From 0dac13d197248bf24ea51cb7911d32286764c0c8 Mon Sep 17 00:00:00 2001 From: Brion Vibber Date: Thu, 18 Feb 2010 21:22:21 +0000 Subject: OStatus refactoring to clean up profile vs feed and fix up subscription issues. PuSH subscription maintenance broken back out to FeedSub, letting Ostatus_profile deal with the profile level (user or group, with unique id URI) --- plugins/OStatus/lib/feeddiscovery.php | 50 +++-- plugins/OStatus/lib/feedmunger.php | 350 ---------------------------------- 2 files changed, 39 insertions(+), 361 deletions(-) delete mode 100644 plugins/OStatus/lib/feedmunger.php (limited to 'plugins/OStatus/lib') diff --git a/plugins/OStatus/lib/feeddiscovery.php b/plugins/OStatus/lib/feeddiscovery.php index 39985fc90..7afb71bdc 100644 --- a/plugins/OStatus/lib/feeddiscovery.php +++ b/plugins/OStatus/lib/feeddiscovery.php @@ -48,6 +48,14 @@ class FeedSubNoFeedException extends FeedSubException { } +class FeedSubBadXmlException extends FeedSubException +{ +} + +class FeedSubNoHubException extends FeedSubException +{ +} + /** * Given a web page or feed URL, discover the final location of the feed * and return its current contents. @@ -57,21 +65,25 @@ class FeedSubNoFeedException extends FeedSubException * if ($feed->discoverFromURL($url)) { * print $feed->uri; * print $feed->type; - * processFeed($feed->body); + * processFeed($feed->feed); // DOMDocument * } */ class FeedDiscovery { public $uri; public $type; - public $body; + public $feed; + /** Post-initialize query helper... */ + public function getLink($rel, $type=null) + { + // @fixme check for non-Atom links in RSS2 feeds as well + return self::getAtomLink($rel, $type); + } - public function feedMunger() + public function getAtomLink($rel, $type=null) { - require_once 'XML/Feed/Parser.php'; - $feed = new XML_Feed_Parser($this->body, false, false, true); // @fixme - return new FeedMunger($feed, $this->uri); + return ActivityUtils::getLink($this->feed->documentElement, $rel, $type); } /** @@ -90,6 +102,7 @@ class FeedDiscovery $client = new HTTPClient(); $response = $client->get($url); } catch (HTTP_Request2_Exception $e) { + common_log(LOG_ERR, __METHOD__ . " Failure for $url - " . $e->getMessage()); throw new FeedSubBadURLException($e); } @@ -107,7 +120,12 @@ class FeedDiscovery return $this->initFromResponse($response); } - + + function discoverFromFeedURL($url) + { + return $this->discoverFromURL($url, false); + } + function initFromResponse($response) { if (!$response->isOk()) { @@ -122,16 +140,26 @@ class FeedDiscovery $type = $response->getHeader('Content-Type'); if (preg_match('!^(text/xml|application/xml|application/(rss|atom)\+xml)!i', $type)) { - $this->uri = $sourceurl; - $this->type = $type; - $this->body = $body; - return true; + return $this->init($sourceurl, $type, $body); } else { common_log(LOG_WARNING, "Unrecognized feed type $type for $sourceurl"); throw new FeedSubUnrecognizedTypeException($type); } } + function init($sourceurl, $type, $body) + { + $feed = new DOMDocument(); + if ($feed->loadXML($body)) { + $this->uri = $sourceurl; + $this->type = $type; + $this->feed = $feed; + return $this->uri; + } else { + throw new FeedSubBadXmlException($url); + } + } + /** * @param string $url source URL, used to resolve relative links * @param string $body HTML body text diff --git a/plugins/OStatus/lib/feedmunger.php b/plugins/OStatus/lib/feedmunger.php deleted file mode 100644 index e8c46de90..000000000 --- a/plugins/OStatus/lib/feedmunger.php +++ /dev/null @@ -1,350 +0,0 @@ -. - */ - -/** - * @package FeedSubPlugin - * @maintainer Brion Vibber - */ - -if (!defined('STATUSNET') && !defined('LACONICA')) { exit(1); } - -class FeedSubPreviewNotice extends Notice -{ - protected $fetched = true; - - function __construct($profile) - { - $this->profile = $profile; - $this->profile_id = 0; - } - - function getProfile() - { - return $this->profile; - } - - function find() - { - return true; - } - - function fetch() - { - $got = $this->fetched; - $this->fetched = false; - return $got; - } -} - -class FeedSubPreviewProfile extends Profile -{ - function getAvatar($width, $height=null) - { - return new FeedSubPreviewAvatar($width, $height, $this->avatar); - } -} - -class FeedSubPreviewAvatar extends Avatar -{ - function __construct($width, $height, $remote) - { - $this->remoteImage = $remote; - } - - function displayUrl() { - return $this->remoteImage; - } -} - -class FeedMunger -{ - /** - * @param XML_Feed_Parser $feed - */ - function __construct($feed, $url=null) - { - $this->feed = $feed; - $this->url = $url; - } - - function ostatusProfile() - { - $profile = new Ostatus_profile(); - $profile->feeduri = $this->url; - $profile->homeuri = $this->feed->link; - $profile->huburi = $this->getHubLink(); - $salmon = $this->getSalmonLink(); - if ($salmon) { - $profile->salmonuri = $salmon; - } - return $profile; - } - - function getAtomLink($item, $attribs=array()) - { - // XML_Feed_Parser gets confused by multiple elements. - $dom = $item->model; - - // Note that RSS feeds would embed an so this should work for both. - /// http://code.google.com/p/pubsubhubbub/wiki/RssFeeds - // - $links = $dom->getElementsByTagNameNS('http://www.w3.org/2005/Atom', 'link'); - for ($i = 0; $i < $links->length; $i++) { - $node = $links->item($i); - if ($node->hasAttributes()) { - $href = $node->attributes->getNamedItem('href'); - if ($href) { - $matches = 0; - foreach ($attribs as $name => $val) { - $attrib = $node->attributes->getNamedItem($name); - if ($attrib && $attrib->value == $val) { - $matches++; - } - } - if ($matches == count($attribs)) { - return $href->value; - } - } - } - } - return false; - } - - function getRssLink($item) - { - // XML_Feed_Parser gets confused by multiple elements. - $dom = $item->model; - - // Note that RSS feeds would embed an so this should work for both. - /// http://code.google.com/p/pubsubhubbub/wiki/RssFeeds - // - $links = $dom->getElementsByTagName('link'); - for ($i = 0; $i < $links->length; $i++) { - $node = $links->item($i); - if (!$node->hasAttributes()) { - return $node->textContent; - } - } - return false; - } - - function getAltLink($item) - { - // Check for an atom link... - $link = $this->getAtomLink($item, array('rel' => 'alternate', 'type' => 'text/html')); - if (!$link) { - $link = $this->getRssLink($item); - } - return $link; - } - - function getHubLink() - { - return $this->getAtomLink($this->feed, array('rel' => 'hub')); - } - - function getSalmonLink() - { - return $this->getAtomLink($this->feed, array('rel' => 'salmon')); - } - - function getSelfLink() - { - return $this->getAtomLink($this->feed, array('rel' => 'self')); - } - - /** - * Get an appropriate avatar image source URL, if available. - * @return mixed string or false - */ - function getAvatar() - { - $logo = $this->feed->logo; - if ($logo) { - return $logo; - } - $icon = $this->feed->icon; - if ($icon) { - return $icon; - } - return common_path('plugins/OStatus/images/48px-Feed-icon.svg.png'); - } - - function profile($preview=false) - { - if ($preview) { - $profile = new FeedSubPreviewProfile(); - } else { - $profile = new Profile(); - } - - // @todo validate/normalize nick? - $profile->nickname = $this->feed->title; - $profile->fullname = $this->feed->title; - $profile->homepage = $this->getAltLink($this->feed); - $profile->bio = $this->feed->description; - $profile->profileurl = $this->getAltLink($this->feed); - - if ($preview) { - $profile->avatar = $this->getAvatar(); - } - - // @todo tags from categories - // @todo lat/lon/location? - - return $profile; - } - - function notice($index=1, $preview=false) - { - $entry = $this->feed->getEntryByOffset($index); - if (!$entry) { - return null; - } - - if ($preview) { - $notice = new FeedSubPreviewNotice($this->profile(true)); - $notice->id = -1; - } else { - $notice = new Notice(); - $notice->profile_id = $this->profileIdForEntry($index); - } - - $link = $this->getAltLink($entry); - if (empty($link)) { - if (preg_match('!^https?://!', $entry->id)) { - $link = $entry->id; - common_log(LOG_DEBUG, "No link on entry, using URL from id: $link"); - } - } - $notice->uri = $link; - $notice->url = $link; - $notice->content = $this->noticeFromEntry($entry); - $notice->rendered = common_render_content($notice->content, $notice); // @fixme this is failing on group posts - $notice->created = common_sql_date($entry->updated); // @fixme - $notice->is_local = Notice::GATEWAY; - $notice->source = 'feed'; - - $location = $this->getLocation($entry); - if ($location) { - if ($location->location_id) { - $notice->location_ns = $location->location_ns; - $notice->location_id = $location->location_id; - } - $notice->lat = $location->lat; - $notice->lon = $location->lon; - } - - return $notice; - } - - function profileIdForEntry($index=1) - { - // hack hack hack - // should get profile for this entry's author... - $feeduri = $this->getSelfLink(); - $remote = Ostatus_profile::staticGet('feeduri', $feeduri); - if ($remote) { - return $remote->profile_id; - } else { - throw new Exception("Can't find feed profile for $feeduri"); - } - } - - /** - * Parse location given as a GeoRSS-simple point, if provided. - * http://www.georss.org/simple - * - * @param feed item $entry - * @return mixed Location or false - */ - function getLocation($entry) - { - $dom = $entry->model; - $points = $dom->getElementsByTagNameNS('http://www.georss.org/georss', 'point'); - - for ($i = 0; $i < $points->length; $i++) { - $point = $points->item(0)->textContent; - $point = str_replace(',', ' ', $point); // per spec "treat commas as whitespace" - $point = preg_replace('/\s+/', ' ', $point); - $point = trim($point); - $coords = explode(' ', $point); - if (count($coords) == 2) { - list($lat, $lon) = $coords; - if (is_numeric($lat) && is_numeric($lon)) { - common_log(LOG_INFO, "Looking up location for $lat $lon from georss"); - return Location::fromLatLon($lat, $lon); - } - } - common_log(LOG_ERR, "Ignoring bogus georss:point value $point"); - } - - return false; - } - - /** - * @param XML_Feed_Type $entry - * @return string notice text, within post size limit - */ - function noticeFromEntry($entry) - { - $max = Notice::maxContent(); - $ellipsis = "\xe2\x80\xa6"; // U+2026 HORIZONTAL ELLIPSIS - $title = $entry->title; - $link = $entry->link; - - // @todo We can get entries like this: - // $cats = $entry->getCategory('category', array(0, true)); - // but it feels like an awful hack. If it's accessible cleanly, - // try adding #hashtags from the categories/tags on a post. - - $title = $entry->title; - $link = $this->getAltLink($entry); - if ($link) { - // Blog post or such... - // @todo Should we force a language here? - $format = _m('New post: "%1$s" %2$s'); - $out = sprintf($format, $title, $link); - - // Trim link if needed... - if (mb_strlen($out) > $max) { - $link = common_shorten_url($link); - $out = sprintf($format, $title, $link); - } - - // Trim title if needed... - if (mb_strlen($out) > $max) { - $used = mb_strlen($out) - mb_strlen($title); - $available = $max - $used - mb_strlen($ellipsis); - $title = mb_substr($title, 0, $available) . $ellipsis; - $out = sprintf($format, $title, $link); - } - } else { - // No link? Consider a bare status update. - if (mb_strlen($title) > $max) { - $available = $max - mb_strlen($ellipsis); - $out = mb_substr($title, 0, $available) . $ellipsis; - } else { - $out = $title; - } - } - - return $out; - } -} -- cgit v1.2.3-54-g00ecf