Crawling PTT

pttCrawler.php

<?php
	require('phpQuery/phpQuery.php');

	$ch = curl_init();

	curl_setopt($ch, CURLOPT_USERAGENT, 'facebookexternalhit/1.1 (+https://www.facebook.com/externalhit_uatext.php)');
	curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);

	$articles = array();

	curl_setopt($ch, CURLOPT_URL, 'https://www.ptt.cc/bbs/Gossiping/index.html');

	$doc = phpQuery::newDocument(curl_exec($ch));
	$entries = $doc['.r-ent'];

	foreach ($entries as $entry) {
		$entry = pq($entry);
		$like = $entry['.nrec span']->html();

		if ($like === '爆' || intval($like) >= 50) {
			$author = $entry['.author']->html();

			if ($author === '-') {
				continue;
			}

			$a = $entry['.title a'];

			$articles[] = array(
					'like' => $like,
					'title' => $a->html(),
					'link' => 'https://www.ptt.cc/' . $a->attr('href'),
					'author' => $author,
					'date' => $entry['.date']->html()
				);
		}
	}
?>