function add_all_url_($url, $old, $numm, $ooo, $site_id, $include_word, $not_include_word)
{
if (!is_url($url)) {
return false;
}
global $db, $config;
$snoopy = new Snoopy();
//国外snoopy程序
$snoopy->fetchlry($url);
$links = $snoopy->resulry;
if (!is_array($links)) {
return;
}
$links = check_wai($links, $numm, $url);
$links = array_values(array_unique($links));
$title = $snoopy->title;
$fulltxt = $snoopy->fulltxt;
$lrymd5 = md5($fulltxt);
$pagesize = $snoopy->pagesize;
$description = $snoopy->description;
$keywords = $snoopy->keywords;
$updatetime = time();
if ($title == "") {
$title = str_cut($fulltxt, 65);
}
//读取url,更新内容
$array = array('lrymd5' => $lrymd5, 'title' => $title, 'fulltxt' => $fulltxt, 'description' => $description, 'keywords' => $keywords, 'pagesize' => $pagesize, 'updatetime' => $updatetime);
$db->update("ve123_links", $array, "url='" . $url . "'");
$all_num = count($links);
$temp_links = array();
$cha_temp = array();
//开始读取 ve123_links_temp 中所有site_id 为$site_id 的url 然后和抓取的 $links 数组比较,将得到的差集创建到 ve123_links_temp 中
$query = $db->query("select url from ve123_links_temp where url like '%" . getdomain($url) . "%'");
while ($row = $db->fetch_array($query)) {
$temp_links[] = rtrim($row[url], "/");
}
$cha_temp = array_diff($links, $temp_links);
foreach ((array) $cha_temp as $value) {
$arral = array('url' => $value, 'site_id' => $site_id);
$db->insert("ve123_links_temp", $arral);
}
//开始读取 ve123_links 中所有site_id 为 $site_id 的url 然后和抓取的 $links 数组比较,将得到的差集创建到 ve123_links 中 合集则输出 已存在了
$query = $db->query("select url from ve123_links where url like '%" . getdomain($url) . "%'");
while ($row = $db->fetch_array($query)) {
$new_links[] = rtrim($row[url], "/");
}
$he_links = array_intersect($links, $new_links);
$he_num = count($he_links);
$cha_links = array_diff($links, $new_links);
$cha_num = count($cha_links);
foreach ((array) $cha_links as $value) {
if (check_include($value, $include_word, $not_include_word)) {
$array = array('url' => $value, 'site_id' => $site_id, 'level' => '1');
$db->insert("ve123_links", $array);
}
}
printLinksReport($cha_num, $all_num, $cl = 0);
echo "<a href=" . $old . " target=_blank>" . $old . "</a>";
ob_flush();
flush();
}
