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         "federated_timeline_removal": 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"]:
 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                         c.execute(
 257                             "update blocks set reason = ? where blocker = ? and blocked = ? and block_level = ?",
 258                             (reason["reason"], blocker, blocked, block_level),
 259                         )
 260             conn.commit()
 261         except Exception as e:
 262             print("error:", e, blocker)
 263     elif software == "mastodon":
 264         print(blocker)
 265         try:
 266             json = get_mastodon_blocks(blocker)
 267             for block_level, blocks in json.items():
 268                 for instance in blocks:
 269                     blocked, blocked_hash, reason = instance.values()
 270                     blocked = tidyup(blocked)
 271                     if blocked.count("*") <= 1:
 272                         c.execute(
 273                             "select hash from instances where hash = ?", (blocked_hash,)
 274                         )
 275                         if c.fetchone() == None:
 276                             c.execute(
 277                                 "insert into instances select ?, ?, ?",
 278                                 (blocked, get_hash(blocked), get_type(blocked)),
 279                             )
 280                     else:
 281                         # Doing the hash search for instance names as well to tidy up DB
 282                         c.execute(
 283                             "select domain from instances where hash = ?", (blocked_hash,)
 284                         )
 285                         searchres = c.fetchone()
 286                         if searchres != None:
 287                             blocked = searchres[0]
 288
 289                     c.execute(
 290                         "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
 291                         (blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level),
 292                     )
 293                     if c.fetchone() == None:
 294                         c.execute(
 295                             "insert into blocks select ?, ?, ?, ?",
 296                             (
 297                                 blocker,
 298                                 blocked if blocked.count("*") <= 1 else blocked_hash,
 299                                 reason,
 300                                 block_level,
 301                             ),
 302                         )
 303             conn.commit()
 304         except Exception as e:
 305             print("error:", e, blocker)
 306     elif software == "friendica" or software == "misskey":
 307         print(blocker)
 308         try:
 309             if software == "friendica":
 310                 json = get_friendica_blocks(blocker)
 311             elif software == "misskey":
 312                 json = get_pisskey_blocks(blocker)
 313             for block_level, blocks in json.items():
 314                 for instance in blocks:
 315                     blocked, reason = instance.values()
 316                     blocked = tidyup(blocked)
 317
 318                     if blocked.count("*") > 0:
 319                         # Some friendica servers also obscure domains without hash
 320                         c.execute(
 321                             "select domain from instances where domain like ? order by rowid limit 1", (blocked.replace("*", "_"),)
 322                         )
 323                         searchres = c.fetchone()
 324                         if searchres != None:
 325                             blocked = searchres[0]
 326
 327                     if blocked.count("?") > 0:
 328                         # Some obscure them with question marks, not sure if that's dependent on version or not
 329                         c.execute(
 330                             "select domain from instances where domain like ? order by rowid limit 1", (blocked.replace("?", "_"),)
 331                         )
 332                         searchres = c.fetchone()
 333                         if searchres != None:
 334                             blocked = searchres[0]
 335
 336                     c.execute(
 337                         "select domain from instances where domain = ?", (blocked,)
 338                     )
 339                     if c.fetchone() == None:
 340                         c.execute(
 341                             "insert into instances select ?, ?, ?",
 342                             (blocked, get_hash(blocked), get_type(blocked)),
 343                         )
 344                     c.execute(
 345                         "select * from blocks where blocker = ? and blocked = ?",
 346                         (blocker, blocked),
 347                     )
 348                     if c.fetchone() == None:
 349                         c.execute(
 350                             "insert into blocks select ?, ?, ?, ?",
 351                             (
 352                                 blocker,
 353                                 blocked,
 354                                 reason,
 355                                 block_level,
 356                             ),
 357                         )
 358             conn.commit()
 359         except Exception as e:
 360             print("error:", e, blocker)
 361     elif software == "gotosocial":
 362         print(blocker)
 363         try:
 364             # Blocks
 365             federation = get(
 366                 f"https://{blocker}/api/v1/instance/peers?filter=suspended", headers=headers, timeout=5
 367             ).json()
 368             for peer in federation:
 369                 blocked = peer["domain"].lower()
 370
 371                 if blocked.count("*") > 0:
 372                     # GTS does not have hashes for obscured domains, so we have to guess it
 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                 c.execute(
 381                     "select domain from instances where domain = ?", (blocked,)
 382                 )
 383                 if c.fetchone() == None:
 384                     c.execute(
 385                         "insert into instances select ?, ?, ?",
 386                         (blocked, get_hash(blocked), get_type(blocked)),
 387                     )
 388                 c.execute(
 389                     "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
 390                     (blocker, blocked, "reject"),
 391                 )
 392                 if c.fetchone() == None:
 393                     c.execute(
 394                         "insert into blocks select ?, ?, ?, ?",
 395                            (blocker, blocked, "", "reject"),
 396                     )
 397
 398                 if "public_comment" in peer:
 399                     reason = peer["public_comment"]
 400                     c.execute(
 401                         "select * from blocks where blocker = ? and blocked = ? and reason != ? and block_level = ?",
 402                         (blocker, blocked, "", "reject"),
 403                     )
 404                     if c.fetchone() == None:
 405                         c.execute(
 406                             "update blocks set reason = ? where blocker = ? and blocked = ? and block_level = ?",
 407                             (reason, blocker, blocked, "reject"),
 408                         )
 409             conn.commit()
 410         except Exception as e:
 411             print("error:", e, blocker)
 412 conn.close()