Change the workflow to get better discovery

Tried to re-structure the workflow of discovery to get more and richer
data and hints.

plugins/OStatus/actions/ostatussub.php

@@ -149,7 +149,7 @@ class OStatusSubAction extends Action
         $fullname = $entity->fullname;
         $homepage = $entity->homepage;
         $location = $entity->location;
-        
+
         if (!$avatar) {
             $avatar = Avatar::defaultImage(AVATAR_PROFILE_SIZE);
         }
@@ -242,7 +242,7 @@ class OStatusSubAction extends Action
             if (Validate::email($this->profile_uri)) {
                 $this->oprofile = Ostatus_profile::ensureWebfinger($this->profile_uri);
             } else if (Validate::uri($this->profile_uri)) {
-                $this->oprofile = Ostatus_profile::ensureProfile($this->profile_uri);
+                $this->oprofile = Ostatus_profile::ensureProfileURL($this->profile_uri);
             } else {
                 $this->error = _m("Sorry, we could not reach that address. Please make sure that the OStatus address is like nickname@example.com or http://example.net/nickname");
                 common_debug('Invalid address format.', __FILE__);
@@ -339,7 +339,6 @@ class OStatusSubAction extends Action
         }
     }
 
-
     /**
      * Handle posts to this form
      *

plugins/OStatus/classes/Ostatus_profile.php

@@ -708,18 +708,122 @@ class Ostatus_profile extends Memcached_DataObject
      * @return Ostatus_profile
      * @throws FeedSubException
      */
-    public static function ensureProfile($profile_uri, $hints=array())
+
+    public static function ensureProfileURL($profile_url, $hints=array())
     {
-        // Get the canonical feed URI and check it
+        $oprofile = self::getFromProfileURL($profile_url);
-        $discover = new FeedDiscovery();
+
-        if (isset($hints['feedurl'])) {
+        if (!empty($oprofile)) {
-            $feeduri = $hints['feedurl'];
+            return $oprofile;
-            $feeduri = $discover->discoverFromFeedURL($feeduri);
-        } else {
-            $feeduri = $discover->discoverFromURL($profile_uri);
-            $hints['feedurl'] = $feeduri;
         }
 
+        $hints['profileurl'] = $profile_url;
+
+        // Fetch the URL
+        // XXX: HTTP caching
+
+        $client = new HTTPClient();
+        $client->setHeader('Accept', 'text/html,application/xhtml+xml');
+        $response = $client->get($profile_url);
+
+        if (!$response->isOk()) {
+            return null;
+        }
+
+        // Check if we have a non-canonical URL
+
+        $finalUrl = $response->getUrl();
+
+        if ($finalUrl != $profile_url) {
+
+            $hints['profileurl'] = $finalUrl;
+
+            $oprofile = self::getFromProfileURL($finalUrl);
+
+            if (!empty($oprofile)) {
+                return $oprofile;
+            }
+        }
+
+        // Try to get some hCard data
+
+        $body = $response->getBody();
+
+        $hcardHints = DiscoveryHints::hcardHints($body, $finalUrl);
+
+        if (!empty($hcardHints)) {
+            $hints = array_merge($hints, $hcardHints);
+        }
+
+        // Check if they've got an LRDD header
+
+        $lrdd = LinkHeader::getLink($response, 'lrdd', 'application/xrd+xml');
+
+        if (!empty($lrdd)) {
+
+            $xrd = Discovery::fetchXrd($lrdd);
+            $xrdHints = DiscoveryHints::fromXRD($xrd);
+
+            $hints = array_merge($hints, $xrdHints);
+        }
+
+        // If discovery found a feedurl (probably from LRDD), use it.
+
+        if (array_key_exists('feedurl', $hints)) {
+            return self::ensureFeedURL($hints['feedurl'], $hints);
+        }
+
+        // Get the feed URL from HTML
+
+        $discover = new FeedDiscovery();
+
+        $feedurl = $discover->discoverFromHTML($finalUrl, $body);
+
+        if (!empty($feedurl)) {
+            $hints['feedurl'] = $feedurl;
+
+            return self::ensureFeedURL($feedurl, $hints);
+        }
+    }
+
+    static function getFromProfileURL($profile_url)
+    {
+        $profile = Profile::staticGet('profileurl', $profile_url);
+
+        if (empty($profile)) {
+            return null;
+        }
+
+        // Is it a known Ostatus profile?
+
+        $oprofile = Ostatus_profile::staticGet('profile_id', $profile->id);
+
+        if (!empty($oprofile)) {
+            return $oprofile;
+        }
+
+        // Is it a local user?
+
+        $user = User::staticGet('id', $profile->id);
+
+        if (!empty($user)) {
+            throw new Exception("'$profile_url' is the profile for local user '{$user->nickname}'.");
+        }
+
+        // Continue discovery; it's a remote profile
+        // for OMB or some other protocol, may also
+        // support OStatus
+
+        return null;
+    }
+
+    public static function ensureFeedURL($feed_url, $hints=array())
+    {
+        $discover = new FeedDiscovery();
+
+        $feeduri = $discover->discoverFromFeedURL($feed_url);
+        $hints['feedurl'] = $feeduri;
+
         $huburi = $discover->getAtomLink('hub');
         $hints['hub'] = $huburi;
         $salmonuri = $discover->getAtomLink(Salmon::NS_REPLIES);
@@ -1303,7 +1407,7 @@ class Ostatus_profile extends Memcached_DataObject
             }
         }
 
-        // First, look it up
+        // Try looking it up
 
         $oprofile = Ostatus_profile::staticGet('uri', 'acct:'.$addr);
 
@@ -1317,7 +1421,7 @@ class Ostatus_profile extends Memcached_DataObject
         $disco = new Discovery();
 
         try {
-            $result = $disco->lookup($addr);
+            $xrd = $disco->lookup($addr);
         } catch (Exception $e) {
             // Save negative cache entry so we don't waste time looking it up again.
             // @fixme distinguish temporary failures?
@@ -1327,38 +1431,26 @@ class Ostatus_profile extends Memcached_DataObject
 
         $hints = array('webfinger' => $addr);
 
-        foreach ($result->links as $link) {
+        $dhints = DiscoveryHints::fromXRD($xrd);
-            switch ($link['rel']) {
-            case Discovery::PROFILEPAGE:
-                $hints['profileurl'] = $profileUrl = $link['href'];
-                break;
-            case Salmon::NS_REPLIES:
-                $hints['salmon'] = $salmonEndpoint = $link['href'];
-                break;
-            case Discovery::UPDATESFROM:
-                $hints['feedurl'] = $feedUrl = $link['href'];
-                break;
-            case Discovery::HCARD:
-                $hcardUrl = $link['href'];
-                break;
-            default:
-                common_log(LOG_NOTICE, "Don't know what to do with rel = '{$link['rel']}'");
-                break;
-            }
-        }
 
-        if (isset($hcardUrl)) {
+        $hints = array_merge($hints, $dhints);
-            $hcardHints = self::slurpHcard($hcardUrl);
+
-            // Note: Webfinger > hcard
+        // If there's an Hcard, let's grab its info
-            $hints = array_merge($hcardHints, $hints);
+
+        if (array_key_exists('hcard', $hints)) {
+            if (!array_key_exists('profileurl', $hints) ||
+                $hints['hcard'] != $hints['profileurl']) {
+                $hcardHints = DiscoveryHints::fromHcardUrl($hints['hcard']);
+                $hints = array_merge($hcardHints, $hints);
+            }
         }
 
         // If we got a feed URL, try that
 
-        if (isset($feedUrl)) {
+        if (array_key_exists('feedurl', $hints)) {
             try {
                 common_log(LOG_INFO, "Discovery on acct:$addr with feed URL $feedUrl");
-                $oprofile = self::ensureProfile($feedUrl, $hints);
+                $oprofile = self::ensureFeedURL($hints['feedurl'], $hints);
                 self::cacheSet(sprintf('ostatus_profile:webfinger:%s', $addr), $oprofile->uri);
                 return $oprofile;
             } catch (Exception $e) {
@@ -1369,10 +1461,10 @@ class Ostatus_profile extends Memcached_DataObject
 
         // If we got a profile page, try that!
 
-        if (isset($profileUrl)) {
+        if (array_key_exists('profileurl', $hints)) {
             try {
                 common_log(LOG_INFO, "Discovery on acct:$addr with profile URL $profileUrl");
-                $oprofile = self::ensureProfile($profileUrl, $hints);
+                $oprofile = self::ensureProfile($hints['profileurl'], $hints);
                 self::cacheSet(sprintf('ostatus_profile:webfinger:%s', $addr), $oprofile->uri);
                 return $oprofile;
             } catch (Exception $e) {
@@ -1384,7 +1476,9 @@ class Ostatus_profile extends Memcached_DataObject
         // XXX: try hcard
         // XXX: try FOAF
 
-        if (isset($salmonEndpoint)) {
+        if (array_key_exists('salmon', $hints)) {
+
+            $salmonEndpoint = $hints['salmon'];
 
             // An account URL, a salmon endpoint, and a dream? Not much to go
             // on, but let's give it a try
@@ -1464,67 +1558,4 @@ class Ostatus_profile extends Memcached_DataObject
 
         return $file;
     }
-
-    protected static function slurpHcard($url)
-    {
-        set_include_path(get_include_path() . PATH_SEPARATOR . INSTALLDIR . '/plugins/OStatus/extlib/hkit/');
-        require_once('hkit.class.php');
-
-        $h	= new hKit;
-
-        // Google Buzz hcards need to be tidied. Probably others too.
-
-        $h->tidy_mode = 'proxy'; // 'proxy', 'exec', 'php' or 'none'
-
-        // Get by URL
-        $hcards = $h->getByURL('hcard', $url);
-
-        if (empty($hcards)) {
-            return array();
-        }
-
-        // @fixme more intelligent guess on multi-hcard pages
-        $hcard = $hcards[0];
-
-        $hints = array();
-
-        $hints['profileurl'] = $url;
-
-        if (array_key_exists('nickname', $hcard)) {
-            $hints['nickname'] = $hcard['nickname'];
-        }
-
-        if (array_key_exists('fn', $hcard)) {
-            $hints['fullname'] = $hcard['fn'];
-        } else if (array_key_exists('n', $hcard)) {
-            $hints['fullname'] = implode(' ', $hcard['n']);
-        }
-
-        if (array_key_exists('photo', $hcard)) {
-            $hints['avatar'] = $hcard['photo'];
-        }
-
-        if (array_key_exists('note', $hcard)) {
-            $hints['bio'] = $hcard['note'];
-        }
-
-        if (array_key_exists('adr', $hcard)) {
-            if (is_string($hcard['adr'])) {
-                $hints['location'] = $hcard['adr'];
-            } else if (is_array($hcard['adr'])) {
-                $hints['location'] = implode(' ', $hcard['adr']);
-            }
-        }
-
-        if (array_key_exists('url', $hcard)) {
-            if (is_string($hcard['url'])) {
-                $hints['homepage'] = $hcard['url'];
-            } else if (is_array($hcard['url'])) {
-                // HACK get the last one; that's how our hcards look
-                $hints['homepage'] = $hcard['url'][count($hcard['url'])-1];
-            }
-        }
-
-        return $hints;
-    }
 }

plugins/OStatus/lib/discovery.php

@@ -40,7 +40,7 @@ class Discovery
     const PROFILEPAGE = 'http://webfinger.net/rel/profile-page';
     const UPDATESFROM = 'http://schemas.google.com/g/2010#updates-from';
     const HCARD = 'http://microformats.org/profile/hcard';
-    
+
     public $methods = array();
 
     public function __construct()
@@ -50,12 +50,11 @@ class Discovery
         $this->registerMethod('Discovery_LRDD_Link_HTML');
     }
 
-
     public function registerMethod($class)
     {
         $this->methods[] = $class;
     }
-    
+
     /**
      * Given a "user id" make sure it's normalized to either a webfinger
      * acct: uri or a profile HTTP URL.
@@ -78,7 +77,7 @@ class Discovery
     public static function isWebfinger($user_id)
     {
         $uri = Discovery::normalize($user_id);
-        
+
         return (substr($uri, 0, 5) == 'acct:');
     }
 
@@ -99,7 +98,7 @@ class Discovery
                 } else {
                     $xrd_uri = $link['href'];
                 }
-                
+
                 $xrd = $this->fetchXrd($xrd_uri);
                 if ($xrd) {
                     return $xrd;
@@ -114,14 +113,13 @@ class Discovery
         if (!is_array($links)) {
             return false;
         }
-        
+
         foreach ($links as $link) {
             if ($link['rel'] == $service) {
                 return $link;
             }
         }
     }
-    
 
     public static function applyTemplate($template, $id)
     {
@@ -130,7 +128,6 @@ class Discovery
         return $template;
     }
 
-    
     public static function fetchXrd($url)
     {
         try {
@@ -171,7 +168,7 @@ class Discovery_LRDD_Host_Meta implements Discovery_LRDD
             if ($xrd->host != $domain) {
                 return false;
             }
-            
+
             return $xrd->links;
         }
     }
@@ -187,7 +184,7 @@ class Discovery_LRDD_Link_Header implements Discovery_LRDD
         } catch (HTTP_Request2_Exception $e) {
             return false;
         }
-             
+
         if ($response->getStatus() != 200) {
             return false;
         }
@@ -196,51 +193,17 @@ class Discovery_LRDD_Link_Header implements Discovery_LRDD
         if (!$link_header) {
             //            return false;
         }
-        
+
         return Discovery_LRDD_Link_Header::parseHeader($link_header);
     }
 
     protected static function parseHeader($header)
     {
-        preg_match('/^<[^>]+>/', $header, $uri_reference);
+        $lh = new LinkHeader($header);
-        //if (empty($uri_reference)) return;
 
-        $links = array();
+        return array('href' => $lh->href,
-        
+                     'rel'  => $lh->rel,
-        $link_uri = trim($uri_reference[0], '<>');
+                     'type' => $lh->type);
-        $link_rel = array();
-        $link_type = null;
-        
-        // remove uri-reference from header
-        $header = substr($header, strlen($uri_reference[0]));
-        
-        // parse link-params
-        $params = explode(';', $header);
-        
-        foreach ($params as $param) {
-            if (empty($param)) continue;
-            list($param_name, $param_value) = explode('=', $param, 2);
-            $param_name = trim($param_name);
-            $param_value = preg_replace('(^"|"$)', '', trim($param_value));
-            
-            // for now we only care about 'rel' and 'type' link params
-            // TODO do something with the other links-params
-            switch ($param_name) {
-            case 'rel':
-                $link_rel = trim($param_value);
-                break;
-                
-            case 'type':
-                $link_type = trim($param_value);
-            }
-        }
-
-        $links[] =  array(
-            'href' => $link_uri,
-            'rel' => $link_rel,
-            'type' => $link_type);
-
-        return $links;
     }
 }
 
@@ -262,49 +225,48 @@ class Discovery_LRDD_Link_HTML implements Discovery_LRDD
         return Discovery_LRDD_Link_HTML::parse($response->getBody());
     }
 
-
     public function parse($html)
     {
         $links = array();
-        
+
         preg_match('/<head(\s[^>]*)?>(.*?)<\/head>/is', $html, $head_matches);
         $head_html = $head_matches[2];
-        
+
         preg_match_all('/<link\s[^>]*>/i', $head_html, $link_matches);
-        
+
         foreach ($link_matches[0] as $link_html) {
             $link_url = null;
             $link_rel = null;
             $link_type = null;
-            
+
             preg_match('/\srel=(("|\')([^\\2]*?)\\2|[^"\'\s]+)/i', $link_html, $rel_matches);
             if ( isset($rel_matches[3]) ) {
                 $link_rel = $rel_matches[3];
             } else if ( isset($rel_matches[1]) ) {
                 $link_rel = $rel_matches[1];
             }
-            
+
             preg_match('/\shref=(("|\')([^\\2]*?)\\2|[^"\'\s]+)/i', $link_html, $href_matches);
             if ( isset($href_matches[3]) ) {
                 $link_uri = $href_matches[3];
             } else if ( isset($href_matches[1]) ) {
                 $link_uri = $href_matches[1];
             }
-            
+
             preg_match('/\stype=(("|\')([^\\2]*?)\\2|[^"\'\s]+)/i', $link_html, $type_matches);
             if ( isset($type_matches[3]) ) {
                 $link_type = $type_matches[3];
             } else if ( isset($type_matches[1]) ) {
                 $link_type = $type_matches[1];
             }
-            
+
             $links[] = array(
                 'href' => $link_url,
                 'rel' => $link_rel,
                 'type' => $link_type,
             );
         }
-        
+
         return $links;
     }
 }

plugins/OStatus/lib/discoveryhints.php

@@ -0,0 +1,182 @@
+<?php
+/*
+ * StatusNet - the distributed open-source microblogging tool
+ * Copyright (C) 2010, StatusNet, Inc.
+ *
+ * Some utilities for generating hint data
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU Affero General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU Affero General Public License for more details.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+class DiscoveryHints {
+
+    static function fromXRD($xrd)
+    {
+        $hints = array();
+
+        foreach ($xrd->links as $link) {
+            switch ($link['rel']) {
+            case Discovery::PROFILEPAGE:
+                $hints['profileurl'] = $link['href'];
+                break;
+            case Salmon::NS_REPLIES:
+                $hints['salmon'] = $link['href'];
+                break;
+            case Discovery::UPDATESFROM:
+                $hints['feedurl'] = $link['href'];
+                break;
+            case Discovery::HCARD:
+                $hints['hcardurl'] = $link['href'];
+                break;
+            default:
+                break;
+            }
+        }
+
+        return $hints;
+    }
+
+    static function fromHcardUrl($url)
+    {
+        $client = new HTTPClient();
+        $client->setHeader('Accept', 'text/html,application/xhtml+xml');
+        $response = $client->get($url);
+
+        if (!$response->isOk()) {
+            return null;
+        }
+
+        return self::hcardHints($response->getBody(),
+                                $response->getUrl());
+    }
+
+    static function hcardHints($body, $url)
+    {
+        common_debug("starting tidy");
+
+        $body = self::_tidy($body);
+
+        common_debug("done with tidy");
+
+        set_include_path(get_include_path() . PATH_SEPARATOR . INSTALLDIR . '/plugins/OStatus/extlib/hkit/');
+        require_once('hkit.class.php');
+
+        $h	= new hKit;
+
+        $hcards = $h->getByString('hcard', $body);
+
+        if (empty($hcards)) {
+            return array();
+        }
+
+        if (count($hcards) == 1) {
+            $hcard = $hcards[0];
+        } else {
+            foreach ($hcards as $try) {
+                if (array_key_exists('url', $try)) {
+                    if (is_string($try['url']) && $try['url'] == $url) {
+                        $hcard = $try;
+                        break;
+                    } else if (is_array($try['url'])) {
+                        foreach ($try['url'] as $tryurl) {
+                            if ($tryurl == $url) {
+                                $hcard = $try;
+                                break 2;
+                            }
+                        }
+                    }
+                }
+            }
+            // last chance; grab the first one
+            if (empty($hcard)) {
+                $hcard = $hcards[0];
+            }
+        }
+
+        $hints = array();
+
+        if (array_key_exists('nickname', $hcard)) {
+            $hints['nickname'] = $hcard['nickname'];
+        }
+
+        if (array_key_exists('fn', $hcard)) {
+            $hints['fullname'] = $hcard['fn'];
+        } else if (array_key_exists('n', $hcard)) {
+            $hints['fullname'] = implode(' ', $hcard['n']);
+        }
+
+        if (array_key_exists('photo', $hcard)) {
+            $hints['avatar'] = $hcard['photo'];
+        }
+
+        if (array_key_exists('note', $hcard)) {
+            $hints['bio'] = $hcard['note'];
+        }
+
+        if (array_key_exists('adr', $hcard)) {
+            if (is_string($hcard['adr'])) {
+                $hints['location'] = $hcard['adr'];
+            } else if (is_array($hcard['adr'])) {
+                $hints['location'] = implode(' ', $hcard['adr']);
+            }
+        }
+
+        if (array_key_exists('url', $hcard)) {
+            if (is_string($hcard['url'])) {
+                $hints['homepage'] = $hcard['url'];
+            } else if (is_array($hcard['url'])) {
+                // HACK get the last one; that's how our hcards look
+                $hints['homepage'] = $hcard['url'][count($hcard['url'])-1];
+            }
+        }
+
+        return $hints;
+    }
+
+    private static function _tidy($body)
+    {
+        if (function_exists('tidy_parse_string')) {
+            common_debug("Tidying with extension");
+            $text = tidy_parse_string($body);
+            $text = tidy_clean_repair($text);
+            return $body;
+        } else if ($fullpath = self::_findProgram('tidy')) {
+            common_debug("Tidying with program $fullpath");
+            $tempfile = tempnam('/tmp', 'snht'); // statusnet hcard tidy
+            file_put_contents($tempfile, $source);
+            exec("$fullpath -utf8 -indent -asxhtml -numeric -bare -quiet $tempfile", $tidy);
+            unlink($tempfile);
+            return implode("\n", $tidy);
+        } else {
+            common_debug("Not tidying.");
+            return $body;
+        }
+    }
+
+    private static function _findProgram($name)
+    {
+        $path = $_ENV['PATH'];
+
+        $parts = explode(':', $path);
+
+        foreach ($parts as $part) {
+            $fullpath = $part . '/' . $name;
+            if (is_executable($fullpath)) {
+                return $fullpath;
+            }
+        }
+
+        return null;
+    }
+}

plugins/OStatus/lib/feeddiscovery.php

@@ -117,7 +117,7 @@ class FeedDiscovery
                 return $this->discoverFromURL($target, false);
             }
         }
-        
+
         return $this->initFromResponse($response);
     }
 
@@ -202,7 +202,7 @@ class FeedDiscovery
             'application/atom+xml' => false,
             'application/rss+xml' => false,
         );
-        
+
         $nodes = $dom->getElementsByTagName('link');
         for ($i = 0; $i < $nodes->length; $i++) {
             $node = $nodes->item($i);

plugins/OStatus/lib/linkheader.php

@@ -0,0 +1,63 @@
+<?php
+
+class LinkHeader
+{
+    var $href;
+    var $rel;
+    var $type;
+
+    function __construct($str)
+    {
+        preg_match('/^<[^>]+>/', $str, $uri_reference);
+        //if (empty($uri_reference)) return;
+
+        $this->uri = trim($uri_reference[0], '<>');
+        $this->rel = array();
+        $this->type = null;
+
+        // remove uri-reference from header
+        $str = substr($str, strlen($uri_reference[0]));
+
+        // parse link-params
+        $params = explode(';', $str);
+
+        foreach ($params as $param) {
+            if (empty($param)) continue;
+            list($param_name, $param_value) = explode('=', $param, 2);
+            $param_name = trim($param_name);
+            $param_value = preg_replace('(^"|"$)', '', trim($param_value));
+
+            // for now we only care about 'rel' and 'type' link params
+            // TODO do something with the other links-params
+            switch ($param_name) {
+            case 'rel':
+                $this->rel = trim($param_value);
+                break;
+
+            case 'type':
+                $this->type = trim($param_value);
+            }
+        }
+    }
+
+    static function getLink($response, $rel=null, $type=null)
+    {
+        $headers = $response->getHeader('Link');
+
+        // Can get an array or string, so try to simplify the path
+        if (!is_array($headers)) {
+            $headers = array($headers);
+        }
+
+        foreach ($headers as $header) {
+            $lh = new LinkHeader($header);
+
+            if ((is_null($rel) || $lh->rel == $rel) &&
+                (is_null($type) || $lh->type == $type)) {
+                return $lh->href;
+            }
+        }
+
+        return null;
+    }
+}