Refonte du code

This commit is contained in:
Yves ASTIER 2013-08-11 13:30:41 +02:00
parent 3c929e0aa6
commit 927b04dfef
20 changed files with 1137 additions and 475 deletions

View file

@ -0,0 +1,35 @@
<?php
/**
* RssBridgeFlickrExplore
* Returns the newest interesting images from http://www.flickr.com/explore
*
* @name Flickr Explore
* @description Returns the latest interesting images from Flickr
*/
class FlickrExploreBridge extends BridgeAbstract{
public function collectData(array $param){
$html = file_get_html('http://www.flickr.com/explore') or $this->returnError('Could not request Flickr.', 404);
foreach($html->find('span.photo_container') as $element) {
$item = new \Item();
$item->uri = 'http://flickr.com'.$element->find('a',0)->href;
$item->thumbnailUri = $element->find('img',0)->getAttribute('data-defer-src');
$item->content = '<a href="' . $item->uri . '"><img src="' . $item->thumbnailUri . '" /></a>'; // FIXME: Filter javascript ?
$item->title = $element->find('a',0)->title;
$this->items[] = $item;
}
}
public function getName(){
return 'Flickr Explore';
}
public function getURI(){
return 'http://www.flickr.com/explore';
}
public function getCacheDuration(){
return 21600; // 6 hours
}
}

View file

@ -0,0 +1,51 @@
<?php
/**
* RssBridgeGoogleMostRecent
* Search Google for most recent pages regarding a specific topic.
* Returns the 100 most recent links in results in past year, sorting by date (most recent first).
* Example:
* http://www.google.com/search?q=sebsauvage&num=100&complete=0&tbs=qdr:y,sbd:1
* complete=0&num=100 : get 100 results
* qdr:y : in past year
* sbd:1 : sort by date (will only work if qdr: is specified)
*
* @name Google search
* @description Returns most recent results from Google search.
* @use1(q="keyword search")
*/
class GoogleSearchBridge extends BridgeAbstract{
public function collectData(array $param){
$html = '';
if (isset($param['q'])) { /* keyword search mode */
$html = file_get_html('http://www.google.com/search?q=' . urlencode($param['q']) . '&num=100&complete=0&tbs=qdr:y,sbd:1') or $this->returnError('No results for this query.', 404);
}
else{
$this->returnError('You must specify a keyword (?q=...).', 400);
}
$emIsRes = $html->find('div[id=ires]',0);
if( !is_null($emIsRes) ){
foreach($emIsRes->find('li[class=g]') as $element) {
$item = new \Item();
$item->uri = $element->find('a[href]',0)->href;
$item->title = $element->find('h3',0)->plaintext;
$item->content = $element->find('span[class=st]',0)->plaintext;
$this->items[] = $item;
}
}
}
public function getName(){
return 'Google search';
}
public function getURI(){
return 'http://google.com';
}
public function getCacheDuration(){
return 1800; // 30 minutes
}
}

50
bridges/TwitterBridge.php Normal file
View file

@ -0,0 +1,50 @@
<?php
/**
* RssBridgeTwitter
* Based on https://github.com/mitsukarenai/twitterbridge-noapi
*
* @name Twitter Bridge
* @description Returns user timelines or keyword search from http://twitter.com without using their API.
* @use1(q="keyword search")
* @use2(u="user timeline mode")
*/
class TwitterBridge extends BridgeAbstract{
public function collectData(array $param){
$html = '';
if (isset($param['q'])) { /* keyword search mode */
$html = file_get_html('http://twitter.com/search/realtime?q='.urlencode($param['q']).'+include:retweets&src=typd') or $this->returnError('No results for this query.', 404);
}
elseif (isset($param['u'])) { /* user timeline mode */
$html = file_get_html('http://twitter.com/'.urlencode($param['u'])) or $this->returnError('Requested username can\'t be found.', 404);
}
else {
$this->returnError('You must specify a keyword (?q=...) or a Twitter username (?u=...).', 400);
}
foreach($html->find('div.tweet') as $tweet) {
$item = new \Item();
$item->username = trim(substr($tweet->find('span.username', 0)->plaintext, 1)); // extract username and sanitize
$item->fullname = $tweet->getAttribute('data-name'); // extract fullname (pseudonym)
$item->avatar = $tweet->find('img', 0)->src; // get avatar link
$item->id = $tweet->getAttribute('data-tweet-id'); // get TweetID
$item->uri = 'https://twitter.com'.$tweet->find('a.details', 0)->getAttribute('href'); // get tweet link
$item->timestamp = $tweet->find('span._timestamp', 0)->getAttribute('data-time'); // extract tweet timestamp
$item->content = str_replace('href="/', 'href="https://twitter.com/', strip_tags($tweet->find('p.tweet-text', 0)->innertext, '<a>')); // extract tweet text
$item->title = $item->fullname . ' (@'. $item->username . ') | ' . $item->content;
$this->items[] = $item;
}
}
public function getName(){
return 'Twitter Bridge';
}
public function getURI(){
return 'http://twitter.com';
}
public function getCacheDuration(){
return 300; // 5 minutes
}
}

View file

@ -1,29 +0,0 @@
<?php
/**
* RssBridgeFlickrExplore
* Returns the newest interesting images from http://www.flickr.com/explore
*
* @name Flickr Explore
* @description Returns the latest interesting images from Flickr
*/
class RssBridgeFlickrExplore extends RssBridgeAbstractClass
{
protected $bridgeName = 'Flickr Explore';
protected $bridgeURI = 'http://www.flickr.com/explore';
protected $bridgeDescription = 'Returns the latest interesting images from Flickr';
protected $cacheDuration = 360; // 6 hours. No need to get more.
protected function collectData($request) {
$html = file_get_html('http://www.flickr.com/explore') or $this->returnError(404, 'could not request Flickr.');
$this->items = Array();
foreach($html->find('span.photo_container') as $element) {
$item['uri'] = 'http://flickr.com'.$element->find('a',0)->href;
$item['thumbnailUri'] = $element->find('img',0)->getAttribute('data-defer-src');
$item['content'] = '<a href="'.$item['uri'].'"><img src="'.$item['thumbnailUri'].'" /></a>'; // FIXME: Filter javascript ?
$item['title'] = $element->find('a',0)->title;
$this->items[] = $item;
}
}
}
$bridge = new RssBridgeFlickrExplore();
$bridge->process();

View file

@ -1,41 +0,0 @@
<?php
/**
* RssBridgeGoogleMostRecent
* Search Google for most recent pages regarding a specific topic.
* Returns the 100 most recent links in results in past year, sorting by date (most recent first).
* Example:
* http://www.google.com/search?q=sebsauvage&num=100&complete=0&tbs=qdr:y,sbd:1
* complete=0&num=100 : get 100 results
* qdr:y : in past year
* sbd:1 : sort by date (will only work if qdr: is specified)
*
* @name Google search
* @description Returns most recent results from Google search.
* @use1(q="keyword search")
*/
class RssBridgeGoogleSearch extends RssBridgeAbstractClass
{
protected $bridgeName = 'Google search';
protected $bridgeURI = 'http://google.com';
protected $bridgeDescription = 'Returns most recent results from Google search.';
protected $cacheDuration = 30; // 30 minutes, otherwise you could get banned by Google, or stumblr upon their captcha.
protected function collectData($request) {
$html = '';
if (isset($request['q'])) { /* keyword search mode */
$html = file_get_html('http://www.google.com/search?q='.urlencode($request['q']).'&num=100&complete=0&tbs=qdr:y,sbd:1') or $this->returnError(404, 'no results for this query.');
} else {
$this->returnError(400, 'You must specify a keyword (?q=...).');
}
$this->items = Array();
foreach($html->find('div[id=ires]',0)->find('li[class=g]') as $element) {
$item['uri'] = $element->find('a[href]',0)->href;
$item['title'] = $element->find('h3',0)->plaintext;
$item['content'] = $element->find('span[class=st]',0)->plaintext;
$this->items[] = $item;
}
}
}
$bridge = new RssBridgeGoogleSearch();
$bridge->process();

View file

@ -1,42 +0,0 @@
<?php
/**
* RssBridgeTwitter
* Based on https://github.com/mitsukarenai/twitterbridge-noapi
*
* @name Twitter Bridge
* @description Returns user timelines or keyword search from http://twitter.com without using their API.
* @use1(q="keyword search")
* @use2(u="user timeline mode")
*/
class RssBridgeTwitter extends RssBridgeAbstractClass
{
protected $bridgeName = 'Twitter Bridge';
protected $bridgeURI = 'http://twitter.com';
protected $bridgeDescription = 'Returns user timelines or keyword search from http://twitter.com without using their API.';
protected $cacheDuration = 5; // 5 minutes
protected function collectData($request) {
$html = '';
if (isset($request['q'])) { /* keyword search mode */
$html = file_get_html('http://twitter.com/search/realtime?q='.urlencode($request['q']).'+include:retweets&src=typd') or $this->returnError(404, 'no results for this query.');
} elseif (isset($request['u'])) { /* user timeline mode */
$html = file_get_html('http://twitter.com/'.urlencode($request['u'])) or $this->returnError(404, 'requested username can\'t be found.');
} else {
$this->returnError(400, 'You must specify a keyword (?q=...) or a Twitter username (?u=...).');
}
$this->items = Array();
foreach($html->find('div.tweet') as $tweet) {
$item['username'] = trim(substr($tweet->find('span.username', 0)->plaintext, 1)); // extract username and sanitize
$item['fullname'] = $tweet->getAttribute('data-name'); // extract fullname (pseudonym)
$item['avatar'] = $tweet->find('img', 0)->src; // get avatar link
$item['id'] = $tweet->getAttribute('data-tweet-id'); // get TweetID
$item['uri'] = 'https://twitter.com'.$tweet->find('a.details', 0)->getAttribute('href'); // get tweet link
$item['timestamp'] = $tweet->find('span._timestamp', 0)->getAttribute('data-time'); // extract tweet timestamp
$item['content'] = str_replace('href="/', 'href="https://twitter.com/', strip_tags($tweet->find('p.tweet-text', 0)->innertext, '<a>')); // extract tweet text
$item['title'] = $item['fullname'] . ' (@'.$item['username'] . ') | ' . $item['content'];
$this->items[] = $item;
}
}
}
$bridge = new RssBridgeTwitter();
$bridge->process();