3 require_once('include/datetime.php');
7 - noscrape for updating contact fields and "last updated"
8 - use /poco/@global for discovering contacts from other servers
9 - Make search for last activity optional
15 * Given a contact-id (minimum), load the PortableContacts friend list for that contact,
16 * and add the entries to the gcontact (Global Contact) table, or update existing entries
17 * if anything (name or photo) has changed.
18 * We use normalised urls for comparison which ignore http vs https and www.domain vs domain
20 * Once the global contact is stored add (if necessary) the contact linkage which associates
21 * the given uid, cid to the global contact entry. There can be many uid/cid combinations
22 * pointing to the same global contact id.
29 function poco_load($cid,$uid = 0,$zcid = 0,$url = null) {
31 require_once("include/html2bbcode.php");
36 if((! $url) || (! $uid)) {
37 $r = q("select `poco`, `uid` from `contact` where `id` = %d limit 1",
52 $url = $url . (($uid) ? '/@me/@all?fields=displayName,urls,photos,updated,network,aboutMe,currentLocation,tags,gender,generation' : '?fields=displayName,urls,photos,updated,network,aboutMe,currentLocation,tags,gender,generation') ;
54 logger('poco_load: ' . $url, LOGGER_DEBUG);
58 logger('poco_load: returns ' . $s, LOGGER_DATA);
60 logger('poco_load: return code: ' . $a->get_curl_code(), LOGGER_DEBUG);
62 if(($a->get_curl_code() > 299) || (! $s))
67 logger('poco_load: json: ' . print_r($j,true),LOGGER_DATA);
69 if(! isset($j->entry))
73 foreach($j->entry as $entry) {
81 $updated = '0000-00-00 00:00:00';
88 $name = $entry->displayName;
90 if(isset($entry->urls)) {
91 foreach($entry->urls as $url) {
92 if($url->type == 'profile') {
93 $profile_url = $url->value;
96 if($url->type == 'webfinger') {
97 $connect_url = str_replace('acct:' , '', $url->value);
102 if(isset($entry->photos)) {
103 foreach($entry->photos as $photo) {
104 if($photo->type == 'profile') {
105 $profile_photo = $photo->value;
111 if(isset($entry->updated))
112 $updated = date("Y-m-d H:i:s", strtotime($entry->updated));
114 if(isset($entry->network))
115 $network = $entry->network;
117 if(isset($entry->currentLocation))
118 $location = $entry->currentLocation;
120 if(isset($entry->aboutMe))
121 $about = html2bbcode($entry->aboutMe);
123 if(isset($entry->gender))
124 $gender = $entry->gender;
126 if(isset($entry->generation) AND ($entry->generation > 0))
127 $generation = ++$entry->generation;
129 if(isset($entry->tags))
130 foreach($entry->tags as $tag)
131 $keywords = implode(", ", $tag);
133 // If you query a Friendica server for its profiles, the network has to be Friendica
134 // To-Do: It could also be a Redmatrix server
136 // $network = NETWORK_DFRN;
138 poco_check($profile_url, $name, $network, $profile_photo, $about, $location, $gender, $keywords, $connect_url, $updated, $generation, $cid, $uid, $zcid);
140 // Update the Friendica contacts. Diaspora is doing it via a message. (See include/diaspora.php)
141 if (($location != "") OR ($about != "") OR ($keywords != "") OR ($gender != ""))
142 q("UPDATE `contact` SET `location` = '%s', `about` = '%s', `keywords` = '%s', `gender` = '%s'
143 WHERE `nurl` = '%s' AND NOT `self` AND `network` = '%s'",
148 dbesc(normalise_link($profile_url)),
149 dbesc(NETWORK_DFRN));
151 logger("poco_load: loaded $total entries",LOGGER_DEBUG);
153 q("DELETE FROM `glink` WHERE `cid` = %d AND `uid` = %d AND `zcid` = %d AND `updated` < UTC_TIMESTAMP - INTERVAL 2 DAY",
161 function poco_check($profile_url, $name, $network, $profile_photo, $about, $location, $gender, $keywords, $connect_url, $updated, $generation, $cid = 0, $uid = 0, $zcid = 0) {
167 // 1: Profiles on this server
168 // 2: Contacts of profiles on this server
169 // 3: Contacts of contacts of profiles on this server
174 if ($profile_url == "")
177 // Don't store the statusnet connector as network
178 // We can't simply set this to NETWORK_OSTATUS since the connector could have fetched posts from friendica as well
179 if ($network == NETWORK_STATUSNET)
182 // The global contacts should contain the original picture, not the cached one
183 if (($generation != 1) AND stristr(normalise_link($profile_photo), normalise_link($a->get_baseurl()."/photo/")))
186 $r = q("SELECT `network` FROM `contact` WHERE `nurl` = '%s' AND `network` != '' AND `network` != '%s' LIMIT 1",
187 dbesc(normalise_link($profile_url)), dbesc(NETWORK_STATUSNET)
190 $network = $r[0]["network"];
192 if (($network == "") OR ($network == NETWORK_OSTATUS)) {
193 $r = q("SELECT `network`, `url` FROM `contact` WHERE `alias` IN ('%s', '%s') AND `network` != '' AND `network` != '%s' LIMIT 1",
194 dbesc($profile_url), dbesc(normalise_link($profile_url)), dbesc(NETWORK_STATUSNET)
197 $network = $r[0]["network"];
198 $profile_url = $r[0]["url"];
202 $x = q("SELECT * FROM `gcontact` WHERE `nurl` = '%s' LIMIT 1",
203 dbesc(normalise_link($profile_url))
207 if (($network == "") AND ($x[0]["network"] != NETWORK_STATUSNET))
208 $network = $x[0]["network"];
210 if ($updated = "0000-00-00 00:00:00")
211 $updated = $x[0]["updated"];
213 $last_contact = $x[0]["last_contact"];
214 $last_failure = $x[0]["last_failure"];
216 $last_contact = "0000-00-00 00:00:00";
217 $last_failure = "0000-00-00 00:00:00";
220 if (($network == "") OR ($name == "") OR ($profile_photo == "")) {
221 require_once("include/Scrape.php");
223 $data = probe_url($profile_url);
224 $network = $data["network"];
225 $name = $data["name"];
226 $profile_url = $data["url"];
227 $profile_photo = $data["photo"];
230 if (count($x) AND ($x[0]["network"] == "") AND ($network != "")) {
231 q("UPDATE `gcontact` SET `network` = '%s' WHERE `nurl` = '%s'",
233 dbesc(normalise_link($profile_url))
237 if (($name == "") OR ($profile_photo == ""))
240 if (!in_array($network, array(NETWORK_DFRN, NETWORK_OSTATUS, NETWORK_DIASPORA)))
243 logger("profile-check generation: ".$generation." Network: ".$network." URL: ".$profile_url." name: ".$name." avatar: ".$profile_photo, LOGGER_DEBUG);
245 if (poco_do_update($updated, $last_contact, $last_failure)) {
246 $last_updated = poco_last_updated($profile_url);
248 $updated = $last_updated;
249 $last_contact = datetime_convert();
250 logger("Last updated for profile ".$profile_url.": ".$updated, LOGGER_DEBUG);
253 q("UPDATE `gcontact` SET `last_contact` = '%s' WHERE `nurl` = '%s'", dbesc($last_contact), dbesc(normalise_link($profile_url)));
255 $last_failure = datetime_convert();
258 q("UPDATE `gcontact` SET `last_failure` = '%s' WHERE `nurl` = '%s'", dbesc($last_failure), dbesc(normalise_link($profile_url)));
265 if (($location == "") AND ($x[0]['location'] != ""))
266 $location = $x[0]['location'];
268 if (($about == "") AND ($x[0]['about'] != ""))
269 $about = $x[0]['about'];
271 if (($gender == "") AND ($x[0]['gender'] != ""))
272 $gender = $x[0]['gender'];
274 if (($keywords == "") AND ($x[0]['keywords'] != ""))
275 $keywords = $x[0]['keywords'];
277 if (($generation == 0) AND ($x[0]['generation'] > 0))
278 $generation = $x[0]['generation'];
280 if($x[0]['name'] != $name || $x[0]['photo'] != $profile_photo || $x[0]['updated'] < $updated) {
281 q("UPDATE `gcontact` SET `name` = '%s', `network` = '%s', `photo` = '%s', `connect` = '%s', `url` = '%s',
282 `updated` = '%s', `location` = '%s', `about` = '%s', `keywords` = '%s', `gender` = '%s', `generation` = %d
283 WHERE (`generation` >= %d OR `generation` = 0) AND `nurl` = '%s'",
286 dbesc($profile_photo),
296 dbesc(normalise_link($profile_url))
300 q("INSERT INTO `gcontact` (`name`,`network`, `url`,`nurl`,`photo`,`connect`, `updated`, `last_contact`, `last_failure`, `location`, `about`, `keywords`, `gender`, `generation`)
301 VALUES ('%s', '%s', '%s', '%s', '%s','%s', '%s', '%s', '%s', '%s', '%s', %d)",
305 dbesc(normalise_link($profile_url)),
306 dbesc($profile_photo),
309 dbesc($last_contact),
310 dbesc($last_failure),
317 $x = q("SELECT * FROM `gcontact` WHERE `nurl` = '%s' LIMIT 1",
318 dbesc(normalise_link($profile_url))
327 $r = q("SELECT * FROM `glink` WHERE `cid` = %d AND `uid` = %d AND `gcid` = %d AND `zcid` = %d LIMIT 1",
334 q("INSERT INTO `glink` (`cid`,`uid`,`gcid`,`zcid`, `updated`) VALUES (%d,%d,%d,%d, '%s') ",
339 dbesc(datetime_convert())
342 q("UPDATE `glink` SET `updated` = '%s' WHERE `cid` = %d AND `uid` = %d AND `gcid` = %d AND `zcid` = %d",
343 dbesc(datetime_convert()),
351 // For unknown reasons there are sometimes duplicates
352 q("DELETE FROM `gcontact` WHERE `nurl` = '%s' AND `id` != %d AND
353 NOT EXISTS (SELECT `gcid` FROM `glink` WHERE `gcid` = `gcontact`.`id`)",
354 dbesc(normalise_link($profile_url)),
361 function poco_last_updated($profile) {
362 $data = probe_url($profile);
364 if (($data["poll"] == "") OR ($data["network"] == NETWORK_FEED))
367 // To-Do: Use noscrape
369 $feedret = z_fetch_url($data["poll"]);
371 if (!$feedret["success"])
374 $doc = new DOMDocument();
375 @$doc->loadXML($feedret["body"]);
377 $xpath = new DomXPath($doc);
378 $xpath->registerNamespace('atom', "http://www.w3.org/2005/Atom");
380 $entries = $xpath->query('/atom:feed/atom:entry');
384 foreach ($entries AS $entry) {
385 $published = $xpath->query('atom:published/text()', $entry)->item(0)->nodeValue;
386 $updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
388 if ($last_updated < $published)
389 $last_updated = $published;
391 if ($last_updated < $updated)
392 $last_updated = $updated;
395 // Maybe there aren't any entries. Then check if it is a valid feed
396 if ($last_updated == "")
397 if ($xpath->query('/atom:feed')->length > 0)
398 $last_updated = "0000-00-00 00:00:00";
400 return($last_updated);
403 function poco_do_update($updated, $last_contact, $last_failure) {
404 $now = strtotime(datetime_convert());
406 if ($updated > $last_contact)
407 $contact_time = strtotime($updated);
409 $contact_time = strtotime($last_contact);
411 $failure_time = strtotime($last_failure);
413 // If the last contact was less than 24 hours then don't update
414 if (($now - $contact_time) < (60 * 60 * 24))
417 // If the last failure was less than 24 hours then don't update
418 if (($now - $failure_time) < (60 * 60 * 24))
421 // If the last contact was less than a week ago and the last failure is older than a week then don't update
422 if ((($now - $contact_time) < (60 * 60 * 24 * 7)) AND ($contact_time > $failure_time))
425 // If the last contact time was more than a week ago, then only try once a week
426 if (($now - $contact_time) > (60 * 60 * 24 * 7) AND ($now - $failure_time) < (60 * 60 * 24 * 7))
429 // If the last contact time was more than a month ago, then only try once a month
430 if (($now - $contact_time) > (60 * 60 * 24 * 30) AND ($now - $failure_time) < (60 * 60 * 24 * 30))
436 function poco_contact_from_body($body, $created, $cid, $uid) {
437 preg_replace_callback("/\[share(.*?)\].*?\[\/share\]/ism",
438 function ($match) use ($created, $cid, $uid){
439 return(sub_poco_from_share($match, $created, $cid, $uid));
443 function sub_poco_from_share($share, $created, $cid, $uid) {
445 preg_match("/profile='(.*?)'/ism", $share[1], $matches);
446 if ($matches[1] != "")
447 $profile = $matches[1];
449 preg_match('/profile="(.*?)"/ism', $share[1], $matches);
450 if ($matches[1] != "")
451 $profile = $matches[1];
456 logger("prepare poco_check for profile ".$profile, LOGGER_DEBUG);
457 poco_check($profile, "", "", "", "", "", "", "", "", $created, 3, $cid, $uid);
460 function poco_store($item) {
463 if ($item['private'])
466 // Or is it from a network where we don't store the global contacts?
467 if (!in_array($item["network"], array(NETWORK_DFRN, NETWORK_DIASPORA, NETWORK_OSTATUS, NETWORK_STATUSNET, "")))
470 // Is it a global copy?
471 $store_gcontact = ($item["uid"] == 0);
473 // Is it a comment on a global copy?
474 if (!$store_gcontact AND ($item["uri"] != $item["parent-uri"])) {
475 $q = q("SELECT `id` FROM `item` WHERE `uri`='%s' AND `uid` = 0", $item["parent-uri"]);
476 $store_gcontact = count($q);
479 if (!$store_gcontact)
482 // "3" means: We don't know this contact directly (Maybe a reshared item)
485 $profile_url = $item["author-link"];
487 // Is it a user from our server?
488 $q = q("SELECT `id` FROM `contact` WHERE `self` AND `nurl` = '%s' LIMIT 1",
489 dbesc(normalise_link($item["author-link"])));
491 logger("Our user (generation 1): ".$item["author-link"], LOGGER_DEBUG);
493 $network = NETWORK_DFRN;
494 } else { // Is it a contact from a user on our server?
495 $q = q("SELECT `network`, `url` FROM `contact` WHERE `uid` != 0 AND `network` != ''
496 AND (`nurl` = '%s' OR `alias` IN ('%s', '%s')) AND `network` != '%s' LIMIT 1",
497 dbesc(normalise_link($item["author-link"])),
498 dbesc(normalise_link($item["author-link"])),
499 dbesc($item["author-link"]),
500 dbesc(NETWORK_STATUSNET));
503 $network = $q[0]["network"];
504 $profile_url = $q[0]["url"];
505 logger("Known contact (generation 2): ".$profile_url, LOGGER_DEBUG);
509 if ($generation == 3)
510 logger("Unknown contact (generation 3): ".$item["author-link"], LOGGER_DEBUG);
512 poco_check($profile_url, $item["author-name"], $network, $item["author-avatar"], "", "", "", "", "", $item["received"], $generation, $item["contact-id"], $item["uid"]);
514 // Maybe its a body with a shared item? Then extract a global contact from it.
515 poco_contact_from_body($item["body"], $item["received"], $item["contact-id"], $item["uid"]);
518 function count_common_friends($uid,$cid) {
520 $r = q("SELECT count(*) as `total`
521 FROM `glink` INNER JOIN `gcontact` on `glink`.`gcid` = `gcontact`.`id`
522 where `glink`.`cid` = %d and `glink`.`uid` = %d
523 and `gcontact`.`nurl` in (select nurl from contact where uid = %d and self = 0 and blocked = 0 and hidden = 0 and id != %d ) ",
530 // logger("count_common_friends: $uid $cid {$r[0]['total']}");
532 return $r[0]['total'];
538 function common_friends($uid,$cid,$start = 0,$limit=9999,$shuffle = false) {
541 $sql_extra = " order by rand() ";
543 $sql_extra = " order by `gcontact`.`name` asc ";
545 $r = q("SELECT `gcontact`.*
546 FROM `glink` INNER JOIN `gcontact` on `glink`.`gcid` = `gcontact`.`id`
547 where `glink`.`cid` = %d and `glink`.`uid` = %d
548 and `gcontact`.`nurl` in (select nurl from contact where uid = %d and self = 0 and blocked = 0 and hidden = 0 and id != %d )
549 $sql_extra limit %d, %d",
563 function count_common_friends_zcid($uid,$zcid) {
565 $r = q("SELECT count(*) as `total`
566 FROM `glink` INNER JOIN `gcontact` on `glink`.`gcid` = `gcontact`.`id`
567 where `glink`.`zcid` = %d
568 and `gcontact`.`nurl` in (select nurl from contact where uid = %d and self = 0 and blocked = 0 and hidden = 0 ) ",
574 return $r[0]['total'];
579 function common_friends_zcid($uid,$zcid,$start = 0, $limit = 9999,$shuffle = false) {
582 $sql_extra = " order by rand() ";
584 $sql_extra = " order by `gcontact`.`name` asc ";
586 $r = q("SELECT `gcontact`.*
587 FROM `glink` INNER JOIN `gcontact` on `glink`.`gcid` = `gcontact`.`id`
588 where `glink`.`zcid` = %d
589 and `gcontact`.`nurl` in (select nurl from contact where uid = %d and self = 0 and blocked = 0 and hidden = 0 )
590 $sql_extra limit %d, %d",
602 function count_all_friends($uid,$cid) {
604 $r = q("SELECT count(*) as `total`
605 FROM `glink` INNER JOIN `gcontact` on `glink`.`gcid` = `gcontact`.`id`
606 where `glink`.`cid` = %d and `glink`.`uid` = %d ",
612 return $r[0]['total'];
618 function all_friends($uid,$cid,$start = 0, $limit = 80) {
620 $r = q("SELECT `gcontact`.*
621 FROM `glink` INNER JOIN `gcontact` on `glink`.`gcid` = `gcontact`.`id`
622 where `glink`.`cid` = %d and `glink`.`uid` = %d
623 order by `gcontact`.`name` asc LIMIT %d, %d ",
635 function suggestion_query($uid, $start = 0, $limit = 80) {
640 $network = array(NETWORK_DFRN);
642 if (get_config('system','diaspora_enabled'))
643 $network[] = NETWORK_DIASPORA;
645 if (!get_config('system','ostatus_disabled'))
646 $network[] = NETWORK_OSTATUS;
648 $sql_network = implode("', '", $network);
649 //$sql_network = "'".$sql_network."', ''";
650 $sql_network = "'".$sql_network."'";
652 $r = q("SELECT count(glink.gcid) as `total`, gcontact.* from gcontact
653 INNER JOIN glink on glink.gcid = gcontact.id
654 where uid = %d and not gcontact.nurl in ( select nurl from contact where uid = %d )
655 and not gcontact.name in ( select name from contact where uid = %d )
656 and not gcontact.id in ( select gcid from gcign where uid = %d )
657 AND `gcontact`.`updated` != '0000-00-00 00:00:00'
658 AND `gcontact`.`last_contact` >= `gcontact`.`last_failure`
659 AND `gcontact`.`network` IN (%s)
660 group by glink.gcid order by gcontact.updated desc,total desc limit %d, %d ",
670 if(count($r) && count($r) >= ($limit -1))
673 $r2 = q("SELECT gcontact.* from gcontact
674 INNER JOIN glink on glink.gcid = gcontact.id
675 where glink.uid = 0 and glink.cid = 0 and glink.zcid = 0 and not gcontact.nurl in ( select nurl from contact where uid = %d )
676 and not gcontact.name in ( select name from contact where uid = %d )
677 and not gcontact.id in ( select gcid from gcign where uid = %d )
678 AND `gcontact`.`updated` != '0000-00-00 00:00:00'
679 AND `gcontact`.`network` IN (%s)
680 order by rand() limit %d, %d ",
690 foreach ($r2 AS $suggestion)
691 $list[$suggestion["nurl"]] = $suggestion;
693 foreach ($r AS $suggestion)
694 $list[$suggestion["nurl"]] = $suggestion;
699 function update_suggestions() {
705 poco_load(0,0,0,$a->get_baseurl() . '/poco');
707 $done[] = $a->get_baseurl() . '/poco';
709 if(strlen(get_config('system','directory_submit_url'))) {
710 $x = fetch_url('http://dir.friendica.com/pubsites');
712 $j = json_decode($x);
714 foreach($j->entries as $entry) {
715 $url = $entry->url . '/poco';
716 if(! in_array($url,$done))
717 poco_load(0,0,0,$entry->url . '/poco');
723 $r = q("select distinct(poco) as poco from contact where network = '%s'",
729 $base = substr($rr['poco'],0,strrpos($rr['poco'],'/'));
730 if(! in_array($base,$done))
731 poco_load(0,0,0,$base);