require_once('library/HTML5/Parser.php');
require_once('include/crypto.php');
+require_once('include/feed.php');
if(! function_exists('scrape_dfrn')) {
function scrape_dfrn($url, $dont_probe = false) {
logger('scrape_dfrn: url=' . $url);
+ // Try to fetch the data from noscrape. This is faster than parsing the HTML
+ $noscrape = str_replace("/hcard/", "/noscrape/", $url);
+ $noscrapejson = fetch_url($noscrape);
+ $noscrapedata = array();
+ if ($noscrapejson) {
+ $noscrapedata = json_decode($noscrapejson, true);
+
+ if (is_array($noscrapedata)) {
+ if ($noscrapedata["nick"] != "")
+ return($noscrapedata);
+ } else
+ $noscrapedata = array();
+ }
+
$s = fetch_url($url);
if(! $s)
}
}
}
-
- return $ret;
+ return array_merge($ret, $noscrapedata);
}}
$a = get_app();
$ret = array();
- $s = fetch_url($url);
+ $cookiejar = tempnam(get_temppath(), 'cookiejar-scrape-feed-');
+ $s = fetch_url($url, false, $redirects, 0, Null, $cookiejar);
+ unlink($cookiejar);
$headers = $a->get_curl_headers();
$code = $a->get_curl_code();
$network = NETWORK_TWITTER;
}
- // Twitter is deactivated since twitter closed its old API
- //$twitter = ((strpos($url,'twitter.com') !== false) ? true : false);
$lastfm = ((strpos($url,'last.fm/user') !== false) ? true : false);
$at_addr = ((strpos($url,'@') !== false) ? true : false);
}
}
+ // Scrape the public key from the hcard.
+ // Diaspora will remove it from the webfinger somewhere in the future.
+ if (($hcard != "") AND ($pubkey == "")) {
+ $ret = scrape_dfrn(($hcard) ? $hcard : $dfrn, true);
+ if (isset($ret["key"])) {
+ $hcard_key = $ret["key"];
+ if(strstr($hcard_key,'RSA '))
+ $pubkey = rsatopem($hcard_key);
+ else
+ $pubkey = $hcard_key;
+ }
+ }
if($diaspora && $diaspora_base && $diaspora_guid) {
- if($mode == PROBE_DIASPORA || ! $notify) {
- $notify = $diaspora_base . 'receive/users/' . $diaspora_guid;
+ $diaspora_notify = $diaspora_base.'receive/users/'.$diaspora_guid;
+
+ if($mode == PROBE_DIASPORA || ! $notify || ($notify == $diaspora_notify)) {
+ $notify = $diaspora_notify;
$batch = $diaspora_base . 'receive/public' ;
}
if(strpos($url,'@'))
$vcard['nick'] = $addr_parts[0];
}
- /* if($twitter) {
- logger('twitter: setup');
- $tid = basename($url);
- $tapi = 'https://api.twitter.com/1/statuses/user_timeline.rss';
- if(intval($tid))
- $poll = $tapi . '?user_id=' . $tid;
- else
- $poll = $tapi . '?screen_name=' . $tid;
- $profile = 'http://twitter.com/#!/' . $tid;
- //$vcard['photo'] = 'https://api.twitter.com/1/users/profile_image/' . $tid;
- $vcard['photo'] = 'https://api.twitter.com/1/users/profile_image?screen_name=' . $tid . '&size=bigger';
- $vcard['nick'] = $tid;
- $vcard['fn'] = $tid;
- } */
-
if($lastfm) {
$profile = $url;
$poll = str_replace(array('www.','last.fm/'),array('','ws.audioscrobbler.com/1.0/'),$url) . '/recenttracks.rss';
if ($connectornetworks)
$check_feed = false;
+
if($check_feed) {
+
$feedret = scrape_feed(($poll) ? $poll : $url);
logger('probe_url: scrape_feed ' . (($poll)? $poll : $url) . ' returns: ' . print_r($feedret,true), LOGGER_DATA);
if(x($feedret,'photo') && (! x($vcard,'photo')))
$vcard['photo'] = $feedret['photo'];
- require_once('library/simplepie/simplepie.inc');
- $feed = new SimplePie();
- $xml = fetch_url($poll);
+
+ $cookiejar = tempnam(get_temppath(), 'cookiejar-scrape-feed-');
+ $xml = fetch_url($poll, false, $redirects, 0, Null, $cookiejar);
+ unlink($cookiejar);
logger('probe_url: fetch feed: ' . $poll . ' returns: ' . $xml, LOGGER_DATA);
- $a = get_app();
- logger('probe_url: scrape_feed: headers: ' . $a->get_curl_headers(), LOGGER_DATA);
+ if ($xml == "") {
+ logger("scrape_feed: XML is empty for feed ".$poll);
+ $network = NETWORK_PHANTOM;
+ } else {
+ $data = feed_import($xml,$dummy1,$dummy2, $dummy3, true);
- // Don't try and parse an empty string
- $feed->set_raw_data(($xml) ? $xml : '<?xml version="1.0" encoding="utf-8" ?><xml></xml>');
+ if (!is_array($data)) {
+ logger("scrape_feed: This doesn't seem to be a feed: ".$poll);
+ $network = NETWORK_PHANTOM;
+ } else {
+ if (($vcard["photo"] == "") AND ($data["header"]["author-avatar"] != ""))
+ $vcard["photo"] = $data["header"]["author-avatar"];
- $feed->init();
- if($feed->error()) {
- logger('probe_url: scrape_feed: Error parsing XML: ' . $feed->error());
- $network = NETWORK_PHANTOM;
- }
+ if (($vcard["fn"] == "") AND ($data["header"]["author-name"] != ""))
+ $vcard["fn"] = $data["header"]["author-name"];
- if(! x($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'],'@'));
-
- $email = unxmlify($author->get_email());
- if(! $profile && $author->get_link())
- $profile = trim(unxmlify($author->get_link()));
- if(! $vcard['photo']) {
- $rawtags = $feed->get_feed_tags( SIMPLEPIE_NAMESPACE_ATOM_10, 'author');
- if($rawtags) {
- $elems = $rawtags[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10];
- if((x($elems,'link')) && ($elems['link'][0]['attribs']['']['rel'] === 'photo'))
- $vcard['photo'] = $elems['link'][0]['attribs']['']['href'];
- }
- }
- // Fetch fullname via poco:displayName
- $pocotags = $feed->get_feed_tags(SIMPLEPIE_NAMESPACE_ATOM_10, 'author');
- if ($pocotags) {
- $elems = $pocotags[0]['child']['http://portablecontacts.net/spec/1.0'];
- if (isset($elems["displayName"]))
- $vcard['fn'] = $elems["displayName"][0]["data"];
- if (isset($elems["preferredUsername"]))
- $vcard['nick'] = $elems["preferredUsername"][0]["data"];
- }
- }
- 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'],'@'));
- $email = unxmlify($author->get_email());
- if(! $profile && $author->get_link())
- $profile = trim(unxmlify($author->get_link()));
- }
- 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']) {
- $rawtags = $item->get_item_tags( SIMPLEPIE_NAMESPACE_ATOM_10, 'author');
- if($rawtags) {
- $elems = $rawtags[0]['child'][SIMPLEPIE_NAMESPACE_ATOM_10];
- if((x($elems,'link')) && ($elems['link'][0]['attribs']['']['rel'] === 'photo'))
- $vcard['photo'] = $elems['link'][0]['attribs']['']['href'];
- }
- }
+ if (($vcard["nick"] == "") AND ($data["header"]["author-nick"] != ""))
+ $vcard["nick"] = $data["header"]["author-nick"];
+
+ if(!$profile AND ($data["header"]["author-link"] != "") AND !in_array($network, array("", NETWORK_FEED)))
+ $profile = $data["header"]["author-link"];
}
}
if (isset($noscrapedata["dfrn-poll"]))
$poll = $noscrapedata["dfrn-poll"];
-// print_r($noscrapedata);
}
}
- if((! $vcard['photo']) && strlen($email))
- $vcard['photo'] = avatar_img($email);
- if($poll === $profile)
- $lnk = $feed->get_permalink();
- if(isset($lnk) && strlen($lnk))
- $profile = $lnk;
-
- if(! $network) {
+ if(! $network)
$network = NETWORK_FEED;
- // If it is a feed, don't take the author name as feed name
- unset($vcard['fn']);
- }
- if(! (x($vcard,'fn')))
- $vcard['fn'] = notags($feed->get_title());
- if(! (x($vcard,'fn')))
- $vcard['fn'] = notags($feed->get_description());
-
- if(strpos($vcard['fn'],'Twitter / ') !== false) {
- $vcard['fn'] = substr($vcard['fn'],strpos($vcard['fn'],'/')+1);
- $vcard['fn'] = trim($vcard['fn']);
- }
+
if(! x($vcard,'nick')) {
$vcard['nick'] = strtolower(notags(unxmlify($vcard['fn'])));
if(strpos($vcard['nick'],' '))
if(! x($vcard,'photo')) {
$a = get_app();
- $vcard['photo'] = $a->get_baseurl() . '/images/person-175.jpg' ;
+ $vcard['photo'] = App::get_baseurl() . '/images/person-175.jpg' ;
}
if(! $profile)
$vcard['fn'] = $url;
if (($notify != "") AND ($poll != "")) {
- $baseurl = matching(normalise_link($notify), normalise_link($poll));
+ $baseurl = matching_url(normalise_link($notify), normalise_link($poll));
- $baseurl2 = matching($baseurl, normalise_link($profile));
+ $baseurl2 = matching_url($baseurl, normalise_link($profile));
if ($baseurl2 != "")
$baseurl = $baseurl2;
}
if (($baseurl == "") AND ($notify != ""))
- $baseurl = matching(normalise_link($profile), normalise_link($notify));
+ $baseurl = matching_url(normalise_link($profile), normalise_link($notify));
if (($baseurl == "") AND ($poll != ""))
- $baseurl = matching(normalise_link($profile), normalise_link($poll));
+ $baseurl = matching_url(normalise_link($profile), normalise_link($poll));
$baseurl = rtrim($baseurl, "/");
return $result;
}
-function matching($part1, $part2) {
- $len = min(strlen($part1), strlen($part2));
+/**
+ * @brief Find the matching part between two url
+ *
+ * @param string $url1
+ * @param string $url2
+ * @return string The matching part
+ */
+function matching_url($url1, $url2) {
+
+ if (($url1 == "") OR ($url2 == ""))
+ return "";
+
+ $url1 = normalise_link($url1);
+ $url2 = normalise_link($url2);
+
+ $parts1 = parse_url($url1);
+ $parts2 = parse_url($url2);
+
+ if (!isset($parts1["host"]) OR !isset($parts2["host"]))
+ return "";
+
+ if ($parts1["scheme"] != $parts2["scheme"])
+ return "";
+
+ if ($parts1["host"] != $parts2["host"])
+ return "";
+
+ if ($parts1["port"] != $parts2["port"])
+ return "";
+
+ $match = $parts1["scheme"]."://".$parts1["host"];
+
+ if ($parts1["port"])
+ $match .= ":".$parts1["port"];
+
+ $pathparts1 = explode("/", $parts1["path"]);
+ $pathparts2 = explode("/", $parts2["path"]);
- $match = "";
- $matching = true;
$i = 0;
- while (($i <= $len) AND $matching) {
- if (substr($part1, $i, 1) == substr($part2, $i, 1))
- $match .= substr($part1, $i, 1);
- else
- $matching = false;
+ $path = "";
+ do {
+ $path1 = $pathparts1[$i];
+ $path2 = $pathparts2[$i];
- $i++;
- }
- return($match);
+ if ($path1 == $path2)
+ $path .= $path1."/";
+
+ } while (($path1 == $path2) AND ($i++ <= count($pathparts1)));
+
+ $match .= $path;
+
+ return normalise_link($match);
}