function parseForDomen($domen)
{
$url = str_replace("{DOMEN}", $domen, $this->templateUrl);
$main = 0;
$rest = 0;
$without = 0;
$res = $this->parseFromUrl($url);
if (PEAR::isError($res)) {
return $res;
}
foreach ($this->items as $link) {
$u = parse_url($link["url"]);
if ((!isset($u["path"]) || $u["path"] == "/") && !isset($u["query"])) {
$main++;
} else {
$rest++;
}
if ($this->withoutOur) {
$host = isset($u["host"]) ? $u["host"] : "";
if (strpos($host, "." . removeWWW($domen)) === false || strcmp($host, $domen) != 0) {
$without++;
}
}
}
return array("main" => $main, "rest" => $rest, "withoutOur" => $without);
}
function koodimonni_mail_from($email)
{
if (endsWith($email, '@localhost') || empty($email)) {
$parsed = parse_url(get_site_url());
$hostname = removeWWW($parsed['host']);
return "no-reply@{$hostname}";
}
}
function HandleCheckTIC()
{
header("Content-type: text/html; charset=windows-1251");
require_once "YandexTICGrabber.php";
$json = new Services_JSON();
$result = array("rowNum" => (int) $_REQUEST["rowNum"], "url" => $_REQUEST["url"]);
$proxyService = new ProxiesService("yandex", SERVER_IP);
$res = $proxyService->getProxyData();
if (PEAR::isError($res)) {
$result["error"] = $res->getMessage();
print $json->encode($result);
exit;
}
if (isset($_REQUEST["delay"]) && $_REQUEST["delay"] == "true") {
if (isset($_REQUEST["delayValue"])) {
sleep((int) $_REQUEST["delayValue"]);
} else {
sleep(DELAY_BEFORE_REQUEST);
}
}
$_REQUEST["url"] = trim($_REQUEST["url"]);
$parser =& new YandexTICGrabber();
$parser->showInfo = false;
$cy = $parser->parseForDomen(removeWWW($_REQUEST["url"]));
$proxyService->increase();
$cyNoWww = $parser->parseForDomen("www." . removeWWW($_REQUEST["url"]));
$proxyService->increase();
$cy = max($cy["num"], $cyNoWWW["num"]);
if ($cy < 0) {
$result["error"] = "Wrong CY";
print $json->encode($result);
exit;
}
$result["num"] = $cy;
print $json->encode($result);
}
function HandleCheckPriceLink()
{
header("Content-type: text/html; charset=windows-1251");
require_once "YandexTICGrabber.php";
require_once "CheckPageRankGrabber.php";
$json = new Services_JSON();
$result = array("url" => $_REQUEST["url"], "rowNum" => (int) $_REQUEST["rowNum"]);
$proxyService = new ProxiesService("yandex", SERVER_IP);
$res = $proxyService->getProxyData();
if (PEAR::isError($res)) {
$result["error"] = $res->getMessage();
print $json->encode($result);
exit;
}
if (isset($_REQUEST["delay"]) && $_REQUEST["delay"] == "true") {
if (isset($_REQUEST["delayValue"])) {
sleep((int) $_REQUEST["delayValue"]);
} else {
sleep(DELAY_BEFORE_REQUEST);
}
}
$_REQUEST["url"] = trim($_REQUEST["url"]);
$parserTic =& new YandexTICGrabber();
$parserTic->showInfo = false;
$resultTic = $parserTic->parseForDomen(removeWWW($_REQUEST["url"]));
$proxyService->increase();
$resultTicNoWww = $parserTic->parseForDomen("www." . removeWWW($_REQUEST["url"]));
$resultTic = max($resultTic["num"], $resultTicNoWww["num"]);
if ($resultTic < 0) {
$result["error"] = "Wrong CY";
print $json->encode($result);
exit;
}
$proxyService->increase();
$parserRank =& new CheckPageRankGrabber();
$parserRank->showInfo = false;
$resultRank = $parserRank->parseForDomen($_REQUEST["url"]);
$pageRank = max($resultRank["www"], $resultRank["nowww"]);
if ($pageRank < 0) {
$result["error"] = "Wrong Page Rank";
print $json->encode($result);
exit;
}
if (strpos($_REQUEST["url"], "http://") === false || strpos($_REQUEST["url"], "http://") > 0) {
$_REQUEST["url"] = "http://" . $_REQUEST["url"];
}
$res = getLinksForUrlDOM($_REQUEST["url"]);
if (PEAR::isError($res)) {
$result["error"] = $res->getMessage();
print $json->encode($result);
exit;
}
list($linksCount, $linksCountNoIndex, $linksCountIn, $linksCountOut, $linkCountOur) = $res;
$lCount = $linksCountOut + $linkCountOur;
switch ((int) $_REQUEST["linktype"]) {
case 1:
$result["num1"] = get_face_value($pageRank, $resultTic, -1);
$result["num"] = get_face_value($pageRank, $resultTic, $linksCountOut + $linkCountOur);
break;
case 2:
$result["num1"] = get_inner_value($pageRank, $resultTic, -1);
$result["num"] = get_inner_value($pageRank, $resultTic, $linksCountOut + $linkCountOur);
break;
}
$result["linksCountOut"] = $lCount;
$result["pr"] = $pageRank;
$result["cy"] = $resultTic;
print $json->encode($result);
}
function getLinksForUrl($url, $useProxy = true)
{
$linksCount = 0;
$linksCountNoIndex = 0;
$linkCountIn = 0;
$linkCountOut = 0;
$linkCountOur = 0;
$params = array();
if ($useProxy) {
$proxyService = new ProxiesService();
$res = $proxyService->getRandomProxyData();
if (PEAR::isError($res)) {
return $res;
}
$params["proxyData"] = $res;
}
$pageContent = UrlUtils::loadPage($url, $params);
if (strlen($pageContent) == 0) {
return new PEAR_Error("Empty content");
}
$pattern = "{<a[^>]+?(href=\".+?\"|href='.+?'|href=.+?\\s)[^>]*>}si";
if (!preg_match_all($pattern, $pageContent, $matches, PREG_SET_ORDER)) {
return array($linksCount, $linksCountNoIndex, $linkCountIn, $linkCountOut, $linkCountOur);
}
$linksCount = sizeof($matches);
$currUrl = parse_url($url);
foreach ($matches as $m) {
$m[1] = strtolower($m[1]);
$url = removeQuotes(str_replace("href=", "", $m[1]));
$u = parse_url($url);
if (!isset($u["host"])) {
$linkCountIn++;
} else {
if (strcmp(removeWWW($u["host"]), removeWWW($currUrl["host"])) == 0) {
$linkCountIn++;
} else {
if (strpos(removeWWW($u["host"]), "." . removeWWW($currUrl["host"])) !== false) {
$linkCountOur++;
} else {
$linkCountOut++;
}
}
}
}
$pageContent = preg_replace("{<noindex>.+?</noindex>}si", "", $pageContent);
preg_match_all($pattern, $pageContent, $matches, PREG_SET_ORDER);
$linksCountNoIndex = sizeof($matches);
return array($linksCount, $linksCountNoIndex, $linkCountIn, $linkCountOut, $linkCountOur);
}
