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