| Serg0007 |
17.06.2010 16:49 |
PHP код:
function check_outlinks($page)
{
$content_page = file_get_contents($page) or die ("<font color=red>Îøèáêà çàïðîñà ñòðàíèöû!</font>");
$content_page = preg_replace("!<NOINDEX>!i", "<noindex>", $content_page);
$content_page = preg_replace("!</NOINDEX>!i", "</noindex>", $content_page);
if (preg_match_all("!<noindex>!Ui", $content_page, $noindex_count))
{
preg_match_all("!</noindex>!Ui", $content_page, $noindexe_count);
$full_count = count($noindex_count[0]) + count($noindexe_count[0]) + 1;
//if (!is_float($full_count/2)) echo "<font color=red>Ñóùåñòâóþò îäèíî÷íûå òýãè <noindex>, âîçìîæíà íåêîððåêòíàÿ ðàáîòà!</font><br><br>";
$noindex = 0;
$n_out = 0;
for ($i=0; $i<$full_count; $i++)
{
$pozition_begin = strpos($content_page, "<noindex>");
$pozition_end = strpos($content_page, "</noindex>");
if ($pozition_end == true && $pozition_begin == true && $pozition_begin < $pozition_end)
{
$content_parts[] = array(substr($content_page, 0, $pozition_begin), $noindex);
$noindex = 1;
$pozition_default = $pozition_begin+9;
$content_page = substr($content_page, $pozition_default);
$n_out++;
}
elseif (($pozition_end == true && $pozition_begin == true && $pozition_end < $pozition_begin) || ($pozition_end == true && $pozition_begin == false))
{
$content_parts[] = array(substr($content_page, 0, $pozition_end), $noindex);
if (--$n_out == 0) $noindex = 0;
$pozition_default = $pozition_end+10;
$content_page = substr($content_page, $pozition_default);
}
elseif ($pozition_end == false && $pozition_begin == false)
{
$content_parts[] = array($content_page, $noindex);
}
}
}
else
$content_parts[] = array($content_page);
$url_real_edit = preg_replace("!http://|https://|www\.!Ui", "", $page);
if (strpos($url_real_edit, "/") == true) $url_real_edit = substr($url_real_edit, 0, strpos($url_real_edit, "/"));
for ($i=0; $i<count($content_parts); $i++)
{
preg_match_all("!<a[^>]+href\s*=\s*[\"']?([^\"']+)[\"']?[^>]*>(.+?)</a>!i", $content_parts[$i][0], $links);
for ($r=0; $r<count($links[0]); $r++)
{
if (preg_match("!^(?:http://|https://)(?:www\.|)(.*)$!i", $links[1][$r], $url_parts))
{
if (preg_match('!^' . $url_real_edit . '!i', $url_parts[1])) continue;
// Îïðåäåëåÿåì ññûëêó è àíêîð
$link = trim($url_parts[0]);
$ankor = $links[2][$r];
// Åñëè êàðòèíêà
if (preg_match ("!<img[^>]+src=[\"']?([^\"']+)[\"']?[^>]*>!i", $links[2][$r], $image_parts))
$ankor = $image_parts[1];
// Äîáàâëÿåì ññûëêó â îñíîâíîé ìàññèâ èñõîäÿùèõ ññûëîê
$result['out'][] = Array($link, $ankor);
// Äîáàâëÿåì ññûëêó â ìàññèâ nofollow
if (preg_match("![\"'\s=]nofollow[\"'\s>]!Ui", $links[0][$r]))
$result['nofollow'][] = Array($link, $ankor);
// Äîáàâëÿåì ññûëêó â ìàññèâ noindex
if ($content_parts[$i][1] == 1)
$result['noindex'][] = Array($link, $ankor);
}
}
}
return $result;
}
if ($_POST['submit'] && $_POST['url_path'] != "")
print_r (check_outlinks($_POST['url_path']));
|