fetch_blocks.py

   1 from requests import get
   2 from requests import post
   3 from hashlib import sha256
   4 import sqlite3
   5 from bs4 import BeautifulSoup
   6 from json import dumps
   7 import re
   8 from time import time
   9
  10 headers = {
  11     "user-agent": "Mozilla/5.0 (Windows NT 10.0; rv:102.0) Gecko/20100101 Firefox/102.0"
  12 }
  13
  14
  15 def get_mastodon_blocks(domain: str) -> dict:
  16     blocks = {
  17         "Suspended servers": [],
  18         "Filtered media": [],
  19         "Limited servers": [],
  20         "Silenced servers": [],
  21     }
  22
  23     translations = {
  24         "Silenced instances": "Silenced servers",
  25         "Suspended instances": "Suspended servers",
  26         "Gesperrte Server": "Suspended servers",
  27         "Gefilterte Medien": "Filtered media",
  28         "Stummgeschaltete Server": "Silenced servers",
  29         "停止済みのサーバー": "Suspended servers",
  30         "メディアを拒否しているサーバー": "Filtered media",
  31         "サイレンス済みのサーバー": "Silenced servers",
  32         "שרתים מושעים": "Suspended servers",
  33         "מדיה מסוננת": "Filtered media",
  34         "שרתים מוגבלים": "Silenced servers",
  35         "Serveurs suspendus": "Suspended servers",
  36         "Médias filtrés": "Filtered media",
  37         "Serveurs limités": "Silenced servers",
  38     }
  39
  40     try:
  41         doc = BeautifulSoup(
  42             get(f"https://{domain}/about/more", headers=headers, timeout=5).text,
  43             "html.parser",
  44         )
  45     except:
  46         return {}
  47
  48     for header in doc.find_all("h3"):
  49         header_text = header.text
  50         if header_text in translations:
  51             header_text = translations[header_text]
  52         if header_text in blocks:
  53             # replaced find_next_siblings with find_all_next to account for instances that e.g. hide lists in dropdown menu
  54             for line in header.find_all_next("table")[0].find_all("tr")[1:]:
  55                 blocks[header_text].append(
  56                     {
  57                         "domain": line.find("span").text,
  58                         "hash": line.find("span")["title"][9:],
  59                         "reason": line.find_all("td")[1].text.strip(),
  60                     }
  61                 )
  62     return {
  63         "reject": blocks["Suspended servers"],
  64         "media_removal": blocks["Filtered media"],
  65         "followers_only": blocks["Limited servers"]
  66         + blocks["Silenced servers"],
  67     }
  68
  69 def get_friendica_blocks(domain: str) -> dict:
  70     blocks = []
  71
  72     try:
  73         doc = BeautifulSoup(
  74             get(f"https://{domain}/friendica", headers=headers, timeout=5).text,
  75             "html.parser",
  76         )
  77     except:
  78         return {}
  79
  80     blocklist = doc.find(id="about_blocklist")
  81     for line in blocklist.find("table").find_all("tr")[1:]:
  82             blocks.append(
  83                 {
  84                     "domain": line.find_all("td")[0].text.strip(),
  85                     "reason": line.find_all("td")[1].text.strip()
  86                 }
  87             )
  88
  89     return {
  90         "reject": blocks
  91     }
  92
  93 def get_pisskey_blocks(domain: str) -> dict:
  94     blocks = {
  95         "suspended": [],
  96         "blocked": []
  97     }
  98
  99     try:
 100         counter = 0
 101         step = 99
 102         while True:
 103             # iterating through all "suspended" (follow-only in its terminology) instances page-by-page, since that troonware doesn't support sending them all at once
 104             try:
 105                 if counter == 0:
 106                     doc = post(f"https://{domain}/api/federation/instances", data=dumps({"sort":"+caughtAt","host":None,"suspended":True,"limit":step}), headers=headers, timeout=5).json()
 107                     if doc == []: raise
 108                 else:
 109                     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()
 110                     if doc == []: raise
 111                 for instance in doc:
 112                     # just in case
 113                     if instance["isSuspended"]:
 114                         blocks["suspended"].append(
 115                             {
 116                                 "domain": instance["host"],
 117                                 # no reason field, nothing
 118                                 "reason": ""
 119                             }
 120                         )
 121                 counter = counter + step
 122             except:
 123                 counter = 0
 124                 break
 125
 126         while True:
 127             # same shit, different asshole ("blocked" aka full suspend)
 128             try:
 129                 if counter == 0:
 130                     doc = post(f"https://{domain}/api/federation/instances", data=dumps({"sort":"+caughtAt","host":None,"blocked":True,"limit":step}), headers=headers, timeout=5).json()
 131                     if doc == []: raise
 132                 else:
 133                     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()
 134                     if doc == []: raise
 135                 for instance in doc:
 136                     if instance["isBlocked"]:
 137                         blocks["blocked"].append(
 138                             {
 139                                 "domain": instance["host"],
 140                                 "reason": ""
 141                             }
 142                         )
 143                 counter = counter + step
 144             except:
 145                 counter = 0
 146                 break
 147
 148         return {
 149             "reject": blocks["blocked"],
 150             "followers_only": blocks["suspended"]
 151         }
 152
 153     except:
 154         return {}
 155
 156 def get_hash(domain: str) -> str:
 157     return sha256(domain.encode("utf-8")).hexdigest()
 158
 159
 160 def get_type(domain: str) -> str:
 161     try:
 162         res = get(f"https://{domain}/nodeinfo/2.1.json", headers=headers, timeout=5)
 163         if res.status_code == 404:
 164             res = get(f"https://{domain}/nodeinfo/2.0", headers=headers, timeout=5)
 165         if res.status_code == 404:
 166             res = get(f"https://{domain}/nodeinfo/2.0.json", headers=headers, timeout=5)
 167         if res.ok and "text/html" in res.headers["content-type"]:
 168             res = get(f"https://{domain}/nodeinfo/2.1", headers=headers, timeout=5)
 169         if res.ok:
 170             if res.json()["software"]["name"] in ["akkoma", "rebased"]:
 171                 return "pleroma"
 172             elif res.json()["software"]["name"] in ["hometown", "ecko"]:
 173                 return "mastodon"
 174             elif res.json()["software"]["name"] in ["calckey", "groundpolis", "foundkey", "cherrypick"]:
 175                 return "misskey"
 176             else:
 177                 return res.json()["software"]["name"]
 178         elif res.status_code == 404:
 179             res = get(f"https://{domain}/api/v1/instance", headers=headers, timeout=5)
 180         if res.ok:
 181             return "mastodon"
 182     except:
 183         return None
 184
 185 def tidyup(domain: str) -> str:
 186     # some retards put their blocks in variable case
 187     domain = domain.lower()
 188     # other retards put the port
 189     domain = re.sub("\:\d+$", "", domain)
 190     # bigger retards put the schema in their blocklist, sometimes even without slashes
 191     domain = re.sub("^https?\:(\/*)", "", domain)
 192     # and trailing slash
 193     domain = re.sub("\/$", "", domain)
 194     # and the @
 195     domain = re.sub("^\@", "", domain)
 196     # the biggest retards of them all try to block individual users
 197     domain = re.sub("(.+)\@", "", domain)
 198     return domain
 199
 200 conn = sqlite3.connect("blocks.db")
 201 c = conn.cursor()
 202
 203 c.execute(
 204     "select domain, software from instances where software in ('pleroma', 'mastodon', 'friendica', 'misskey', 'gotosocial')"
 205 )
 206
 207 for blocker, software in c.fetchall():
 208     blocker = tidyup(blocker)
 209     if software == "pleroma":
 210         print(blocker)
 211         try:
 212             # Blocks
 213             federation = get(
 214                 f"https://{blocker}/nodeinfo/2.1.json", headers=headers, timeout=5
 215             ).json()["metadata"]["federation"]
 216             if "mrf_simple" in federation:
 217                 for block_level, blocks in (
 218                     {**federation["mrf_simple"],
 219                     **{"quarantined_instances": federation["quarantined_instances"]}}
 220                 ).items():
 221                     for blocked in blocks:
 222                         blocked = tidyup(blocked)
 223                         if blocked == "":
 224                             continue
 225                         if blocked.count("*") > 1:
 226                             # -ACK!-oma also started obscuring domains without hash
 227                             c.execute(
 228                                 "select domain from instances where domain like ? order by rowid limit 1", (blocked.replace("*", "_"),)
 229                             )
 230                             searchres = c.fetchone()
 231                             if searchres != None:
 232                                 blocked = searchres[0]
 233
 234                         c.execute(
 235                             "select domain from instances where domain = ?", (blocked,)
 236                         )
 237                         if c.fetchone() == None:
 238                             c.execute(
 239                                 "insert into instances select ?, ?, ?",
 240                                 (blocked, get_hash(blocked), get_type(blocked)),
 241                             )
 242                         timestamp = int(time())
 243                         c.execute(
 244                             "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
 245                             (blocker, blocked, block_level),
 246                         )
 247                         if c.fetchone() == None:
 248                             c.execute(
 249                                 "insert into blocks select ?, ?, '', ?, ?, ?",
 250                                 (blocker, blocked, block_level, timestamp, timestamp),
 251                             )
 252                         else:
 253                             c.execute(
 254                                 "update blocks set last_seen = ? where blocker = ? and blocked = ? and block_level = ?",
 255                                 (timestamp, blocker, blocked, block_level)
 256                             )
 257             conn.commit()
 258             # Reasons
 259             if "mrf_simple_info" in federation:
 260                 for block_level, info in (
 261                     {**federation["mrf_simple_info"],
 262                     **(federation["quarantined_instances_info"]
 263                     if "quarantined_instances_info" in federation
 264                     else {})}
 265                 ).items():
 266                     for blocked, reason in info.items():
 267                         blocked = tidyup(blocked)
 268                         if blocked == "":
 269                             continue
 270                         if blocked.count("*") > 1:
 271                             # same domain guess as above, but for reasons field
 272                             c.execute(
 273                                 "select domain from instances where domain like ? order by rowid limit 1", (blocked.replace("*", "_"),)
 274                             )
 275                             searchres = c.fetchone()
 276                             if searchres != None:
 277                                 blocked = searchres[0]
 278                         c.execute(
 279                             "update blocks set reason = ? where blocker = ? and blocked = ? and block_level = ? and reason = ''",
 280                             (reason["reason"], blocker, blocked, block_level),
 281                         )
 282             conn.commit()
 283         except Exception as e:
 284             print("error:", e, blocker)
 285     elif software == "mastodon":
 286         print(blocker)
 287         try:
 288             # json endpoint for newer mastodongs
 289             try:
 290                 json = {
 291                     "reject": [],
 292                     "media_removal": [],
 293                     "followers_only": [],
 294                     "report_removal": []
 295                 }
 296                 blocks = get(
 297                     f"https://{blocker}/api/v1/instance/domain_blocks", headers=headers, timeout=5
 298                 ).json()
 299                 for block in blocks:
 300                     entry = {'domain': block['domain'], 'hash': block['digest'], 'reason': block['comment']}
 301                     if block['severity'] == 'suspend':
 302                         json['reject'].append(entry)
 303                     elif block['severity'] == 'silence':
 304                         json['followers_only'].append(entry)
 305                     elif block['severity'] == 'reject_media':
 306                         json['media_removal'].append(entry)
 307                     elif block['severity'] == 'reject_reports':
 308                         json['report_removal'].append(entry)
 309             except:
 310                 json = get_mastodon_blocks(blocker)
 311
 312             for block_level, blocks in json.items():
 313                 for instance in blocks:
 314                     blocked, blocked_hash, reason = instance.values()
 315                     blocked = tidyup(blocked)
 316                     if blocked.count("*") <= 1:
 317                         c.execute(
 318                             "select hash from instances where hash = ?", (blocked_hash,)
 319                         )
 320                         if c.fetchone() == None:
 321                             c.execute(
 322                                 "insert into instances select ?, ?, ?",
 323                                 (blocked, get_hash(blocked), get_type(blocked)),
 324                             )
 325                     else:
 326                         # Doing the hash search for instance names as well to tidy up DB
 327                         c.execute(
 328                             "select domain from instances where hash = ?", (blocked_hash,)
 329                         )
 330                         searchres = c.fetchone()
 331                         if searchres != None:
 332                             blocked = searchres[0]
 333
 334                     timestamp = int(time())
 335                     c.execute(
 336                         "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
 337                         (blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level),
 338                     )
 339                     if c.fetchone() == None:
 340                         c.execute(
 341                             "insert into blocks select ?, ?, ?, ?, ?, ?",
 342                             (
 343                                 blocker,
 344                                 blocked if blocked.count("*") <= 1 else blocked_hash,
 345                                 reason,
 346                                 block_level,
 347                                 timestamp,
 348                                 timestamp,
 349                             ),
 350                         )
 351                     else:
 352                         c.execute(
 353                             "update blocks set last_seen = ? where blocker = ? and blocked = ? and block_level = ?",
 354                             (timestamp, blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level),
 355                         )
 356                     if reason != '':
 357                         c.execute(
 358                             "update blocks set reason = ? where blocker = ? and blocked = ? and block_level = ? and reason = ''",
 359                             (reason, blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level),
 360                         )
 361             conn.commit()
 362         except Exception as e:
 363             print("error:", e, blocker)
 364     elif software == "friendica" or software == "misskey":
 365         print(blocker)
 366         try:
 367             if software == "friendica":
 368                 json = get_friendica_blocks(blocker)
 369             elif software == "misskey":
 370                 json = get_pisskey_blocks(blocker)
 371             for block_level, blocks in json.items():
 372                 for instance in blocks:
 373                     blocked, reason = instance.values()
 374                     blocked = tidyup(blocked)
 375
 376                     if blocked.count("*") > 0:
 377                         # Some friendica servers also obscure domains without hash
 378                         c.execute(
 379                             "select domain from instances where domain like ? order by rowid limit 1", (blocked.replace("*", "_"),)
 380                         )
 381                         searchres = c.fetchone()
 382                         if searchres != None:
 383                             blocked = searchres[0]
 384
 385                     if blocked.count("?") > 0:
 386                         # Some obscure them with question marks, not sure if that's dependent on version or not
 387                         c.execute(
 388                             "select domain from instances where domain like ? order by rowid limit 1", (blocked.replace("?", "_"),)
 389                         )
 390                         searchres = c.fetchone()
 391                         if searchres != None:
 392                             blocked = searchres[0]
 393
 394                     c.execute(
 395                         "select domain from instances where domain = ?", (blocked,)
 396                     )
 397                     if c.fetchone() == None:
 398                         c.execute(
 399                             "insert into instances select ?, ?, ?",
 400                             (blocked, get_hash(blocked), get_type(blocked)),
 401                         )
 402
 403                     timestamp = int(time())
 404                     c.execute(
 405                         "select * from blocks where blocker = ? and blocked = ? and reason = ?",
 406                         (blocker, blocked, reason),
 407                     )
 408                     if c.fetchone() == None:
 409                         c.execute(
 410                             "insert into blocks select ?, ?, ?, ?, ?, ?",
 411                             (
 412                                 blocker,
 413                                 blocked,
 414                                 reason,
 415                                 block_level,
 416                                 timestamp,
 417                                 timestamp
 418                             ),
 419                         )
 420                     else:
 421                         c.execute(
 422                             "update blocks set last_seen = ? where blocker = ? and blocked = ? and block_level = ?",
 423                             (timestamp, blocker, blocked, block_level),
 424                         )
 425                     if reason != '':
 426                         c.execute(
 427                             "update blocks set reason = ? where blocker = ? and blocked = ? and block_level = ? and reason = ''",
 428                             (reason, blocker, blocked, block_level),
 429                         )
 430             conn.commit()
 431         except Exception as e:
 432             print("error:", e, blocker)
 433     elif software == "gotosocial":
 434         print(blocker)
 435         try:
 436             # Blocks
 437             federation = get(
 438                 f"https://{blocker}/api/v1/instance/peers?filter=suspended", headers=headers, timeout=5
 439             ).json()
 440             for peer in federation:
 441                 blocked = peer["domain"].lower()
 442
 443                 if blocked.count("*") > 0:
 444                     # GTS does not have hashes for obscured domains, so we have to guess it
 445                     c.execute(
 446                         "select domain from instances where domain like ? order by rowid limit 1", (blocked.replace("*", "_"),)
 447                     )
 448                     searchres = c.fetchone()
 449                     if searchres != None:
 450                         blocked = searchres[0]
 451
 452                 c.execute(
 453                     "select domain from instances where domain = ?", (blocked,)
 454                 )
 455                 if c.fetchone() == None:
 456                     c.execute(
 457                         "insert into instances select ?, ?, ?",
 458                         (blocked, get_hash(blocked), get_type(blocked)),
 459                     )
 460                 c.execute(
 461                     "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
 462                     (blocker, blocked, "reject"),
 463                 )
 464                 timestamp = int(time())
 465                 if c.fetchone() == None:
 466                     c.execute(
 467                         "insert into blocks select ?, ?, ?, ?, ?, ?",
 468                            (blocker, blocked, "", "reject", timestamp, timestamp),
 469                     )
 470                 else:
 471                     c.execute(
 472                         "update blocks set last_seen = ? where blocker = ? and blocked = ? and block_level = ?",
 473                         (timestamp, blocker, blocked, "reject"),
 474                     )
 475                 if "public_comment" in peer:
 476                     reason = peer["public_comment"]
 477                     c.execute(
 478                         "update blocks set reason = ? where blocker = ? and blocked = ? and block_level = ? and reason = ''",
 479                         (reason, blocker, blocked, "reject"),
 480                     )
 481             conn.commit()
 482         except Exception as e:
 483             print("error:", e, blocker)
 484 conn.close()