diff options
Diffstat (limited to 'include/Scrape.php')
| -rw-r--r-- | include/Scrape.php | 192 |
1 files changed, 191 insertions, 1 deletions
diff --git a/include/Scrape.php b/include/Scrape.php index 002fdfb35..e5c10a9be 100644 --- a/include/Scrape.php +++ b/include/Scrape.php @@ -260,4 +260,194 @@ function scrape_feed($url) { } return $ret; -}}
\ No newline at end of file +}} + + +function probe_url($url) { + + $result = array(); + + if(! $url) + return $result; + + $diaspora = false; + $email_conversant = false; + + if($url) { + $links = lrdd($url); + + if(count($links)) { + foreach($links as $link) { + if($link['@attributes']['rel'] === NAMESPACE_DFRN) + $dfrn = unamp($link['@attributes']['href']); + if($link['@attributes']['rel'] === 'salmon') + $notify = unamp($link['@attributes']['href']); + if($link['@attributes']['rel'] === NAMESPACE_FEED) + $poll = unamp($link['@attributes']['href']); + if($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard') + $hcard = unamp($link['@attributes']['href']); + if($link['@attributes']['rel'] === 'http://webfinger.net/rel/profile-page') + $profile = unamp($link['@attributes']['href']); + if($link['@attributes']['rel'] === 'http://joindiaspora.com/seed_location') + $diaspora = true; + } + + // Status.Net can have more than one profile URL. We need to match the profile URL + // to a contact on incoming messages to prevent spam, and we won't know which one + // to match. So in case of two, one of them is stored as an alias. Only store URL's + // and not webfinger user@host aliases. If they've got more than two non-email style + // aliases, let's hope we're lucky and get one that matches the feed author-uri because + // otherwise we're screwed. + + foreach($links as $link) { + if($link['@attributes']['rel'] === 'alias') { + if(strpos($link['@attributes']['href'],'@') === false) { + if(isset($profile)) { + if($link['@attributes']['href'] !== $profile) + $alias = unamp($link['@attributes']['href']); + } + else + $profile = unamp($link['@attributes']['href']); + } + } + } + } + else { + if((strpos($orig_url,'@')) && validate_email($orig_url)) { + $email_conversant = true; + } + } + } + + if(strlen($dfrn)) { + $ret = scrape_dfrn($dfrn); + if(is_array($ret) && x($ret,'dfrn-request')) { + $network = NETWORK_DFRN; + $request = $ret['dfrn-request']; + $confirm = $ret['dfrn-confirm']; + $notify = $ret['dfrn-notify']; + $poll = $ret['dfrn-poll']; + } + } + + if($network !== NETWORK_DFRN) { + $network = NETWORK_OSTATUS; + $priority = 0; + + if($hcard) { + $vcard = scrape_vcard($hcard); + + // Google doesn't use absolute url in profile photos + + if((x($vcard,'photo')) && substr($vcard['photo'],0,1) == '/') { + $h = @parse_url($hcard); + if($h) + $vcard['photo'] = $h['scheme'] . '://' . $h['host'] . $vcard['photo']; + } + + logger('probe_url: scrape_vcard: ' . print_r($vcard,true), LOGGER_DATA); + } + + if(! $profile) { + if($diaspora) + $profile = $hcard; + else + $profile = $url; + } + + if(! x($vcard,'fn')) + if(x($vcard,'nick')) + $vcard['fn'] = $vcard['nick']; + + if((! isset($vcard)) && (! $poll)) { + + $ret = scrape_feed($url); + logger('probe_url: scrape_feed returns: ' . print_r($ret,true), LOGGER_DATA); + if(count($ret) && ($ret['feed_atom'] || $ret['feed_rss'])) { + $poll = ((x($ret,'feed_atom')) ? unamp($ret['feed_atom']) : unamp($ret['feed_rss'])); + $vcard = array(); + if(x($ret,'photo')) + $vcard['photo'] = $ret['photo']; + require_once('simplepie/simplepie.inc'); + $feed = new SimplePie(); + $xml = fetch_url($poll); + + $feed->set_raw_data($xml); + + $feed->init(); + + if(! x($vcard,'photo')) + $vcard['photo'] = $feed->get_image_url(); + $author = $feed->get_author(); + if($author) { + $vcard['fn'] = unxmlify(trim($author->get_name())); + if(! $vcard['fn']) + $vcard['fn'] = trim(unxmlify($author->get_email())); + if(strpos($vcard['fn'],'@') !== false) + $vcard['fn'] = substr($vcard['fn'],0,strpos($vcard['fn'],'@')); + $vcard['nick'] = strtolower(notags(unxmlify($vcard['fn']))); + if(strpos($vcard['nick'],' ')) + $vcard['nick'] = trim(substr($vcard['nick'],0,strpos($vcard['nick'],' '))); + $email = unxmlify($author->get_email()); + } + else { + $item = $feed->get_item(0); + if($item) { + $author = $item->get_author(); + if($author) { + $vcard['fn'] = trim(unxmlify($author->get_name())); + if(! $vcard['fn']) + $vcard['fn'] = trim(unxmlify($author->get_email())); + if(strpos($vcard['fn'],'@') !== false) + $vcard['fn'] = substr($vcard['fn'],0,strpos($vcard['fn'],'@')); + $vcard['nick'] = strtolower(unxmlify($vcard['fn'])); + if(strpos($vcard['nick'],' ')) + $vcard['nick'] = trim(substr($vcard['nick'],0,strpos($vcard['nick'],' '))); + $email = unxmlify($author->get_email()); + } + if(! $vcard['photo']) { + $rawmedia = $item->get_item_tags('http://search.yahoo.com/mrss/','thumbnail'); + if($rawmedia && $rawmedia[0]['attribs']['']['url']) + $vcard['photo'] = unxmlify($rawmedia[0]['attribs']['']['url']); + } + } + } + if((! $vcard['photo']) && strlen($email)) + $vcard['photo'] = gravatar_img($email); + if($poll === $profile) + $lnk = $feed->get_permalink(); + if(isset($lnk) && strlen($lnk)) + $profile = $lnk; + if(! (x($vcard,'fn'))) + $vcard['fn'] = notags($feed->get_title()); + if(! (x($vcard,'fn'))) + $vcard['fn'] = notags($feed->get_description()); + $network = 'feed'; + $priority = 2; + } + } + } + + if(! x($vcard,'photo')) { + $a = get_app(); + $vcard['photo'] = $a->get_baseurl() . '/images/default-profile.jpg' ; + } + $vcard['fn'] = notags($vcard['fn']); + $vcard['nick'] = notags($vcard['nick']); + + $result['name'] = $vcard['fn']; + $result['nick'] = $vcard['nick']; + $result['url'] = $profile; + $result['notify'] = $notify; + $result['poll'] = $poll; + $result['request'] = $request; + $result['confirm'] = $confirm; + $result['photo'] = $vcard['photo']; + $result['priority'] = $priority; + $result['network'] = $network; + $result['alias'] = $alias; + + logger('probe_url: ' . print_r($result,true), LOGGER_DEBUG); + + return $result; +} |