relay/actor.py

   1 import aiohttp
   2 import aiohttp.web
   3 import asyncio
   4 import logging
   5 import uuid
   6 import re
   7 import simplejson as json
   8 import cgi
   9 import datetime
  10
  11 from urllib.parse import urlsplit
  12 from Crypto.PublicKey import RSA
  13 from cachetools import LFUCache
  14
  15 from . import app, CONFIG
  16 from .database import DATABASE
  17 from .http_debug import http_debug
  18 from .remote_actor import fetch_actor
  19 from .http_signatures import sign_headers, generate_body_digest
  20
  21
  22 # generate actor keys if not present
  23 if "actorKeys" not in DATABASE:
  24     logging.info("No actor keys present, generating 4096-bit RSA keypair.")
  25
  26     privkey = RSA.generate(4096)
  27     pubkey = privkey.publickey()
  28
  29     DATABASE["actorKeys"] = {
  30         "publicKey": pubkey.exportKey('PEM').decode('utf-8'),
  31         "privateKey": privkey.exportKey('PEM').decode('utf-8')
  32     }
  33
  34
  35 PRIVKEY = RSA.importKey(DATABASE["actorKeys"]["privateKey"])
  36 PUBKEY = PRIVKEY.publickey()
  37 AP_CONFIG = CONFIG['ap']
  38 CACHE_SIZE = CONFIG.get('cache-size', 16384)
  39 CACHE = LFUCache(CACHE_SIZE)
  40
  41 sem = asyncio.Semaphore(500)
  42
  43
  44 async def actor(request):
  45     data = {
  46         "@context": "https://www.w3.org/ns/activitystreams",
  47         "endpoints": {
  48             "sharedInbox": "https://{}/inbox".format(request.host)
  49         },
  50         "followers": "https://{}/followers".format(request.host),
  51         "following": "https://{}/following".format(request.host),
  52         "inbox": "https://{}/inbox".format(request.host),
  53         "name": "ActivityRelay",
  54         "type": "Application",
  55         "id": "https://{}/actor".format(request.host),
  56         "publicKey": {
  57             "id": "https://{}/actor#main-key".format(request.host),
  58             "owner": "https://{}/actor".format(request.host),
  59             "publicKeyPem": DATABASE["actorKeys"]["publicKey"]
  60         },
  61         "summary": "ActivityRelay bot",
  62         "preferredUsername": "relay",
  63         "url": "https://{}/actor".format(request.host)
  64     }
  65     return aiohttp.web.json_response(data)
  66
  67
  68 app.router.add_get('/actor', actor)
  69 get_actor_inbox = lambda actor: actor.get('endpoints', {}).get('sharedInbox', actor['inbox'])
  70
  71
  72 async def push_message_to_actor(actor, message, our_key_id):
  73     inbox = get_actor_inbox(actor)
  74     url = urlsplit(inbox)
  75
  76     # XXX: Digest
  77     data = json.dumps(message)
  78     headers = {
  79         '(request-target)': 'post {}'.format(url.path),
  80         'Content-Length': str(len(data)),
  81         'Content-Type': 'application/activity+json',
  82         'User-Agent': 'ActivityRelay',
  83         'Host': url.netloc,
  84         'Digest': 'SHA-256={}'.format(generate_body_digest(data)),
  85         'Date': datetime.datetime.utcnow().strftime('%a, %d %b %Y %H:%M:%S GMT')
  86     }
  87     headers['signature'] = sign_headers(headers, PRIVKEY, our_key_id)
  88     headers.pop('(request-target)')
  89     headers.pop('Host')
  90
  91     logging.debug('%r >> %r', inbox, message)
  92
  93     global sem
  94     async with sem:
  95         try:
  96             async with aiohttp.ClientSession(trace_configs=[http_debug()]) as session:
  97                 async with session.post(inbox, data=data, headers=headers) as resp:
  98                     if resp.status == 202:
  99                         return
 100                     resp_payload = await resp.text()
 101                     logging.debug('%r >> resp %r', inbox, resp_payload)
 102         except Exception as e:
 103             logging.info('Caught %r while pushing to %r.', e, inbox)
 104
 105
 106 async def follow_remote_actor(actor_uri):
 107     actor = await fetch_actor(actor_uri)
 108
 109     if not actor:
 110         logging.info('failed to fetch actor at: %r', actor_uri)
 111         return
 112
 113     if AP_CONFIG['whitelist_enabled'] is True and urlsplit(actor_uri).hostname not in AP_CONFIG['whitelist']:
 114         logging.info('refusing to follow non-whitelisted actor: %r', actor_uri)
 115         return
 116
 117     logging.info('following: %r', actor_uri)
 118
 119     message = {
 120         "@context": "https://www.w3.org/ns/activitystreams",
 121         "type": "Follow",
 122         "to": [actor['id']],
 123         "object": actor['id'],
 124         "id": "https://{}/activities/{}".format(AP_CONFIG['host'], uuid.uuid4()),
 125         "actor": "https://{}/actor".format(AP_CONFIG['host'])
 126     }
 127     await push_message_to_actor(actor, message, "https://{}/actor#main-key".format(AP_CONFIG['host']))
 128
 129
 130 async def unfollow_remote_actor(actor_uri):
 131     actor = await fetch_actor(actor_uri)
 132     if not actor:
 133         logging.info('failed to fetch actor at: %r', actor_uri)
 134         return
 135
 136     logging.info('unfollowing: %r', actor_uri)
 137
 138     message = {
 139         "@context": "https://www.w3.org/ns/activitystreams",
 140         "type": "Undo",
 141         "to": [actor['id']],
 142         "object": {
 143              "type": "Follow",
 144              "object": actor_uri,
 145              "actor": actor['id'],
 146              "id": "https://{}/activities/{}".format(AP_CONFIG['host'], uuid.uuid4())
 147         },
 148         "id": "https://{}/activities/{}".format(AP_CONFIG['host'], uuid.uuid4()),
 149         "actor": "https://{}/actor".format(AP_CONFIG['host'])
 150     }
 151     await push_message_to_actor(actor, message, "https://{}/actor#main-key".format(AP_CONFIG['host']))
 152
 153
 154 tag_re = re.compile(r'(<!--.*?-->|<[^>]*>)')
 155 def strip_html(data):
 156     no_tags = tag_re.sub('', data)
 157     return cgi.escape(no_tags)
 158
 159
 160 def distill_inboxes(actor, object_id):
 161     global DATABASE
 162
 163     origin_hostname = urlsplit(object_id).hostname
 164
 165     inbox = get_actor_inbox(actor)
 166     targets = [target for target in DATABASE.get('relay-list', []) if target != inbox]
 167     targets = [target for target in targets if urlsplit(target).hostname != origin_hostname]
 168     hostnames = [urlsplit(target).hostname for target in targets]
 169
 170     assert inbox not in targets
 171     assert origin_hostname not in hostnames
 172
 173     return targets
 174
 175
 176 def distill_object_id(activity):
 177     logging.debug('>> determining object ID for %r', activity['object'])
 178     obj = activity['object']
 179
 180     if isinstance(obj, str):
 181         return obj
 182
 183     return obj['id']
 184
 185
 186 async def handle_relay(actor, data, request):
 187     global CACHE
 188
 189     object_id = distill_object_id(data)
 190
 191     if object_id in CACHE:
 192         logging.debug('>> already relayed %r as %r', object_id, CACHE[object_id])
 193         return
 194
 195     activity_id = "https://{}/activities/{}".format(request.host, uuid.uuid4())
 196
 197     message = {
 198         "@context": "https://www.w3.org/ns/activitystreams",
 199         "type": "Announce",
 200         "to": ["https://{}/followers".format(request.host)],
 201         "actor": "https://{}/actor".format(request.host),
 202         "object": object_id,
 203         "id": activity_id
 204     }
 205
 206     logging.debug('>> relay: %r', message)
 207
 208     inboxes = distill_inboxes(actor, object_id)
 209
 210     futures = [push_message_to_actor({'inbox': inbox}, message, 'https://{}/actor#main-key'.format(request.host)) for inbox in inboxes]
 211     asyncio.ensure_future(asyncio.gather(*futures))
 212
 213     CACHE[object_id] = activity_id
 214
 215
 216 async def handle_forward(actor, data, request):
 217     object_id = distill_object_id(data)
 218
 219     logging.debug('>> Relay %r', data)
 220
 221     inboxes = distill_inboxes(actor, object_id)
 222
 223     futures = [
 224         push_message_to_actor(
 225             {'inbox': inbox},
 226             data,
 227             'https://{}/actor#main-key'.format(request.host))
 228         for inbox in inboxes]
 229     asyncio.ensure_future(asyncio.gather(*futures))
 230
 231
 232 async def handle_follow(actor, data, request):
 233     global DATABASE
 234
 235     following = DATABASE.get('relay-list', [])
 236     inbox = get_actor_inbox(actor)
 237
 238     if urlsplit(inbox).hostname in AP_CONFIG['blocked_instances']:
 239         return
 240
 241     if inbox not in following:
 242         following += [inbox]
 243         DATABASE['relay-list'] = following
 244
 245         asyncio.ensure_future(follow_remote_actor(actor['id']))
 246
 247     message = {
 248         "@context": "https://www.w3.org/ns/activitystreams",
 249         "type": "Accept",
 250         "to": [actor["id"]],
 251         "actor": "https://{}/actor".format(request.host),
 252
 253         # this is wrong per litepub, but mastodon < 2.4 is not compliant with that profile.
 254         "object": {
 255             "type": "Follow",
 256             "id": data["id"],
 257             "object": "https://{}/actor".format(request.host),
 258             "actor": actor["id"]
 259         },
 260
 261         "id": "https://{}/activities/{}".format(request.host, uuid.uuid4()),
 262     }
 263
 264     asyncio.ensure_future(push_message_to_actor(actor, message, 'https://{}/actor#main-key'.format(request.host)))
 265
 266
 267 async def handle_undo(actor, data, request):
 268     global DATABASE
 269
 270     child = data['object']
 271     if child['type'] == 'Follow':
 272         following = DATABASE.get('relay-list', [])
 273
 274         inbox = get_actor_inbox(actor)
 275
 276         if inbox in following:
 277             following.remove(inbox)
 278             DATABASE['relay-list'] = following
 279
 280         await unfollow_remote_actor(actor['id'])
 281
 282
 283 processors = {
 284     'Announce': handle_relay,
 285     'Create': handle_relay,
 286     'Delete': handle_forward,
 287     'Follow': handle_follow,
 288     'Undo': handle_undo,
 289     'Update': handle_forward,
 290 }
 291
 292
 293 async def inbox(request):
 294     data = await request.json()
 295     instance = urlsplit(data['actor']).hostname
 296
 297     if 'actor' not in data or not request['validated']:
 298         raise aiohttp.web.HTTPUnauthorized(body='access denied', content_type='text/plain')
 299
 300     elif data['type'] != 'Follow' and 'https://{}/inbox'.format(instance) not in DATABASE['relay-list']:
 301         raise aiohttp.web.HTTPUnauthorized(body='access denied', content_type='text/plain')
 302
 303     elif AP_CONFIG['whitelist_enabled'] is True and instance not in AP_CONFIG['whitelist']:
 304         raise aiohttp.web.HTTPUnauthorized(body='access denied', content_type='text/plain')
 305
 306     actor = await fetch_actor(data["actor"])
 307     actor_uri = 'https://{}/actor'.format(request.host)
 308
 309     logging.debug(">> payload %r", data)
 310
 311     processor = processors.get(data['type'], None)
 312     if processor:
 313         await processor(actor, data, request)
 314
 315     return aiohttp.web.Response(body=b'{}', content_type='application/activity+json')
 316
 317 app.router.add_post('/inbox', inbox)