickeau 48c3437056SNickeau // The redirect type 49c3437056SNickeau const REDIRECT_TRANSPARENT_METHOD = 'transparent'; // was (Id) 50c3437056SNickeau // For permanent, see https://developers.google.com/search/docs/advanced/crawling/301-redirects 51c3437056SNickeau const REDIRECT_PERMANENT_METHOD = 'permanent'; // was `Http` (301) 52c3437056SNickeau const REDIRECT_NOTFOUND_METHOD = "notfound"; // 404 (See other) (when best page name is calculated) 53c3437056SNickeau 54c3437056SNickeau public const PERMANENT_REDIRECT_CANONICAL = "permanent:redirect"; 55c3437056SNickeau 56c3437056SNickeau // Where the target id value comes from 57c3437056SNickeau const TARGET_ORIGIN_WELL_KNOWN = 'well-known'; 58c3437056SNickeau const TARGET_ORIGIN_PAGE_RULES = 'pageRules'; 59c3437056SNickeau /** 60c3437056SNickeau * Named Permalink (canonical) 61c3437056SNickeau */ 62c3437056SNickeau const TARGET_ORIGIN_CANONICAL = 'canonical'; 63c3437056SNickeau const TARGET_ORIGIN_ALIAS = 'alias'; 64c3437056SNickeau /** 65c3437056SNickeau * Identifier Permalink (full page id) 66c3437056SNickeau */ 67c3437056SNickeau const TARGET_ORIGIN_PERMALINK = "permalink"; 68c3437056SNickeau /** 69c3437056SNickeau * Extended Permalink (abbreviated page id at the end) 70c3437056SNickeau */ 71c3437056SNickeau const TARGET_ORIGIN_PERMALINK_EXTENDED = "extendedPermalink"; 72c3437056SNickeau const TARGET_ORIGIN_START_PAGE = 'startPage'; 73c3437056SNickeau const TARGET_ORIGIN_BEST_PAGE_NAME = 'bestPageName'; 74c3437056SNickeau const TARGET_ORIGIN_BEST_NAMESPACE = 'bestNamespace'; 75c3437056SNickeau const TARGET_ORIGIN_SEARCH_ENGINE = 'searchEngine'; 76c3437056SNickeau const TARGET_ORIGIN_BEST_END_PAGE_NAME = 'bestEndPageName'; 77c3437056SNickeau const TARGET_ORIGIN_SHADOW_BANNED = "shadowBanned"; 78c3437056SNickeau 79c3437056SNickeau 80c3437056SNickeau // The constant parameters 81c3437056SNickeau const GO_TO_SEARCH_ENGINE = 'GoToSearchEngine'; 82c3437056SNickeau const GO_TO_BEST_NAMESPACE = 'GoToBestNamespace'; 83c3437056SNickeau const GO_TO_BEST_PAGE_NAME = 'GoToBestPageName'; 84c3437056SNickeau const GO_TO_BEST_END_PAGE_NAME = 'GoToBestEndPageName'; 85c3437056SNickeau const GO_TO_NS_START_PAGE = 'GoToNsStartPage'; 86c3437056SNickeau const GO_TO_EDIT_MODE = 'GoToEditMode'; 87c3437056SNickeau const NOTHING = 'Nothing'; 88c3437056SNickeau 89c3437056SNickeau /** @var string - a name used in log and other places */ 90c3437056SNickeau const NAME = 'Url Manager'; 91c3437056SNickeau const CANONICAL = 'router'; 92c3437056SNickeau const PAGE_404 = "<html lang=\"en\"><body></body></html>"; 93c3437056SNickeau const REFRESH_HEADER_NAME = "Refresh"; 94c3437056SNickeau const REFRESH_HEADER_PREFIX = self::REFRESH_HEADER_NAME . ': 0;url='; 9504fd306cSNickeau const LOCATION_HEADER_PREFIX = HttpResponse::LOCATION_HEADER_NAME . ": "; 96c3437056SNickeau public const URL_MANAGER_NAME = "Router"; 97c3437056SNickeau 98c3437056SNickeau 99c3437056SNickeau /** 100c3437056SNickeau * @var PageRules 101c3437056SNickeau */ 102c3437056SNickeau private $pageRules; 103c3437056SNickeau 104c3437056SNickeau 105c3437056SNickeau function __construct() 106c3437056SNickeau { 107c3437056SNickeau // enable direct access to language strings 108c3437056SNickeau // ie $this->lang 109c3437056SNickeau $this->setupLocale(); 110c3437056SNickeau 111c3437056SNickeau } 112c3437056SNickeau 113c3437056SNickeau /** 11404fd306cSNickeau * @param string $refreshHeader 115c3437056SNickeau * @return false|string 116c3437056SNickeau */ 11704fd306cSNickeau public static function getUrlFromRefresh(string $refreshHeader) 118c3437056SNickeau { 119c3437056SNickeau return substr($refreshHeader, strlen(action_plugin_combo_router::REFRESH_HEADER_PREFIX)); 120c3437056SNickeau } 121c3437056SNickeau 122c3437056SNickeau public static function getUrlFromLocation($refreshHeader) 123c3437056SNickeau { 124c3437056SNickeau return substr($refreshHeader, strlen(action_plugin_combo_router::LOCATION_HEADER_PREFIX)); 125c3437056SNickeau } 126c3437056SNickeau 127c3437056SNickeau /** 128c3437056SNickeau * @return array|mixed|string|string[] 129c3437056SNickeau * 130c3437056SNickeau * Unfortunately, DOKUWIKI_STARTED is not the first event 131c3437056SNickeau * The id may have been changed by 132c3437056SNickeau * {@link action_plugin_combo_metalang::load_lang()} 133c3437056SNickeau * function, that's why we have this function 134c3437056SNickeau * to get the original requested id 135c3437056SNickeau */ 136c3437056SNickeau private static function getOriginalIdFromRequest() 137c3437056SNickeau { 138f49666c9Sgerardnico $originalId = $_GET["id"] ?? null; 139f49666c9Sgerardnico if ($originalId === null) { 140f49666c9Sgerardnico return null; 141f49666c9Sgerardnico } 14204fd306cSNickeau return str_replace("/", WikiPath::NAMESPACE_SEPARATOR_DOUBLE_POINT, $originalId); 143c3437056SNickeau } 144c3437056SNickeau 145c3437056SNickeau /** 146c3437056SNickeau * Determine if the request should be banned based on the id 147c3437056SNickeau * 148c3437056SNickeau * @param string $id 149c3437056SNickeau * @return bool 150c3437056SNickeau * 151c3437056SNickeau * See also {@link https://perishablepress.com/7g-firewall/#features} 152c3437056SNickeau * for blocking rules on http request data such as: 153c3437056SNickeau * * query_string 154c3437056SNickeau * * user_agent, 155c3437056SNickeau * * remote host 156c3437056SNickeau */ 157c3437056SNickeau public static function isShadowBanned(string $id): bool 158c3437056SNickeau { 159c3437056SNickeau /** 160c3437056SNickeau * ie 161c3437056SNickeau * wp-json:api:flutter_woo:config_file 162c3437056SNickeau * wp-content:plugins:wpdiscuz:themes:default:style-rtl.css 163c3437056SNickeau * wp-admin 164c3437056SNickeau * 2020:wp-includes:wlwmanifest.xml 165c3437056SNickeau * wp-content:start 166c3437056SNickeau * wp-admin:css:start 167c3437056SNickeau * sito:wp-includes:wlwmanifest.xml 168c3437056SNickeau * site:wp-includes:wlwmanifest.xml 169c3437056SNickeau * cms:wp-includes:wlwmanifest.xml 170c3437056SNickeau * test:wp-includes:wlwmanifest.xml 171c3437056SNickeau * media:wp-includes:wlwmanifest.xml 172c3437056SNickeau * wp2:wp-includes:wlwmanifest.xml 173c3437056SNickeau * 2019:wp-includes:wlwmanifest.xml 174c3437056SNickeau * shop:wp-includes:wlwmanifest.xml 175c3437056SNickeau * wp1:wp-includes:wlwmanifest.xml 176c3437056SNickeau * news:wp-includes:wlwmanifest.xml 177c3437056SNickeau * 2018:wp-includes:wlwmanifest.xml 178c3437056SNickeau */ 179c3437056SNickeau if (strpos($id, 'wp-') !== false) { 180c3437056SNickeau return true; 181c3437056SNickeau } 182c3437056SNickeau 183c3437056SNickeau /** 184c3437056SNickeau * db:oracle:long_or_1_utl_inaddr.get_host_address_chr_33_chr_126_chr_33_chr_65_chr_66_chr_67_chr_49_chr_52_chr_53_chr_90_chr_81_chr_54_chr_50_chr_68_chr_87_chr_81_chr_65_chr_70_chr_80_chr_79_chr_73_chr_89_chr_67_chr_70_chr_68_chr_33_chr_126_chr_33 185c3437056SNickeau * db:oracle:999999.9:union:all:select_null:from_dual 186c3437056SNickeau * db:oracle:999999.9:union:all:select_null:from_dual_and_0_0 187c3437056SNickeau */ 188c3437056SNickeau if (preg_match('/_chr_|_0_0/', $id) === 1) { 189c3437056SNickeau return true; 190c3437056SNickeau } 191c3437056SNickeau 192c3437056SNickeau 193c3437056SNickeau /** 194c3437056SNickeau * ie 195c3437056SNickeau * git:objects: 196c3437056SNickeau * git:refs:heads:stable 197c3437056SNickeau * git:logs:refs:heads:main 198c3437056SNickeau * git:logs:refs:heads:stable 199c3437056SNickeau * git:hooks:pre-push.sample 200c3437056SNickeau * git:hooks:pre-receive.sample 201c3437056SNickeau */ 202c3437056SNickeau if (strpos($id, "git:") === 0) { 203c3437056SNickeau return true; 204c3437056SNickeau } 205c3437056SNickeau 206c3437056SNickeau return false; 207c3437056SNickeau 208c3437056SNickeau } 209c3437056SNickeau 210c3437056SNickeau /** 211c3437056SNickeau * @param string $id 212c3437056SNickeau * @return bool 213c3437056SNickeau * well-known:traffic-advice = https://github.com/buettner/private-prefetch-proxy/blob/main/traffic-advice.md 214c3437056SNickeau * .well-known/security.txt, id=well-known:security.txt = https://securitytxt.org/ 215c3437056SNickeau * well-known:dnt-policy.txt 216c3437056SNickeau */ 217c3437056SNickeau public static function isWellKnownFile(string $id): bool 218c3437056SNickeau { 219c3437056SNickeau return strpos($id, "well-known") === 0; 220c3437056SNickeau } 221c3437056SNickeau 222c3437056SNickeau 223c3437056SNickeau function register(Doku_Event_Handler $controller) 224c3437056SNickeau { 225c3437056SNickeau 22604fd306cSNickeau if (SiteConfig::getConfValue(self::ROUTER_ENABLE_CONF, 1)) { 22704fd306cSNickeau 228c3437056SNickeau /** 229c3437056SNickeau * This will call the function {@link action_plugin_combo_router::_router()} 230c3437056SNickeau * The event is not DOKUWIKI_STARTED because this is not the first one 231c3437056SNickeau * 232c3437056SNickeau * https://www.dokuwiki.org/devel:event:init_lang_load 233c3437056SNickeau */ 234c3437056SNickeau $controller->register_hook('DOKUWIKI_STARTED', 23504fd306cSNickeau 'BEFORE', 236c3437056SNickeau $this, 237c3437056SNickeau 'router', 238c3437056SNickeau array()); 239c3437056SNickeau 240c3437056SNickeau /** 241c3437056SNickeau * This is the real first call of Dokuwiki 242c3437056SNickeau * Unfortunately, it does not create the environment 243c3437056SNickeau * We just ban to spare server resources 244c3437056SNickeau * 245c3437056SNickeau * https://www.dokuwiki.org/devel:event:init_lang_load 246c3437056SNickeau */ 247c3437056SNickeau $controller->register_hook('INIT_LANG_LOAD', 'BEFORE', $this, 'ban', array()); 248c3437056SNickeau 249c3437056SNickeau } 250c3437056SNickeau 251c3437056SNickeau 252c3437056SNickeau } 253c3437056SNickeau 254c3437056SNickeau /** 255c3437056SNickeau * 256c3437056SNickeau * We have created a spacial ban function that is 257c3437056SNickeau * called before the first function 258c3437056SNickeau * {@link action_plugin_combo_metalang::load_lang()} 259c3437056SNickeau * to spare CPU. 260c3437056SNickeau * 261c3437056SNickeau * @param $event 262c3437056SNickeau * @throws Exception 263c3437056SNickeau */ 264c3437056SNickeau function ban(&$event) 265c3437056SNickeau { 266c3437056SNickeau 267c3437056SNickeau $id = self::getOriginalIdFromRequest(); 268*06ecf9e7Sgerardnico if ($id === null) { 269*06ecf9e7Sgerardnico return; 270*06ecf9e7Sgerardnico } 27104fd306cSNickeau $page = MarkupPath::createMarkupFromId($id); 27204fd306cSNickeau if (!FileSystems::exists($page)) { 273c3437056SNickeau // Well known 274c3437056SNickeau if (self::isWellKnownFile($id)) { 275c3437056SNickeau $this->logRedirection($id, "", self::TARGET_ORIGIN_WELL_KNOWN, self::REDIRECT_NOTFOUND_METHOD); 27604fd306cSNickeau ExecutionContext::getActualOrCreateFromEnv() 27704fd306cSNickeau ->response() 27804fd306cSNickeau ->setStatus(HttpResponseStatus::NOT_FOUND) 27904fd306cSNickeau ->end(); 280c3437056SNickeau return; 281c3437056SNickeau } 282c3437056SNickeau 283c3437056SNickeau // Shadow banned 284c3437056SNickeau if (self::isShadowBanned($id)) { 28504fd306cSNickeau $webSiteHomePage = Site::getIndexPageName(); 286c3437056SNickeau $this->executeTransparentRedirect($webSiteHomePage, self::TARGET_ORIGIN_SHADOW_BANNED); 287c3437056SNickeau } 288c3437056SNickeau } 289c3437056SNickeau } 290c3437056SNickeau 291c3437056SNickeau /** 292c3437056SNickeau * @param $event Doku_Event 293c3437056SNickeau * @param $param 294c3437056SNickeau * @return void 295c3437056SNickeau * @throws Exception 296c3437056SNickeau */ 297c3437056SNickeau function router(&$event, $param) 298c3437056SNickeau { 299c3437056SNickeau 30004fd306cSNickeau /** 30104fd306cSNickeau * Just the {@link ExecutionContext::SHOW_ACTION} 30204fd306cSNickeau * may be redirected 30304fd306cSNickeau */ 30404fd306cSNickeau $executionContext = ExecutionContext::getActualOrCreateFromEnv(); 30504fd306cSNickeau if ($executionContext->getExecutingAction() !== ExecutionContext::SHOW_ACTION) { 30604fd306cSNickeau return; 30704fd306cSNickeau } 308c3437056SNickeau 30954743e42Sgerardnico $urlRewrite = Site::getUrlRewrite(); 31054743e42Sgerardnico if ($urlRewrite == UrlRewrite::VALUE_DOKU_REWRITE) { 31154743e42Sgerardnico UrlRewrite::sendErrorMessage(); 31254743e42Sgerardnico return; 31354743e42Sgerardnico } 314c3437056SNickeau 315c3437056SNickeau global $ID; 316c3437056SNickeau 317c3437056SNickeau /** 318c3437056SNickeau * Without SQLite, this module does not work further 319c3437056SNickeau */ 32004fd306cSNickeau try { 32104fd306cSNickeau Sqlite::createOrGetSqlite(); 32204fd306cSNickeau } catch (ExceptionSqliteNotAvailable $e) { 323c3437056SNickeau return; 324c3437056SNickeau } 325c3437056SNickeau 32604fd306cSNickeau $this->pageRules = new PageRules(); 32704fd306cSNickeau 32804fd306cSNickeau 329c3437056SNickeau /** 330c3437056SNickeau * Unfortunately, DOKUWIKI_STARTED is not the first event 331c3437056SNickeau * The id may have been changed by 33204fd306cSNickeau * {@link action_plugin_combo_lang::load_lang()} 333c3437056SNickeau * function, that's why we check against the {@link $_REQUEST} 334c3437056SNickeau * and not the global ID 335c3437056SNickeau */ 336c3437056SNickeau $originalId = self::getOriginalIdFromRequest(); 337c3437056SNickeau 338c3437056SNickeau /** 339c3437056SNickeau * Page is an existing id ? 340c3437056SNickeau */ 34104fd306cSNickeau $requestedMarkupPath = MarkupPath::createMarkupFromId($ID); 34204fd306cSNickeau if (FileSystems::exists($requestedMarkupPath)) { 343c3437056SNickeau 344c3437056SNickeau /** 345c3437056SNickeau * If this is not the root home page 346c3437056SNickeau * and if the canonical id is the not the same, 347c3437056SNickeau * and if this is not a historical page (revision) 348c3437056SNickeau * redirect 349c3437056SNickeau */ 350c3437056SNickeau if ( 35104fd306cSNickeau $originalId !== $requestedMarkupPath->getUrlId() // The id may have been changed 35204fd306cSNickeau && $ID != Site::getIndexPageName() 353c3437056SNickeau && !isset($_REQUEST["rev"]) 354c3437056SNickeau ) { 3554cadd4f8SNickeau /** 3564cadd4f8SNickeau * TODO: When saving for the first time, the page is not stored in the database 3574cadd4f8SNickeau * but that's not the case actually 3584cadd4f8SNickeau */ 35904fd306cSNickeau $databasePageRow = $requestedMarkupPath->getDatabasePage(); 36004fd306cSNickeau if ($databasePageRow->exists()) { 36104fd306cSNickeau /** 36204fd306cSNickeau * A move may leave the database in a bad state, 36304fd306cSNickeau * unfortunately (ie page is not in index, unable to update, ...) 36404fd306cSNickeau * We test therefore if the database page id exists 36504fd306cSNickeau */ 36604fd306cSNickeau $targetPageId = $databasePageRow->getFromRow("id"); 36704fd306cSNickeau $targetPath = WikiPath::createMarkupPathFromId($targetPageId); 36804fd306cSNickeau if (FileSystems::exists($targetPath)) { 369c3437056SNickeau $this->executePermanentRedirect( 37004fd306cSNickeau $requestedMarkupPath->getCanonicalUrl()->toAbsoluteUrlString(), 371c3437056SNickeau self::TARGET_ORIGIN_PERMALINK_EXTENDED 372c3437056SNickeau ); 373c3437056SNickeau } 3744cadd4f8SNickeau } 37504fd306cSNickeau } 376c3437056SNickeau return; 377c3437056SNickeau } 378c3437056SNickeau 379c3437056SNickeau 380c3437056SNickeau $identifier = $ID; 381c3437056SNickeau 382c3437056SNickeau 383c3437056SNickeau /** 384c3437056SNickeau * Page Id Website / root Permalink ? 385c3437056SNickeau */ 38604fd306cSNickeau $shortPageId = PageUrlPath::getShortEncodedPageIdFromUrlId($requestedMarkupPath->getPathObject()->getLastNameWithoutExtension()); 387c3437056SNickeau if ($shortPageId !== null) { 388c3437056SNickeau $pageId = PageUrlPath::decodePageId($shortPageId); 38904fd306cSNickeau if ($requestedMarkupPath->getParent() === null && $pageId !== null) { 39004fd306cSNickeau $page = DatabasePageRow::createFromPageId($pageId)->getMarkupPath(); 391c3437056SNickeau if ($page !== null && $page->exists()) { 39282a60d03SNickeau $this->executePermanentRedirect( 39304fd306cSNickeau $page->getCanonicalUrl()->toAbsoluteUrlString(), 39482a60d03SNickeau self::TARGET_ORIGIN_PERMALINK 39582a60d03SNickeau ); 396c3437056SNickeau } 397c3437056SNickeau } 398c3437056SNickeau 399c3437056SNickeau /** 400c3437056SNickeau * Page Id Abbr ? 401c3437056SNickeau * {@link PageUrlType::CONF_CANONICAL_URL_TYPE} 402c3437056SNickeau */ 403c3437056SNickeau if ( 404c3437056SNickeau $pageId !== null 405c3437056SNickeau ) { 40604fd306cSNickeau $page = DatabasePageRow::createFromPageIdAbbr($pageId)->getMarkupPath(); 407c3437056SNickeau if ($page === null) { 408c3437056SNickeau // or the length of the abbr has changed 40904fd306cSNickeau $canonicalDatabasePage = new DatabasePageRow(); 41004fd306cSNickeau $row = $canonicalDatabasePage->getDatabaseRowFromAttribute("substr(" . PageId::PROPERTY_NAME . ", 1, " . strlen($pageId) . ")", $pageId); 4114cadd4f8SNickeau if ($row !== null) { 41204fd306cSNickeau $canonicalDatabasePage->setRow($row); 41304fd306cSNickeau $page = $canonicalDatabasePage->getMarkupPath(); 414c3437056SNickeau } 415c3437056SNickeau } 416c3437056SNickeau if ($page !== null && $page->exists()) { 417c3437056SNickeau /** 418c3437056SNickeau * If the url canonical id has changed, we show it 419c3437056SNickeau * to the writer by performing a permanent redirect 420c3437056SNickeau */ 421c3437056SNickeau if ($identifier != $page->getUrlId()) { 422c3437056SNickeau // Google asks for a redirect 423c3437056SNickeau // https://developers.google.com/search/docs/advanced/crawling/301-redirects 424c3437056SNickeau // People access your site through several different URLs. 425c3437056SNickeau // If, for example, your home page can be reached in multiple ways 426c3437056SNickeau // (for instance, http://example.com/home, http://home.example.com, or http://www.example.com), 427c3437056SNickeau // it's a good idea to pick one of those URLs as your preferred (canonical) destination, 428c3437056SNickeau // and use redirects to send traffic from the other URLs to your preferred URL. 429c3437056SNickeau $this->executePermanentRedirect( 43004fd306cSNickeau $page->getCanonicalUrl()->toAbsoluteUrlString(), 431c3437056SNickeau self::TARGET_ORIGIN_PERMALINK_EXTENDED 432c3437056SNickeau ); 433c3437056SNickeau return; 434c3437056SNickeau } 4354cadd4f8SNickeau 43604fd306cSNickeau $this->executeTransparentRedirect($page->getWikiId(), self::TARGET_ORIGIN_PERMALINK_EXTENDED); 437c3437056SNickeau return; 438c3437056SNickeau 439c3437056SNickeau } 440c3437056SNickeau // permanent url not yet in the database 441c3437056SNickeau // Other permanent such as permanent canonical ? 442c3437056SNickeau // We let the process go with the new identifier 443c3437056SNickeau 444c3437056SNickeau } 445c3437056SNickeau 446c3437056SNickeau } 447c3437056SNickeau 448c3437056SNickeau // Global variable needed in the process 449c3437056SNickeau global $conf; 450c3437056SNickeau 451c3437056SNickeau /** 452c3437056SNickeau * Identifier is a Canonical ? 453c3437056SNickeau */ 45404fd306cSNickeau $canonicalDatabasePage = DatabasePageRow::createFromCanonical($identifier); 45504fd306cSNickeau $canonicalPage = $canonicalDatabasePage->getMarkupPath(); 45604fd306cSNickeau if ($canonicalPage !== null && $canonicalPage->exists()) { 45782a60d03SNickeau /** 45882a60d03SNickeau * Does the canonical url is canonical name based 45982a60d03SNickeau * ie {@link PageUrlType::CONF_VALUE_CANONICAL_PATH} 46082a60d03SNickeau */ 46104fd306cSNickeau if ($canonicalPage->getUrlId() === $identifier) { 46282a60d03SNickeau $res = $this->executeTransparentRedirect( 46304fd306cSNickeau $canonicalPage->getWikiId(), 46482a60d03SNickeau self::TARGET_ORIGIN_CANONICAL 46582a60d03SNickeau ); 46682a60d03SNickeau } else { 46782a60d03SNickeau $res = $this->executePermanentRedirect( 46804fd306cSNickeau $canonicalPage->getWikiId(), // not the url because, it allows to add url query redirection property 46982a60d03SNickeau self::TARGET_ORIGIN_CANONICAL 47082a60d03SNickeau ); 47182a60d03SNickeau } 472c3437056SNickeau if ($res) { 473c3437056SNickeau return; 474c3437056SNickeau } 475c3437056SNickeau } 476c3437056SNickeau 477c3437056SNickeau /** 478c3437056SNickeau * Identifier is an alias 479c3437056SNickeau */ 48004fd306cSNickeau $aliasRequestedPage = DatabasePageRow::createFromAlias($identifier)->getMarkupPath(); 481c3437056SNickeau if ( 48204fd306cSNickeau $aliasRequestedPage !== null 48304fd306cSNickeau && $aliasRequestedPage->exists() 484c3437056SNickeau // The build alias is the file system metadata alias 485c3437056SNickeau // it may be null if the replication in the database was not successful 48604fd306cSNickeau && $aliasRequestedPage->getBuildAlias() !== null 487c3437056SNickeau ) { 48804fd306cSNickeau $buildAlias = $aliasRequestedPage->getBuildAlias(); 489c3437056SNickeau switch ($buildAlias->getType()) { 490c3437056SNickeau case AliasType::REDIRECT: 49104fd306cSNickeau $res = $this->executePermanentRedirect($aliasRequestedPage->getCanonicalUrl()->toAbsoluteUrlString(), self::TARGET_ORIGIN_ALIAS); 492c3437056SNickeau if ($res) { 493c3437056SNickeau return; 494c3437056SNickeau } 495c3437056SNickeau break; 496c3437056SNickeau case AliasType::SYNONYM: 49704fd306cSNickeau $res = $this->executeTransparentRedirect($aliasRequestedPage->getWikiId(), self::TARGET_ORIGIN_ALIAS); 498c3437056SNickeau if ($res) { 499c3437056SNickeau return; 500c3437056SNickeau } 501c3437056SNickeau break; 502c3437056SNickeau default: 503c3437056SNickeau LogUtility::msg("The alias type ({$buildAlias->getType()}) is unknown. A permanent redirect was performed for the alias $identifier"); 50404fd306cSNickeau $res = $this->executePermanentRedirect($aliasRequestedPage->getCanonicalUrl()->toAbsoluteUrlString(), self::TARGET_ORIGIN_ALIAS); 505c3437056SNickeau if ($res) { 506c3437056SNickeau return; 507c3437056SNickeau } 508c3437056SNickeau break; 509c3437056SNickeau } 510c3437056SNickeau } 511c3437056SNickeau 512c3437056SNickeau 513c3437056SNickeau // If there is a redirection defined in the page rules 514c3437056SNickeau $result = $this->processingPageRules(); 515c3437056SNickeau if ($result) { 516c3437056SNickeau // A redirection has occurred 517c3437056SNickeau // finish the process 518c3437056SNickeau return; 519c3437056SNickeau } 520c3437056SNickeau 521c3437056SNickeau /** 522c3437056SNickeau * 523c3437056SNickeau * There was no redirection found, redirect to edit mode if writer 524c3437056SNickeau * 525c3437056SNickeau */ 526c3437056SNickeau if (Identity::isWriter() && $this->getConf(self::GO_TO_EDIT_MODE) == 1) { 527c3437056SNickeau 528c3437056SNickeau $this->gotToEditMode($event); 529c3437056SNickeau // Stop here 530c3437056SNickeau return; 531c3437056SNickeau 532c3437056SNickeau } 533c3437056SNickeau 53404fd306cSNickeau /** 535c3437056SNickeau * We are still a reader, the redirection does not exist the user is not allowed to edit the page (public of other) 536c3437056SNickeau */ 537c3437056SNickeau if ($this->getConf('ActionReaderFirst') == self::NOTHING) { 538c3437056SNickeau return; 539c3437056SNickeau } 540c3437056SNickeau 541c3437056SNickeau // We are reader and their is no redirection set, we apply the algorithm 542c3437056SNickeau $readerAlgorithms = array(); 543c3437056SNickeau $readerAlgorithms[0] = $this->getConf('ActionReaderFirst'); 544c3437056SNickeau $readerAlgorithms[1] = $this->getConf('ActionReaderSecond'); 545c3437056SNickeau $readerAlgorithms[2] = $this->getConf('ActionReaderThird'); 546c3437056SNickeau 547c3437056SNickeau while ( 548c3437056SNickeau ($algorithm = array_shift($readerAlgorithms)) != null 549c3437056SNickeau ) { 550c3437056SNickeau 551c3437056SNickeau switch ($algorithm) { 552c3437056SNickeau 553c3437056SNickeau case self::NOTHING: 554c3437056SNickeau return; 555c3437056SNickeau 556c3437056SNickeau case self::GO_TO_BEST_END_PAGE_NAME: 557c3437056SNickeau 55804fd306cSNickeau /** 55904fd306cSNickeau * @var MarkupPath $bestEndPage 56004fd306cSNickeau */ 56104fd306cSNickeau list($bestEndPage, $method) = RouterBestEndPage::process($requestedMarkupPath); 56204fd306cSNickeau if ($bestEndPage != null) { 563c3437056SNickeau $res = false; 564c3437056SNickeau switch ($method) { 565c3437056SNickeau case self::REDIRECT_PERMANENT_METHOD: 56604fd306cSNickeau $res = $this->executePermanentRedirect($bestEndPage->getWikiId(), self::TARGET_ORIGIN_BEST_END_PAGE_NAME); 567c3437056SNickeau break; 568c3437056SNickeau case self::REDIRECT_NOTFOUND_METHOD: 56904fd306cSNickeau $res = $this->performNotFoundRedirect($bestEndPage->getWikiId(), self::TARGET_ORIGIN_BEST_END_PAGE_NAME); 570c3437056SNickeau break; 571c3437056SNickeau default: 572c3437056SNickeau LogUtility::msg("This redirection method ($method) was not expected for the redirection algorithm ($algorithm)"); 573c3437056SNickeau } 574c3437056SNickeau if ($res) { 575c3437056SNickeau // Redirection has succeeded 576c3437056SNickeau return; 577c3437056SNickeau } 578c3437056SNickeau } 579c3437056SNickeau break; 580c3437056SNickeau 581c3437056SNickeau case self::GO_TO_NS_START_PAGE: 582c3437056SNickeau 583c3437056SNickeau // Start page with the conf['start'] parameter 584c3437056SNickeau $startPage = getNS($identifier) . ':' . $conf['start']; 585c3437056SNickeau if (page_exists($startPage)) { 586c3437056SNickeau $res = $this->performNotFoundRedirect($startPage, self::TARGET_ORIGIN_START_PAGE); 587c3437056SNickeau if ($res) { 588c3437056SNickeau return; 589c3437056SNickeau } 590c3437056SNickeau } 591c3437056SNickeau 592c3437056SNickeau // Start page with the same name than the namespace 593c3437056SNickeau $startPage = getNS($identifier) . ':' . curNS($identifier); 594c3437056SNickeau if (page_exists($startPage)) { 595c3437056SNickeau $res = $this->performNotFoundRedirect($startPage, self::TARGET_ORIGIN_START_PAGE); 596c3437056SNickeau if ($res) { 597c3437056SNickeau return; 598c3437056SNickeau } 599c3437056SNickeau } 600c3437056SNickeau break; 601c3437056SNickeau 602c3437056SNickeau case self::GO_TO_BEST_PAGE_NAME: 603c3437056SNickeau 604c3437056SNickeau $bestPageId = null; 605c3437056SNickeau 606c3437056SNickeau $bestPage = $this->getBestPage($identifier); 607c3437056SNickeau $bestPageId = $bestPage['id']; 608c3437056SNickeau $scorePageName = $bestPage['score']; 609c3437056SNickeau 610c3437056SNickeau // Get Score from a Namespace 611c3437056SNickeau $bestNamespace = $this->scoreBestNamespace($identifier); 612c3437056SNickeau $bestNamespaceId = $bestNamespace['namespace']; 613c3437056SNickeau $namespaceScore = $bestNamespace['score']; 614c3437056SNickeau 615c3437056SNickeau // Compare the two score 616c3437056SNickeau if ($scorePageName > 0 or $namespaceScore > 0) { 617c3437056SNickeau if ($scorePageName > $namespaceScore) { 618c3437056SNickeau $this->performNotFoundRedirect($bestPageId, self::TARGET_ORIGIN_BEST_PAGE_NAME); 619c3437056SNickeau } else { 620c3437056SNickeau $this->performNotFoundRedirect($bestNamespaceId, self::TARGET_ORIGIN_BEST_PAGE_NAME); 621c3437056SNickeau } 622c3437056SNickeau return; 623c3437056SNickeau } 624c3437056SNickeau break; 625c3437056SNickeau 626c3437056SNickeau case self::GO_TO_BEST_NAMESPACE: 627c3437056SNickeau 628c3437056SNickeau $scoreNamespace = $this->scoreBestNamespace($identifier); 629c3437056SNickeau $bestNamespaceId = $scoreNamespace['namespace']; 630c3437056SNickeau $score = $scoreNamespace['score']; 631c3437056SNickeau 632c3437056SNickeau if ($score > 0) { 633c3437056SNickeau $this->performNotFoundRedirect($bestNamespaceId, self::TARGET_ORIGIN_BEST_NAMESPACE); 634c3437056SNickeau return; 635c3437056SNickeau } 636c3437056SNickeau break; 637c3437056SNickeau 638c3437056SNickeau case self::GO_TO_SEARCH_ENGINE: 639c3437056SNickeau 640c3437056SNickeau $this->redirectToSearchEngine(); 641c3437056SNickeau 642c3437056SNickeau return; 643c3437056SNickeau 644c3437056SNickeau // End Switch Action 645c3437056SNickeau } 646c3437056SNickeau 647c3437056SNickeau // End While Action 648c3437056SNickeau } 649c3437056SNickeau 650c3437056SNickeau 651c3437056SNickeau } 652c3437056SNickeau 653c3437056SNickeau 654c3437056SNickeau /** 655c3437056SNickeau * getBestNamespace 656c3437056SNickeau * Return a list with 'BestNamespaceId Score' 657c3437056SNickeau * @param $id 658c3437056SNickeau * @return array 659c3437056SNickeau */ 660c3437056SNickeau private 661c3437056SNickeau function scoreBestNamespace($id) 662c3437056SNickeau { 663c3437056SNickeau 664c3437056SNickeau global $conf; 665c3437056SNickeau 666c3437056SNickeau // Parameters 667c3437056SNickeau $pageNameSpace = getNS($id); 668c3437056SNickeau 669c3437056SNickeau // If the page has an existing namespace start page take it, other search other namespace 670c3437056SNickeau $startPageNameSpace = $pageNameSpace . ":"; 671c3437056SNickeau $dateAt = ''; 672c3437056SNickeau // $startPageNameSpace will get a full path (ie with start or the namespace 673c3437056SNickeau resolve_pageid($pageNameSpace, $startPageNameSpace, $exists, $dateAt, true); 674c3437056SNickeau if (page_exists($startPageNameSpace)) { 675c3437056SNickeau $nameSpaces = array($startPageNameSpace); 676c3437056SNickeau } else { 677c3437056SNickeau $nameSpaces = ft_pageLookup($conf['start']); 678c3437056SNickeau } 679c3437056SNickeau 680c3437056SNickeau // Parameters and search the best namespace 681c3437056SNickeau $pathNames = explode(':', $pageNameSpace); 682c3437056SNickeau $bestNbWordFound = 0; 683c3437056SNickeau $bestNamespaceId = ''; 684c3437056SNickeau foreach ($nameSpaces as $nameSpace) { 685c3437056SNickeau 686c3437056SNickeau $nbWordFound = 0; 687c3437056SNickeau foreach ($pathNames as $pathName) { 688c3437056SNickeau if (strlen($pathName) > 2) { 689c3437056SNickeau $nbWordFound = $nbWordFound + substr_count($nameSpace, $pathName); 690c3437056SNickeau } 691c3437056SNickeau } 692c3437056SNickeau if ($nbWordFound > $bestNbWordFound) { 693c3437056SNickeau // Take only the smallest namespace 694c3437056SNickeau if (strlen($nameSpace) < strlen($bestNamespaceId) or $nbWordFound > $bestNbWordFound) { 695c3437056SNickeau $bestNbWordFound = $nbWordFound; 696c3437056SNickeau $bestNamespaceId = $nameSpace; 697c3437056SNickeau } 698c3437056SNickeau } 699c3437056SNickeau } 700c3437056SNickeau 701c3437056SNickeau $startPageFactor = $this->getConf('WeightFactorForStartPage'); 702c3437056SNickeau $nameSpaceFactor = $this->getConf('WeightFactorForSameNamespace'); 703c3437056SNickeau if ($bestNbWordFound > 0) { 704c3437056SNickeau $bestNamespaceScore = $bestNbWordFound * $nameSpaceFactor + $startPageFactor; 705c3437056SNickeau } else { 706c3437056SNickeau $bestNamespaceScore = 0; 707c3437056SNickeau } 708c3437056SNickeau 709c3437056SNickeau 710c3437056SNickeau return array( 711c3437056SNickeau 'namespace' => $bestNamespaceId, 712c3437056SNickeau 'score' => $bestNamespaceScore 713c3437056SNickeau ); 714c3437056SNickeau 715c3437056SNickeau } 716c3437056SNickeau 717c3437056SNickeau /** 718c3437056SNickeau * @param $event 719c3437056SNickeau */ 720c3437056SNickeau private 721c3437056SNickeau function gotToEditMode(&$event) 722c3437056SNickeau { 723c3437056SNickeau global $ACT; 724c3437056SNickeau $ACT = 'edit'; 725c3437056SNickeau 726c3437056SNickeau } 727c3437056SNickeau 728c3437056SNickeau 729c3437056SNickeau /** 730c3437056SNickeau * Redirect to an internal page ie: 731c3437056SNickeau * * on the same domain 732c3437056SNickeau * * no HTTP redirect 733c3437056SNickeau * * id rewrite 734c3437056SNickeau * @param string $targetPageId - target page id 735c3437056SNickeau * @param string $targetOriginId - the source of the target (redirect) 736c3437056SNickeau * @return bool - return true if the user has the permission and that the redirect was done 737c3437056SNickeau * @throws Exception 738c3437056SNickeau */ 739c3437056SNickeau private 740c3437056SNickeau function executeTransparentRedirect(string $targetPageId, string $targetOriginId): bool 741c3437056SNickeau { 742c3437056SNickeau /** 743c3437056SNickeau * Because we set the ID globally for the ID redirect 74404fd306cSNickeau * we make sure that this is not a {@link MarkupPath} 745c3437056SNickeau * object otherwise we got an error in the {@link \ComboStrap\AnalyticsMenuItem} 746c3437056SNickeau * because the constructor takes it {@link \dokuwiki\Menu\Item\AbstractItem} 747c3437056SNickeau */ 748c3437056SNickeau if (is_object($targetPageId)) { 749c3437056SNickeau $class = get_class($targetPageId); 750c3437056SNickeau LogUtility::msg("The parameters targetPageId ($targetPageId) is an object of the class ($class) and it should be a page id"); 751c3437056SNickeau } 752c3437056SNickeau 753c3437056SNickeau if (is_object($targetOriginId)) { 754c3437056SNickeau $class = get_class($targetOriginId); 755c3437056SNickeau LogUtility::msg("The parameters targetOriginId ($targetOriginId) is an object of the class ($class) and it should be a page id"); 756c3437056SNickeau } 757c3437056SNickeau 758c3437056SNickeau // If the user does not have the right to see the target page 759c3437056SNickeau // don't do anything 760c3437056SNickeau if (!(Identity::isReader($targetPageId))) { 761c3437056SNickeau return false; 762c3437056SNickeau } 763c3437056SNickeau 764c3437056SNickeau // Change the id 765c3437056SNickeau global $ID; 766c3437056SNickeau global $INFO; 767c3437056SNickeau $sourceId = $ID; 768c3437056SNickeau $ID = $targetPageId; 76904fd306cSNickeau if (isset($_REQUEST["id"])) { 77004fd306cSNickeau $_REQUEST["id"] = $targetPageId; 77104fd306cSNickeau } 77204fd306cSNickeau if (isset($_GET["id"])) { 77304fd306cSNickeau $_GET["id"] = $targetPageId; 77404fd306cSNickeau } 7754cadd4f8SNickeau 776c3437056SNickeau /** 7774cadd4f8SNickeau * Refresh the $INFO data 7784cadd4f8SNickeau * 7794cadd4f8SNickeau * the info attributes are used elsewhere 7804cadd4f8SNickeau * 'id': for the sidebar 7814cadd4f8SNickeau * 'exist' : for the meta robot = noindex,follow, see {@link tpl_metaheaders()} 7824cadd4f8SNickeau * 'rev' : for the edit button to be sure that the page is still the same 783c3437056SNickeau */ 7844cadd4f8SNickeau $INFO = pageinfo(); 785c3437056SNickeau 786c3437056SNickeau /** 787c3437056SNickeau * Not compatible with 788c3437056SNickeau * https://www.dokuwiki.org/config:send404 is enabled 789c3437056SNickeau * 790c3437056SNickeau * This check happens before that dokuwiki is started 791c3437056SNickeau * and send an header in doku.php 792c3437056SNickeau * 793c3437056SNickeau * We send a warning 794c3437056SNickeau */ 795c3437056SNickeau global $conf; 796c3437056SNickeau if ($conf['send404'] == true) { 797c3437056SNickeau LogUtility::msg("The <a href=\"https://www.dokuwiki.org/config:send404\">dokuwiki send404 configuration</a> is on and should be disabled when using the url manager", LogUtility::LVL_MSG_ERROR, self::CANONICAL); 798c3437056SNickeau } 799c3437056SNickeau 800c3437056SNickeau // Redirection 801c3437056SNickeau $this->logRedirection($sourceId, $targetPageId, $targetOriginId, self::REDIRECT_TRANSPARENT_METHOD); 802c3437056SNickeau 803c3437056SNickeau return true; 804c3437056SNickeau 805c3437056SNickeau } 806c3437056SNickeau 80704fd306cSNickeau private function executePermanentRedirect(string $targetIdOrUrl, $targetOrigin): bool 808c3437056SNickeau { 80904fd306cSNickeau return $this->executeHttpRedirect($targetIdOrUrl, $targetOrigin, self::REDIRECT_PERMANENT_METHOD); 810c3437056SNickeau } 811c3437056SNickeau 812c3437056SNickeau /** 813c3437056SNickeau * The general HTTP Redirect method to an internal page 814c3437056SNickeau * where the redirection method decide which type of redirection 81504fd306cSNickeau * @param string $targetIdOrUrl - a dokuwiki id or an url 816c3437056SNickeau * @param string $targetOrigin - the origin of the target (the algorithm used to get the target origin) 817c3437056SNickeau * @param string $method - the redirection method 818c3437056SNickeau */ 819c3437056SNickeau private 82004fd306cSNickeau function executeHttpRedirect(string $targetIdOrUrl, string $targetOrigin, string $method): bool 821c3437056SNickeau { 822c3437056SNickeau 823c3437056SNickeau global $ID; 824c3437056SNickeau 825c3437056SNickeau 826c3437056SNickeau // Log the redirections 82704fd306cSNickeau $this->logRedirection($ID, $targetIdOrUrl, $targetOrigin, $method); 828c3437056SNickeau 829c3437056SNickeau 83004fd306cSNickeau // An http external url ? 83104fd306cSNickeau try { 83204fd306cSNickeau $isValid = Url::createFromString($targetIdOrUrl)->isHttpUrl(); 83304fd306cSNickeau } catch (ExceptionBadSyntax|ExceptionBadArgument $e) { 83404fd306cSNickeau $isValid = false; 83504fd306cSNickeau } 83604fd306cSNickeau 83782a60d03SNickeau // If there is a bug in the isValid function for an internal url 83882a60d03SNickeau // We get a loop. 83982a60d03SNickeau // The Url becomes the id, the id is unknown and we do a redirect again 84082a60d03SNickeau // 84182a60d03SNickeau // We check then if the target starts with the base url 84282a60d03SNickeau // if this is the case, it's valid 84304fd306cSNickeau if (!$isValid && strpos($targetIdOrUrl, DOKU_URL) === 0) { 84482a60d03SNickeau $isValid = true; 84582a60d03SNickeau } 84682a60d03SNickeau if ($isValid) { 847c3437056SNickeau 848c3437056SNickeau // defend against HTTP Response Splitting 849c3437056SNickeau // https://owasp.org/www-community/attacks/HTTP_Response_Splitting 85004fd306cSNickeau $targetUrl = stripctl($targetIdOrUrl); 851c3437056SNickeau 852c3437056SNickeau } else { 853c3437056SNickeau 854c3437056SNickeau 855c3437056SNickeau // Explode the page ID and the anchor (#) 85604fd306cSNickeau $link = explode('#', $targetIdOrUrl, 2); 857c3437056SNickeau 858c3437056SNickeau // Query String to pass the message 859c3437056SNickeau $urlParams = []; 860c3437056SNickeau if ($targetOrigin != self::TARGET_ORIGIN_PERMALINK) { 861c3437056SNickeau $urlParams = array( 862c3437056SNickeau action_plugin_combo_routermessage::ORIGIN_PAGE => $ID, 863c3437056SNickeau action_plugin_combo_routermessage::ORIGIN_TYPE => $targetOrigin 864c3437056SNickeau ); 865c3437056SNickeau } 866c3437056SNickeau 867c3437056SNickeau // if this is search engine redirect 868c3437056SNickeau if ($targetOrigin == self::TARGET_ORIGIN_SEARCH_ENGINE) { 869c3437056SNickeau $replacementPart = array(':', '_', '-'); 870c3437056SNickeau $query = str_replace($replacementPart, ' ', $ID); 871c3437056SNickeau $urlParams["do"] = "search"; 872c3437056SNickeau $urlParams["q"] = $query; 873c3437056SNickeau } 874c3437056SNickeau 875c3437056SNickeau $targetUrl = wl($link[0], $urlParams, true, '&'); 876c3437056SNickeau // %3A back to : 877c3437056SNickeau $targetUrl = str_replace("%3A", ":", $targetUrl); 878c3437056SNickeau if ($link[1]) { 879c3437056SNickeau $targetUrl .= '#' . rawurlencode($link[1]); 880c3437056SNickeau } 881c3437056SNickeau 882c3437056SNickeau } 883c3437056SNickeau 884c3437056SNickeau /** 885c3437056SNickeau * The dokuwiki function {@link send_redirect()} 886c3437056SNickeau * set the `Location header` and in php, the header function 887c3437056SNickeau * in this case change the status code to 302 Arghhhh. 888c3437056SNickeau * The code below is adapted from this function {@link send_redirect()} 889c3437056SNickeau */ 890c3437056SNickeau global $MSG; // are there any undisplayed messages? keep them in session for display 891c3437056SNickeau if (isset($MSG) && count($MSG) && !defined('NOSESSION')) { 892c3437056SNickeau //reopen session, store data and close session again 893c3437056SNickeau @session_start(); 894c3437056SNickeau $_SESSION[DOKU_COOKIE]['msg'] = $MSG; 895c3437056SNickeau } 896c3437056SNickeau session_write_close(); // always close the session 897c3437056SNickeau 898c3437056SNickeau switch ($method) { 899c3437056SNickeau case self::REDIRECT_PERMANENT_METHOD: 90004fd306cSNickeau ExecutionContext::getActualOrCreateFromEnv() 90104fd306cSNickeau ->response() 90204fd306cSNickeau ->setStatus(HttpResponseStatus::PERMANENT_REDIRECT) 903c3437056SNickeau ->addHeader(self::LOCATION_HEADER_PREFIX . $targetUrl) 90404fd306cSNickeau ->end(); 905c3437056SNickeau return true; 906c3437056SNickeau case self::REDIRECT_NOTFOUND_METHOD: 907c3437056SNickeau 908c3437056SNickeau // Empty 404 body to not get the standard 404 page of the browser 909c3437056SNickeau // but a blank page to avoid a sort of FOUC. 910c3437056SNickeau // ie the user see a page briefly 91104fd306cSNickeau ExecutionContext::getActualOrCreateFromEnv() 91204fd306cSNickeau ->response() 91304fd306cSNickeau ->setStatus(HttpResponseStatus::NOT_FOUND) 914c3437056SNickeau ->addHeader(self::REFRESH_HEADER_PREFIX . $targetUrl) 91504fd306cSNickeau ->setBody(self::PAGE_404, Mime::getHtml()) 91604fd306cSNickeau ->end(); 917c3437056SNickeau return true; 918c3437056SNickeau 919c3437056SNickeau default: 920c3437056SNickeau LogUtility::msg("The method ($method) is not an http redirection"); 921c3437056SNickeau return false; 922c3437056SNickeau } 923c3437056SNickeau 924c3437056SNickeau 925c3437056SNickeau } 926c3437056SNickeau 927c3437056SNickeau /** 928c3437056SNickeau * @param $id 929c3437056SNickeau * @return array 930c3437056SNickeau */ 931c3437056SNickeau private 932c3437056SNickeau function getBestPage($id): array 933c3437056SNickeau { 934c3437056SNickeau 935c3437056SNickeau // The return parameters 936c3437056SNickeau $bestPageId = null; 937c3437056SNickeau $scorePageName = null; 938c3437056SNickeau 939c3437056SNickeau // Get Score from a page 940c3437056SNickeau $pageName = noNS($id); 941c3437056SNickeau $pagesWithSameName = ft_pageLookup($pageName); 942c3437056SNickeau if (count($pagesWithSameName) > 0) { 943c3437056SNickeau 944c3437056SNickeau // Search same namespace in the page found than in the Id page asked. 945c3437056SNickeau $bestNbWordFound = 0; 946c3437056SNickeau 947c3437056SNickeau 948c3437056SNickeau $wordsInPageSourceId = explode(':', $id); 949c3437056SNickeau foreach ($pagesWithSameName as $targetPageId => $title) { 950c3437056SNickeau 951c3437056SNickeau // Nb of word found in the target page id 952c3437056SNickeau // that are in the source page id 953c3437056SNickeau $nbWordFound = 0; 954c3437056SNickeau foreach ($wordsInPageSourceId as $word) { 955c3437056SNickeau $nbWordFound = $nbWordFound + substr_count($targetPageId, $word); 956c3437056SNickeau } 957c3437056SNickeau 958c3437056SNickeau if ($bestPageId == null) { 959c3437056SNickeau 960c3437056SNickeau $bestNbWordFound = $nbWordFound; 961c3437056SNickeau $bestPageId = $targetPageId; 962c3437056SNickeau 963c3437056SNickeau } else { 964c3437056SNickeau 965c3437056SNickeau if ($nbWordFound >= $bestNbWordFound && strlen($bestPageId) > strlen($targetPageId)) { 966c3437056SNickeau 967c3437056SNickeau $bestNbWordFound = $nbWordFound; 968c3437056SNickeau $bestPageId = $targetPageId; 969c3437056SNickeau 970c3437056SNickeau } 971c3437056SNickeau 972c3437056SNickeau } 973c3437056SNickeau 974c3437056SNickeau } 975c3437056SNickeau $scorePageName = $this->getConf('WeightFactorForSamePageName') + ($bestNbWordFound - 1) * $this->getConf('WeightFactorForSameNamespace'); 976c3437056SNickeau return array( 977c3437056SNickeau 'id' => $bestPageId, 978c3437056SNickeau 'score' => $scorePageName); 979c3437056SNickeau } 980c3437056SNickeau return array( 981c3437056SNickeau 'id' => $bestPageId, 982c3437056SNickeau 'score' => $scorePageName 983c3437056SNickeau ); 984c3437056SNickeau 985c3437056SNickeau } 986c3437056SNickeau 987c3437056SNickeau 988c3437056SNickeau /** 989c3437056SNickeau * Redirect to the search engine 990c3437056SNickeau */ 991c3437056SNickeau private 992c3437056SNickeau function redirectToSearchEngine() 993c3437056SNickeau { 994c3437056SNickeau 995c3437056SNickeau global $ID; 996c3437056SNickeau $this->performNotFoundRedirect($ID, self::TARGET_ORIGIN_SEARCH_ENGINE); 997c3437056SNickeau 998c3437056SNickeau } 999c3437056SNickeau 1000c3437056SNickeau 1001c3437056SNickeau /** 1002c3437056SNickeau * 1003c3437056SNickeau * * For a conf file, it will update the Redirection Action Data as Referrer, Count Of Redirection, Redirection Date 1004c3437056SNickeau * * For a SQlite database, it will add a row into the log 1005c3437056SNickeau * 1006c3437056SNickeau * @param string $sourcePageId 1007c3437056SNickeau * @param $targetPageId 1008c3437056SNickeau * @param $algorithmic 1009c3437056SNickeau * @param $method - http or rewrite 1010c3437056SNickeau */ 1011c3437056SNickeau function logRedirection(string $sourcePageId, $targetPageId, $algorithmic, $method) 1012c3437056SNickeau { 1013c3437056SNickeau 1014c3437056SNickeau $row = array( 1015c3437056SNickeau "TIMESTAMP" => date("c"), 1016c3437056SNickeau "SOURCE" => $sourcePageId, 1017c3437056SNickeau "TARGET" => $targetPageId, 1018c3437056SNickeau "REFERRER" => $_SERVER['HTTP_REFERER'], 1019c3437056SNickeau "TYPE" => $algorithmic, 1020c3437056SNickeau "METHOD" => $method 1021c3437056SNickeau ); 1022c3437056SNickeau $request = Sqlite::createOrGetBackendSqlite() 1023c3437056SNickeau ->createRequest() 1024c3437056SNickeau ->setTableRow('redirections_log', $row); 1025c3437056SNickeau try { 1026c3437056SNickeau $request 1027c3437056SNickeau ->execute(); 102804fd306cSNickeau } catch (ExceptionCompile $e) { 1029c3437056SNickeau LogUtility::msg("Redirection Log Insert Error. {$e->getMessage()}"); 1030c3437056SNickeau } finally { 1031c3437056SNickeau $request->close(); 1032c3437056SNickeau } 1033c3437056SNickeau 1034c3437056SNickeau 1035c3437056SNickeau } 1036c3437056SNickeau 1037c3437056SNickeau /** 1038c3437056SNickeau * This function check if there is a redirection declared 1039c3437056SNickeau * in the redirection table 1040c3437056SNickeau * @return bool - true if a rewrite or redirection occurs 1041c3437056SNickeau * @throws Exception 1042c3437056SNickeau */ 1043c3437056SNickeau private function processingPageRules(): bool 1044c3437056SNickeau { 1045c3437056SNickeau global $ID; 1046c3437056SNickeau 1047c3437056SNickeau $calculatedTarget = null; 1048c3437056SNickeau $ruleMatcher = null; // Used in a warning message if the target page does not exist 1049c3437056SNickeau // Known redirection in the table 1050c3437056SNickeau // Get the page from redirection data 1051c3437056SNickeau $rules = $this->pageRules->getRules(); 1052c3437056SNickeau foreach ($rules as $rule) { 1053c3437056SNickeau 1054c3437056SNickeau $ruleMatcher = strtolower($rule[PageRules::MATCHER_NAME]); 1055c3437056SNickeau $ruleTarget = $rule[PageRules::TARGET_NAME]; 1056c3437056SNickeau 1057c3437056SNickeau // Glob to Rexgexp 105804fd306cSNickeau $regexpPattern = '/' . str_replace("*", "(.*)", $ruleMatcher) . '/i'; 1059c3437056SNickeau 1060c3437056SNickeau // Match ? 1061c3437056SNickeau // https://www.php.net/manual/en/function.preg-match.php 106286867397Sgerardnico $pregMatchResult = @preg_match($regexpPattern, $ID, $matches); 106386867397Sgerardnico if ($pregMatchResult === false) { 106486867397Sgerardnico // The `if` to take into account this problem 106586867397Sgerardnico // PHP Warning: preg_match(): Unknown modifier 'd' in /opt/www/datacadamia.com/lib/plugins/combo/action/router.php on line 972 106686867397Sgerardnico LogUtility::log2file("processing Page Rules An error occurred with the pattern ($regexpPattern)", LogUtility::LVL_MSG_WARNING); 106786867397Sgerardnico return false; 106886867397Sgerardnico } 106986867397Sgerardnico if ($pregMatchResult) { 1070c3437056SNickeau $calculatedTarget = $ruleTarget; 1071c3437056SNickeau foreach ($matches as $key => $match) { 1072c3437056SNickeau if ($key == 0) { 1073c3437056SNickeau continue; 1074c3437056SNickeau } else { 1075c3437056SNickeau $calculatedTarget = str_replace('$' . $key, $match, $calculatedTarget); 1076c3437056SNickeau } 1077c3437056SNickeau } 1078c3437056SNickeau break; 1079c3437056SNickeau } 1080c3437056SNickeau } 1081c3437056SNickeau 1082c3437056SNickeau if ($calculatedTarget == null) { 1083c3437056SNickeau return false; 1084c3437056SNickeau } 1085c3437056SNickeau 1086c3437056SNickeau // If this is an external redirect (other domain) 108704fd306cSNickeau try { 108804fd306cSNickeau $isHttpUrl = Url::createFromString($calculatedTarget)->isHttpUrl(); 108904fd306cSNickeau } catch (ExceptionBadSyntax $e) { 109004fd306cSNickeau $isHttpUrl = false; 109104fd306cSNickeau } 109204fd306cSNickeau if ($isHttpUrl) { 1093c3437056SNickeau $this->executeHttpRedirect($calculatedTarget, self::TARGET_ORIGIN_PAGE_RULES, self::REDIRECT_PERMANENT_METHOD); 1094c3437056SNickeau return true; 1095c3437056SNickeau } 1096c3437056SNickeau 1097c3437056SNickeau // If the page exist 1098c3437056SNickeau if (page_exists($calculatedTarget)) { 1099c3437056SNickeau 1100c3437056SNickeau // This is DokuWiki Id and should always be lowercase 1101c3437056SNickeau // The page rule may have change that 1102c3437056SNickeau $calculatedTarget = strtolower($calculatedTarget); 1103d0abe7fcSgerardnico $res = $this->executeHttpRedirect($calculatedTarget, self::TARGET_ORIGIN_PAGE_RULES, self::REDIRECT_PERMANENT_METHOD); 1104c3437056SNickeau if ($res) { 1105c3437056SNickeau return true; 1106c3437056SNickeau } else { 1107c3437056SNickeau return false; 1108c3437056SNickeau } 1109c3437056SNickeau 1110c3437056SNickeau } else { 1111c3437056SNickeau 1112c3437056SNickeau LogUtility::msg("The calculated target page ($calculatedTarget) (for the non-existing page `$ID` with the matcher `$ruleMatcher`) does not exist", LogUtility::LVL_MSG_ERROR); 1113c3437056SNickeau return false; 1114c3437056SNickeau 1115c3437056SNickeau } 1116c3437056SNickeau 1117c3437056SNickeau } 1118c3437056SNickeau 1119c3437056SNickeau private function performNotFoundRedirect(string $targetId, string $origin): bool 1120c3437056SNickeau { 1121c3437056SNickeau return $this->executeHttpRedirect($targetId, $origin, self::REDIRECT_NOTFOUND_METHOD); 1122c3437056SNickeau } 1123c3437056SNickeau 1124c3437056SNickeau 1125c3437056SNickeau} 1126