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($noscrapedata);
+ else
+ unset($noscrapedata["nick"]);
+ } else
+ $noscrapedata = array();
+ }
+
$s = fetch_url($url);
- if(! $s)
+ if (!$s)
return $ret;
if (!$dont_probe) {
}
}
}
-
- return $ret;
+ return array_merge($ret, $noscrapedata);
}}
$result = array();
- if(! $url)
+ if (!$url)
return $result;
$result = Cache::get("probe_url:".$mode.":".$url);
return $result;
}
+ $original_url = $url;
$network = null;
$diaspora = false;
$diaspora_base = '';
else
$links = lrdd($url);
- if(count($links)) {
+ if ((count($links) == 0) AND strstr($url, "/index.php")) {
+ $url = str_replace("/index.php", "", $url);
+ $links = lrdd($url);
+ }
+
+ if (count($links)) {
$has_lrdd = true;
logger('probe_url: found lrdd links: ' . print_r($links,true), LOGGER_DATA);
// aliases, let's hope we're lucky and get one that matches the feed author-uri because
// otherwise we're screwed.
+ $backup_alias = "";
+
foreach($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']);
+ $alias_url = $link['@attributes']['href'];
+
+ if(($alias_url !== $profile) AND ($backup_alias == "") AND
+ ($alias_url !== str_replace("/index.php", "", $profile)))
+ $backup_alias = $alias_url;
+
+ if(($alias_url !== $profile) AND !strstr($alias_url, "index.php") AND
+ ($alias_url !== str_replace("/index.php", "", $profile)))
+ $alias = $alias_url;
}
else
$profile = unamp($link['@attributes']['href']);
}
}
+ if ($alias == "")
+ $alias = $backup_alias;
+
// If the profile is different from the url then the url is abviously an alias
if (($alias == "") AND ($profile != "") AND !$at_addr AND (normalise_link($profile) != normalise_link($url)))
$alias = $url;
if($diaspora && $diaspora_base && $diaspora_guid) {
$diaspora_notify = $diaspora_base.'receive/users/'.$diaspora_guid;
- if($mode == PROBE_DIASPORA || ! $notify || ($notify == $diaspora_notify)) {
+ if($mode == PROBE_DIASPORA || !$notify || ($notify == $diaspora_notify)) {
$notify = $diaspora_notify;
$batch = $diaspora_base . 'receive/public' ;
}
if (($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)))
+ if ($network == NETWORK_OSTATUS) {
+ if ($data["header"]["author-id"] != "")
+ $alias = $data["header"]["author-id"];
+
+ if ($data["header"]["author-link"] != "")
+ $profile = $data["header"]["author-link"];
+
+ } elseif(!$profile AND ($data["header"]["author-link"] != "") AND !in_array($network, array("", NETWORK_FEED)))
$profile = $data["header"]["author-link"];
}
}
$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($poll));
+
+ if (substr($baseurl, -10) == "/index.php")
+ $baseurl = str_replace("/index.php", "", $baseurl);
+
+ if ($network == "")
+ $network = NETWORK_PHANTOM;
$baseurl = rtrim($baseurl, "/");
require_once('include/bbcode.php');
$address = GetProfileUsername($url, "", true);
$result2 = probe_url($address, $mode, ++$level);
- if ($result2['network'] != "")
+ if (!in_array($result2['network'], array("", NETWORK_PHANTOM, NETWORK_FEED)))
$result = $result2;
}
if (($result['network'] == NETWORK_FEED) AND ($result['baseurl'] != "") AND ($result['nick'] != "")) {
$addr = $result['nick'].'@'.str_replace("http://", "", $result['baseurl']);
$result2 = probe_url($addr, $mode, ++$level);
- if (($result2['network'] != "") AND ($result2['network'] != NETWORK_FEED))
+ if (!in_array($result2['network'], array("", NETWORK_PHANTOM, NETWORK_FEED)))
$result = $result2;
}
+
+ // Quickfix for Hubzilla systems with enabled OStatus plugin
+ if (($result['network'] == NETWORK_DIASPORA) AND ($result["batch"] == "")) {
+ $result2 = probe_url($url, PROBE_DIASPORA, ++$level);
+ if ($result2['network'] == NETWORK_DIASPORA) {
+ $addr = $result["addr"];
+ $result = $result2;
+
+ if (($result["addr"] == "") AND ($addr != ""))
+ $result["addr"] = $addr;
+ }
+ }
}
// Only store into the cache if the value seems to be valid
- if ($result['network'] != NETWORK_PHANTOM)
- Cache::set("probe_url:".$mode.":".$url,serialize($result), CACHE_DAY);
+ if ($result['network'] != NETWORK_PHANTOM) {
+ Cache::set("probe_url:".$mode.":".$original_url,serialize($result), CACHE_DAY);
+
+ /// @todo temporary fix - we need a real contact update function that updates only changing fields
+ /// The biggest problem is the avatar picture that could have a reduced image size.
+ /// It should only be updated if the existing picture isn't existing anymore.
+ if (($result['network'] != NETWORK_FEED) AND ($mode == PROBE_NORMAL) AND
+ $result["name"] AND $result["nick"] AND $result["url"] AND $result["addr"] AND $result["poll"])
+ q("UPDATE `contact` SET `name` = '%s', `nick` = '%s', `url` = '%s', `addr` = '%s',
+ `notify` = '%s', `poll` = '%s', `alias` = '%s', `success_update` = '%s'
+ WHERE `nurl` = '%s' AND NOT `self` AND `uid` = 0",
+ dbesc($result["name"]),
+ dbesc($result["nick"]),
+ dbesc($result["url"]),
+ dbesc($result["addr"]),
+ dbesc($result["notify"]),
+ dbesc($result["poll"]),
+ dbesc($result["alias"]),
+ dbesc(datetime_convert()),
+ dbesc(normalise_link($result['url']))
+ );
+ }
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);
}
projects / friendica.git / blobdiff