function crawl_one_page($url)
{
$nims = array();
$kraw = new Crawler($url);
$kraw->go2linewhere('------------------------------------------');
$kraw->go2linewhere('------------------------------------------');
$kraw->readline();
while ($kraw->strpos('------------------------------------------') === false) {
$nims[] = $kraw->getbetween(' ', ' ');
$kraw->readline();
}
$kraw->close();
return $nims;
}
function crawl_1_page($start_url)
{
global $prefix;
global $bas;
$cr = new Crawler($start_url);
/*
echo $cr->readline();flush();
while(!feof($cr->stream)) {
}
exit(0);
*/
$cr->go2linewhere('headerSelect');
$cr->readline();
$line = $cr->curline;
$cr->close();
$chap = Crawler::cutfromlast1($start_url, '/');
if (strpos($chap, '.') === false) {
$chap = Crawler::n($chap, 3);
} else {
$a = explode('.', $chap);
$a[0] = Crawler::n($a[0], 3);
$chap = implode('.', $a);
}
$pi = 1;
// page i
$ledak = explode('value="', $line);
$pages = array();
for ($i = 1; $i < count($ledak); $i++) {
$uurl = Crawler::cutuntil($ledak[$i], '"');
$key = Crawler::cutfromlast1($uurl, '/');
$pages[$key] = strpos($uurl, 'http://') === 0 ? $uurl : $bas . $uurl;
}
//print_r($pages);flush();
$results = array();
foreach ($pages as $pagenum => $new_url) {
$berhasil = false;
while (!$berhasil) {
$cr = new Crawler($new_url);
//echo "URL:$new_url<br/>\n";flush();
$cr->go2linewhere('id="readerPage"');
if ($cr->strpos('<img src="') === false) {
$cr->readline();
}
$line = $cr->curline;
$cr->close();
$img_url = Crawler::extract($line, '<img src="', '"');
//echo "IMG:$img_url<br/>\n";flush();
$real_name = basename($img_url);
$ext = Crawler::cutfromlast($img_url, '.');
//$filename = $prefix . '-' . $chap . '-' . Crawler::n($pagenum, 2) . $ext;
$filename = $prefix . '-' . $chap . '-' . urldecode($real_name);
$val = $img_url;
if (!empty($val)) {
$berhasil = true;
$key = $filename;
$results[$filename] = $img_url;
?>
<a href="<?php
echo $val;
?>
"><?php
echo $key;
?>
</a><br />
<?php
flush();
} else {
}
}
}
}
<?php
require_once "crawler.php";
$base = 'http://www.ez-wallpaper.org';
$berhenti = 0;
$url = $base;
while (!$berhenti) {
echo "\nURL:{$url}\n";
$c = new Crawler($url);
$c->readline();
while ($line = $c->readline()) {
if ($c->strpos('nodeTitle') !== false) {
$href = $c->getbetween('<a href="', '"');
$c2 = new Crawler($base . $href);
$c2->go2linewhere('pageTitle');
$title = $c2->getbetween('>', '<');
$c2->go2linewhere('node_images');
$ledak = explode('<a href="', $c2->curline);
for ($i = 1; $i < count($ledak); $i++) {
$ahref = substr($ledak[$i], 0, strpos($ledak[$i], '"'));
echo "<a href='{$ahref}'>{$title}</a><br />\n";
}
//echo $c2->curline;
$c2->close();
} else {
if ($c->strpos('Go to next page') !== false) {
echo "\nADA NEXT\n";
$url = $base . $c->getbetweenlast('</span><a href="', '"');
break;
} else {
if ($c->strpos('Go to previous page') !== false) {
require_once "crawler.php";
$start_date = '2009-03-10';
$base_url = 'http://www.dilbert.com';
$middle_url = '/strips/comic/';
extract($_GET);
$selesai = false;
$url = $base_url . $middle_url . $start_date;
while (!$selesai) {
$ada_next = false;
$c = new Crawler($url);
echo "URL is {$url}<br />\n";
flush();
$c->go2lineor(array('STR_Content', 'STR_Prev'));
//echo "go2lineor selesai\n";flush();
if ($c->strpos('STR_Prev') !== false) {
//masih ada next
$ada_next = true;
$url = $base_url . $c->getbetween('<a href="', '"');
$c->go2linewhere('STR_Content');
} else {
$ada_next = false;
$selesai = true;
}
$c->readline();
$img = $c->getbetween('<img src="', '"');
echo "<a href='{$base_url}{$img}'>{$start_date}</a><br />\n";
$start_date = Crawler::extract($url, 'comic/', '/');
$c->close();
echo "Closed\n";
flush();
