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     # NOISY-DEBUG: print("DEBUG: json[]:", type(json))
 157     if json is None or len(json) == 0:
 158         print("WARNING: Failed fetching nodeinfo from domain:", domain)
 159         json = fetch_wellknown_nodeinfo(domain)
 160
 161     # NOISY-DEBUG: print("DEBUG: Returning json[]:", type(json))
 162     return json
 163
 164 def fetch_wellknown_nodeinfo(domain: str) -> list:
 165     # NOISY-DEBUG: print("DEBUG: Fetching .well-known info for domain:", domain)
 166     json = {}
 167
 168     try:
 169         res = reqto.get(f"https://{domain}/.well-known/nodeinfo", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
 170         # NOISY-DEBUG: print("DEBUG: domain,res.ok:", domain, res.ok)
 171         if res.ok and res.json() is not None:
 172             nodeinfo = res.json()
 173             # NOISY-DEBUG: print("DEBUG: Found entries:", len(nodeinfo), domain)
 174             if "links" in nodeinfo:
 175                 # NOISY-DEBUG: print("DEBUG: Found links in nodeinfo():", len(nodeinfo["links"]))
 176                 for link in nodeinfo["links"]:
 177                     # NOISY-DEBUG: print("DEBUG: rel,href:", link["rel"], link["href"])
 178                     if link["rel"] == "http://nodeinfo.diaspora.software/ns/schema/2.0":
 179                         # NOISY-DEBUG: print("DEBUG: Fetching nodeinfo from:", link["href"])
 180                         res = reqto.get(link["href"])
 181                         # NOISY-DEBUG: print("DEBUG: href,res.ok:", link["href"], res.ok)
 182                         if res.ok and res.json() is not None:
 183                             # NOISY-DEBUG: print("DEBUG: Found JSON nodeinfo():", len(res.json()))
 184                             json = res.json()
 185                             break
 186
 187     except:
 188         print("WARNING: Failed fetching .well-known info:", domain)
 189         pass
 190
 191     # NOISY-DEBUG: print("DEBUG: Returning json[]:", type(json))
 192     return json
 193
 194 def determine_software(domain: str) -> str:
 195     # NOISY-DEBUG: print("DEBUG: Determining software for domain:", domain)
 196     software = None
 197
 198     json = fetch_nodeinfo(domain)
 199     # NOISY-DEBUG: print("DEBUG: json[]:", type(json))
 200
 201     if json is None or len(json) == 0:
 202         # NOISY-DEBUG: print("DEBUG: Could not determine software type:", domain)
 203         return None
 204
 205     # NOISY-DEBUG: print("DEBUG: json():", len(json), json)
 206     if "software" not in json or "name" not in json["software"]:
 207         print("WARNING: JSON response does not include [software][name], guessing ...")
 208         found = 0
 209         for element in {"uri", "title", "description", "email", "version", "urls", "stats", "thumbnail", "languages", "contact_account"}:
 210             if element in json:
 211                 found = found + 1
 212
 213         # NOISY-DEBUG: print("DEBUG: Found elements:", found)
 214         if found == len(json):
 215             # NOISY-DEBUG: print("DEBUG: Maybe is Mastodon:", domain)
 216             return "mastodon"
 217
 218         print("WARNING: Cannot guess software type:", domain, found, len(json))
 219         return None
 220
 221     software = tidyup(json["software"]["name"])
 222
 223     # NOISY-DEBUG: print("DEBUG: tidyup software:", software)
 224     if software in ["akkoma", "rebased"]:
 225         # NOISY-DEBUG: print("DEBUG: Setting pleroma:", domain, software)
 226         software = "pleroma"
 227     elif software in ["hometown", "ecko"]:
 228         # NOISY-DEBUG: print("DEBUG: Setting mastodon:", domain, software)
 229         software = "mastodon"
 230     elif software in ["calckey", "groundpolis", "foundkey", "cherrypick"]:
 231         # NOISY-DEBUG: print("DEBUG: Setting misskey:", domain, software)
 232         software = "misskey"
 233     elif software.find("/") > 0:
 234         print("WARNING: Spliting of path:", software)
 235         software = software.split("/")[-1];
 236
 237     if software == "":
 238         print("WARNING: tidyup() left no software name behind:", domain)
 239         software = None
 240
 241     # NOISY-DEBUG: print("DEBUG: Returning domain,software:", domain, software)
 242     return software
 243
 244 def update_block_reason(reason: str, blocker: str, blocked: str, block_level: str):
 245     # NOISY: # NOISY-DEBUG: print("DEBUG: Updating block reason:", reason, blocker, blocked, block_level)
 246     try:
 247         cursor.execute(
 248             "UPDATE blocks SET reason = ?, last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ? AND reason = ''",
 249             (
 250                 reason,
 251                 time.time(),
 252                 blocker,
 253                 blocked,
 254                 block_level
 255             ),
 256         )
 257
 258     except:
 259         print("ERROR: failed SQL query:", reason, blocker, blocked, block_level)
 260         sys.exit(255)
 261
 262 def update_last_seen(blocker: str, blocked: str, block_level: str):
 263     # NOISY: # NOISY-DEBUG: print("DEBUG: Updating last_seen for:", blocker, blocked, block_level)
 264     try:
 265         cursor.execute(
 266             "UPDATE blocks SET last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ?",
 267             (
 268                 time.time(),
 269                 blocker,
 270                 blocked,
 271                 block_level
 272             )
 273         )
 274
 275     except:
 276         print("ERROR: failed SQL query:", last_seen, blocker, blocked, block_level)
 277         sys.exit(255)
 278
 279 def block_instance(blocker: str, blocked: str, reason: str, block_level: str):
 280     # NOISY-DEBUG: print("DEBUG: blocker,blocked,reason,block_level:", blocker, blocked, reason, block_level)
 281     if blocker.find("@") > 0:
 282         print("WARNING: Bad blocker:", blocker)
 283         raise
 284     elif blocked.find("@") > 0:
 285         print("WARNING: Bad blocked:", blocked)
 286         raise
 287
 288     print("INFO: New block:", blocker, blocked, reason, block_level, first_added, last_seen)
 289     try:
 290         cursor.execute(
 291             "INSERT INTO blocks (blocker, blocked, reason, block_level, first_added, last_seen) VALUES(?, ?, ?, ?, ?, ?)",
 292              (
 293                  blocker,
 294                  blocked,
 295                  reason,
 296                  block_level,
 297                  time.time(),
 298                  time.time()
 299              ),
 300         )
 301
 302     except:
 303         print("ERROR: failed SQL query:", blocker, blocked, reason, block_level, first_added, last_seen)
 304         sys.exit(255)
 305
 306 def add_instance(domain: str, origin: str, originator: str):
 307     # NOISY-DEBUG: print("DEBUG: domain,origin:", domain, origin, originator)
 308     if domain.find("@") > 0:
 309         print("WARNING: Bad domain name:", domain)
 310         raise
 311     elif origin is not None and origin.find("@") > 0:
 312         print("WARNING: Bad origin name:", origin)
 313         raise
 314
 315     software = determine_software(domain)
 316     # NOISY-DEBUG: print("DEBUG: Determined software:", software)
 317
 318     print(f"INFO: Adding new instance {domain} (origin: {origin})")
 319     try:
 320         cursor.execute(
 321             "INSERT INTO instances (domain, origin, originator, hash, software, first_seen) VALUES (?, ?, ?, ?, ?, ?)",
 322             (
 323                domain,
 324                origin,
 325                originator,
 326                get_hash(domain),
 327                software,
 328                time.time()
 329             ),
 330         )
 331
 332     except:
 333         print("ERROR: failed SQL query:", domain)
 334         sys.exit(255)
 335     else:
 336         # NOISY-DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain)
 337         update_last_nodeinfo(domain)
 338
 339 def send_bot_post(instance: str, blocks: dict):
 340     message = instance + " has blocked the following instances:\n\n"
 341     truncated = False
 342
 343     if len(blocks) > 20:
 344         truncated = True
 345         blocks = blocks[0 : 19]
 346
 347     for block in blocks:
 348         if block["reason"] == None or block["reason"] == '':
 349             message = message + block["blocked"] + " with unspecified reason\n"
 350         else:
 351             if len(block["reason"]) > 420:
 352                 block["reason"] = block["reason"][0:419] + "[…]"
 353
 354             message = message + block["blocked"] + ' for "' + block["reason"].replace("@", "@\u200b") + '"\n'
 355
 356     if truncated:
 357         message = message + "(the list has been truncated to the first 20 entries)"
 358
 359     botheaders = {**headers, **{"Authorization": "Bearer " + config["bot_token"]}}
 360
 361     req = reqto.post(f"{config['bot_instance']}/api/v1/statuses",
 362         data={"status":message, "visibility":config['bot_visibility'], "content_type":"text/plain"},
 363         headers=botheaders, timeout=10).json()
 364
 365     return True
 366
 367 def get_mastodon_blocks(domain: str) -> dict:
 368     # NOISY-DEBUG: print("DEBUG: Fetching mastodon blocks from domain:", domain)
 369     blocks = {
 370         "Suspended servers": [],
 371         "Filtered media"   : [],
 372         "Limited servers"  : [],
 373         "Silenced servers" : [],
 374     }
 375
 376     translations = {
 377         "Silenced instances": "Silenced servers",
 378         "Suspended instances": "Suspended servers",
 379         "Gesperrte Server": "Suspended servers",
 380         "Gefilterte Medien": "Filtered media",
 381         "Stummgeschaltete Server": "Silenced servers",
 382         "停止済みのサーバー": "Suspended servers",
 383         "メディアを拒否しているサーバー": "Filtered media",
 384         "サイレンス済みのサーバー": "Silenced servers",
 385         "שרתים מושעים": "Suspended servers",
 386         "מדיה מסוננת": "Filtered media",
 387         "שרתים מוגבלים": "Silenced servers",
 388         "Serveurs suspendus": "Suspended servers",
 389         "Médias filtrés": "Filtered media",
 390         "Serveurs limités": "Silenced servers",
 391     }
 392
 393     try:
 394         doc = BeautifulSoup(
 395             reqto.get(f"https://{domain}/about/more", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
 396             "html.parser",
 397         )
 398     except:
 399         print("ERROR: Cannot fetch from domain:", domain)
 400         return {}
 401
 402     for header in doc.find_all("h3"):
 403         header_text = header.text
 404
 405         if header_text in translations:
 406             header_text = translations[header_text]
 407
 408         if header_text in blocks:
 409             # replaced find_next_siblings with find_all_next to account for instances that e.g. hide lists in dropdown menu
 410             for line in header.find_all_next("table")[0].find_all("tr")[1:]:
 411                 blocks[header_text].append(
 412                     {
 413                         "domain": tidyup(line.find("span").text),
 414                         "hash"  : tidyup(line.find("span")["title"][9:]),
 415                         "reason": tidyup(line.find_all("td")[1].text),
 416                     }
 417                 )
 418
 419     # NOISY-DEBUG: print("DEBUG: Returning blocks for domain:", domain)
 420     return {
 421         "reject"        : blocks["Suspended servers"],
 422         "media_removal" : blocks["Filtered media"],
 423         "followers_only": blocks["Limited servers"] + blocks["Silenced servers"],
 424     }
 425
 426 def get_friendica_blocks(domain: str) -> dict:
 427     # NOISY-DEBUG: print("DEBUG: Fetching friendica blocks from domain:", domain)
 428     blocks = []
 429
 430     try:
 431         doc = BeautifulSoup(
 432             reqto.get(f"https://{domain}/friendica", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
 433             "html.parser",
 434         )
 435     except:
 436         print("WARNING: Failed to fetch /friendica from domain:", domain)
 437         return {}
 438
 439     blocklist = doc.find(id="about_blocklist")
 440
 441     # Prevents exceptions:
 442     if blocklist is None:
 443         # NOISY-DEBUG: print("DEBUG: Instance has no block list:", domain)
 444         return {}
 445
 446     for line in blocklist.find("table").find_all("tr")[1:]:
 447         blocks.append({
 448             "domain": tidyup(line.find_all("td")[0].text),
 449             "reason": tidyup(line.find_all("td")[1].text)
 450         })
 451
 452     # NOISY-DEBUG: print("DEBUG: Returning blocks() for domain:", domain, len(blocks))
 453     return {
 454         "reject": blocks
 455     }
 456
 457 def get_misskey_blocks(domain: str) -> dict:
 458     # NOISY-DEBUG: print("DEBUG: Fetching misskey blocks from domain:", domain)
 459     blocks = {
 460         "suspended": [],
 461         "blocked": []
 462     }
 463
 464     try:
 465         counter = 0
 466         step = 99
 467         while True:
 468             # iterating through all "suspended" (follow-only in its terminology)
 469             # instances page-by-page, since that troonware doesn't support
 470             # sending them all at once
 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                         "suspended": 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                         "suspended": 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                     # just in case
 497                     if instance["isSuspended"]:
 498                         blocks["suspended"].append(
 499                             {
 500                                 "domain": tidyup(instance["host"]),
 501                                 # no reason field, nothing
 502                                 "reason": ""
 503                             }
 504                         )
 505
 506                 if len(doc) < step:
 507                     # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
 508                     break
 509
 510                 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
 511                 counter = counter + step
 512
 513             except:
 514                 print("WARNING: Caught error, exiting loop:", domain)
 515                 counter = 0
 516                 break
 517
 518         while True:
 519             # same shit, different asshole ("blocked" aka full suspend)
 520             try:
 521                 if counter == 0:
 522                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 523                     doc = post_json_api(domain,"/api/federation/instances", json.dumps({
 524                         "sort"   : "+caughtAt",
 525                         "host"   : None,
 526                         "blocked": True,
 527                         "limit"  : step
 528                     }))
 529                 else:
 530                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 531                     doc = post_json_api(domain,"/api/federation/instances", json.dumps({
 532                         "sort"   : "+caughtAt",
 533                         "host"   : None,
 534                         "blocked": True,
 535                         "limit"  : step,
 536                         "offset" : counter-1
 537                     }))
 538
 539                 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
 540                 if len(doc) == 0:
 541                     # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
 542                     break
 543
 544                 for instance in doc:
 545                     if instance["isBlocked"]:
 546                         blocks["blocked"].append({
 547                             "domain": tidyup(instance["host"]),
 548                             "reason": ""
 549                         })
 550
 551                 if len(doc) < step:
 552                     # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
 553                     break
 554
 555                 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
 556                 counter = counter + step
 557
 558             except:
 559                 counter = 0
 560                 break
 561
 562         # NOISY-DEBUG: print("DEBUG: Returning for domain,blocked(),suspended():", domain, len(blocks["blocked"]), len(blocks["suspended"]))
 563         return {
 564             "reject"        : blocks["blocked"],
 565             "followers_only": blocks["suspended"]
 566         }
 567
 568     except:
 569         print("WARNING: API request failed for domain:", domain)
 570         return {}
 571
 572 def tidyup(string: str) -> str:
 573     # some retards put their blocks in variable case
 574     string = string.lower().strip()
 575
 576     # other retards put the port
 577     string = re.sub("\:\d+$", "", string)
 578
 579     # bigger retards put the schema in their blocklist, sometimes even without slashes
 580     string = re.sub("^https?\:(\/*)", "", string)
 581
 582     # and trailing slash
 583     string = re.sub("\/$", "", string)
 584
 585     # and the @
 586     string = re.sub("^\@", "", string)
 587
 588     # the biggest retards of them all try to block individual users
 589     string = re.sub("(.+)\@", "", string)
 590
 591     return string