boyska/rss-bridge
1
0
Fork 0
forked from blallo/rss-bridge
Code Releases Activity

Merge pull request #276 from ORelio/master

Browse source 
[JapanExpo] Allow fetching full content
This commit is contained in:
Mitsu 2016-06-25 13:27:29 +02:00 committed by GitHub
commit d27929668f
1 changed files with 97 additions and 22 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

113
bridges/JapanExpoBridge.php
View file

@ -2,30 +2,105 @@
class JapanExpoBridge extends BridgeAbstract{
public function loadMetadatas() {
$this->maintainer = "Ginko";
$this->name = "JapanExpo";
$this->uri = "http://www.japan-expo-paris.com/fr/actualites";
$this->description = "Returns most recent results from Japan Expo actualités.";
$this->update = "2016-01-19";
$this->maintainer = 'Ginko';
$this->name = 'JapanExpo';
$this->uri = 'http://www.japan-expo-paris.com/fr/actualites';
$this->description = 'Returns most recent entries from Japan Expo actualités.';
$this->update = '2016-06-12';
$this->parameters[] =
'[
{
"name" : "Mode",
"type" : "list",
"identifier" : "mode",
"values" :
[
{
"name" : "Titles only",
"value" : "light"
},
{
"name" : "Full Contents",
"value" : "full"
}
]
}
]';
}
public function collectData(array $param) {
$link = 'http://www.japan-expo-paris.com/fr/actualites';
$html = file_get_html($link) or $this->returnError('Could not request JapanExpo. for : ' . $link , 404);
foreach($html->find('a._tile2') as $element) {
$item = new Item();
$item->uri = $element->href;
$item->title = $element->find('span._title', 0)->plaintext;
$style = $element->find('img.rspvimgset', 0)->style;
preg_match('/url\(([^)]+)\)/', $style, $match);
$item->content = "<img src=".$match[1]."></img><br>".$element->find('span.date', 0)->plaintext;
$this->items[] = $item;
function french_pubdate_to_timestamp($date_to_parse) {
return strtotime(
strtr(
strtolower(str_replace('Publié le ', '', $date_to_parse)),
array(
'janvier' => 'jan',
'février' => 'feb',
'mars' => 'march',
'avril' => 'apr',
'mai' => 'may',
'juin' => 'jun',
'juillet' => 'jul',
'août' => 'aug',
'septembre' => 'sep',
'octobre' => 'oct',
'novembre' => 'nov',
'décembre' => 'dec'
)
)
);
}
$convert_article_images = function ($matches) {
if (is_array($matches) && count($matches) > 1) {
return '<img src="'.$matches[1].'" />';
}
};
$link = 'http://www.japan-expo-paris.com/fr/actualites';
$html = file_get_html($link) or $this->returnError('Could not request JapanExpo: '.$link , 500);
$fullcontent = (!empty($param['mode']) && $param['mode'] == 'full');
$count = 0;
foreach ($html->find('a._tile2') as $element) {
$url = $element->href;
$thumbnail = 'http://s.japan-expo.com/katana/images/JES049/paris.png';
preg_match('/url\(([^)]+)\)/', $element->find('img.rspvimgset', 0)->style, $img_search_result);
if (count($img_search_result) >= 2)
$thumbnail = trim($img_search_result[1], "'");
if ($fullcontent) {
if ($count < 5) {
$article_html = file_get_html($url) or $this->returnError('Could not request JapanExpo: '.$url , 500);
$header = $article_html->find('header.pageHeadBox', 0);
$timestamp = strtotime($header->find('time', 0)->datetime);
$title_html = $header->find('div.section', 0)->next_sibling();
$title = $title_html->plaintext;
$headings = $title_html->next_sibling()->outertext;
$article = $article_html->find('div.content', 0)->innertext;
$article = preg_replace_callback('/<img [^>]+ style="[^\(]+\(\'([^\']+)\'[^>]+>/i', $convert_article_images, $article);
$content = $headings.$article;
} else {
break;
}
} else {
$date_text = $element->find('span.date', 0)->plaintext;
$timestamp = french_pubdate_to_timestamp($date_text);
$title = trim($element->find('span._title', 0)->plaintext);
$content = '<img src="'.$thumbnail.'"></img><br />'.$date_text.'<br /><a href="'.$url.'">Lire l\'article</a>';
}
$item = new \Item();
$item->uri = $url;
$item->title = $title;
$item->timestamp = $timestamp;
$item->thumbnailUri = $thumbnail;
$item->content = $content;
$this->items[] = $item;
$count++;
}
}
public function getName(){
@ -37,6 +112,6 @@ class JapanExpoBridge extends BridgeAbstract{
}
public function getCacheDuration(){
return 86400; // 1 day
return 14400; // 4 hours
}
}