public function parse($string)
{
$parsed = array();
$parsedClean = array();
// first find all <a> tags
$parsed = $this->subparser->parse($string);
//X_Debug::i("Atoms found: ".count($parsed));
foreach ($parsed as $found) {
try {
$hoster = $this->helper->findHoster($found['href']);
$parsedClean[] = array('hoster' => $hoster, 'url' => $found['href'], 'label' => strip_tags($found['label']));
} catch (Exception $e) {
/* invalid href, ignored */
/*X_Debug::i("No hoster for link: {$found['href']}");*/
}
}
return $parsedClean;
}
public function parse($string)
{
$parsed = array();
$parsedClean = array();
// first find all <a> tags
//@file_put_contents(sys_get_temp_dir()."/last-http-request.html", $string);
$parsed = $this->subparser->parse($string);
//X_Debug::i("Atoms found: ".count($parsed));
foreach ($parsed as $found) {
try {
$hoster = $this->helper->findHoster($found['href']);
$parsedClean[] = array('hoster' => $hoster, 'url' => $found['href'], 'label' => strip_tags($found['label']));
//X_Debug::i("Valid link: {$found['href']}");
} catch (Exception $e) {
/* invalid href, ignored */
/*
if ( X_Env::startWith($found['href'], 'http://') )
X_Debug::i("No hoster for link: {$found['href']}");
*/
}
}
return $parsedClean;
}
/**
* (non-PHPdoc)
* @see X_PageParser_Parser::parse()
*/
public function parse($string)
{
switch ($this->mode) {
case self::MODE_TITLES:
return $this->parseTitles($string);
case self::MODE_EPISODES:
return $this->parseEpisodes($string);
case self::MODE_LINKS:
return $this->parseLinks($string);
case self::MODE_NEXTPAGE:
return $this->parseNextPage($string);
default:
throw new Exception("Invalid mode: {" . $this->mode . "}");
}
// dead code
// first, find the thumb url:
$match = array();
$thumb = false;
preg_match(self::PATTERN_THUMBNAIL, $string, $match);
// avoid relative thumbnails
if (count($match) > 0 && X_Env::startWith($match['image'], 'http://')) {
$thumb = $match['image'];
X_Debug::i("Thumbnail found: {$thumb}");
}
/* @var $hosterHelper X_VlcShares_Plugins_Helper_Hoster */
$hosterHelper = X_VlcShares_Plugins::helpers()->hoster();
$subparser = X_PageParser_Parser_HosterLinks::factory($hosterHelper, X_PageParser_Parser_Preg::factory(self::PATTERN_LINKS, X_PageParser_Parser_Preg::PREG_MATCH_ALL, PREG_SET_ORDER));
$links = $subparser->parse($string);
X_Debug::i("Valid hosters link found: " . count($links));
// clean results and reformat them
$cleanedLinks = array();
foreach ($links as $link) {
$cLink = array();
$cLink['hosterId'] = $link['hoster']->getId();
$cLink['videoId'] = $link['hoster']->getResourceId($link['url']);
$cLink['label'] = strip_tags($link['label']);
$cLink['link'] = "{$cLink['hosterId']}:{$cLink['videoId']}";
$cLink['thumbnail'] = $thumb;
$cleanedLinks[] = $cLink;
}
return $cleanedLinks;
}
private function _fetchGroupUpdates(X_Page_ItemList_PItem $items, $resourceType, $pageN = 1)
{
$page = X_PageParser_Page::getPage(self::URL_UPDATES_INDEX, X_PageParser_Parser_Preg::factory('/<p align="center"><a href="http:\\/\\/www\\.streaming-online\\.biz\\/(?P<id>[^\\/]+)\\/">(?P<label>.+?)<\\/a><\\/p>/', X_PageParser_Parser_Preg::PREG_MATCH_ALL, PREG_SET_ORDER));
$this->preparePageLoader($page);
$parsed = $page->getParsed();
// $parsed format = array(array('image', 'category', 'id', 'label'),..)
foreach ($parsed as $pItem) {
//$group = $pItem['category'];
$group = 'telefilm';
$label = trim($pItem['label']);
$id = $pItem['id'];
X_Debug::i("Parsed item: " . var_export(array($id, $group, $label), true));
$item = new X_Page_Item_PItem($this->getId() . "-{$resourceType}-{$group}-{$id}", $label);
$item->setIcon('/images/icons/folder_32.png')->setType(X_Page_Item_PItem::TYPE_CONTAINER)->setCustom(__CLASS__ . ':location', "{$resourceType}/{$pageN}/{$group}/{$id}")->setDescription(APPLICATION_ENV == 'development' ? "{$resourceType}/{$pageN}/{$group}/{$id}" : null)->setLink(array('l' => X_Env::encode("{$resourceType}/{$pageN}/{$group}/{$id}")), 'default', false);
$items->append($item);
}
}
/**
* (non-PHPdoc)
* @see X_PageParser_Parser::parse()
*/
public function parse($string)
{
// first, find the thumb url:
$match = array();
$thumb = false;
/*
preg_match(self::PATTERN_THUMBNAIL, $string, $match);
// avoid relative thumbnails
if ( count($match) > 0 && X_Env::startWith($match['image'], 'http://') ) {
$thumb = $match['image'];
X_Debug::i("Thumbnail found: {$thumb}");
}
*/
/* @var $hosterHelper X_VlcShares_Plugins_Helper_Hoster */
$hosterHelper = X_VlcShares_Plugins::helpers()->hoster();
/*
$subparser = X_PageParser_Parser_HosterLinks::factory($hosterHelper,
X_PageParser_Parser_Preg::factory(self::PATTERN_LINKS, X_PageParser_Parser_Preg::PREG_MATCH_ALL, PREG_SET_ORDER)
);
$links = $subparser->parse($string);
X_Debug::i("Valid hosters link found: ".count($links));
*/
$subparser = X_PageParser_Parser_Preg::factory(self::PATTERN_LINKS, X_PageParser_Parser_Preg::PREG_MATCH_ALL, PREG_SET_ORDER);
$links = $subparser->parse($string);
X_Debug::i("Step1 links: " . count($links));
// clean results and reformat them
$cleanedLinks = array();
$pre = false;
$startLine = false;
foreach ($links as $link) {
$cLink = array();
// first check if valid
try {
$hoster = $hosterHelper->findHoster($link['href']);
// link is valid
if ($link['startline']) {
if ($link['pre']) {
$pre = $link['pre'];
} else {
$pre = false;
}
}
$cLink['hosterId'] = $hoster->getId();
$cLink['videoId'] = $hoster->getResourceId($link['href']);
$cLink['label'] = $pre . " " . trim(strip_tags($link['label']));
$cLink['link'] = "{$cLink['hosterId']}:{$cLink['videoId']}";
//$cLink['thumbnail'] = $thumb;
$cleanedLinks[] = $cLink;
} catch (Exception $e) {
// the link is invalid
// so if it's a startline, the label has to be added to the eventual "pre" (if any).
if ($link['startline']) {
if ($link['pre']) {
$pre = $link['pre'];
} else {
$pre = $link['label'];
}
}
}
}
X_Debug::i("Step2 links: " . count($cleanedLinks));
return $cleanedLinks;
}
