/**
* Returns the normalized form of the given page title, using the
* normalization rules of the given site. If the given title is a redirect,
* the redirect weill be resolved and the redirect target is returned.
*
* @note This actually makes an API request to the remote site, so beware
* that this function is slow and depends on an external service.
*
* @see Site::normalizePageName
*
* @since 1.27
*
* @param string $pageName
* @param string $apiUrl
*
* @return string
* @throws \MWException
*/
public function normalizePageName($pageName, $apiUrl)
{
// Check if we have strings as arguments.
if (!is_string($pageName)) {
throw new \MWException('$pageName must be a string');
}
// Go on call the external site
// Make sure the string is normalized into NFC (due to T42017)
// but do nothing to the whitespaces, that should work appropriately.
// @see https://phabricator.wikimedia.org/T42017
$pageName = Validator::cleanUp($pageName);
// Build the args for the specific call
$args = ['action' => 'query', 'prop' => 'info', 'redirects' => true, 'converttitles' => true, 'format' => 'json', 'titles' => $pageName];
$url = wfAppendQuery($apiUrl, $args);
// Go on call the external site
// @todo we need a good way to specify a timeout here.
$ret = $this->http->get($url, [], __METHOD__);
if ($ret === false) {
wfDebugLog("MediaWikiSite", "call to external site failed: {$url}");
return false;
}
$data = FormatJson::decode($ret, true);
if (!is_array($data)) {
wfDebugLog("MediaWikiSite", "call to <{$url}> returned bad json: " . $ret);
return false;
}
$page = static::extractPageRecord($data, $pageName);
if (isset($page['missing'])) {
wfDebugLog("MediaWikiSite", "call to <{$url}> returned a marker for a missing page title! " . $ret);
return false;
}
if (isset($page['invalid'])) {
wfDebugLog("MediaWikiSite", "call to <{$url}> returned a marker for an invalid page title! " . $ret);
return false;
}
if (!isset($page['title'])) {
wfDebugLog("MediaWikiSite", "call to <{$url}> did not return a page title! " . $ret);
return false;
}
return $page['title'];
}
/**
* The ultimate convenience function! Clean up invalid UTF-8 sequences,
* and convert to normal form C, canonical composition.
*
* Fast return for pure ASCII strings; some lesser optimizations for
* strings containing only known-good characters. Not as fast as toNFC().
*
* @param string $string a UTF-8 string
* @return string a clean, shiny, normalized UTF-8 string
*/
static function cleanUp($string)
{
return Validator::cleanUp($string);
}
/**
* Normalize string into NFC by using the cleanup method from UtfNormal.
*
* @param string $inputString The actual string to process.
*
* @return string
*/
public function cleanupToNFC($inputString)
{
$cleaned = $inputString;
$cleaned = $this->trimBadChars($cleaned);
$cleaned = Validator::cleanUp($cleaned);
return $cleaned;
}
