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/2.0",
 134        f"https://{domain}/nodeinfo/2.0.json",
 135        f"https://{domain}/nodeinfo/2.1",
 136        f"https://{domain}/api/v1/instance"
 137     ]
 138
 139     json = None
 140     for request in requests:
 141         # NOISY-DEBUG: print("DEBUG: Fetching request:", request)
 142         res = reqto.get(request, headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
 143
 144         # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json()))
 145         if res.ok and res.json() is not None:
 146             # NOISY-DEBUG: print("DEBUG: Success:", request)
 147             json = res.json()
 148             break
 149         elif not res.ok or res.status_code >= 400:
 150             # NOISY-DEBUG: print("DEBUG: Failed fetching nodeinfo from domain:", domain)
 151             update_last_error(domain, res)
 152             continue
 153
 154     if json is None:
 155         print("WARNING: Failed fetching nodeinfo from domain:", domain)
 156
 157     # NOISY-DEBUG: print("DEBUG: Updating last_nodeinfo for domain:", domain)
 158     update_last_nodeinfo(domain)
 159
 160     # NOISY-DEBUG: print("DEBUG: Returning json():", len(json))
 161     return json
 162
 163 def determine_software(domain: str) -> str:
 164     # NOISY-DEBUG: print("DEBUG: Determining software for domain:", domain)
 165     software = None
 166
 167     try:
 168         json = fetch_nodeinfo(domain)
 169
 170         # NOISY-DEBUG: print("DEBUG: json():", len(json))
 171         software = tidyup(json["software"]["name"])
 172
 173         # NOISY-DEBUG: print("DEBUG: BEFORE software:", software)
 174         if software in ["akkoma", "rebased"]:
 175             # NOISY-DEBUG: print("DEBUG: Setting pleroma:", domain, software)
 176             software = "pleroma"
 177         elif software in ["hometown", "ecko"]:
 178             # NOISY-DEBUG: print("DEBUG: Setting mastodon:", domain, software)
 179             software = "mastodon"
 180         elif software in ["calckey", "groundpolis", "foundkey", "cherrypick"]:
 181             # NOISY-DEBUG: print("DEBUG: Setting misskey:", domain, software)
 182             software = "misskey"
 183         elif software.find("/") > 0:
 184             print("WARNING: Spliting of path:", software)
 185             software = software.split("/")[-1];
 186
 187         if software == "":
 188             print("WARNING: tidyup() left no software name behind:", domain)
 189             software = None
 190         # NOISY-DEBUG: print("DEBUG: AFTER software:", software)
 191     except:
 192         print("WARNING: Could not determine software type:", domain)
 193
 194     # NOISY-DEBUG: print("DEBUG: Returning domain,software:", domain, software)
 195     return software
 196
 197 def update_block_reason(reason: str, blocker: str, blocked: str, block_level: str):
 198     # NOISY: print("--- Updating block reason:", reason, blocker, blocked, block_level)
 199     try:
 200         cursor.execute(
 201             "UPDATE blocks SET reason = ?, last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ? AND reason = ''",
 202             (
 203                 reason,
 204                 time.time(),
 205                 blocker,
 206                 blocked,
 207                 block_level
 208             ),
 209         )
 210
 211     except:
 212         print("ERROR: failed SQL query:", reason, blocker, blocked, block_level)
 213         sys.exit(255)
 214
 215 def update_last_seen(blocker: str, blocked: str, block_level: str):
 216     # NOISY: print("--- Updating last_seen for:", blocker, blocked, block_level)
 217     try:
 218         cursor.execute(
 219             "UPDATE blocks SET last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ?",
 220             (
 221                 time.time(),
 222                 blocker,
 223                 blocked,
 224                 block_level
 225             )
 226         )
 227
 228     except:
 229         print("ERROR: failed SQL query:", last_seen, blocker, blocked, block_level)
 230         sys.exit(255)
 231
 232 def block_instance(blocker: str, blocked: str, reason: str, block_level: str):
 233     # NOISY-DEBUG: print("DEBUG: blocker,blocked,reason,block_level:", blocker, blocked, reason, block_level)
 234     if blocker.find("@") > 0:
 235         print("WARNING: Bad blocker:", blocker)
 236         raise
 237     elif blocked.find("@") > 0:
 238         print("WARNING: Bad blocked:", blocked)
 239         raise
 240
 241     print("--- New block:", blocker, blocked, reason, block_level, first_added, last_seen)
 242     try:
 243         cursor.execute(
 244             "INSERT INTO blocks (blocker, blocked, reason, block_level, first_added, last_seen) VALUES(?, ?, ?, ?, ?, ?)",
 245              (
 246                  blocker,
 247                  blocked,
 248                  reason,
 249                  block_level,
 250                  time.time(),
 251                  time.time()
 252              ),
 253         )
 254
 255     except:
 256         print("ERROR: failed SQL query:", blocker, blocked, reason, block_level, first_added, last_seen)
 257         sys.exit(255)
 258
 259 def add_instance(domain: str, origin: str, originator: str):
 260     # NOISY-DEBUG: print("DEBUG: domain,origin:", domain, origin, originator)
 261     if domain.find("@") > 0:
 262         print("WARNING: Bad domain name:", domain)
 263         raise
 264     elif origin is not None and origin.find("@") > 0:
 265         print("WARNING: Bad origin name:", origin)
 266         raise
 267
 268     print(f"--- Adding new instance {domain} (origin: {origin})")
 269     try:
 270         cursor.execute(
 271             "INSERT INTO instances (domain, origin, originator, hash, software, first_seen) VALUES (?, ?, ?, ?, ?, ?)",
 272             (
 273                domain,
 274                origin,
 275                originator,
 276                get_hash(domain),
 277                determine_software(domain),
 278                time.time()
 279             ),
 280         )
 281
 282     except:
 283         print("ERROR: failed SQL query:", domain)
 284         sys.exit(255)
 285
 286 def send_bot_post(instance: str, blocks: dict):
 287     message = instance + " has blocked the following instances:\n\n"
 288     truncated = False
 289
 290     if len(blocks) > 20:
 291         truncated = True
 292         blocks = blocks[0 : 19]
 293
 294     for block in blocks:
 295         if block["reason"] == None or block["reason"] == '':
 296             message = message + block["blocked"] + " with unspecified reason\n"
 297         else:
 298             if len(block["reason"]) > 420:
 299                 block["reason"] = block["reason"][0:419] + "[…]"
 300
 301             message = message + block["blocked"] + ' for "' + block["reason"].replace("@", "@\u200b") + '"\n'
 302
 303     if truncated:
 304         message = message + "(the list has been truncated to the first 20 entries)"
 305
 306     botheaders = {**headers, **{"Authorization": "Bearer " + config["bot_token"]}}
 307
 308     req = reqto.post(f"{config['bot_instance']}/api/v1/statuses",
 309         data={"status":message, "visibility":config['bot_visibility'], "content_type":"text/plain"},
 310         headers=botheaders, timeout=10).json()
 311
 312     return True
 313
 314 def get_mastodon_blocks(domain: str) -> dict:
 315     # NOISY-DEBUG: print("DEBUG: Fetching mastodon blocks from domain:", domain)
 316     blocks = {
 317         "Suspended servers": [],
 318         "Filtered media": [],
 319         "Limited servers": [],
 320         "Silenced servers": [],
 321     }
 322
 323     translations = {
 324         "Silenced instances": "Silenced servers",
 325         "Suspended instances": "Suspended servers",
 326         "Gesperrte Server": "Suspended servers",
 327         "Gefilterte Medien": "Filtered media",
 328         "Stummgeschaltete Server": "Silenced servers",
 329         "停止済みのサーバー": "Suspended servers",
 330         "メディアを拒否しているサーバー": "Filtered media",
 331         "サイレンス済みのサーバー": "Silenced servers",
 332         "שרתים מושעים": "Suspended servers",
 333         "מדיה מסוננת": "Filtered media",
 334         "שרתים מוגבלים": "Silenced servers",
 335         "Serveurs suspendus": "Suspended servers",
 336         "Médias filtrés": "Filtered media",
 337         "Serveurs limités": "Silenced servers",
 338     }
 339
 340     try:
 341         doc = BeautifulSoup(
 342             reqto.get(f"https://{domain}/about/more", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
 343             "html.parser",
 344         )
 345     except:
 346         print("ERROR: Cannot fetch from domain:", domain)
 347         return {}
 348
 349     for header in doc.find_all("h3"):
 350         header_text = header.text
 351
 352         if header_text in translations:
 353             header_text = translations[header_text]
 354
 355         if header_text in blocks:
 356             # replaced find_next_siblings with find_all_next to account for instances that e.g. hide lists in dropdown menu
 357             for line in header.find_all_next("table")[0].find_all("tr")[1:]:
 358                 blocks[header_text].append(
 359                     {
 360                         "domain": line.find("span").text,
 361                         "hash": line.find("span")["title"][9:],
 362                         "reason": line.find_all("td")[1].text.strip(),
 363                     }
 364                 )
 365
 366     # NOISY-DEBUG: print("DEBUG: Returning blocks for domain:", domain)
 367     return {
 368         "reject": blocks["Suspended servers"],
 369         "media_removal": blocks["Filtered media"],
 370         "followers_only": blocks["Limited servers"] + blocks["Silenced servers"],
 371     }
 372
 373 def get_friendica_blocks(domain: str) -> dict:
 374     # NOISY-DEBUG: print("DEBUG: Fetching friendica blocks from domain:", domain)
 375     blocks = []
 376
 377     try:
 378         doc = BeautifulSoup(
 379             reqto.get(f"https://{domain}/friendica", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
 380             "html.parser",
 381         )
 382     except:
 383         print("WARNING: Failed to fetch /friendica from domain:", domain)
 384         return {}
 385
 386     blocklist = doc.find(id="about_blocklist")
 387
 388     # Prevents exceptions:
 389     if blocklist is None:
 390         # NOISY-DEBUG: print("DEBUG: Instance has no block list:", domain)
 391         return {}
 392
 393     for line in blocklist.find("table").find_all("tr")[1:]:
 394         blocks.append({
 395             "domain": line.find_all("td")[0].text.strip(),
 396             "reason": line.find_all("td")[1].text.strip()
 397         })
 398
 399     # NOISY-DEBUG: print("DEBUG: Returning blocks() for domain:", domain, len(blocks))
 400     return {
 401         "reject": blocks
 402     }
 403
 404 def get_misskey_blocks(domain: str) -> dict:
 405     # NOISY-DEBUG: print("DEBUG: Fetching misskey blocks from domain:", domain)
 406     blocks = {
 407         "suspended": [],
 408         "blocked": []
 409     }
 410
 411     try:
 412         counter = 0
 413         step = 99
 414         while True:
 415             # iterating through all "suspended" (follow-only in its terminology)
 416             # instances page-by-page, since that troonware doesn't support
 417             # sending them all at once
 418             try:
 419                 if counter == 0:
 420                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 421                     doc = post_json_api(domain, "/api/federation/instances/", json.dumps({
 422                         "sort"     : "+caughtAt",
 423                         "host"     : None,
 424                         "suspended": True,
 425                         "limit"    : step
 426                     }))
 427                 else:
 428                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 429                     doc = post_json_api(domain, "/api/federation/instances/", json.dumps({
 430                         "sort"     : "+caughtAt",
 431                         "host"     : None,
 432                         "suspended": True,
 433                         "limit"    : step,
 434                         "offset"   : counter-1
 435                     }))
 436
 437                 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
 438                 if len(doc) == 0:
 439                     # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
 440                     break
 441
 442                 for instance in doc:
 443                     # just in case
 444                     if instance["isSuspended"]:
 445                         blocks["suspended"].append(
 446                             {
 447                                 "domain": instance["host"],
 448                                 # no reason field, nothing
 449                                 "reason": ""
 450                             }
 451                         )
 452
 453                 if len(doc) < step:
 454                     # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
 455                     break
 456
 457                 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
 458                 counter = counter + step
 459
 460             except:
 461                 print("WARNING: Caught error, exiting loop:", domain)
 462                 counter = 0
 463                 break
 464
 465         while True:
 466             # same shit, different asshole ("blocked" aka full suspend)
 467             try:
 468                 if counter == 0:
 469                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 470                     doc = post_json_api(domain,"/api/federation/instances", json.dumps({
 471                         "sort"   : "+caughtAt",
 472                         "host"   : None,
 473                         "blocked": True,
 474                         "limit"  : step
 475                     }))
 476                 else:
 477                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 478                     doc = post_json_api(domain,"/api/federation/instances", json.dumps({
 479                         "sort"   : "+caughtAt",
 480                         "host"   : None,
 481                         "blocked": True,
 482                         "limit"  : step,
 483                         "offset" : counter-1
 484                     }))
 485
 486                 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
 487                 if len(doc) == 0:
 488                     # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
 489                     break
 490
 491                 for instance in doc:
 492                     if instance["isBlocked"]:
 493                         blocks["blocked"].append({
 494                             "domain": instance["host"],
 495                             "reason": ""
 496                         })
 497
 498                 if len(doc) < step:
 499                     # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
 500                     break
 501
 502                 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
 503                 counter = counter + step
 504
 505             except:
 506                 counter = 0
 507                 break
 508
 509         # NOISY-DEBUG: print("DEBUG: Returning for domain,blocked(),suspended():", domain, len(blocks["blocked"]), len(blocks["suspended"]))
 510         return {
 511             "reject"        : blocks["blocked"],
 512             "followers_only": blocks["suspended"]
 513         }
 514
 515     except:
 516         print("WARNING: API request failed for domain:", domain)
 517         return {}
 518
 519 def tidyup(domain: str) -> str:
 520     # some retards put their blocks in variable case
 521     domain = domain.lower()
 522
 523     # other retards put the port
 524     domain = re.sub("\:\d+$", "", domain)
 525
 526     # bigger retards put the schema in their blocklist, sometimes even without slashes
 527     domain = re.sub("^https?\:(\/*)", "", domain)
 528
 529     # and trailing slash
 530     domain = re.sub("\/$", "", domain)
 531
 532     # and the @
 533     domain = re.sub("^\@", "", domain)
 534
 535     # the biggest retards of them all try to block individual users
 536     domain = re.sub("(.+)\@", "", domain)
 537
 538     return domain