AP users: rename [is_]homepage => [is_]web_url, move into Protocol subclasses

for #512

activitypub.py

@@ -54,6 +54,24 @@ class ActivityPub(User, Protocol):
     """
     LABEL = 'activitypub'
 
+    def web_url(self):
+        """Returns this user's web URL aka web_url, eg 'https://foo.com/'."""
+        return util.get_url(self.actor_as2) or self.ap_actor()
+
+    def ap_address(self):
+        """Returns this user's ActivityPub address, eg '@foo.com@foo.com'."""
+        if self.direct:
+            return f'@{self.username()}@{self.key.id()}'
+        else:
+            return f'@{self.key.id()}@{request.host}'
+
+    def ap_actor(self, rest=None):
+        """Returns this user's ActivityPub/AS2 actor id URL.
+
+        Eg 'https://fed.brid.gy/foo.com'
+        """
+        return self.key.id()
+
     @classmethod
     def send(cls, obj, url, log_data=True):
         """Delivers an activity to an inbox URL."""
@@ -373,13 +391,13 @@ def postprocess_as2(activity, target=None, wrap=True):
         activity['object'] = target_id
     elif not id:
         obj['id'] = util.get_first(obj, 'url') or target_id
-    elif g.user and g.user.is_homepage(id):
+    elif g.user and g.user.is_web_url(id):
         obj['id'] = g.user.ap_actor()
     elif g.external_user:
         obj['id'] = redirect_wrap(g.external_user)
 
     # for Accepts
-    if g.user and g.user.is_homepage(obj.get('object')):
+    if g.user and g.user.is_web_url(obj.get('object')):
         obj['object'] = g.user.ap_actor()
     elif g.external_user and g.external_user == obj.get('object'):
         obj['object'] = redirect_wrap(g.external_user)
@@ -469,11 +487,11 @@ def postprocess_as2_actor(actor, wrap=True):
     if not actor:
         return actor
     elif isinstance(actor, str):
-        if g.user and g.user.is_homepage(actor):
+        if g.user and g.user.is_web_url(actor):
             return g.user.ap_actor()
         return redirect_wrap(actor)
 
-    url = g.user.homepage if g.user else None
+    url = g.user.web_url() if g.user else None
     urls = util.get_list(actor, 'url')
     if not urls and url:
       urls = [url]
@@ -483,7 +501,7 @@ def postprocess_as2_actor(actor, wrap=True):
         urls[0] = redirect_wrap(urls[0])
 
     id = actor.get('id')
-    if g.user and (not id or g.user.is_homepage(id)):
+    if g.user and (not id or g.user.is_web_url(id)):
         actor['id'] = g.user.ap_actor()
     elif g.external_user and (not id or id == g.external_user):
         actor['id'] = redirect_wrap(g.external_user)

common.py

@@ -95,7 +95,7 @@ def pretty_link(url, text=None, **kwargs):
     text: str
     kwargs: passed through to :func:`webutil.util.pretty_link`
   """
-  if g.user and g.user.is_homepage(url):
+  if g.user and g.user.is_web_url(url):
     return g.user.user_page_link()
 
   if text is None:

models.py

@@ -103,11 +103,6 @@ class User(StringIdModel, metaclass=ProtocolUserMeta):
         """Try to prevent instantiation. Use subclasses instead."""
         raise NotImplementedError()
 
-    # TODO(#512): move this and is_homepage to web.py?
-    @property
-    def homepage(self):
-        return f'https://{self.key.id()}/'
-
     def _post_put_hook(self, future):
         logger.info(f'Wrote {self.key}')
 
@@ -200,16 +195,48 @@ class User(StringIdModel, metaclass=ProtocolUserMeta):
         logger.info(f'Defaulting username to key id {id}')
         return id
 
+    def web_url(self):
+        """Returns this user's web URL aka web_url, eg 'https://foo.com/'.
+
+        To be implemented by subclasses.
+
+        Returns:
+          str
+        """
+        raise NotImplementedError()
+
+    def is_web_url(self, url):
+        """Returns True if the given URL is this user's web URL (web_url).
+
+        Args:
+          url: str
+
+        Returns:
+          boolean
+        """
+        if not url:
+            return False
+
+        url = url.strip().rstrip('/')
+        parsed_url = urllib.parse.urlparse(url)
+        if parsed_url.scheme not in ('http', 'https', ''):
+            return False
+
+        this = self.web_url().rstrip('/')
+        parsed_this = urllib.parse.urlparse(this)
+
+        return (url == this or url == parsed_this.netloc or
+                parsed_url[1:] == parsed_this[1:])  # ignore http vs https
+
     def ap_address(self):
         """Returns this user's ActivityPub address, eg '@me@foo.com'.
 
         To be implemented by subclasses.
+
+        Returns:
+          str
         """
         raise NotImplementedError()
-        if self.direct:
-            return f'@{self.username()}@{self.key.id()}'
-        else:
-            return f'@{self.key.id()}@{request.host}'
 
     def ap_actor(self, rest=None):
         """Returns this user's ActivityPub/AS2 actor id.
@@ -217,33 +244,14 @@ class User(StringIdModel, metaclass=ProtocolUserMeta):
         Eg 'https://fed.brid.gy/ap/bluesky/foo.com'
 
         To be implemented by subclasses.
+
+        Args:
+          rest: str, optional, appended to URL path
+
+        Returns:
+          str
         """
         raise NotImplementedError()
-        if self.direct or rest:
-            # special case Web users to skip /ap/web/ prefix, for backward compatibility
-            url = common.host_url(self.key.id() if self.LABEL == 'web'
-                                  else f'/ap{self.user_page_path()}')
-            if rest:
-                url += f'/{rest}'
-            return url
-        # TODO(#512): drop once we fetch site if web user doesn't already exist
-        else:
-            return redirect_wrap(self.homepage)
-
-    def is_homepage(self, url):
-        """Returns True if the given URL points to this user's home page."""
-        if not url:
-            return False
-
-        url = url.strip().rstrip('/')
-        if url == self.key.id():
-            return True
-
-        parsed = urllib.parse.urlparse(url)
-        return (parsed.netloc == self.key.id()
-                and parsed.scheme in ('', 'http', 'https')
-                and not parsed.path and not parsed.query
-                and not parsed.params and not parsed.fragment)
 
     def user_page_path(self, rest=None):
         """Returns the user's Bridgy Fed user page path."""

pages.py

@@ -131,7 +131,7 @@ def feed(protocol, domain):
 
     actor = {
       'displayName': domain,
-      'url': g.user.homepage,
+      'url': g.user.web_url(),
     }
     title = f'Bridgy Fed feed for {domain}'
 
@@ -277,7 +277,7 @@ def nodeinfo():
             'name': 'bridgy-fed',
             'version': os.getenv('GAE_VERSION'),
             'repository': 'https://github.com/snarfed/bridgy-fed',
-            'homepage': 'https://fed.brid.gy/',
+            'web_url': 'https://fed.brid.gy/',
         },
         'protocols': [
             'activitypub',

tests/test_activitypub.py

@@ -278,11 +278,11 @@ class ActivityPubTest(TestCase):
         self.assertTrue(type.startswith(as2.CONTENT_TYPE), type)
         self.assertEqual({
             'preferredUsername': 'fake.com',
-            'id': 'http://bf/fake/ap',
-            'inbox': 'http://bf/fake/ap/inbox',
-            'outbox': 'http://bf/fake/ap/outbox',
-            'following': 'http://bf/fake/ap/following',
-            'followers': 'http://bf/fake/ap/followers',
+            'id': 'http://bf/fake.com/ap',
+            'inbox': 'http://bf/fake.com/ap/inbox',
+            'outbox': 'http://bf/fake.com/ap/outbox',
+            'following': 'http://bf/fake.com/ap/following',
+            'followers': 'http://bf/fake.com/ap/followers',
             'endpoints': {
                 'sharedInbox': 'http://localhost/ap/sharedInbox',
             },
@@ -1259,8 +1259,8 @@ class ActivityPubUtilsTest(TestCase):
     def setUp(self):
         super().setUp()
         self.request_context.push()
-        g.user = self.make_user('user.com', has_hcard=True,
-                                actor_as2=ACTOR)
+        g.user = self.make_user('user.com', has_hcard=True, actor_as2=ACTOR)
+
     def tearDown(self):
         self.request_context.pop()
         super().tearDown()
@@ -1302,16 +1302,16 @@ class ActivityPubUtilsTest(TestCase):
             'actor': {
                 'id': 'baj',
                 'preferredUsername': 'site',
-                'url': 'http://localhost/r/https://site/',
+                'url': 'http://localhost/r/https://site',
             },
             'attributedTo': [{
                 'id': 'bar',
                 'preferredUsername': 'site',
-                'url': 'http://localhost/r/https://site/',
+                'url': 'http://localhost/r/https://site',
             }, {
                 'id': 'baz',
                 'preferredUsername': 'site',
-                'url': 'http://localhost/r/https://site/',
+                'url': 'http://localhost/r/https://site',
             }],
             'to': [as2.PUBLIC_AUDIENCE],
         }, activitypub.postprocess_as2({
@@ -1542,3 +1542,21 @@ class ActivityPubUtilsTest(TestCase):
                     activitypub.postprocess_as2(obj),
                     activitypub.postprocess_as2(activitypub.postprocess_as2(obj)),
                     ignore=['to'])
+
+    def test_ap_actor(self):
+        user = self.make_user('http://foo/actor', cls=ActivityPub)
+        self.assertEqual('http://foo/actor', user.ap_actor())
+
+    def test_ap_address(self):
+        user = self.make_user('http://foo/actor', cls=ActivityPub)
+        self.assertEqual('http://foo/actor', user.ap_actor())
+
+    def test_web_url(self):
+        user = self.make_user('http://foo/actor', cls=ActivityPub)
+        self.assertEqual('http://foo/actor', user.web_url())
+
+        user.actor_as2 = copy.deepcopy(ACTOR)  # no url
+        self.assertEqual('http://foo/actor', user.web_url())
+
+        user.actor_as2['url'] = ['http://my/url']
+        self.assertEqual('http://my/url', user.web_url())

tests/test_models.py

@@ -86,6 +86,14 @@ class UserTest(TestCase):
         g.user.actor_as2 = ACTOR
         self.assertEqual('<a class="h-card u-author" href="/web/y.z"><img src="https://user.com/me.jpg" class="profile"> Mrs. ☕ Foo</a>', g.user.user_page_link())
 
+    def test_is_web_url(self):
+        for url in 'y.z', '//y.z', 'http://y.z', 'https://y.z':
+            self.assertTrue(g.user.is_web_url(url), url)
+
+        for url in (None, '', 'user', 'com', 'com.user', 'ftp://y.z',
+                    'https://user', '://y.z'):
+            self.assertFalse(g.user.is_web_url(url), url)
+
 
 class ObjectTest(TestCase):
     def setUp(self):

tests/test_web.py

@@ -1501,16 +1501,31 @@ http://this/404s
             'preferredUsername': 'user.com',
         })
 
-    def test_homepage(self, _, __):
-        self.assertEqual('https://user.com/', self.user.homepage)
+    def test_web_url(self, _, __):
+        self.assertEqual('https://user.com/', self.user.web_url())
 
-    def test_is_homepage(self, _, __):
-        for url in 'user.com', '//user.com', 'http://user.com', 'https://user.com':
-            self.assertTrue(self.user.is_homepage(url), url)
+    def test_ap_address(self, *_):
+        self.assertEqual('@user.com@user.com', g.user.ap_address())
 
-        for url in (None, '', 'user', 'com', 'com.user', 'ftp://user.com',
-                    'https://user', '://user.com'):
-            self.assertFalse(self.user.is_homepage(url), url)
+        g.user.actor_as2 = {'type': 'Person'}
+        self.assertEqual('@user.com@user.com', g.user.ap_address())
+
+        g.user.actor_as2 = {'url': 'http://foo'}
+        self.assertEqual('@user.com@user.com', g.user.ap_address())
+
+        g.user.actor_as2 = {'url': ['http://foo', 'acct:bar@foo', 'acct:baz@user.com']}
+        self.assertEqual('@baz@user.com', g.user.ap_address())
+
+        g.user.direct = False
+        self.assertEqual('@user.com@localhost', g.user.ap_address())
+
+    def test_ap_actor(self, *_):
+        self.assertEqual('http://localhost/user.com', g.user.ap_actor())
+
+        g.user.direct = False
+        self.assertEqual('http://localhost/r/https://user.com/', g.user.ap_actor())
+
+        self.assertEqual('http://localhost/user.com/inbox', g.user.ap_actor('inbox'))
 
     def test_check_web_site(self, mock_get, _):
         redir = 'http://localhost/.well-known/webfinger?resource=acct:user.com@user.com'
@@ -1547,29 +1562,6 @@ http://this/404s
         self.assertTrue(get_flashed_messages()[0].startswith(
             "Couldn't connect to https://orig/: "))
 
-    def test_ap_address(self, *_):
-        self.assertEqual('@user.com@user.com', g.user.ap_address())
-
-        g.user.actor_as2 = {'type': 'Person'}
-        self.assertEqual('@user.com@user.com', g.user.ap_address())
-
-        g.user.actor_as2 = {'url': 'http://foo'}
-        self.assertEqual('@user.com@user.com', g.user.ap_address())
-
-        g.user.actor_as2 = {'url': ['http://foo', 'acct:bar@foo', 'acct:baz@user.com']}
-        self.assertEqual('@baz@user.com', g.user.ap_address())
-
-        g.user.direct = False
-        self.assertEqual('@user.com@localhost', g.user.ap_address())
-
-    def test_ap_actor(self, *_):
-        self.assertEqual('http://localhost/user.com', g.user.ap_actor())
-
-        g.user.direct = False
-        self.assertEqual('http://localhost/r/https://user.com/', g.user.ap_actor())
-
-        self.assertEqual('http://localhost/user.com/inbox', g.user.ap_actor('inbox'))
-
 
 @patch('requests.post')
 @patch('requests.get')

tests/testutil.py

@@ -44,11 +44,14 @@ class Fake(User, protocol.Protocol):
     # in-order list of ids
     fetched = []
 
+    def web_url(self):
+        return f'https://{self.key.id()}'
+
     def ap_address(self):
-        return '@fake@fake'
+        return f'@{self.key.id()}@fake'
 
     def ap_actor(self, rest=None):
-        return 'http://bf/fake/ap' + (f'/{rest}' if rest else '')
+        return f'http://bf/{self.key.id()}/ap' + (f'/{rest}' if rest else '')
 
     @classmethod
     def send(cls, obj, url, log_data=True):

web.py

@@ -49,6 +49,10 @@ class Web(User, Protocol):
     def _get_kind(cls):
         return 'MagicKey'
 
+    def web_url(self):
+        """Returns this user's web URL aka web_url, eg 'https://foo.com/'."""
+        return f'https://{self.key.id()}/'
+
     def ap_address(self):
         """Returns this user's ActivityPub address, eg '@foo.com@foo.com'.
 
@@ -74,7 +78,7 @@ class Web(User, Protocol):
             return url
 
         # TODO(#512): drop once we fetch site if web user doesn't already exist
-        return common.redirect_wrap(self.homepage)
+        return common.redirect_wrap(self.web_url())
 
     def verify(self):
         """Fetches site a couple ways to check for redirects and h-card.
@@ -94,7 +98,7 @@ class Web(User, Protocol):
             root_site = f'https://{root}/'
             try:
                 resp = util.requests_get(root_site, gateway=False)
-                if resp.ok and self.is_homepage(resp.url):
+                if resp.ok and self.is_web_url(resp.url):
                     logger.info(f'{root_site} redirects to {resp.url} ; using {root} instead')
                     root_user = Web.get_or_create(root)
                     self.use_instead = root_user.key
@@ -108,7 +112,7 @@ class Web(User, Protocol):
         self.has_redirects = False
         self.redirects_error = None
         try:
-            url = urljoin(self.homepage, path)
+            url = urljoin(self.web_url(), path)
             resp = util.requests_get(url, gateway=False)
             domain_urls = ([f'https://{domain}/' for domain in common.DOMAINS] +
                            [common.host_url()])
@@ -129,7 +133,7 @@ class Web(User, Protocol):
 
         # check home page
         try:
-            obj = Web.load(self.homepage, gateway=True)
+            obj = Web.load(self.web_url(), gateway=True)
             self.actor_as2 = activitypub.postprocess_as2(as2.from_as1(obj.as1))
             self.has_hcard = True
         except (BadRequest, NotFound):
@@ -168,12 +172,12 @@ class Web(User, Protocol):
           check_backlink: bool, optional, whether to require a link to Bridgy Fed
         """
         url = obj.key.id()
-        is_homepage = ((g.user and g.user.is_homepage(url)) or
+        is_web_url = ((g.user and g.user.is_web_url(url)) or
                        (g.external_user and g.external_user == url))
 
 
         require_backlink = None
-        if check_backlink or (check_backlink is None and not is_homepage):
+        if check_backlink or (check_backlink is None and not is_web_url):
             require_backlink = common.host_url().rstrip('/')
 
         try:
@@ -186,8 +190,8 @@ class Web(User, Protocol):
             error(f'id {urlparse(url).fragment} not found in {url}')
 
         # find mf2 item
-        if is_homepage:
-            logger.info(f"{url} is user's homepage")
+        if is_web_url:
+            logger.info(f"{url} is user's web url")
             entry = mf2util.representative_hcard(parsed, parsed['url'])
             logger.info(f'Representative h-card: {json_dumps(entry, indent=2)}')
             if not entry:
@@ -200,7 +204,7 @@ class Web(User, Protocol):
         # store final URL in mf2 object, and also default url property to it,
         # since that's the fallback for AS1/AS2 id
         entry['url'] = parsed['url']
-        if is_homepage:
+        if is_web_url:
             entry.setdefault('rel-urls', {}).update(parsed.get('rel-urls', {}))
         props = entry.setdefault('properties', {})
         props.setdefault('url', [parsed['url']])
@@ -209,7 +213,7 @@ class Web(User, Protocol):
         # run full authorship algorithm if necessary: https://indieweb.org/authorship
         # duplicated in microformats2.json_to_object
         author = util.get_first(props, 'author')
-        if not isinstance(author, dict) and not is_homepage:
+        if not isinstance(author, dict) and not is_web_url:
             logger.info(f'Fetching full authorship for author {author}')
             author = mf2util.find_author({'items': [entry]}, hentry=entry,
                                          fetch_mf2_func=util.fetch_mf2)
@@ -329,7 +333,7 @@ def webmention_interactive():
     """
     try:
         webmention_external()
-        flash(f'Updating fediverse profile from <a href="{g.user.homepage}">{g.user.key.id()}</a>...')
+        flash(f'Updating fediverse profile from <a href="{g.user.web_url()}">{g.user.key.id()}</a>...')
     except HTTPException as e:
         flash(util.linkify(str(e.description), pretty=True))
 
@@ -379,14 +383,14 @@ def webmention_task():
         # set actor to user
         props = obj.mf2['properties']
         author_urls = microformats2.get_string_urls(props.get('author', []))
-        if author_urls and not g.user.is_homepage(author_urls[0]):
+        if author_urls and not g.user.is_web_url(author_urls[0]):
             logger.info(f'Overriding author {author_urls[0]} with {g.user.ap_actor()}')
             props['author'] = [g.user.ap_actor()]
 
     logger.info(f'Converted to AS1: {obj.type}: {json_dumps(obj.as1, indent=2)}')
 
     # if source is home page, send an actor Update to followers' instances
-    if g.user.is_homepage(obj.key.id()):
+    if g.user.is_web_url(obj.key.id()):
         obj.put()
         actor_as1 = {
             **obj.as1,

webfinger.py

@@ -54,20 +54,17 @@ class Actor(flask_util.XrdOrJrd):
             error(f'No user or web site found for {domain}', status=404)
 
         actor = g.user.to_as1() or {}
-        homepage = g.user.homepage
-        handle = g.user.ap_address()
-
         logger.info(f'Generating WebFinger data for {domain}')
         logger.info(f'AS1 actor: {actor}')
         urls = util.dedupe_urls(util.get_list(actor, 'urls') +
                                 util.get_list(actor, 'url') +
-                                [homepage])
+                                [g.user.web_url()])
         logger.info(f'URLs: {urls}')
         canonical_url = urls[0]
 
         # generate webfinger content
         data = util.trim_nulls({
-            'subject': 'acct:' + handle.lstrip('@'),
+            'subject': 'acct:' + g.user.ap_address().lstrip('@'),
             'aliases': urls,
             'links':
             [{