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