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 get_hash(domain: str) -> str:
  31     # NOISY-DEBUG: print("DEBUG: Calculating hash for domain:", domain)
  32     return sha256(domain.encode("utf-8")).hexdigest()
  33
  34 def update_last_blocked(domain: str):
  35     # NOISY-DEBUG: print("DEBUG: Updating last_blocked for domain", domain)
  36
  37     try:
  38         c.execute("UPDATE instances SET last_blocked = ? WHERE domain = ?", [
  39             time.time(),
  40             domain
  41         ])
  42
  43     except:
  44         print("ERROR: failed SQL query:", domain)
  45         sys.exit(255)
  46
  47 def update_last_error(domain: str, res: any):
  48     # NOISY-DEBUG: print("DEBUG: domain,res.status_code", domain, res.status_code)
  49
  50     try:
  51         c.execute("UPDATE instances SET last_status_code = ? WHERE domain = ?", [
  52             res.status_code,
  53             domain
  54         ])
  55
  56     except:
  57         print("ERROR: failed SQL query:", domain)
  58         sys.exit(255)
  59
  60 def update_last_nodeinfo(domain: str):
  61     # NOISY-DEBUG: print("DEBUG: Updating last_nodeinfo for domain:", domain)
  62
  63     try:
  64         c.execute("UPDATE instances SET last_nodeinfo = ? WHERE domain = ?", [
  65             time.time(),
  66             domain
  67         ])
  68
  69     except:
  70         print("ERROR: failed SQL query:", domain)
  71         sys.exit(255)
  72
  73 def get_peers(domain: str) -> list:
  74     # NOISY-DEBUG: print("DEBUG: Getting peers for domain:", domain)
  75     peers = None
  76
  77     try:
  78         res = reqto.get(f"https://{domain}/api/v1/instance/peers", headers=headers, timeout=5)
  79
  80         if not res.ok or res.status_code >= 400:
  81             print("WARNING: Cannot fetch peers:", domain)
  82             update_last_error(domain, res)
  83         else:
  84             # NOISY-DEBUG: print("DEBUG: Querying API was successful:", domain, len(res.json()))
  85             peers = res.json()
  86
  87     except:
  88         print("WARNING: Some error during get():", domain)
  89
  90     update_last_nodeinfo(domain)
  91
  92     # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers))
  93     return peers
  94
  95 def post_json_api(domain: str, path: str, data: str) -> list:
  96     try:
  97         # NOISY-DEBUG: print("DEBUG: Sending POST to domain,path,data:", domain, path, data)
  98         res = reqto.post(f"https://{domain}{path}", data=data, headers=headers, timeout=5)
  99
 100         if not res.ok or res.status_code >= 400:
 101             print("WARNING: Cannot query JSON API:", domain, path, data, res.status_code)
 102             update_last_error(domain, res)
 103             raise
 104
 105         update_last_nodeinfo(domain)
 106         json = res.json()
 107     except:
 108         print("WARNING: Some error during post():", domain, path, data)
 109
 110     # NOISY-DEBUG: print("DEBUG: Returning json():", len(json))
 111     return json
 112
 113 def fetch_nodeinfo(domain: str) -> list:
 114     # NOISY-DEBUG: print("DEBUG: Fetching nodeinfo from domain:", domain)
 115
 116     requests = [
 117        f"https://{domain}/nodeinfo/2.1.json",
 118        f"https://{domain}/nodeinfo/2.0",
 119        f"https://{domain}/nodeinfo/2.0.json",
 120        f"https://{domain}/nodeinfo/2.1",
 121        f"https://{domain}/api/v1/instance"
 122     ]
 123
 124     json = None
 125     for request in requests:
 126         # NOISY-DEBUG: print("DEBUG: Fetching request:", request)
 127         res = reqto.get(request, headers=headers, timeout=5)
 128
 129         # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json()))
 130         if res.ok and res.json() is not None:
 131             # NOISY-DEBUG: print("DEBUG: Success:", request)
 132             json = res.json()
 133             break
 134         elif not res.ok or res.status_code >= 400:
 135             # NOISY-DEBUG: print("DEBUG: Failed fetching nodeinfo from domain:", domain)
 136             update_last_error(domain, res)
 137             continue
 138
 139     if json is None:
 140         print("WARNING: Failed fetching nodeinfo from domain:", domain)
 141
 142     # NOISY-DEBUG: print("DEBUG: Updating last_nodeinfo for domain:", domain)
 143     update_last_nodeinfo(domain)
 144
 145     # NOISY-DEBUG: print("DEBUG: Returning json():", len(json))
 146     return json
 147
 148 def determine_software(domain: str) -> str:
 149     # NOISY-DEBUG: print("DEBUG: Determining software for domain:", domain)
 150     software = None
 151
 152     try:
 153         json = fetch_nodeinfo(domain)
 154         # NOISY-DEBUG: print("DEBUG: json():", len(json))
 155
 156         if json["software"]["name"] in ["akkoma", "rebased"]:
 157             # NOISY-DEBUG: print("DEBUG: Setting pleroma:", domain, json["software"]["name"])
 158             software = "pleroma"
 159         elif json["software"]["name"] in ["hometown", "ecko"]:
 160             # NOISY-DEBUG: print("DEBUG: Setting mastodon:", domain, json["software"]["name"])
 161             software = "mastodon"
 162         elif json["software"]["name"] in ["calckey", "groundpolis", "foundkey", "cherrypick"]:
 163             # NOISY-DEBUG: print("DEBUG: Setting misskey:", domain, json["software"]["name"])
 164             software = "misskey"
 165         else:
 166             # NOISY-DEBUG: print("DEBUG: Using name:", domain, json["software"]["name"])
 167             software = json["software"]["name"]
 168
 169     except:
 170         print("WARNING: Could not determine software type:", domain)
 171
 172     # NOISY-DEBUG: print("DEBUG: Returning domain,software:", domain, software)
 173     return software
 174
 175 def update_block_reason(reason: str, blocker: str, blocked: str, block_level: str):
 176     # NOISY: print("--- Updating block reason:", reason, blocker, blocked, block_level)
 177     try:
 178         c.execute(
 179             "UPDATE blocks SET reason = ? WHERE blocker = ? AND blocked = ? AND block_level = ? AND reason = ''",
 180             (
 181                 reason,
 182                 blocker,
 183                 blocked,
 184                 block_level
 185             ),
 186         )
 187
 188     except:
 189         print("ERROR: failed SQL query:", reason, blocker, blocked, block_level)
 190         sys.exit(255)
 191
 192 def update_last_seen(last_seen: int, blocker: str, blocked: str, block_level: str):
 193     # NOISY: print("--- Updating last_seen:", last_seen, blocker, blocked, block_level)
 194     try:
 195         c.execute(
 196             "UPDATE blocks SET last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ?",
 197             (
 198                 last_seen,
 199                 blocker,
 200                 blocked,
 201                 block_level
 202             )
 203         )
 204
 205     except:
 206         print("ERROR: failed SQL query:", last_seen, blocker, blocked, block_level)
 207         sys.exit(255)
 208
 209 def block_instance(blocker: str, blocked: str, reason: str, block_level: str, first_added: int, last_seen: int):
 210     # NOISY-DEBUG: print("DEBUG: blocker,blocked,reason,block_level,first_added,last_seen:", blocker, blocked, reason, block_level, first_added, last_seen)
 211     if blocker.find("@") > 0:
 212         print("WARNING: Bad blocker:", blocker)
 213         raise
 214     elif blocked.find("@") > 0:
 215         print("WARNING: Bad blocked:", blocked)
 216         raise
 217
 218     print("--- New block:", blocker, blocked, reason, block_level, first_added, last_seen)
 219     try:
 220         c.execute(
 221             "INSERT INTO blocks SELECT ?, ?, ?, ?, ?, ?",
 222              (
 223                  blocker,
 224                  blocked,
 225                  reason,
 226                  block_level,
 227                  first_added,
 228                  last_seen
 229              ),
 230         )
 231
 232     except:
 233         print("ERROR: failed SQL query:", blocker, blocked, reason, block_level, first_added, last_seen)
 234         sys.exit(255)
 235
 236 def add_instance(domain: str):
 237     # NOISY-DEBUG: print("DEBUG: domain:", domain)
 238     if domain.find("@") > 0:
 239         print("WARNING: Bad domain name:", domain)
 240         raise
 241
 242     print("--- Adding new instance:", domain)
 243     try:
 244         c.execute(
 245             "INSERT INTO instances (domain,hash,software) VALUES (?, ?, ?)",
 246             (
 247                domain,
 248                get_hash(domain),
 249                determine_software(domain)
 250             ),
 251         )
 252
 253     except:
 254         print("ERROR: failed SQL query:", domain)
 255         sys.exit(255)
 256
 257 def send_bot_post(instance: str, blocks: dict):
 258     message = instance + " has blocked the following instances:\n\n"
 259     truncated = False
 260
 261     if len(blocks) > 20:
 262         truncated = True
 263         blocks = blocks[0 : 19]
 264
 265     for block in blocks:
 266         if block["reason"] == None or block["reason"] == '':
 267             message = message + block["blocked"] + " with unspecified reason\n"
 268         else:
 269             if len(block["reason"]) > 420:
 270                 block["reason"] = block["reason"][0:419] + "[…]"
 271
 272             message = message + block["blocked"] + ' for "' + block["reason"].replace("@", "@\u200b") + '"\n'
 273
 274     if truncated:
 275         message = message + "(the list has been truncated to the first 20 entries)"
 276
 277     botheaders = {**headers, **{"Authorization": "Bearer " + config["bot_token"]}}
 278
 279     req = reqto.post(f"{config['bot_instance']}/api/v1/statuses",
 280         data={"status":message, "visibility":config['bot_visibility'], "content_type":"text/plain"},
 281         headers=botheaders, timeout=10).json()
 282
 283     return True
 284
 285 def get_mastodon_blocks(domain: str) -> dict:
 286     # NOISY-DEBUG: print("DEBUG: Fetching mastodon blocks from domain:", domain)
 287     blocks = {
 288         "Suspended servers": [],
 289         "Filtered media": [],
 290         "Limited servers": [],
 291         "Silenced servers": [],
 292     }
 293
 294     translations = {
 295         "Silenced instances": "Silenced servers",
 296         "Suspended instances": "Suspended servers",
 297         "Gesperrte Server": "Suspended servers",
 298         "Gefilterte Medien": "Filtered media",
 299         "Stummgeschaltete Server": "Silenced servers",
 300         "停止済みのサーバー": "Suspended servers",
 301         "メディアを拒否しているサーバー": "Filtered media",
 302         "サイレンス済みのサーバー": "Silenced servers",
 303         "שרתים מושעים": "Suspended servers",
 304         "מדיה מסוננת": "Filtered media",
 305         "שרתים מוגבלים": "Silenced servers",
 306         "Serveurs suspendus": "Suspended servers",
 307         "Médias filtrés": "Filtered media",
 308         "Serveurs limités": "Silenced servers",
 309     }
 310
 311     try:
 312         doc = BeautifulSoup(
 313             reqto.get(f"https://{domain}/about/more", headers=headers, timeout=5).text,
 314             "html.parser",
 315         )
 316     except:
 317         print("ERROR: Cannot fetch from domain:", domain)
 318         return {}
 319
 320     for header in doc.find_all("h3"):
 321         header_text = header.text
 322
 323         if header_text in translations:
 324             header_text = translations[header_text]
 325
 326         if header_text in blocks:
 327             # replaced find_next_siblings with find_all_next to account for instances that e.g. hide lists in dropdown menu
 328             for line in header.find_all_next("table")[0].find_all("tr")[1:]:
 329                 blocks[header_text].append(
 330                     {
 331                         "domain": line.find("span").text,
 332                         "hash": line.find("span")["title"][9:],
 333                         "reason": line.find_all("td")[1].text.strip(),
 334                     }
 335                 )
 336
 337     # NOISY-DEBUG: print("DEBUG: Returning blocks for domain:", domain)
 338     return {
 339         "reject": blocks["Suspended servers"],
 340         "media_removal": blocks["Filtered media"],
 341         "followers_only": blocks["Limited servers"] + blocks["Silenced servers"],
 342     }
 343
 344 def get_friendica_blocks(domain: str) -> dict:
 345     # NOISY-DEBUG: print("DEBUG: Fetching friendica blocks from domain:", domain)
 346     blocks = []
 347
 348     try:
 349         doc = BeautifulSoup(
 350             reqto.get(f"https://{domain}/friendica", headers=headers, timeout=5).text,
 351             "html.parser",
 352         )
 353     except:
 354         print("WARNING: Failed to fetch /friendica from domain:", domain)
 355         return {}
 356
 357     blocklist = doc.find(id="about_blocklist")
 358
 359     # Prevents exceptions:
 360     if blocklist is None:
 361         # NOISY-DEBUG: print("DEBUG: Instance has no block list:", domain)
 362         return {}
 363
 364     for line in blocklist.find("table").find_all("tr")[1:]:
 365         blocks.append({
 366             "domain": line.find_all("td")[0].text.strip(),
 367             "reason": line.find_all("td")[1].text.strip()
 368         })
 369
 370     # NOISY-DEBUG: print("DEBUG: Returning blocks() for domain:", domain, len(blocks))
 371     return {
 372         "reject": blocks
 373     }
 374
 375 def get_misskey_blocks(domain: str) -> dict:
 376     # NOISY-DEBUG: print("DEBUG: Fetching misskey blocks from domain:", domain)
 377     blocks = {
 378         "suspended": [],
 379         "blocked": []
 380     }
 381
 382     try:
 383         counter = 0
 384         step = 99
 385         while True:
 386             # iterating through all "suspended" (follow-only in its terminology)
 387             # instances page-by-page, since that troonware doesn't support
 388             # sending them all at once
 389             try:
 390                 if counter == 0:
 391                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 392                     doc = post_json_api(domain, "/api/federation/instances/", json.dumps({
 393                         "sort": "+caughtAt",
 394                         "host": None,
 395                         "suspended": True,
 396                         "limit": step
 397                     }))
 398                 else:
 399                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 400                     doc = post_json_api(domain, "/api/federation/instances/", json.dumps({
 401                         "sort": "+caughtAt",
 402                         "host": None,
 403                         "suspended": True,
 404                         "limit": step,
 405                         "offset": counter-1
 406                     }))
 407
 408                 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
 409                 if len(doc) == 0:
 410                     # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
 411                     break
 412
 413                 for instance in doc:
 414                     # just in case
 415                     if instance["isSuspended"]:
 416                         blocks["suspended"].append(
 417                             {
 418                                 "domain": instance["host"],
 419                                 # no reason field, nothing
 420                                 "reason": ""
 421                             }
 422                         )
 423
 424                 if len(doc) < step:
 425                     # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
 426                     break
 427
 428                 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
 429                 counter = counter + step
 430
 431             except:
 432                 print("WARNING: Caught error, exiting loop:", domain)
 433                 counter = 0
 434                 break
 435
 436         while True:
 437             # same shit, different asshole ("blocked" aka full suspend)
 438             try:
 439                 if counter == 0:
 440                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 441                     doc = post_json_api(domain,"/api/federation/instances", json.dumps({
 442                         "sort": "+caughtAt",
 443                         "host": None,
 444                         "blocked": True,
 445                         "limit": step
 446                     }))
 447                 else:
 448                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 449                     doc = post_json_api(domain,"/api/federation/instances", json.dumps({
 450                         "sort": "+caughtAt",
 451                         "host": None,
 452                         "blocked": True,
 453                         "limit": step,
 454                         "offset": counter-1
 455                     }))
 456
 457                 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
 458                 if len(doc) == 0:
 459                     # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
 460                     break
 461
 462                 for instance in doc:
 463                     if instance["isBlocked"]:
 464                         blocks["blocked"].append({
 465                             "domain": instance["host"],
 466                             "reason": ""
 467                         })
 468
 469                 if len(doc) < step:
 470                     # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
 471                     break
 472
 473                 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
 474                 counter = counter + step
 475
 476             except:
 477                 counter = 0
 478                 break
 479
 480         # NOISY-DEBUG: print("DEBUG: Returning for domain,blocked(),suspended():", domain, len(blocks["blocked"]), len(blocks["suspended"]))
 481         return {
 482             "reject": blocks["blocked"],
 483             "followers_only": blocks["suspended"]
 484         }
 485
 486     except:
 487         print("WARNING: API request failed for domain:", domain)
 488         return {}
 489
 490 def tidyup(domain: str) -> str:
 491     # some retards put their blocks in variable case
 492     domain = domain.lower()
 493
 494     # other retards put the port
 495     domain = re.sub("\:\d+$", "", domain)
 496
 497     # bigger retards put the schema in their blocklist, sometimes even without slashes
 498     domain = re.sub("^https?\:(\/*)", "", domain)
 499
 500     # and trailing slash
 501     domain = re.sub("\/$", "", domain)
 502
 503     # and the @
 504     domain = re.sub("^\@", "", domain)
 505
 506     # the biggest retards of them all try to block individual users
 507     domain = re.sub("(.+)\@", "", domain)
 508
 509     return domain