/**
* Fetches the location from the hostip.info web service
*
* @param string $ip
*/
function get_location($location_map)
{
$city = '';
$state = '';
$country = '';
$country_code = '';
$latitude = '';
$longitude = '';
// check to see if ip is in map
if (array_key_exists('ip_address', $location_map) && !empty($location_map['ip_address']) && empty($location_map['country'])) {
// check to see if ip is valid and not a private address
if (filter_var($location_map['ip_address'], FILTER_VALIDATE_IP, FILTER_FLAG_IPV4 | FILTER_FLAG_NO_PRIV_RANGE)) {
// create crawler
$crawler = new owa_http();
$crawler->read_timeout = owa_coreAPI::getSetting('base', 'ws_timeout');
// hit web service
$crawler->fetch(sprintf($this->ws_url, $location_map['ip_address']));
owa_coreAPI::debug(sprintf("HostIp web service response code: %s", $crawler->crawler->response_code));
$location = $crawler->crawler->results;
// replace delimiter
$location = str_replace("\n", "|", $location);
// convert string to array
$loc_array = explode("|", $location);
$result = array();
// convert array to multi dimensional array
foreach ($loc_array as $k => $v) {
if (!empty($v)) {
list($name, $value) = explode(":", $v, 2);
$result[$name] = $value;
}
}
// parse the city line of response
if (isset($result['City']) && !empty($result['City'])) {
// lowercase
$result['City'] = strtolower($result['City']);
// explode into array
$city_array = explode(',', $result['City']);
// city name is always first
$city = $city_array[0];
// if there is a second element then it's a state
if (isset($city_array[1])) {
$state = $city_array[1];
}
}
// parse country line of response
if (isset($result['Country']) && !empty($result['Country'])) {
//lowercase
$result['Country'] = strtolower($result['Country']);
// set country
$country_parts = explode('(', trim($result['Country']));
$country = $country_parts[0];
// if there is a second element then it's a country code.
if (isset($country_parts[1])) {
$country_code = substr($country_code, 0, -1);
}
// debug
owa_coreAPI::debug('Parse of Hostip country string: ' . $result['Country'] . ' c: ' . $country . ' cc: ' . $country_code);
}
// set latitude
if (isset($result['Latitude']) && !empty($result['Latitude'])) {
$latitude = $result['Latitude'];
}
// set longitude
if (isset($result['Longitude']) && !empty($result['Longitude'])) {
$longitude = $result['Longitude'];
}
}
// fail safe checks for empty, unknown or private adddress labels
// check to make sure values are not "private address" contain "unknown" or "xx"
if (empty($city) || strpos($city, 'private') || strpos($city, 'unknown')) {
$city = '(not set)';
}
// check state
if (empty($state) || strpos($state, 'private') || strpos($state, 'unknown')) {
$state = '(not set)';
}
// check country
if (empty($country) || strpos($country, 'unknown') || strpos($country, 'private')) {
$country = '(not set)';
}
// check country code
if (empty($country_code) || strpos($country_code, 'xx') || strpos($country_code, 'unknown') || strpos($country_code, 'private')) {
$country_code = '(not set)';
}
$location_map['city'] = strtolower(trim($city));
$location_map['state'] = strtolower(trim($state));
$location_map['country'] = strtolower(trim($country));
$location_map['country_code'] = strtoupper(trim($country_code));
$location_map['latitude'] = trim($latitude);
$location_map['longitude'] = trim($longitude);
// log headers if status is not a 200
if (isset($crawler->response_code) && !strpos($crawler->response_code, '200')) {
owa_coreAPI::debug(sprintf("HostIp web service response headers: %s", print_r($crawler->crawler->headers, true)));
}
}
return $location_map;
}
/**
* Notify Event Handler
*
* @param unknown_type $event
* @access public
*/
function notify($event)
{
$medium = $event->get('medium');
// if there is no session referer then return
if (!$event->get('session_referer')) {
return OWA_EHS_EVENT_HANDLED;
}
// Make entity
$r = owa_coreAPI::entityFactory('base.referer');
$r->load($event->get('referer_id'));
if (!$r->wasPersisted()) {
// Set id
if ($event->get('referer_id')) {
$r->set('id', $event->get('referer_id'));
} else {
$r->set('id', $r->generateId($event->get('session_referer')));
}
// set referer url
$r->set('url', $event->get('session_referer'));
// Set site
$url = owa_lib::parse_url($event->get('session_referer'));
$r->set('site', $url['host']);
if ($medium === 'organic-search') {
$r->set('is_searchengine', true);
}
// set title. this will be updated later by the crawler.
$r->set('page_title', '(not set)');
// Crawl and analyze refering page
if (owa_coreAPI::getSetting('base', 'fetch_refering_page_info') && $medium != 'organic-search') {
//owa_coreAPI::debug('hello from logReferer');
$crawler = new owa_http();
//$crawler->fetch($this->params['HTTP_REFERER']);
$res = $crawler->getRequest($event->get('session_referer'));
owa_coreAPI::debug('http request response: ' . print_r($res, true));
//Extract Title
$title = trim($crawler->extract_title());
if ($title) {
$r->set('page_title', owa_lib::utf8Encode($title));
}
$se = $r->get('is_searchengine');
//Extract anchortext and page snippet but not if it's a search engine...
if ($se != true) {
$snippet = $crawler->extract_anchor_snippet($event->get('inbound_page_url'));
if ($snippet) {
if (function_exists('iconv')) {
$snippet = iconv('UTF-8', 'UTF-8//TRANSLIT', $snippet);
}
$r->set('snippet', $snippet);
}
$anchortext = $crawler->anchor_info['anchor_text'];
if ($anchortext) {
$r->set('refering_anchortext', owa_lib::utf8Encode($anchortext));
}
}
}
// Persist to database
$ret = $r->create();
if ($ret) {
return OWA_EHS_EVENT_HANDLED;
} else {
return OWA_EHS_EVENT_FAILED;
}
} else {
owa_coreAPI::debug('Not Persisting. Referrer already exists.');
return OWA_EHS_EVENT_HANDLED;
}
}
