Merge pull request #2424 from annando/1603-scrape-alias

Scrape: Always take the first alias
This commit is contained in:
Tobias Diekershoff 2016-04-02 09:48:31 +02:00
commit c30306e085

View File

@ -365,6 +365,7 @@ function probe_url($url, $mode = PROBE_NORMAL, $level = 1) {
return $result; return $result;
} }
$original_url = $url;
$network = null; $network = null;
$diaspora = false; $diaspora = false;
$diaspora_base = ''; $diaspora_base = '';
@ -393,6 +394,11 @@ function probe_url($url, $mode = PROBE_NORMAL, $level = 1) {
else else
$links = lrdd($url); $links = lrdd($url);
if ((count($links) == 0) AND strstr($url, "/index.php")) {
$url = str_replace("/index.php", "", $url);
$links = lrdd($url);
}
if (count($links)) { if (count($links)) {
$has_lrdd = true; $has_lrdd = true;
@ -440,12 +446,21 @@ function probe_url($url, $mode = PROBE_NORMAL, $level = 1) {
// aliases, let's hope we're lucky and get one that matches the feed author-uri because // aliases, let's hope we're lucky and get one that matches the feed author-uri because
// otherwise we're screwed. // otherwise we're screwed.
$backup_alias = "";
foreach($links as $link) { foreach($links as $link) {
if($link['@attributes']['rel'] === 'alias') { if($link['@attributes']['rel'] === 'alias') {
if(strpos($link['@attributes']['href'],'@') === false) { if(strpos($link['@attributes']['href'],'@') === false) {
if(isset($profile)) { if(isset($profile)) {
if($link['@attributes']['href'] !== $profile) $alias_url = $link['@attributes']['href'];
$alias = unamp($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 else
$profile = unamp($link['@attributes']['href']); $profile = unamp($link['@attributes']['href']);
@ -453,6 +468,9 @@ function probe_url($url, $mode = PROBE_NORMAL, $level = 1) {
} }
} }
if ($alias == "")
$alias = $backup_alias;
// If the profile is different from the url then the url is abviously an alias // If the profile is different from the url then the url is abviously an alias
if (($alias == "") AND ($profile != "") AND !$at_addr AND (normalise_link($profile) != normalise_link($url))) if (($alias == "") AND ($profile != "") AND !$at_addr AND (normalise_link($profile) != normalise_link($url)))
$alias = $url; $alias = $url;
@ -769,6 +787,9 @@ function probe_url($url, $mode = PROBE_NORMAL, $level = 1) {
if (($baseurl == "") AND ($poll != "")) if (($baseurl == "") AND ($poll != ""))
$baseurl = matching_url(normalise_link($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);
$baseurl = rtrim($baseurl, "/"); $baseurl = rtrim($baseurl, "/");
if(strpos($url,'@') AND ($addr == "") AND ($network == NETWORK_DFRN)) if(strpos($url,'@') AND ($addr == "") AND ($network == NETWORK_DFRN))
@ -816,8 +837,24 @@ function probe_url($url, $mode = PROBE_NORMAL, $level = 1) {
} }
// Only store into the cache if the value seems to be valid // Only store into the cache if the value seems to be valid
if ($result['network'] != NETWORK_PHANTOM) if ($result['network'] != NETWORK_PHANTOM) {
Cache::set("probe_url:".$mode.":".$url,serialize($result), CACHE_DAY); 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["addr"] AND $result["name"] AND $result["nick"])
q("UPDATE `contact` SET `addr` = '%s', `alias` = '%s', `name` = '%s', `nick` = '%s',
`success_update` = '%s' WHERE `nurl` = '%s' AND NOT `self` AND `uid` = 0",
dbesc($result["addr"]),
dbesc($result["alias"]),
dbesc($result["name"]),
dbesc($result["nick"]),
dbesc(datetime_convert()),
dbesc(normalise_link($result['url']))
);
}
return $result; return $result;
} }