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