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 conn = sqlite3.connect("blocks.db")
  28 c = conn.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         c.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)
  61
  62     try:
  63         c.execute("UPDATE instances SET last_status_code = ?, last_error_details = ?, last_updated = ? WHERE domain = ?", [
  64             res.status_code,
  65             res.text,
  66             time.time(),
  67             domain
  68         ])
  69
  70     except:
  71         print("ERROR: failed SQL query:", domain)
  72         sys.exit(255)
  73
  74 def update_last_nodeinfo(domain: str):
  75     # NOISY-DEBUG: print("DEBUG: Updating last_nodeinfo for domain:", domain)
  76
  77     try:
  78         c.execute("UPDATE instances SET last_nodeinfo = ?, last_updated = ? WHERE domain = ?", [
  79             time.time(),
  80             time.time(),
  81             domain
  82         ])
  83
  84     except:
  85         print("ERROR: failed SQL query:", domain)
  86         sys.exit(255)
  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["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     try:
 112         # NOISY-DEBUG: print("DEBUG: Sending POST to domain,path,data:", domain, path, data)
 113         res = reqto.post(f"https://{domain}{path}", data=data, headers=headers, timeout=config["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["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         # NOISY-DEBUG: print("DEBUG: json():", len(json))
 170
 171         if json["software"]["name"] in ["akkoma", "rebased"]:
 172             # NOISY-DEBUG: print("DEBUG: Setting pleroma:", domain, json["software"]["name"])
 173             software = "pleroma"
 174         elif json["software"]["name"] in ["hometown", "ecko"]:
 175             # NOISY-DEBUG: print("DEBUG: Setting mastodon:", domain, json["software"]["name"])
 176             software = "mastodon"
 177         elif json["software"]["name"] in ["calckey", "groundpolis", "foundkey", "cherrypick"]:
 178             # NOISY-DEBUG: print("DEBUG: Setting misskey:", domain, json["software"]["name"])
 179             software = "misskey"
 180         else:
 181             # NOISY-DEBUG: print("DEBUG: Using name:", domain, json["software"]["name"])
 182             software = json["software"]["name"].lower()
 183
 184     except:
 185         print("WARNING: Could not determine software type:", domain)
 186
 187     # NOISY-DEBUG: print("DEBUG: Returning domain,software:", domain, software)
 188     return software
 189
 190 def update_block_reason(reason: str, blocker: str, blocked: str, block_level: str):
 191     # NOISY: print("--- Updating block reason:", reason, blocker, blocked, block_level)
 192     try:
 193         c.execute(
 194             "UPDATE blocks SET reason = ?, last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ? AND reason = ''",
 195             (
 196                 reason,
 197                 time.time(),
 198                 blocker,
 199                 blocked,
 200                 block_level
 201             ),
 202         )
 203
 204     except:
 205         print("ERROR: failed SQL query:", reason, blocker, blocked, block_level)
 206         sys.exit(255)
 207
 208 def update_last_seen(blocker: str, blocked: str, block_level: str):
 209     # NOISY: print("--- Updating last_seen for:", blocker, blocked, block_level)
 210     try:
 211         c.execute(
 212             "UPDATE blocks SET last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ?",
 213             (
 214                 time.time(),
 215                 blocker,
 216                 blocked,
 217                 block_level
 218             )
 219         )
 220
 221     except:
 222         print("ERROR: failed SQL query:", last_seen, blocker, blocked, block_level)
 223         sys.exit(255)
 224
 225 def block_instance(blocker: str, blocked: str, reason: str, block_level: str):
 226     # NOISY-DEBUG: print("DEBUG: blocker,blocked,reason,block_level:", blocker, blocked, reason, block_level)
 227     if blocker.find("@") > 0:
 228         print("WARNING: Bad blocker:", blocker)
 229         raise
 230     elif blocked.find("@") > 0:
 231         print("WARNING: Bad blocked:", blocked)
 232         raise
 233
 234     print("--- New block:", blocker, blocked, reason, block_level, first_added, last_seen)
 235     try:
 236         c.execute(
 237             "INSERT INTO blocks (blocker, blocked, reason, block_level, first_added, last_seen) VALUES(?, ?, ?, ?, ?, ?)",
 238              (
 239                  blocker,
 240                  blocked,
 241                  reason,
 242                  block_level,
 243                  time.time(),
 244                  time.time()
 245              ),
 246         )
 247
 248     except:
 249         print("ERROR: failed SQL query:", blocker, blocked, reason, block_level, first_added, last_seen)
 250         sys.exit(255)
 251
 252 def add_instance(domain: str, origin: str, originator: str):
 253     # NOISY-DEBUG: print("DEBUG: domain,origin:", domain, origin, originator)
 254     if domain.find("@") > 0:
 255         print("WARNING: Bad domain name:", domain)
 256         raise
 257     elif origin is not None and origin.find("@") > 0:
 258         print("WARNING: Bad origin name:", origin)
 259         raise
 260
 261     print(f"--- Adding new instance {domain} (origin: {origin})")
 262     try:
 263         c.execute(
 264             "INSERT INTO instances (domain, origin, originator, hash, software, first_seen) VALUES (?, ?, ?, ?, ?, ?)",
 265             (
 266                domain,
 267                origin,
 268                originator,
 269                get_hash(domain),
 270                determine_software(domain),
 271                time.time()
 272             ),
 273         )
 274
 275     except:
 276         print("ERROR: failed SQL query:", domain)
 277         sys.exit(255)
 278
 279 def send_bot_post(instance: str, blocks: dict):
 280     message = instance + " has blocked the following instances:\n\n"
 281     truncated = False
 282
 283     if len(blocks) > 20:
 284         truncated = True
 285         blocks = blocks[0 : 19]
 286
 287     for block in blocks:
 288         if block["reason"] == None or block["reason"] == '':
 289             message = message + block["blocked"] + " with unspecified reason\n"
 290         else:
 291             if len(block["reason"]) > 420:
 292                 block["reason"] = block["reason"][0:419] + "[…]"
 293
 294             message = message + block["blocked"] + ' for "' + block["reason"].replace("@", "@\u200b") + '"\n'
 295
 296     if truncated:
 297         message = message + "(the list has been truncated to the first 20 entries)"
 298
 299     botheaders = {**headers, **{"Authorization": "Bearer " + config["bot_token"]}}
 300
 301     req = reqto.post(f"{config['bot_instance']}/api/v1/statuses",
 302         data={"status":message, "visibility":config['bot_visibility'], "content_type":"text/plain"},
 303         headers=botheaders, timeout=10).json()
 304
 305     return True
 306
 307 def get_mastodon_blocks(domain: str) -> dict:
 308     # NOISY-DEBUG: print("DEBUG: Fetching mastodon blocks from domain:", domain)
 309     blocks = {
 310         "Suspended servers": [],
 311         "Filtered media": [],
 312         "Limited servers": [],
 313         "Silenced servers": [],
 314     }
 315
 316     translations = {
 317         "Silenced instances": "Silenced servers",
 318         "Suspended instances": "Suspended servers",
 319         "Gesperrte Server": "Suspended servers",
 320         "Gefilterte Medien": "Filtered media",
 321         "Stummgeschaltete Server": "Silenced servers",
 322         "停止済みのサーバー": "Suspended servers",
 323         "メディアを拒否しているサーバー": "Filtered media",
 324         "サイレンス済みのサーバー": "Silenced servers",
 325         "שרתים מושעים": "Suspended servers",
 326         "מדיה מסוננת": "Filtered media",
 327         "שרתים מוגבלים": "Silenced servers",
 328         "Serveurs suspendus": "Suspended servers",
 329         "Médias filtrés": "Filtered media",
 330         "Serveurs limités": "Silenced servers",
 331     }
 332
 333     try:
 334         doc = BeautifulSoup(
 335             reqto.get(f"https://{domain}/about/more", headers=headers, timeout=config["timeout"]).text,
 336             "html.parser",
 337         )
 338     except:
 339         print("ERROR: Cannot fetch from domain:", domain)
 340         return {}
 341
 342     for header in doc.find_all("h3"):
 343         header_text = header.text
 344
 345         if header_text in translations:
 346             header_text = translations[header_text]
 347
 348         if header_text in blocks:
 349             # replaced find_next_siblings with find_all_next to account for instances that e.g. hide lists in dropdown menu
 350             for line in header.find_all_next("table")[0].find_all("tr")[1:]:
 351                 blocks[header_text].append(
 352                     {
 353                         "domain": line.find("span").text,
 354                         "hash": line.find("span")["title"][9:],
 355                         "reason": line.find_all("td")[1].text.strip(),
 356                     }
 357                 )
 358
 359     # NOISY-DEBUG: print("DEBUG: Returning blocks for domain:", domain)
 360     return {
 361         "reject": blocks["Suspended servers"],
 362         "media_removal": blocks["Filtered media"],
 363         "followers_only": blocks["Limited servers"] + blocks["Silenced servers"],
 364     }
 365
 366 def get_friendica_blocks(domain: str) -> dict:
 367     # NOISY-DEBUG: print("DEBUG: Fetching friendica blocks from domain:", domain)
 368     blocks = []
 369
 370     try:
 371         doc = BeautifulSoup(
 372             reqto.get(f"https://{domain}/friendica", headers=headers, timeout=config["timeout"]).text,
 373             "html.parser",
 374         )
 375     except:
 376         print("WARNING: Failed to fetch /friendica from domain:", domain)
 377         return {}
 378
 379     blocklist = doc.find(id="about_blocklist")
 380
 381     # Prevents exceptions:
 382     if blocklist is None:
 383         # NOISY-DEBUG: print("DEBUG: Instance has no block list:", domain)
 384         return {}
 385
 386     for line in blocklist.find("table").find_all("tr")[1:]:
 387         blocks.append({
 388             "domain": line.find_all("td")[0].text.strip(),
 389             "reason": line.find_all("td")[1].text.strip()
 390         })
 391
 392     # NOISY-DEBUG: print("DEBUG: Returning blocks() for domain:", domain, len(blocks))
 393     return {
 394         "reject": blocks
 395     }
 396
 397 def get_misskey_blocks(domain: str) -> dict:
 398     # NOISY-DEBUG: print("DEBUG: Fetching misskey blocks from domain:", domain)
 399     blocks = {
 400         "suspended": [],
 401         "blocked": []
 402     }
 403
 404     try:
 405         counter = 0
 406         step = 99
 407         while True:
 408             # iterating through all "suspended" (follow-only in its terminology)
 409             # instances page-by-page, since that troonware doesn't support
 410             # sending them all at once
 411             try:
 412                 if counter == 0:
 413                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 414                     doc = post_json_api(domain, "/api/federation/instances/", json.dumps({
 415                         "sort": "+caughtAt",
 416                         "host": None,
 417                         "suspended": True,
 418                         "limit": step
 419                     }))
 420                 else:
 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                         "offset": counter-1
 428                     }))
 429
 430                 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
 431                 if len(doc) == 0:
 432                     # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
 433                     break
 434
 435                 for instance in doc:
 436                     # just in case
 437                     if instance["isSuspended"]:
 438                         blocks["suspended"].append(
 439                             {
 440                                 "domain": instance["host"],
 441                                 # no reason field, nothing
 442                                 "reason": ""
 443                             }
 444                         )
 445
 446                 if len(doc) < step:
 447                     # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
 448                     break
 449
 450                 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
 451                 counter = counter + step
 452
 453             except:
 454                 print("WARNING: Caught error, exiting loop:", domain)
 455                 counter = 0
 456                 break
 457
 458         while True:
 459             # same shit, different asshole ("blocked" aka full suspend)
 460             try:
 461                 if counter == 0:
 462                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 463                     doc = post_json_api(domain,"/api/federation/instances", json.dumps({
 464                         "sort": "+caughtAt",
 465                         "host": None,
 466                         "blocked": True,
 467                         "limit": step
 468                     }))
 469                 else:
 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                         "offset": counter-1
 477                     }))
 478
 479                 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
 480                 if len(doc) == 0:
 481                     # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
 482                     break
 483
 484                 for instance in doc:
 485                     if instance["isBlocked"]:
 486                         blocks["blocked"].append({
 487                             "domain": instance["host"],
 488                             "reason": ""
 489                         })
 490
 491                 if len(doc) < step:
 492                     # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
 493                     break
 494
 495                 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
 496                 counter = counter + step
 497
 498             except:
 499                 counter = 0
 500                 break
 501
 502         # NOISY-DEBUG: print("DEBUG: Returning for domain,blocked(),suspended():", domain, len(blocks["blocked"]), len(blocks["suspended"]))
 503         return {
 504             "reject": blocks["blocked"],
 505             "followers_only": blocks["suspended"]
 506         }
 507
 508     except:
 509         print("WARNING: API request failed for domain:", domain)
 510         return {}
 511
 512 def tidyup(domain: str) -> str:
 513     # some retards put their blocks in variable case
 514     domain = domain.lower()
 515
 516     # other retards put the port
 517     domain = re.sub("\:\d+$", "", domain)
 518
 519     # bigger retards put the schema in their blocklist, sometimes even without slashes
 520     domain = re.sub("^https?\:(\/*)", "", domain)
 521
 522     # and trailing slash
 523     domain = re.sub("\/$", "", domain)
 524
 525     # and the @
 526     domain = re.sub("^\@", "", domain)
 527
 528     # the biggest retards of them all try to block individual users
 529     domain = re.sub("(.+)\@", "", domain)
 530
 531     return domain