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