public function testFindNoLinksInHTML()
{
$html = '<div><p>Hello World</p></div>';
$mf2 = Mf2\parse($html, 'http://source.example.net/');
$links = IndieWeb\MentionClientTest::findOutgoingLinks($mf2);
$this->assertEquals(array(), $links);
}
public function testMultipleHCardsURLNoRelMe()
{
$html = '<div class="h-card"><a href="http://example.com/" class="u-url">Example</a></div>
<div class="h-card"><a href="http://example.com/" class="u-url">Example</a></div>';
$parsed = Mf2\parse($html);
$representative = Mf2\HCard\representative($parsed, 'http://example.com/');
$this->assertFalse($representative);
}
public function get_outgoing_links(Request $request, Response $response)
{
if (!$this->_is_logged_in($request, $response)) {
return $response;
}
$sourceURL = $request->get('url');
$client = new IndieWeb\MentionClient();
$source = $this->http->get($sourceURL);
$parsed = \Mf2\parse($source['body'], $sourceURL);
$links = array_values($client->findOutgoingLinks($parsed));
// Remove the source URL from the list if present
$links = array_filter($links, function ($link) use($sourceURL) {
return $link != $sourceURL;
});
$response->headers->set('Content-Type', 'application/json');
$response->setContent(json_encode(['links' => $links]));
return $response;
}
private static function mf2parse($content, $url)
{
$data = array();
$host = extract_domain_name($url);
switch ($host) {
case 'twitter.com':
$parsed = Mf2\Shim\parseTwitter($content, $url);
break;
default:
$parsed = Mf2\parse($content, $url);
}
if (mf2_cleaner::isMicroformatCollection($parsed)) {
$entries = mf2_cleaner::findMicroformatsByType($parsed, 'h-entry');
if ($entries) {
$entry = $entries[0];
if (mf2_cleaner::isMicroformat($entry)) {
foreach ($entry['properties'] as $key => $value) {
$data[$key] = mf2_cleaner::getPlaintext($entry, $key);
}
$data['published'] = mf2_cleaner::getPublished($entry);
$data['updated'] = mf2_cleaner::getUpdated($entry);
$data['name'] = mf2_cleaner::getPlaintext($entry, 'name');
$data['content'] = mf2_cleaner::getHtml($entry, 'content');
$data['summary'] = mf2_cleaner::getHtml($entry, 'summary');
$data['name'] = trim(preg_replace('/https?:\\/\\/([^ ]+|$)/', '', $data['name']));
$author = mf2_cleaner::getAuthor($entry);
if ($author) {
$data['author'] = array();
foreach ($author['properties'] as $key => $value) {
$data['author'][$key] = mf2_cleaner::getPlaintext($author, $key);
}
$data['author'] = array_filter($data['author']);
}
}
}
}
return array_filter($data);
}
/**
* main worker function: reads all bookmarks sorted by owner;
* gets them, processes them, sends the new entries
*
*/
public function worker()
{
static::debug('worker started', 6);
$args = array('orderby' => 'owner', 'order' => 'ASC', 'limit' => -1);
$bookmarks = get_bookmarks($args);
$currowner = $owner = false;
foreach ($bookmarks as $bookmark) {
/* print_r ($bookmark);
stdClass Object
(
[link_id] => 1
[link_url] => http://devopsreactions.tumblr.com/
[link_name] => http://devopsreactions.tumblr.com/
[link_image] =>
[link_target] =>
[link_description] =>
[link_visible] => Y
[link_owner] => 1
[link_rating] => 0
[link_updated] => 0000-00-00 00:00:00
[link_rel] =>
[link_notes] =>
[link_rss] => http://devopsreactions.tumblr.com/rss
)
*/
//if ( $currowner != $bookmark->link_owner) {
$currowner = $bookmark->link_owner;
$owner = get_userdata($currowner);
$owner = $owner->data;
/* print_r ( $owner );
stdClass Object
(
[ID] => 1
[user_login] => cadeyrn
[user_pass] => $P$B.QI9GiDNyfXC7S75jJ4pcrQjx3awy/
[user_nicename] => cadeyrn
[user_email] => hello@petermolnar.eu
[user_url] => https://petermolnar.eu/
[user_registered] => 2014-04-28 21:36:35
[user_activation_key] =>
[user_status] => 0
[display_name] => Peter Molnar
)
*/
//}
$export_yaml[$owner->user_nicename][] = array('name' => $bookmark->link_name, 'url' => $bookmark->link_url, 'rss' => $bookmark->link_rss, 'description' => $bookmark->link_description, 'lastfetched' => $bookmark->link_updated);
if (!empty($bookmark->link_rss)) {
$this->do_rss($bookmark, $owner);
} else {
static::debug('Switcing into HTML mode', 6);
$url = htmlspecialchars_decode($bookmark->link_url);
static::debug(" fetching {$url}", 6);
$q = wp_remote_get($url);
if (is_wp_error($q)) {
static::debug(' something went wrong: ' . $q->get_error_message(), 4);
continue;
}
if (!is_array($q)) {
continue;
}
if (!isset($q['headers']) || !is_array($q['headers'])) {
continue;
}
if (!isset($q['body']) || empty($q['body'])) {
continue;
}
$ctype = isset($q['headers']['content-type']) ? $q['headers']['content-type'] : 'text/html';
if ($ctype == "application/json") {
static::debug(" content is json", 6);
$content = json_decode($q['body'], true);
} else {
static::debug(" content is html", 6);
$content = Mf2\parse($q['body'], $url);
}
static::debug(" sending it to mf parser", 6);
$this->parse_mf($bookmark, $owner, $content);
}
}
if (function_exists('yaml_emit')) {
$flatroot = WP_CONTENT_DIR . DIRECTORY_SEPARATOR . 'flat';
if (!is_dir($flatroot)) {
if (!mkdir($flatroot)) {
static::debug_log('Failed to create ' . $flatroot . ', exiting YAML creation', 4);
}
}
foreach ($export_yaml as $owner => $bookmarks) {
$export = yaml_emit($bookmarks, YAML_UTF8_ENCODING);
$export_file = $flatroot . DIRECTORY_SEPARATOR . 'bookmarks_' . $owner . '.yml';
file_put_contents($export_file, $export);
}
}
}
/**
* try to get contents of webmention originator
*
* @param int $post_id ID of post
* @param string $source Originator URL
* @param string $target Target URL
*
* @return bool|array false on error; plain array or Mf2 parsed (and
* flattened ) array of remote content on success
*/
protected static function try_receive_remote($post_id, $source, $target)
{
$content = false;
$q = static::_wp_remote_get($source);
if (false === $q) {
return false;
}
$targets = array($target, wp_get_shortlink($post_id), get_permalink($post_id));
$found = false;
foreach ($targets as $k => $t) {
$t = preg_replace('/https?:\\/\\/(?:www.)?/', '', $t);
$t = preg_replace('/#.*/', '', $t);
$t = untrailingslashit($t);
//$targets[ $k ] = $t;
if (!stristr($q['body'], $t)) {
$found = true;
}
}
// check if source really links to target
// this could be a temporary error, so we'll retry later this one as well
if (false == $found) {
static::debug(" missing link to {$t} in remote body", 6);
return false;
}
$ctype = isset($q['headers']['content-type']) ? $q['headers']['content-type'] : 'text/html';
if ("text/plain" == $ctype) {
static::debug(" interesting, plain text webmention. I'm not prepared for this yet", 6);
return false;
} elseif ($ctype == "application/json") {
static::debug(" content is JSON", 6);
$content = json_decode($q['body'], true);
} else {
static::debug(" content is (probably) html, trying to parse it with MF2", 6);
try {
$content = Mf2\parse($q['body'], $source);
} catch (Exception $e) {
static::debug(" parsing MF2 failed: " . $e->getMessage(), 4);
return false;
}
$content = static::flatten_mf2_array($content);
}
return $content;
}
public function testHasRel()
{
$html = '<div class="h-card"><a href="http://example.com/" class="u-url" rel="me"></a></div>';
$parsed = Mf2\parse($html);
$this->assertTrue(Mf2\HCard\has_rel($parsed, 'me', 'http://example.com/'));
}
public function sendMentions($sourceURL, $sourceBody = false)
{
if ($sourceBody) {
$this->_sourceBody = $sourceBody;
$this->_links = self::findOutgoingLinks($sourceBody);
} else {
$body = static::_get($sourceURL);
$this->_sourceBody = $body['body'];
$parsed = \Mf2\parse($this->_sourceBody, $sourceURL);
$this->_links = self::findOutgoingLinks($parsed);
}
$totalAccepted = 0;
foreach ($this->_links as $target) {
self::_debug("Checking {$target} for webmention and pingback endpoints");
if ($this->sendFirstSupportedMention($sourceURL, $target)) {
$totalAccepted++;
}
}
return $totalAccepted;
}
private function getViewModel($mention)
{
$source_base = $this->getBaseUrl($mention['source']);
return new \Aruna\PostViewModel(\Mf2\parse($mention['html'], $source_base));
}
public function get_outgoing_links(Request $request, Response $response)
{
if (!$this->_is_logged_in($request, $response)) {
return $response;
}
$sourceURL = $request->get('url');
$client = new IndieWeb\MentionClient();
$source = $this->http->get($sourceURL);
$parsed = \Mf2\parse($source['body'], $sourceURL);
$links = $client->findOutgoingLinks($parsed);
$response->headers->set('Content-Type', 'application/json');
$response->setContent(json_encode(['links' => array_values($links)]));
return $response;
}
