fetch_blocks.py

   1 import reqto
   2 import time
   3 import bs4
   4 import itertools
   5 import re
   6 import fba
   7
   8 fba.c.execute(
   9     "SELECT domain, software FROM instances WHERE software IN ('pleroma', 'mastodon', 'friendica', 'misskey', 'gotosocial') AND (last_blocked IS NULL OR last_blocked < ?) ORDER BY rowid DESC", [time.time() - fba.config["recheck_block"]]
  10 )
  11
  12 for blocker, software in fba.c.fetchall():
  13     # NOISY-DEBUG: print("DEBUG: BEFORE-blocker,software:", blocker, software)
  14     blockdict = []
  15     blocker = fba.tidyup(blocker)
  16     # NOISY-DEBUG: print("DEBUG: AFTER-blocker,software:", blocker, software)
  17
  18     if blocker == "":
  19         print("WARNING: blocker is now empty!")
  20         continue
  21
  22     fba.update_last_blocked(blocker)
  23
  24     if software == "pleroma":
  25         print("INFO: blocker:", blocker)
  26         try:
  27             # Blocks
  28             json = fba.fetch_nodeinfo(blocker)
  29             if json is None:
  30                 print("WARNING: Could not fetch nodeinfo from blocker:", blocker)
  31                 continue
  32
  33             federation = json["metadata"]["federation"]
  34
  35             if "enabled" in federation:
  36                 # NOISY-DEBUG: print("DEBUG: Instance has no block list to analyze:", blocker)
  37                 continue
  38
  39             if "mrf_simple" in federation:
  40                 for block_level, blocks in (
  41                     {**federation["mrf_simple"],
  42                     **{"quarantined_instances": federation["quarantined_instances"]}}
  43                 ).items():
  44                     # NOISY-DEBUG: print("DEBUG: block_level, blocks():", block_level, len(blocks))
  45                     block_level = fba.tidyup(block_level)
  46                     # NOISY-DEBUG: print("DEBUG: BEFORE block_level:", block_level)
  47
  48                     if block_level == "":
  49                         print("WARNING: block_level is now empty!")
  50                         continue
  51
  52                     for blocked in blocks:
  53                         # NOISY-DEBUG: print("DEBUG: BEFORE blocked:", blocked)
  54                         blocked = fba.tidyup(blocked)
  55                         # NOISY-DEBUG: print("DEBUG: AFTER blocked:", blocked)
  56
  57                         if blocked == "":
  58                             print("WARNING: blocked is empty after fba.tidyup():", blocker, block_level)
  59                             continue
  60
  61                         if blocked.count("*") > 1:
  62                             # -ACK!-oma also started obscuring domains without hash
  63                             fba.c.execute(
  64                                 "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", [blocked.replace("*", "_")]
  65                             )
  66                             searchres = fba.c.fetchone()
  67                             # NOISY-DEBUG: print("DEBUG: searchres[]:", type(searchres))
  68                             if searchres != None:
  69                                 blocked = searchres[0]
  70                                 # NOISY-DEBUG: print("DEBUG: Looked up domain:", blocked)
  71
  72                         # NOISY-DEBUG: print("DEBUG: Looking up instance by domain:", blocked)
  73                         fba.c.execute(
  74                             "SELECT domain FROM instances WHERE domain = ?", [blocked]
  75                         )
  76
  77                         if fba.c.fetchone() == None:
  78                             # NOISY-DEBUG: print("DEBUG: Domain wasn't found, adding:", blocked, blocker)
  79                             fba.add_instance(blocked, blocker, argv[0])
  80
  81                         fba.c.execute(
  82                             "SELECT * FROM blocks WHERE blocker = ? AND blocked = ? AND block_level = ?",
  83                             (
  84                                blocker,
  85                                blocked,
  86                                block_level
  87                            ),
  88                         )
  89
  90                         if fba.c.fetchone() == None:
  91                             # NOISY-DEBUG: print("DEBUG: Blocking:", blocker, blocked, block_level)
  92                             fba.block_instance(blocker, blocked, "unknown", block_level)
  93
  94                             if block_level == "reject":
  95                                 # NOISY-DEBUG: print("DEBUG: Adding to blockdict:", blocked)
  96                                 blockdict.append(
  97                                     {
  98                                         "blocked": blocked,
  99                                         "reason": None
 100                                     })
 101                         else:
 102                             # NOISY-DEBUG: print("DEBUG: Updating last_seen:", blocker, blocked, block_level)
 103                             fba.update_last_seen(blocker, blocked, block_level)
 104
 105             fba.conn.commit()
 106
 107             # Reasons
 108             if "mrf_simple_info" in federation:
 109                 # NOISY-DEBUG: print("DEBUG: Found mrf_simple_info:", blocker)
 110                 for block_level, info in (
 111                     {**federation["mrf_simple_info"],
 112                     **(federation["quarantined_instances_info"]
 113                     if "quarantined_instances_info" in federation
 114                     else {})}
 115                 ).items():
 116                     # NOISY-DEBUG: print("DEBUG: block_level, info.items():", block_level, len(info.items()))
 117                     block_level = fba.tidyup(block_level)
 118                     # NOISY-DEBUG: print("DEBUG: BEFORE block_level:", block_level)
 119
 120                     if block_level == "":
 121                         print("WARNING: block_level is now empty!")
 122                         continue
 123
 124                     for blocked, reason in info.items():
 125                         # NOISY-DEBUG: print("DEBUG: BEFORE blocked:", blocked)
 126                         blocked = fba.tidyup(blocked)
 127                         # NOISY-DEBUG: print("DEBUG: AFTER blocked:", blocked)
 128
 129                         if blocked == "":
 130                             print("WARNING: blocked is empty after fba.tidyup():", blocker, block_level)
 131                             continue
 132                         elif blocked.count("*") > 1:
 133                             # same domain guess as above, but for reasons field
 134                             fba.c.execute(
 135                                 "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", [blocked.replace("*", "_")]
 136                             )
 137                             searchres = fba.c.fetchone()
 138
 139                             if searchres != None:
 140                                 blocked = searchres[0]
 141
 142                         # NOISY-DEBUG: print("DEBUG: Updating block reason:", blocker, blocked, reason["reason"])
 143                         fba.update_block_reason(reason["reason"], blocker, blocked, block_level)
 144
 145                         for entry in blockdict:
 146                             if entry["blocked"] == blocked:
 147                                 # NOISY-DEBUG: print("DEBUG: Updating entry reason:", blocked)
 148                                 entry["reason"] = reason["reason"]
 149
 150             fba.conn.commit()
 151         except Exception as e:
 152             print("error:", e, blocker, software)
 153     elif software == "mastodon":
 154         print("INFO: blocker:", blocker)
 155         try:
 156             # json endpoint for newer mastodongs
 157             try:
 158                 json = {
 159                     "reject": [],
 160                     "media_removal": [],
 161                     "followers_only": [],
 162                     "report_removal": []
 163                 }
 164
 165                 # handling CSRF, I've saw at least one server requiring it to access the endpoint
 166                 # NOISY-DEBUG: print("DEBUG: Fetching meta:", blocker)
 167                 meta = bs4.BeautifulSoup(
 168                     reqto.get(f"https://{blocker}/about", headers=fba.headers, timeout=fba.config["timeout"]).text,
 169                     "html.parser",
 170                 )
 171                 try:
 172                     csrf = meta.find("meta", attrs={"name": "csrf-token"})["content"]
 173                     # NOISY-DEBUG: print("DEBUG: Adding CSRF token:", blocker, csrf)
 174                     reqheaders = {**fba.headers, **{"x-csrf-token": csrf}}
 175                 except:
 176                     # NOISY-DEBUG: print("DEBUG: No CSRF token found, using normal headers:", blocker)
 177                     reqheaders = fba.headers
 178
 179                 # NOISY-DEBUG: print("DEBUG: Quering API domain_blocks:", blocker)
 180                 blocks = reqto.get(f"https://{blocker}/api/v1/instance/domain_blocks", headers=reqheaders, timeout=fba.config["timeout"]).json()
 181
 182                 # NOISY-DEBUG: print("DEBUG: blocks():", len(blocks))
 183                 for block in blocks:
 184                     entry = {
 185                         'domain': block['domain'],
 186                         'hash': block['digest'],
 187                         'reason': block['comment']
 188                     }
 189
 190                     # NOISY-DEBUG: print("DEBUG: severity,domain,hash,comment:", block['severity'], block['domain'], block['digest'], block['comment'])
 191                     if block['severity'] == 'suspend':
 192                         json['reject'].append(entry)
 193                     elif block['severity'] == 'silence':
 194                         json['followers_only'].append(entry)
 195                     elif block['severity'] == 'reject_media':
 196                         json['media_removal'].append(entry)
 197                     elif block['severity'] == 'reject_reports':
 198                         json['report_removal'].append(entry)
 199                     else:
 200                         print("WARNING: Unknown severity:", block['severity'], block['domain'])
 201             except:
 202                 # NOISY-DEBUG: print("DEBUG: Failed, Trying mastodon-specific fetches:", blocker)
 203                 json = fba.get_mastodon_blocks(blocker)
 204
 205             # NOISY-DEBUG: print("DEBUG: json.items():", blocker, len(json.items()))
 206             for block_level, blocks in json.items():
 207                 # NOISY-DEBUG: print("DEBUG: blocker,block_level,blocks():", blocker, block_level, len(blocks))
 208                 block_level = fba.tidyup(block_level)
 209                 # NOISY-DEBUG: print("DEBUG: AFTER-block_level:", block_level)
 210                 if block_level == "":
 211                     print("WARNING: block_level is empty, blocker:", blocker)
 212                     continue
 213
 214                 for instance in blocks:
 215                     blocked, blocked_hash, reason = instance.values()
 216                     # NOISY-DEBUG: print("DEBUG: blocked,hash,reason:", blocked, blocked_hash, reason)
 217                     blocked = fba.tidyup(blocked)
 218                     # NOISY-DEBUG: print("DEBUG: AFTER-blocked:", blocked)
 219
 220                     if blocked == "":
 221                         print("WARNING: blocked is empty:", blocker)
 222                         continue
 223                     elif blocked.count("*") < 1:
 224                         # No obsfucation for this instance
 225                         fba.c.execute(
 226                             "SELECT hash FROM instances WHERE domain = ? LIMIT 1", [blocked]
 227                         )
 228
 229                         if fba.c.fetchone() == None:
 230                             # NOISY-DEBUG: print("DEBUG: Hash wasn't found, adding:", blocked, blocker)
 231                             fba.add_instance(blocked, blocker, argv[0])
 232                     else:
 233                         # Doing the hash search for instance names as well to tidy up DB
 234                         fba.c.execute(
 235                             "SELECT domain FROM instances WHERE hash = ? LIMIT 1", [blocked_hash]
 236                         )
 237                         searchres = fba.c.fetchone()
 238
 239                         if searchres != None:
 240                             # NOISY-DEBUG: print("DEBUG: Updating domain: ", searchres[0])
 241                             blocked = searchres[0]
 242
 243                     fba.c.execute(
 244                         "SELECT * FROM blocks WHERE blocker = ? AND blocked = ? AND block_level = ?",
 245                         (
 246                             blocker,
 247                             blocked if blocked.count("*") <= 1 else blocked_hash,
 248                             block_level
 249                         ),
 250                     )
 251
 252                     if fba.c.fetchone() == None:
 253                         fba.block_instance(blocker, blocked if blocked.count("*") <= 1 else blocked_hash, reason, block_level)
 254
 255                         if block_level == "reject":
 256                             blockdict.append(
 257                                 {
 258                                     "blocked": blocked,
 259                                     "reason": reason
 260                                 })
 261                     else:
 262                         fba.update_last_seen(blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level)
 263
 264                     if reason != '':
 265                         # NOISY-DEBUG: print("DEBUG: Updating block reason:", blocker, blocked, reason)
 266                         fba.update_block_reason(reason, blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level)
 267
 268             fba.conn.commit()
 269         except Exception as e:
 270             print("error:", e, blocker, software)
 271     elif software == "friendica" or software == "misskey":
 272         print("INFO: blocker:", blocker)
 273         try:
 274             if software == "friendica":
 275                 json = fba.get_friendica_blocks(blocker)
 276             elif software == "misskey":
 277                 json = fba.get_misskey_blocks(blocker)
 278
 279             for block_level, blocks in json.items():
 280                 # NOISY-DEBUG: print("DEBUG: blocker,block_level,blocks():", blocker, block_level, len(blocks))
 281                 block_level = fba.tidyup(block_level)
 282                 # NOISY-DEBUG: print("DEBUG: AFTER-block_level:", block_level)
 283                 if block_level == "":
 284                     print("WARNING: block_level is empty, blocker:", blocker)
 285                     continue
 286
 287                 for instance in blocks:
 288                     blocked, reason = instance.values()
 289                     # NOISY-DEBUG: print("DEBUG: BEFORE-blocked:", blocked)
 290                     blocked = fba.tidyup(blocked)
 291                     # NOISY-DEBUG: print("DEBUG: AFTER-blocked:", blocked)
 292
 293                     if blocked == "":
 294                         print("WARNING: blocked is empty:", blocker)
 295                         continue
 296                     if blocked.count("*") > 0:
 297                         # Some friendica servers also obscure domains without hash
 298                         fba.c.execute(
 299                             "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", [blocked.replace("*", "_")]
 300                         )
 301                         searchres = fba.c.fetchone()
 302                         if searchres != None:
 303                             blocked = searchres[0]
 304
 305                     if blocked.count("?") > 0:
 306                         # Some obscure them with question marks, not sure if that's dependent on version or not
 307                         fba.c.execute(
 308                             "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", [blocked.replace("?", "_")]
 309                         )
 310                         searchres = fba.c.fetchone()
 311                         if searchres != None:
 312                             blocked = searchres[0]
 313
 314                     # NOISY-DEBUG: print("DEBUG: AFTER-blocked:", blocked)
 315                     fba.c.execute(
 316                         "SELECT domain FROM instances WHERE domain = ?", [blocked]
 317                     )
 318
 319                     if fba.c.fetchone() == None:
 320                         # NOISY-DEBUG: print("DEBUG: Hash wasn't found, adding:", blocked, blocker)
 321                         fba.add_instance(blocked, blocker)
 322
 323                     fba.c.execute(
 324                         "SELECT * FROM blocks WHERE blocker = ? AND blocked = ?",
 325                         (blocker, blocked),
 326                     )
 327                     if fba.c.fetchone() == None:
 328                         fba.block_instance(blocker, blocked, reason, block_level)
 329
 330                         if block_level == "reject":
 331                             blockdict.append(
 332                                 {
 333                                     "blocked": blocked,
 334                                     "reason": reason
 335                                 })
 336                     else:
 337                         fba.update_last_seen(blocker, blocked, block_level)
 338
 339                     if reason != '':
 340                         # NOISY-DEBUG: print("DEBUG: Updating block reason:", blocker, blocked, reason)
 341                         fba.update_block_reason(reason, blocker, blocked, block_level)
 342
 343             fba.conn.commit()
 344         except Exception as e:
 345             print("error:", e, blocker, software)
 346     elif software == "gotosocial":
 347         print("INFO: blocker:", blocker)
 348         try:
 349             # Blocks
 350             federation = reqto.get(f"https://{blocker}/api/v1/instance/peers?filter=suspended", headers=fba.headers, timeout=fba.config["timeout"]).json()
 351
 352             if (federation == None):
 353                 print("WARNING: No valid response:", blocker);
 354             elif "error" in federation:
 355                 print("WARNING: API returned error:", federation["error"])
 356             else:
 357                 # NOISY-DEBUG: print("DEBUG: Checking fenderation():", len(federation))
 358                 for peer in federation:
 359                     blocked = peer["domain"].lower()
 360                     # NOISY-DEBUG: print("DEBUG: BEFORE-blocked:", blocked)
 361                     blocked = fba.tidyup(blocked)
 362                     # NOISY-DEBUG: print("DEBUG: AFTER-blocked:", blocked)
 363
 364                     if blocked == "":
 365                         print("WARNING: blocked is empty:", blocker)
 366                         continue
 367                     if blocked.count("*") > 0:
 368                         # GTS does not have hashes for obscured domains, so we have to guess it
 369                         fba.c.execute(
 370                             "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", [blocked.replace("*", "_")]
 371                         )
 372                         searchres = fba.c.fetchone()
 373
 374                         if searchres != None:
 375                             blocked = searchres[0]
 376
 377                     fba.c.execute(
 378                         "SELECT domain FROM instances WHERE domain = ?", [blocked]
 379                     )
 380
 381                     if fba.c.fetchone() == None:
 382                         # NOISY-DEBUG: print("DEBUG: Updating block reason:", blocker, blocked, reason)
 383                         fba.update_block_reason(reason, blocker, blocked, block_level)
 384
 385                     fba.c.execute(
 386                         "SELECT * FROM blocks WHERE blocker = ? AND blocked = ? AND block_level = ?",
 387                         (
 388                             blocker,
 389                             blocked,
 390                             "reject"
 391                         ),
 392                     )
 393
 394                     if fba.c.fetchone() == None:
 395                         fba.block_instance(blocker, blocked, "unknown", "reject")
 396
 397                         blockdict.append(
 398                             {
 399                                 "blocked": blocked,
 400                                 "reason": None
 401                             })
 402                     else:
 403                         fba.update_last_seen(blocker, blocked, "reject")
 404
 405                     if "public_comment" in peer:
 406                         reason = peer["public_comment"]
 407                         # NOISY-DEBUG: print("DEBUG: Updating block reason:", blocker, blocked, reason)
 408                         fba.update_block_reason(reason, blocker, blocked, "reject")
 409
 410                         for entry in blockdict:
 411                             if entry["blocked"] == blocked:
 412                                 entry["reason"] = reason
 413
 414                 fba.conn.commit()
 415         except Exception as e:
 416             print("error:", e, blocker, software)
 417     else:
 418         print("WARNING: Unknown software:", software)
 419
 420     if fba.config["bot_enabled"] and len(blockdict) > 0:
 421         send_bot_post(blocker, blockdict)
 422
 423     blockdict = []
 424
 425 fba.conn.close()