ids.translate_handle: convert _ and ~ chars to - for Bluesky

hopefully fixes #982

atproto.py

@@ -258,14 +258,13 @@ def create_for(cls, user):
         # PDS URL shouldn't include trailing slash!
         # https://atproto.com/specs/did#did-documents
         pds_url = common.host_url().rstrip('/') if DEBUG else cls.PDS_URL
-        logger.info(f'Creating new did:plc for {user.key} {pds_url}')
-        did_plc = did.create_plc(user.handle_as('atproto'),
-                                 pds_url=pds_url, post_fn=util.requests_post)
+        handle = user.handle_as('atproto')
+        logger.info(f'Creating new did:plc for {user.key} {handle} {pds_url}')
+        did_plc = did.create_plc(handle, pds_url=pds_url, post_fn=util.requests_post)
 
         Object.get_or_create(did_plc.did, raw=did_plc.doc)
         # TODO: move this to ATProto.get_or_create?
         add(user.copies, Target(uri=did_plc.did, protocol='atproto'))
-        handle = user.handle_as('atproto')
 
         # create _atproto DNS record for handle resolution
         # https://atproto.com/specs/handle#handle-resolution

ids.py

@@ -32,6 +32,17 @@
 # Maps string domain to string subdomain (bsky, fed, or web).
 _NON_WEB_SUBDOMAIN_SITES = None
 
+# Webfinger allows all sorts of characters that ATProto handles don't,
+# notably _ and ~. Map those to -.
+# https://www.rfc-editor.org/rfc/rfc7565.html#section-7
+# https://atproto.com/specs/handle
+# https://github.com/snarfed/bridgy-fed/issues/982
+# https://github.com/swicg/activitypub-webfinger/issues/9
+TO_ATPROTO_CHARS = {
+    '_': '-',
+    '~': '-',
+}
+
 
 def web_ap_base_domain(user_domain):
     """Returns the full Bridgy Fed domain to use for a given Web user.
@@ -167,6 +178,10 @@ def translate_handle(*, handle, from_, to, enhanced):
             return f'@{handle}@{domain}'
 
         case _, 'atproto' | 'nostr':
+            if to.LABEL == 'atproto':
+                for from_char, to_char in TO_ATPROTO_CHARS.items():
+                    handle = handle.replace(from_char, to_char)
+
             handle = handle.lstrip('@').replace('@', '.')
             if enhanced or handle == PRIMARY_DOMAIN or handle in PROTOCOL_DOMAINS:
                 return handle

tests/test_ids.py

@@ -120,14 +120,17 @@ def test_translate_handle(self):
             (Web, 'user.com', ActivityPub, '@user.com@web.brid.gy'),
             (Web, 'user.com', ATProto, 'user.com.web.brid.gy'),
             (Web, 'user.com', Fake, 'fake:handle:user.com'),
+            (Web, 'u_se-r.com', Fake, 'fake:handle:u_se-r.com'),
             (Web, 'user.com', Web, 'user.com'),
 
             (ActivityPub, '@user@instance', ActivityPub, '@user@instance'),
             (ActivityPub, '@user@instance', ATProto, 'user.instance.ap.brid.gy'),
+            (ActivityPub, '@u_se~r@instance', ATProto, 'u-se-r.instance.ap.brid.gy'),
             (ActivityPub, '@user@instance', Fake, 'fake:handle:@user@instance'),
             (ActivityPub, '@user@instance', Web, 'https://instance/@user'),
 
             (ATProto, 'user.com', ActivityPub, '@user.com@bsky.brid.gy'),
+            (ATProto, 'u-se-r.com', ActivityPub, '@u-se-r.com@bsky.brid.gy'),
             (ATProto, 'user.com', ATProto, 'user.com'),
             (ATProto, 'user.com', Fake, 'fake:handle:user.com'),
             (ATProto, 'user.com', Web, 'user.com'),

web.py

@@ -26,7 +26,7 @@
 import common
 from common import add, DOMAIN_RE, PRIMARY_DOMAIN, PROTOCOL_DOMAINS, SUPERDOMAIN
 from flask_app import app, cache
-from ids import translate_handle, translate_object_id, translate_user_id
+from ids import translate_object_id, translate_user_id
 from models import Follower, Object, PROTOCOLS, Target, User
 from protocol import Protocol