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, originator: str):
 237     # NOISY-DEBUG: print("DEBUG: domain,originator:", domain, originator)
 238     if domain.find("@") > 0:
 239         print("WARNING: Bad domain name:", domain)
 240         raise
 241     elif originator is not None and originator.find("@") > 0:
 242         print("WARNING: Bad originator name:", originator)
 243         raise
 244
 245     print("--- Adding new instance:", domain, originator)
 246     try:
 247         c.execute(
 248             "INSERT INTO instances (domain,originator,hash,software) VALUES (?, ?, ?, ?)",
 249             (
 250                domain,
 251                originator,
 252                get_hash(domain),
 253                determine_software(domain)
 254             ),
 255         )
 256
 257     except:
 258         print("ERROR: failed SQL query:", domain)
 259         sys.exit(255)
 260
 261 def send_bot_post(instance: str, blocks: dict):
 262     message = instance + " has blocked the following instances:\n\n"
 263     truncated = False
 264
 265     if len(blocks) > 20:
 266         truncated = True
 267         blocks = blocks[0 : 19]
 268
 269     for block in blocks:
 270         if block["reason"] == None or block["reason"] == '':
 271             message = message + block["blocked"] + " with unspecified reason\n"
 272         else:
 273             if len(block["reason"]) > 420:
 274                 block["reason"] = block["reason"][0:419] + "[…]"
 275
 276             message = message + block["blocked"] + ' for "' + block["reason"].replace("@", "@\u200b") + '"\n'
 277
 278     if truncated:
 279         message = message + "(the list has been truncated to the first 20 entries)"
 280
 281     botheaders = {**headers, **{"Authorization": "Bearer " + config["bot_token"]}}
 282
 283     req = reqto.post(f"{config['bot_instance']}/api/v1/statuses",
 284         data={"status":message, "visibility":config['bot_visibility'], "content_type":"text/plain"},
 285         headers=botheaders, timeout=10).json()
 286
 287     return True
 288
 289 def get_mastodon_blocks(domain: str) -> dict:
 290     # NOISY-DEBUG: print("DEBUG: Fetching mastodon blocks from domain:", domain)
 291     blocks = {
 292         "Suspended servers": [],
 293         "Filtered media": [],
 294         "Limited servers": [],
 295         "Silenced servers": [],
 296     }
 297
 298     translations = {
 299         "Silenced instances": "Silenced servers",
 300         "Suspended instances": "Suspended servers",
 301         "Gesperrte Server": "Suspended servers",
 302         "Gefilterte Medien": "Filtered media",
 303         "Stummgeschaltete Server": "Silenced servers",
 304         "停止済みのサーバー": "Suspended servers",
 305         "メディアを拒否しているサーバー": "Filtered media",
 306         "サイレンス済みのサーバー": "Silenced servers",
 307         "שרתים מושעים": "Suspended servers",
 308         "מדיה מסוננת": "Filtered media",
 309         "שרתים מוגבלים": "Silenced servers",
 310         "Serveurs suspendus": "Suspended servers",
 311         "Médias filtrés": "Filtered media",
 312         "Serveurs limités": "Silenced servers",
 313     }
 314
 315     try:
 316         doc = BeautifulSoup(
 317             reqto.get(f"https://{domain}/about/more", headers=headers, timeout=5).text,
 318             "html.parser",
 319         )
 320     except:
 321         print("ERROR: Cannot fetch from domain:", domain)
 322         return {}
 323
 324     for header in doc.find_all("h3"):
 325         header_text = header.text
 326
 327         if header_text in translations:
 328             header_text = translations[header_text]
 329
 330         if header_text in blocks:
 331             # replaced find_next_siblings with find_all_next to account for instances that e.g. hide lists in dropdown menu
 332             for line in header.find_all_next("table")[0].find_all("tr")[1:]:
 333                 blocks[header_text].append(
 334                     {
 335                         "domain": line.find("span").text,
 336                         "hash": line.find("span")["title"][9:],
 337                         "reason": line.find_all("td")[1].text.strip(),
 338                     }
 339                 )
 340
 341     # NOISY-DEBUG: print("DEBUG: Returning blocks for domain:", domain)
 342     return {
 343         "reject": blocks["Suspended servers"],
 344         "media_removal": blocks["Filtered media"],
 345         "followers_only": blocks["Limited servers"] + blocks["Silenced servers"],
 346     }
 347
 348 def get_friendica_blocks(domain: str) -> dict:
 349     # NOISY-DEBUG: print("DEBUG: Fetching friendica blocks from domain:", domain)
 350     blocks = []
 351
 352     try:
 353         doc = BeautifulSoup(
 354             reqto.get(f"https://{domain}/friendica", headers=headers, timeout=5).text,
 355             "html.parser",
 356         )
 357     except:
 358         print("WARNING: Failed to fetch /friendica from domain:", domain)
 359         return {}
 360
 361     blocklist = doc.find(id="about_blocklist")
 362
 363     # Prevents exceptions:
 364     if blocklist is None:
 365         # NOISY-DEBUG: print("DEBUG: Instance has no block list:", domain)
 366         return {}
 367
 368     for line in blocklist.find("table").find_all("tr")[1:]:
 369         blocks.append({
 370             "domain": line.find_all("td")[0].text.strip(),
 371             "reason": line.find_all("td")[1].text.strip()
 372         })
 373
 374     # NOISY-DEBUG: print("DEBUG: Returning blocks() for domain:", domain, len(blocks))
 375     return {
 376         "reject": blocks
 377     }
 378
 379 def get_misskey_blocks(domain: str) -> dict:
 380     # NOISY-DEBUG: print("DEBUG: Fetching misskey blocks from domain:", domain)
 381     blocks = {
 382         "suspended": [],
 383         "blocked": []
 384     }
 385
 386     try:
 387         counter = 0
 388         step = 99
 389         while True:
 390             # iterating through all "suspended" (follow-only in its terminology)
 391             # instances page-by-page, since that troonware doesn't support
 392             # sending them all at once
 393             try:
 394                 if counter == 0:
 395                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 396                     doc = post_json_api(domain, "/api/federation/instances/", json.dumps({
 397                         "sort": "+caughtAt",
 398                         "host": None,
 399                         "suspended": True,
 400                         "limit": step
 401                     }))
 402                 else:
 403                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 404                     doc = post_json_api(domain, "/api/federation/instances/", json.dumps({
 405                         "sort": "+caughtAt",
 406                         "host": None,
 407                         "suspended": True,
 408                         "limit": step,
 409                         "offset": counter-1
 410                     }))
 411
 412                 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
 413                 if len(doc) == 0:
 414                     # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
 415                     break
 416
 417                 for instance in doc:
 418                     # just in case
 419                     if instance["isSuspended"]:
 420                         blocks["suspended"].append(
 421                             {
 422                                 "domain": instance["host"],
 423                                 # no reason field, nothing
 424                                 "reason": ""
 425                             }
 426                         )
 427
 428                 if len(doc) < step:
 429                     # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
 430                     break
 431
 432                 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
 433                 counter = counter + step
 434
 435             except:
 436                 print("WARNING: Caught error, exiting loop:", domain)
 437                 counter = 0
 438                 break
 439
 440         while True:
 441             # same shit, different asshole ("blocked" aka full suspend)
 442             try:
 443                 if counter == 0:
 444                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 445                     doc = post_json_api(domain,"/api/federation/instances", json.dumps({
 446                         "sort": "+caughtAt",
 447                         "host": None,
 448                         "blocked": True,
 449                         "limit": step
 450                     }))
 451                 else:
 452                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 453                     doc = post_json_api(domain,"/api/federation/instances", json.dumps({
 454                         "sort": "+caughtAt",
 455                         "host": None,
 456                         "blocked": True,
 457                         "limit": step,
 458                         "offset": counter-1
 459                     }))
 460
 461                 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
 462                 if len(doc) == 0:
 463                     # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
 464                     break
 465
 466                 for instance in doc:
 467                     if instance["isBlocked"]:
 468                         blocks["blocked"].append({
 469                             "domain": instance["host"],
 470                             "reason": ""
 471                         })
 472
 473                 if len(doc) < step:
 474                     # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
 475                     break
 476
 477                 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
 478                 counter = counter + step
 479
 480             except:
 481                 counter = 0
 482                 break
 483
 484         # NOISY-DEBUG: print("DEBUG: Returning for domain,blocked(),suspended():", domain, len(blocks["blocked"]), len(blocks["suspended"]))
 485         return {
 486             "reject": blocks["blocked"],
 487             "followers_only": blocks["suspended"]
 488         }
 489
 490     except:
 491         print("WARNING: API request failed for domain:", domain)
 492         return {}
 493
 494 def tidyup(domain: str) -> str:
 495     # some retards put their blocks in variable case
 496     domain = domain.lower()
 497
 498     # other retards put the port
 499     domain = re.sub("\:\d+$", "", domain)
 500
 501     # bigger retards put the schema in their blocklist, sometimes even without slashes
 502     domain = re.sub("^https?\:(\/*)", "", domain)
 503
 504     # and trailing slash
 505     domain = re.sub("\/$", "", domain)
 506
 507     # and the @
 508     domain = re.sub("^\@", "", domain)
 509
 510     # the biggest retards of them all try to block individual users
 511     domain = re.sub("(.+)\@", "", domain)
 512
 513     return domain