3 require_once('include/datetime.php');
4 require_once("include/Scrape.php");
8 - noscrape for updating contact fields and "last updated"
9 - use /poco/@global for discovering contacts from other servers
10 - Make search for last activity optional
11 - only export contacts via poco where update is higher than failure
12 - check your own contacts in some way as well
18 * Given a contact-id (minimum), load the PortableContacts friend list for that contact,
19 * and add the entries to the gcontact (Global Contact) table, or update existing entries
20 * if anything (name or photo) has changed.
21 * We use normalised urls for comparison which ignore http vs https and www.domain vs domain
23 * Once the global contact is stored add (if necessary) the contact linkage which associates
24 * the given uid, cid to the global contact entry. There can be many uid/cid combinations
25 * pointing to the same global contact id.
32 function poco_load($cid,$uid = 0,$zcid = 0,$url = null) {
34 require_once("include/html2bbcode.php");
39 if((! $url) || (! $uid)) {
40 $r = q("select `poco`, `uid` from `contact` where `id` = %d limit 1",
55 $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') ;
57 logger('poco_load: ' . $url, LOGGER_DEBUG);
61 logger('poco_load: returns ' . $s, LOGGER_DATA);
63 logger('poco_load: return code: ' . $a->get_curl_code(), LOGGER_DEBUG);
65 if(($a->get_curl_code() > 299) || (! $s))
70 logger('poco_load: json: ' . print_r($j,true),LOGGER_DATA);
72 if(! isset($j->entry))
76 foreach($j->entry as $entry) {
84 $updated = '0000-00-00 00:00:00';
91 $name = $entry->displayName;
93 if(isset($entry->urls)) {
94 foreach($entry->urls as $url) {
95 if($url->type == 'profile') {
96 $profile_url = $url->value;
99 if($url->type == 'webfinger') {
100 $connect_url = str_replace('acct:' , '', $url->value);
105 if(isset($entry->photos)) {
106 foreach($entry->photos as $photo) {
107 if($photo->type == 'profile') {
108 $profile_photo = $photo->value;
114 if(isset($entry->updated))
115 $updated = date("Y-m-d H:i:s", strtotime($entry->updated));
117 if(isset($entry->network))
118 $network = $entry->network;
120 if(isset($entry->currentLocation))
121 $location = $entry->currentLocation;
123 if(isset($entry->aboutMe))
124 $about = html2bbcode($entry->aboutMe);
126 if(isset($entry->gender))
127 $gender = $entry->gender;
129 if(isset($entry->generation) AND ($entry->generation > 0))
130 $generation = ++$entry->generation;
132 if(isset($entry->tags))
133 foreach($entry->tags as $tag)
134 $keywords = implode(", ", $tag);
136 // If you query a Friendica server for its profiles, the network has to be Friendica
137 // To-Do: It could also be a Redmatrix server
139 // $network = NETWORK_DFRN;
141 poco_check($profile_url, $name, $network, $profile_photo, $about, $location, $gender, $keywords, $connect_url, $updated, $generation, $cid, $uid, $zcid);
143 // Update the Friendica contacts. Diaspora is doing it via a message. (See include/diaspora.php)
144 if (($location != "") OR ($about != "") OR ($keywords != "") OR ($gender != ""))
145 q("UPDATE `contact` SET `location` = '%s', `about` = '%s', `keywords` = '%s', `gender` = '%s'
146 WHERE `nurl` = '%s' AND NOT `self` AND `network` = '%s'",
151 dbesc(normalise_link($profile_url)),
152 dbesc(NETWORK_DFRN));
154 logger("poco_load: loaded $total entries",LOGGER_DEBUG);
156 q("DELETE FROM `glink` WHERE `cid` = %d AND `uid` = %d AND `zcid` = %d AND `updated` < UTC_TIMESTAMP - INTERVAL 2 DAY",
164 function poco_check($profile_url, $name, $network, $profile_photo, $about, $location, $gender, $keywords, $connect_url, $updated, $generation, $cid = 0, $uid = 0, $zcid = 0) {
170 // 1: Profiles on this server
171 // 2: Contacts of profiles on this server
172 // 3: Contacts of contacts of profiles on this server
177 if ($profile_url == "")
180 $orig_updated = $updated;
182 // Don't store the statusnet connector as network
183 // We can't simply set this to NETWORK_OSTATUS since the connector could have fetched posts from friendica as well
184 if ($network == NETWORK_STATUSNET)
187 // The global contacts should contain the original picture, not the cached one
188 if (($generation != 1) AND stristr(normalise_link($profile_photo), normalise_link($a->get_baseurl()."/photo/")))
191 $r = q("SELECT `network` FROM `contact` WHERE `nurl` = '%s' AND `network` != '' AND `network` != '%s' LIMIT 1",
192 dbesc(normalise_link($profile_url)), dbesc(NETWORK_STATUSNET)
195 $network = $r[0]["network"];
197 if (($network == "") OR ($network == NETWORK_OSTATUS)) {
198 $r = q("SELECT `network`, `url` FROM `contact` WHERE `alias` IN ('%s', '%s') AND `network` != '' AND `network` != '%s' LIMIT 1",
199 dbesc($profile_url), dbesc(normalise_link($profile_url)), dbesc(NETWORK_STATUSNET)
202 $network = $r[0]["network"];
203 $profile_url = $r[0]["url"];
207 $x = q("SELECT * FROM `gcontact` WHERE `nurl` = '%s' LIMIT 1",
208 dbesc(normalise_link($profile_url))
212 if (($network == "") AND ($x[0]["network"] != NETWORK_STATUSNET))
213 $network = $x[0]["network"];
215 if ($updated == "0000-00-00 00:00:00")
216 $updated = $x[0]["updated"];
218 $last_contact = $x[0]["last_contact"];
219 $last_failure = $x[0]["last_failure"];
221 $last_contact = "0000-00-00 00:00:00";
222 $last_failure = "0000-00-00 00:00:00";
225 if (($network == "") OR ($name == "") OR ($profile_photo == "")) {
226 $data = probe_url($profile_url);
227 $network = $data["network"];
228 $name = $data["name"];
229 $profile_url = $data["url"];
230 $profile_photo = $data["photo"];
233 if (count($x) AND ($x[0]["network"] == "") AND ($network != "")) {
234 q("UPDATE `gcontact` SET `network` = '%s' WHERE `nurl` = '%s'",
236 dbesc(normalise_link($profile_url))
240 if (($name == "") OR ($profile_photo == ""))
243 if (!in_array($network, array(NETWORK_DFRN, NETWORK_OSTATUS, NETWORK_DIASPORA)))
246 logger("profile-check generation: ".$generation." Network: ".$network." URL: ".$profile_url." name: ".$name." avatar: ".$profile_photo, LOGGER_DEBUG);
248 // Only fetch last update manually if it wasn't provided
249 if (($orig_updated == "0000-00-00 00:00:00") AND poco_do_update($updated, $last_contact, $last_failure)) {
250 $last_updated = poco_last_updated($profile_url);
252 $updated = $last_updated;
253 $last_contact = datetime_convert();
254 logger("Last updated for profile ".$profile_url.": ".$updated, LOGGER_DEBUG);
257 q("UPDATE `gcontact` SET `last_contact` = '%s' WHERE `nurl` = '%s'", dbesc($last_contact), dbesc(normalise_link($profile_url)));
259 $last_failure = datetime_convert();
262 q("UPDATE `gcontact` SET `last_failure` = '%s' WHERE `nurl` = '%s'", dbesc($last_failure), dbesc(normalise_link($profile_url)));
269 if (($location == "") AND ($x[0]['location'] != ""))
270 $location = $x[0]['location'];
272 if (($about == "") AND ($x[0]['about'] != ""))
273 $about = $x[0]['about'];
275 if (($gender == "") AND ($x[0]['gender'] != ""))
276 $gender = $x[0]['gender'];
278 if (($keywords == "") AND ($x[0]['keywords'] != ""))
279 $keywords = $x[0]['keywords'];
281 if (($generation == 0) AND ($x[0]['generation'] > 0))
282 $generation = $x[0]['generation'];
284 if($x[0]['name'] != $name || $x[0]['photo'] != $profile_photo || $x[0]['updated'] < $updated) {
285 q("UPDATE `gcontact` SET `name` = '%s', `network` = '%s', `photo` = '%s', `connect` = '%s', `url` = '%s',
286 `updated` = '%s', `location` = '%s', `about` = '%s', `keywords` = '%s', `gender` = '%s', `generation` = %d
287 WHERE (`generation` >= %d OR `generation` = 0) AND `nurl` = '%s'",
290 dbesc($profile_photo),
300 dbesc(normalise_link($profile_url))
304 q("INSERT INTO `gcontact` (`name`,`network`, `url`,`nurl`,`photo`,`connect`, `updated`, `last_contact`, `last_failure`, `location`, `about`, `keywords`, `gender`, `generation`)
305 VALUES ('%s', '%s', '%s', '%s', '%s','%s', '%s', '%s', '%s', '%s', '%s', %d)",
309 dbesc(normalise_link($profile_url)),
310 dbesc($profile_photo),
313 dbesc($last_contact),
314 dbesc($last_failure),
321 $x = q("SELECT * FROM `gcontact` WHERE `nurl` = '%s' LIMIT 1",
322 dbesc(normalise_link($profile_url))
331 $r = q("SELECT * FROM `glink` WHERE `cid` = %d AND `uid` = %d AND `gcid` = %d AND `zcid` = %d LIMIT 1",
338 q("INSERT INTO `glink` (`cid`,`uid`,`gcid`,`zcid`, `updated`) VALUES (%d,%d,%d,%d, '%s') ",
343 dbesc(datetime_convert())
346 q("UPDATE `glink` SET `updated` = '%s' WHERE `cid` = %d AND `uid` = %d AND `gcid` = %d AND `zcid` = %d",
347 dbesc(datetime_convert()),
355 // For unknown reasons there are sometimes duplicates
356 q("DELETE FROM `gcontact` WHERE `nurl` = '%s' AND `id` != %d AND
357 NOT EXISTS (SELECT `gcid` FROM `glink` WHERE `gcid` = `gcontact`.`id`)",
358 dbesc(normalise_link($profile_url)),
365 function poco_last_updated($profile) {
366 $data = probe_url($profile);
368 if (($data["poll"] == "") OR ($data["network"] == NETWORK_FEED))
371 // To-Do: Use noscrape
373 $feedret = z_fetch_url($data["poll"]);
375 if (!$feedret["success"])
378 $doc = new DOMDocument();
379 @$doc->loadXML($feedret["body"]);
381 $xpath = new DomXPath($doc);
382 $xpath->registerNamespace('atom', "http://www.w3.org/2005/Atom");
384 $entries = $xpath->query('/atom:feed/atom:entry');
388 foreach ($entries AS $entry) {
389 $published = $xpath->query('atom:published/text()', $entry)->item(0)->nodeValue;
390 $updated = $xpath->query('atom:updated/text()', $entry)->item(0)->nodeValue;
392 if ($last_updated < $published)
393 $last_updated = $published;
395 if ($last_updated < $updated)
396 $last_updated = $updated;
399 // Maybe there aren't any entries. Then check if it is a valid feed
400 if ($last_updated == "")
401 if ($xpath->query('/atom:feed')->length > 0)
402 $last_updated = "0000-00-00 00:00:00";
404 return($last_updated);
407 function poco_do_update($updated, $last_contact, $last_failure) {
408 $now = strtotime(datetime_convert());
410 if ($updated > $last_contact)
411 $contact_time = strtotime($updated);
413 $contact_time = strtotime($last_contact);
415 $failure_time = strtotime($last_failure);
417 // If the last contact was less than 24 hours then don't update
418 if (($now - $contact_time) < (60 * 60 * 24))
421 // If the last failure was less than 24 hours then don't update
422 if (($now - $failure_time) < (60 * 60 * 24))
425 // If the last contact was less than a week ago and the last failure is older than a week then don't update
426 if ((($now - $contact_time) < (60 * 60 * 24 * 7)) AND ($contact_time > $failure_time))
429 // If the last contact time was more than a week ago, then only try once a week
430 if (($now - $contact_time) > (60 * 60 * 24 * 7) AND ($now - $failure_time) < (60 * 60 * 24 * 7))
433 // If the last contact time was more than a month ago, then only try once a month
434 if (($now - $contact_time) > (60 * 60 * 24 * 30) AND ($now - $failure_time) < (60 * 60 * 24 * 30))
440 function poco_contact_from_body($body, $created, $cid, $uid) {
441 preg_replace_callback("/\[share(.*?)\].*?\[\/share\]/ism",
442 function ($match) use ($created, $cid, $uid){
443 return(sub_poco_from_share($match, $created, $cid, $uid));
447 function sub_poco_from_share($share, $created, $cid, $uid) {
449 preg_match("/profile='(.*?)'/ism", $share[1], $matches);
450 if ($matches[1] != "")
451 $profile = $matches[1];
453 preg_match('/profile="(.*?)"/ism', $share[1], $matches);
454 if ($matches[1] != "")
455 $profile = $matches[1];
460 logger("prepare poco_check for profile ".$profile, LOGGER_DEBUG);
461 poco_check($profile, "", "", "", "", "", "", "", "", $created, 3, $cid, $uid);
464 function poco_store($item) {
467 if ($item['private'])
470 // Or is it from a network where we don't store the global contacts?
471 if (!in_array($item["network"], array(NETWORK_DFRN, NETWORK_DIASPORA, NETWORK_OSTATUS, NETWORK_STATUSNET, "")))
474 // Is it a global copy?
475 $store_gcontact = ($item["uid"] == 0);
477 // Is it a comment on a global copy?
478 if (!$store_gcontact AND ($item["uri"] != $item["parent-uri"])) {
479 $q = q("SELECT `id` FROM `item` WHERE `uri`='%s' AND `uid` = 0", $item["parent-uri"]);
480 $store_gcontact = count($q);
483 if (!$store_gcontact)
486 // "3" means: We don't know this contact directly (Maybe a reshared item)
489 $profile_url = $item["author-link"];
491 // Is it a user from our server?
492 $q = q("SELECT `id` FROM `contact` WHERE `self` AND `nurl` = '%s' LIMIT 1",
493 dbesc(normalise_link($item["author-link"])));
495 logger("Our user (generation 1): ".$item["author-link"], LOGGER_DEBUG);
497 $network = NETWORK_DFRN;
498 } else { // Is it a contact from a user on our server?
499 $q = q("SELECT `network`, `url` FROM `contact` WHERE `uid` != 0 AND `network` != ''
500 AND (`nurl` = '%s' OR `alias` IN ('%s', '%s')) AND `network` != '%s' LIMIT 1",
501 dbesc(normalise_link($item["author-link"])),
502 dbesc(normalise_link($item["author-link"])),
503 dbesc($item["author-link"]),
504 dbesc(NETWORK_STATUSNET));
507 $network = $q[0]["network"];
508 $profile_url = $q[0]["url"];
509 logger("Known contact (generation 2): ".$profile_url, LOGGER_DEBUG);
513 if ($generation == 3)
514 logger("Unknown contact (generation 3): ".$item["author-link"], LOGGER_DEBUG);
516 poco_check($profile_url, $item["author-name"], $network, $item["author-avatar"], "", "", "", "", "", $item["received"], $generation, $item["contact-id"], $item["uid"]);
518 // Maybe its a body with a shared item? Then extract a global contact from it.
519 poco_contact_from_body($item["body"], $item["received"], $item["contact-id"], $item["uid"]);
522 function count_common_friends($uid,$cid) {
524 $r = q("SELECT count(*) as `total`
525 FROM `glink` INNER JOIN `gcontact` on `glink`.`gcid` = `gcontact`.`id`
526 where `glink`.`cid` = %d and `glink`.`uid` = %d
527 and `gcontact`.`nurl` in (select nurl from contact where uid = %d and self = 0 and blocked = 0 and hidden = 0 and id != %d ) ",
534 // logger("count_common_friends: $uid $cid {$r[0]['total']}");
536 return $r[0]['total'];
542 function common_friends($uid,$cid,$start = 0,$limit=9999,$shuffle = false) {
545 $sql_extra = " order by rand() ";
547 $sql_extra = " order by `gcontact`.`name` asc ";
549 $r = q("SELECT `gcontact`.*
550 FROM `glink` INNER JOIN `gcontact` on `glink`.`gcid` = `gcontact`.`id`
551 where `glink`.`cid` = %d and `glink`.`uid` = %d
552 and `gcontact`.`nurl` in (select nurl from contact where uid = %d and self = 0 and blocked = 0 and hidden = 0 and id != %d )
553 $sql_extra limit %d, %d",
567 function count_common_friends_zcid($uid,$zcid) {
569 $r = q("SELECT count(*) as `total`
570 FROM `glink` INNER JOIN `gcontact` on `glink`.`gcid` = `gcontact`.`id`
571 where `glink`.`zcid` = %d
572 and `gcontact`.`nurl` in (select nurl from contact where uid = %d and self = 0 and blocked = 0 and hidden = 0 ) ",
578 return $r[0]['total'];
583 function common_friends_zcid($uid,$zcid,$start = 0, $limit = 9999,$shuffle = false) {
586 $sql_extra = " order by rand() ";
588 $sql_extra = " order by `gcontact`.`name` asc ";
590 $r = q("SELECT `gcontact`.*
591 FROM `glink` INNER JOIN `gcontact` on `glink`.`gcid` = `gcontact`.`id`
592 where `glink`.`zcid` = %d
593 and `gcontact`.`nurl` in (select nurl from contact where uid = %d and self = 0 and blocked = 0 and hidden = 0 )
594 $sql_extra limit %d, %d",
606 function count_all_friends($uid,$cid) {
608 $r = q("SELECT count(*) as `total`
609 FROM `glink` INNER JOIN `gcontact` on `glink`.`gcid` = `gcontact`.`id`
610 where `glink`.`cid` = %d and `glink`.`uid` = %d ",
616 return $r[0]['total'];
622 function all_friends($uid,$cid,$start = 0, $limit = 80) {
624 $r = q("SELECT `gcontact`.*
625 FROM `glink` INNER JOIN `gcontact` on `glink`.`gcid` = `gcontact`.`id`
626 where `glink`.`cid` = %d and `glink`.`uid` = %d
627 order by `gcontact`.`name` asc LIMIT %d, %d ",
639 function suggestion_query($uid, $start = 0, $limit = 80) {
644 $network = array(NETWORK_DFRN);
646 if (get_config('system','diaspora_enabled'))
647 $network[] = NETWORK_DIASPORA;
649 if (!get_config('system','ostatus_disabled'))
650 $network[] = NETWORK_OSTATUS;
652 $sql_network = implode("', '", $network);
653 //$sql_network = "'".$sql_network."', ''";
654 $sql_network = "'".$sql_network."'";
656 $r = q("SELECT count(glink.gcid) as `total`, gcontact.* from gcontact
657 INNER JOIN glink on glink.gcid = gcontact.id
658 where uid = %d and not gcontact.nurl in ( select nurl from contact where uid = %d )
659 and not gcontact.name in ( select name from contact where uid = %d )
660 and not gcontact.id in ( select gcid from gcign where uid = %d )
661 AND `gcontact`.`updated` != '0000-00-00 00:00:00'
662 AND `gcontact`.`last_contact` >= `gcontact`.`last_failure`
663 AND `gcontact`.`network` IN (%s)
664 group by glink.gcid order by gcontact.updated desc,total desc limit %d, %d ",
674 if(count($r) && count($r) >= ($limit -1))
677 $r2 = q("SELECT gcontact.* from gcontact
678 INNER JOIN glink on glink.gcid = gcontact.id
679 where glink.uid = 0 and glink.cid = 0 and glink.zcid = 0 and not gcontact.nurl in ( select nurl from contact where uid = %d )
680 and not gcontact.name in ( select name from contact where uid = %d )
681 and not gcontact.id in ( select gcid from gcign where uid = %d )
682 AND `gcontact`.`updated` != '0000-00-00 00:00:00'
683 AND `gcontact`.`network` IN (%s)
684 order by rand() limit %d, %d ",
694 foreach ($r2 AS $suggestion)
695 $list[$suggestion["nurl"]] = $suggestion;
697 foreach ($r AS $suggestion)
698 $list[$suggestion["nurl"]] = $suggestion;
703 function update_suggestions() {
709 poco_load(0,0,0,$a->get_baseurl() . '/poco');
711 $done[] = $a->get_baseurl() . '/poco';
713 if(strlen(get_config('system','directory_submit_url'))) {
714 $x = fetch_url('http://dir.friendica.com/pubsites');
716 $j = json_decode($x);
718 foreach($j->entries as $entry) {
719 $url = $entry->url . '/poco';
720 if(! in_array($url,$done))
721 poco_load(0,0,0,$entry->url . '/poco');
727 $r = q("select distinct(poco) as poco from contact where network = '%s'",
733 $base = substr($rr['poco'],0,strrpos($rr['poco'],'/'));
734 if(! in_array($base,$done))
735 poco_load(0,0,0,$base);