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