private function collect_images($url, $dir)
{
if (strpos($url, '/idol.')) {
$base = 'https://idol.sankakucomplex.com';
} else {
$base = 'https://chan.sankakucomplex.com';
}
$page = $this->page_from;
$id = 1;
$Turl = Text::create($url);
do {
if ($page > $this->page_to) {
break;
}
$purl = $url . '&page=' . $page;
echo "{$purl}\n";
do {
$P = new Page($purl, array('become_firefox' => true));
$T = new Text($P->content());
sleep(3);
// 429 too many requests
} while ($T->contain('429 Too many requests'));
$a = $T->extract_to_array('href="', '"');
foreach ($a as $i => $e) {
$E = new Text($e);
if (!$E->contain('/post/show')) {
unset($a[$i]);
}
}
if (!count($a)) {
break;
}
foreach ($a as $i => $e) {
$E = new Text($e);
$kurl = $base . $e;
echo "{$kurl}\n";
flush();
do {
$P = new Page($kurl, array('become_firefox' => true));
$T = new Text($P->content());
sleep(3);
// 429 too many requests
} while ($T->contain('429 Too many requests'));
$P->go_line('id=highres');
$img = $P->curr_line()->cut_between('href="', '"');
/*if ($img->contain('.webm')) {
echo "This is WEBM\n";
} else*/
if ($img->to_s()) {
$this->download_if_not_exist($img, $dir, $id);
$id++;
} else {
echo "No id=highres\n";
}
}
$page++;
} while (true);
}
public function get_info($base)
{
$p = new Page($base);
$p->go_line('<div class="divThickBorder" style="padding:7px">');
$raw = $p->next_line()->dup();
$list = array();
foreach (explode('<tr>', $raw->to_s()) as $line) {
$tline = new Text($line);
if ($tline->contain('href="')) {
$href = $tline->dup()->cut_between('href="', '"')->to_s();
$desc = $tline->dup()->cut_between('">', '</a')->to_s();
preg_match('/([\\.\\d]+) :/', $desc, $m);
$infix = $m[1];
$list[] = array('url' => $href, 'desc' => strip_tags($desc), 'infix' => $infix);
}
}
return $list;
}
public function crawl_chapter($v)
{
$ifx = Text::create($v['infix'])->pad(3)->to_s();
$p = new Page($v['url']);
$p->go_line('var pages');
$json = $p->curr_line()->cut_between(' = ', ';');
$list = json_decode($json);
foreach ($list as $page) {
$purl = new Text($page->url);
$name = new Text($page->filename);
if ($name->strlen() < 15) {
$name = $this->prefix . '-' . $ifx . '-' . $name;
}
if ($purl->contain('resize_img.php')) {
$purl = $purl->cut_between('resize_img.php?url=', '&width');
}
echo "<a href='{$purl}'>{$name}</a><br/>\n";
}
}
function idol_sankaku2($base_url, $from, $to)
{
$base = 'http://idol.sankakucomplex.com';
for ($i = $from; $i <= $to; $i++) {
$P = new Page($base_url . '&page=' . $i);
$T = new Text($P->content());
$a = $T->extract_to_array('href="', '"');
foreach ($a as $e) {
$E = new Text($e);
if ($E->contain('/post/show')) {
$url = $base . $e;
$P = new Page($url);
$P->go_line('id="highres"');
$img = $P->curr_line()->cut_between('href="', '"')->to_s();
$P->reset_line();
$P->go_line('id="post_old_tags"');
$tag = $P->curr_line()->cut_between('value="', '"')->substr(0, 150)->to_s();
// max 100 karakter
echo "<a href='{$img}'>{$tag}</a><br />\n";
}
}
}
}
public function action_update_tags()
{
$hundred_mangas = Model::factory('Hmanga')->order_by_asc('id')->find_many();
foreach ($hundred_mangas as $m) {
echo "ID: {$m->id} URL is: {$m->url}<br>\n";
echo "Old tags: {$m->tags}<br>\n";
$real_url = self::$base . $m->url;
$p = new Page($real_url);
$content = new Text($p->content());
// hack: sometimes old urls gone
if ($content->contain('<title>Error Message</title>')) {
$m->delete();
echo "DELETED {$m->url} is gone<br>\n";
continue;
}
$tags = $this->grab_all_tags($p);
$tags = '#' . implode('#', $tags) . '#';
echo "New tags: {$tags}<br>\n";
if ($tags !== $m->tags) {
echo "Tags are different! Save!<br>\n";
$m->tags = $tags;
$m->save();
}
}
}
function deviantart($gal)
{
$next = true;
preg_match('/:\\/\\/([^\\/]+)\\//', $gal, $m);
$base = 'http://' . $m[1];
do {
echo "{$gal}<br>";
$p = new Page($gal);
$p->go_line('id="gmi-ResourceStream"');
$raw = explode('<a class="thumb', $p->curr_line()->to_s());
for ($i = 1, $l = count($raw); $i < $l; $i++) {
$line = new Text($raw[$i]);
$name = $line->dup()->cut_between('<b>', '</')->to_s();
if ($line->contain('super_fullimg=')) {
$src = $line->dup()->cut_between('super_fullimg="', '"')->to_s();
} else {
$src = $line->dup()->cut_between('super_img="', '"')->to_s();
}
echo "<a href='{$src}'>{$name}</a><br>\n";
}
// ada next?
$p->go_line('<li class="next">');
if ($p->curr_line()->contain('<li class="next"><a class="away" href="')) {
$part = $p->curr_line()->dup()->cut_between('<li class="next"><a class="away" href="', '"');
$gal = $base . $part;
} else {
$next = false;
}
} while ($next);
}
function sankakucomplex($url)
{
if (strpos($url, '/idol.')) {
$base = 'https://idol.sankakucomplex.com';
} else {
$base = 'https://chan.sankakucomplex.com';
}
$page = 1;
$tag = uniqid();
$Turl = Text::create($url);
if ($Turl->contain('tags=')) {
$tag = $Turl->cut_after('tags=')->urldecode()->to_s();
}
do {
if (isset($_GET['limit'])) {
if ($page > $_GET['limit']) {
break;
}
}
$purl = $url . '&page=' . $page;
echo "{$purl}<br>\n";
do {
$P = new Page($purl, array('become_firefox' => true));
$T = new Text($P->content());
sleep(3);
// 429 too many requests
} while ($T->contain('429 Too many requests'));
$a = $T->extract_to_array('href="', '"');
foreach ($a as $i => $e) {
$E = new Text($e);
if (!$E->contain('/post/show')) {
unset($a[$i]);
}
}
if (!count($a)) {
break;
}
foreach ($a as $i => $e) {
$E = new Text($e);
$kurl = $base . $e;
echo "{$kurl}<br>\n";
flush();
do {
$P = new Page($kurl, array('become_firefox' => true));
$T = new Text($P->content());
sleep(3);
// 429 too many requests
} while ($T->contain('429 Too many requests'));
// $P->go_line('id="highres"');
if (isset($_GET['hires'])) {
$P->go_line('id=highres');
} else {
$P->go_line('id=lowres');
}
if ($P->end_of_line()) {
$P->reset_line();
$P->go_line('id=highres');
}
$img = $P->curr_line()->cut_between('href="', '"')->to_s();
// $P->reset_line();
// $P->go_line('id="post_old_tags"');
// $tag = $P->curr_line()->cut_between('value="', '"')->substring(0, 150)->to_s(); // max 100 karakter
if ($img) {
echo "<a href='{$img}'>{$tag}</a><br />\n";
flush();
} else {
echo "This is flash<br />\n";
}
}
$page++;
} while (true);
}
public function action_fill_reference()
{
$p = new Page('http://www.hbrowse.com/advance');
$raw = new Text($p->content());
$names = $raw->extract_to_array('name="', '"');
$names = array_unique($names);
$ref = array();
foreach ($names as $raw_name) {
$name = new Text($raw_name);
if ($name->contain('_')) {
$key = $name->cut_before('_')->to_s();
$val = $name->cut_after('_')->to_s();
$ref[$key][] = $val;
}
}
// insert
foreach ($ref as $key => $val) {
$reference = ORM::for_table('reference')->create();
$reference->id = $key;
$reference->val = '#' . implode('#', $val) . '#';
$reference->save();
}
}
