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 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                 blocks = get(
 301                     f"https://{blocker}/api/v1/instance/domain_blocks", headers=headers, timeout=5
 302                 ).json()
 303                 for block in blocks:
 304                     entry = {'domain': block['domain'], 'hash': block['digest'], 'reason': block['comment']}
 305                     if block['severity'] == 'suspend':
 306                         json['reject'].append(entry)
 307                     elif block['severity'] == 'silence':
 308                         json['followers_only'].append(entry)
 309                     elif block['severity'] == 'reject_media':
 310                         json['media_removal'].append(entry)
 311                     elif block['severity'] == 'reject_reports':
 312                         json['report_removal'].append(entry)
 313             except:
 314                 json = get_mastodon_blocks(blocker)
 315
 316             for block_level, blocks in json.items():
 317                 for instance in blocks:
 318                     blocked, blocked_hash, reason = instance.values()
 319                     blocked = tidyup(blocked)
 320                     if blocked.count("*") <= 1:
 321                         c.execute(
 322                             "select hash from instances where hash = ?", (blocked_hash,)
 323                         )
 324                         if c.fetchone() == None:
 325                             c.execute(
 326                                 "insert into instances select ?, ?, ?",
 327                                 (blocked, get_hash(blocked), get_type(blocked)),
 328                             )
 329                     else:
 330                         # Doing the hash search for instance names as well to tidy up DB
 331                         c.execute(
 332                             "select domain from instances where hash = ?", (blocked_hash,)
 333                         )
 334                         searchres = c.fetchone()
 335                         if searchres != None:
 336                             blocked = searchres[0]
 337
 338                     timestamp = int(time())
 339                     c.execute(
 340                         "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
 341                         (blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level),
 342                     )
 343                     if c.fetchone() == None:
 344                         c.execute(
 345                             "insert into blocks select ?, ?, ?, ?, ?, ?",
 346                             (
 347                                 blocker,
 348                                 blocked if blocked.count("*") <= 1 else blocked_hash,
 349                                 reason,
 350                                 block_level,
 351                                 timestamp,
 352                                 timestamp,
 353                             ),
 354                         )
 355                     else:
 356                         c.execute(
 357                             "update blocks set last_seen = ? where blocker = ? and blocked = ? and block_level = ?",
 358                             (timestamp, blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level),
 359                         )
 360                     if reason != '':
 361                         c.execute(
 362                             "update blocks set reason = ? where blocker = ? and blocked = ? and block_level = ? and reason = ''",
 363                             (reason, blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level),
 364                         )
 365             conn.commit()
 366         except Exception as e:
 367             print("error:", e, blocker)
 368     elif software == "friendica" or software == "misskey":
 369         print(blocker)
 370         try:
 371             if software == "friendica":
 372                 json = get_friendica_blocks(blocker)
 373             elif software == "misskey":
 374                 json = get_pisskey_blocks(blocker)
 375             for block_level, blocks in json.items():
 376                 for instance in blocks:
 377                     blocked, reason = instance.values()
 378                     blocked = tidyup(blocked)
 379
 380                     if blocked.count("*") > 0:
 381                         # Some friendica servers also obscure domains without hash
 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                     if blocked.count("?") > 0:
 390                         # Some obscure them with question marks, not sure if that's dependent on version or not
 391                         c.execute(
 392                             "select domain from instances where domain like ? order by rowid limit 1", (blocked.replace("?", "_"),)
 393                         )
 394                         searchres = c.fetchone()
 395                         if searchres != None:
 396                             blocked = searchres[0]
 397
 398                     c.execute(
 399                         "select domain from instances where domain = ?", (blocked,)
 400                     )
 401                     if c.fetchone() == None:
 402                         c.execute(
 403                             "insert into instances select ?, ?, ?",
 404                             (blocked, get_hash(blocked), get_type(blocked)),
 405                         )
 406
 407                     timestamp = int(time())
 408                     c.execute(
 409                         "select * from blocks where blocker = ? and blocked = ? and reason = ?",
 410                         (blocker, blocked, reason),
 411                     )
 412                     if c.fetchone() == None:
 413                         c.execute(
 414                             "insert into blocks select ?, ?, ?, ?, ?, ?",
 415                             (
 416                                 blocker,
 417                                 blocked,
 418                                 reason,
 419                                 block_level,
 420                                 timestamp,
 421                                 timestamp
 422                             ),
 423                         )
 424                     else:
 425                         c.execute(
 426                             "update blocks set last_seen = ? where blocker = ? and blocked = ? and block_level = ?",
 427                             (timestamp, blocker, blocked, block_level),
 428                         )
 429                     if reason != '':
 430                         c.execute(
 431                             "update blocks set reason = ? where blocker = ? and blocked = ? and block_level = ? and reason = ''",
 432                             (reason, blocker, blocked, block_level),
 433                         )
 434             conn.commit()
 435         except Exception as e:
 436             print("error:", e, blocker)
 437     elif software == "gotosocial":
 438         print(blocker)
 439         try:
 440             # Blocks
 441             federation = get(
 442                 f"https://{blocker}/api/v1/instance/peers?filter=suspended", headers=headers, timeout=5
 443             ).json()
 444             for peer in federation:
 445                 blocked = peer["domain"].lower()
 446
 447                 if blocked.count("*") > 0:
 448                     # GTS does not have hashes for obscured domains, so we have to guess it
 449                     c.execute(
 450                         "select domain from instances where domain like ? order by rowid limit 1", (blocked.replace("*", "_"),)
 451                     )
 452                     searchres = c.fetchone()
 453                     if searchres != None:
 454                         blocked = searchres[0]
 455
 456                 c.execute(
 457                     "select domain from instances where domain = ?", (blocked,)
 458                 )
 459                 if c.fetchone() == None:
 460                     c.execute(
 461                         "insert into instances select ?, ?, ?",
 462                         (blocked, get_hash(blocked), get_type(blocked)),
 463                     )
 464                 c.execute(
 465                     "select * from blocks where blocker = ? and blocked = ? and block_level = ?",
 466                     (blocker, blocked, "reject"),
 467                 )
 468                 timestamp = int(time())
 469                 if c.fetchone() == None:
 470                     c.execute(
 471                         "insert into blocks select ?, ?, ?, ?, ?, ?",
 472                            (blocker, blocked, "", "reject", timestamp, timestamp),
 473                     )
 474                 else:
 475                     c.execute(
 476                         "update blocks set last_seen = ? where blocker = ? and blocked = ? and block_level = ?",
 477                         (timestamp, blocker, blocked, "reject"),
 478                     )
 479                 if "public_comment" in peer:
 480                     reason = peer["public_comment"]
 481                     c.execute(
 482                         "update blocks set reason = ? where blocker = ? and blocked = ? and block_level = ? and reason = ''",
 483                         (reason, blocker, blocked, "reject"),
 484                     )
 485             conn.commit()
 486         except Exception as e:
 487             print("error:", e, blocker)
 488 conn.close()