/**
* Return true if the content filtering is enabled
*
* @access public
* @return boolean
*/
public function isFilteringEnabled()
{
if ($this->config === null) {
return $this->enable_filter;
}
return $this->config->getContentFiltering($this->enable_filter);
}
public function testOverrideFilters()
{
$data = '<iframe src="http://www.kickstarter.com/projects/lefnire/habitrpg-mobile/widget/video.html" height="480" width="640" frameborder="0"></iframe>';
$expected = '<iframe src="https://www.kickstarter.com/projects/lefnire/habitrpg-mobile/widget/video.html" height="480" width="640" frameborder="0"></iframe>';
$f = Filter::html($data, 'http://blabla');
$f->attribute->setIframeWhitelist(array('http://www.kickstarter.com'));
$this->assertEquals($expected, $f->execute());
$data = '<iframe src="http://www.youtube.com/bla" height="480" width="640" frameborder="0"></iframe>';
$f = Filter::html($data, 'http://blabla');
$f->attribute->setIframeWhitelist(array('http://www.kickstarter.com'));
$this->assertEmpty($f->execute());
$config = new Config();
$config->setFilterWhitelistedTags(array('p' => array('title')));
$f = Filter::html('<p>Test<strong>boo</strong></p>', 'http://blabla');
$f->setConfig($config);
$this->assertEquals('<p>Testboo</p>', $f->execute());
}
/**
* Get the list of folders that contains rules.
*
* @return array
*/
public function getRulesFolders()
{
$folders = array(__DIR__ . '/../Rules');
if ($this->config !== null && $this->config->getGrabberRulesFolder() !== null) {
$folders[] = $this->config->getGrabberRulesFolder();
}
return $folders;
}
function addProxyToTags($html, $website, $proxy_images, $cloak_referrer)
{
if ($html === '' || !$cloak_referrer && !$proxy_images || !$cloak_referrer && $proxy_images && !Helper\isSecureConnection()) {
// only proxy enabled, but not connected via HTTPS
return $html;
}
$config = new PicoFeedConfig();
$config->setFilterImageProxyUrl('?action=proxy&url=%s');
if (!$cloak_referrer && $proxy_images) {
// image proxy mode only: https links do not need to be proxied, since
// they do not trigger mixed content warnings.
$config->setFilterImageProxyProtocol('http');
} elseif (!$proxy_images && $cloak_referrer && Helper\isSecureConnection()) {
// cloaking mode only: if a request from a HTTPS connection to a HTTP
// connection is made, the referrer will be omitted by the browser.
// Only the referrer for HTTPS to HTTPs requests needs to be cloaked.
$config->setFilterImageProxyProtocol('https');
}
$filter = Filter::html($html, $website);
$filter->setConfig($config);
return $filter->execute();
}
/**
* Get a parser instance
*
* @access public
* @param string $url Site url
* @param string $content Feed content
* @param string $encoding HTTP encoding
* @return \PicoFeed\Parser\Parser
*/
public function getParser($url, $content, $encoding)
{
$format = $this->detectFormat($content);
if (empty($format)) {
throw new UnsupportedFeedFormatException('Unable to detect feed format');
}
$className = '\\PicoFeed\\Parser\\' . $format;
$parser = new $className($content, $encoding, $url);
$parser->setHashAlgo($this->config->getParserHashAlgo());
$parser->setTimezone($this->config->getTimezone());
$parser->setConfig($this->config);
return $parser;
}
public function testGetRulesFolders()
{
// No custom path
$loader = new RuleLoader(new Config());
$dirs = $loader->getRulesFolders();
$this->assertNotEmpty($dirs);
$this->assertCount(1, $dirs);
$this->assertTrue(strpos($dirs[0], '/../Rules') !== false);
// Custom path
$config = new Config();
$config->setGrabberRulesFolder('/foobar/rules');
$loader = new RuleLoader($config);
$dirs = $loader->getRulesFolders();
$this->assertNotEmpty($dirs);
$this->assertCount(2, $dirs);
$this->assertTrue(strpos($dirs[0], '/../Rules') !== false);
$this->assertEquals('/foobar/rules', $dirs[1]);
// No custom path with empty config object
$loader = new RuleLoader(new Config());
$dirs = $loader->getRulesFolders();
$this->assertNotEmpty($dirs);
$this->assertCount(1, $dirs);
$this->assertTrue(strpos($dirs[0], '/../Rules') !== false);
}
/**
* Set config object.
*
* @param \PicoFeed\Config\Config $config Config instance
*
* @return \PicoFeed\Filter\Html
*/
public function setConfig($config)
{
$this->config = $config;
if ($this->config !== null) {
$this->attribute->setImageProxyCallback($this->config->getFilterImageProxyCallback());
$this->attribute->setImageProxyUrl($this->config->getFilterImageProxyUrl());
$this->attribute->setImageProxyProtocol($this->config->getFilterImageProxyProtocol());
$this->attribute->setIframeWhitelist($this->config->getFilterIframeWhitelist(array()));
$this->attribute->setIntegerAttributes($this->config->getFilterIntegerAttributes(array()));
$this->attribute->setAttributeOverrides($this->config->getFilterAttributeOverrides(array()));
$this->attribute->setRequiredAttributes($this->config->getFilterRequiredAttributes(array()));
$this->attribute->setMediaBlacklist($this->config->getFilterMediaBlacklist(array()));
$this->attribute->setMediaAttributes($this->config->getFilterMediaAttributes(array()));
$this->attribute->setSchemeWhitelist($this->config->getFilterSchemeWhitelist(array()));
$this->attribute->setWhitelistedAttributes($this->config->getFilterWhitelistedTags(array()));
$this->tag->setWhitelistedTags(array_keys($this->config->getFilterWhitelistedTags(array())));
}
return $this;
}
/**
* Download the HTML content.
*
* @return bool
*/
public function download()
{
if (!empty($this->url)) {
// Clear everything
$this->html = '';
$this->content = '';
$this->encoding = '';
try {
$client = Client::getInstance();
$client->setConfig($this->config);
$client->setTimeout($this->config->getGrabberTimeout());
$client->setUserAgent($this->config->getGrabberUserAgent());
$client->execute($this->url);
$this->url = $client->getUrl();
$this->html = $client->getContent();
$this->encoding = $client->getEncoding();
return true;
} catch (ClientException $e) {
Logger::setMessage(get_called_class() . ': ' . $e->getMessage());
}
}
return false;
}
/**
* Set config object.
*
* @param \PicoFeed\Config\Config $config Config instance
*
* @return \PicoFeed\Client\Client
*/
public function setConfig(Config $config)
{
if ($config !== null) {
$this->setTimeout($config->getClientTimeout());
$this->setUserAgent($config->getClientUserAgent());
$this->setMaxRedirections($config->getMaxRedirections());
$this->setMaxBodySize($config->getMaxBodySize());
$this->setProxyHostname($config->getProxyHostname());
$this->setProxyPort($config->getProxyPort());
$this->setProxyUsername($config->getProxyUsername());
$this->setProxyPassword($config->getProxyPassword());
}
return $this;
}
function get_reader_config()
{
$config = new ReaderConfig();
$config->setTimezone(get('timezone'));
// Client
$config->setClientTimeout(HTTP_TIMEOUT);
$config->setClientUserAgent(HTTP_USER_AGENT);
// Grabber
$config->setGrabberRulesFolder(RULES_DIRECTORY);
// Proxy
$config->setProxyHostname(PROXY_HOSTNAME);
$config->setProxyPort(PROXY_PORT);
$config->setProxyUsername(PROXY_USERNAME);
$config->setProxyPassword(PROXY_PASSWORD);
// Filter
$config->setFilterIframeWhitelist(get_iframe_whitelist());
if ((bool) get('debug_mode')) {
Logger::enable();
}
// Parser
$config->setParserHashAlgo('crc32b');
return $config;
}
public function __construct(array $urlParams = [])
{
parent::__construct('news', $urlParams);
// files
$this->registerFileContents('checksums', 'checksum.json');
$this->registerFileContents('info', 'info.xml');
// parameters
$this->registerParameter('exploreDir', __DIR__ . '/../explore');
$this->registerParameter('configFile', 'config.ini');
// factories
$this->registerFactory(ItemMapper::class, MapperFactory::class);
/**
* App config parser
*/
$this->registerService(AppConfig::class, function ($c) {
$config = new AppConfig($c->query(INavigationManager::class), $c->query(IURLGenerator::class));
$config->loadConfig($c->query('info'));
return $config;
});
/**
* Core
*/
$this->registerService('LoggerParameters', function ($c) {
return ['app' => $c->query('AppName')];
});
$this->registerService('databaseType', function ($c) {
return $c->query(IConfig::class)->getSystemValue('dbtype');
});
$this->registerService('ConfigView', function ($c) {
$fs = $c->query(IRootFolder::class);
$path = 'news/config';
if ($fs->nodeExists($path)) {
return $fs->get($path);
} else {
return $fs->newFolder($path);
}
});
$this->registerService(Config::class, function ($c) {
$config = new Config($c->query('ConfigView'), $c->query(ILogger::class), $c->query('LoggerParameters'));
$config->read($c->query('configFile'), true);
return $config;
});
$this->registerService(HTMLPurifier::class, function ($c) {
$directory = $c->query(IConfig::class)->getSystemValue('datadirectory') . '/news/cache/purifier';
if (!is_dir($directory)) {
mkdir($directory, 0770, true);
}
$config = HTMLPurifier_Config::createDefault();
$config->set('HTML.ForbiddenAttributes', 'class');
$config->set('Cache.SerializerPath', $directory);
$config->set('HTML.SafeIframe', true);
$config->set('URI.SafeIframeRegexp', '%^https://(?:www\\.)?(' . 'youtube(?:-nocookie)?.com/embed/|' . 'player.vimeo.com/video/)%');
//allow YouTube and Vimeo
$def = $config->getHTMLDefinition(true);
$def->addAttribute('iframe', 'allowfullscreen', 'Bool');
return new HTMLPurifier($config);
});
/**
* Fetchers
*/
$this->registerService(PicoFeedConfig::class, function ($c) {
// FIXME: move this into a separate class for testing?
$config = $c->query(Config::class);
$appConfig = $c->query(AppConfig::class);
$proxy = $c->query(ProxyConfigParser::class);
$userAgent = 'ownCloud News/' . $appConfig->getConfig('version') . ' (+https://owncloud.org/; 1 subscriber;)';
$pico = new PicoFeedConfig();
$pico->setClientUserAgent($userAgent)->setClientTimeout($config->getFeedFetcherTimeout())->setMaxRedirections($config->getMaxRedirects())->setMaxBodySize($config->getMaxSize())->setParserHashAlgo('md5');
// proxy settings
$proxySettings = $proxy->parse();
$host = $proxySettings['host'];
$port = $proxySettings['port'];
$user = $proxySettings['user'];
$password = $proxySettings['password'];
if ($host) {
$pico->setProxyHostname($host);
if ($port) {
$pico->setProxyPort($port);
}
}
if ($user) {
$pico->setProxyUsername($user)->setProxyPassword($password);
}
return $pico;
});
$this->registerService(Fetcher::class, function ($c) {
$fetcher = new Fetcher();
// register fetchers in order, the most generic fetcher should be
// the last one
$fetcher->registerFetcher($c->query(YoutubeFetcher::class));
$fetcher->registerFetcher($c->query(FeedFetcher::class));
return $fetcher;
});
}
public function testsetFilterImageProxyCallback()
{
$config = new Config();
$config->setFilterImageProxyCallback(function ($image_url) {
$key = hash_hmac('sha1', $image_url, 'secret');
return 'https://mypublicproxy/' . $key . '/' . rawurlencode($image_url);
});
$f = Filter::html('<p>Image <img src="/image.png" alt="My Image"/></p>', 'http://foo');
$f->setConfig($config);
$this->assertEquals('<p>Image <img src="https://mypublicproxy/4924964043f3119b3cf2b07b1922d491bcc20092/' . rawurlencode('http://foo/image.png') . '" alt="My Image"/></p>', $f->execute());
}
public function __construct(array $urlParams = [])
{
parent::__construct('news', $urlParams);
$container = $this->getContainer();
/**
* Mappers
*/
$container->registerService('OCA\\News\\Db\\ItemMapper', function ($c) {
return $c->query('OCA\\News\\Db\\MapperFactory')->getItemMapper($c->query('OCP\\IDBConnection'));
});
/**
* App config parser
*/
$container->registerService('OCA\\News\\Config\\AppConfig', function ($c) {
$config = new AppConfig($c->query('OCP\\INavigationManager'), $c->query('OCP\\IURLGenerator'));
$config->loadConfig(__DIR__ . '/info.xml');
return $config;
});
/**
* Core
*/
$container->registerService('LoggerParameters', function ($c) {
return ['app' => $c->query('AppName')];
});
$container->registerService('DatabaseType', function ($c) {
return $c->query('OCP\\IConfig')->getSystemValue('dbtype');
});
/**
* Utility
*/
$container->registerService('ConfigPath', function () {
return 'config.ini';
});
$container->registerService('ConfigView', function ($c) {
$fs = $c->query('OCP\\Files\\IRootFolder');
$path = 'news/config';
if ($fs->nodeExists($path)) {
return $fs->get($path);
} else {
return $fs->newFolder($path);
}
});
$container->registerService('OCA\\News\\Config\\Config', function ($c) {
$config = new Config($c->query('ConfigView'), $c->query('OCP\\ILogger'), $c->query('LoggerParameters'));
$config->read($c->query('ConfigPath'), true);
return $config;
});
$container->registerService('HTMLPurifier', function ($c) {
$directory = $c->query('OCP\\IConfig')->getSystemValue('datadirectory') . '/news/cache/purifier';
if (!is_dir($directory)) {
mkdir($directory, 0770, true);
}
$config = HTMLPurifier_Config::createDefault();
$config->set('HTML.ForbiddenAttributes', 'class');
$config->set('Cache.SerializerPath', $directory);
$config->set('HTML.SafeIframe', true);
$config->set('URI.SafeIframeRegexp', '%^https://(?:www\\.)?(' . 'youtube(?:-nocookie)?.com/embed/|' . 'player.vimeo.com/video/)%');
//allow YouTube and Vimeo
return new HTMLPurifier($config);
});
$container->registerService('OCA\\News\\ArticleEnhancer\\Enhancer', function ($c) {
$enhancer = new Enhancer();
// register simple enhancers from config json file
$xpathEnhancerConfig = file_get_contents(__DIR__ . '/../articleenhancer/xpathenhancers.json');
$xpathEnhancerConfig = json_decode($xpathEnhancerConfig, true);
foreach ($xpathEnhancerConfig as $feed => $config) {
$articleEnhancer = new XPathArticleEnhancer($c->query('OCA\\News\\Utility\\PicoFeedClientFactory'), $config);
$enhancer->registerEnhancer($feed, $articleEnhancer);
}
$regexEnhancerConfig = file_get_contents(__DIR__ . '/../articleenhancer/regexenhancers.json');
$regexEnhancerConfig = json_decode($regexEnhancerConfig, true);
foreach ($regexEnhancerConfig as $feed => $config) {
foreach ($config as $matchArticleUrl => $regex) {
$articleEnhancer = new RegexArticleEnhancer($matchArticleUrl, $regex);
$enhancer->registerEnhancer($feed, $articleEnhancer);
}
}
return $enhancer;
});
/**
* Fetchers
*/
$container->registerService('PicoFeed\\Config\\Config', function ($c) {
// FIXME: move this into a separate class for testing?
$config = $c->query('OCA\\News\\Config\\Config');
$appConfig = $c->query('OCA\\News\\Config\\AppConfig');
$proxy = $c->query('OCA\\News\\Utility\\ProxyConfigParser');
$userAgent = 'ownCloud News/' . $appConfig->getConfig('version') . ' (+https://owncloud.org/; 1 subscriber;)';
$pico = new PicoFeedConfig();
$pico->setClientUserAgent($userAgent)->setClientTimeout($config->getFeedFetcherTimeout())->setMaxRedirections($config->getMaxRedirects())->setMaxBodySize($config->getMaxSize())->setParserHashAlgo('md5');
// proxy settings
$proxySettings = $proxy->parse();
$host = $proxySettings['host'];
$port = $proxySettings['port'];
$user = $proxySettings['user'];
$password = $proxySettings['password'];
if ($host) {
$pico->setProxyHostname($host);
if ($port) {
$pico->setProxyPort($port);
}
}
if ($user) {
$pico->setProxyUsername($user)->setProxyPassword($password);
}
return $pico;
});
$container->registerService('OCA\\News\\Fetcher\\Fetcher', function ($c) {
$fetcher = new Fetcher();
// register fetchers in order
// the most generic fetcher should be the last one
$fetcher->registerFetcher($c->query('OCA\\News\\Fetcher\\YoutubeFetcher'));
$fetcher->registerFetcher($c->query('OCA\\News\\Fetcher\\FeedFetcher'));
return $fetcher;
});
$container->registerService('OCA\\News\\Explore\\RecommendedSites', function () {
return new RecommendedSites(__DIR__ . '/../explore');
});
}
/**
* Constructor.
*
* @param \PicoFeed\Config\Config $config Config class instance
*/
public function __construct(Config $config = null)
{
$this->config = $config ?: new Config();
Logger::setTimezone($this->config->getTimezone());
}
/**
* Check if a tag is on the whitelist.
*
* @param string $tag Tag name
*
* @return bool
*/
public function isAllowedTag($tag)
{
return in_array($tag, array_merge($this->tag_whitelist, array_keys($this->config->getFilterWhitelistedTags(array()))));
}
