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