1c3437056SNickeau<?php 2c3437056SNickeau 3c3437056SNickeau 4c3437056SNickeauuse ComboStrap\DatabasePageRow; 511d09b86Sgerardnicouse ComboStrap\DokuwikiId; 604fd306cSNickeauuse ComboStrap\ExceptionBadArgument; 704fd306cSNickeauuse ComboStrap\ExceptionBadSyntax; 804fd306cSNickeauuse ComboStrap\ExceptionCompile; 904fd306cSNickeauuse ComboStrap\ExceptionSqliteNotAvailable; 1004fd306cSNickeauuse ComboStrap\ExecutionContext; 1104fd306cSNickeauuse ComboStrap\FileSystems; 12c3437056SNickeauuse ComboStrap\HttpResponse; 1304fd306cSNickeauuse ComboStrap\HttpResponseStatus; 14c3437056SNickeauuse ComboStrap\Identity; 15c3437056SNickeauuse ComboStrap\LogUtility; 1604fd306cSNickeauuse ComboStrap\MarkupPath; 1704fd306cSNickeauuse ComboStrap\Meta\Field\AliasType; 18c3437056SNickeauuse ComboStrap\Mime; 19c3437056SNickeauuse ComboStrap\PageId; 20c3437056SNickeauuse ComboStrap\PageRules; 21c3437056SNickeauuse ComboStrap\PageUrlPath; 2282a60d03SNickeauuse ComboStrap\PageUrlType; 2304fd306cSNickeauuse ComboStrap\RouterBestEndPage; 24c3437056SNickeauuse ComboStrap\Site; 2504fd306cSNickeauuse ComboStrap\SiteConfig; 26c3437056SNickeauuse ComboStrap\Sqlite; 2704fd306cSNickeauuse ComboStrap\Web\Url; 2811d09b86Sgerardnicouse ComboStrap\Web\UrlEndpoint; 2954743e42Sgerardnicouse ComboStrap\Web\UrlRewrite; 3004fd306cSNickeauuse ComboStrap\WikiPath; 31c3437056SNickeau 3204fd306cSNickeaurequire_once(__DIR__ . '/../vendor/autoload.php'); 33c3437056SNickeau 34c3437056SNickeau/** 35c3437056SNickeau * Class action_plugin_combo_url 36c3437056SNickeau * 37c3437056SNickeau * The actual URL manager 38c3437056SNickeau * 39c3437056SNickeau * 40c3437056SNickeau */ 41c3437056SNickeauclass action_plugin_combo_router extends DokuWiki_Action_Plugin 42c3437056SNickeau{ 43c3437056SNickeau 44c3437056SNickeau /** 45c3437056SNickeau * @deprecated 46c3437056SNickeau */ 47c3437056SNickeau const URL_MANAGER_ENABLE_CONF = "enableUrlManager"; 48c3437056SNickeau const ROUTER_ENABLE_CONF = "enableRouter"; 49c3437056SNickeau 50c3437056SNickeau // The redirect type 51c3437056SNickeau const REDIRECT_TRANSPARENT_METHOD = 'transparent'; // was (Id) 52c3437056SNickeau // For permanent, see https://developers.google.com/search/docs/advanced/crawling/301-redirects 53c3437056SNickeau const REDIRECT_PERMANENT_METHOD = 'permanent'; // was `Http` (301) 54c3437056SNickeau const REDIRECT_NOTFOUND_METHOD = "notfound"; // 404 (See other) (when best page name is calculated) 55c3437056SNickeau 56c3437056SNickeau public const PERMANENT_REDIRECT_CANONICAL = "permanent:redirect"; 57c3437056SNickeau 58c3437056SNickeau // Where the target id value comes from 59c3437056SNickeau const TARGET_ORIGIN_WELL_KNOWN = 'well-known'; 60c3437056SNickeau const TARGET_ORIGIN_PAGE_RULES = 'pageRules'; 61c3437056SNickeau /** 62c3437056SNickeau * Named Permalink (canonical) 63c3437056SNickeau */ 64c3437056SNickeau const TARGET_ORIGIN_CANONICAL = 'canonical'; 65c3437056SNickeau const TARGET_ORIGIN_ALIAS = 'alias'; 66c3437056SNickeau /** 67c3437056SNickeau * Identifier Permalink (full page id) 68c3437056SNickeau */ 69c3437056SNickeau const TARGET_ORIGIN_PERMALINK = "permalink"; 70c3437056SNickeau /** 71c3437056SNickeau * Extended Permalink (abbreviated page id at the end) 72c3437056SNickeau */ 73c3437056SNickeau const TARGET_ORIGIN_PERMALINK_EXTENDED = "extendedPermalink"; 74c3437056SNickeau const TARGET_ORIGIN_START_PAGE = 'startPage'; 75c3437056SNickeau const TARGET_ORIGIN_BEST_PAGE_NAME = 'bestPageName'; 76c3437056SNickeau const TARGET_ORIGIN_BEST_NAMESPACE = 'bestNamespace'; 77c3437056SNickeau const TARGET_ORIGIN_SEARCH_ENGINE = 'searchEngine'; 78c3437056SNickeau const TARGET_ORIGIN_BEST_END_PAGE_NAME = 'bestEndPageName'; 79c3437056SNickeau const TARGET_ORIGIN_SHADOW_BANNED = "shadowBanned"; 80c3437056SNickeau 81c3437056SNickeau 82c3437056SNickeau // The constant parameters 83c3437056SNickeau const GO_TO_SEARCH_ENGINE = 'GoToSearchEngine'; 84c3437056SNickeau const GO_TO_BEST_NAMESPACE = 'GoToBestNamespace'; 85c3437056SNickeau const GO_TO_BEST_PAGE_NAME = 'GoToBestPageName'; 86c3437056SNickeau const GO_TO_BEST_END_PAGE_NAME = 'GoToBestEndPageName'; 87c3437056SNickeau const GO_TO_NS_START_PAGE = 'GoToNsStartPage'; 88c3437056SNickeau const GO_TO_EDIT_MODE = 'GoToEditMode'; 89c3437056SNickeau const NOTHING = 'Nothing'; 90c3437056SNickeau 91c3437056SNickeau /** @var string - a name used in log and other places */ 92c3437056SNickeau const NAME = 'Url Manager'; 93c3437056SNickeau const CANONICAL = 'router'; 94c3437056SNickeau const PAGE_404 = "<html lang=\"en\"><body></body></html>"; 95c3437056SNickeau const REFRESH_HEADER_NAME = "Refresh"; 96c3437056SNickeau const REFRESH_HEADER_PREFIX = self::REFRESH_HEADER_NAME . ': 0;url='; 9704fd306cSNickeau const LOCATION_HEADER_PREFIX = HttpResponse::LOCATION_HEADER_NAME . ": "; 98c3437056SNickeau public const URL_MANAGER_NAME = "Router"; 99c3437056SNickeau 100c3437056SNickeau 101c3437056SNickeau /** 102c3437056SNickeau * @var PageRules 103c3437056SNickeau */ 104c3437056SNickeau private $pageRules; 105c3437056SNickeau 106c3437056SNickeau 107c3437056SNickeau function __construct() 108c3437056SNickeau { 109c3437056SNickeau // enable direct access to language strings 110c3437056SNickeau // ie $this->lang 111c3437056SNickeau $this->setupLocale(); 112c3437056SNickeau 113c3437056SNickeau } 114c3437056SNickeau 115c3437056SNickeau /** 11604fd306cSNickeau * @param string $refreshHeader 117c3437056SNickeau * @return false|string 118c3437056SNickeau */ 11904fd306cSNickeau public static function getUrlFromRefresh(string $refreshHeader) 120c3437056SNickeau { 121c3437056SNickeau return substr($refreshHeader, strlen(action_plugin_combo_router::REFRESH_HEADER_PREFIX)); 122c3437056SNickeau } 123c3437056SNickeau 124c3437056SNickeau public static function getUrlFromLocation($refreshHeader) 125c3437056SNickeau { 126c3437056SNickeau return substr($refreshHeader, strlen(action_plugin_combo_router::LOCATION_HEADER_PREFIX)); 127c3437056SNickeau } 128c3437056SNickeau 129c3437056SNickeau /** 130*ad79af66SNico * @return string|null 131c3437056SNickeau * 132c3437056SNickeau * Unfortunately, DOKUWIKI_STARTED is not the first event 133c3437056SNickeau * The id may have been changed by 134*ad79af66SNico * {@link action_plugin_combo_lang::load_lang()} 135c3437056SNickeau * function, that's why we have this function 136c3437056SNickeau * to get the original requested id 137c3437056SNickeau */ 138*ad79af66SNico private static function getOriginalIdFromRequest(): ?string 139c3437056SNickeau { 140*ad79af66SNico // and not $_GET["id"] otherwise we may get a `/` 141*ad79af66SNico $originalId = getID(); 142f49666c9Sgerardnico if ($originalId === null) { 143f49666c9Sgerardnico return null; 144f49666c9Sgerardnico } 14504fd306cSNickeau return str_replace("/", WikiPath::NAMESPACE_SEPARATOR_DOUBLE_POINT, $originalId); 146c3437056SNickeau } 147c3437056SNickeau 148c3437056SNickeau /** 149c3437056SNickeau * Determine if the request should be banned based on the id 150c3437056SNickeau * 151c3437056SNickeau * @param string $id 152c3437056SNickeau * @return bool 153c3437056SNickeau * 154c3437056SNickeau * See also {@link https://perishablepress.com/7g-firewall/#features} 155c3437056SNickeau * for blocking rules on http request data such as: 156c3437056SNickeau * * query_string 157c3437056SNickeau * * user_agent, 158c3437056SNickeau * * remote host 159c3437056SNickeau */ 160c3437056SNickeau public static function isShadowBanned(string $id): bool 161c3437056SNickeau { 162c3437056SNickeau /** 163c3437056SNickeau * ie 164c3437056SNickeau * wp-json:api:flutter_woo:config_file 165c3437056SNickeau * wp-content:plugins:wpdiscuz:themes:default:style-rtl.css 166c3437056SNickeau * wp-admin 167c3437056SNickeau * 2020:wp-includes:wlwmanifest.xml 168c3437056SNickeau * wp-content:start 169c3437056SNickeau * wp-admin:css:start 170c3437056SNickeau * sito:wp-includes:wlwmanifest.xml 171c3437056SNickeau * site:wp-includes:wlwmanifest.xml 172c3437056SNickeau * cms:wp-includes:wlwmanifest.xml 173c3437056SNickeau * test:wp-includes:wlwmanifest.xml 174c3437056SNickeau * media:wp-includes:wlwmanifest.xml 175c3437056SNickeau * wp2:wp-includes:wlwmanifest.xml 176c3437056SNickeau * 2019:wp-includes:wlwmanifest.xml 177c3437056SNickeau * shop:wp-includes:wlwmanifest.xml 178c3437056SNickeau * wp1:wp-includes:wlwmanifest.xml 179c3437056SNickeau * news:wp-includes:wlwmanifest.xml 180c3437056SNickeau * 2018:wp-includes:wlwmanifest.xml 181c3437056SNickeau */ 182c3437056SNickeau if (strpos($id, 'wp-') !== false) { 183c3437056SNickeau return true; 184c3437056SNickeau } 185c3437056SNickeau 186c3437056SNickeau /** 187c3437056SNickeau * 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 188c3437056SNickeau * db:oracle:999999.9:union:all:select_null:from_dual 189c3437056SNickeau * db:oracle:999999.9:union:all:select_null:from_dual_and_0_0 190c3437056SNickeau */ 191c3437056SNickeau if (preg_match('/_chr_|_0_0/', $id) === 1) { 192c3437056SNickeau return true; 193c3437056SNickeau } 194c3437056SNickeau 195c3437056SNickeau 196c3437056SNickeau /** 197c3437056SNickeau * ie 198c3437056SNickeau * git:objects: 199c3437056SNickeau * git:refs:heads:stable 200c3437056SNickeau * git:logs:refs:heads:main 201c3437056SNickeau * git:logs:refs:heads:stable 202c3437056SNickeau * git:hooks:pre-push.sample 203c3437056SNickeau * git:hooks:pre-receive.sample 204c3437056SNickeau */ 205c3437056SNickeau if (strpos($id, "git:") === 0) { 206c3437056SNickeau return true; 207c3437056SNickeau } 208c3437056SNickeau 209c3437056SNickeau return false; 210c3437056SNickeau 211c3437056SNickeau } 212c3437056SNickeau 213c3437056SNickeau /** 214c3437056SNickeau * @param string $id 215c3437056SNickeau * @return bool 216c3437056SNickeau * well-known:traffic-advice = https://github.com/buettner/private-prefetch-proxy/blob/main/traffic-advice.md 217c3437056SNickeau * .well-known/security.txt, id=well-known:security.txt = https://securitytxt.org/ 218c3437056SNickeau * well-known:dnt-policy.txt 219c3437056SNickeau */ 220c3437056SNickeau public static function isWellKnownFile(string $id): bool 221c3437056SNickeau { 222c3437056SNickeau return strpos($id, "well-known") === 0; 223c3437056SNickeau } 224c3437056SNickeau 225c3437056SNickeau 226c3437056SNickeau function register(Doku_Event_Handler $controller) 227c3437056SNickeau { 228c3437056SNickeau 22904fd306cSNickeau if (SiteConfig::getConfValue(self::ROUTER_ENABLE_CONF, 1)) { 23004fd306cSNickeau 231c3437056SNickeau /** 232c3437056SNickeau * This will call the function {@link action_plugin_combo_router::_router()} 233c3437056SNickeau * The event is not DOKUWIKI_STARTED because this is not the first one 234c3437056SNickeau * 235c3437056SNickeau * https://www.dokuwiki.org/devel:event:init_lang_load 236c3437056SNickeau */ 237c3437056SNickeau $controller->register_hook('DOKUWIKI_STARTED', 23804fd306cSNickeau 'BEFORE', 239c3437056SNickeau $this, 240c3437056SNickeau 'router', 241c3437056SNickeau array()); 242c3437056SNickeau 243c3437056SNickeau /** 244c3437056SNickeau * This is the real first call of Dokuwiki 245c3437056SNickeau * Unfortunately, it does not create the environment 246c3437056SNickeau * We just ban to spare server resources 247c3437056SNickeau * 248c3437056SNickeau * https://www.dokuwiki.org/devel:event:init_lang_load 249c3437056SNickeau */ 250c3437056SNickeau $controller->register_hook('INIT_LANG_LOAD', 'BEFORE', $this, 'ban', array()); 251c3437056SNickeau 252c3437056SNickeau } 253c3437056SNickeau 254c3437056SNickeau 255c3437056SNickeau } 256c3437056SNickeau 257c3437056SNickeau /** 258c3437056SNickeau * 259c3437056SNickeau * We have created a spacial ban function that is 260c3437056SNickeau * called before the first function 261c3437056SNickeau * {@link action_plugin_combo_metalang::load_lang()} 262c3437056SNickeau * to spare CPU. 263c3437056SNickeau * 264c3437056SNickeau * @param $event 265c3437056SNickeau * @throws Exception 266c3437056SNickeau */ 267c3437056SNickeau function ban(&$event) 268c3437056SNickeau { 269c3437056SNickeau 270c3437056SNickeau $id = self::getOriginalIdFromRequest(); 27106ecf9e7Sgerardnico if ($id === null) { 27206ecf9e7Sgerardnico return; 27306ecf9e7Sgerardnico } 27404fd306cSNickeau $page = MarkupPath::createMarkupFromId($id); 27504fd306cSNickeau if (!FileSystems::exists($page)) { 276c3437056SNickeau // Well known 277c3437056SNickeau if (self::isWellKnownFile($id)) { 278c3437056SNickeau $this->logRedirection($id, "", self::TARGET_ORIGIN_WELL_KNOWN, self::REDIRECT_NOTFOUND_METHOD); 27904fd306cSNickeau ExecutionContext::getActualOrCreateFromEnv() 28004fd306cSNickeau ->response() 28104fd306cSNickeau ->setStatus(HttpResponseStatus::NOT_FOUND) 28204fd306cSNickeau ->end(); 283c3437056SNickeau return; 284c3437056SNickeau } 285c3437056SNickeau 286c3437056SNickeau // Shadow banned 287c3437056SNickeau if (self::isShadowBanned($id)) { 28804fd306cSNickeau $webSiteHomePage = Site::getIndexPageName(); 289c3437056SNickeau $this->executeTransparentRedirect($webSiteHomePage, self::TARGET_ORIGIN_SHADOW_BANNED); 290c3437056SNickeau } 291c3437056SNickeau } 292c3437056SNickeau } 293c3437056SNickeau 294c3437056SNickeau /** 295c3437056SNickeau * @param $event Doku_Event 296c3437056SNickeau * @param $param 297c3437056SNickeau * @return void 298c3437056SNickeau * @throws Exception 299c3437056SNickeau */ 300c3437056SNickeau function router(&$event, $param) 301c3437056SNickeau { 302c3437056SNickeau 30304fd306cSNickeau /** 30404fd306cSNickeau * Just the {@link ExecutionContext::SHOW_ACTION} 30504fd306cSNickeau * may be redirected 30604fd306cSNickeau */ 30704fd306cSNickeau $executionContext = ExecutionContext::getActualOrCreateFromEnv(); 30804fd306cSNickeau if ($executionContext->getExecutingAction() !== ExecutionContext::SHOW_ACTION) { 30904fd306cSNickeau return; 31004fd306cSNickeau } 311c3437056SNickeau 31254743e42Sgerardnico $urlRewrite = Site::getUrlRewrite(); 31354743e42Sgerardnico if ($urlRewrite == UrlRewrite::VALUE_DOKU_REWRITE) { 31454743e42Sgerardnico UrlRewrite::sendErrorMessage(); 31554743e42Sgerardnico return; 31654743e42Sgerardnico } 317c3437056SNickeau 318c3437056SNickeau global $ID; 319c3437056SNickeau 320c3437056SNickeau /** 321c3437056SNickeau * Without SQLite, this module does not work further 322c3437056SNickeau */ 32304fd306cSNickeau try { 32404fd306cSNickeau Sqlite::createOrGetSqlite(); 32504fd306cSNickeau } catch (ExceptionSqliteNotAvailable $e) { 326c3437056SNickeau return; 327c3437056SNickeau } 328c3437056SNickeau 32904fd306cSNickeau $this->pageRules = new PageRules(); 33004fd306cSNickeau 33104fd306cSNickeau 332c3437056SNickeau /** 333c3437056SNickeau * Unfortunately, DOKUWIKI_STARTED is not the first event 334c3437056SNickeau * The id may have been changed by 33504fd306cSNickeau * {@link action_plugin_combo_lang::load_lang()} 336c3437056SNickeau * function, that's why we check against the {@link $_REQUEST} 337c3437056SNickeau * and not the global ID 338c3437056SNickeau */ 339c3437056SNickeau $originalId = self::getOriginalIdFromRequest(); 340c3437056SNickeau 341c3437056SNickeau /** 342c3437056SNickeau * Page is an existing id ? 343c3437056SNickeau */ 34404fd306cSNickeau $requestedMarkupPath = MarkupPath::createMarkupFromId($ID); 34504fd306cSNickeau if (FileSystems::exists($requestedMarkupPath)) { 346c3437056SNickeau 347c3437056SNickeau /** 348c3437056SNickeau * If this is not the root home page 349c3437056SNickeau * and if the canonical id is the not the same, 350c3437056SNickeau * and if this is not a historical page (revision) 351c3437056SNickeau * redirect 352c3437056SNickeau */ 353c3437056SNickeau if ( 35404fd306cSNickeau $originalId !== $requestedMarkupPath->getUrlId() // The id may have been changed 35504fd306cSNickeau && $ID != Site::getIndexPageName() 356c3437056SNickeau && !isset($_REQUEST["rev"]) 357c3437056SNickeau ) { 3584cadd4f8SNickeau /** 3594cadd4f8SNickeau * TODO: When saving for the first time, the page is not stored in the database 3604cadd4f8SNickeau * but that's not the case actually 3614cadd4f8SNickeau */ 36204fd306cSNickeau $databasePageRow = $requestedMarkupPath->getDatabasePage(); 36304fd306cSNickeau if ($databasePageRow->exists()) { 36404fd306cSNickeau /** 36504fd306cSNickeau * A move may leave the database in a bad state, 36604fd306cSNickeau * unfortunately (ie page is not in index, unable to update, ...) 36704fd306cSNickeau * We test therefore if the database page id exists 36804fd306cSNickeau */ 36904fd306cSNickeau $targetPageId = $databasePageRow->getFromRow("id"); 37004fd306cSNickeau $targetPath = WikiPath::createMarkupPathFromId($targetPageId); 37104fd306cSNickeau if (FileSystems::exists($targetPath)) { 372c3437056SNickeau $this->executePermanentRedirect( 37304fd306cSNickeau $requestedMarkupPath->getCanonicalUrl()->toAbsoluteUrlString(), 374c3437056SNickeau self::TARGET_ORIGIN_PERMALINK_EXTENDED 375c3437056SNickeau ); 376c3437056SNickeau } 3774cadd4f8SNickeau } 37804fd306cSNickeau } 379c3437056SNickeau return; 380c3437056SNickeau } 381c3437056SNickeau 382c3437056SNickeau 383c3437056SNickeau $identifier = $ID; 384c3437056SNickeau 385c3437056SNickeau 386c3437056SNickeau /** 3871089b853Sgerardnico * Page Id in the url 388c3437056SNickeau */ 38904fd306cSNickeau $shortPageId = PageUrlPath::getShortEncodedPageIdFromUrlId($requestedMarkupPath->getPathObject()->getLastNameWithoutExtension()); 3901089b853Sgerardnico if ($shortPageId != null) { 391c3437056SNickeau $pageId = PageUrlPath::decodePageId($shortPageId); 3921089b853Sgerardnico } else { 3931089b853Sgerardnico /** 3941089b853Sgerardnico * Permalink with id 3951089b853Sgerardnico */ 3961089b853Sgerardnico $pageId = PageUrlPath::decodePageId($identifier); 3971089b853Sgerardnico } 3981089b853Sgerardnico if ($pageId !== null) { 3991089b853Sgerardnico 4001089b853Sgerardnico if ($requestedMarkupPath->getParent() === null) { 40104fd306cSNickeau $page = DatabasePageRow::createFromPageId($pageId)->getMarkupPath(); 402c3437056SNickeau if ($page !== null && $page->exists()) { 40382a60d03SNickeau $this->executePermanentRedirect( 40404fd306cSNickeau $page->getCanonicalUrl()->toAbsoluteUrlString(), 40582a60d03SNickeau self::TARGET_ORIGIN_PERMALINK 40682a60d03SNickeau ); 4071089b853Sgerardnico return; 408c3437056SNickeau } 409c3437056SNickeau } 410c3437056SNickeau 411c3437056SNickeau /** 412c3437056SNickeau * Page Id Abbr ? 413c3437056SNickeau * {@link PageUrlType::CONF_CANONICAL_URL_TYPE} 414c3437056SNickeau */ 41504fd306cSNickeau $page = DatabasePageRow::createFromPageIdAbbr($pageId)->getMarkupPath(); 416c3437056SNickeau if ($page === null) { 417c3437056SNickeau // or the length of the abbr has changed 41804fd306cSNickeau $canonicalDatabasePage = new DatabasePageRow(); 41904fd306cSNickeau $row = $canonicalDatabasePage->getDatabaseRowFromAttribute("substr(" . PageId::PROPERTY_NAME . ", 1, " . strlen($pageId) . ")", $pageId); 4204cadd4f8SNickeau if ($row !== null) { 42104fd306cSNickeau $canonicalDatabasePage->setRow($row); 42204fd306cSNickeau $page = $canonicalDatabasePage->getMarkupPath(); 423c3437056SNickeau } 424c3437056SNickeau } 425c3437056SNickeau if ($page !== null && $page->exists()) { 426c3437056SNickeau /** 427c3437056SNickeau * If the url canonical id has changed, we show it 428c3437056SNickeau * to the writer by performing a permanent redirect 429c3437056SNickeau */ 430c3437056SNickeau if ($identifier != $page->getUrlId()) { 431c3437056SNickeau // Google asks for a redirect 432c3437056SNickeau // https://developers.google.com/search/docs/advanced/crawling/301-redirects 433c3437056SNickeau // People access your site through several different URLs. 434c3437056SNickeau // If, for example, your home page can be reached in multiple ways 435c3437056SNickeau // (for instance, http://example.com/home, http://home.example.com, or http://www.example.com), 436c3437056SNickeau // it's a good idea to pick one of those URLs as your preferred (canonical) destination, 437c3437056SNickeau // and use redirects to send traffic from the other URLs to your preferred URL. 438c3437056SNickeau $this->executePermanentRedirect( 43904fd306cSNickeau $page->getCanonicalUrl()->toAbsoluteUrlString(), 440c3437056SNickeau self::TARGET_ORIGIN_PERMALINK_EXTENDED 441c3437056SNickeau ); 442c3437056SNickeau return; 443c3437056SNickeau } 4444cadd4f8SNickeau 44504fd306cSNickeau $this->executeTransparentRedirect($page->getWikiId(), self::TARGET_ORIGIN_PERMALINK_EXTENDED); 446c3437056SNickeau return; 447c3437056SNickeau 448c3437056SNickeau } 449c3437056SNickeau // permanent url not yet in the database 450c3437056SNickeau // Other permanent such as permanent canonical ? 451c3437056SNickeau // We let the process go with the new identifier 452c3437056SNickeau 453c3437056SNickeau } 454c3437056SNickeau 455c3437056SNickeau // Global variable needed in the process 456c3437056SNickeau global $conf; 457c3437056SNickeau 458c3437056SNickeau /** 459c3437056SNickeau * Identifier is a Canonical ? 460c3437056SNickeau */ 46104fd306cSNickeau $canonicalDatabasePage = DatabasePageRow::createFromCanonical($identifier); 46204fd306cSNickeau $canonicalPage = $canonicalDatabasePage->getMarkupPath(); 46304fd306cSNickeau if ($canonicalPage !== null && $canonicalPage->exists()) { 46482a60d03SNickeau /** 46582a60d03SNickeau * Does the canonical url is canonical name based 46682a60d03SNickeau * ie {@link PageUrlType::CONF_VALUE_CANONICAL_PATH} 46782a60d03SNickeau */ 46804fd306cSNickeau if ($canonicalPage->getUrlId() === $identifier) { 46982a60d03SNickeau $res = $this->executeTransparentRedirect( 47004fd306cSNickeau $canonicalPage->getWikiId(), 47182a60d03SNickeau self::TARGET_ORIGIN_CANONICAL 47282a60d03SNickeau ); 47382a60d03SNickeau } else { 47482a60d03SNickeau $res = $this->executePermanentRedirect( 47504fd306cSNickeau $canonicalPage->getWikiId(), // not the url because, it allows to add url query redirection property 47682a60d03SNickeau self::TARGET_ORIGIN_CANONICAL 47782a60d03SNickeau ); 47882a60d03SNickeau } 479c3437056SNickeau if ($res) { 480c3437056SNickeau return; 481c3437056SNickeau } 482c3437056SNickeau } 483c3437056SNickeau 484c3437056SNickeau /** 485c3437056SNickeau * Identifier is an alias 486c3437056SNickeau */ 48704fd306cSNickeau $aliasRequestedPage = DatabasePageRow::createFromAlias($identifier)->getMarkupPath(); 488c3437056SNickeau if ( 48904fd306cSNickeau $aliasRequestedPage !== null 49004fd306cSNickeau && $aliasRequestedPage->exists() 491c3437056SNickeau // The build alias is the file system metadata alias 492c3437056SNickeau // it may be null if the replication in the database was not successful 49304fd306cSNickeau && $aliasRequestedPage->getBuildAlias() !== null 494c3437056SNickeau ) { 49504fd306cSNickeau $buildAlias = $aliasRequestedPage->getBuildAlias(); 496c3437056SNickeau switch ($buildAlias->getType()) { 497c3437056SNickeau case AliasType::REDIRECT: 49804fd306cSNickeau $res = $this->executePermanentRedirect($aliasRequestedPage->getCanonicalUrl()->toAbsoluteUrlString(), self::TARGET_ORIGIN_ALIAS); 499c3437056SNickeau if ($res) { 500c3437056SNickeau return; 501c3437056SNickeau } 502c3437056SNickeau break; 503c3437056SNickeau case AliasType::SYNONYM: 50404fd306cSNickeau $res = $this->executeTransparentRedirect($aliasRequestedPage->getWikiId(), self::TARGET_ORIGIN_ALIAS); 505c3437056SNickeau if ($res) { 506c3437056SNickeau return; 507c3437056SNickeau } 508c3437056SNickeau break; 509c3437056SNickeau default: 510c3437056SNickeau LogUtility::msg("The alias type ({$buildAlias->getType()}) is unknown. A permanent redirect was performed for the alias $identifier"); 51104fd306cSNickeau $res = $this->executePermanentRedirect($aliasRequestedPage->getCanonicalUrl()->toAbsoluteUrlString(), self::TARGET_ORIGIN_ALIAS); 512c3437056SNickeau if ($res) { 513c3437056SNickeau return; 514c3437056SNickeau } 515c3437056SNickeau break; 516c3437056SNickeau } 517c3437056SNickeau } 518c3437056SNickeau 519c3437056SNickeau 520c3437056SNickeau // If there is a redirection defined in the page rules 521c3437056SNickeau $result = $this->processingPageRules(); 522c3437056SNickeau if ($result) { 523c3437056SNickeau // A redirection has occurred 524c3437056SNickeau // finish the process 525c3437056SNickeau return; 526c3437056SNickeau } 527c3437056SNickeau 528c3437056SNickeau /** 529c3437056SNickeau * 530c3437056SNickeau * There was no redirection found, redirect to edit mode if writer 531c3437056SNickeau * 532c3437056SNickeau */ 533c3437056SNickeau if (Identity::isWriter() && $this->getConf(self::GO_TO_EDIT_MODE) == 1) { 534c3437056SNickeau 535c3437056SNickeau $this->gotToEditMode($event); 536c3437056SNickeau // Stop here 537c3437056SNickeau return; 538c3437056SNickeau 539c3437056SNickeau } 540c3437056SNickeau 54104fd306cSNickeau /** 542c3437056SNickeau * We are still a reader, the redirection does not exist the user is not allowed to edit the page (public of other) 543c3437056SNickeau */ 544c3437056SNickeau if ($this->getConf('ActionReaderFirst') == self::NOTHING) { 545c3437056SNickeau return; 546c3437056SNickeau } 547c3437056SNickeau 548c3437056SNickeau // We are reader and their is no redirection set, we apply the algorithm 549c3437056SNickeau $readerAlgorithms = array(); 550c3437056SNickeau $readerAlgorithms[0] = $this->getConf('ActionReaderFirst'); 551c3437056SNickeau $readerAlgorithms[1] = $this->getConf('ActionReaderSecond'); 552c3437056SNickeau $readerAlgorithms[2] = $this->getConf('ActionReaderThird'); 553c3437056SNickeau 554c3437056SNickeau while ( 555c3437056SNickeau ($algorithm = array_shift($readerAlgorithms)) != null 556c3437056SNickeau ) { 557c3437056SNickeau 558c3437056SNickeau switch ($algorithm) { 559c3437056SNickeau 560c3437056SNickeau case self::NOTHING: 561c3437056SNickeau return; 562c3437056SNickeau 563c3437056SNickeau case self::GO_TO_BEST_END_PAGE_NAME: 564c3437056SNickeau 56504fd306cSNickeau /** 56604fd306cSNickeau * @var MarkupPath $bestEndPage 56704fd306cSNickeau */ 56804fd306cSNickeau list($bestEndPage, $method) = RouterBestEndPage::process($requestedMarkupPath); 569ea801fd8Sgerardnico if ($bestEndPage != null && $bestEndPage->getWikiId() !== $requestedMarkupPath->getWikiId()) { 570c3437056SNickeau $res = false; 571c3437056SNickeau switch ($method) { 572c3437056SNickeau case self::REDIRECT_PERMANENT_METHOD: 57304fd306cSNickeau $res = $this->executePermanentRedirect($bestEndPage->getWikiId(), self::TARGET_ORIGIN_BEST_END_PAGE_NAME); 574c3437056SNickeau break; 575c3437056SNickeau case self::REDIRECT_NOTFOUND_METHOD: 57604fd306cSNickeau $res = $this->performNotFoundRedirect($bestEndPage->getWikiId(), self::TARGET_ORIGIN_BEST_END_PAGE_NAME); 577c3437056SNickeau break; 578c3437056SNickeau default: 579c3437056SNickeau LogUtility::msg("This redirection method ($method) was not expected for the redirection algorithm ($algorithm)"); 580c3437056SNickeau } 581c3437056SNickeau if ($res) { 582c3437056SNickeau // Redirection has succeeded 583c3437056SNickeau return; 584c3437056SNickeau } 585c3437056SNickeau } 586c3437056SNickeau break; 587c3437056SNickeau 588c3437056SNickeau case self::GO_TO_NS_START_PAGE: 589c3437056SNickeau 590c3437056SNickeau // Start page with the conf['start'] parameter 591c3437056SNickeau $startPage = getNS($identifier) . ':' . $conf['start']; 592c3437056SNickeau if (page_exists($startPage)) { 593c3437056SNickeau $res = $this->performNotFoundRedirect($startPage, self::TARGET_ORIGIN_START_PAGE); 594c3437056SNickeau if ($res) { 595c3437056SNickeau return; 596c3437056SNickeau } 597c3437056SNickeau } 598c3437056SNickeau 599c3437056SNickeau // Start page with the same name than the namespace 600c3437056SNickeau $startPage = getNS($identifier) . ':' . curNS($identifier); 601c3437056SNickeau if (page_exists($startPage)) { 602c3437056SNickeau $res = $this->performNotFoundRedirect($startPage, self::TARGET_ORIGIN_START_PAGE); 603c3437056SNickeau if ($res) { 604c3437056SNickeau return; 605c3437056SNickeau } 606c3437056SNickeau } 607c3437056SNickeau break; 608c3437056SNickeau 609c3437056SNickeau case self::GO_TO_BEST_PAGE_NAME: 610c3437056SNickeau 611c3437056SNickeau $bestPageId = null; 612c3437056SNickeau 613c3437056SNickeau $bestPage = $this->getBestPage($identifier); 614c3437056SNickeau $bestPageId = $bestPage['id']; 615c3437056SNickeau $scorePageName = $bestPage['score']; 616c3437056SNickeau 617c3437056SNickeau // Get Score from a Namespace 618c3437056SNickeau $bestNamespace = $this->scoreBestNamespace($identifier); 619c3437056SNickeau $bestNamespaceId = $bestNamespace['namespace']; 620c3437056SNickeau $namespaceScore = $bestNamespace['score']; 621c3437056SNickeau 622c3437056SNickeau // Compare the two score 623c3437056SNickeau if ($scorePageName > 0 or $namespaceScore > 0) { 624c3437056SNickeau if ($scorePageName > $namespaceScore) { 625c3437056SNickeau $this->performNotFoundRedirect($bestPageId, self::TARGET_ORIGIN_BEST_PAGE_NAME); 626c3437056SNickeau } else { 627c3437056SNickeau $this->performNotFoundRedirect($bestNamespaceId, self::TARGET_ORIGIN_BEST_PAGE_NAME); 628c3437056SNickeau } 629c3437056SNickeau return; 630c3437056SNickeau } 631c3437056SNickeau break; 632c3437056SNickeau 633c3437056SNickeau case self::GO_TO_BEST_NAMESPACE: 634c3437056SNickeau 635c3437056SNickeau $scoreNamespace = $this->scoreBestNamespace($identifier); 636c3437056SNickeau $bestNamespaceId = $scoreNamespace['namespace']; 637c3437056SNickeau $score = $scoreNamespace['score']; 638c3437056SNickeau 639c3437056SNickeau if ($score > 0) { 640c3437056SNickeau $this->performNotFoundRedirect($bestNamespaceId, self::TARGET_ORIGIN_BEST_NAMESPACE); 641c3437056SNickeau return; 642c3437056SNickeau } 643c3437056SNickeau break; 644c3437056SNickeau 645c3437056SNickeau case self::GO_TO_SEARCH_ENGINE: 646c3437056SNickeau 647c3437056SNickeau $this->redirectToSearchEngine(); 648c3437056SNickeau 649c3437056SNickeau return; 650c3437056SNickeau 651c3437056SNickeau // End Switch Action 652c3437056SNickeau } 653c3437056SNickeau 654c3437056SNickeau // End While Action 655c3437056SNickeau } 656c3437056SNickeau 657c3437056SNickeau 658c3437056SNickeau } 659c3437056SNickeau 660c3437056SNickeau 661c3437056SNickeau /** 662c3437056SNickeau * getBestNamespace 663c3437056SNickeau * Return a list with 'BestNamespaceId Score' 664c3437056SNickeau * @param $id 665c3437056SNickeau * @return array 666c3437056SNickeau */ 667c3437056SNickeau private 668c3437056SNickeau function scoreBestNamespace($id) 669c3437056SNickeau { 670c3437056SNickeau 671c3437056SNickeau global $conf; 672c3437056SNickeau 673c3437056SNickeau // Parameters 674c3437056SNickeau $pageNameSpace = getNS($id); 675c3437056SNickeau 676c3437056SNickeau // If the page has an existing namespace start page take it, other search other namespace 677c3437056SNickeau $startPageNameSpace = $pageNameSpace . ":"; 678c3437056SNickeau $dateAt = ''; 679c3437056SNickeau // $startPageNameSpace will get a full path (ie with start or the namespace 680c3437056SNickeau resolve_pageid($pageNameSpace, $startPageNameSpace, $exists, $dateAt, true); 681c3437056SNickeau if (page_exists($startPageNameSpace)) { 682c3437056SNickeau $nameSpaces = array($startPageNameSpace); 683c3437056SNickeau } else { 684c3437056SNickeau $nameSpaces = ft_pageLookup($conf['start']); 685c3437056SNickeau } 686c3437056SNickeau 687c3437056SNickeau // Parameters and search the best namespace 688c3437056SNickeau $pathNames = explode(':', $pageNameSpace); 689c3437056SNickeau $bestNbWordFound = 0; 690c3437056SNickeau $bestNamespaceId = ''; 691c3437056SNickeau foreach ($nameSpaces as $nameSpace) { 692c3437056SNickeau 693c3437056SNickeau $nbWordFound = 0; 694c3437056SNickeau foreach ($pathNames as $pathName) { 695c3437056SNickeau if (strlen($pathName) > 2) { 696c3437056SNickeau $nbWordFound = $nbWordFound + substr_count($nameSpace, $pathName); 697c3437056SNickeau } 698c3437056SNickeau } 699c3437056SNickeau if ($nbWordFound > $bestNbWordFound) { 700c3437056SNickeau // Take only the smallest namespace 701c3437056SNickeau if (strlen($nameSpace) < strlen($bestNamespaceId) or $nbWordFound > $bestNbWordFound) { 702c3437056SNickeau $bestNbWordFound = $nbWordFound; 703c3437056SNickeau $bestNamespaceId = $nameSpace; 704c3437056SNickeau } 705c3437056SNickeau } 706c3437056SNickeau } 707c3437056SNickeau 708c3437056SNickeau $startPageFactor = $this->getConf('WeightFactorForStartPage'); 709c3437056SNickeau $nameSpaceFactor = $this->getConf('WeightFactorForSameNamespace'); 710c3437056SNickeau if ($bestNbWordFound > 0) { 711c3437056SNickeau $bestNamespaceScore = $bestNbWordFound * $nameSpaceFactor + $startPageFactor; 712c3437056SNickeau } else { 713c3437056SNickeau $bestNamespaceScore = 0; 714c3437056SNickeau } 715c3437056SNickeau 716c3437056SNickeau 717c3437056SNickeau return array( 718c3437056SNickeau 'namespace' => $bestNamespaceId, 719c3437056SNickeau 'score' => $bestNamespaceScore 720c3437056SNickeau ); 721c3437056SNickeau 722c3437056SNickeau } 723c3437056SNickeau 724c3437056SNickeau /** 725c3437056SNickeau * @param $event 726c3437056SNickeau */ 727c3437056SNickeau private 728c3437056SNickeau function gotToEditMode(&$event) 729c3437056SNickeau { 730c3437056SNickeau global $ACT; 731c3437056SNickeau $ACT = 'edit'; 732c3437056SNickeau 733c3437056SNickeau } 734c3437056SNickeau 735c3437056SNickeau 736c3437056SNickeau /** 737c3437056SNickeau * Redirect to an internal page ie: 738c3437056SNickeau * * on the same domain 739c3437056SNickeau * * no HTTP redirect 740c3437056SNickeau * * id rewrite 741c3437056SNickeau * @param string $targetPageId - target page id 742c3437056SNickeau * @param string $targetOriginId - the source of the target (redirect) 743c3437056SNickeau * @return bool - return true if the user has the permission and that the redirect was done 744c3437056SNickeau * @throws Exception 745c3437056SNickeau */ 746c3437056SNickeau private 747c3437056SNickeau function executeTransparentRedirect(string $targetPageId, string $targetOriginId): bool 748c3437056SNickeau { 749c3437056SNickeau /** 750c3437056SNickeau * Because we set the ID globally for the ID redirect 75104fd306cSNickeau * we make sure that this is not a {@link MarkupPath} 752c3437056SNickeau * object otherwise we got an error in the {@link \ComboStrap\AnalyticsMenuItem} 753c3437056SNickeau * because the constructor takes it {@link \dokuwiki\Menu\Item\AbstractItem} 754c3437056SNickeau */ 755c3437056SNickeau if (is_object($targetPageId)) { 756c3437056SNickeau $class = get_class($targetPageId); 757c3437056SNickeau LogUtility::msg("The parameters targetPageId ($targetPageId) is an object of the class ($class) and it should be a page id"); 758c3437056SNickeau } 759c3437056SNickeau 760c3437056SNickeau if (is_object($targetOriginId)) { 761c3437056SNickeau $class = get_class($targetOriginId); 762c3437056SNickeau LogUtility::msg("The parameters targetOriginId ($targetOriginId) is an object of the class ($class) and it should be a page id"); 763c3437056SNickeau } 764c3437056SNickeau 765c3437056SNickeau // If the user does not have the right to see the target page 766c3437056SNickeau // don't do anything 767c3437056SNickeau if (!(Identity::isReader($targetPageId))) { 768c3437056SNickeau return false; 769c3437056SNickeau } 770c3437056SNickeau 771c3437056SNickeau // Change the id 772c3437056SNickeau global $ID; 773c3437056SNickeau global $INFO; 774c3437056SNickeau $sourceId = $ID; 775c3437056SNickeau $ID = $targetPageId; 77604fd306cSNickeau if (isset($_REQUEST["id"])) { 77704fd306cSNickeau $_REQUEST["id"] = $targetPageId; 77804fd306cSNickeau } 77904fd306cSNickeau if (isset($_GET["id"])) { 78004fd306cSNickeau $_GET["id"] = $targetPageId; 78104fd306cSNickeau } 7824cadd4f8SNickeau 783c3437056SNickeau /** 7844cadd4f8SNickeau * Refresh the $INFO data 7854cadd4f8SNickeau * 7864cadd4f8SNickeau * the info attributes are used elsewhere 7874cadd4f8SNickeau * 'id': for the sidebar 7884cadd4f8SNickeau * 'exist' : for the meta robot = noindex,follow, see {@link tpl_metaheaders()} 7894cadd4f8SNickeau * 'rev' : for the edit button to be sure that the page is still the same 790c3437056SNickeau */ 7914cadd4f8SNickeau $INFO = pageinfo(); 792c3437056SNickeau 793c3437056SNickeau /** 794c3437056SNickeau * Not compatible with 795c3437056SNickeau * https://www.dokuwiki.org/config:send404 is enabled 796c3437056SNickeau * 797c3437056SNickeau * This check happens before that dokuwiki is started 798c3437056SNickeau * and send an header in doku.php 799c3437056SNickeau * 800c3437056SNickeau * We send a warning 801c3437056SNickeau */ 802c3437056SNickeau global $conf; 803c3437056SNickeau if ($conf['send404'] == true) { 804c3437056SNickeau 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); 805c3437056SNickeau } 806c3437056SNickeau 807c3437056SNickeau // Redirection 808c3437056SNickeau $this->logRedirection($sourceId, $targetPageId, $targetOriginId, self::REDIRECT_TRANSPARENT_METHOD); 809c3437056SNickeau 810c3437056SNickeau return true; 811c3437056SNickeau 812c3437056SNickeau } 813c3437056SNickeau 81404fd306cSNickeau private function executePermanentRedirect(string $targetIdOrUrl, $targetOrigin): bool 815c3437056SNickeau { 81604fd306cSNickeau return $this->executeHttpRedirect($targetIdOrUrl, $targetOrigin, self::REDIRECT_PERMANENT_METHOD); 817c3437056SNickeau } 818c3437056SNickeau 819c3437056SNickeau /** 820c3437056SNickeau * The general HTTP Redirect method to an internal page 821c3437056SNickeau * where the redirection method decide which type of redirection 82204fd306cSNickeau * @param string $targetIdOrUrl - a dokuwiki id or an url 823c3437056SNickeau * @param string $targetOrigin - the origin of the target (the algorithm used to get the target origin) 824c3437056SNickeau * @param string $method - the redirection method 825c3437056SNickeau */ 826c3437056SNickeau private 82704fd306cSNickeau function executeHttpRedirect(string $targetIdOrUrl, string $targetOrigin, string $method): bool 828c3437056SNickeau { 829c3437056SNickeau 830c3437056SNickeau global $ID; 831c3437056SNickeau 832c3437056SNickeau 833c3437056SNickeau // Log the redirections 83404fd306cSNickeau $this->logRedirection($ID, $targetIdOrUrl, $targetOrigin, $method); 835c3437056SNickeau 836c3437056SNickeau 83704fd306cSNickeau // An http external url ? 83804fd306cSNickeau try { 8395b0932efSgerardnico $isHttpUrl = Url::createFromString($targetIdOrUrl)->isHttpUrl(); 84004fd306cSNickeau } catch (ExceptionBadSyntax|ExceptionBadArgument $e) { 8415b0932efSgerardnico $isHttpUrl = false; 84204fd306cSNickeau } 84304fd306cSNickeau 84482a60d03SNickeau // If there is a bug in the isValid function for an internal url 84582a60d03SNickeau // We get a loop. 84682a60d03SNickeau // The Url becomes the id, the id is unknown and we do a redirect again 84782a60d03SNickeau // 84882a60d03SNickeau // We check then if the target starts with the base url 84982a60d03SNickeau // if this is the case, it's valid 8505b0932efSgerardnico if (!$isHttpUrl && strpos($targetIdOrUrl, DOKU_URL) === 0) { 8515b0932efSgerardnico $isHttpUrl = true; 85282a60d03SNickeau } 8535b0932efSgerardnico if ($isHttpUrl) { 854c3437056SNickeau 855c3437056SNickeau // defend against HTTP Response Splitting 856c3437056SNickeau // https://owasp.org/www-community/attacks/HTTP_Response_Splitting 85704fd306cSNickeau $targetUrl = stripctl($targetIdOrUrl); 858c3437056SNickeau 859c3437056SNickeau } else { 860c3437056SNickeau 861c3437056SNickeau 862c3437056SNickeau // Explode the page ID and the anchor (#) 86304fd306cSNickeau $link = explode('#', $targetIdOrUrl, 2); 864c3437056SNickeau 86511d09b86Sgerardnico $url = UrlEndpoint::createDokuUrl(); 866c3437056SNickeau 8675b0932efSgerardnico $urlParams = []; 868c3437056SNickeau // if this is search engine redirect 869c3437056SNickeau if ($targetOrigin == self::TARGET_ORIGIN_SEARCH_ENGINE) { 870c3437056SNickeau $replacementPart = array(':', '_', '-'); 871c3437056SNickeau $query = str_replace($replacementPart, ' ', $ID); 87211d09b86Sgerardnico $url->setQueryParameter(ExecutionContext::DO_ATTRIBUTE, ExecutionContext::SEARCH_ACTION); 87311d09b86Sgerardnico $url->setQueryParameter("q", $query); 874c3437056SNickeau } 875c3437056SNickeau 8765b0932efSgerardnico /** 8775b0932efSgerardnico * Doing a permanent redirect with a added query string 8785b0932efSgerardnico * create a new page url on the search engine 8795b0932efSgerardnico * 8805b0932efSgerardnico * ie 8815b0932efSgerardnico * http://host/page 8825b0932efSgerardnico * is not the same 8835b0932efSgerardnico * than 8845b0932efSgerardnico * http://host/page?whatever 8855b0932efSgerardnico * 8865b0932efSgerardnico * We can't pass query string otherwise, we get 88749b8fb24Sgerardnico * the SEO warning / error 8885b0932efSgerardnico * `Alternative page with proper canonical tag` 88911d09b86Sgerardnico * 89011d09b86Sgerardnico * Use HTTP X header for debug 8915b0932efSgerardnico */ 8925b0932efSgerardnico if ($method !== self::REDIRECT_PERMANENT_METHOD) { 89311d09b86Sgerardnico $url->setQueryParameter(action_plugin_combo_routermessage::ORIGIN_PAGE, $ID); 89411d09b86Sgerardnico $url->setQueryParameter(action_plugin_combo_routermessage::ORIGIN_TYPE, $targetOrigin); 8955b0932efSgerardnico } 8965b0932efSgerardnico 89711d09b86Sgerardnico $id = $link[0]; 89811d09b86Sgerardnico $url->setQueryParameter(DokuwikiId::DOKUWIKI_ID_ATTRIBUTE, $id); 89970bbd7f1Sgerardnico if (array_key_exists(1, $link)) { 90011d09b86Sgerardnico $url->setFragment($link[1]); 901c3437056SNickeau } 90211d09b86Sgerardnico $targetUrl = $url->toAbsoluteUrlString(); 903c3437056SNickeau 904c3437056SNickeau } 905c3437056SNickeau 906c3437056SNickeau /** 907c3437056SNickeau * The dokuwiki function {@link send_redirect()} 908c3437056SNickeau * set the `Location header` and in php, the header function 909c3437056SNickeau * in this case change the status code to 302 Arghhhh. 910c3437056SNickeau * The code below is adapted from this function {@link send_redirect()} 911c3437056SNickeau */ 912c3437056SNickeau global $MSG; // are there any undisplayed messages? keep them in session for display 913c3437056SNickeau if (isset($MSG) && count($MSG) && !defined('NOSESSION')) { 914c3437056SNickeau //reopen session, store data and close session again 915c3437056SNickeau @session_start(); 916c3437056SNickeau $_SESSION[DOKU_COOKIE]['msg'] = $MSG; 917c3437056SNickeau } 918c3437056SNickeau session_write_close(); // always close the session 919c3437056SNickeau 920c3437056SNickeau switch ($method) { 9215b0932efSgerardnico 922c3437056SNickeau case self::REDIRECT_PERMANENT_METHOD: 92304fd306cSNickeau ExecutionContext::getActualOrCreateFromEnv() 92404fd306cSNickeau ->response() 92504fd306cSNickeau ->setStatus(HttpResponseStatus::PERMANENT_REDIRECT) 926c3437056SNickeau ->addHeader(self::LOCATION_HEADER_PREFIX . $targetUrl) 92704fd306cSNickeau ->end(); 928c3437056SNickeau return true; 9295b0932efSgerardnico 930c3437056SNickeau case self::REDIRECT_NOTFOUND_METHOD: 931c3437056SNickeau 9325b0932efSgerardnico 933c3437056SNickeau // Empty 404 body to not get the standard 404 page of the browser 934c3437056SNickeau // but a blank page to avoid a sort of FOUC. 935c3437056SNickeau // ie the user see a page briefly 93604fd306cSNickeau ExecutionContext::getActualOrCreateFromEnv() 93704fd306cSNickeau ->response() 93804fd306cSNickeau ->setStatus(HttpResponseStatus::NOT_FOUND) 939c3437056SNickeau ->addHeader(self::REFRESH_HEADER_PREFIX . $targetUrl) 94004fd306cSNickeau ->setBody(self::PAGE_404, Mime::getHtml()) 94104fd306cSNickeau ->end(); 942c3437056SNickeau return true; 943c3437056SNickeau 944c3437056SNickeau default: 945c3437056SNickeau LogUtility::msg("The method ($method) is not an http redirection"); 946c3437056SNickeau return false; 947c3437056SNickeau } 948c3437056SNickeau 949c3437056SNickeau 950c3437056SNickeau } 951c3437056SNickeau 952c3437056SNickeau /** 953c3437056SNickeau * @param $id 954c3437056SNickeau * @return array 955c3437056SNickeau */ 956c3437056SNickeau private 957c3437056SNickeau function getBestPage($id): array 958c3437056SNickeau { 959c3437056SNickeau 960c3437056SNickeau // The return parameters 961c3437056SNickeau $bestPageId = null; 962c3437056SNickeau $scorePageName = null; 963c3437056SNickeau 964c3437056SNickeau // Get Score from a page 965c3437056SNickeau $pageName = noNS($id); 966c3437056SNickeau $pagesWithSameName = ft_pageLookup($pageName); 967c3437056SNickeau if (count($pagesWithSameName) > 0) { 968c3437056SNickeau 969c3437056SNickeau // Search same namespace in the page found than in the Id page asked. 970c3437056SNickeau $bestNbWordFound = 0; 971c3437056SNickeau 972c3437056SNickeau 973c3437056SNickeau $wordsInPageSourceId = explode(':', $id); 974c3437056SNickeau foreach ($pagesWithSameName as $targetPageId => $title) { 975c3437056SNickeau 976c3437056SNickeau // Nb of word found in the target page id 977c3437056SNickeau // that are in the source page id 978c3437056SNickeau $nbWordFound = 0; 979c3437056SNickeau foreach ($wordsInPageSourceId as $word) { 980c3437056SNickeau $nbWordFound = $nbWordFound + substr_count($targetPageId, $word); 981c3437056SNickeau } 982c3437056SNickeau 983c3437056SNickeau if ($bestPageId == null) { 984c3437056SNickeau 985c3437056SNickeau $bestNbWordFound = $nbWordFound; 986c3437056SNickeau $bestPageId = $targetPageId; 987c3437056SNickeau 988c3437056SNickeau } else { 989c3437056SNickeau 990c3437056SNickeau if ($nbWordFound >= $bestNbWordFound && strlen($bestPageId) > strlen($targetPageId)) { 991c3437056SNickeau 992c3437056SNickeau $bestNbWordFound = $nbWordFound; 993c3437056SNickeau $bestPageId = $targetPageId; 994c3437056SNickeau 995c3437056SNickeau } 996c3437056SNickeau 997c3437056SNickeau } 998c3437056SNickeau 999c3437056SNickeau } 1000c3437056SNickeau $scorePageName = $this->getConf('WeightFactorForSamePageName') + ($bestNbWordFound - 1) * $this->getConf('WeightFactorForSameNamespace'); 1001c3437056SNickeau return array( 1002c3437056SNickeau 'id' => $bestPageId, 1003c3437056SNickeau 'score' => $scorePageName); 1004c3437056SNickeau } 1005c3437056SNickeau return array( 1006c3437056SNickeau 'id' => $bestPageId, 1007c3437056SNickeau 'score' => $scorePageName 1008c3437056SNickeau ); 1009c3437056SNickeau 1010c3437056SNickeau } 1011c3437056SNickeau 1012c3437056SNickeau 1013c3437056SNickeau /** 1014c3437056SNickeau * Redirect to the search engine 1015c3437056SNickeau */ 1016c3437056SNickeau private 1017c3437056SNickeau function redirectToSearchEngine() 1018c3437056SNickeau { 1019c3437056SNickeau 1020c3437056SNickeau global $ID; 1021c3437056SNickeau $this->performNotFoundRedirect($ID, self::TARGET_ORIGIN_SEARCH_ENGINE); 1022c3437056SNickeau 1023c3437056SNickeau } 1024c3437056SNickeau 1025c3437056SNickeau 1026c3437056SNickeau /** 1027c3437056SNickeau * 1028c3437056SNickeau * * For a conf file, it will update the Redirection Action Data as Referrer, Count Of Redirection, Redirection Date 1029c3437056SNickeau * * For a SQlite database, it will add a row into the log 1030c3437056SNickeau * 1031c3437056SNickeau * @param string $sourcePageId 1032c3437056SNickeau * @param $targetPageId 1033c3437056SNickeau * @param $algorithmic 1034c3437056SNickeau * @param $method - http or rewrite 1035c3437056SNickeau */ 1036c3437056SNickeau function logRedirection(string $sourcePageId, $targetPageId, $algorithmic, $method) 1037c3437056SNickeau { 1038c3437056SNickeau 1039c3437056SNickeau $row = array( 1040c3437056SNickeau "TIMESTAMP" => date("c"), 1041c3437056SNickeau "SOURCE" => $sourcePageId, 1042c3437056SNickeau "TARGET" => $targetPageId, 104370bbd7f1Sgerardnico "REFERRER" => $_SERVER['HTTP_REFERER'] ?? null, 1044c3437056SNickeau "TYPE" => $algorithmic, 1045c3437056SNickeau "METHOD" => $method 1046c3437056SNickeau ); 1047c3437056SNickeau $request = Sqlite::createOrGetBackendSqlite() 1048c3437056SNickeau ->createRequest() 1049c3437056SNickeau ->setTableRow('redirections_log', $row); 1050c3437056SNickeau try { 1051c3437056SNickeau $request 1052c3437056SNickeau ->execute(); 105304fd306cSNickeau } catch (ExceptionCompile $e) { 1054c3437056SNickeau LogUtility::msg("Redirection Log Insert Error. {$e->getMessage()}"); 1055c3437056SNickeau } finally { 1056c3437056SNickeau $request->close(); 1057c3437056SNickeau } 1058c3437056SNickeau 1059c3437056SNickeau 1060c3437056SNickeau } 1061c3437056SNickeau 1062c3437056SNickeau /** 1063c3437056SNickeau * This function check if there is a redirection declared 1064c3437056SNickeau * in the redirection table 1065c3437056SNickeau * @return bool - true if a rewrite or redirection occurs 1066c3437056SNickeau * @throws Exception 1067c3437056SNickeau */ 1068c3437056SNickeau private function processingPageRules(): bool 1069c3437056SNickeau { 1070c3437056SNickeau global $ID; 1071c3437056SNickeau 1072c3437056SNickeau $calculatedTarget = null; 1073c3437056SNickeau $ruleMatcher = null; // Used in a warning message if the target page does not exist 1074c3437056SNickeau // Known redirection in the table 1075c3437056SNickeau // Get the page from redirection data 1076c3437056SNickeau $rules = $this->pageRules->getRules(); 1077c3437056SNickeau foreach ($rules as $rule) { 1078c3437056SNickeau 1079c3437056SNickeau $ruleMatcher = strtolower($rule[PageRules::MATCHER_NAME]); 1080c3437056SNickeau $ruleTarget = $rule[PageRules::TARGET_NAME]; 1081c3437056SNickeau 1082c3437056SNickeau // Glob to Rexgexp 108304fd306cSNickeau $regexpPattern = '/' . str_replace("*", "(.*)", $ruleMatcher) . '/i'; 1084c3437056SNickeau 1085c3437056SNickeau // Match ? 1086c3437056SNickeau // https://www.php.net/manual/en/function.preg-match.php 108786867397Sgerardnico $pregMatchResult = @preg_match($regexpPattern, $ID, $matches); 108886867397Sgerardnico if ($pregMatchResult === false) { 108986867397Sgerardnico // The `if` to take into account this problem 109086867397Sgerardnico // PHP Warning: preg_match(): Unknown modifier 'd' in /opt/www/datacadamia.com/lib/plugins/combo/action/router.php on line 972 109186867397Sgerardnico LogUtility::log2file("processing Page Rules An error occurred with the pattern ($regexpPattern)", LogUtility::LVL_MSG_WARNING); 109286867397Sgerardnico return false; 109386867397Sgerardnico } 109486867397Sgerardnico if ($pregMatchResult) { 1095c3437056SNickeau $calculatedTarget = $ruleTarget; 1096c3437056SNickeau foreach ($matches as $key => $match) { 1097c3437056SNickeau if ($key == 0) { 1098c3437056SNickeau continue; 1099c3437056SNickeau } else { 1100c3437056SNickeau $calculatedTarget = str_replace('$' . $key, $match, $calculatedTarget); 1101c3437056SNickeau } 1102c3437056SNickeau } 1103c3437056SNickeau break; 1104c3437056SNickeau } 1105c3437056SNickeau } 1106c3437056SNickeau 1107c3437056SNickeau if ($calculatedTarget == null) { 1108c3437056SNickeau return false; 1109c3437056SNickeau } 1110c3437056SNickeau 1111c3437056SNickeau // If this is an external redirect (other domain) 111204fd306cSNickeau try { 111304fd306cSNickeau $isHttpUrl = Url::createFromString($calculatedTarget)->isHttpUrl(); 111404fd306cSNickeau } catch (ExceptionBadSyntax $e) { 111504fd306cSNickeau $isHttpUrl = false; 111604fd306cSNickeau } 111704fd306cSNickeau if ($isHttpUrl) { 1118c3437056SNickeau $this->executeHttpRedirect($calculatedTarget, self::TARGET_ORIGIN_PAGE_RULES, self::REDIRECT_PERMANENT_METHOD); 1119c3437056SNickeau return true; 1120c3437056SNickeau } 1121c3437056SNickeau 1122c3437056SNickeau // If the page exist 1123c3437056SNickeau if (page_exists($calculatedTarget)) { 1124c3437056SNickeau 1125c3437056SNickeau // This is DokuWiki Id and should always be lowercase 1126c3437056SNickeau // The page rule may have change that 1127c3437056SNickeau $calculatedTarget = strtolower($calculatedTarget); 1128d0abe7fcSgerardnico $res = $this->executeHttpRedirect($calculatedTarget, self::TARGET_ORIGIN_PAGE_RULES, self::REDIRECT_PERMANENT_METHOD); 1129c3437056SNickeau if ($res) { 1130c3437056SNickeau return true; 1131c3437056SNickeau } else { 1132c3437056SNickeau return false; 1133c3437056SNickeau } 1134c3437056SNickeau 1135c3437056SNickeau } else { 1136c3437056SNickeau 1137c3437056SNickeau LogUtility::msg("The calculated target page ($calculatedTarget) (for the non-existing page `$ID` with the matcher `$ruleMatcher`) does not exist", LogUtility::LVL_MSG_ERROR); 1138c3437056SNickeau return false; 1139c3437056SNickeau 1140c3437056SNickeau } 1141c3437056SNickeau 1142c3437056SNickeau } 1143c3437056SNickeau 1144c3437056SNickeau private function performNotFoundRedirect(string $targetId, string $origin): bool 1145c3437056SNickeau { 1146c3437056SNickeau return $this->executeHttpRedirect($targetId, $origin, self::REDIRECT_NOTFOUND_METHOD); 1147c3437056SNickeau } 1148c3437056SNickeau 1149c3437056SNickeau 1150c3437056SNickeau} 1151