2015-05-05 21:18:12 +03:00
|
|
|
<?php
|
2015-11-05 14:20:11 +03:00
|
|
|
|
2022-07-01 16:10:30 +03:00
|
|
|
class ReporterreBridge extends BridgeAbstract
|
|
|
|
{
|
Fix coding style missed by phpbcf (#2901)
$ composer require --dev friendsofphp/php-cs-fixer
$ echo >.php-cs-fixer.dist.php "<?php
$finder = PhpCsFixer\Finder::create()
->in(__DIR__);
$rules = [
'@PSR12' => true,
// '@PSR12:risky' => true,
'@PHP74Migration' => true,
// '@PHP74Migration:risky' => true,
// buggy, duplicates existing comment sometimes
'no_break_comment' => false,
'array_syntax' => true,
'lowercase_static_reference' => true,
'visibility_required' => false,
// Too much noise
'binary_operator_spaces' => false,
'heredoc_indentation' => false,
'trailing_comma_in_multiline' => false,
];
$config = new PhpCsFixer\Config();
return $config
->setRules($rules)
// ->setRiskyAllowed(true)
->setFinder($finder);
"
$ vendor/bin/php-cs-fixer --version
PHP CS Fixer 3.8.0 BerSzcz against war! by Fabien Potencier and Dariusz Ruminski.
PHP runtime: 8.1.7
$ vendor/bin/php-cs-fixer fix
$ rm .php-cs-fixer.cache
$ vendor/bin/php-cs-fixer fix
2022-07-08 14:00:52 +03:00
|
|
|
const MAINTAINER = 'nyutag';
|
|
|
|
const NAME = 'Reporterre Bridge';
|
|
|
|
const URI = 'https://www.reporterre.net/';
|
|
|
|
const DESCRIPTION = 'Returns the newest articles.';
|
2015-11-05 14:20:11 +03:00
|
|
|
|
2022-07-01 16:10:30 +03:00
|
|
|
private function extractContent($url)
|
|
|
|
{
|
|
|
|
$html2 = getSimpleHTMLDOM($url);
|
|
|
|
$html2 = defaultLinkTo($html2, self::URI);
|
2016-08-04 13:34:40 +03:00
|
|
|
|
2022-07-01 16:10:30 +03:00
|
|
|
foreach ($html2->find('div[style=text-align:justify]') as $e) {
|
|
|
|
$text = $e->outertext;
|
|
|
|
}
|
2016-08-04 13:34:40 +03:00
|
|
|
|
2022-07-01 16:10:30 +03:00
|
|
|
$html2->clear();
|
|
|
|
unset($html2);
|
2016-08-04 13:34:19 +03:00
|
|
|
|
2022-07-01 16:10:30 +03:00
|
|
|
$text = strip_tags($text, '<p><br><a><img>');
|
|
|
|
return $text;
|
|
|
|
}
|
2015-05-05 21:18:12 +03:00
|
|
|
|
2022-07-01 16:10:30 +03:00
|
|
|
public function collectData()
|
|
|
|
{
|
|
|
|
$html = getSimpleHTMLDOM(self::URI . 'spip.php?page=backend');
|
|
|
|
$limit = 0;
|
2015-05-05 21:18:12 +03:00
|
|
|
|
2022-07-01 16:10:30 +03:00
|
|
|
foreach ($html->find('item') as $element) {
|
|
|
|
if ($limit < 5) {
|
|
|
|
$item = [];
|
|
|
|
$item['title'] = html_entity_decode($element->find('title', 0)->plaintext);
|
|
|
|
$item['timestamp'] = strtotime($element->find('dc:date', 0)->plaintext);
|
|
|
|
$item['uri'] = $element->find('guid', 0)->innertext;
|
|
|
|
$item['content'] = html_entity_decode($this->extractContent($item['uri']));
|
|
|
|
$this->items[] = $item;
|
|
|
|
$limit++;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2015-05-05 21:18:12 +03:00
|
|
|
}
|