Check the global contacts every 90 days/Better handling in item creation

diff --git a/include/feed.php b/include/feed.php
index 293de3cc965ed6b892599db7edd17c5aad6e12d4..15e5f65366d939d7f351a5a3b4fe09e292b6a151 100644 (file)
--- a/include/feed.php
+++ b/include/feed.php
@@ -47,7 +47,7 @@ function feed_import($xml,$importer,&$contact, &$hub, $simulate = false) {
        }
 
        // Is it Atom?
-       if ($xpath->query('/atom:feed/atom:entry')->length > 0) {
+       if ($xpath->query('/atom:feed')->length > 0) {
                $alternate = $xpath->query("atom:link[@rel='alternate']")->item(0)->attributes;
                if (is_object($alternate))
                        foreach($alternate AS $attributes)

diff --git a/include/gprobe.php b/include/gprobe.php
index bd8503835b346eae7a4b04f337edfead95f65986..91b9b16e3306077ba2f0c574e6c46e0ce5fe066a 100644 (file)
--- a/include/gprobe.php
+++ b/include/gprobe.php
@@ -39,7 +39,7 @@ function gprobe_run(&$argv, &$argc){
 
        logger("gprobe start for ".normalise_link($url), LOGGER_DEBUG);
 
-       if(! count($r)) {
+       if (!count($r)) {
 
                // Is it a DDoS attempt?
                $urlparts = parse_url($url);
@@ -65,9 +65,11 @@ function gprobe_run(&$argv, &$argc){
                        dbesc(normalise_link($url))
                );
        }
-       if(count($r))
-               if ($r[0]["network"] == NETWORK_DFRN)
+       if(count($r)) {
+               // Check for accessibility and do a poco discovery
+               if (poco_last_updated($r[0]['url'], true) AND ($r[0]["network"] == NETWORK_DFRN))
                        poco_load(0,0,$r[0]['id'], str_replace('/profile/','/poco/',$r[0]['url']));
+       }
 
        logger("gprobe end for ".normalise_link($url), LOGGER_DEBUG);
        return;

diff --git a/include/socgraph.php b/include/socgraph.php
index 889d16ee6588530ab4e73db56f51c9ba4e17282e..0acd2793aa21e3351d8ceb365142d03da917d760 100644 (file)
--- a/include/socgraph.php
+++ b/include/socgraph.php
@@ -404,6 +404,11 @@ function poco_last_updated($profile, $force = false) {
        else
                $server_url = poco_detect_server($profile);
 
+       if (!in_array($gcontacts[0]["network"], array(NETWORK_DFRN, NETWORK_DIASPORA, NETWORK_FEED, NETWORK_OSTATUS, ""))) {
+               logger("Profile ".$profile.": Network type ".$gcontacts[0]["network"]." can't be checked", LOGGER_DEBUG);
+               return false;
+       }
+
        if ($server_url != "") {
                if (!poco_check_server($server_url, $gcontacts[0]["network"], $force)) {
 
@@ -411,6 +416,7 @@ function poco_last_updated($profile, $force = false) {
                                q("UPDATE `gcontact` SET `last_failure` = '%s' WHERE `nurl` = '%s'",
                                        dbesc(datetime_convert()), dbesc(normalise_link($profile)));
 
+                       logger("Profile ".$profile.": Server ".$server_url." wasn't reachable.", LOGGER_DEBUG);
                        return false;
                }
 
@@ -426,7 +432,7 @@ function poco_last_updated($profile, $force = false) {
                        q("UPDATE `gcontact` SET `network` = '%s' WHERE `nurl` = '%s'",
                                dbesc($server[0]["network"]), dbesc(normalise_link($profile)));
                else
-                       return;
+                       return false;
        }
 
        // noscrape is really fast so we don't cache the call.
@@ -477,19 +483,32 @@ function poco_last_updated($profile, $force = false) {
                                        unset($noscrape["dfrn-notify"]);
                                        unset($noscrape["dfrn-poll"]);
 
+                                       // Set the date of the last contact
+                                       /// @todo By now the function "update_gcontact" doesn't work with this field
+                                       //$contact["last_contact"] = datetime_convert();
+
                                        $contact = array_merge($contact, $noscrape);
 
                                        update_gcontact($contact);
 
-                                       return $noscrape["updated"];
+                                       if (trim($noscrape["updated"]) != "") {
+                                               q("UPDATE `gcontact` SET `last_contact` = '%s' WHERE `nurl` = '%s'",
+                                                       dbesc(datetime_convert()), dbesc(normalise_link($profile)));
+
+                                               logger("Profile ".$profile." was last updated at ".$noscrape["updated"]." (noscrape)", LOGGER_DEBUG);
+
+                                               return $noscrape["updated"];
+                                       }
                                }
                        }
                }
        }
 
        // If we only can poll the feed, then we only do this once a while
-       if (!$force AND !poco_do_update($gcontacts[0]["created"], $gcontacts[0]["updated"], $gcontacts[0]["last_failure"],  $gcontacts[0]["last_contact"]))
+       if (!$force AND !poco_do_update($gcontacts[0]["created"], $gcontacts[0]["updated"], $gcontacts[0]["last_failure"],  $gcontacts[0]["last_contact"])) {
+               logger("Profile ".$profile." was last updated at ".$gcontacts[0]["updated"]." (cached)", LOGGER_DEBUG);
                return $gcontacts[0]["updated"];
+       }
 
        $data = probe_url($profile);
 
@@ -508,12 +527,15 @@ function poco_last_updated($profile, $force = false) {
 
                poco_last_updated($data["url"], $force);
 
+               logger("Profile ".$profile." was deleted", LOGGER_DEBUG);
                return false;
        }
 
        if (($data["poll"] == "") OR (in_array($data["network"], array(NETWORK_FEED, NETWORK_PHANTOM)))) {
                q("UPDATE `gcontact` SET `last_failure` = '%s' WHERE `nurl` = '%s'",
                        dbesc(datetime_convert()), dbesc(normalise_link($profile)));
+
+               logger("Profile ".$profile." wasn't reachable (profile)", LOGGER_DEBUG);
                return false;
        }
 
@@ -539,6 +561,8 @@ function poco_last_updated($profile, $force = false) {
        if (!$feedret["success"]) {
                q("UPDATE `gcontact` SET `last_failure` = '%s' WHERE `nurl` = '%s'",
                        dbesc(datetime_convert()), dbesc(normalise_link($profile)));
+
+               logger("Profile ".$profile." wasn't reachable (no feed)", LOGGER_DEBUG);
                return false;
        }
 
@@ -575,6 +599,8 @@ function poco_last_updated($profile, $force = false) {
                q("UPDATE `gcontact` SET `generation` = 9 WHERE `nurl` = '%s'",
                        dbesc(normalise_link($profile)));
 
+       logger("Profile ".$profile." was last updated at ".$last_updated, LOGGER_DEBUG);
+
        return($last_updated);
 }
 
@@ -1411,6 +1437,7 @@ function fix_alternate_contact_address(&$contact) {
 function get_gcontact_id($contact) {
 
        $gcontact_id = 0;
+       $doprobing = false;
 
        if (in_array($contact["network"], array(NETWORK_PHANTOM))) {
                logger("Invalid network for contact url ".$contact["url"]." - Called by: ".App::callstack(), LOGGER_DEBUG);
@@ -1427,12 +1454,21 @@ function get_gcontact_id($contact) {
        if (in_array($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",
+       $r = q("SELECT `id`, `last_contact`, `last_failure`, `network` FROM `gcontact` WHERE `nurl` = '%s' ORDER BY `id` LIMIT 2",
                dbesc(normalise_link($contact["url"])));
 
-       if ($r)
+       if ($r) {
                $gcontact_id = $r[0]["id"];
-       else {
+
+               // Update every 90 days
+               if (in_array($r[0]["network"], array(NETWORK_DFRN, NETWORK_DIASPORA, NETWORK_OSTATUS, ""))) {
+                       $last_failure_str = $r[0]["last_failure"];
+                       $last_failure = strtotime($r[0]["last_failure"]);
+                       $last_contact_str = $r[0]["last_contact"];
+                       $last_contact = strtotime($r[0]["last_contact"]);
+                       $doprobing = (((time() - $last_contact) > (90 * 86400)) AND ((time() - $last_failure) > (90 * 86400)));
+               }
+       } else {
                q("INSERT INTO `gcontact` (`name`, `nick`, `addr` , `network`, `url`, `nurl`, `photo`, `created`, `updated`, `location`, `about`, `generation`)
                        VALUES ('%s', '%s', '%s', '%s', '%s', '%s', '%s', '%s', '%s', '%s', '%s', %d)",
                        dbesc($contact["name"]),
@@ -1449,20 +1485,21 @@ function get_gcontact_id($contact) {
                        intval($contact["generation"])
                );
 
-               $r = q("SELECT `id` FROM `gcontact` WHERE `nurl` = '%s' ORDER BY `id` LIMIT 2",
+               $r = q("SELECT `id`, `network` FROM `gcontact` WHERE `nurl` = '%s' ORDER BY `id` LIMIT 2",
                        dbesc(normalise_link($contact["url"])));
 
                if ($r) {
                        $gcontact_id = $r[0]["id"];
 
-                       // Complete newly added contacts from "probable" accounts
-                       if (in_array($contact["network"], array(NETWORK_DFRN, NETWORK_OSTATUS, NETWORK_DIASPORA, NETWORK_FEED))) {
-                               logger("Probing ".$contact["url"], LOGGER_DEBUG);
-                               proc_run('php', 'include/gprobe.php', bin2hex($contact["url"]));
-                       }
+                       $doprobing = in_array($r[0]["network"], array(NETWORK_DFRN, NETWORK_DIASPORA, NETWORK_OSTATUS, ""));
                }
        }
 
+       if ($doprobing) {
+               logger("Last Contact: ". $last_contact_str." - Last Failure: ".$last_failure_str." - Checking: ".$contact["url"], LOGGER_DEBUG);
+               proc_run('php', 'include/gprobe.php', bin2hex($contact["url"]));
+       }
+
        if ((count($r) > 1) AND ($gcontact_id > 0) AND ($contact["url"] != ""))
         q("DELETE FROM `gcontact` WHERE `nurl` = '%s' AND `id` != %d",
                dbesc(normalise_link($contact["url"])),

diff --git a/mod/item.php b/mod/item.php
index 60e6931583426f08bc91f2827972fbecd40be94d..10877b00ac9e96fee73c623c29498b6d6af60821 100644 (file)
--- a/mod/item.php
+++ b/mod/item.php
@@ -1121,29 +1121,43 @@ function handle_tag($a, &$body, &$inform, &$str_tags, $profile_uid, $tag, $netwo
                        // Is it in format @user@domain.tld or @http://domain.tld/...?
 
                        // First check the contact table for the address
-                       $r = q("SELECT `id`, `url`, `nick`, `name`, `alias`, `network`, `notify` FROM `contact` WHERE `addr` = '%s' AND `uid` = %d LIMIT 1",
+                       $r = q("SELECT `id`, `url`, `nick`, `name`, `alias`, `network`, `notify` FROM `contact`
+                               WHERE `addr` = '%s' AND `uid` = %d AND
+                                       (`network` != '%s' OR (`notify` != '' AND `alias` != ''))
+                               LIMIT 1",
                                        dbesc($name),
-                                       intval($profile_uid)
+                                       intval($profile_uid),
+                                       dbesc(NETWORK_OSTATUS)
                        );
 
                        // Then check in the contact table for the url
                        if (!$r)
-                               $r = q("SELECT `id`, `url`, `nick`, `name`, `alias`, `notify`, `network` FROM `contact` WHERE `nurl` = '%s' AND `uid` = %d LIMIT 1",
+                               $r = q("SELECT `id`, `url`, `nick`, `name`, `alias`, `network`, `notify` FROM `contact`
+                                       WHERE `nurl` = '%s' AND `uid` = %d AND
+                                               (`network` != '%s' OR (`notify` != '' AND `alias` != ''))
+                                       LIMIT 1",
                                                dbesc(normalise_link($name)),
-                                               intval($profile_uid)
+                                               intval($profile_uid),
+                                               dbesc(NETWORK_OSTATUS)
                                );
 
                        // Then check in the global contacts for the address
                        if (!$r)
-                               $r = q("SELECT `url`, `name`, `nick`, `network`, `alias`, `notify` FROM `gcontact` WHERE `addr` = '%s' LIMIT 1", dbesc($name));
+                               $r = q("SELECT `url`, `nick`, `name`, `alias`, `network`, `notify` FROM `gcontact`
+                                       WHERE `addr` = '%s' AND (`network` != '%s' OR (`notify` != '' AND `alias` != ''))
+                                       LIMIT 1",
+                                               dbesc($name),
+                                               dbesc(NETWORK_OSTATUS)
+                               );
 
                        // Then check in the global contacts for the url
                        if (!$r)
-                               $r = q("SELECT `url`, `name`, `nick`, `network`, `alias`, `notify` FROM `gcontact` WHERE `nurl` = '%s' LIMIT 1", dbesc(normalise_link($name)));
-
-                       // If the data isn't complete then refetch the data
-                       if ($r AND ($r[0]["network"] == NETWORK_OSTATUS) AND (($r[0]["notify"] == "") OR ($r[0]["alias"] == "")))
-                               $r = false;
+                               $r = q("SELECT `url`, `nick`, `name`, `alias`, `network`, `notify` FROM `gcontact`
+                                       WHERE `nurl` = '%s' AND (`network` != '%s' OR (`notify` != '' AND `alias` != ''))
+                                       LIMIT 1",
+                                               dbesc(normalise_link($name)),
+                                               dbesc(NETWORK_OSTATUS)
+                               );
 
                        if (!$r) {
                                $probed = probe_url($name);