[IsoHuntBridge] code simplification

Signed-off-by: Pierre Mazière <pierre.maziere@gmx.com>
This commit is contained in:
Pierre Mazière 2016-08-29 02:38:54 +02:00
parent b5c432d66c
commit f13f44a682

View file

@ -1,483 +1,471 @@
<?php <?php
class IsoHuntBridge extends BridgeAbstract{ class IsoHuntBridge extends BridgeAbstract{
public $maintainer = 'logmanoriginal'; public $maintainer = 'logmanoriginal';
public $name = 'isoHunt Bridge'; // Is replaced later! public $name = 'isoHunt Bridge';
public $uri = 'https://isohunt.to'; // Is replaced later! public $uri = 'https://isohunt.to/';
public $description = 'Returns the latest results by category or search result'; public $description = 'Returns the latest results by category or search result';
public $parameters = array( public $parameters = array(
/* /*
* Get feeds for one of the "latest" categories * Get feeds for one of the "latest" categories
* Notice: The categories "News" and "Top Searches" are received from the main page * Notice: The categories "News" and "Top Searches" are received from the main page
* Elements are sorted by name ascending! * Elements are sorted by name ascending!
*/ */
'By "Latest" category' => array( 'By "Latest" category' => array(
'latest_category'=>array( 'latest_category'=>array(
'name'=>'Latest category', 'name'=>'Latest category',
'type'=>'list', 'type'=>'list',
'required'=>true, 'required'=>true,
'title'=>'Select your category', 'title'=>'Select your category',
'defaultValue'=>'News', 'defaultValue'=>'news',
'values'=>array( 'values'=>array(
'Hot Torrents'=>'hot_torrents', 'Hot Torrents'=>'hot_torrents',
'News'=>'news', 'News'=>'news',
'Releases'=>'releases', 'Releases'=>'releases',
'Torrents'=>'torrents' 'Torrents'=>'torrents'
)
)
),
/*
* Get feeds for one of the "torrent" categories
* Make sure to add new categories also to get_torrent_category_index($)!
* Elements are sorted by name ascending!
*/
'By "Torrent" category' => array(
'torrent_category'=>array(
'name'=>'Torrent category',
'type'=>'list',
'required'=>true,
'title'=>'Select your category',
'defaultValue'=>'Anime',
'values'=>array(
'Adult'=>'adult',
'Anime'=>'anime',
'Books'=>'books',
'Games'=>'games',
'Movies'=>'movies',
'Music'=>'music',
'Other'=>'other',
'Series & TV'=>'series_tv',
'Software'=>'software'
)
),
'torrent_popularity'=>array(
'name'=>'Sort by popularity',
'type'=>'checkbox',
'title'=>'Activate to receive results by popularity'
)
),
/*
* Get feeds for a specific search request
*/
'Search torrent by name' => array(
'search_name'=>array(
'name'=>'Name',
'required'=>true,
'title'=>'Insert your search query',
'exampleValue'=>'Bridge'
),
'search_category'=>array(
'name'=>'Category',
'type'=>'list',
'title'=>'Select your category',
'defaultValue'=>'All',
'values'=>array(
'Adult'=>'adult',
'All'=>'all',
'Anime'=>'anime',
'Books'=>'books',
'Games'=>'games',
'Movies'=>'movies',
'Music'=>'music',
'Other'=>'other',
'Series & TV'=>'series_tv',
'Software'=>'software'
)
)
) )
); )
),
public function collectData(){ /*
$request_path = '/'; // We'll request the main page by default * Get feeds for one of the "torrent" categories
* Make sure to add new categories also to get_torrent_category_index($)!
* Elements are sorted by name ascending!
*/
'By "Torrent" category' => array(
'torrent_category'=>array(
'name'=>'Torrent category',
'type'=>'list',
'required'=>true,
'title'=>'Select your category',
'defaultValue'=>'anime',
'values'=>array(
'Adult'=>'adult',
'Anime'=>'anime',
'Books'=>'books',
'Games'=>'games',
'Movies'=>'movies',
'Music'=>'music',
'Other'=>'other',
'Series & TV'=>'series_tv',
'Software'=>'software'
)
),
'torrent_popularity'=>array(
'name'=>'Sort by popularity',
'type'=>'checkbox',
'title'=>'Activate to receive results by popularity'
)
),
if($this->getInput('latest_category')){ // Requesting one of the latest categories /*
$this->request_latest_category($this->getInput('latest_category')); * Get feeds for a specific search request
} elseif($this->getInput('torrent_category')){ // Requesting one of the torrent categories */
$order_popularity = false; 'Search torrent by name' => array(
'search_name'=>array(
'name'=>'Name',
'required'=>true,
'title'=>'Insert your search query',
'exampleValue'=>'Bridge'
),
'search_category'=>array(
'name'=>'Category',
'type'=>'list',
'title'=>'Select your category',
'defaultValue'=>'all',
'values'=>array(
'Adult'=>'adult',
'All'=>'all',
'Anime'=>'anime',
'Books'=>'books',
'Games'=>'games',
'Movies'=>'movies',
'Music'=>'music',
'Other'=>'other',
'Series & TV'=>'series_tv',
'Software'=>'software'
)
)
)
);
if($this->getInput('torrent_popularity')) public function getURI(){
$order_popularity = $this->getInput('torrent_popularity'); $uri=$this->uri;
switch($this->queriedContext){
case 'By "Latest" category':
switch($this->getInput('latest_category')){
case 'hot_torrents':
$uri .= 'statistic/hot/torrents';
break;
case 'news':
break;
case 'releases':
$uri .= 'releases.php';
break;
case 'torrents':
$uri .= 'latest.php';
break;
}
break;
$this->request_torrent_category($this->getInput('torrent_category'), $order_popularity); case 'By "Torrent" category':
} else if($this->getInput('search_name')){ // Requesting search $uri .= $this->build_category_uri(
if($this->getInput('search_category')) $this->getInput('torrent_category'),
$this->request_search($this->getInput('search_name'), $this->getInput('search_category')); $this->getInput('torrent_popularity')
else );
$this->request_search($this->getInput('search_name')); break;
} else {
$this->returnClientError('Unknown request!'); case 'Search torrent by name':
} $category=$this->getInput('search_category');
$uri .= $this->build_category_uri($category);
if($category!=='movies')
$uri .= '&ihq=' . urlencode($this->getInput('search_name'));
break;
} }
public function getCacheDuration(){ return $uri;
return 300; // 5 minutes }
public function getName(){
switch($this->queriedContext){
case 'By "Latest" category':
$categoryName =
array_search(
$this->getInput('latest_category'),
$this->parameters['By "Latest" category']['latest_category']['values']
);
$name = 'Latest '.$categoryName.' - ' . $this->name;
break;
case 'By "Torrent" category':
$categoryName =
array_search(
$this->getInput('torrent_category'),
$this->parameters['By "Torrent" category']['torrent_category']['values']
);
$name = 'Category: ' . $categoryName . ' - ' . $this->name;
break;
case 'Search torrent by name':
$categoryName =
array_search(
$this->getInput('search_category'),
$this->parameters['Search torrent by name']['search_category']['values']
);
$name = 'Search: "' . $this->getInput('search_name') . '" in category: ' . $categoryName . ' - ' . $this->name;
break;
} }
#region Helper functions for "By "Torrent" category" return $name;
}
private function request_torrent_category($category, $order_popularity){
$category_name = $this->get_torrent_category_name($category);
$category_index = $this->get_torrent_category_index($category);
$this->name = 'Category: ' . $category_name . ' - ' . $this->name; public function collectData(){
$this->uri .= $this->build_category_uri($category_index, $order_popularity); $html = $this->load_html($this->getURI());
$html = $this->load_html($this->uri); switch($this->queriedContext){
case 'By "Latest" category':
if(strtolower(trim($category)) === 'movies') // This one is special (content wise) switch($this->getInput('latest_category')){
$this->get_movie_torrents($html); case 'hot_torrents':
else $this->get_latest_hot_torrents($html);
$this->get_latest_torrents($html); break;
} case 'news':
$this->get_latest_news($html);
private function get_torrent_category_name($category){ break;
$parameter = $this->parameters['By "Torrent" category']; case 'releases':
$languages = $parameter['torrent_category']['values']; case 'torrents':
foreach($languages as $name=>$value)
if(strtolower(trim($value)) === strtolower(trim($category)))
return $name;
return 'Unknown category';
}
private function get_torrent_category_index($category){
switch(strtolower(trim($category))){
case 'anime': return 1;
case 'software' : return 2;
case 'games' : return 3;
case 'adult' : return 4;
case 'movies' : return 5;
case 'music' : return 6;
case 'other' : return 7;
case 'series_tv' : return 8;
case 'books': return 9;
default: return 0;
}
}
#endregion
private function request_latest_category($category){
switch($category){
case 'hot_torrents': // This is a special case! (that's why return)
$this->name = 'Latest hot torrents - ' . $this->name;
$this->uri .= '/statistic/hot/torrents';
$html = $this->load_html($this->uri);
$this->get_latest_hot_torrents($html);
return;
case 'news': // This is a special case! (that's why return)
$this->name = 'Latest news - ' . $this->name;
$this->uri .= '/';
$html = $this->load_html($this->uri);
$this->get_latest_news($html);
return;
case 'releases':
$this->name = 'Latest releases - ' . $this->name;
$this->uri .= '/releases.php';
break;
case 'torrents':
$this->name = 'Latest torrents - ' . $this->name;
$this->uri .= '/latest.php';
break;
default: // No category applies
$this->returnClientError('Undefined category: ' . $category . '!');
}
$html = $this->load_html($this->uri);
$this->get_latest_torrents($html); $this->get_latest_torrents($html);
break;
}
break;
case 'By "Torrent" category':
if($this->getInput('torrent_category') === 'movies'){
// This one is special (content wise)
$this->get_movie_torrents($html);
}else{
$this->get_latest_torrents($html);
}
break;
case 'Search torrent by name':
if( $this->getInput('search_category') === 'movies'){
// This one is special (content wise)
$this->get_movie_torrents($html);
} else {
$this->get_latest_torrents($html);
}
break;
}
}
public function getCacheDuration(){
return 300; // 5 minutes
}
#region Helper functions for "Movie Torrents"
private function get_movie_torrents($html){
$container = $html->find('div#w0', 0);
if(!$container)
$this->returnServerError('Unable to find torrent container!');
$torrents = $container->find('article');
if(!$torrents)
$this->returnServerError('Unable to find torrents!');
foreach($torrents as $torrent){
$anchor = $torrent->find('a', 0);
if(!$anchor)
$this->returnServerError('Unable to find anchor!');
$date = $torrent->find('small', 0);
if(!$date)
$this->returnServerError('Unable to find date!');
$item = array();
$item['uri'] = $this->fix_relative_uri($anchor->href);
$item['title'] = $anchor->title;
// $item['author'] =
$item['timestamp'] = strtotime($date->plaintext);
$item['content'] = $this->fix_relative_uri($torrent->innertext);
$this->items[] = $item;
}
}
#endregion
#region Helper functions for "Latest Hot Torrents"
private function get_latest_hot_torrents($html){
$container = $html->find('div#serps', 0);
if(!$container)
$this->returnServerError('Unable to find torrent container!');
$torrents = $container->find('tr');
if(!$torrents)
$this->returnServerError('Unable to find torrents!');
// Remove first element (header row)
$torrents = array_slice($torrents, 1);
foreach($torrents as $torrent){
$cell = $torrent->find('td', 0);
if(!$cell)
$this->returnServerError('Unable to find cell!');
$element = $cell->find('a', 0);
if(!$element)
$this->returnServerError('Unable to find element!');
$item = array();
$item['uri'] = $element->href;
$item['title'] = $element->plaintext;
// $item['author'] =
// $item['timestamp'] =
// $item['content'] =
$this->items[] = $item;
}
}
#endregion
#region Helper functions for "Latest News"
private function get_latest_news($html){
$container = $html->find('div#postcontainer', 0);
if(!$container)
$this->returnServerError('Unable to find post container!');
$posts = $container->find('div.index-post');
if(!$posts)
$this->returnServerError('Unable to find posts!');
foreach($posts as $post){
$item = array();
$item['uri'] = $this->latest_news_extract_uri($post);
$item['title'] = $this->latest_news_extract_title($post);
$item['author'] = $this->latest_news_extract_author($post);
$item['timestamp'] = $this->latest_news_extract_timestamp($post);
$item['content'] = $this->latest_news_extract_content($post);
$this->items[] = $item;
}
}
private function latest_news_extract_author($post){
$author = $post->find('small', 0);
if(!$author)
$this->returnServerError('Unable to find author!');
// The author is hidden within a string like: 'Posted by {author} on {date}'
preg_match('/Posted\sby\s(.*)\son/i', $author->innertext, $matches);
return $matches[1];
}
private function latest_news_extract_timestamp($post){
$date = $post->find('small', 0);
if(!$date)
$this->returnServerError('Unable to find date!');
// The date is hidden within a string like: 'Posted by {author} on {date}'
preg_match('/Posted\sby\s.*\son\s(.*)/i', $date->innertext, $matches);
$timestamp = strtotime($matches[1]);
// Make sure date is not in the future (dates are given like 'Nov. 20' without year)
if($timestamp > time()){
$timestamp = strtotime('-1 year', $timestamp);
} }
#region Helper functions for "Search torrent by name" return $timestamp;
}
private function request_search($name, $category = 'all'){ private function latest_news_extract_title($post){
$category_name = $this->get_search_category_name($category); $title = $post->find('a', 0);
$category_index = $this->get_search_category_index($category); if(!$title)
$this->returnServerError('Unable to find title!');
$this->name = 'Search: "' . $name . '" in category: ' . $category_name . ' - ' . $this->name; return $title->plaintext;
$this->uri .= $this->build_category_uri($category_index); }
if(strtolower(trim($category)) === 'movies'){ // This one is special (content wise) private function latest_news_extract_uri($post){
$html = $this->load_html($this->uri); $uri = $post->find('a', 0);
$this->get_movie_torrents($html); if(!$uri)
} else { $this->returnServerError('Unable to find uri!');
$this->uri .= '&ihq=' . urlencode($name);
$html = $this->load_html($this->uri); return $uri->href;
$this->get_latest_torrents($html); }
}
private function latest_news_extract_content($post){
$content = $post->find('div', 0);
if(!$content)
$this->returnServerError('Unable to find content!');
// Remove <h2>...</h2> (title)
foreach($content->find('h2') as $element){
$element->outertext = '';
} }
private function get_search_category_name($category){ // Remove <small>...</small> (author)
$parameter = $this->parameters['Search torrent by name']; foreach($content->find('small') as $element){
$languages = $parameter['search_category']['values']; $element->outertext = '';
foreach($languages as $name=>$value)
if(strtolower(trim($value)) === strtolower(trim($category)))
return $name;
return 'Unknown category';
} }
private function get_search_category_index($category){ return $content->innertext;
switch(strtolower(trim($category))){ }
case 'all': return 0;
default: return $this->get_torrent_category_index($category); // Uses the same index #endregion
}
#region Helper functions for "Latest Torrents", "Latest Releases" and "Torrent Category"
private function get_latest_torrents($html){
$container = $html->find('div#serps', 0);
if(!$container)
$this->returnServerError('Unable to find torrent container!');
$torrents = $container->find('tr[data-key]');
if(!$torrents)
$this->returnServerError('Unable to find torrents!');
foreach($torrents as $torrent){
$item = array();
$item['uri'] = $this->latest_torrents_extract_uri($torrent);
$item['title'] = $this->latest_torrents_extract_title($torrent);
$item['author'] = $this->latest_torrents_extract_author($torrent);
$item['timestamp'] = $this->latest_torrents_extract_timestamp($torrent);
$item['content'] = ''; // There is no valuable content
$this->items[] = $item;
}
}
private function latest_torrents_extract_title($torrent){
$cell = $torrent->find('td.title-row', 0);
if(!$cell)
$this->returnServerError('Unable to find title cell!');
$title = $cell->find('span', 0);
if(!$title)
$this->returnServerError('Unable to find title!');
return $title->plaintext;
}
private function latest_torrents_extract_uri($torrent){
$cell = $torrent->find('td.title-row', 0);
if(!$cell)
$this->returnServerError('Unable to find title cell!');
$uri = $cell->find('a', 0);
if(!$uri)
$this->returnServerError('Unable to find uri!');
return $this->fix_relative_uri($uri->href);
}
private function latest_torrents_extract_author($torrent){
$cell = $torrent->find('td.user-row', 0);
if(!$cell)
return; // No author
$user = $cell->find('a', 0);
if(!$user)
$this->returnServerError('Unable to find user!');
return $user->plaintext;
}
private function latest_torrents_extract_timestamp($torrent){
$cell = $torrent->find('td.date-row', 0);
if(!$cell)
$this->returnServerError('Unable to find date cell!');
return strtotime('-' . $cell->plaintext, time());
}
#endregion
#region Generic helper functions
private function load_html($uri){
$html = $this->getSimpleHTMLDOM($uri);
if(!$html)
$this->returnServerError('Unable to load ' . $uri . '!');
return $html;
}
private function fix_relative_uri($uri){
return preg_replace('/\//i', $this->uri, $uri, 1);
}
private function build_category_uri($category, $order_popularity = false){
switch($category){
case 'anime': $index = 1; break;
case 'software' : $index = 2; break;
case 'games' : $index = 3; break;
case 'adult' : $index = 4; break;
case 'movies' : $index = 5; break;
case 'music' : $index = 6; break;
case 'other' : $index = 7; break;
case 'series_tv' : $index = 8; break;
case 'books': $index = 9; break;
case 'all':
default: $index = 0; break;
} }
#endregion return 'torrents/?iht=' . $index . '&ihs=' . ($order_popularity ? 1 : 0) . '&age=0';
}
#region Helper functions for "Movie Torrents" #endregion
private function get_movie_torrents($html){
$container = $html->find('div#w0', 0);
if(!$container)
$this->returnServerError('Unable to find torrent container!');
$torrents = $container->find('article');
if(!$torrents)
$this->returnServerError('Unable to find torrents!');
foreach($torrents as $torrent){
$anchor = $torrent->find('a', 0);
if(!$anchor)
$this->returnServerError('Unable to find anchor!');
$date = $torrent->find('small', 0);
if(!$date)
$this->returnServerError('Unable to find date!');
$item = array();
$item['uri'] = $this->fix_relative_uri($anchor->href);
$item['title'] = $anchor->title;
// $item['author'] =
$item['timestamp'] = strtotime($date->plaintext);
$item['content'] = $this->fix_relative_uri($torrent->innertext);
$this->items[] = $item;
}
}
#endregion
#region Helper functions for "Latest Hot Torrents"
private function get_latest_hot_torrents($html){
$container = $html->find('div#serps', 0);
if(!$container)
$this->returnServerError('Unable to find torrent container!');
$torrents = $container->find('tr');
if(!$torrents)
$this->returnServerError('Unable to find torrents!');
// Remove first element (header row)
$torrents = array_slice($torrents, 1);
foreach($torrents as $torrent){
$cell = $torrent->find('td', 0);
if(!$cell)
$this->returnServerError('Unable to find cell!');
$element = $cell->find('a', 0);
if(!$element)
$this->returnServerError('Unable to find element!');
$item = array();
$item['uri'] = $element->href;
$item['title'] = $element->plaintext;
// $item['author'] =
// $item['timestamp'] =
// $item['content'] =
$this->items[] = $item;
}
}
#endregion
#region Helper functions for "Latest News"
private function get_latest_news($html){
$container = $html->find('div#postcontainer', 0);
if(!$container)
$this->returnServerError('Unable to find post container!');
$posts = $container->find('div.index-post');
if(!$posts)
$this->returnServerError('Unable to find posts!');
foreach($posts as $post){
$item = array();
$item['uri'] = $this->latest_news_extract_uri($post);
$item['title'] = $this->latest_news_extract_title($post);
$item['author'] = $this->latest_news_extract_author($post);
$item['timestamp'] = $this->latest_news_extract_timestamp($post);
$item['content'] = $this->latest_news_extract_content($post);
$this->items[] = $item;
}
}
private function latest_news_extract_author($post){
$author = $post->find('small', 0);
if(!$author)
$this->returnServerError('Unable to find author!');
// The author is hidden within a string like: 'Posted by {author} on {date}'
preg_match('/Posted\sby\s(.*)\son/i', $author->innertext, $matches);
return $matches[1];
}
private function latest_news_extract_timestamp($post){
$date = $post->find('small', 0);
if(!$date)
$this->returnServerError('Unable to find date!');
// The date is hidden within a string like: 'Posted by {author} on {date}'
preg_match('/Posted\sby\s.*\son\s(.*)/i', $date->innertext, $matches);
$timestamp = strtotime($matches[1]);
// Make sure date is not in the future (dates are given like 'Nov. 20' without year)
if($timestamp > time()){
$timestamp = strtotime('-1 year', $timestamp);
}
return $timestamp;
}
private function latest_news_extract_title($post){
$title = $post->find('a', 0);
if(!$title)
$this->returnServerError('Unable to find title!');
return $title->plaintext;
}
private function latest_news_extract_uri($post){
$uri = $post->find('a', 0);
if(!$uri)
$this->returnServerError('Unable to find uri!');
return $uri->href;
}
private function latest_news_extract_content($post){
$content = $post->find('div', 0);
if(!$content)
$this->returnServerError('Unable to find content!');
// Remove <h2>...</h2> (title)
foreach($content->find('h2') as $element){
$element->outertext = '';
}
// Remove <small>...</small> (author)
foreach($content->find('small') as $element){
$element->outertext = '';
}
return $content->innertext;
}
#endregion
#region Helper functions for "Latest Torrents", "Latest Releases" and "Torrent Category"
private function get_latest_torrents($html){
$container = $html->find('div#serps', 0);
if(!$container)
$this->returnServerError('Unable to find torrent container!');
$torrents = $container->find('tr[data-key]');
if(!$torrents)
$this->returnServerError('Unable to find torrents!');
foreach($torrents as $torrent){
$item = array();
$item['uri'] = $this->latest_torrents_extract_uri($torrent);
$item['title'] = $this->latest_torrents_extract_title($torrent);
$item['author'] = $this->latest_torrents_extract_author($torrent);
$item['timestamp'] = $this->latest_torrents_extract_timestamp($torrent);
$item['content'] = ''; // There is no valuable content
$this->items[] = $item;
}
}
private function latest_torrents_extract_title($torrent){
$cell = $torrent->find('td.title-row', 0);
if(!$cell)
$this->returnServerError('Unable to find title cell!');
$title = $cell->find('span', 0);
if(!$title)
$this->returnServerError('Unable to find title!');
return $title->plaintext;
}
private function latest_torrents_extract_uri($torrent){
$cell = $torrent->find('td.title-row', 0);
if(!$cell)
$this->returnServerError('Unable to find title cell!');
$uri = $cell->find('a', 0);
if(!$uri)
$this->returnServerError('Unable to find uri!');
return $this->fix_relative_uri($uri->href);
}
private function latest_torrents_extract_author($torrent){
$cell = $torrent->find('td.user-row', 0);
if(!$cell)
return; // No author
$user = $cell->find('a', 0);
if(!$user)
$this->returnServerError('Unable to find user!');
return $user->plaintext;
}
private function latest_torrents_extract_timestamp($torrent){
$cell = $torrent->find('td.date-row', 0);
if(!$cell)
$this->returnServerError('Unable to find date cell!');
return strtotime('-' . $cell->plaintext, time());
}
#endregion
#region Generic helper functions
private function load_html($uri){
$html = $this->getSimpleHTMLDOM($uri);
if(!$html)
$this->returnServerError('Unable to load ' . $uri . '!');
return $html;
}
private function fix_relative_uri($uri){
return preg_replace('/\//i', 'https://isohunt.to/', $uri, 1);
}
private function build_category_uri($index, $order_popularity = false){
return '/torrents/?iht=' . $index . '&ihs=' . ($order_popularity ? 1 : 0) . '&age=0';
}
#endregion
} }