From 05e3768e6a833d99a5180d4306d68f59e2d8f8c9 Mon Sep 17 00:00:00 2001 From: Evan Prodromou Date: Fri, 19 Mar 2010 09:48:39 -0500 Subject: [PATCH] Parse RSS items as activities First steps to parsing RSS items as activities. RSS feeds don't seem to have enough data to make good remote profiles, but this may work with some "hints". --- lib/activity.php | 272 ++++++++++++++++++++++++++++++----- tests/ActivityParseTests.php | 95 +++++++++++- 2 files changed, 333 insertions(+), 34 deletions(-) diff --git a/lib/activity.php b/lib/activity.php index c67d090f72..5b304020d6 100644 --- a/lib/activity.php +++ b/lib/activity.php @@ -643,38 +643,11 @@ class ActivityObject ); if ($element->tagName == 'author') { - - $this->type = self::PERSON; // XXX: is this fair? - $this->title = $this->_childContent($element, self::NAME); - $this->id = $this->_childContent($element, self::URI); - - if (empty($this->id)) { - $email = $this->_childContent($element, self::EMAIL); - if (!empty($email)) { - // XXX: acct: ? - $this->id = 'mailto:'.$email; - } - } - + $this->_fromAuthor($element); + } else if ($element->tagName == 'item') { + $this->_fromRssItem($element); } else { - - $this->type = $this->_childContent($element, Activity::OBJECTTYPE, - Activity::SPEC); - - if (empty($this->type)) { - $this->type = ActivityObject::NOTE; - } - - $this->id = $this->_childContent($element, self::ID); - $this->title = $this->_childContent($element, self::TITLE); - $this->summary = $this->_childContent($element, self::SUMMARY); - - $this->source = $this->_getSource($element); - - $this->content = ActivityUtils::getContent($element); - - $this->link = ActivityUtils::getPermalink($element); - + $this->_fromAtomEntry($element); } // Some per-type attributes... @@ -697,6 +670,72 @@ class ActivityObject } } + private function _fromAuthor($element) + { + $this->type = self::PERSON; // XXX: is this fair? + $this->title = $this->_childContent($element, self::NAME); + $this->id = $this->_childContent($element, self::URI); + + if (empty($this->id)) { + $email = $this->_childContent($element, self::EMAIL); + if (!empty($email)) { + // XXX: acct: ? + $this->id = 'mailto:'.$email; + } + } + } + + private function _fromAtomEntry($element) + { + $this->type = $this->_childContent($element, Activity::OBJECTTYPE, + Activity::SPEC); + + if (empty($this->type)) { + $this->type = ActivityObject::NOTE; + } + + $this->id = $this->_childContent($element, self::ID); + $this->title = $this->_childContent($element, self::TITLE); + $this->summary = $this->_childContent($element, self::SUMMARY); + + $this->source = $this->_getSource($element); + + $this->content = ActivityUtils::getContent($element); + + $this->link = ActivityUtils::getPermalink($element); + } + + // @fixme rationalize with Activity::_fromRssItem() + + private function _fromRssItem($item) + { + $this->title = ActivityUtils::childContent($item, ActivityObject::TITLE, Activity::RSS); + + $contentEl = ActivityUtils::child($item, ActivityUtils::CONTENT, Activity::CONTENTNS); + + if (!empty($contentEl)) { + $this->content = htmlspecialchars_decode($contentEl->textContent, ENT_QUOTES); + } else { + $descriptionEl = ActivityUtils::child($item, Activity::DESCRIPTION, Activity::RSS); + if (!empty($descriptionEl)) { + $this->content = htmlspecialchars_decode($descriptionEl->textContent, ENT_QUOTES); + } + } + + $this->link = ActivityUtils::childContent($item, ActivityUtils::LINK, Activity::RSS); + + $guidEl = ActivityUtils::child($item, Activity::GUID, Activity::RSS); + + if (!empty($guidEl)) { + $this->id = $guidEl->textContent; + + if ($guidEl->hasAttribute('isPermaLink')) { + // overwrites + $this->link = $this->id; + } + } + } + private function _childContent($element, $tag, $namespace=ActivityUtils::ATOM) { return ActivityUtils::childContent($element, $tag, $namespace); @@ -1051,6 +1090,21 @@ class Activity const PUBLISHED = 'published'; const UPDATED = 'updated'; + const RSS = null; // no namespace! + + const PUBDATE = 'pubDate'; + const DESCRIPTION = 'description'; + const GUID = 'guid'; + const SELF = 'self'; + const IMAGE = 'image'; + const URL = 'url'; + + const DC = 'http://purl.org/dc/elements/1.1/'; + + const CREATOR = 'creator'; + + const CONTENTNS = 'http://purl.org/rss/1.0/modules/content/'; + public $actor; // an ActivityObject public $verb; // a string (the URL) public $object; // an ActivityObject @@ -1081,8 +1135,6 @@ class Activity return; } - $this->entry = $entry; - // Insist on a feed's root DOMElement; don't allow a DOMDocument if ($feed instanceof DOMDocument) { throw new ClientException( @@ -1090,8 +1142,22 @@ class Activity ); } + $this->entry = $entry; $this->feed = $feed; + if ($entry->namespaceURI == Activity::ATOM && + $entry->localName == 'entry') { + $this->_fromAtomEntry($entry, $feed); + } else if ($entry->namespaceURI == Activity::RSS && + $entry->localName == 'item') { + $this->_fromRssItem($entry, $feed); + } else { + throw new Exception("Unknown DOM element: {$entry->namespaceURI} {$entry->localName}"); + } + } + + function _fromAtomEntry($entry, $feed) + { $pubEl = $this->_child($entry, self::PUBLISHED, self::ATOM); if (!empty($pubEl)) { @@ -1177,6 +1243,69 @@ class Activity } } + function _fromRssItem($item, $rss) + { + $verbEl = $this->_child($item, self::VERB); + + if (!empty($verbEl)) { + $this->verb = trim($verbEl->textContent); + } else { + $this->verb = ActivityVerb::POST; + // XXX: do other implied stuff here + } + + $pubDateEl = $this->_child($item, self::PUBDATE, self::RSS); + + if (!empty($pubDateEl)) { + $this->time = strtotime($pubDateEl->textContent); + } + + $authorEl = $this->_child($item, self::AUTHOR, self::RSS); + + if (!empty($authorEl)) { + $this->actor = $this->_fromRssAuthor($authorEl); + } else { + $dcCreatorEl = $this->_child($item, self::CREATOR, self::DC); + if (!empty($dcCreatorEl)) { + $this->actor = $this->_fromDcCreator($dcCreatorEl); + } else if (!empty($rss)) { + $this->actor = $this->_fromRss($rss); + } + } + + $this->title = ActivityUtils::childContent($item, ActivityObject::TITLE, self::RSS); + + $contentEl = ActivityUtils::child($item, ActivityUtils::CONTENT, self::CONTENTNS); + + if (!empty($contentEl)) { + $this->content = htmlspecialchars_decode($contentEl->textContent, ENT_QUOTES); + } else { + $descriptionEl = ActivityUtils::child($item, self::DESCRIPTION, self::RSS); + if (!empty($descriptionEl)) { + $this->content = htmlspecialchars_decode($descriptionEl->textContent, ENT_QUOTES); + } + } + + $this->link = ActivityUtils::childContent($item, ActivityUtils::LINK, self::RSS); + + // @fixme enclosures + // @fixme thumbnails... maybe + + $guidEl = ActivityUtils::child($item, self::GUID, self::RSS); + + if (!empty($guidEl)) { + $this->id = $guidEl->textContent; + + if ($guidEl->hasAttribute('isPermaLink') && $guidEl->getAttribute('isPermaLink') != 'false') { + // overwrites + $this->link = $this->id; + } + } + + $this->object = new ActivityObject($item); + $this->context = new ActivityContext($item); + } + /** * Returns an Atom based on this activity * @@ -1249,6 +1378,83 @@ class Activity return $xs->getString(); } + function _fromRssAuthor($el) + { + $text = $el->textContent; + + if (preg_match('/^(.*?) \((.*)\)$/', $text, $match)) { + $email = $match[1]; + $name = $match[2]; + } else if (preg_match('/^(.*?) <(.*)>$/', $text, $match)) { + $name = $match[1]; + $email = $match[2]; + } else if (preg_match('/.*@.*/', $text)) { + $email = $text; + $name = null; + } else { + $name = $text; + $email = null; + } + + // Not really enough info + + $actor = new ActivityObject(); + + $actor->element = $el; + + $actor->type = ActivityObject::PERSON; + $actor->title = $name; + + if (!empty($email)) { + $actor->id = 'mailto:'.$email; + } + + return $actor; + } + + function _fromDcCreator($el) + { + // Not really enough info + + $text = $el->textContent; + + $actor = new ActivityObject(); + + $actor->element = $el; + + $actor->title = $text; + $actor->type = ActivityObject::PERSON; + + return $actor; + } + + function _fromRss($el) + { + $actor = new ActivityObject(); + + $actor->element = $el; + + $actor->type = ActivityObject::PERSON; // @fixme guess better + + $actor->title = ActivityUtils::childContent($el, ActivityObject::TITLE, self::RSS); + $actor->link = ActivityUtils::childContent($el, ActivityUtils::LINK, self::RSS); + $actor->id = ActivityUtils::getLink($el, self::SELF); + + $desc = ActivityUtils::childContent($el, self::DESCRIPTION, self::RSS); + + if (!empty($desc)) { + $actor->content = htmlspecialchars_decode($desc, ENT_QUOTES); + } + + $imageEl = ActivityUtils::child($el, self::IMAGE, self::RSS); + + if (!empty($imageEl)) { + $actor->avatarLinks[] = ActivityUtils::childContent($imageEl, self::URL, self::RSS); + } + + return $actor; + } + private function _child($element, $tag, $namespace=self::SPEC) { return ActivityUtils::child($element, $tag, $namespace); diff --git a/tests/ActivityParseTests.php b/tests/ActivityParseTests.php index 7bf9cec7c4..b6980a6bb9 100644 --- a/tests/ActivityParseTests.php +++ b/tests/ActivityParseTests.php @@ -138,9 +138,38 @@ class ActivityParseTests extends PHPUnit_Framework_TestCase $this->assertEquals($poco->urls[0]->value, 'http://example.com/blog.html'); $this->assertEquals($poco->urls[0]->primary, 'true'); $this->assertEquals($act->actor->geopoint, '37.7749295 -122.4194155'); - } + public function testExample6() + { + global $_example6; + + $dom = DOMDocument::loadXML($_example6); + + $rss = $dom->documentElement; + + $channels = $dom->getElementsByTagName('channel'); + + $channel = $channels->item(0); + + $items = $channel->getElementsByTagName('item'); + + $item = $items->item(0); + + $act = new Activity($item, $channel); + + $this->assertEquals($act->verb, ActivityVerb::POST); + + $this->assertEquals($act->id, 'http://en.blog.wordpress.com/?p=3857'); + $this->assertEquals($act->link, 'http://en.blog.wordpress.com/2010/03/03/rub-a-dub-dub-in-the-pubsubhubbub/'); + $this->assertEquals($act->title, 'Rub-a-Dub-Dub in the PubSubHubbub'); + $this->assertEquals($act->time, 1267634892); + + $actor = $act->actor; + + $this->assertFalse(empty($actor)); + $this->assertEquals($actor->title, "Joseph Scott"); + } } $_example1 = << EXAMPLE5; + +$_example6 = << + + + + WordPress.com News + + http://en.blog.wordpress.com + The latest news on WordPress.com and the WordPress community. + Thu, 18 Mar 2010 23:25:35 +0000 + + http://wordpress.com/ + en + hourly + 1 + + + http://www.gravatar.com/blavatar/e6392390e3bcfadff3671c5a5653d95b?s=96&d=http://s2.wp.com/i/buttonw-com.png + WordPress.com News + http://en.blog.wordpress.com + + + +EXAMPLE6; +