"""Handles requests for WebFinger endpoints.
* https://webfinger.net/
* https://tools.ietf.org/html/rfc7033
"""
from datetime import timedelta
import logging
import re
from urllib.parse import urljoin, urlparse
from flask import render_template, request
from granary import as2
from webutil import flask_util, util
from webutil.flask_util import error, flash, Found
from webutil.util import json_dumps, json_loads
from requests import RequestException
import activitypub
import common
from common import CACHE_CONTROL, CACHE_CONTROL_VARY_ACCEPT
import domains
from domains import (
LOCAL_DOMAINS,
PRIMARY_DOMAIN,
PROTOCOL_DOMAINS,
subdomain_wrap,
SUPERDOMAIN,
)
from flask_app import app
import memcache
import models
from protocol import Protocol
import web
from web import Web
SUBSCRIBE_LINK_REL = 'http://ostatus.org/schema/1.0/subscribe'
logger = logging.getLogger(__name__)
[docs]
class Webfinger(flask_util.XrdOrJrd):
"""Serves a user's WebFinger profile.
Supports both JRD and XRD; defaults to JRD.
https://tools.ietf.org/html/rfc7033#section-4
"""
@memcache.memoize(expire=timedelta(hours=1),
key=lambda self: (request.url, request.headers.get('Accept')))
@flask_util.headers(CACHE_CONTROL_VARY_ACCEPT)
def dispatch_request(self):
return super().dispatch_request()
def template_prefix(self):
return 'webfinger_user'
def template_vars(self):
# logger.debug(f'Headers: {list(request.headers.items())}')
resource = flask_util.get_required_param('resource').strip()
resource = resource.removeprefix(domains.host_url())
# handle Bridgy Fed actor URLs, eg https://fed.brid.gy/snarfed.org
host = util.domain_from_link(domains.host_url())
if resource in ('', '/', f'acct:{host}', f'acct:@{host}'):
error('Expected other domain, not *.brid.gy')
proto = None
try:
username, server = util.parse_acct_uri(resource)
id = server
proto = Protocol.for_bridgy_subdomain(id, fed='web')
if proto:
id = username
except ValueError:
id = username = server = urlparse(resource).netloc or resource
if id == PRIMARY_DOMAIN or id in PROTOCOL_DOMAINS:
proto = Web
elif not proto:
proto = Protocol.for_request(fed='web')
if not proto:
error(f"Couldn't determine protocol for f{resource}")
logger.debug(f'Protocol {proto.LABEL}, user id {id}')
try:
user = models.load_user(id, proto=proto)
except RuntimeError as e:
error(f'No {proto.LABEL} user found for {id}', status=404)
if (not user.is_enabled(activitypub.ActivityPub)
or (proto == Web and username not in (user.key.id(), user.username()))):
error(f'No {proto.LABEL} user found for {id}', status=404)
ap_handle = user.handle_as('activitypub')
if not ap_handle:
error(f'{proto.LABEL} user {id} has no handle', status=404)
# backward compatibility for initial Web users whose AP actor ids are on
# fed.brid.gy, not web.brid.gy
subdomain = request.host.split('.')[0]
if (user.LABEL == 'web'
and subdomain not in (LOCAL_DOMAINS + (user.ap_subdomain,))):
url = urljoin(f'https://{user.ap_subdomain}{SUPERDOMAIN}/',
request.full_path)
raise Found(location=url)
actor = user.obj.as1 if user.obj and user.obj.as1 else {}
logger.debug(f'Generating WebFinger data for {user.key.id()}')
actor_id = user.id_as(activitypub.ActivityPub)
logger.debug(f'AS1 actor: {actor_id}')
urls = util.dedupe_urls(util.get_list(actor, 'urls') +
util.get_list(actor, 'url') +
[user.web_url()])
logger.debug(f'URLs: {urls}')
canonical_url = urls[0]
# generate webfinger content
data = util.trim_nulls({
'subject': 'acct:' + ap_handle.lstrip('@'),
'aliases': urls,
'links':
[{
'rel': 'http://webfinger.net/rel/profile-page',
'type': 'text/html',
'href': url,
} for url in urls if util.is_web(url)] +
[{
'rel': 'http://webfinger.net/rel/avatar',
'href': url,
} for url in util.get_urls(actor, 'image')] +
[{
'rel': 'canonical_uri',
'type': 'text/html',
'href': canonical_url,
},
# ActivityPub
#
# include two self links, one for each AP content type, since some
# fediverse servers (eg Pleroma, Akkoma) are unnecessarily picky
# about which one they use from XRD vs JRD.
# https://github.com/snarfed/bridgy-fed/issues/995
{
'rel': 'self',
'type': as2.CONTENT_TYPE_LD_PROFILE,
'href': actor_id,
}, {
'rel': 'self',
'type': as2.CONTENT_TYPE,
'href': actor_id,
}, {
# AP reads this and sharedInbox from the AS2 actor, not
# webfinger, so strictly speaking, it's probably not needed here.
'rel': 'inbox',
'type': as2.CONTENT_TYPE_LD_PROFILE,
'href': actor_id + '/inbox',
}, {
# https://www.w3.org/TR/activitypub/#sharedInbox
'rel': 'sharedInbox',
'type': as2.CONTENT_TYPE_LD_PROFILE,
'href': subdomain_wrap(proto, '/ap/sharedInbox'),
},
# remote follow
# https://socialhub.activitypub.rocks/t/what-is-the-current-spec-for-remote-follow/2020/11?u=snarfed
# https://github.com/snarfed/bridgy-fed/issues/60#issuecomment-1325589750
{
'rel': 'http://ostatus.org/schema/1.0/subscribe',
# always use fed.brid.gy for UI pages, not protocol subdomain
# TODO: switch to:
# 'template': domains.host_url(user.user_page_path('?url={uri}')),
# the problem is that user_page_path() uses handle_or_id, which uses
# custom username instead of domain, which may not be unique
'template': f'https://{PRIMARY_DOMAIN}' +
user.user_page_path('?url={uri}'),
}]
})
# logger.info(f'Returning WebFinger data: {json_dumps(data, indent=2)}')
return data
[docs]
def fetch(addr):
"""Fetches and returns an address's WebFinger data.
On failure, flashes a message and returns None.
TODO: switch to raising exceptions instead of flashing messages and
returning None
Args:
addr (str): a Webfinger-compatible address, eg ``@x@y``, ``acct:x@y``, or
``https://x/y``
Returns:
dict: fetched WebFinger data, or None on error
"""
addr = addr.strip().strip('@')
split = addr.split('@')
if len(split) == 2:
addr_domain = split[1]
resource = f'acct:{addr}'
elif util.is_web(addr):
addr_domain = util.domain_from_link(addr, minimize=False)
resource = addr
else:
flash('Enter a fediverse address in @user@domain.social format')
return None
try:
resp = util.requests_get(
f'https://{addr_domain}/.well-known/webfinger?resource={resource}')
except RequestException as e:
flash(f"Couldn't fetch {addr}: {e}")
return None
except BaseException as e:
if util.is_connection_failure(e):
flash(f"Couldn't connect to {addr_domain}")
return None
raise
if not resp.ok:
flash(f'WebFinger on {addr_domain} returned HTTP {resp.status_code}')
return None
try:
data = resp.json()
except ValueError as e:
logger.warning(f'Got {e}', exc_info=True)
flash(f'WebFinger on {addr_domain} returned non-JSON')
return None
logger.debug(f'Got WebFinger for {addr}')
return data
[docs]
def fetch_actor_url(addr):
"""Fetches and returns a WebFinger address's ActivityPub actor URL.
On failure, flashes a message and returns None.
Args:
addr (str): a Webfinger-compatible address, eg ``@x@y``, ``acct:x@y``, or
``https://x/y``
Returns:
str: ActivityPub actor URL, or None on error or not found
"""
data = fetch(addr)
if not data:
return None
for link in data.get('links', []):
type = link.get('type', '').split(';')[0]
if link.get('rel') == 'self' and type in as2.CONTENT_TYPES:
return link.get('href')
app.add_url_rule('/.well-known/webfinger', view_func=Webfinger.as_view('webfinger'))
app.add_url_rule('/.well-known/host-meta', view_func=HostMeta.as_view('hostmeta'))
app.add_url_rule('/.well-known/host-meta.json', view_func=HostMeta.as_view('hostmeta-json'))