Merge e5df116765 into 0ed24c19b3

Bumps [idna](https://github.com/kjd/idna) from 3.6 to 3.7.
- [Release notes](https://github.com/kjd/idna/releases)
- [Changelog](https://github.com/kjd/idna/blob/master/HISTORY.rst)
- [Commits](https://github.com/kjd/idna/compare/v3.6...v3.7)

---
updated-dependencies:
- dependency-name: idna
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>

activitypub.py

@@ -54,6 +54,8 @@ _INSTANCE_ACTOR = None
 # populated in User.status
 WEB_OPT_OUT_DOMAINS = None
 
+FEDI_URL_RE = re.compile(r'https://[^/]+/(@|users/)([^/@]+)(@[^/@]+)?(/(?:statuses/)?[0-9]+)?')
+
 
 def instance_actor():
     global _INSTANCE_ACTOR
@@ -226,8 +228,8 @@ class ActivityPub(User, Protocol):
             logger.info(f'Skipping sending to blocklisted {url}')
             return False
 
-        activity = to_cls.convert(obj, from_user=from_user,
-                                  orig_obj=to_cls.convert(orig_obj))
+        orig_obj = to_cls.convert(orig_obj, from_user=from_user)
+        activity = to_cls.convert(obj, from_user=from_user, orig_obj=orig_obj)
 
         return signed_post(url, data=activity, from_user=from_user).ok
 
@@ -359,7 +361,8 @@ class ActivityPub(User, Protocol):
             return {}
 
         from_proto = PROTOCOLS.get(obj.source_protocol)
-        if from_proto and not common.is_enabled(cls, from_proto):
+        user_id = from_user.key.id() if from_user and from_user.key else None
+        if from_proto and not common.is_enabled(cls, from_proto, handle_or_id=user_id):
             error(f'{cls.LABEL} <=> {from_proto.LABEL} not enabled')
 
         if obj.as2:
@@ -828,7 +831,7 @@ def actor(handle_or_id):
     cls = Protocol.for_request(fed='web')
     if not cls:
         error(f"Couldn't determine protocol", status=404)
-    elif not common.is_enabled(cls, ActivityPub):
+    elif not common.is_enabled(cls, ActivityPub, handle_or_id=handle_or_id):
         error(f'{cls.LABEL} <=> activitypub not enabled')
     elif cls.LABEL == 'web' and request.path.startswith('/ap/'):
         # we started out with web users' AP ids as fed.brid.gy/[domain], so we

atproto.py

@@ -43,7 +43,18 @@ logger = logging.getLogger(__name__)
 
 arroba.server.storage = DatastoreStorage(ndb_client=ndb_client)
 
-LEXICONS = Client('https://unused').defs
+appview = Client(f'https://{os.environ["APPVIEW_HOST"]}',
+                 headers={'User-Agent': USER_AGENT})
+LEXICONS = appview.defs
+
+# https://atproto.com/guides/applications#record-types
+COLLECTION_TO_TYPE = {
+  'app.bsky.actor.profile': 'profile',
+  'app.bsky.feed.like': 'like',
+  'app.bsky.feed.post': 'post',
+  'app.bsky.feed.repost': 'repost',
+  'app.bsky.graph.follow': 'follow',
+}
 
 DNS_GCP_PROJECT = 'brid-gy'
 DNS_ZONE = 'brid-gy'
@@ -52,6 +63,22 @@ logger.info(f'Using GCP DNS project {DNS_GCP_PROJECT} zone {DNS_ZONE}')
 dns_client = dns.Client(project=DNS_GCP_PROJECT)
 
 
+def did_to_handle(did):
+    """Resolves a DID to a handle _if_ we have the DID doc stored locally.
+
+    Args:
+      did (str)
+
+    Returns:
+      str: handle, or None
+    """
+    if did_obj := ATProto.load(did, did_doc=True):
+        if aka := util.get_first(did_obj.raw, 'alsoKnownAs', ''):
+            handle, _, _ = parse_at_uri(aka)
+            if handle:
+                return handle
+
+
 class ATProto(User, Protocol):
     """AT Protocol class.
 
@@ -59,6 +86,7 @@ class ATProto(User, Protocol):
     https://atproto.com/specs/did
     """
     ABBREV = 'atproto'
+    # TODO: add second bsky label? inject into PROTOCOLS?
     PHRASE = 'Bluesky'
     LOGO_HTML = '<img src="/static/atproto_logo.png">'
     PDS_URL = f'https://{ABBREV}{common.SUPERDOMAIN}/'
@@ -82,11 +110,7 @@ class ATProto(User, Protocol):
     @ndb.ComputedProperty
     def handle(self):
         """Returns handle if the DID document includes one, otherwise None."""
-        if did_obj := ATProto.load(self.key.id(), did_doc=True):
-            if aka := util.get_first(did_obj.raw, 'alsoKnownAs', ''):
-                handle, _, _ = parse_at_uri(aka)
-                if handle:
-                    return handle
+        return did_to_handle(self.key.id())
 
     def web_url(self):
         return bluesky.Bluesky.user_url(self.handle_or_id())
@@ -243,7 +267,7 @@ class ATProto(User, Protocol):
         initial_writes = None
         if user.obj and user.obj.as1:
             # create user profile
-            profile = cls.convert(user.obj, fetch_blobs=True)
+            profile = cls.convert(user.obj, fetch_blobs=True, from_user=user)
             profile.setdefault('labels', {'$type': 'com.atproto.label.defs#selfLabels'})
             profile['labels'].setdefault('values', []).append({
                 'val' : f'bridged-from-bridgy-fed-{user.LABEL}',
@@ -302,7 +326,7 @@ class ATProto(User, Protocol):
 
         # convert to Bluesky record; short circuits on error
         try:
-            record = to_cls.convert(base_obj, fetch_blobs=True)
+            record = to_cls.convert(base_obj, fetch_blobs=True, from_user=from_user)
         except ValueError as e:
             logger.info(f'Skipping due to {e}')
             return False
@@ -354,8 +378,7 @@ class ATProto(User, Protocol):
                 assert copy_did == did
                 assert coll == type
 
-            logger.info(f'Storing ATProto {action} {type} {rkey}: ' +
-                        json_dumps(dag_json.encode(record).decode(), indent=2))
+            logger.info(f'Storing ATProto {action} {type} {rkey}: {dag_json.encode(record).decode()}')
             repo.apply_writes([Write(action=action, collection=type, rkey=rkey,
                                      record=record)])
 
@@ -431,10 +454,8 @@ class ATProto(User, Protocol):
             assert repo.startswith('did:')
             obj.key = ndb.Key(Object, id.replace(f'at://{handle}', f'at://{repo}'))
 
-        client = Client(f'https://{os.environ["APPVIEW_HOST"]}',
-                        headers={'User-Agent': USER_AGENT})
         try:
-            ret = client.com.atproto.repo.getRecord(
+            ret = appview.com.atproto.repo.getRecord(
                 repo=repo, collection=collection, rkey=rkey)
         except RequestException as e:
             util.interpret_http_exception(e)
@@ -462,7 +483,8 @@ class ATProto(User, Protocol):
           dict: JSON object
         """
         from_proto = PROTOCOLS.get(obj.source_protocol)
-        if from_proto and not common.is_enabled(cls, from_proto):
+        user_id = from_user.key.id() if from_user and from_user.key else None
+        if from_proto and not common.is_enabled(cls, from_proto, handle_or_id=user_id):
             error(f'{cls.LABEL} <=> {from_proto.LABEL} not enabled')
 
         if obj.bsky:
@@ -481,6 +503,11 @@ class ATProto(User, Protocol):
                         blobs[url] = blob.as_object()
 
         ret = bluesky.from_as1(cls.translate_ids(obj.as1), blobs=blobs)
+            # TODO: uncomment this and pass through client eventually? would be
+            # nice to start reusing granary's resolving handles and CIDs, but we
+            # do much of that ourselves here in BF beforehand, so granary ends
+            # up duplicating those network requests
+            # client=appview)
 
         # fill in CIDs from Objects
         def populate_cid(strong_ref):
@@ -521,6 +548,8 @@ def poll_notifications():
                               for cls in set(PROTOCOLS.values())
                               if cls and cls != ATProto))
 
+    # this client needs to be request-local because we set its service token
+    # below per user that we're polling
     client = Client(f'https://{os.environ["APPVIEW_HOST"]}',
                     headers={'User-Agent': USER_AGENT})
 
@@ -579,6 +608,8 @@ def poll_posts():
                               for cls in set(PROTOCOLS.values())
                               if cls and cls != ATProto))
 
+    # this client needs to be request-local because we set its service token
+    # below per user that we're polling
     client = Client(f'https://{os.environ["APPVIEW_HOST"]}',
                     headers={'User-Agent': USER_AGENT})
 

common.py

@@ -83,6 +83,15 @@ util.set_user_agent(USER_AGENT)
 TASKS_LOCATION = 'us-central1'
 RUN_TASKS_INLINE = False  # overridden by unit tests
 
+USER_ALLOWLIST = (
+    'snarfed.org',
+    'did:plc:fdme4gb7mu7zrie7peay7tst',
+    'snarfed.bsky.social',
+    'did:plc:3ljmtyyjqcjee2kpewgsifvb',
+    'https://indieweb.social/users/snarfed',
+    '@snarfed@indieweb.social',
+)
+
 
 def base64_to_long(x):
     """Converts from URL safe base64 encoding to long integer.
@@ -255,12 +264,13 @@ def add(seq, val):
         seq.append(val)
 
 
-def is_enabled(proto_a, proto_b):
+def is_enabled(proto_a, proto_b, handle_or_id=None):
     """Returns True if bridging the two input protocols is enabled, False otherwise.
 
     Args:
       proto_a (Protocol subclass)
       proto_b (Protocol subclass)
+      handle_or_id (str): optional user handle or id
 
     Returns:
       bool:
@@ -273,6 +283,9 @@ def is_enabled(proto_a, proto_b):
     if DEBUG and ('fake' in labels or 'other' in labels):
         return True
 
+    if handle_or_id in USER_ALLOWLIST:
+        return True
+
     return labels in ENABLED_BRIDGES
 
 

ids.py

@@ -8,6 +8,7 @@ from urllib.parse import urljoin, urlparse
 
 from flask import request
 from google.cloud.ndb.query import FilterNode, Query
+from granary.bluesky import BSKY_APP_URL_RE, web_url_to_at_uri
 from oauth_dropins.webutil import util
 
 from common import subdomain_wrap, LOCAL_DOMAINS, PRIMARY_DOMAIN, SUPERDOMAIN
@@ -24,7 +25,7 @@ _FED_SUBDOMAIN_SITES = None
 
 
 def web_ap_base_domain(user_domain):
-    """Returns the full Bridgy Fed domain to user for a given Web user.
+    """Returns the full Bridgy Fed domain to use for a given Web user.
 
     Specifically, returns ``http://localhost/` if we're running locally,
     ``https://fed.brid.gy/`` if the given Web user has ``ap_subdomain='fed'``,
@@ -55,6 +56,8 @@ def web_ap_base_domain(user_domain):
 def translate_user_id(*, id, from_proto, to_proto):
     """Translate a user id from one protocol to another.
 
+    TODO: unify with :func:`translate_object_id`.
+
     Args:
       id (str)
       from_proto (protocol.Protocol)
@@ -72,6 +75,20 @@ def translate_user_id(*, id, from_proto, to_proto):
         # home page; replace with domain
         id = parsed.netloc
 
+    # bsky.app profile URL to DID
+    if to_proto.LABEL == 'atproto':
+        if match := BSKY_APP_URL_RE.match(id):
+            repo = match.group('id')
+            if repo.startswith('did:'):
+                return repo
+
+            from atproto import ATProto
+            try:
+                return ATProto.handle_to_id(repo)
+            except (AssertionError, ValueError) as e:
+                logger.warning(e)
+                return None
+
     if from_proto == to_proto:
         return id
 
@@ -161,6 +178,8 @@ def translate_handle(*, handle, from_proto, to_proto, enhanced):
 def translate_object_id(*, id, from_proto, to_proto):
     """Translates a user handle from one protocol to another.
 
+    TODO: unify with :func:`translate_user_id`.
+
     Args:
       id (str)
       from_proto (protocol.Protocol)
@@ -172,6 +191,22 @@ def translate_object_id(*, id, from_proto, to_proto):
     assert id and from_proto and to_proto
     assert from_proto.owns_id(id) is not False or from_proto.LABEL == 'ui'
 
+    # bsky.app profile URL to DID
+    if to_proto.LABEL == 'atproto':
+        if match := BSKY_APP_URL_RE.match(id):
+            repo = match.group('id')
+            handle = None
+            if not repo.startswith('did:'):
+                handle = repo
+                from atproto import ATProto
+                try:
+                    repo = ATProto.handle_to_id(repo)
+                except (AssertionError, ValueError) as e:
+                    logger.warning(e)
+                    return None
+
+            return web_url_to_at_uri(id, handle=handle, did=repo)
+
     if from_proto == to_proto:
         return id
 

models.py

@@ -15,7 +15,7 @@ from Crypto.PublicKey import RSA
 from flask import g, request
 from google.cloud import ndb
 from granary import as1, as2, atom, bluesky, microformats2
-from granary.bluesky import BSKY_APP_URL_RE
+from granary.bluesky import AT_URI_PATTERN, BSKY_APP_URL_RE
 from granary.source import html_to_text
 from oauth_dropins.webutil import util
 from oauth_dropins.webutil.appengine_info import DEBUG
@@ -28,8 +28,8 @@ from oauth_dropins.webutil.models import (
 from oauth_dropins.webutil.util import json_dumps, json_loads
 
 import common
-from common import add, base64_to_long, long_to_base64, unwrap
-import ids
+from common import add, base64_to_long, DOMAIN_RE, long_to_base64, unwrap
+from ids import translate_handle, translate_object_id, translate_user_id
 
 # maps string label to Protocol subclass. populated by ProtocolUserMeta.
 # seed with old and upcoming protocols that don't have their own classes (yet).
@@ -241,7 +241,7 @@ class User(StringIdModel, metaclass=ProtocolUserMeta):
 
         ATProto = PROTOCOLS['atproto']
         if propagate and cls.LABEL != 'atproto' and not user.get_copy(ATProto):
-            if common.is_enabled(cls, ATProto):
+            if common.is_enabled(cls, ATProto, handle_or_id=id):
                 ATProto.create_for(user)
             else:
                 logger.info(f'{cls.LABEL} <=> atproto not enabled, skipping')
@@ -359,8 +359,8 @@ class User(StringIdModel, metaclass=ProtocolUserMeta):
         if not handle:
             return None
 
-        return ids.translate_handle(handle=handle, from_proto=self.__class__,
-                                    to_proto=to_proto, enhanced=False)
+        return translate_handle(handle=handle, from_proto=self.__class__,
+                                to_proto=to_proto, enhanced=False)
 
     def id_as(self, to_proto):
         """Returns this user's id in a different protocol.
@@ -374,8 +374,8 @@ class User(StringIdModel, metaclass=ProtocolUserMeta):
         if isinstance(to_proto, str):
             to_proto = PROTOCOLS[to_proto]
 
-        return ids.translate_user_id(id=self.key.id(), from_proto=self.__class__,
-                                     to_proto=to_proto)
+        return translate_user_id(id=self.key.id(), from_proto=self.__class__,
+                                 to_proto=to_proto)
 
     def handle_or_id(self):
         """Returns handle if we know it, otherwise id."""
@@ -606,7 +606,7 @@ class Object(StringIdModel):
         def use_urls_as_ids(obj):
             """If id field is missing or not a URL, use the url field."""
             id = obj.get('id')
-            if not id or not util.is_web(id):
+            if not id or not (util.is_web(id) or re.match(DOMAIN_RE, id)):
                 if url := util.get_url(obj):
                     obj['id'] = url
 
@@ -922,6 +922,8 @@ class Object(StringIdModel):
           ``at://did:plc:abc/app.bsky.feed.post/123`` => ``https://mas.to/@user/456``.
         * Bridgy Fed subdomain URLs to the ids embedded inside them, eg
           ``https://atproto.brid.gy/ap/did:plc:xyz`` => ``did:plc:xyz``
+        * ATProto bsky.app URLs to their DIDs or `at://` URIs, eg
+          ``https://bsky.app/profile/a.com`` => ``did:plc:123``
 
         ...in these AS1 fields, in place:
 
@@ -937,6 +939,9 @@ class Object(StringIdModel):
 
         :meth:`protocol.Protocol.translate_ids` is partly the inverse of this.
         Much of the same logic is duplicated there!
+
+        TODO: unify with :meth:`normalize_ids`,
+        :meth:`protocol.Protocol.normalize_ids`.
         """
         if not self.as1:
             return
@@ -974,7 +979,7 @@ class Object(StringIdModel):
                 if copy.protocol in (self_proto.LABEL, self_proto.ABBREV):
                     origs[copy.uri] = obj.key.id()
 
-        logger.debug(f'Replacing copies with originals: {origs}')
+        logger.debug(f'Resolving {self_proto.LABEL} ids; originals: {origs}')
         replaced = False
 
         def replace(val):
@@ -1002,12 +1007,87 @@ class Object(StringIdModel):
                     obj[field] = obj[field][0]
 
         outer_obj['object'] = replace(inner_obj)
+
         if util.trim_nulls(outer_obj['object']).keys() == {'id'}:
             outer_obj['object'] = outer_obj['object']['id']
 
         if replaced:
             self.our_as1 = util.trim_nulls(outer_obj)
 
+    def normalize_ids(self):
+        """Normalizes ids to their protocol's canonical representation, if any.
+
+        For example, normalizes ATProto ``https://bsky.app/...`` URLs to DIDs
+        for profiles, ``at://`` URIs for posts.
+
+        Modifies this object in place.
+
+        TODO: unify with :meth:`resolve_ids`, :meth:`Protocol.translate_ids`.
+        """
+        from protocol import Protocol
+
+        if not self.as1:
+            return
+
+        logger.debug(f'Normalizing ids')
+        outer_obj = copy.deepcopy(self.as1)
+        inner_objs = as1.get_objects(outer_obj)
+        replaced = False
+
+        def replace(val, translate_fn):
+            nonlocal replaced
+
+            orig = val.get('id') if isinstance(val, dict) else val
+            if not orig:
+                return val
+
+            proto = Protocol.for_id(orig, remote=False)
+            if not proto:
+                return val
+
+            translated = translate_fn(id=orig, from_proto=proto, to_proto=proto)
+            if translated and translated != orig:
+                logger.info(f'Normalized {proto.LABEL} id {orig} to {translated}')
+                replaced = True
+                if isinstance(val, dict):
+                    val['id'] = translated
+                    return val
+                else:
+                    return translated
+
+            return val
+
+        # actually replace ids
+        for obj in [outer_obj] + inner_objs:
+            for tag in as1.get_objects(obj, 'tags'):
+                if tag.get('objectType') == 'mention':
+                    tag['url'] = replace(tag.get('url'), translate_user_id)
+            for field in ['actor', 'author', 'inReplyTo']:
+                fn = translate_object_id if field == 'inReplyTo' else translate_user_id
+                obj[field] = [replace(val, fn) for val in util.get_list(obj, field)]
+                if len(obj[field]) == 1:
+                    obj[field] = obj[field][0]
+
+        outer_obj['object'] = []
+        for inner_obj in inner_objs:
+            translate_fn = (translate_user_id
+                            if (as1.object_type(inner_obj) in as1.ACTOR_TYPES
+                                or as1.object_type(outer_obj) in
+                                ('follow', 'stop-following'))
+                            else translate_object_id)
+
+            got = replace(inner_obj, translate_fn)
+            if isinstance(got, dict) and util.trim_nulls(got).keys() == {'id'}:
+                got = got['id']
+
+            outer_obj['object'].append(got)
+
+        if len(outer_obj['object']) == 1:
+            outer_obj['object'] = outer_obj['object'][0]
+
+        if replaced:
+            self.our_as1 = util.trim_nulls(outer_obj)
+
 
 class Follower(ndb.Model):
     """A follower of a Bridgy Fed user."""
@@ -1192,23 +1272,32 @@ def fetch_objects(query, by=None, user=None):
                 'url': id,
             })
         elif url:
-            # heuristics for sniffing Mastodon and similar fediverse URLs and
-            # converting them to more friendly @-names
+            # heuristics for sniffing URLs and converting them to more friendly
+            # phrases and user handles.
             # TODO: standardize this into granary.as2 somewhere?
             if not content:
-                fedi_url = re.match(
-                    r'https://[^/]+/(@|users/)([^/@]+)(@[^/@]+)?(/(?:statuses/)?[0-9]+)?', url)
-                if fedi_url:
-                    content = '@' + fedi_url.group(2)
-                    if fedi_url.group(4):
-                        content += "'s post"
+                from activitypub import FEDI_URL_RE
+                from atproto import COLLECTION_TO_TYPE, did_to_handle
 
-            if not content:
-                if bsky_url := BSKY_APP_URL_RE.match(url):
-                    if handle := bsky_url.group('id'):  # or DID
-                        content = '@' + handle
-                        if bsky_url.group('tid'):
-                            content += "'s post"
+                if match := FEDI_URL_RE.match(url):
+                    content = '@' + match.group(2)
+                    if match.group(4):
+                        content += "'s post"
+                elif match := BSKY_APP_URL_RE.match(url):
+                    id = match.group('id')
+                    if id.startswith('did:'):
+                        id = ATdid_to_handle(id) or id
+                    content = '@' + id
+                    if match.group('tid'):
+                        content += "'s post"
+                elif match := AT_URI_PATTERN.match(url):
+                    id = match.group('repo')
+                    if id.startswith('did:'):
+                        id = did_to_handle(id) or id
+                    content = '@' + id
+                    if coll := match.group('collection'):
+                        content += f"'s {COLLECTION_TO_TYPE.get(coll) or 'post'}"
+                    url = bluesky.at_uri_to_web_url(url)
 
             content = common.pretty_link(url, text=content, user=user)
 

pages.py

@@ -60,17 +60,16 @@ def load_user(protocol, id):
       :class:`werkzeug.exceptions.HTTPException` on error or redirect
     """
     assert id
-    if protocol == 'ap' and not id.startswith('@'):
-        id = '@' + id
 
     cls = PROTOCOLS[protocol]
+
+    if cls.ABBREV == 'ap' and not id.startswith('@'):
+        id = '@' + id
     user = cls.get_by_id(id)
 
-    if protocol != 'web':
+    if cls.ABBREV != 'web':
         if not user:
-            user = cls.query(OR(cls.handle == id,
-                                  cls.readable_id == id),
-                               ).get()
+            user = cls.query(OR(cls.handle == id, cls.handle == id)).get()
             if user and user.use_instead:
                 user = user.use_instead.get()
 
@@ -80,7 +79,7 @@ def load_user(protocol, id):
     elif user and id != user.key.id():  # use_instead redirect
         error('', status=302, location=user.user_page_path())
 
-    if user and (user.direct or protocol == 'web'):
+    if user and (user.direct or cls.ABBREV != 'ap'):
         assert not user.use_instead
         return user
 

protocol.py

@@ -226,19 +226,18 @@ class Protocol:
 
     @cached(LRUCache(20000), lock=Lock())
     @staticmethod
-    def for_id(id):
+    def for_id(id, remote=True):
         """Returns the protocol for a given id.
 
-        May incur expensive side effects like fetching the id itself over the
-        network or other discovery.
-
         Args:
           id (str)
+          remote (bool): whether to perform expensive side effects like fetching
+            the id itself over the network, or other discovery.
 
         Returns:
-          Protocol subclass: matching protocol, or None if no known protocol
-          owns this id
-       """
+          Protocol subclass: matching protocol, or None if no single known
+          protocol definitively owns this id
+        """
         logger.info(f'Determining protocol for id {id}')
         if not id:
             return None
@@ -273,7 +272,10 @@ class Protocol:
             logger.info(f'  {obj.key} owned by source_protocol {obj.source_protocol}')
             return PROTOCOLS[obj.source_protocol]
 
-        # step 4: fetch over the network
+        # step 4: fetch over the network, if necessary
+        if not remote:
+            return None
+
         for protocol in candidates:
             logger.info(f'Trying {protocol.LABEL}')
             try:
@@ -474,7 +476,7 @@ class Protocol:
 
     @classmethod
     def translate_ids(to_cls, obj):
-        """Wraps ids and actors in an AS1 object in subdomain convert URLs.
+        """Translates all ids in an AS1 object to a specific protocol.
 
         Infers source protocol for each id value separately.
 
@@ -497,6 +499,9 @@ class Protocol:
         This is the inverse of :meth:`models.Object.resolve_ids`. Much of the
         same logic is duplicated there!
 
+        TODO: unify with :meth:`Object.resolve_ids`,
+        :meth:`protocol.Protocol.normalize_ids`.
+
         Args:
           to_proto (Protocol subclass)
           obj (dict): AS1 object or activity (not :class:`models.Object`!)
@@ -792,6 +797,7 @@ class Protocol:
             from_user = from_cls.get_or_create(id=from_key.id(), obj=from_obj)
 
             logger.info(f'Follow {from_id} => {to_id}')
+
             to_cls = Protocol.for_id(to_id)
             if not to_cls:
                 error(f"Couldn't determine protocol for {to_id}")
@@ -1172,6 +1178,7 @@ class Protocol:
             return None
 
         obj.resolve_ids()
+        obj.normalize_ids()
 
         if obj.new is False:
             obj.changed = obj.activity_changed(orig_as1)

requirements.txt

@@ -15,7 +15,7 @@ beautifulsoup4==4.12.3
 bech32==1.2.0
 brevity==0.2.17
 cachetools==5.3.0
-cbor2==5.6.2
+cbor2==5.6.3
 certifi==2024.2.2
 cffi==1.16.0
 charset-normalizer==3.3.2
@@ -26,12 +26,12 @@ dag-cbor==0.3.3
 Deprecated==1.2.14
 dnspython==2.6.1
 domain2idna==1.12.0
-ecdsa==0.18.0
+ecdsa==0.19.0
 extras==1.0.0
 feedgen==1.0.0
 feedparser==6.0.11
 fixtures==4.1.0
-Flask==3.0.2
+Flask==3.0.3
 Flask-Caching==2.1.0
 flask-gae-static==1.0
 flask-sock==0.7.0
@@ -56,7 +56,7 @@ html2text==2024.2.26
 html5lib==1.1
 humanfriendly==10.0
 humanize==4.9.0
-idna==3.6
+idna==3.7
 iterators==0.2.0
 itsdangerous==2.1.2
 Jinja2==3.1.3
@@ -89,7 +89,7 @@ pytz==2024.1
 PyYAML==6.0.1
 redis==5.0.3
 requests==2.31.0
-requests-oauthlib==1.4.0
+requests-oauthlib==2.0.0
 rsa==4.9
 sgmllib3k==1.0.0
 simple-websocket==1.0.0

tests/test_atproto.py

@@ -9,7 +9,6 @@ from arroba.datastore_storage import AtpBlock, AtpRemoteBlob, AtpRepo, Datastore
 from arroba.did import encode_did_key
 from arroba.repo import Repo
 import arroba.util
-import dns.resolver
 from dns.resolver import NXDOMAIN
 from flask import g
 from google.cloud.tasks_v2.types import Task
@@ -137,7 +136,7 @@ class ATProtoTest(TestCase):
         self.make_user('did:plc:user', cls=ATProto)
         self.assertEqual('did:plc:user', ATProto.handle_to_id('han.dull'))
 
-    @patch('dns.resolver.resolve', side_effect=dns.resolver.NXDOMAIN())
+    @patch('dns.resolver.resolve', side_effect=NXDOMAIN())
     # resolving handle, HTTPS method, not found
     @patch('requests.get', return_value=requests_response('', status=404))
     def test_handle_to_id_not_found(self, *_):
@@ -192,7 +191,7 @@ class ATProtoTest(TestCase):
             id='https://bsky.app/profile/did:plc:user/post/123'))
         self.assertEqual('https://some.pds', got)
 
-    @patch('dns.resolver.resolve', side_effect=dns.resolver.NXDOMAIN())
+    @patch('dns.resolver.resolve', side_effect=NXDOMAIN())
     @patch('requests.get', side_effect=[
         # resolving handle, HTTPS method
         requests_response('did:plc:user', content_type='text/plain'),
@@ -297,7 +296,7 @@ class ATProtoTest(TestCase):
             with self.assertRaises(AssertionError):
                 ATProto.fetch(Object(id=uri))
 
-    @patch('dns.resolver.resolve', side_effect=dns.resolver.NXDOMAIN())
+    @patch('dns.resolver.resolve', side_effect=NXDOMAIN())
     @patch('requests.get', return_value=requests_response(status=404))
     def test_fetch_resolve_handle_fails(self, mock_get, _):
         obj = Object(id='at://bad.com/app.bsky.feed.post/789')
@@ -313,7 +312,7 @@ class ATProtoTest(TestCase):
                                     bsky=ACTOR_PROFILE_BSKY)
         self.assert_entities_equal(profile, ATProto.load('did:plc:user'))
 
-    @patch('dns.resolver.resolve', side_effect=dns.resolver.NXDOMAIN())
+    @patch('dns.resolver.resolve', side_effect=NXDOMAIN())
     @patch('requests.get', side_effect=[
         # resolving handle, HTTPS method
         requests_response('did:plc:user', content_type='text/plain'),
@@ -430,13 +429,13 @@ class ATProtoTest(TestCase):
             'inReplyTo': 'at://did:plc:bob/app.bsky.feed.post/tid',
         })))
 
-    @patch('dns.resolver.resolve', side_effect=dns.resolver.NXDOMAIN())
+    @patch('dns.resolver.resolve', side_effect=NXDOMAIN())
     @patch('requests.get', side_effect=[
         # resolving handle, HTTPS method
         requests_response('did:plc:user', content_type='text/plain'),
         # AppView getRecord
         requests_response({
-            'uri': 'at://did:plc:bob/app.bsky.feed.post/tid',
+            'uri': 'at://did:plc:user/app.bsky.feed.post/tid',
             'cid': 'my sidd',
             'value': {
                 '$type': 'app.bsky.feed.post',
@@ -464,7 +463,7 @@ class ATProtoTest(TestCase):
             'https://api.bsky-sandbox.dev/xrpc/com.atproto.repo.getRecord?repo=did%3Aplc%3Auser&collection=app.bsky.feed.post&rkey=tid',
             json=None, data=None, headers=ANY)
 
-    @patch('dns.resolver.resolve', side_effect=dns.resolver.NXDOMAIN())
+    @patch('dns.resolver.resolve', side_effect=NXDOMAIN())
     # resolving handle, HTTPS method
     @patch('requests.get', return_value=requests_response(status=404))
     def test_convert_populate_cid_fetch_remote_record_bad_handle(self, _, __):
@@ -536,6 +535,39 @@ class ATProtoTest(TestCase):
             'image': [{'url': 'http://my/pic'}],
         }), fetch_blobs=True))
 
+    # resolveHandle
+    @patch('requests.get', return_value=requests_response({'did': 'did:plc:user'}))
+    def test_convert_resolve_mention_handle(self, mock_get):
+        self.store_object(id='did:plc:user', raw=DID_DOC)
+
+        self.assertEqual({
+            '$type': 'app.bsky.feed.post',
+            'createdAt': '2022-01-02T03:04:05.000Z',
+            'text': 'hi @han.dull hows it going',
+            'facets': [{
+                '$type': 'app.bsky.richtext.facet',
+                'features': [{
+                    '$type': 'app.bsky.richtext.facet#mention',
+                    'did': 'did:plc:user',
+                }],
+                'index': {
+                    'byteEnd': 12,
+                    'byteStart': 3,
+                },
+            }],
+        }, ATProto.convert(Object(our_as1={
+            # this mention has the DID in url, and it will also be extracted
+            # from the link in content. make sure we merge the two and don't end
+            # up with a duplicate mention of the DID or a mention of the handle.
+            'objectType': 'note',
+            'content': 'hi <a href="https://bsky.app/profile/han.dull">@han.dull</a> hows it going',
+            'tags': [{
+                'objectType': 'mention',
+                'url': 'did:plc:user',
+                'displayName': '@han.dull'
+            }],
+        })))
+
     def test_convert_protocols_not_enabled(self):
         obj = Object(our_as1={'foo': 'bar'}, source_protocol='activitypub')
         with self.assertRaises(BadRequest):
@@ -1024,13 +1056,6 @@ class ATProtoTest(TestCase):
                     'cid': 'my sidd',
                 },
             },
-            'facets': [{
-                '$type': 'app.bsky.richtext.facet',
-                'features': [{
-                    '$type': 'app.bsky.richtext.facet#mention',
-                    'did': 'did:alice',
-                }],
-            }],
         }, record)
 
         at_uri = f'at://did:plc:user/app.bsky.feed.post/{last_tid}'

tests/test_common.py

@@ -109,3 +109,10 @@ class CommonTest(TestCase):
         self.assertTrue(common.is_enabled(ATProto, Web))
         self.assertTrue(common.is_enabled(Fake, OtherFake))
         self.assertFalse(common.is_enabled(ATProto, ActivityPub))
+
+        self.assertFalse(common.is_enabled(
+            ATProto, ActivityPub, handle_or_id='unknown'))
+        self.assertTrue(common.is_enabled(
+            ATProto, ActivityPub, handle_or_id='snarfed.org'))
+        self.assertTrue(common.is_enabled(
+            ATProto, ActivityPub, handle_or_id='did:plc:fdme4gb7mu7zrie7peay7tst'))

tests/test_ids.py

@@ -19,11 +19,21 @@ class IdsTest(TestCase):
         Fake(id='fake:user',
              copies=[Target(uri='did:plc:789', protocol='atproto')]).put()
 
+        # DID doc and ATProto, used to resolve handle in bsky.app URL
+        did = self.store_object(id='did:plc:123', raw={
+            'id': 'did:plc:123',
+            'alsoKnownAs': ['at://user.com'],
+        })
+        ATProto(id='did:plc:123', obj_key=did.key).put()
+
         for from_, id, to, expected in [
             (ActivityPub, 'https://inst/user', ActivityPub, 'https://inst/user'),
             (ActivityPub, 'https://inst/user', ATProto, 'did:plc:456'),
             (ActivityPub, 'https://inst/user', Fake, 'fake:u:https://inst/user'),
             (ActivityPub, 'https://inst/user', Web, 'https://inst/user'),
+            (ActivityPub, 'https://bsky.app/profile/user.com', ATProto, 'did:plc:123'),
+            (ActivityPub, 'https://bsky.app/profile/did:plc:123',
+             ATProto, 'did:plc:123'),
             (ATProto, 'did:plc:456', ATProto, 'did:plc:456'),
             # copies
             (ATProto, 'did:plc:123', Web, 'user.com'),
@@ -33,6 +43,8 @@ class IdsTest(TestCase):
             (ATProto, 'did:plc:x', Web, 'https://atproto.brid.gy/web/did:plc:x'),
             (ATProto, 'did:plc:x', ActivityPub, 'https://atproto.brid.gy/ap/did:plc:x'),
             (ATProto, 'did:plc:x', Fake, 'fake:u:did:plc:x'),
+            (ATProto, 'https://bsky.app/profile/user.com', ATProto, 'did:plc:123'),
+            (ATProto, 'https://bsky.app/profile/did:plc:123', ATProto, 'did:plc:123'),
             (Fake, 'fake:user', ActivityPub, 'https://fa.brid.gy/ap/fake:user'),
             (Fake, 'fake:user', ATProto, 'did:plc:789'),
             (Fake, 'fake:user', Fake, 'fake:user'),
@@ -41,6 +53,8 @@ class IdsTest(TestCase):
             (Web, 'https://user.com/', ActivityPub, 'http://localhost/user.com'),
             (Web, 'user.com', ATProto, 'did:plc:123'),
             (Web, 'https://user.com', ATProto, 'did:plc:123'),
+            (Web, 'https://bsky.app/profile/user.com', ATProto, 'did:plc:123'),
+            (Web, 'https://bsky.app/profile/did:plc:123', ATProto, 'did:plc:123'),
             (Web, 'user.com', Fake, 'fake:u:user.com'),
             (Web, 'user.com', Web, 'user.com'),
             (Web, 'https://user.com/', Web, 'user.com'),
@@ -132,6 +146,13 @@ class IdsTest(TestCase):
         self.store_object(id='fake:post',
                           copies=[Target(uri='at://did/fa/post', protocol='atproto')])
 
+        # DID doc and ATProto, used to resolve handle in bsky.app URL
+        did = self.store_object(id='did:plc:123', raw={
+            'id': 'did:plc:123',
+            'alsoKnownAs': ['at://user.com'],
+        })
+        ATProto(id='did:plc:123', obj_key=did.key).put()
+
         for from_, id, to, expected in [
             (ActivityPub, 'https://inst/post', ActivityPub, 'https://inst/post'),
             (ActivityPub, 'https://inst/post', ATProto, 'at://did/ap/post'),
@@ -147,6 +168,10 @@ class IdsTest(TestCase):
             (ATProto, 'did:plc:x', Web, 'https://atproto.brid.gy/convert/web/did:plc:x'),
             (ATProto, 'did:plc:x', ActivityPub, 'https://atproto.brid.gy/convert/ap/did:plc:x'),
             (ATProto, 'did:plc:x', Fake, 'fake:o:atproto:did:plc:x'),
+            (ATProto, 'https://bsky.app/profile/user.com/post/456',
+             ATProto, 'at://did:plc:123/app.bsky.feed.post/456'),
+            (ATProto, 'https://bsky.app/profile/did:plc:123/post/456',
+             ATProto, 'at://did:plc:123/app.bsky.feed.post/456'),
             (Fake, 'fake:post',
              ActivityPub, 'https://fa.brid.gy/convert/ap/fake:post'),
             (Fake, 'fake:post', ATProto, 'at://did/fa/post'),

tests/test_integrations.py

@@ -9,9 +9,10 @@ from oauth_dropins.webutil.testutil import requests_response
 from activitypub import ActivityPub
 import app
 from atproto import ATProto
+from dns.resolver import NXDOMAIN
 from granary.tests.test_bluesky import ACTOR_PROFILE_BSKY
 import hub
-from models import Target
+from models import Object, Target
 from web import Web
 
 from .testutil import ATPROTO_KEY, TestCase
@@ -23,6 +24,11 @@ DID_DOC = {
     'id': 'did:plc:alice',
     'alsoKnownAs': ['at://alice.com'],
 }
+PROFILE_GETRECORD = {
+    'uri': 'at://did:plc:alice/app.bsky.actor.profile/self',
+    'cid': 'alice sidd',
+    'value': test_atproto.ACTOR_PROFILE_BSKY,
+}
 
 
 class IntegrationTests(TestCase):
@@ -92,11 +98,7 @@ class IntegrationTests(TestCase):
                 }],
             }),
             # ATProto getRecord of alice's profile
-            requests_response({
-                'uri': 'at://did:plc:alice/app.bsky.actor.profile/self',
-                'cid': 'alice sidd',
-                'value': test_atproto.ACTOR_PROFILE_BSKY,
-            }),
+            requests_response(PROFILE_GETRECORD),
         ]
 
         resp = self.post('/queue/atproto-poll-notifs', client=hub.app.test_client())
@@ -164,11 +166,7 @@ class IntegrationTests(TestCase):
                 }],
             }),
             # ATProto getRecord of alice's profile
-            requests_response({
-                'uri': 'at://did:plc:alice/app.bsky.actor.profile/self',
-                'cid': 'alice sidd',
-                'value': test_atproto.ACTOR_PROFILE_BSKY,
-            }),
+            requests_response(PROFILE_GETRECORD),
             # webmention discovery
             test_web.WEBMENTION_REL_LINK,
         ]
@@ -181,3 +179,94 @@ class IntegrationTests(TestCase):
             'source': 'https://atproto.brid.gy/convert/web/at://did:plc:alice/app.bsky.graph.follow/123',
             'target': 'https://bob.com/',
         }, allow_redirects=False, headers={'Accept': '*/*'})
+
+    @patch('dns.resolver.resolve', side_effect=NXDOMAIN())
+    @patch('oauth_dropins.webutil.appengine_config.tasks_client.create_task')
+    @patch('requests.post', side_effect=[
+           requests_response('OK'),  # create DID
+    ])
+    @patch('requests.get', side_effect = [
+        # webmention source page, follow HTML
+        requests_response("""\
+<html>
+<body class="h-entry">
+<a class="u-url" href="https://bob.com/follow"></a>
+<a class="u-follow-of" href="https://bsky.app/profile/alice.com"></a>
+<a href="http://localhost/"></a>
+</body>
+</html>
+"""),
+        # https://bob.com/ , for authorship
+        requests_response("""\
+<html>
+<body class="h-card">
+<a class="p-name u-url" rel="me" href="https://bob.com/">Bob</a>
+</body>
+</html>
+"""),
+        # alice.com handle resolution, HTTPS method
+        requests_response('did:plc:alice', content_type='text/plain'),
+        # alice profile
+        requests_response(PROFILE_GETRECORD),
+        # alice DID
+        requests_response(DID_DOC),
+        # alice profile
+        requests_response(PROFILE_GETRECORD),
+    ])
+    def test_web_follow_to_atproto(self, mock_get, mock_post, _, __):
+        """Incoming webmention for a web follow of an ATProto bsky.app profile URL.
+
+        Web user bob.com
+        ATProto user alice.com (did:plc:alice)
+        Follow is HTML with mf2 u-follow-of of https://bsky.app/profile/alice.com
+        """
+        bob = self.make_user(id='bob.com', cls=Web, obj_mf2={
+            'type': ['h-card'],
+            'properties': {
+                'url': ['https://bob.com/'],
+                'name': ['Bob'],
+            },
+        })
+
+        # send webmention
+        resp = self.post('/webmention', data={
+            'source': 'https://bob.com/follow',
+            'target': 'http://localhost/',
+        })
+        self.assertEqual(202, resp.status_code)
+
+        # check results
+        bob = bob.key.get()
+        self.assertEqual(1, len(bob.copies))
+        self.assertEqual('atproto', bob.copies[0].protocol)
+        bob_did = bob.copies[0].uri
+
+        self.assertEqual({
+            'type': ['h-entry'],
+            'properties': {
+                'url': ['https://bob.com/follow'],
+                'follow-of': ['https://bsky.app/profile/alice.com'],
+                'name': [''],
+                'author': [{
+                    'type': ['h-card'],
+                    'properties': {
+                        'name': ['Bob'],
+                        'url': ['https://bob.com/'],
+                    },
+                }],
+            },
+        }, Object.get_by_id('https://bob.com/follow').mf2)
+
+        storage = DatastoreStorage()
+        repo = storage.load_repo('bob.com.web.brid.gy')
+        self.assertEqual(bob_did, repo.did)
+
+        records = repo.get_contents()
+        self.assertEqual(['app.bsky.actor.profile', 'app.bsky.graph.follow'],
+                         list(records.keys()))
+        self.assertEqual(['self'], list(records['app.bsky.actor.profile'].keys()))
+        self.assertEqual([{
+            '$type': 'app.bsky.graph.follow',
+            'subject': 'did:plc:alice',
+            'createdAt': '2022-01-02T03:04:05.000Z',
+        }], list(records['app.bsky.graph.follow'].values()))

tests/test_models.py

@@ -681,6 +681,14 @@ class ObjectTest(TestCase):
         self.assertNotIn('id', obj.as1['actor'])
         self.assertEqual(['c', 'd'], obj.as1['object'])
 
+        obj = Object(mf2={
+            'properties': {
+                'uid': ['z.com'],
+                'url': ['x'],
+            },
+        })
+        self.assertEqual('z.com', obj.as1['id'])
+
     def test_validate_id(self):
         # DID repo ids
         Object(id='at://did:plc:123/app.bsky.feed.post/abc').put()
@@ -877,6 +885,65 @@ class ObjectTest(TestCase):
             },
         }, obj.our_as1)
 
+    def test_normalize_ids_empty(self):
+        obj = Object()
+        obj.normalize_ids()
+        self.assertIsNone(obj.as1)
+
+    def test_normalize_ids_follow_atproto(self):
+        # for ATProto handle resolution
+        self.store_object(id='did:plc:user', raw=DID_DOC)
+        alice = self.make_user(id='did:plc:user', cls=ATProto)
+
+        obj = Object(our_as1={
+            'objectType': 'activity',
+            'verb': 'follow',
+            'actor': 'https://bsky.app/profile/did:plc:123',
+            'object': 'https://bsky.app/profile/han.dull',
+        })
+        obj.normalize_ids()
+        self.assert_equals({
+            'objectType': 'activity',
+            'verb': 'follow',
+            'actor': 'did:plc:123',
+            'object': 'did:plc:user',
+        }, obj.our_as1)
+
+    def test_normalize_ids_reply(self):
+        # for ATProto handle resolution
+        self.store_object(id='did:plc:user', raw=DID_DOC)
+        self.make_user(id='did:plc:user', cls=ATProto)
+
+        obj = Object(our_as1={
+            'objectType': 'activity',
+            'verb': 'post',
+            'object': {
+                'id': 'https://bsky.app/profile/han.dull/post/456',
+                'objectType': 'note',
+                'inReplyTo': 'https://bsky.app/profile/did:plc:123/post/789',
+                'author': 'https://bsky.app/profile/han.dull',
+                'tags': [{
+                    'objectType': 'mention',
+                    'url': 'https://bsky.app/profile/did:plc:123',
+                }],
+            },
+        })
+        obj.normalize_ids()
+        self.assert_equals({
+            'objectType': 'activity',
+            'verb': 'post',
+            'object': {
+                'id': 'at://did:plc:user/app.bsky.feed.post/456',
+                'objectType': 'note',
+                'inReplyTo': 'at://did:plc:123/app.bsky.feed.post/789',
+                'author': 'did:plc:user',
+                'tags': [{
+                    'objectType': 'mention',
+                    'url': 'did:plc:123',
+                }],
+            },
+        }, obj.our_as1)
+
     def test_get_originals(self):
         self.assertEqual([], models.get_originals(['foo', 'did:plc:bar']))
 

tests/test_pages.py

@@ -99,11 +99,12 @@ class PagesTest(TestCase):
         self.assert_equals(404, got.status_code)
 
     def test_user_not_direct(self):
-        fake = self.make_user('fake:foo', cls=Fake)
-        fake.direct = False
-        fake.put()
-
+        fake = self.make_user('fake:foo', cls=Fake, direct=False)
         got = self.client.get('/fake/fake:foo')
+        self.assert_equals(200, got.status_code)
+
+        fake = self.make_user('http://fo/o', cls=ActivityPub, direct=False)
+        got = self.client.get('/ap/@o@fo')
         self.assert_equals(404, got.status_code)
 
     def test_user_opted_out(self):

tests/test_protocol.py

@@ -93,7 +93,8 @@ class ProtocolTest(TestCase):
                 ('https://ap.brid.gy/foo/bar', ActivityPub),
                 ('https://web.brid.gy/foo/bar', Web),
         ]:
-            self.assertEqual(expected, Protocol.for_id(id))
+            self.assertEqual(expected, Protocol.for_id(id, remote=False))
+            self.assertEqual(expected, Protocol.for_id(id, remote=True))
 
     def test_for_id_true_overrides_none(self):
         class Greedy(Protocol, User):
@@ -137,6 +138,11 @@ class ProtocolTest(TestCase):
         self.assertIsNone(Protocol.for_id('http://web.site/'))
         self.assertIn(self.req('http://web.site/'), mock_get.mock_calls)
 
+    @patch('requests.get')
+    def test_for_id_web_remote_false(self, mock_get):
+        self.assertIsNone(Protocol.for_id('http://web.site/', remote=False))
+        mock_get.assert_not_called()
+
     def test_for_handle_deterministic(self):
         for handle, expected in [
             (None, (None, None)),

tests/test_web.py

@@ -947,6 +947,7 @@ class WebTest(TestCase):
                            type='share',
                            object_ids=['https://mas.to/toot/id'],
                            labels=['user', 'activity', 'notification', 'feed'],
+                           ignore=['our_as1'],
                            )
 
     def test_link_rel_alternate_as2(self, mock_get, mock_post):
@@ -1005,6 +1006,7 @@ class WebTest(TestCase):
                            type='like',
                            labels=['activity', 'user'],
                            status='ignored',
+                           ignore=['our_as1'],
                            )
 
     def test_post_type_discovery_multiple_types(self, mock_get, mock_post):
@@ -1124,6 +1126,7 @@ class WebTest(TestCase):
                            notify=[ndb.Key('ActivityPub', 'https://mas.to/author')],
                            delivered=['https://mas.to/inbox'],
                            status='complete',
+                           ignore=['our_as1'],
                            )
 
     def test_create_non_domain_author(self, mock_get, mock_post):
@@ -1316,16 +1319,19 @@ class WebTest(TestCase):
 
         self.assert_deliveries(mock_post, ['https://mas.to/inbox'], FOLLOW_AS2)
 
+        follow_as1 = copy.deepcopy(FOLLOW_AS1)
+        follow_as1['actor']['id'] = 'https://user.com/'
         obj = self.assert_object('https://user.com/follow',
                                  users=[self.user.key],
                                  notify=[self.mrs_foo],
                                  source_protocol='web',
                                  status='complete',
-                                 mf2=FOLLOW_MF2,
+                                 our_as1=follow_as1,
                                  delivered=['https://mas.to/inbox'],
                                  type='follow',
                                  object_ids=['https://mas.to/mrs-foo'],
                                  labels=['user', 'activity', 'notification'],
+                                 ignore=['our_as1'],
                                  )
 
         to = self.assert_user(ActivityPub, 'https://mas.to/mrs-foo', obj_as2={
@@ -1412,7 +1418,8 @@ class WebTest(TestCase):
                            delivered=['https://mas.to/inbox'],
                            type='follow',
                            object_ids=['https://mas.to/mrs-foo'],
-                           labels=['user', 'activity', 'notification',],
+                           labels=['user', 'activity', 'notification'],
+                           ignore=['our_as1'],
                            )
 
         followers = Follower.query().fetch()
@@ -1476,7 +1483,8 @@ class WebTest(TestCase):
                                  object_ids=['https://mas.to/mrs-foo',
                                              'https://mas.to/mr-biff'],
                                  labels=['user', 'activity', 'notification',],
-                                 )
+                                 ignore=['our_as1'],
+                                  )
 
         followers = Follower.query().fetch()
         self.assertEqual(2, len(followers))
@@ -1590,6 +1598,7 @@ class WebTest(TestCase):
                            type='follow',
                            object_ids=['https://mas.to/mrs-foo'],
                            labels=['user', 'activity', 'notification',],
+                           ignore=['our_as1'],
                            )
 
     def test_repost_twitter_blocklisted(self, *mocks):

tests/testutil.py

@@ -11,6 +11,7 @@ from unittest.mock import ANY, call
 from urllib.parse import urlencode
 import warnings
 
+from arroba import did
 import arroba.util
 from arroba.util import datetime_to_tid
 from bs4 import MarkupResemblesLocatorWarning
@@ -197,6 +198,9 @@ class TestCase(unittest.TestCase, testutil.Asserts):
         protocol.Protocol.for_id.cache.clear()
         common.webmention_discover.cache.clear()
         User.count_followers.cache.clear()
+        did.resolve_handle.cache.clear()
+        did.resolve_plc.cache.clear()
+        did.resolve_web.cache.clear()
 
         for cls in Fake, OtherFake:
             cls.fetchable = {}
@@ -299,8 +303,8 @@ class TestCase(unittest.TestCase, testutil.Asserts):
                                            mf2=obj_mf2, source_protocol=cls.LABEL
                                            ).key
 
+        kwargs.setdefault('direct', True)
         user = cls(id=id,
-                   direct=True,
                    mod=global_user.mod,
                    public_exponent=global_user.public_exponent,
                    private_exponent=global_user.private_exponent,

web.py

@@ -537,7 +537,8 @@ class Web(User, Protocol):
         obj_as1 = obj.as1
         from_proto = PROTOCOLS.get(obj.source_protocol)
         if from_proto:
-            if not common.is_enabled(cls, from_proto):
+            user_id = from_user.key.id() if from_user and from_user.key else None
+            if not common.is_enabled(cls, from_proto, handle_or_id=user_id):
                 error(f'{cls.LABEL} <=> {from_proto.LABEL} not enabled')
 
             # fill in author/actor if available