/**
* Parses a validation response from a CAS server to see if the returning CAS request is valid
*
* @param string $strResults xml or plain text response from cas server
* @return bool true if valid, false otherwise
* @exception throws exception if cannot parse response or invalid version
*/
private function isValid()
{
// values from the request
$strTicket = $this->request->getProperty("ticket");
// configuration settings
$configCasValidate = $this->registry->getConfig("CAS_VALIDATE", true);
// figure out which type of response this is based on the service url
$arrURL = explode("/", $configCasValidate);
$service = array_pop($arrURL);
// now get it!
$strUrl = $configCasValidate . "?ticket=" . $strTicket . "&service=" . urlencode($this->validate_url);
$strResults = Xerxes_Framework_Parser::request($strUrl);
// validate is plain text
if ($service == "validate") {
$arrMessage = explode("\n", $strResults);
if (count($arrMessage) >= 2) {
if ($arrMessage[0] == "yes") {
return $arrMessage[1];
}
} else {
throw new Exception("Could not parse CAS validation response.");
}
} elseif ($service == "serviceValidate" || $service == "proxyValidate") {
// these are XML based
$objXml = new DOMDocument();
$objXml->loadXML($strResults);
$strCasNamespace = "http://www.yale.edu/tp/cas";
$objUser = $objXml->getElementsByTagNameNS($strCasNamespace, "user")->item(0);
$objFailure = $objXml->getElementsByTagNameNS($strCasNamespace, "authenticationFailure")->item(0);
if ($objUser != null) {
if ($objUser->nodeValue != "") {
return $objUser->nodeValue;
} else {
throw new Exception("CAS validation response missing username value");
}
} elseif ($objFailure != null) {
// see if error, rather than failed authentication
if ($objFailure->getAttribute("code") == "INVALID_REQUEST") {
throw new Exception("Invalid request to CAS server: " . $objFailure->nodeValue);
}
} else {
throw new Exception("Could not parse CAS validation response.");
}
} else {
throw new Exception("Unsupported CAS version.");
}
// if we got this far, the request was invalid
return false;
}
public function doExecute()
{
$configToken = $this->registry->getConfig("BX_TOKEN", false);
if ($configToken != null) {
$configBX = $this->registry->getConfig("BX_SERVICE_URL", false, "http://recommender.service.exlibrisgroup.com/service");
$configLinkResolver = $this->registry->getConfig("LINK_RESOLVER_ADDRESS", true);
$configSID = $this->registry->getConfig("APPLICATION_SID", false, "calstate.edu:xerxes");
$configMaxRecords = $this->registry->getConfig("BX_MAX_RECORDS", false, "10");
$configMinRelevance = $this->registry->getConfig("BX_MIN_RELEVANCE", false, "0");
$open_url = $this->request->getData("//openurl_kev_co");
$url = $configBX . "/recommender/openurl?token={$configToken}&{$open_url}&res_dat=source=global&threshold={$configMinRelevance}&maxRecords={$configMaxRecords}";
try {
$xml = Xerxes_Framework_Parser::request($url, 4);
if ($xml == "") {
throw new Exception("No response from bx service");
}
} catch (Exception $e) {
trigger_error("Could not get result from bX service: " . $e->getTraceAsString(), E_USER_WARNING);
return 1;
}
// header("Content-type: text/xml"); echo $xml; exit;
$doc = new Xerxes_BxRecord_Document();
$doc->loadXML($xml);
$objXml = new DOMDocument();
$objXml->loadXML("<recommendations />");
$results = $doc->records();
$x = 0;
if (count($results) > 1) {
foreach ($doc->records() as $record) {
// first one is the same document
if ($x == 0) {
$x++;
continue;
}
$objRecord = $objXml->createElement("record");
$objXml->documentElement->appendChild($objRecord);
$objImport = $objXml->importNode($record->toXML()->documentElement, true);
$objRecord->appendChild($objImport);
$strOpenURL = $record->getOpenURL($configLinkResolver, $configSID);
$objOpenURL = $objXml->createElement("url_open", Xerxes_Framework_Parser::escapeXML($strOpenURL));
$objRecord->appendChild($objOpenURL);
}
}
$this->request->addDocument($objXml);
}
return 1;
}
/**
* Checks the spelling of the supplied query and offers spelling suggestion
*
* @param string $strQuery supplied query
* @param string $strYahooId yahoo access id
* @param string $strAltLocation an alternate location for the Yahoo spell check, in case you need to balanace
* @return string spelling suggesting
*/
public function checkSpelling($strQuery, $strYahooId, $strAltLocation = "")
{
return "";
if ($strAltLocation != "") {
$this->url = $strAltLocation . "?appid=" . $strYahooId . "&query=" . urlencode($strQuery);
} else {
$this->url = "http://api.search.yahoo.com/WebSearchService/V1/spellingSuggestion?appid=" . $strYahooId . "&query=" . urlencode($strQuery);
}
$strResponse = Xerxes_Framework_Parser::request($this->url);
$objSpelling = new DOMDocument();
$objSpelling->loadXML($strResponse);
if ($objSpelling->getElementsByTagName("Result")->item(0) != null) {
return $objSpelling->getElementsByTagName("Result")->item(0)->nodeValue;
} else {
return "";
}
}
public function doExecute()
{
$configMemory = $this->registry->getConfig("HARVEST_MEMORY_LIMIT", false, "500M");
ini_set("memory_limit", $configMemory);
echo "\n\nSFX INSTITUTIONAL HOLDINGS POPULATION \n\n";
// You can define the export file on sfx as having an instance extension, so
// give the client the opportunity to define that here
$strInstance = $this->request->getProperty("instance");
if ($strInstance != "") {
$strInstance = "-" . $strInstance;
}
// construct the address to Google Scholar institutional
// holdings file on SFX. Either SFX specific config, or
// general link resolver config.
$configSfx = $this->registry->getConfig("ALTERNATE_FULLTEXT_HARVEST_ADDRESS", false, $this->registry->getConfig("LINK_RESOLVER_ADDRESS", false));
if (!$configSfx) {
throw new Exception("Can not run populate action, no link resolver address configured. " . "Need config ALTERNATE_FULLTEXT_HARVEST_ADDRESS or LINK_RESOLVER_ADDRESS.");
}
// fire-up a transaction with the database
$objData = new Xerxes_DataMap();
$objData->beginTransaction();
// clear old data
echo " Flushing SFX fulltext table . . . ";
$objData->clearFullText();
echo "done.\n";
// try to get the data from sfx
$done = false;
$x = 0;
while ($done == false) {
$x++;
$strUrl = $configSfx . "/cgi/public/get_file.cgi?file=institutional_holding" . $strInstance . '-.prt' . str_pad($x, 2, '0', STR_PAD_LEFT) . ".xml";
echo " Pulling down SFX inst holding file ({$x}) . . . ";
try {
$strResponse = Xerxes_Framework_Parser::request($strUrl);
$objXml = new SimpleXMLElement($strResponse);
} catch (Exception $e) {
if ($x == 1) {
throw new Exception("cannot get institutional holding file from sfx: '{$strUrl}'. " . "If this is the correct SFX server address, make sure your SFX allows access to " . "institutional holding file from this IP address in config/get_file_restriction.config " . "on SFX server.");
}
$done = true;
}
echo "done.\n";
if (!$done) {
echo " Processing file . . . ";
$objItems = $objXml->xpath("//item[@type != 'other']");
if ($objItems == false) {
throw new Exception("could not find items in inst holding file.");
}
echo "done.\n";
echo " Adding to database . . . ";
foreach ($objItems as $objItem) {
foreach ($objItem->coverage as $objCoverage) {
$objFullText = new Xerxes_Data_Fulltext();
$objFullText->issn = (string) $objItem->issn;
$objFullText->issn = str_replace("-", "", $objFullText->issn);
$objFullText->title = (string) $objItem->title;
$objFullText->title = urlencode($objFullText->title);
$objFullText->title = substr(Xerxes_Framework_Parser::strtolower($objFullText->title), 0, 100);
$objFullText->startdate = (int) $objCoverage->from->year;
$objFullText->enddate = (int) $objCoverage->to->year;
if ($objFullText->enddate == 0) {
$objFullText->enddate = 9999;
}
$objFullText->embargo = (int) $objCoverage->embargo->days_not_available;
$objFullText->updated = date("YmdHis");
// add it to the database
$objData->addFulltext($objFullText);
}
}
echo "done.\n";
}
}
echo " Commiting changes . . . ";
$objData->commit();
echo "done.\n";
return 1;
}
public function checkSpelling()
{
$registry = Xerxes_Framework_Registry::getInstance();
$strAltYahoo = $registry->getConfig("ALTERNATE_YAHOO_LOCATION", false);
$configYahooID = $registry->getConfig("YAHOO_ID", false, "calstate");
$spell_return = array();
// we'll return this one
for ($x = 0; $x < count($this->query_list); $x++) {
$term = $this->query_list[$x];
$url = "";
if ($strAltYahoo != "") {
$url = $strAltYahoo;
} else {
$url = "http://api.search.yahoo.com/WebSearchService/V1/spellingSuggestion";
}
$url .= "?appid=" . $configYahooID . "&query=" . urlencode($term->phrase);
$strResponse = Xerxes_Framework_Parser::request($url);
$objSpelling = new DOMDocument();
$objSpelling->loadXML($strResponse);
if ($objSpelling->getElementsByTagName("Result")->item(0) != null) {
$term->spell_correct = $objSpelling->getElementsByTagName("Result")->item(0)->nodeValue;
$spell_return[$term->id] = $term->spell_correct;
}
// also put it here so we can return it
$this->query_list[$x] = $term;
}
return $spell_return;
}
/**
* Fetches and normalize the API data
*
* @param string $url url of patron dump or pint test
* @return array patron data
*/
private function getContent($url)
{
$arrRawData = array();
$arrData = array();
// get the data and strip out html tags
$strResponse = Xerxes_Framework_Parser::request($url);
$strResponse = trim(strip_tags($strResponse));
if ($strResponse == "") {
throw new Exception("Could not connect to Innovative Patron API");
} else {
// cycle thru each line in the response, splitting each
// on the equal sign into an associative array
$arrRawData = explode("\n", $strResponse);
foreach ($arrRawData as $strLine) {
$arrLine = explode("=", $strLine);
// strip out the code, leaving just the attribute name
$arrLine[0] = preg_replace("/\\[[^\\]]{1,}\\]/", "", $arrLine[0]);
$arrData[trim($arrLine[0])] = trim($arrLine[1]);
}
}
return $arrData;
}
/**
* Fetch the data from Metalib and check for errors
*
* @param string $url url of the request
*/
private function getResponse($url, $timeout = null, $retry = 0)
{
// metalib takes little care to ensure propper encoding of its xml, so we will set
// recover to true here in order to allow libxml to recover from errors and continue
// processing the document
$objXml = new DOMDocument();
$objXml->recover = true;
// time out, should only be defined for search-related functions
$strResponse = Xerxes_Framework_Parser::request($url, $timeout);
if ($strResponse == "") {
throw new Exception("Cannot process search at this time.");
}
$objXml->loadXML($strResponse);
// no response?
if ($objXml->documentElement == null) {
throw new Exception("cannot connect to metalib server");
}
// error in response
if ($objXml->getElementsByTagName("error_code") != null) {
// for easier handling
$xml = simplexml_import_dom($objXml->documentElement);
foreach ($xml->xpath("//global_error|//local_error") as $error) {
$error_code = (int) $error->error_code;
$error_text = (string) $error->error_text;
// now examine them
// metalib session has timed out!
if ($error_code == 0151) {
// this particular metalib error message is confusing to end-users, so make it more clear
$error_text = "The Metalib session has expired";
// also try to re-up the session in the event metalib was restarted or something
if ($retry == 0) {
$this->session = $this->session();
return $this->getResponse($url, $timeout, 1);
}
}
// these are just warnings
if ($error_code == 2114 || $error_code == 6039 || $error_code == 6033 || $error_code == 6034 || $error_code == 6023 || $error_code == 134 || $error_code == 6022) {
trigger_error("Metalib warning ({$error_code}): {$error_text}", E_USER_WARNING);
} else {
throw new Exception("Metalib exception ({$error_code}): {$error_text}");
}
}
}
return $objXml;
}
