fix_channel_import_content #8

streams commented 2025年03月05日 21:20:30 +01:00

This line is probably not needed. In HTTPSig.php::L186-181 the query and fragment are filtered out.

This line is probably not needed. In HTTPSig.php::L186-181 the query and fragment are filtered out.

streams commented 2025年03月05日 22:06:53 +01:00

$r['hubloc'] should return a single hubloc record; the best identity we can use. Originally this was the case with zot/nomad where a single actor could have records for identities on multiple networks. In forte we no longer check for zot/nomad, so the 'best' record will be one with hubloc_network = 'nomadic' -- if it's available.

The solution here is to fix hubloc_id_addr_query() so it will return the correct hubloc record(s) for the keyId, which should be the case already. Ignore the cached entry from get_webfinger_key(), which may not exist depending on the value of $force. If there isn't a hubloc after doing the webfinger lookup, we don't have the record and couldn't discover it.

The issue here is that we may have an apgateway address (https://$something/.well-known/apgateway/did:key:....), and hubloc_id_addr_query() won't resolve it. If this happens, hubloc_id_addr_query() needs to be fixed. The ActorId class can be used to normalize the id we received and $actorId->getId() will return either a URL or a DID. Then you might end up with something like this:

function hubloc_id_addr_query($s, $limit = 0) {
 if ($limit) {
 $qlimit = 'limit ' . intval($limit);
 }
 $actorId = new ActorId($s);
 $id = $actorId->getId();
 $r = q(
 "select * from xchan left join hubloc on xchan_hash = hubloc_hash where ( hubloc_addr = '%s' or hubloc_id_url = '%s' or hubloc_hash = '%s' or xchan_epubkey = '%s') and hubloc_deleted = 0 order by hubloc_id desc $qlimit",
 dbesc(str_replace('acct:', '', $s)), // we still need to match the non-normalised id against webfinger addresses
 dbesc($id), // use the results from ActorId for any other id forms. 
 dbesc($id),
 dbesc(str_replace('did:key:', '', $id)) // this is a fallback for some special cases where the nomadic id is in transition
 );
 return $r;
 }

This might need a bit of additional tweaking, but should return all the hublocs for the record in nearly cases.

$r['hubloc'] should return a single hubloc record; the best identity we can use. Originally this was the case with zot/nomad where a single actor could have records for identities on multiple networks. In forte we no longer check for zot/nomad, so the 'best' record will be one with hubloc_network = 'nomadic' -- if it's available. The solution here is to fix hubloc_id_addr_query() so it will return the correct hubloc record(s) for the keyId, which should be the case already. Ignore the cached entry from get_webfinger_key(), which may not exist depending on the value of $force. If there isn't a hubloc after doing the webfinger lookup, we don't have the record and couldn't discover it. The issue here is that we may have an apgateway address (https://$something/.well-known/apgateway/did:key:....), and hubloc_id_addr_query() won't resolve it. If this happens, hubloc_id_addr_query() needs to be fixed. The ActorId class can be used to normalize the id we received and $actorId->getId() will return either a URL or a DID. Then you might end up with something like this: ``` function hubloc_id_addr_query($s, $limit = 0) { if ($limit) { $qlimit = 'limit ' . intval($limit); } $actorId = new ActorId($s); $id = $actorId->getId(); $r = q( "select * from xchan left join hubloc on xchan_hash = hubloc_hash where ( hubloc_addr = '%s' or hubloc_id_url = '%s' or hubloc_hash = '%s' or xchan_epubkey = '%s') and hubloc_deleted = 0 order by hubloc_id desc $qlimit", dbesc(str_replace('acct:', '', $s)), // we still need to match the non-normalised id against webfinger addresses dbesc($id), // use the results from ActorId for any other id forms. dbesc($id), dbesc(str_replace('did:key:', '', $id)) // this is a fallback for some special cases where the nomadic id is in transition ); return $r; } ``` This might need a bit of additional tweaking, but should return all the hublocs for the record in nearly cases.

streams commented 2025年03月05日 22:09:13 +01:00

With the above fixes to hubloc_id_addr_query(), the original signer should work here.

With the above fixes to hubloc_id_addr_query(), the original signer should work here.

streams commented 2025年03月05日 22:14:16 +01:00

With the above changes, this function should be reverted. It has been working fine for a long time. You can also revert get_zotfinger_key() - which shouldn't be used any more. I'll remove it in a future update.

With the above changes, this function should be reverted. It has been working fine for a long time. You can also revert get_zotfinger_key() - which shouldn't be used any more. I'll remove it in a future update.