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 nodeinfos = [
  24     "http://nodeinfo.diaspora.software/ns/schema/2.1",
  25     "http://nodeinfo.diaspora.software/ns/schema/2.0",
  26     "http://nodeinfo.diaspora.software/ns/schema/1.1",
  27     "http://nodeinfo.diaspora.software/ns/schema/1.0",
  28 ]
  29
  30 headers = {
  31     "user-agent": config["useragent"]
  32 }
  33
  34 connection = sqlite3.connect("blocks.db")
  35 cursor = connection.cursor()
  36
  37 def is_blacklisted(domain: str) -> bool:
  38     blacklisted = False
  39     for peer in blacklist:
  40         if peer in domain:
  41             blacklisted = True
  42
  43     return blacklisted
  44
  45 def get_hash(domain: str) -> str:
  46     return sha256(domain.encode("utf-8")).hexdigest()
  47
  48 def update_last_blocked(domain: str):
  49     # NOISY-DEBUG: print("DEBUG: Updating last_blocked for domain", domain)
  50     try:
  51         cursor.execute("UPDATE instances SET last_blocked = ?, last_updated = ? WHERE domain = ?", [
  52             time.time(),
  53             time.time(),
  54             domain
  55         ])
  56
  57     except:
  58         print("ERROR: failed SQL query:", domain)
  59         sys.exit(255)
  60
  61 def update_last_error(domain: str, res: any):
  62     # NOISY-DEBUG: print("DEBUG: domain,res.status_code:", domain, res.status_code, res.reason)
  63     try:
  64         cursor.execute("UPDATE instances SET last_status_code = ?, last_error_details = ?, last_updated = ? WHERE domain = ?", [
  65             res.status_code,
  66             res.reason,
  67             time.time(),
  68             domain
  69         ])
  70
  71     except:
  72         print("ERROR: failed SQL query:", domain)
  73         sys.exit(255)
  74
  75 def update_last_nodeinfo(domain: str):
  76     # NOISY-DEBUG: print("DEBUG: Updating last_nodeinfo for domain:", domain)
  77     try:
  78         cursor.execute("UPDATE instances SET last_nodeinfo = ?, last_updated = ? WHERE domain = ?", [
  79             time.time(),
  80             time.time(),
  81             domain
  82         ])
  83
  84     except:
  85         print("ERROR: failed SQL query:", domain)
  86         sys.exit(255)
  87
  88     connection.commit()
  89
  90 def get_peers(domain: str) -> list:
  91     # NOISY-DEBUG: print("DEBUG: Getting peers for domain:", domain)
  92     peers = None
  93
  94     try:
  95         res = reqto.get(f"https://{domain}/api/v1/instance/peers", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
  96
  97         if not res.ok or res.status_code >= 400:
  98             print("WARNING: Cannot fetch peers:", domain)
  99             update_last_error(domain, res)
 100         else:
 101             # NOISY-DEBUG: print("DEBUG: Querying API was successful:", domain, len(res.json()))
 102             peers = res.json()
 103
 104     except:
 105         print("WARNING: Some error during get():", domain)
 106
 107     update_last_nodeinfo(domain)
 108
 109     # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers))
 110     return peers
 111
 112 def post_json_api(domain: str, path: str, data: str) -> list:
 113     # NOISY-DEBUG: print("DEBUG: Sending POST to domain,path,data:", domain, path, data)
 114     json = {}
 115     try:
 116         res = reqto.post(f"https://{domain}{path}", data=data, headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
 117
 118         if not res.ok or res.status_code >= 400:
 119             print("WARNING: Cannot query JSON API:", domain, path, data, res.status_code)
 120             update_last_error(domain, res)
 121             raise
 122
 123         update_last_nodeinfo(domain)
 124         json = res.json()
 125     except:
 126         print("WARNING: Some error during post():", domain, path, data)
 127
 128     # NOISY-DEBUG: print("DEBUG: Returning json():", len(json))
 129     return json
 130
 131 def fetch_nodeinfo(domain: str) -> list:
 132     # NOISY-DEBUG: print("DEBUG: Fetching nodeinfo from domain:", domain)
 133
 134     nodeinfo = fetch_wellknown_nodeinfo(domain)
 135     # NOISY-DEBUG: print("DEBUG: nodeinfo:", len(nodeinfo))
 136
 137     if len(nodeinfo) > 0:
 138         # NOISY-DEBUG: print("DEBUG: Returning auto-discovered nodeinfo:", len(nodeinfo))
 139         return nodeinfo
 140
 141     requests = [
 142        f"https://{domain}/nodeinfo/2.1.json",
 143        f"https://{domain}/nodeinfo/2.1",
 144        f"https://{domain}/nodeinfo/2.0.json",
 145        f"https://{domain}/nodeinfo/2.0",
 146        f"https://{domain}/nodeinfo/1.0",
 147        f"https://{domain}/api/v1/instance"
 148     ]
 149
 150     json = {}
 151     for request in requests:
 152         try:
 153             # NOISY-DEBUG: print("DEBUG: Fetching request:", request)
 154             res = reqto.get(request, headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
 155
 156             # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json()))
 157             if res.ok and res.json() is not None:
 158                 # NOISY-DEBUG: print("DEBUG: Success:", request)
 159                 json = res.json()
 160                 break
 161             elif not res.ok or res.status_code >= 400:
 162                 # NOISY-DEBUG: print("DEBUG: Failed fetching nodeinfo from domain:", domain)
 163                 update_last_error(domain, res)
 164                 continue
 165
 166         except:
 167             # NOISY-DEBUG: print("DEBUG: Cannot fetch API request:", request)
 168             pass
 169
 170     # NOISY-DEBUG: print("DEBUG: json[]:", type(json))
 171     if json is None or len(json) == 0:
 172         print("WARNING: Failed fetching nodeinfo from domain:", domain)
 173
 174     # NOISY-DEBUG: print("DEBUG: Returning json[]:", type(json))
 175     return json
 176
 177 def fetch_wellknown_nodeinfo(domain: str) -> list:
 178     # NOISY-DEBUG: print("DEBUG: Fetching .well-known info for domain:", domain)
 179     json = {}
 180
 181     try:
 182         res = reqto.get(f"https://{domain}/.well-known/nodeinfo", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
 183         # NOISY-DEBUG: print("DEBUG: domain,res.ok:", domain, res.ok)
 184         if res.ok and res.json() is not None:
 185             nodeinfo = res.json()
 186             # NOISY-DEBUG: print("DEBUG: Found entries:", len(nodeinfo), domain)
 187             if "links" in nodeinfo:
 188                 # NOISY-DEBUG: print("DEBUG: Found links in nodeinfo():", len(nodeinfo["links"]))
 189                 for link in nodeinfo["links"]:
 190                     # NOISY-DEBUG: print("DEBUG: rel,href:", link["rel"], link["href"])
 191                     if link["rel"] in nodeinfos:
 192                         # NOISY-DEBUG: print("DEBUG: Fetching nodeinfo from:", link["href"])
 193                         res = reqto.get(link["href"])
 194                         # NOISY-DEBUG: print("DEBUG: href,res.ok,res.status_code:", link["href"], res.ok, res.status_code)
 195                         if res.ok and res.json() is not None:
 196                             # NOISY-DEBUG: print("DEBUG: Found JSON nodeinfo():", len(res.json()))
 197                             json = res.json()
 198                             break
 199                     else:
 200                         print("WARNING: Unknown 'rel' value:", domain, link["rel"])
 201             else:
 202                 print("WARNING: nodeinfo does not contain 'links':", domain)
 203
 204     except:
 205         print("WARNING: Failed fetching .well-known info:", domain)
 206         pass
 207
 208     # NOISY-DEBUG: print("DEBUG: Returning json[]:", type(json))
 209     return json
 210
 211 def determine_software(domain: str) -> str:
 212     # NOISY-DEBUG: print("DEBUG: Determining software for domain:", domain)
 213     software = None
 214
 215     json = fetch_nodeinfo(domain)
 216     # NOISY-DEBUG: print("DEBUG: json[]:", type(json))
 217
 218     if json is None or len(json) == 0:
 219         # NOISY-DEBUG: print("DEBUG: Could not determine software type:", domain)
 220         return None
 221
 222     # NOISY-DEBUG: print("DEBUG: json():", len(json), json)
 223     if "software" not in json or "name" not in json["software"]:
 224         print("WARNING: JSON response does not include [software][name], guessing ...")
 225         found = 0
 226         for element in {"uri", "title", "description", "email", "version", "urls", "stats", "thumbnail", "languages", "contact_account"}:
 227             if element in json:
 228                 found = found + 1
 229
 230         # NOISY-DEBUG: print("DEBUG: Found elements:", found)
 231         if found == len(json):
 232             # NOISY-DEBUG: print("DEBUG: Maybe is Mastodon:", domain)
 233             return "mastodon"
 234
 235         print("WARNING: Cannot guess software type:", domain, found, len(json))
 236         return None
 237
 238     software = tidyup(json["software"]["name"])
 239
 240     # NOISY-DEBUG: print("DEBUG: tidyup software:", software)
 241     if software in ["akkoma", "rebased"]:
 242         # NOISY-DEBUG: print("DEBUG: Setting pleroma:", domain, software)
 243         software = "pleroma"
 244     elif software in ["hometown", "ecko"]:
 245         # NOISY-DEBUG: print("DEBUG: Setting mastodon:", domain, software)
 246         software = "mastodon"
 247     elif software in ["calckey", "groundpolis", "foundkey", "cherrypick"]:
 248         # NOISY-DEBUG: print("DEBUG: Setting misskey:", domain, software)
 249         software = "misskey"
 250     elif software.find("/") > 0:
 251         print("WARNING: Spliting of path:", software)
 252         software = software.split("/")[-1];
 253     elif software.find("|") > 0:
 254         print("WARNING: Spliting of path:", software)
 255         software = software.split("|")[0].strip();
 256
 257     if software == "":
 258         print("WARNING: tidyup() left no software name behind:", domain)
 259         software = None
 260
 261     # NOISY-DEBUG: print("DEBUG: Returning domain,software:", domain, software)
 262     return software
 263
 264 def update_block_reason(reason: str, blocker: str, blocked: str, block_level: str):
 265     # NOISY: # NOISY-DEBUG: print("DEBUG: Updating block reason:", reason, blocker, blocked, block_level)
 266     try:
 267         cursor.execute(
 268             "UPDATE blocks SET reason = ?, last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ? AND reason = ''",
 269             (
 270                 reason,
 271                 time.time(),
 272                 blocker,
 273                 blocked,
 274                 block_level
 275             ),
 276         )
 277
 278     except:
 279         print("ERROR: failed SQL query:", reason, blocker, blocked, block_level)
 280         sys.exit(255)
 281
 282 def update_last_seen(blocker: str, blocked: str, block_level: str):
 283     # NOISY: # NOISY-DEBUG: print("DEBUG: Updating last_seen for:", blocker, blocked, block_level)
 284     try:
 285         cursor.execute(
 286             "UPDATE blocks SET last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ?",
 287             (
 288                 time.time(),
 289                 blocker,
 290                 blocked,
 291                 block_level
 292             )
 293         )
 294
 295     except:
 296         print("ERROR: failed SQL query:", last_seen, blocker, blocked, block_level)
 297         sys.exit(255)
 298
 299 def block_instance(blocker: str, blocked: str, reason: str, block_level: str):
 300     # NOISY-DEBUG: print("DEBUG: blocker,blocked,reason,block_level:", blocker, blocked, reason, block_level)
 301     if blocker.find("@") > 0:
 302         print("WARNING: Bad blocker:", blocker)
 303         raise
 304     elif blocked.find("@") > 0:
 305         print("WARNING: Bad blocked:", blocked)
 306         raise
 307
 308     print("INFO: New block:", blocker, blocked, reason, block_level, first_added, last_seen)
 309     try:
 310         cursor.execute(
 311             "INSERT INTO blocks (blocker, blocked, reason, block_level, first_added, last_seen) VALUES(?, ?, ?, ?, ?, ?)",
 312              (
 313                  blocker,
 314                  blocked,
 315                  reason,
 316                  block_level,
 317                  time.time(),
 318                  time.time()
 319              ),
 320         )
 321
 322     except:
 323         print("ERROR: failed SQL query:", blocker, blocked, reason, block_level, first_added, last_seen)
 324         sys.exit(255)
 325
 326 def add_instance(domain: str, origin: str, originator: str):
 327     # NOISY-DEBUG: print("DEBUG: domain,origin:", domain, origin, originator)
 328     if domain.find("@") > 0:
 329         print("WARNING: Bad domain name:", domain)
 330         raise
 331     elif origin is not None and origin.find("@") > 0:
 332         print("WARNING: Bad origin name:", origin)
 333         raise
 334
 335     software = determine_software(domain)
 336     # NOISY-DEBUG: print("DEBUG: Determined software:", software)
 337
 338     print(f"INFO: Adding new instance {domain} (origin: {origin})")
 339     try:
 340         cursor.execute(
 341             "INSERT INTO instances (domain, origin, originator, hash, software, first_seen) VALUES (?, ?, ?, ?, ?, ?)",
 342             (
 343                domain,
 344                origin,
 345                originator,
 346                get_hash(domain),
 347                software,
 348                time.time()
 349             ),
 350         )
 351
 352     except:
 353         print("ERROR: failed SQL query:", domain)
 354         sys.exit(255)
 355     else:
 356         # NOISY-DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain)
 357         update_last_nodeinfo(domain)
 358
 359 def send_bot_post(instance: str, blocks: dict):
 360     message = instance + " has blocked the following instances:\n\n"
 361     truncated = False
 362
 363     if len(blocks) > 20:
 364         truncated = True
 365         blocks = blocks[0 : 19]
 366
 367     for block in blocks:
 368         if block["reason"] == None or block["reason"] == '':
 369             message = message + block["blocked"] + " with unspecified reason\n"
 370         else:
 371             if len(block["reason"]) > 420:
 372                 block["reason"] = block["reason"][0:419] + "[…]"
 373
 374             message = message + block["blocked"] + ' for "' + block["reason"].replace("@", "@\u200b") + '"\n'
 375
 376     if truncated:
 377         message = message + "(the list has been truncated to the first 20 entries)"
 378
 379     botheaders = {**headers, **{"Authorization": "Bearer " + config["bot_token"]}}
 380
 381     req = reqto.post(f"{config['bot_instance']}/api/v1/statuses",
 382         data={"status":message, "visibility":config['bot_visibility'], "content_type":"text/plain"},
 383         headers=botheaders, timeout=10).json()
 384
 385     return True
 386
 387 def get_mastodon_blocks(domain: str) -> dict:
 388     # NOISY-DEBUG: print("DEBUG: Fetching mastodon blocks from domain:", domain)
 389     blocks = {
 390         "Suspended servers": [],
 391         "Filtered media"   : [],
 392         "Limited servers"  : [],
 393         "Silenced servers" : [],
 394     }
 395
 396     translations = {
 397         "Silenced instances": "Silenced servers",
 398         "Suspended instances": "Suspended servers",
 399         "Gesperrte Server": "Suspended servers",
 400         "Gefilterte Medien": "Filtered media",
 401         "Stummgeschaltete Server": "Silenced servers",
 402         "停止済みのサーバー": "Suspended servers",
 403         "メディアを拒否しているサーバー": "Filtered media",
 404         "サイレンス済みのサーバー": "Silenced servers",
 405         "שרתים מושעים": "Suspended servers",
 406         "מדיה מסוננת": "Filtered media",
 407         "שרתים מוגבלים": "Silenced servers",
 408         "Serveurs suspendus": "Suspended servers",
 409         "Médias filtrés": "Filtered media",
 410         "Serveurs limités": "Silenced servers",
 411     }
 412
 413     try:
 414         doc = BeautifulSoup(
 415             reqto.get(f"https://{domain}/about/more", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
 416             "html.parser",
 417         )
 418     except:
 419         print("ERROR: Cannot fetch from domain:", domain)
 420         return {}
 421
 422     for header in doc.find_all("h3"):
 423         header_text = header.text
 424
 425         if header_text in translations:
 426             header_text = translations[header_text]
 427
 428         if header_text in blocks:
 429             # replaced find_next_siblings with find_all_next to account for instances that e.g. hide lists in dropdown menu
 430             for line in header.find_all_next("table")[0].find_all("tr")[1:]:
 431                 blocks[header_text].append(
 432                     {
 433                         "domain": tidyup(line.find("span").text),
 434                         "hash"  : tidyup(line.find("span")["title"][9:]),
 435                         "reason": tidyup(line.find_all("td")[1].text),
 436                     }
 437                 )
 438
 439     # NOISY-DEBUG: print("DEBUG: Returning blocks for domain:", domain)
 440     return {
 441         "reject"        : blocks["Suspended servers"],
 442         "media_removal" : blocks["Filtered media"],
 443         "followers_only": blocks["Limited servers"] + blocks["Silenced servers"],
 444     }
 445
 446 def get_friendica_blocks(domain: str) -> dict:
 447     # NOISY-DEBUG: print("DEBUG: Fetching friendica blocks from domain:", domain)
 448     blocks = []
 449
 450     try:
 451         doc = BeautifulSoup(
 452             reqto.get(f"https://{domain}/friendica", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
 453             "html.parser",
 454         )
 455     except:
 456         print("WARNING: Failed to fetch /friendica from domain:", domain)
 457         return {}
 458
 459     blocklist = doc.find(id="about_blocklist")
 460
 461     # Prevents exceptions:
 462     if blocklist is None:
 463         # NOISY-DEBUG: print("DEBUG: Instance has no block list:", domain)
 464         return {}
 465
 466     for line in blocklist.find("table").find_all("tr")[1:]:
 467         blocks.append({
 468             "domain": tidyup(line.find_all("td")[0].text),
 469             "reason": tidyup(line.find_all("td")[1].text)
 470         })
 471
 472     # NOISY-DEBUG: print("DEBUG: Returning blocks() for domain:", domain, len(blocks))
 473     return {
 474         "reject": blocks
 475     }
 476
 477 def get_misskey_blocks(domain: str) -> dict:
 478     # NOISY-DEBUG: print("DEBUG: Fetching misskey blocks from domain:", domain)
 479     blocks = {
 480         "suspended": [],
 481         "blocked": []
 482     }
 483
 484     try:
 485         counter = 0
 486         step = 99
 487         while True:
 488             # iterating through all "suspended" (follow-only in its terminology)
 489             # instances page-by-page, since that troonware doesn't support
 490             # sending them all at once
 491             try:
 492                 if counter == 0:
 493                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 494                     doc = post_json_api(domain, "/api/federation/instances/", json.dumps({
 495                         "sort"     : "+caughtAt",
 496                         "host"     : None,
 497                         "suspended": True,
 498                         "limit"    : step
 499                     }))
 500                 else:
 501                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 502                     doc = post_json_api(domain, "/api/federation/instances/", json.dumps({
 503                         "sort"     : "+caughtAt",
 504                         "host"     : None,
 505                         "suspended": True,
 506                         "limit"    : step,
 507                         "offset"   : counter-1
 508                     }))
 509
 510                 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
 511                 if len(doc) == 0:
 512                     # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
 513                     break
 514
 515                 for instance in doc:
 516                     # just in case
 517                     if instance["isSuspended"]:
 518                         blocks["suspended"].append(
 519                             {
 520                                 "domain": tidyup(instance["host"]),
 521                                 # no reason field, nothing
 522                                 "reason": ""
 523                             }
 524                         )
 525
 526                 if len(doc) < step:
 527                     # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
 528                     break
 529
 530                 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
 531                 counter = counter + step
 532
 533             except:
 534                 print("WARNING: Caught error, exiting loop:", domain)
 535                 counter = 0
 536                 break
 537
 538         while True:
 539             # same shit, different asshole ("blocked" aka full suspend)
 540             try:
 541                 if counter == 0:
 542                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 543                     doc = post_json_api(domain,"/api/federation/instances", json.dumps({
 544                         "sort"   : "+caughtAt",
 545                         "host"   : None,
 546                         "blocked": True,
 547                         "limit"  : step
 548                     }))
 549                 else:
 550                     # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 551                     doc = post_json_api(domain,"/api/federation/instances", json.dumps({
 552                         "sort"   : "+caughtAt",
 553                         "host"   : None,
 554                         "blocked": True,
 555                         "limit"  : step,
 556                         "offset" : counter-1
 557                     }))
 558
 559                 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
 560                 if len(doc) == 0:
 561                     # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
 562                     break
 563
 564                 for instance in doc:
 565                     if instance["isBlocked"]:
 566                         blocks["blocked"].append({
 567                             "domain": tidyup(instance["host"]),
 568                             "reason": ""
 569                         })
 570
 571                 if len(doc) < step:
 572                     # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
 573                     break
 574
 575                 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
 576                 counter = counter + step
 577
 578             except:
 579                 counter = 0
 580                 break
 581
 582         # NOISY-DEBUG: print("DEBUG: Returning for domain,blocked(),suspended():", domain, len(blocks["blocked"]), len(blocks["suspended"]))
 583         return {
 584             "reject"        : blocks["blocked"],
 585             "followers_only": blocks["suspended"]
 586         }
 587
 588     except:
 589         print("WARNING: API request failed for domain:", domain)
 590         return {}
 591
 592 def tidyup(string: str) -> str:
 593     # some retards put their blocks in variable case
 594     string = string.lower().strip()
 595
 596     # other retards put the port
 597     string = re.sub("\:\d+$", "", string)
 598
 599     # bigger retards put the schema in their blocklist, sometimes even without slashes
 600     string = re.sub("^https?\:(\/*)", "", string)
 601
 602     # and trailing slash
 603     string = re.sub("\/$", "", string)
 604
 605     # and the @
 606     string = re.sub("^\@", "", string)
 607
 608     # the biggest retards of them all try to block individual users
 609     string = re.sub("(.+)\@", "", string)
 610
 611     return string