fetch_blocks.py

   1 from reqto import get
   2 from reqto import post
   3 from hashlib import sha256
   4 import sqlite3
   5 from bs4 import BeautifulSoup
   6 from json import dumps
   7 from json import loads
   8 import re
   9 from time import time
  10
  11 with open("config.json") as f:
  12     config = loads(f.read())
  13
  14 headers = {
  15     "user-agent": config["useragent"]
  16 }
  17
  18
  19 def get_mastodon_blocks(domain: str) -> dict:
  20     blocks = {
  21         "Suspended servers": [],
  22         "Filtered media": [],
  23         "Limited servers": [],
  24         "Silenced servers": [],
  25     }
  26
  27     translations = {
  28         "Silenced instances": "Silenced servers",
  29         "Suspended instances": "Suspended servers",
  30         "Gesperrte Server": "Suspended servers",
  31         "Gefilterte Medien": "Filtered media",
  32         "Stummgeschaltete Server": "Silenced servers",
  33         "停止済みのサーバー": "Suspended servers",
  34         "メディアを拒否しているサーバー": "Filtered media",
  35         "サイレンス済みのサーバー": "Silenced servers",
  36         "שרתים מושעים": "Suspended servers",
  37         "מדיה מסוננת": "Filtered media",
  38         "שרתים מוגבלים": "Silenced servers",
  39         "Serveurs suspendus": "Suspended servers",
  40         "Médias filtrés": "Filtered media",
  41         "Serveurs limités": "Silenced servers",
  42     }
  43
  44     try:
  45         doc = BeautifulSoup(
  46             get(f"https://{domain}/about/more", headers=headers, timeout=5).text,
  47             "html.parser",
  48         )
  49     except:
  50         return {}
  51
  52     for header in doc.find_all("h3"):
  53         header_text = header.text
  54         if header_text in translations:
  55             header_text = translations[header_text]
  56         if header_text in blocks:
  57             # replaced find_next_siblings with find_all_next to account for instances that e.g. hide lists in dropdown menu
  58             for line in header.find_all_next("table")[0].find_all("tr")[1:]:
  59                 blocks[header_text].append(
  60                     {
  61                         "domain": line.find("span").text,
  62                         "hash": line.find("span")["title"][9:],
  63                         "reason": line.find_all("td")[1].text.strip(),
  64                     }
  65                 )
  66     return {
  67         "reject": blocks["Suspended servers"],
  68         "media_removal": blocks["Filtered media"],
  69         "followers_only": blocks["Limited servers"]
  70         + blocks["Silenced servers"],
  71     }
  72
  73 def get_friendica_blocks(domain: str) -> dict:
  74     blocks = []
  75
  76     try:
  77         doc = BeautifulSoup(
  78             get(f"https://{domain}/friendica", headers=headers, timeout=5).text,
  79             "html.parser",
  80         )
  81     except:
  82         return {}
  83
  84     blocklist = doc.find(id="about_blocklist")
  85     for line in blocklist.find("table").find_all("tr")[1:]:
  86             blocks.append(
  87                 {
  88                     "domain": line.find_all("td")[0].text.strip(),
  89                     "reason": line.find_all("td")[1].text.strip()
  90                 }
  91             )
  92
  93     return {
  94         "reject": blocks
  95     }
  96
  97 def get_pisskey_blocks(domain: str) -> dict:
  98     blocks = {
  99         "suspended": [],
 100         "blocked": []
 101     }
 102
 103     try:
 104         counter = 0
 105         step = 99
 106         while True:
 107             # iterating through all "suspended" (follow-only in its terminology) instances page-by-page, since that troonware doesn't support sending them all at once
 108             try:
 109                 if counter == 0:
 110                     doc = post(f"https://{domain}/api/federation/instances", data=dumps({"sort":"+caughtAt","host":None,"suspended":True,"limit":step}), headers=headers, timeout=5).json()
 111                     if doc == []: raise
 112                 else:
 113                     doc = post(f"https://{domain}/api/federation/instances", data=dumps({"sort":"+caughtAt","host":None,"suspended":True,"limit":step,"offset":counter-1}), headers=headers, timeout=5).json()
 114                     if doc == []: raise
 115                 for instance in doc:
 116                     # just in case
 117                     if instance["isSuspended"]:
 118                         blocks["suspended"].append(
 119                             {
 120                                 "domain": instance["host"],
 121                                 # no reason field, nothing
 122                                 "reason": ""
 123                             }
 124                         )
 125                 counter = counter + step
 126             except:
 127                 counter = 0
 128                 break
 129
 130         while True:
 131             # same shit, different asshole ("blocked" aka full suspend)
 132             try:
 133                 if counter == 0:
 134                     doc = post(f"https://{domain}/api/federation/instances", data=dumps({"sort":"+caughtAt","host":None,"blocked":True,"limit":step}), headers=headers, timeout=5).json()
 135                     if doc == []: raise
 136                 else:
 137                     doc = post(f"https://{domain}/api/federation/instances", data=dumps({"sort":"+caughtAt","host":None,"blocked":True,"limit":step,"offset":counter-1}), headers=headers, timeout=5).json()
 138                     if doc == []: raise
 139                 for instance in doc:
 140                     if instance["isBlocked"]:
 141                         blocks["blocked"].append(
 142                             {
 143                                 "domain": instance["host"],
 144                                 "reason": ""
 145                             }
 146                         )
 147                 counter = counter + step
 148             except:
 149                 counter = 0
 150                 break
 151
 152         return {
 153             "reject": blocks["blocked"],
 154             "followers_only": blocks["suspended"]
 155         }
 156
 157     except:
 158         return {}
 159
 160 def get_hash(domain: str) -> str:
 161     return sha256(domain.encode("utf-8")).hexdigest()
 162
 163
 164 def get_type(domain: str) -> str:
 165     try:
 166         res = get(f"https://{domain}/nodeinfo/2.1.json", headers=headers, timeout=5)
 167         if res.status_code == 404:
 168             res = get(f"https://{domain}/nodeinfo/2.0", headers=headers, timeout=5)
 169         if res.status_code == 404:
 170             res = get(f"https://{domain}/nodeinfo/2.0.json", headers=headers, timeout=5)
 171         if res.ok and "text/html" in res.headers["content-type"]:
 172             res = get(f"https://{domain}/nodeinfo/2.1", headers=headers, timeout=5)
 173         if res.ok:
 174             if res.json()["software"]["name"] in ["akkoma", "rebased"]:
 175                 return "pleroma"
 176             elif res.json()["software"]["name"] in ["hometown", "ecko"]:
 177                 return "mastodon"
 178             elif res.json()["software"]["name"] in ["calckey", "groundpolis", "foundkey", "cherrypick"]:
 179                 return "misskey"
 180             else:
 181                 return res.json()["software"]["name"]
 182         elif res.status_code == 404:
 183             res = get(f"https://{domain}/api/v1/instance", headers=headers, timeout=5)
 184         if res.ok:
 185             return "mastodon"
 186     except:
 187         return None
 188
 189 def tidyup(domain: str) -> str:
 190     # some retards put their blocks in variable case
 191     domain = domain.lower()
 192     # other retards put the port
 193     domain = re.sub("\:\d+$", "", domain)
 194     # bigger retards put the schema in their blocklist, sometimes even without slashes
 195     domain = re.sub("^https?\:(\/*)", "", domain)
 196     # and trailing slash
 197     domain = re.sub("\/$", "", domain)
 198     # and the @
 199     domain = re.sub("^\@", "", domain)
 200     # the biggest retards of them all try to block individual users
 201     domain = re.sub("(.+)\@", "", domain)
 202     return domain
 203
 204 conn = sqlite3.connect("blocks.db")
 205 c = conn.cursor()
 206
 207 c.execute(
 208     "select domain, software from instances where software in ('pleroma', 'mastodon', 'friendica', 'misskey', 'gotosocial')"
 209 )
 210
 211 for blocker, software in c.fetchall():
 212     blocker = tidyup(blocker)
 213     if software == "pleroma":
 214         print(blocker)
 215         try:
 216             # Blocks
 217             federation = get(
 218                 f"https://{blocker}/nodeinfo/2.1.json", headers=headers, timeout=5
 219             ).json()["metadata"]["federation"]
 220             if "mrf_simple" in federation:
 221                 for block_level, blocks in (
 222                     {**federation["mrf_simple"],
 223                     **{"quarantined_instances": federation["quarantined_instances"]}}
 224                 ).items():
 225                     for blocked in blocks:
 226                         blocked = tidyup(blocked)
 227                         if blocked == "":
 228                             continue
 229                         if blocked.count("*") > 1:
 230                             # -ACK!-oma also started obscuring domains without hash
 231                             c.execute(
 232                                 "select domain from instances where domain like ? order by rowid limit 1", (blocked.replace("*", "_"),)
 233                             )
 234                             searchres = c.fetchone()
 235                             if searchres != None:
 236                                 blocked = searchres[0]
 237
 238                         c.execute(
 239                             "select domain from instances where domain = ?", (blocked,)
 240                         )
 241                         if c.fetchone() == None:
 242                             c.execute(
 243                                 "insert into instances select ?, ?, ?",
 244                                 (blocked, get_hash(blocked), get_type(blocked)),
 245                             )
 246                         timestamp = int(time())
 247                         c.execute(
 248                             "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
 249                             (blocker, blocked, block_level),
 250                         )
 251                         if c.fetchone() == None:
 252                             c.execute(
 253                                 "insert into blocks select ?, ?, '', ?, ?, ?",
 254                                 (blocker, blocked, block_level, timestamp, timestamp),
 255                             )
 256                         else:
 257                             c.execute(
 258                                 "update blocks set last_seen = ? where blocker = ? and blocked = ? and block_level = ?",
 259                                 (timestamp, blocker, blocked, block_level)
 260                             )
 261             conn.commit()
 262             # Reasons
 263             if "mrf_simple_info" in federation:
 264                 for block_level, info in (
 265                     {**federation["mrf_simple_info"],
 266                     **(federation["quarantined_instances_info"]
 267                     if "quarantined_instances_info" in federation
 268                     else {})}
 269                 ).items():
 270                     for blocked, reason in info.items():
 271                         blocked = tidyup(blocked)
 272                         if blocked == "":
 273                             continue
 274                         if blocked.count("*") > 1:
 275                             # same domain guess as above, but for reasons field
 276                             c.execute(
 277                                 "select domain from instances where domain like ? order by rowid limit 1", (blocked.replace("*", "_"),)
 278                             )
 279                             searchres = c.fetchone()
 280                             if searchres != None:
 281                                 blocked = searchres[0]
 282                         c.execute(
 283                             "update blocks set reason = ? where blocker = ? and blocked = ? and block_level = ? and reason = ''",
 284                             (reason["reason"], blocker, blocked, block_level),
 285                         )
 286             conn.commit()
 287         except Exception as e:
 288             print("error:", e, blocker)
 289     elif software == "mastodon":
 290         print(blocker)
 291         try:
 292             # json endpoint for newer mastodongs
 293             try:
 294                 json = {
 295                     "reject": [],
 296                     "media_removal": [],
 297                     "followers_only": [],
 298                     "report_removal": []
 299                 }
 300
 301                 # handling CSRF, I've saw at least one server requiring it to access the endpoint
 302                 meta = BeautifulSoup(
 303                     get(f"https://{blocker}/about", headers=headers, timeout=5).text,
 304                     "html.parser",
 305                 )
 306                 try:
 307                     csrf = meta.find("meta", attrs={"name": "csrf-token"})["content"]
 308                     reqheaders = {**headers, **{"x-csrf-token": csrf}}
 309                 except:
 310                     reqheaders = headers
 311
 312                 blocks = get(
 313                     f"https://{blocker}/api/v1/instance/domain_blocks", headers=reqheaders, timeout=5
 314                 ).json()
 315                 for block in blocks:
 316                     entry = {'domain': block['domain'], 'hash': block['digest'], 'reason': block['comment']}
 317                     if block['severity'] == 'suspend':
 318                         json['reject'].append(entry)
 319                     elif block['severity'] == 'silence':
 320                         json['followers_only'].append(entry)
 321                     elif block['severity'] == 'reject_media':
 322                         json['media_removal'].append(entry)
 323                     elif block['severity'] == 'reject_reports':
 324                         json['report_removal'].append(entry)
 325             except:
 326                 json = get_mastodon_blocks(blocker)
 327
 328             for block_level, blocks in json.items():
 329                 for instance in blocks:
 330                     blocked, blocked_hash, reason = instance.values()
 331                     blocked = tidyup(blocked)
 332                     if blocked.count("*") <= 1:
 333                         c.execute(
 334                             "select hash from instances where hash = ?", (blocked_hash,)
 335                         )
 336                         if c.fetchone() == None:
 337                             c.execute(
 338                                 "insert into instances select ?, ?, ?",
 339                                 (blocked, get_hash(blocked), get_type(blocked)),
 340                             )
 341                     else:
 342                         # Doing the hash search for instance names as well to tidy up DB
 343                         c.execute(
 344                             "select domain from instances where hash = ?", (blocked_hash,)
 345                         )
 346                         searchres = c.fetchone()
 347                         if searchres != None:
 348                             blocked = searchres[0]
 349
 350                     timestamp = int(time())
 351                     c.execute(
 352                         "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
 353                         (blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level),
 354                     )
 355                     if c.fetchone() == None:
 356                         c.execute(
 357                             "insert into blocks select ?, ?, ?, ?, ?, ?",
 358                             (
 359                                 blocker,
 360                                 blocked if blocked.count("*") <= 1 else blocked_hash,
 361                                 reason,
 362                                 block_level,
 363                                 timestamp,
 364                                 timestamp,
 365                             ),
 366                         )
 367                     else:
 368                         c.execute(
 369                             "update blocks set last_seen = ? where blocker = ? and blocked = ? and block_level = ?",
 370                             (timestamp, blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level),
 371                         )
 372                     if reason != '':
 373                         c.execute(
 374                             "update blocks set reason = ? where blocker = ? and blocked = ? and block_level = ? and reason = ''",
 375                             (reason, blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level),
 376                         )
 377             conn.commit()
 378         except Exception as e:
 379             print("error:", e, blocker)
 380     elif software == "friendica" or software == "misskey":
 381         print(blocker)
 382         try:
 383             if software == "friendica":
 384                 json = get_friendica_blocks(blocker)
 385             elif software == "misskey":
 386                 json = get_pisskey_blocks(blocker)
 387             for block_level, blocks in json.items():
 388                 for instance in blocks:
 389                     blocked, reason = instance.values()
 390                     blocked = tidyup(blocked)
 391
 392                     if blocked.count("*") > 0:
 393                         # Some friendica servers also obscure domains without hash
 394                         c.execute(
 395                             "select domain from instances where domain like ? order by rowid limit 1", (blocked.replace("*", "_"),)
 396                         )
 397                         searchres = c.fetchone()
 398                         if searchres != None:
 399                             blocked = searchres[0]
 400
 401                     if blocked.count("?") > 0:
 402                         # Some obscure them with question marks, not sure if that's dependent on version or not
 403                         c.execute(
 404                             "select domain from instances where domain like ? order by rowid limit 1", (blocked.replace("?", "_"),)
 405                         )
 406                         searchres = c.fetchone()
 407                         if searchres != None:
 408                             blocked = searchres[0]
 409
 410                     c.execute(
 411                         "select domain from instances where domain = ?", (blocked,)
 412                     )
 413                     if c.fetchone() == None:
 414                         c.execute(
 415                             "insert into instances select ?, ?, ?",
 416                             (blocked, get_hash(blocked), get_type(blocked)),
 417                         )
 418
 419                     timestamp = int(time())
 420                     c.execute(
 421                         "select * from blocks where blocker = ? and blocked = ? and reason = ?",
 422                         (blocker, blocked, reason),
 423                     )
 424                     if c.fetchone() == None:
 425                         c.execute(
 426                             "insert into blocks select ?, ?, ?, ?, ?, ?",
 427                             (
 428                                 blocker,
 429                                 blocked,
 430                                 reason,
 431                                 block_level,
 432                                 timestamp,
 433                                 timestamp
 434                             ),
 435                         )
 436                     else:
 437                         c.execute(
 438                             "update blocks set last_seen = ? where blocker = ? and blocked = ? and block_level = ?",
 439                             (timestamp, blocker, blocked, block_level),
 440                         )
 441                     if reason != '':
 442                         c.execute(
 443                             "update blocks set reason = ? where blocker = ? and blocked = ? and block_level = ? and reason = ''",
 444                             (reason, blocker, blocked, block_level),
 445                         )
 446             conn.commit()
 447         except Exception as e:
 448             print("error:", e, blocker)
 449     elif software == "gotosocial":
 450         print(blocker)
 451         try:
 452             # Blocks
 453             federation = get(
 454                 f"https://{blocker}/api/v1/instance/peers?filter=suspended", headers=headers, timeout=5
 455             ).json()
 456             for peer in federation:
 457                 blocked = peer["domain"].lower()
 458
 459                 if blocked.count("*") > 0:
 460                     # GTS does not have hashes for obscured domains, so we have to guess it
 461                     c.execute(
 462                         "select domain from instances where domain like ? order by rowid limit 1", (blocked.replace("*", "_"),)
 463                     )
 464                     searchres = c.fetchone()
 465                     if searchres != None:
 466                         blocked = searchres[0]
 467
 468                 c.execute(
 469                     "select domain from instances where domain = ?", (blocked,)
 470                 )
 471                 if c.fetchone() == None:
 472                     c.execute(
 473                         "insert into instances select ?, ?, ?",
 474                         (blocked, get_hash(blocked), get_type(blocked)),
 475                     )
 476                 c.execute(
 477                     "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
 478                     (blocker, blocked, "reject"),
 479                 )
 480                 timestamp = int(time())
 481                 if c.fetchone() == None:
 482                     c.execute(
 483                         "insert into blocks select ?, ?, ?, ?, ?, ?",
 484                            (blocker, blocked, "", "reject", timestamp, timestamp),
 485                     )
 486                 else:
 487                     c.execute(
 488                         "update blocks set last_seen = ? where blocker = ? and blocked = ? and block_level = ?",
 489                         (timestamp, blocker, blocked, "reject"),
 490                     )
 491                 if "public_comment" in peer:
 492                     reason = peer["public_comment"]
 493                     c.execute(
 494                         "update blocks set reason = ? where blocker = ? and blocked = ? and block_level = ? and reason = ''",
 495                         (reason, blocker, blocked, "reject"),
 496                     )
 497             conn.commit()
 498         except Exception as e:
 499             print("error:", e, blocker)
 500 conn.close()