fba.py

   1 from bs4 import BeautifulSoup
   2 from hashlib import sha256
   3
   4 import reqto
   5 import re
   6 import sqlite3
   7 import json
   8 import sys
   9 import time
  10
  11 with open("config.json") as f:
  12     config = json.loads(f.read())
  13
  14 blacklist = [
  15     "activitypub-troll.cf",
  16     "gab.best",
  17     "4chan.icu",
  18     "social.shrimpcam.pw",
  19     "mastotroll.netz.org",
  20     "ngrok.io",
  21 ]
  22
  23 headers = {
  24     "user-agent": config["useragent"]
  25 }
  26
  27 connection = sqlite3.connect("blocks.db")
  28 cursor = connection.cursor()
  29
  30 def is_blacklisted(domain: str) -> bool:
  31     blacklisted = False
  32     for peer in blacklist:
  33         if peer in domain:
  34             blacklisted = True
  35
  36     return blacklisted
  37
  38 def get_hash(domain: str) -> str:
  39     return sha256(domain.encode("utf-8")).hexdigest()
  40
  41 def update_last_blocked(domain: str):
  42     # NOISY-DEBUG: print("DEBUG: Updating last_blocked for domain", domain)
  43     try:
  44         cursor.execute("UPDATE instances SET last_blocked = ?, last_updated = ? WHERE domain = ?", [
  45             time.time(),
  46             time.time(),
  47             domain
  48         ])
  49
  50     except:
  51         print("ERROR: failed SQL query:", domain)
  52         sys.exit(255)
  53
  54 def update_last_error(domain: str, res: any):
  55     # NOISY-DEBUG: print("DEBUG: domain,res.status_code:", domain, res.status_code, res.reason)
  56     try:
  57         cursor.execute("UPDATE instances SET last_status_code = ?, last_error_details = ?, last_updated = ? WHERE domain = ?", [
  58             res.status_code,
  59             res.reason,
  60             time.time(),
  61             domain
  62         ])
  63
  64     except:
  65         print("ERROR: failed SQL query:", domain)
  66         sys.exit(255)
  67
  68 def update_last_nodeinfo(domain: str):
  69     # NOISY-DEBUG: print("DEBUG: Updating last_nodeinfo for domain:", domain)
  70     try:
  71         cursor.execute("UPDATE instances SET last_nodeinfo = ?, last_updated = ? WHERE domain = ?", [
  72             time.time(),
  73             time.time(),
  74             domain
  75         ])
  76
  77     except:
  78         print("ERROR: failed SQL query:", domain)
  79         sys.exit(255)
  80
  81     connection.commit()
  82
  83 def get_peers(domain: str) -> list:
  84     # NOISY-DEBUG: print("DEBUG: Getting peers for domain:", domain)
  85     peers = None
  86
  87     try:
  88         res = reqto.get(f"https://{domain}/api/v1/instance/peers", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
  89
  90         if not res.ok or res.status_code >= 400:
  91             print("WARNING: Cannot fetch peers:", domain)
  92             update_last_error(domain, res)
  93         else:
  94             # NOISY-DEBUG: print("DEBUG: Querying API was successful:", domain, len(res.json()))
  95             peers = res.json()
  96
  97     except:
  98         print("WARNING: Some error during get():", domain)
  99
 100     update_last_nodeinfo(domain)
 101
 102     # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers))
 103     return peers
 104
 105 def post_json_api(domain: str, path: str, data: str) -> list:
 106     # NOISY-DEBUG: print("DEBUG: Sending POST to domain,path,data:", domain, path, data)
 107     json = {}
 108     try:
 109         res = reqto.post(f"https://{domain}{path}", data=data, headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
 110
 111         if not res.ok or res.status_code >= 400:
 112             print("WARNING: Cannot query JSON API:", domain, path, data, res.status_code)
 113             update_last_error(domain, res)
 114             raise
 115
 116         update_last_nodeinfo(domain)
 117         json = res.json()
 118     except:
 119         print("WARNING: Some error during post():", domain, path, data)
 120
 121     # NOISY-DEBUG: print("DEBUG: Returning json():", len(json))
 122     return json
 123
 124 def fetch_nodeinfo(domain: str) -> list:
 125     # NOISY-DEBUG: print("DEBUG: Fetching nodeinfo from domain:", domain)
 126
 127     requests = [
 128        f"https://{domain}/nodeinfo/2.1.json",
 129        f"https://{domain}/nodeinfo/2.1",
 130        f"https://{domain}/nodeinfo/2.0.json",
 131        f"https://{domain}/nodeinfo/2.0",
 132        f"https://{domain}/nodeinfo/1.0",
 133        f"https://{domain}/api/v1/instance"
 134     ]
 135
 136     json = {}
 137     for request in requests:
 138         try:
 139             # NOISY-DEBUG: print("DEBUG: Fetching request:", request)
 140             res = reqto.get(request, headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
 141
 142             # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json()))
 143             if res.ok and res.json() is not None:
 144                 # NOISY-DEBUG: print("DEBUG: Success:", request)
 145                 json = res.json()
 146                 break
 147             elif not res.ok or res.status_code >= 400:
 148                 # NOISY-DEBUG: print("DEBUG: Failed fetching nodeinfo from domain:", domain)
 149                 update_last_error(domain, res)
 150                 continue
 151
 152         except:
 153             # NOISY-DEBUG: print("DEBUG: Cannot fetch API request:", request)
 154             pass
 155
 156     if json is None:
 157         print("WARNING: Failed fetching nodeinfo from domain:", domain)
 158
 159     # NOISY-DEBUG: print("DEBUG: Returning json[]:", type(json))
 160     return json
 161
 162 def determine_software(domain: str) -> str:
 163     # NOISY-DEBUG: print("DEBUG: Determining software for domain:", domain)
 164     software = None
 165
 166     json = fetch_nodeinfo(domain)
 167     if len(json) == 0:
 168         print("DEBUG: Could not determine software type:", domain)
 169         return None
 170
 171     # NOISY-DEBUG: print("DEBUG: json():", len(json))
 172     software = tidyup(json["software"]["name"])
 173
 174     # NOISY-DEBUG: print("DEBUG: tidyup software:", software)
 175     if software in ["akkoma", "rebased"]:
 176         # NOISY-DEBUG: print("DEBUG: Setting pleroma:", domain, software)
 177         software = "pleroma"
 178     elif software in ["hometown", "ecko"]:
 179         # NOISY-DEBUG: print("DEBUG: Setting mastodon:", domain, software)
 180         software = "mastodon"
 181     elif software in ["calckey", "groundpolis", "foundkey", "cherrypick"]:
 182         # NOISY-DEBUG: print("DEBUG: Setting misskey:", domain, software)
 183         software = "misskey"
 184     elif software.find("/") > 0:
 185         print("WARNING: Spliting of path:", software)
 186         software = software.split("/")[-1];
 187
 188     if software == "":
 189         print("WARNING: tidyup() left no software name behind:", domain)
 190         software = None
 191
 192     # NOISY-DEBUG: print("DEBUG: Returning domain,software:", domain, software)
 193     return software
 194
 195 def update_block_reason(reason: str, blocker: str, blocked: str, block_level: str):
 196     # NOISY: print("--- Updating block reason:", reason, blocker, blocked, block_level)
 197     try:
 198         cursor.execute(
 199             "UPDATE blocks SET reason = ?, last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ? AND reason = ''",
 200             (
 201                 reason,
 202                 time.time(),
 203                 blocker,
 204                 blocked,
 205                 block_level
 206             ),
 207         )
 208
 209     except:
 210         print("ERROR: failed SQL query:", reason, blocker, blocked, block_level)
 211         sys.exit(255)
 212
 213 def update_last_seen(blocker: str, blocked: str, block_level: str):
 214     # NOISY: print("--- Updating last_seen for:", blocker, blocked, block_level)
 215     try:
 216         cursor.execute(
 217             "UPDATE blocks SET last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ?",
 218             (
 219                 time.time(),
 220                 blocker,
 221                 blocked,
 222                 block_level
 223             )
 224         )
 225
 226     except:
 227         print("ERROR: failed SQL query:", last_seen, blocker, blocked, block_level)
 228         sys.exit(255)
 229
 230 def block_instance(blocker: str, blocked: str, reason: str, block_level: str):
 231     # NOISY-DEBUG: print("DEBUG: blocker,blocked,reason,block_level:", blocker, blocked, reason, block_level)
 232     if blocker.find("@") > 0:
 233         print("WARNING: Bad blocker:", blocker)
 234         raise
 235     elif blocked.find("@") > 0:
 236         print("WARNING: Bad blocked:", blocked)
 237         raise
 238
 239     print("--- New block:", blocker, blocked, reason, block_level, first_added, last_seen)
 240     try:
 241         cursor.execute(
 242             "INSERT INTO blocks (blocker, blocked, reason, block_level, first_added, last_seen) VALUES(?, ?, ?, ?, ?, ?)",
 243              (
 244                  blocker,
 245                  blocked,
 246                  reason,
 247                  block_level,
 248                  time.time(),
 249                  time.time()
 250              ),
 251         )
 252
 253     except:
 254         print("ERROR: failed SQL query:", blocker, blocked, reason, block_level, first_added, last_seen)
 255         sys.exit(255)
 256
 257 def add_instance(domain: str, origin: str, originator: str):
 258     # NOISY-DEBUG: print("DEBUG: domain,origin:", domain, origin, originator)
 259     if domain.find("@") > 0:
 260         print("WARNING: Bad domain name:", domain)
 261         raise
 262     elif origin is not None and origin.find("@") > 0:
 263         print("WARNING: Bad origin name:", origin)
 264         raise
 265
 266     software = determine_software(domain)
 267     # NOISY-DEBUG: print("DEBUG: Determined software:", software)
 268
 269     print(f"--- Adding new instance {domain} (origin: {origin})")
 270     try:
 271         cursor.execute(
 272             "INSERT INTO instances (domain, origin, originator, hash, software, first_seen) VALUES (?, ?, ?, ?, ?, ?)",
 273             (
 274                domain,
 275                origin,
 276                originator,
 277                get_hash(domain),
 278                software,
 279                time.time()
 280             ),
 281         )
 282
 283     except:
 284         print("ERROR: failed SQL query:", domain)
 285         sys.exit(255)
 286     else:
 287         # NOISY-DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain)
 288         update_last_nodeinfo(domain)
 289
 290 def send_bot_post(instance: str, blocks: dict):
 291     message = instance + " has blocked the following instances:\n\n"
 292     truncated = False
 293
 294     if len(blocks) > 20:
 295         truncated = True
 296         blocks = blocks[0 : 19]
 297
 298     for block in blocks:
 299         if block["reason"] == None or block["reason"] == '':
 300             message = message + block["blocked"] + " with unspecified reason\n"
 301         else:
 302             if len(block["reason"]) > 420:
 303                 block["reason"] = block["reason"][0:419] + "[…]"
 304
 305             message = message + block["blocked"] + ' for "' + block["reason"].replace("@", "@\u200b") + '"\n'
 306
 307     if truncated:
 308         message = message + "(the list has been truncated to the first 20 entries)"
 309
 310     botheaders = {**headers, **{"Authorization": "Bearer " + config["bot_token"]}}
 311
 312     req = reqto.post(f"{config['bot_instance']}/api/v1/statuses",
 313         data={"status":message, "visibility":config['bot_visibility'], "content_type":"text/plain"},
 314         headers=botheaders, timeout=10).json()
 315
 316     return True
 317
 318 def get_mastodon_blocks(domain: str) -> dict:
 319     # NOISY-DEBUG: print("DEBUG: Fetching mastodon blocks from domain:", domain)
 320     blocks = {
 321         "Suspended servers": [],
 322         "Filtered media"   : [],
 323         "Limited servers"  : [],
 324         "Silenced servers" : [],
 325     }
 326
 327     translations = {
 328         "Silenced instances": "Silenced servers",
 329         "Suspended instances": "Suspended servers",
 330         "Gesperrte Server": "Suspended servers",
 331         "Gefilterte Medien": "Filtered media",
 332         "Stummgeschaltete Server": "Silenced servers",
 333         "停止済みのサーバー": "Suspended servers",
 334         "メディアを拒否しているサーバー": "Filtered media",
 335         "サイレンス済みのサーバー": "Silenced servers",
 336         "שרתים מושעים": "Suspended servers",
 337         "מדיה מסוננת": "Filtered media",
 338         "שרתים מוגבלים": "Silenced servers",
 339         "Serveurs suspendus": "Suspended servers",
 340         "Médias filtrés": "Filtered media",
 341         "Serveurs limités": "Silenced servers",
 342     }
 343
 344     try:
 345         doc = BeautifulSoup(
 346             reqto.get(f"https://{domain}/about/more", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
 347             "html.parser",
 348         )
 349     except:
 350         print("ERROR: Cannot fetch from domain:", domain)
 351         return {}
 352
 353     for header in doc.find_all("h3"):
 354         header_text = header.text
 355
 356         if header_text in translations:
 357             header_text = translations[header_text]
 358
 359         if header_text in blocks:
 360             # replaced find_next_siblings with find_all_next to account for instances that e.g. hide lists in dropdown menu
 361             for line in header.find_all_next("table")[0].find_all("tr")[1:]:
 362                 blocks[header_text].append(
 363                     {
 364                         "domain": tidyup(line.find("span").text),
 365                         "hash"  : tidyup(line.find("span")["title"][9:]),
 366                         "reason": tidyup(line.find_all("td")[1].text),
 367                     }
 368                 )
 369
 370     # NOISY-DEBUG: print("DEBUG: Returning blocks for domain:", domain)
 371     return {
 372         "reject"        : blocks["Suspended servers"],
 373         "media_removal" : blocks["Filtered media"],
 374         "followers_only": blocks["Limited servers"] + blocks["Silenced servers"],
 375     }
 376
 377 def get_friendica_blocks(domain: str) -> dict:
 378     # NOISY-DEBUG: print("DEBUG: Fetching friendica blocks from domain:", domain)
 379     blocks = []
 380
 381     try:
 382         doc = BeautifulSoup(
 383             reqto.get(f"https://{domain}/friendica", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
 384             "html.parser",
 385         )
 386     except:
 387         print("WARNING: Failed to fetch /friendica from domain:", domain)
 388         return {}
 389
 390     blocklist = doc.find(id="about_blocklist")
 391
 392     # Prevents exceptions:
 393     if blocklist is None:
 394         # NOISY-DEBUG: print("DEBUG: Instance has no block list:", domain)
 395         return {}
 396
 397     for line in blocklist.find("table").find_all("tr")[1:]:
 398         blocks.append({
 399             "domain": tidyup(line.find_all("td")[0].text),
 400             "reason": tidyup(line.find_all("td")[1].text)
 401         })
 402
 403     # NOISY-DEBUG: print("DEBUG: Returning blocks() for domain:", domain, len(blocks))
 404     return {
 405         "reject": blocks
 406     }
 407
 408 def get_misskey_blocks(domain: str) -> dict:
 409     # NOISY-DEBUG: print("DEBUG: Fetching misskey blocks from domain:", domain)
 410     blocks = {
 411         "suspended": [],
 412         "blocked": []
 413     }
 414
 415     try:
 416         counter = 0
 417         step = 99
 418         while True:
 419             # iterating through all "suspended" (follow-only in its terminology)
 420             # instances page-by-page, since that troonware doesn't support
 421             # sending them all at once
 422             try:
 423                 if counter == 0:
 424                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 425                     doc = post_json_api(domain, "/api/federation/instances/", json.dumps({
 426                         "sort"     : "+caughtAt",
 427                         "host"     : None,
 428                         "suspended": True,
 429                         "limit"    : step
 430                     }))
 431                 else:
 432                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 433                     doc = post_json_api(domain, "/api/federation/instances/", json.dumps({
 434                         "sort"     : "+caughtAt",
 435                         "host"     : None,
 436                         "suspended": True,
 437                         "limit"    : step,
 438                         "offset"   : counter-1
 439                     }))
 440
 441                 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
 442                 if len(doc) == 0:
 443                     # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
 444                     break
 445
 446                 for instance in doc:
 447                     # just in case
 448                     if instance["isSuspended"]:
 449                         blocks["suspended"].append(
 450                             {
 451                                 "domain": tidyup(instance["host"]),
 452                                 # no reason field, nothing
 453                                 "reason": ""
 454                             }
 455                         )
 456
 457                 if len(doc) < step:
 458                     # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
 459                     break
 460
 461                 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
 462                 counter = counter + step
 463
 464             except:
 465                 print("WARNING: Caught error, exiting loop:", domain)
 466                 counter = 0
 467                 break
 468
 469         while True:
 470             # same shit, different asshole ("blocked" aka full suspend)
 471             try:
 472                 if counter == 0:
 473                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 474                     doc = post_json_api(domain,"/api/federation/instances", json.dumps({
 475                         "sort"   : "+caughtAt",
 476                         "host"   : None,
 477                         "blocked": True,
 478                         "limit"  : step
 479                     }))
 480                 else:
 481                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 482                     doc = post_json_api(domain,"/api/federation/instances", json.dumps({
 483                         "sort"   : "+caughtAt",
 484                         "host"   : None,
 485                         "blocked": True,
 486                         "limit"  : step,
 487                         "offset" : counter-1
 488                     }))
 489
 490                 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
 491                 if len(doc) == 0:
 492                     # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
 493                     break
 494
 495                 for instance in doc:
 496                     if instance["isBlocked"]:
 497                         blocks["blocked"].append({
 498                             "domain": tidyup(instance["host"]),
 499                             "reason": ""
 500                         })
 501
 502                 if len(doc) < step:
 503                     # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
 504                     break
 505
 506                 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
 507                 counter = counter + step
 508
 509             except:
 510                 counter = 0
 511                 break
 512
 513         # NOISY-DEBUG: print("DEBUG: Returning for domain,blocked(),suspended():", domain, len(blocks["blocked"]), len(blocks["suspended"]))
 514         return {
 515             "reject"        : blocks["blocked"],
 516             "followers_only": blocks["suspended"]
 517         }
 518
 519     except:
 520         print("WARNING: API request failed for domain:", domain)
 521         return {}
 522
 523 def tidyup(string: str) -> str:
 524     # some retards put their blocks in variable case
 525     string = string.lower().strip()
 526
 527     # other retards put the port
 528     string = re.sub("\:\d+$", "", string)
 529
 530     # bigger retards put the schema in their blocklist, sometimes even without slashes
 531     string = re.sub("^https?\:(\/*)", "", string)
 532
 533     # and trailing slash
 534     string = re.sub("\/$", "", string)
 535
 536     # and the @
 537     string = re.sub("^\@", "", string)
 538
 539     # the biggest retards of them all try to block individual users
 540     string = re.sub("(.+)\@", "", string)
 541
 542     return string