/**
*
* @param string $url
* @param array $title
* @param string $task
* @param int $limit
* @param int $limitstart
* @return sefurl
*/
public static function sefGetLocation($nonSefUrl, &$title, $task = null, $limit = null, $limitstart = null, $langParam = null, $showall = null, $suppressPagination = false)
{
try {
$shPageInfo =& Sh404sefFactory::getPageInfo();
$sefConfig =& Sh404sefFactory::getConfig();
$lang = empty($langParam) ? $shPageInfo->currentLanguageTag : $langParam;
// shumisha : try to avoid duplicate content on multilingual sites by always adding &lang=xx to url (stored in DB).
// warning : must add &lang=xx only if it does not exists already
if (!strpos($nonSefUrl, 'lang=')) {
$shSepString = substr($nonSefUrl, -9) == 'index.php' ? '?' : '&';
$nonSefUrl .= $shSepString . 'lang=' . shGetIsoCodeFromName($lang);
}
// make sure url is consistent
$nonSefUrl = str_replace('&', '&', $nonSefUrl);
// detect multipage homepage
$shMultiPageHomePageFlag = shIsHomepage($nonSefUrl);
// get all the slugs ready for being urls bits
$tempSefUrl = array();
foreach ($title as $titlestring) {
$decodedTitletring = urldecode($titlestring);
$tempSefUrl[] = titleToLocation($decodedTitletring);
}
// now build the URL
$tempSefUrl = implode("/", $tempSefUrl);
// remove duplicate /
$tempSefUrl = ShlSystem_Strings::pr('/\\/{2,}/u', '/', $tempSefUrl);
// and truncate to max length, according to param
$tempSefUrl = JString::substr($tempSefUrl, 0, sh404SEF_MAX_SEF_URL_LENGTH);
// trim to max length V 1.2.4.t
// if URL is empty, and unless this is a paginated home page, or home page in non-default language, stop there
if (empty($tempSefUrl)) {
if ((!shIsMultilingual() || shIsMultilingual() && shIsDefaultlang($lang)) && !$sefConfig->addFile && !$shMultiPageHomePageFlag) {
//
return '';
}
// if location is empty, and not multilingual site, or multilingual, but this is default language, then there is nothing to add to url
}
// we have a valid SEF url, built with the data ($title) sent
// by plugin. Now we want to check if it's already in the db
// and add it if not
// first, we search the memory cache for the non-sef url
// as it is faster than looking up the db
$finalSefUrl = '';
$sefUrlType = Sh404sefHelperCache::getSefUrlFromCache($nonSefUrl, $finalSefUrl);
// if non-sef was not found in cache - or found, but it was a 404 last time we saw it -
// we should continue and try adding it
if ($sefUrlType == sh404SEF_URLTYPE_NONE || $sefUrlType == sh404SEF_URLTYPE_404) {
$finalSefUrl = false;
// non-sef was not found in cache, let's look up the database
if ($sefUrlType == sh404SEF_URLTYPE_NONE) {
$finalSefUrl = ShlDbHelper::selectResult('#__sh404sef_urls', 'oldurl', array('newurl' => $nonSefUrl));
}
// we found the sef url in database, we're done
if (!empty($finalSefUrl)) {
return $finalSefUrl;
}
// the non-sef url is not in memory cache, nor in database
// that's a new one, we need to finalize its sef (add pagination and language information)
// After finalizing it, we'll also check that sef is not in the db
// as it can already be there, associated with another non-sef (ie: a duplicate)
// Either way we'll add it in the db, but mark it as a duplicate if needed
// add pagination information, unless we were instructed by extension plugin not to
// find if we should separate pagination info from sef with a / or not
if (!empty($tempSefUrl)) {
$shSeparator = JString::substr($tempSefUrl, -1) == '/' ? '' : '/';
} else {
$shSeparator = '';
}
$finalSefUrl = $suppressPagination ? $tempSefUrl : shAddPaginationInfo($limit, $limitstart, $showall, 1, $nonSefUrl, $tempSefUrl, $shSeparator);
// v 1.2.4.t
// if home page, we don't record anything, just return "home page"
if ($shMultiPageHomePageFlag && '/' . $finalSefUrl == $tempSefUrl && (!shIsMultilingual() || shIsMultilingual() && shIsDefaultLang($lang))) {
// but this is default language
// this is start page of multipage homepage, return home or forced home
if (!empty($sefConfig->shForcedHomePage)) {
return str_replace($shPageInfo->getDefaultFrontLiveSite() . '/', '', $sefConfig->shForcedHomePage);
} else {
return '';
}
}
// add language information
// first, remove languages in non-sef, to see if we're on homepage
// as handling is sligthly different for homepage
$v1 = shCleanUpLang($nonSefUrl);
$v2 = shCleanUpLang($shPageInfo->homeLink);
if ($v1 == $v2 || $v1 == 'index.php') {
// check if this is homepage
if (shIsMultilingual() && !shIsDefaultLang($lang)) {
// if homepage in not-default-language, then add language code regardless of user settings
// as we otherwise would not be able to switch language on the frontpage
$finalSefUrl = shGetIsoCodeFromName($lang) . '/';
} else {
$finalSefUrl = '';
}
} else {
// not on homepage, insert lang code based on user setting
$option = shGetURLVar($nonSefUrl, 'option', '');
if (shInsertIsoCodeInUrl($option, $lang)) {
// insert language code based on param
// pass URL lang info, as may not be current lang
$finalSefUrl = shGetIsoCodeFromName($lang) . '/' . $finalSefUrl;
// must be forced lang, not default
}
}
// after adding pagination part of SEF, and adding language code
// the new SEF url is now complete and we can try adding to it cache and db
if ($finalSefUrl != '') {
$dburl = null;
$dbUrlId = null;
$nonSefUrlType = sh404SEF_URLTYPE_NONE;
// search the memory cache for this new sef
if ($sefConfig->shUseURLCache) {
$nonSefUrlType = Sh404sefHelperCache::getNonSefUrlFromCache($finalSefUrl, $dburl);
}
$newMaxRank = 0;
// if the new SEF was not found in memory cache, or if it was found but
// we're set to record duplicates, we search for it in the database
if ($sefConfig->shRecordDuplicates || $nonSefUrlType == sh404SEF_URLTYPE_NONE) {
$dbUrlList = ShlDbHelper::selectObjectList('#__sh404sef_urls', array('id', 'newurl', 'rank', 'dateadd'), array('oldurl' => $finalSefUrl), $aWhereData = array(), $orderBy = array('rank'));
if (count($dbUrlList) > 0) {
$dburl = $dbUrlList[0]->newurl;
$dbUrlId = $dbUrlList[0]->id;
if (empty($dburl)) {
// V 1.2.4.t url was found in DB, but was a 404
$nonSefUrlType = sh404SEF_URLTYPE_404;
} else {
$newMaxRank = $dbUrlList[count($dbUrlList) - 1]->rank + 1;
$nonSefUrlType = $dbUrlList[0]->dateadd == '0000-00-00' ? sh404SEF_URLTYPE_AUTO : sh404SEF_URLTYPE_CUSTOM;
}
}
}
if ($nonSefUrlType != sh404SEF_URLTYPE_NONE && $nonSefUrlType != sh404SEF_URLTYPE_404) {
// we found the SEF, one or more times in the db, in records which do have a non-sef attached
$isDuplicate = $dburl != $nonSefUrl;
// This is a duplicate so we must indert it with incremented rank;
if (is_null($dburl) || $isDuplicate && $sefConfig->shRecordDuplicates) {
// shAddSefUrlToDBAndCache( $nonSefUrl, $finalSefUrl, ($isDuplicate ? $newMaxRank : 0), $nonSefUrlType);
$dateAdd = $nonSefUrlType == sh404SEF_URLTYPE_AUTO ? '0000-00-00' : date("Y-m-d");
ShlDbHelper::insert('#__sh404sef_urls', array('oldurl' => $finalSefUrl, 'newurl' => $nonSefUrl, 'rank' => $isDuplicate ? $newMaxRank : 0, 'dateadd' => $dateAdd));
// store new sef/non-sef pair in memory cache
Sh404sefHelperCache::addSefUrlToCache($nonSefUrl, $finalSefUrl, $nonSefUrlType);
// create shURL : get a shURL model, and ask url creation
$model = ShlMvcModel_Base::getInstance('pageids', 'Sh404sefModel');
$model->createPageId($finalSefUrl, $nonSefUrl);
}
} else {
// we haven't found the non-sef/sef pair, but maybe there is a record for
// a 404 with that SEF. If so, we will "upgrade" the 404 record to a
// normal non-sef/sef pair
$dbUrlId = empty($dbUrlId) ? 0 : intval($dbUrlId);
if ($sefConfig->shLog404Errors) {
if ($nonSefUrlType == sh404SEF_URLTYPE_404 && !empty($dbUrlId)) {
// we already have seen that it is a 404
$id = $dbUrlId;
} elseif ($nonSefUrlType == sh404SEF_URLTYPE_404) {
$id = ShlDbHelper::selectResult('#__sh404sef_urls', 'id', array('oldurl' => $finalSefUrl, 'newurl' => ''));
} else {
$id = null;
}
} else {
$id = null;
// if we are not logging 404 errors, then no need to check for
}
// previous hit of this page.
if (!empty($id)) {
// we found a 404 record matching the SEF url just created. We'll update that record
// instead of creating a new one
// need to update dateadd to 0, as otherwise this sef/non-sef pair will be seen as custom
// this makes all such 404 errors 'disappear' from the 404 log, but no other solution
ShlDbHelper::updateIn('#__sh404sef_urls', array('newurl' => $nonSefUrl, 'dateadd' => '0000-00-00'), 'id', array($id));
Sh404sefHelperCache::addSefUrlToCache($nonSefUrl, $finalSefUrl, sh404SEF_URLTYPE_AUTO);
} else {
// standard case: creation of a totally new sef/non-sef pair
ShlDbHelper::insert('#__sh404sef_urls', array('oldurl' => $finalSefUrl, 'newurl' => $nonSefUrl, 'rank' => 0, 'dateadd' => '0000-00-00'));
// store new sef/non-sef pair in memory cache
Sh404sefHelperCache::addSefUrlToCache($nonSefUrl, $finalSefUrl, sh404SEF_URLTYPE_AUTO);
// create shURL : get a shURL model, and ask url creation
$model = ShlMvcModel_Base::getInstance('pageids', 'Sh404sefModel');
$model->createPageId($finalSefUrl, $nonSefUrl);
}
}
}
}
} catch (Exception $e) {
$finalSefUrl = '';
ShlSystem_Log::error('sh404sef', '%s::%s::%d: %s', __CLASS__, __METHOD__, __LINE__, $e->getMessage());
}
return $finalSefUrl;
}
function shDoTitleTags(&$buffer)
{
// Replace TITLE and DESCRIPTION and KEYWORDS
if (empty($buffer)) {
return;
}
global $shCustomTitleTag, $shCustomDescriptionTag, $shCustomKeywordsTag, $shCustomRobotsTag, $shCustomLangTag, $shCanonicalTag;
$database = ShlDbHelper::getDb();
$shPageInfo =& Sh404sefFactory::getPageInfo();
$sefConfig =& Sh404sefFactory::getConfig();
$document = JFactory::getDocument();
$headData = $document->getHeadData();
// V 1.2.4.t protect against error if using shCustomtags without sh404SEF activated
// this should not happen, so we simply do nothing
if (!isset($sefConfig) || empty($shPageInfo->currentNonSefUrl)) {
return;
}
// check if there is a manually created set of tags from tags file
// need to get them from DB
if ($sefConfig->shMetaManagementActivated) {
shIncludeMetaPlugin();
// is this homepage ? set flag for future use
// V 1.2.4.t make sure we have lang info and properly sorted params
if (!preg_match('/(&|\\?)lang=[a-zA-Z]{2,3}/iuU', $shPageInfo->currentNonSefUrl)) {
// no lang string, let's add default
$shTemp = explode('-', $shPageInfo->currentLanguageTag);
$shLangTemp = $shTemp[0] ? $shTemp[0] : 'en';
$shPageInfo->currentNonSefUrl .= '&lang=' . $shLangTemp;
}
$nonSef = shGetCurrentNonSef();
$isHome = $nonSef == shSortUrl(shCleanUpAnchor(Sh404sefFactory::getPageInfo()->homeLink));
$shCustomTags = shGetCustomMetaData($isHome ? sh404SEF_HOMEPAGE_CODE : $nonSef);
// J! 2.5 finder canonical handling/hack
$highlight = shGetURLVar($nonSef, 'highlight', null);
if (!empty($highlight) && empty($shCanonicalTag)) {
$searchCanoNonSef = str_replace('?highlight=' . $highlight, '', $nonSef);
$searchCanoNonSef = str_replace('&highlight=' . $highlight, '', $searchCanoNonSef);
$shCanonicalTag = JRoute::_($searchCanoNonSef);
}
$tagsToInsert = '';
// group new tags insertion, better perf
if (!empty($shCustomTags)) {
$shCustomTitleTag = !empty($shCustomTags->metatitle) ? $shCustomTags->metatitle : $shCustomTitleTag;
$shCustomDescriptionTag = !empty($shCustomTags->metadesc) ? $shCustomTags->metadesc : $shCustomDescriptionTag;
$shCustomKeywordsTag = !empty($shCustomTags->metakey) ? $shCustomTags->metakey : $shCustomKeywordsTag;
$shCustomRobotsTag = !empty($shCustomTags->metarobots) ? $shCustomTags->metarobots : $shCustomRobotsTag;
$shCustomLangTag = !empty($shCustomTags->metalang) ? $shCustomTags->metalang : $shCustomLangTag;
$shCanonicalTag = !empty($shCustomTags->canonical) ? $shCustomTags->canonical : $shCanonicalTag;
}
// then insert them in page
if (empty($shCustomTitleTag)) {
$shCustomTitleTag = $document->getTitle();
}
if (!empty($shCustomTitleTag)) {
$prepend = $isHome ? '' : $sefConfig->prependToPageTitle;
$append = $isHome ? '' : $sefConfig->appendToPageTitle;
$shPageInfo->pageTitle = htmlspecialchars(shCleanUpTitle($prepend . $shCustomTitleTag . $append), ENT_COMPAT, 'UTF-8');
$buffer = ShlSystem_Strings::pr('/\\<\\s*title\\s*\\>.*\\<\\s*\\/title\\s*\\>/isuU', '<title>' . $shPageInfo->pageTitle . '</title>', $buffer);
$buffer = ShlSystem_Strings::pr('/\\<\\s*meta\\s+name\\s*=\\s*"title.*\\/\\>/isuU', '', $buffer);
// remove any title meta
}
if (!is_null($shCustomDescriptionTag)) {
$t = htmlspecialchars(shCleanUpDesc($shCustomDescriptionTag), ENT_COMPAT, 'UTF-8');
$shPageInfo->pageDescription = ShlSystem_Strings::pr('#\\$([0-9]*)#u', '\\\\$${1}', $t);
if (strpos($buffer, '<meta name="description" content=') !== false) {
$buffer = ShlSystem_Strings::pr('/\\<\\s*meta\\s+name\\s*=\\s*"description.*\\/\\>/isUu', '<meta name="description" content="' . $shPageInfo->pageDescription . '" />', $buffer);
} else {
$tagsToInsert .= "\n" . '<meta name="description" content="' . $shPageInfo->pageDescription . '" />';
}
} else {
// read Joomla! description if none set by us
if (empty($shPageInfo->pageDescription)) {
$shPageInfo->pageDescription = empty($headData['description']) ? '' : htmlspecialchars(shCleanUpDesc($headData['description']), ENT_COMPAT, 'UTF-8');
}
}
if (!is_null($shCustomKeywordsTag)) {
$t = htmlspecialchars(shCleanUpDesc($shCustomKeywordsTag), ENT_COMPAT, 'UTF-8');
$shPageInfo->pageKeywords = ShlSystem_Strings::pr('#\\$([0-9]*)#u', '\\\\$${1}', $t);
if (strpos($buffer, '<meta name="keywords" content=') !== false) {
$buffer = ShlSystem_Strings::pr('/\\<\\s*meta\\s+name\\s*=\\s*"keywords.*\\/\\>/isUu', '<meta name="keywords" content="' . $shPageInfo->pageKeywords . '" />', $buffer);
} else {
$tagsToInsert .= "\n" . '<meta name="keywords" content="' . $shPageInfo->pageKeywords . '" />';
}
} else {
// read Joomla! description if none set by us
if (empty($shPageInfo->pageKeywords)) {
$shPageInfo->pageKeywords = empty($headData['metaTags']['standard']['keywords']) ? '' : htmlspecialchars(shCleanUpDesc($headData['metaTags']['standard']['keywords']), ENT_COMPAT, 'UTF-8');
}
}
if (!is_null($shCustomRobotsTag)) {
$shPageInfo->pageRobotsTag = $shCustomRobotsTag;
if (strpos($buffer, '<meta name="robots" content="') !== false) {
$buffer = ShlSystem_Strings::pr('/\\<\\s*meta\\s+name\\s*=\\s*"robots.*\\/\\>/isUu', '<meta name="robots" content="' . $shCustomRobotsTag . '" />', $buffer);
} else {
if (!empty($shCustomRobotsTag)) {
$tagsToInsert .= "\n" . '<meta name="robots" content="' . $shCustomRobotsTag . '" />';
}
}
} else {
// read Joomla! description if none set by us
if (empty($shPageInfo->pageRobotsTag)) {
$shPageInfo->pageRobotsTag = empty($headData['metaTags']['standard']['robots']) ? '' : htmlspecialchars(shCleanUpDesc($headData['metaTags']['standard']['robots']), ENT_COMPAT, 'UTF-8');
}
}
if (!is_null($shCustomLangTag)) {
$shLang = $shCustomLangTag;
$shPageInfo->pageLangTag = $shCustomLangTag;
if (strpos($buffer, '<meta http-equiv="Content-Language"') !== false) {
$buffer = ShlSystem_Strings::pr('/\\<\\s*meta\\s+http-equiv\\s*=\\s*"Content-Language".*\\/\\>/isUu', '<meta http-equiv="Content-Language" content="' . $shCustomLangTag . '" />', $buffer);
} else {
$tagsToInsert .= "\n" . '<meta http-equiv="Content-Language" content="' . $shCustomLangTag . '" />';
}
}
// custom handling of canonical
$canonicalPattern = '/\\<\\s*link[^>]+rel\\s*=\\s*"canonical[^>]+\\/\\>/isUu';
$matches = array();
$canonicalCount = preg_match_all($canonicalPattern, $buffer, $matches);
// more than one canonical already: kill them all
if ($canonicalCount > 1 && Sh404sefFactory::getConfig()->removeOtherCanonicals) {
$buffer = ShlSystem_Strings::pr($canonicalPattern, '', $buffer);
$canonicalCount = 0;
}
// more than one and J3: must be the one inserted by J3 SEF plugin
if ($canonicalCount > 0 && Sh404sefFactory::getConfig()->removeOtherCanonicals && version_compare(JVERSION, '3.0', 'ge')) {
// kill it, if asked to
$buffer = ShlSystem_Strings::pr($canonicalPattern, '', $buffer);
$canonicalCount = 0;
}
// if there' a custom canonical for that page, insert it, or replace any existing ones
if (!empty($shCanonicalTag) && $canonicalCount == 0) {
// insert a new canonical
$tagsToInsert .= "\n" . '<link href="' . htmlspecialchars($shCanonicalTag, ENT_COMPAT, 'UTF-8') . '" rel="canonical" />' . "\n";
} else {
if (!empty($shCanonicalTag)) {
// replace existing canonical
$buffer = ShlSystem_Strings::pr($canonicalPattern, '<link href="' . htmlspecialchars($shCanonicalTag, ENT_COMPAT, 'UTF-8') . '" rel="canonical" />', $buffer);
}
}
// insert all tags in one go
if (!empty($tagsToInsert)) {
$buffer = shInsertCustomTagInBuffer($buffer, '<head>', 'after', $tagsToInsert, 'first');
}
// remove Generator tag
if ($sefConfig->shRemoveGeneratorTag) {
$buffer = ShlSystem_Strings::pr('/<meta\\s*name="generator"\\s*content=".*\\/>/isUu', '', $buffer);
}
// put <h1> tags around content elements titles
if ($sefConfig->shPutH1Tags) {
if (strpos($buffer, 'class="componentheading') !== false) {
$buffer = ShlSystem_Strings::pr('/<div class="componentheading([^>]*)>\\s*(.*)\\s*<\\/div>/isUu', '<div class="componentheading$1><h1>$2</h1></div>', $buffer);
$buffer = ShlSystem_Strings::pr('/<td class="contentheading([^>]*)>\\s*(.*)\\s*<\\/td>/isUu', '<td class="contentheading$1><h2>$2</h2></td>', $buffer);
} else {
// replace contentheading by h1
$buffer = ShlSystem_Strings::pr('/<td class="contentheading([^>]*)>\\s*(.*)\\s*<\\/td>/isUu', '<td class="contentheading$1><h1>$2</h1></td>', $buffer);
}
}
// version x : if multiple h1 headings, replace them by h2
if ($sefConfig->shMultipleH1ToH2 && substr_count(JString::strtolower($buffer), '<h1>') > 1) {
$buffer = str_replace('<h1>', '<h2>', $buffer);
$buffer = str_replace('<H1>', '<h2>', $buffer);
$buffer = str_replace('</h1>', '</h2>', $buffer);
$buffer = str_replace('</H1>', '</h2>', $buffer);
}
// V 1.3.1 : replace outbounds links by internal redirects
if (sh404SEF_REDIRECT_OUTBOUND_LINKS) {
$tmp = preg_replace_callback('/<\\s*a\\s*href\\s*=\\s*"(.*)"/isUu', 'shDoRedirectOutboundLinksCallback', $buffer);
if (empty($tmp)) {
ShlSystem_Log::error('shlib', '%s::%s::%d: %s', __CLASS__, __METHOD__, __LINE__, 'RegExp failed: invalid character on page ' . Sh404sefFactory::getPageInfo()->currentSefUrl);
} else {
$buffer = $tmp;
}
}
// V 1.3.1 : add symbol to outbounds links
if ($sefConfig->shInsertOutboundLinksImage) {
$tmp = preg_replace_callback("/<\\s*a\\s*href\\s*=\\s*(\"|').*(\"|')\\s*>.*<\\/a>/isUu", 'shDoInsertOutboundLinksImageCallback', $buffer);
if (empty($tmp)) {
ShlSystem_Log::error('shlib', '%s::%s::%d: %s', __CLASS__, __METHOD__, __LINE__, 'RegExp failed: invalid character on page ' . Sh404sefFactory::getPageInfo()->currentSefUrl);
} else {
$buffer = $tmp;
}
}
// fix homepage link when using Joomfish in non default languages, error in joomla mainmenu helper
/*
if (sh404SEF_PROTECT_AGAINST_BAD_NON_DEFAULT_LANGUAGE_MENU_HOMELINK && !shIsDefaultLang( $shPageInfo->currentLanguageTag)) {
$badHomeLink = preg_quote(JURI::base());
$targetLang = explode( '-', $shPageInfo->currentLanguageTag);
$goodHomeLink = rtrim(JURI::base(), '/') . $sefConfig->shRewriteStrings[$sefConfig->shRewriteMode] . $targetLang[0] . '/';
$buffer = preg_replace( '#<div class="module_menu(.*)href="' . $badHomeLink . '"#isU',
'<div class="module_menu$1href="' . $goodHomeLink . '"', $buffer);
$buffer = preg_replace( '#<div class="moduletable_menu(.*)href="' . $badHomeLink . '"#isU',
'<div class="moduletable_menu$1href="' . $goodHomeLink . '"', $buffer);
}
*/
// all done
return $buffer;
}
}
function shIsCurrentPageHome()
{
$currentPage = shSortUrl(ShlSystem_Strings::pr('/(&|\\?)lang=[a-zA-Z]{2,3}/iu', '', empty($_SERVER['QUERY_STRING']) ? '' : $_SERVER['QUERY_STRING']));
// V 1.2.4.t
$currentPage = JString::ltrim(str_replace('index.php', '', $currentPage), '/');
$currentPage = JString::ltrim($currentPage, '?');
$shHomePage = ShlSystem_Strings::pr('/(&|\\?)lang=[a-zA-Z]{2,3}/iu', '', Sh404sefFactory::getPageInfo()->homeLink);
$shHomePage = JString::ltrim(str_replace('index.php', '', $shHomePage), '/');
$shHomePage = JString::ltrim($shHomePage, '?');
return $currentPage == $shHomePage;
}
