Merge pull request #2467 from annando/1604-gcontact-cleanup

Some more checks that the gcontact table contains valid data

include/Scrape.php

@@ -799,6 +799,9 @@ function probe_url($url, $mode = PROBE_NORMAL, $level = 1) {
 	if (substr($baseurl, -10) == "/index.php")
 		$baseurl = str_replace("/index.php", "", $baseurl);
 
+	if ($network == "")
+		$network = NETWORK_PHANTOM;
+
 	$baseurl = rtrim($baseurl, "/");
 
 	if(strpos($url,'@') AND ($addr == "") AND ($network == NETWORK_DFRN))
@@ -846,7 +849,7 @@ function probe_url($url, $mode = PROBE_NORMAL, $level = 1) {
 	}
 
 	// Only store into the cache if the value seems to be valid
-	if (!in_array($result['network'], array(NETWORK_PHANTOM, ""))) {
+	if ($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

include/gprobe.php

@@ -58,7 +58,7 @@ function gprobe_run(&$argv, &$argc){
 		if (is_null($result))
 			Cache::set("gprobe:".$urlparts["host"],serialize($arr));
 
-		if (!in_array($result["network"], array(NETWORK_FEED, NETWORK_PHANTOM)))
+		if (!in_array($arr["network"], array(NETWORK_FEED, NETWORK_PHANTOM)))
 			update_gcontact($arr);
 
 		$r = q("SELECT `id`, `url`, `network` FROM `gcontact` WHERE `nurl` = '%s' ORDER BY `id` LIMIT 1",

include/socgraph.php

@@ -1374,9 +1374,30 @@ function clean_contact_url($url) {
         if (isset($parts["path"]))
                 $new_url .= $parts["path"];
 
+	if ($new_url != $url)
+		logger("Cleaned contact url ".$url." to ".$new_url." - Called by: ".App::callstack(), LOGGER_DEBUG);
+
         return $new_url;
 }
 
+/**
+ * @brief Replace alternate OStatus user format with the primary one
+ *
+ * @param arr $contact contact array (called by reference)
+ */
+function fix_alternate_contact_address(&$contact) {
+	if (($contact["network"] == NETWORK_OSTATUS) AND poco_alternate_ostatus_url($contact["url"])) {
+	        $data = probe_url($contact["url"]);
+		if ($contact["network"] == NETWORK_OSTATUS) {
+			logger("Fix primary url from ".$contact["url"]." to ".$data["url"]." - Called by: ".App::callstack(), LOGGER_DEBUG);
+			$contact["url"] = $data["url"];
+			$contact["addr"] = $data["addr"];
+			$contact["alias"] = $data["alias"];
+			$contact["server_url"] = $data["baseurl"];
+		}
+	}
+}
+
 /**
  * @brief Fetch the gcontact id, add an entry if not existed
  *
@@ -1387,14 +1408,20 @@ function get_gcontact_id($contact) {
 
 	$gcontact_id = 0;
 
-	if ($contact["network"] == NETWORK_PHANTOM)
+	if (in_array($contact["network"], array(NETWORK_PHANTOM))) {
+		logger("Invalid network for contact url ".$contact["url"]." - Called by: ".App::callstack(), LOGGER_DEBUG);
 		return false;
+	}
 
 	if ($contact["network"] == NETWORK_STATUSNET)
 		$contact["network"] = NETWORK_OSTATUS;
 
+	// Replace alternate OStatus user format with the primary one
+	fix_alternate_contact_address($contact);
+
 	// Remove unwanted parts from the contact url (e.g. "?zrl=...")
-	$contact["url"] = clean_contact_url($contact["url"]);
+	if (in_array($contact["network"], array(NETWORK_DFRN, NETWORK_DIASPORA, NETWORK_OSTATUS)))
+		$contact["url"] = clean_contact_url($contact["url"]);
 
 	$r = q("SELECT `id` FROM `gcontact` WHERE `nurl` = '%s' ORDER BY `id` LIMIT 2",
 		dbesc(normalise_link($contact["url"])));
@@ -1477,6 +1504,9 @@ function update_gcontact($contact) {
 	if ($contact["network"] == NETWORK_STATUSNET)
 		$contact["network"] = NETWORK_OSTATUS;
 
+	// Replace alternate OStatus user format with the primary one
+	fix_alternate_contact_address($contact);
+
 	if (!isset($contact["updated"]))
 		$contact["updated"] = datetime_convert();
 
@@ -1571,8 +1601,10 @@ function update_gcontact($contact) {
 function update_gcontact_from_probe($url) {
 	$data = probe_url($url);
 
-	if ($data["network"] == NETWORK_PHANTOM)
+	if (in_array($data["network"], array(NETWORK_PHANTOM))) {
+		logger("Invalid network for contact url ".$data["url"]." - Called by: ".App::callstack(), LOGGER_DEBUG);
 		return;
+	}
 
 	update_gcontact($data);
 }

mod/item.php

@@ -1147,7 +1147,7 @@ function handle_tag($a, &$body, &$inform, &$str_tags, $profile_uid, $tag, $netwo
 
 			if (!$r) {
 				$probed = probe_url($name);
-				if (isset($probed["url"])) {
+				if ($result['network'] != NETWORK_PHANTOM) {
 					update_gcontact($probed);
 					$r = q("SELECT `url`, `name`, `nick`, `network`, `alias`, `notify` FROM `gcontact` WHERE `nurl` = '%s' LIMIT 1",
 						dbesc(normalise_link($probed["url"])));