[GoogleGroupsBridge] Add new bridge for Google Groups

bridges/GoogleGroupsBridge.php

@@ -0,0 +1,73 @@
+<?php
+
+class GoogleGroupsBridge extends XPathAbstract {
+	const NAME = 'Google Groups Bridge';
+	const DESCRIPTION = 'Returns the latest posts on a Google Group';
+	const URI = 'https://groups.google.com';
+	const MAINTAINER = 'Yaman Qalieh';
+	const PARAMETERS = array( array(
+		'group' => array(
+			'name' => 'Group id',
+			'title' => 'The string that follows /g/ in the URL',
+			'exampleValue' => 'announce',
+			'required' => true
+		),
+		'account' => array(
+			'name' => 'Account id',
+			'title' => 'Some Google groups have an additional id following /a/ in the URL',
+			'exampleValue' => 'mozilla.org',
+			'required' => false
+		)
+	));
+	const CACHE_TIMEOUT = 3600;
+
+	const TEST_DETECT_PARAMETERS = array(
+		'https://groups.google.com/a/mozilla.org/g/announce' => array(
+			'account' => 'mozilla.org', 'group' => 'announce'
+		),
+		'https://groups.google.com/g/ansible-project' => array(
+			'account' => null, 'group' => 'ansible-project'
+		),
+	);
+
+	const XPATH_EXPRESSION_ITEM = '//div[@class="yhgbKd"]';
+	const XPATH_EXPRESSION_ITEM_TITLE = './/span[@class="o1DPKc"]';
+	const XPATH_EXPRESSION_ITEM_CONTENT = './/span[@class="WzoK"]';
+	const XPATH_EXPRESSION_ITEM_URI = './/a[@class="ZLl54"]/@href';
+	const XPATH_EXPRESSION_ITEM_AUTHOR = './/span[@class="z0zUgf"][last()]';
+	const XPATH_EXPRESSION_ITEM_TIMESTAMP = './/div[@class="tRlaM"]';
+	const XPATH_EXPRESSION_ITEM_ENCLOSURES = '';
+	const XPATH_EXPRESSION_ITEM_CATEGORIES = '';
+	const SETTING_FIX_ENCODING = true;
+
+	protected function getSourceUrl() {
+		$source = self::URI;
+
+		$account = $this->getInput('account');
+		if($account) {
+			$source = $source . '/a/' . $account;
+		}
+		return $source . '/g/' . $this->getInput('group');
+	}
+
+	public function collectData() {
+		parent::collectData();
+
+		# There is duplication in the original item url, so this fixes that
+		$replace_len = strlen(self::getSourceUrl()) - strlen($this->getInput('group')) - 1;
+		foreach ($this->items as $item) {
+			$item->setURI(substr_replace($item->getURI(), self::URI, 0, $replace_len));
+		}
+	}
+
+	const URL_REGEX = '/^https:\/\/groups.google.com(?:\/a\/(?<account>\S+))?(?:\/g\/(?<group>\S+))/';
+
+	public function detectParameters($url) {
+		$params = array();
+		if(preg_match(self::URL_REGEX, $url, $matches)) {
+			$params['group'] = $matches['group'];
+			$params['account'] = $matches['account'];
+		}
+		return $params;
+	}
+}