We don't need a precise test here because the value comes from the website
tracked source code and the URLs may look very strange.
| public static isLookLikeUrl ( string $url ) : boolean | ||
| $url | string | |
| return | boolean | |
/**
* Returns an array containing the following information:
* - referer_type
* - direct -- absence of referrer URL OR referrer URL has the same host
* - site -- based on the referrer URL
* - search_engine -- based on the referrer URL
* - campaign -- based on campaign URL parameter
*
* - referer_name
* - ()
* - piwik.net -- site host name
* - google.fr -- search engine host name
* - adwords-search -- campaign name
*
* - referer_keyword
* - ()
* - ()
* - my keyword
* - my paid keyword
* - ()
* - ()
*
* - referer_url : the same for all the referrer types
*
* @param string $referrerUrl must be URL Encoded
* @param string $currentUrl
* @param int $idSite
* @return array
*/
public function getReferrerInformation($referrerUrl, $currentUrl, $idSite)
{
$this->idsite = $idSite;
// default values for the referer_* fields
$referrerUrl = Common::unsanitizeInputValue($referrerUrl);
if (!empty($referrerUrl) && !UrlHelper::isLookLikeUrl($referrerUrl)) {
$referrerUrl = '';
}
$currentUrl = PageUrl::cleanupUrl($currentUrl);
$this->referrerUrl = $referrerUrl;
$this->referrerUrlParse = @parse_url($this->referrerUrl);
$this->currentUrlParse = @parse_url($currentUrl);
$this->typeReferrerAnalyzed = Common::REFERRER_TYPE_DIRECT_ENTRY;
$this->nameReferrerAnalyzed = '';
$this->keywordReferrerAnalyzed = '';
$this->referrerHost = '';
if (isset($this->referrerUrlParse['host'])) {
$this->referrerHost = $this->referrerUrlParse['host'];
}
$referrerDetected = $this->detectReferrerCampaign();
if (!$referrerDetected) {
if ($this->detectReferrerDirectEntry() || $this->detectReferrerSearchEngine()) {
$referrerDetected = true;
}
}
if (!empty($this->referrerHost) && !$referrerDetected) {
$this->typeReferrerAnalyzed = Common::REFERRER_TYPE_WEBSITE;
$this->nameReferrerAnalyzed = Common::mb_strtolower($this->referrerHost);
}
$referrerInformation = array('referer_type' => $this->typeReferrerAnalyzed, 'referer_name' => $this->nameReferrerAnalyzed, 'referer_keyword' => $this->keywordReferrerAnalyzed, 'referer_url' => $this->referrerUrl);
return $referrerInformation;
}
/**
* Check that the array of URLs are valid URLs
*
* @param array $urls
* @throws Exception if any of the urls is not valid
*/
public function checkUrls($urls)
{
$urls = $this->cleanParameterUrls($urls);
foreach ($urls as $url) {
if (!UrlHelper::isLookLikeUrl($url)) {
throw new Exception(sprintf(Piwik::translate('SitesManager_ExceptionInvalidUrl'), $url));
}
}
}
public function render()
{
$idSite = Common::getRequestVar('idSite');
$site = new Site($idSite);
$url = urldecode(Common::getRequestVar('url', '', 'string'));
if (!empty($url) && strpos($url, 'http://') !== 0 && strpos($url, 'https://') !== 0) {
$url = 'http://' . $url;
}
if (empty($url) || !UrlHelper::isLookLikeUrl($url)) {
$url = $site->getMainUrl();
}
$dataTable = API::getInstance()->getRank($url);
/** @var \Piwik\DataTable\Renderer\Php $renderer */
$renderer = Renderer::factory('php');
$renderer->setSerialize(false);
return $this->renderTemplate('getRank', array('urlToRank' => Url::getHostFromUrl($url), 'ranks' => $renderer->render($dataTable)));
}
/**
* Main Plugin Index
*
* @return mixed
* @throws \Exception
*/
public function index()
{
Piwik::checkUserHasSomeAdminAccess();
if (isset($_SERVER['REQUEST_METHOD']) && 'POST' == $_SERVER['REQUEST_METHOD']) {
// Cannot use Common::getRequestVar, because the function remove whitespaces and newline breaks
$postedSiteData = isset($_POST['site']) ? $_POST['site'] : null;
if (is_array($postedSiteData) && count($postedSiteData) > 0) {
foreach ($postedSiteData as $id => $site) {
if (!isset($site['css'], $site['file'])) {
continue;
}
// Check URL
if (!UrlHelper::isLookLikeUrl($site['file'])) {
$site['file'] = null;
}
API::getInstance()->saveSite($id, $site['css'], $site['file']);
}
// Redirect to, clear POST vars
$this->redirectToIndex('CustomOptOut', 'index');
return;
}
}
$view = new View('@CustomOptOut/index.twig');
Site::clearCache();
if (Piwik::hasUserSuperUserAccess()) {
$sitesRaw = APISiteManager::getInstance()->getAllSites();
} else {
$sitesRaw = APISiteManager::getInstance()->getSitesWithAdminAccess();
}
// Gets sites after Site.setSite hook was called
$sites = array_values(Site::getSites());
if (count($sites) != count($sitesRaw)) {
throw new \Exception("One or more website are missing or invalid.");
}
foreach ($sites as &$site) {
$site['alias_urls'] = APISiteManager::getInstance()->getSiteUrlsFromId($site['idsite']);
}
$view->adminSites = $sites;
$view->adminSitesCount = count($sites);
$view->language = LanguagesManager::getLanguageCodeForCurrentUser();
$view->isEditorEnabled = API::getInstance()->isCssEditorEnabled();
$view->editorTheme = API::getInstance()->getEditorTheme();
$view->showOldLinks = false;
$this->setBasicVariablesView($view);
return $view->render();
}
public function getReferrerUrl()
{
if ($this->getReferrerType() == 'search') {
if ($this->details['referer_keyword'] == API::LABEL_KEYWORD_NOT_DEFINED) {
return 'http://piwik.org/faq/general/#faq_144';
} elseif ($this->getReferrerName() == 'Google' && strpos($this->details['referer_url'], '/url')) {
$refUrl = @parse_url($this->details['referer_url']);
if (isset($refUrl['host'])) {
$url = SearchEngine::getInstance()->getBackLinkFromUrlAndKeyword('http://google.com', $this->getKeyword());
$url = str_replace('google.com', $refUrl['host'], $url);
return $url;
}
}
}
if (UrlHelper::isLookLikeUrl($this->details['referer_url'])) {
return $this->details['referer_url'];
}
return null;
}
/**
* Returns an array containing the following information:
* - referer_type
* - direct -- absence of referrer URL OR referrer URL has the same host
* - site -- based on the referrer URL
* - search_engine -- based on the referrer URL
* - campaign -- based on campaign URL parameter
*
* - referer_name
* - ()
* - piwik.net -- site host name
* - google.fr -- search engine host name
* - adwords-search -- campaign name
*
* - referer_keyword
* - ()
* - ()
* - my keyword
* - my paid keyword
* - ()
* - ()
*
* - referer_url : the same for all the referrer types
*
* @param string $referrerUrl must be URL Encoded
* @param string $currentUrl
* @param int $idSite
* @return array
*/
protected function getReferrerInformation($referrerUrl, $currentUrl, $idSite, Request $request)
{
$cacheKey = $referrerUrl . $currentUrl . $idSite;
if (isset(self::$cachedReferrer[$cacheKey])) {
return self::$cachedReferrer[$cacheKey];
}
$this->idsite = $idSite;
// default values for the referer_* fields
$referrerUrl = Common::unsanitizeInputValue($referrerUrl);
if (!empty($referrerUrl) && !UrlHelper::isLookLikeUrl($referrerUrl)) {
$referrerUrl = '';
}
$currentUrl = PageUrl::cleanupUrl($currentUrl);
$this->referrerUrl = $referrerUrl;
$this->referrerUrlParse = @parse_url($this->referrerUrl);
$this->currentUrlParse = @parse_url($currentUrl);
$this->typeReferrerAnalyzed = Common::REFERRER_TYPE_DIRECT_ENTRY;
$this->nameReferrerAnalyzed = '';
$this->keywordReferrerAnalyzed = '';
$this->referrerHost = '';
if (isset($this->referrerUrlParse['host'])) {
$this->referrerHost = $this->referrerUrlParse['host'];
}
$referrerDetected = $this->detectReferrerCampaign($request);
if (!$referrerDetected) {
if ($this->detectReferrerDirectEntry() || $this->detectReferrerSearchEngine()) {
$referrerDetected = true;
}
}
if (!$referrerDetected && !empty($this->referrerHost)) {
$this->typeReferrerAnalyzed = Common::REFERRER_TYPE_WEBSITE;
$this->nameReferrerAnalyzed = Common::mb_strtolower($this->referrerHost);
$urlsByHost = $this->getCachedUrlsByHostAndIdSite();
$directEntry = new SiteUrls();
$path = $directEntry->getPathMatchingUrl($this->referrerUrlParse, $urlsByHost);
if (!empty($path) && $path !== '/') {
$this->nameReferrerAnalyzed .= rtrim($path, '/');
}
}
$referrerInformation = array('referer_type' => $this->typeReferrerAnalyzed, 'referer_name' => $this->nameReferrerAnalyzed, 'referer_keyword' => $this->keywordReferrerAnalyzed, 'referer_url' => $this->referrerUrl);
self::$cachedReferrer[$cacheKey] = $referrerInformation;
return $referrerInformation;
}
function getRank()
{
$idSite = Common::getRequestVar('idSite');
$site = new Site($idSite);
$url = urldecode(Common::getRequestVar('url', '', 'string'));
if (!empty($url) && strpos($url, 'http://') !== 0 && strpos($url, 'https://') !== 0) {
$url = 'http://' . $url;
}
if (empty($url) || !UrlHelper::isLookLikeUrl($url)) {
$url = $site->getMainUrl();
}
$dataTable = API::getInstance()->getRank($url);
$view = new View('@SEO/getRank');
$view->urlToRank = RankChecker::extractDomainFromUrl($url);
/** @var \Piwik\DataTable\Renderer\Php $renderer */
$renderer = Renderer::factory('php');
$renderer->setSerialize(false);
$view->ranks = $renderer->render($dataTable);
return $view->render();
}
/**
* Output redirection page instead of linking directly to avoid
* exposing the referrer on the Piwik demo.
*
* @internal param string $url (via $_GET)
*/
public function redirect()
{
$url = Common::getRequestVar('url', '', 'string', $_GET);
if (!UrlHelper::isLookLikeUrl($url)) {
die('Please check the &url= parameter: it should to be a valid URL');
}
// validate referrer
$referrer = Url::getReferrer();
if (empty($referrer) || !Url::isLocalUrl($referrer)) {
die('Invalid Referrer detected - This means that your web browser is not sending the "Referrer URL" which is
required to proceed with the redirect. Verify your browser settings and add-ons, to check why your browser
is not sending this referrer.
<br/><br/>You can access the page at: ' . $url);
}
// mask visits to *.piwik.org
if (!self::isPiwikUrl($url)) {
Piwik::checkUserHasSomeViewAccess();
}
Common::sendHeader('Content-Type: text/html; charset=utf-8');
echo '<html><head><meta http-equiv="refresh" content="0;url=' . $url . '" /></head></html>';
exit;
}
public static function getHostFromUrl($url)
{
if (!UrlHelper::isLookLikeUrl($url)) {
$url = "http://" . $url;
}
return parse_url($url, PHP_URL_HOST);
}
/**
* Tests if the URL is a valid URL
*
* @param string $url
* @return bool
*/
private function isValidUrl($url)
{
return UrlHelper::isLookLikeUrl($url);
}
public static function getTrustedHostsFromConfig()
{
$trustedHosts = @Config::getInstance()->General['trusted_hosts'];
if (!is_array($trustedHosts)) {
return array();
}
foreach ($trustedHosts as &$trustedHost) {
// Case user wrote in the config, http://example.com/test instead of example.com
if (UrlHelper::isLookLikeUrl($trustedHost)) {
$trustedHost = parse_url($trustedHost, PHP_URL_HOST);
}
}
return $trustedHosts;
}
public static function getTrustedHostsFromConfig()
{
$hosts = self::getHostsFromConfig('General', 'trusted_hosts');
// Case user wrote in the config, http://example.com/test instead of example.com
foreach ($hosts as &$host) {
if (UrlHelper::isLookLikeUrl($host)) {
$host = parse_url($host, PHP_URL_HOST);
}
}
return $hosts;
}
public static function getUrlIfLookValid($url)
{
$url = PageUrl::cleanupString($url);
if (!UrlHelper::isLookLikeUrl($url)) {
Common::printDebug("WARNING: URL looks invalid and is discarded");
return false;
}
return $url;
}
private function initPiwikHost($piwikUrl = false)
{
// If core:archive command run as a web cron, we use the current hostname+path
if (empty($piwikUrl)) {
if (!empty(self::$url)) {
$piwikUrl = self::$url;
} else {
// example.org/piwik/
$piwikUrl = SettingsPiwik::getPiwikUrl();
}
}
if (!$piwikUrl) {
$this->logFatalErrorUrlExpected();
}
if (!\Piwik\UrlHelper::isLookLikeUrl($piwikUrl)) {
// try adding http:// in case it's missing
$piwikUrl = "http://" . $piwikUrl;
}
if (!\Piwik\UrlHelper::isLookLikeUrl($piwikUrl)) {
$this->logFatalErrorUrlExpected();
}
// ensure there is a trailing slash
if ($piwikUrl[strlen($piwikUrl) - 1] != '/' && !Common::stringEndsWith($piwikUrl, 'index.php')) {
$piwikUrl .= '/';
}
$this->initConfigObject($piwikUrl);
if (Config::getInstance()->General['force_ssl'] == 1) {
$piwikUrl = str_replace('http://', 'https://', $piwikUrl);
}
if (!Common::stringEndsWith($piwikUrl, 'index.php')) {
$piwikUrl .= 'index.php';
}
$this->piwikUrl = $piwikUrl;
}
/**
* Redirects the user to the specified URL.
*
* @param string $url
* @api
*/
public static function redirectToUrl($url)
{
if (UrlHelper::isLookLikeUrl($url) || strpos($url, 'index.php') === 0) {
@header("Location: {$url}");
} else {
echo "Invalid URL to redirect to.";
}
exit;
}
private function createTrackToCustomSiteUrlSetting()
{
return $this->makeSetting('customSiteUrl', $default = '', FieldConfig::TYPE_STRING, function (FieldConfig $field) {
$field->title = 'Piwik Url';
$field->uiControl = FieldConfig::UI_CONTROL_TEXT;
$field->uiControlAttributes = array('placeHolder' => 'eg. http://example.com/piwik');
$field->introduction = 'Send anonymize usage data to a custom Piwik';
$field->description = '';
$field->validate = function ($value, $setting) {
if (empty($value)) {
return;
}
if (!UrlHelper::isLookLikeUrl($value)) {
throw new Exception("URL '{$value}' seems to be not a valid URL");
}
// TODO should we check if URL exists and is valid?!? might not work if instance is not connected to internet
};
$field->transform = function ($value) {
if (empty($value)) {
return '';
}
if (!Common::stringEndsWith($value, '/piwik.php')) {
if (!Common::stringEndsWith($value, '/')) {
$value .= '/';
}
$value .= 'piwik.php';
}
return $value;
};
});
}
/**
* @dataProvider getUrls
* @group Core
*/
public function testIsUrl($url, $isValid)
{
$this->assertEquals($isValid, UrlHelper::isLookLikeUrl($url), "{$url} failed test");
}
private function initPiwikHost()
{
// If archive.php run as a web cron, we use the current hostname+path
if (!Common::isPhpCliMode()) {
// example.org/piwik/misc/cron/
$piwikUrl = Common::sanitizeInputValue(Url::getCurrentUrlWithoutFileName());
// example.org/piwik/
$piwikUrl = $piwikUrl . "../../";
} else {
// If archive.php run as CLI/shell we require the piwik url to be set
$piwikUrl = $this->isParameterSet("url", true);
if (!$piwikUrl) {
$this->logFatalErrorUrlExpected();
}
if (!\Piwik\UrlHelper::isLookLikeUrl($piwikUrl)) {
// try adding http:// in case it's missing
$piwikUrl = "http://" . $piwikUrl;
}
if (!\Piwik\UrlHelper::isLookLikeUrl($piwikUrl)) {
$this->logFatalErrorUrlExpected();
}
// ensure there is a trailing slash
if ($piwikUrl[strlen($piwikUrl) - 1] != '/') {
$piwikUrl .= '/';
}
}
$this->initConfigObject($piwikUrl);
if (Config::getInstance()->General['force_ssl'] == 1) {
$piwikUrl = str_replace('http://', 'https://', $piwikUrl);
}
$this->piwikUrl = $piwikUrl . "index.php";
}
private function createTrackToCustomSiteUrlSetting()
{
$this->customPiwikSiteUrl = new SystemSetting('customSiteUrl', 'Piwik Url');
$this->customPiwikSiteUrl->readableByCurrentUser = true;
$this->customPiwikSiteUrl->uiControlType = static::CONTROL_TEXT;
$this->customPiwikSiteUrl->uiControlAttributes = array('placeHolder' => 'eg. http://example.com/piwik');
$this->customPiwikSiteUrl->introduction = 'Send anonymize usage data to a custom Piwik';
$this->customPiwikSiteUrl->description = '';
$this->customPiwikSiteUrl->defaultValue = '';
$this->customPiwikSiteUrl->validate = function ($value, $setting) {
if (empty($value)) {
return;
}
if (!UrlHelper::isLookLikeUrl($value)) {
throw new Exception("URL '{$value}' seems to be not a valid URL");
}
// TODO should we check if URL exists and is valid?!? might not work if instance is not connected to internet
};
$this->customPiwikSiteUrl->transform = function ($value) {
if (empty($value)) {
return '';
}
if (!Common::stringEndsWith($value, '/piwik.php')) {
if (!Common::stringEndsWith($value, '/')) {
$value .= '/';
}
$value .= 'piwik.php';
}
return $value;
};
$this->addSetting($this->customPiwikSiteUrl);
}