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                         timestamp = int(time.time())
  82                         fba.c.execute(
  83                             "SELECT * FROM blocks WHERE blocker = ? AND blocked = ? AND block_level = ?",
  84                             (
  85                                blocker,
  86                                blocked,
  87                                block_level
  88                            ),
  89                         )
  90
  91                         if fba.c.fetchone() == None:
  92                             # NOISY-DEBUG: print("DEBUG: Blocking:", blocker, blocked, block_level)
  93                             fba.block_instance(blocker, blocked, "unknown", block_level, timestamp, timestamp)
  94
  95                             if block_level == "reject":
  96                                 # NOISY-DEBUG: print("DEBUG: Adding to blockdict:", blocked)
  97                                 blockdict.append(
  98                                     {
  99                                         "blocked": blocked,
 100                                         "reason": None
 101                                     })
 102                         else:
 103                             # NOISY-DEBUG: print("DEBUG: Updating last_seen:", blocker, blocked, block_level)
 104                             fba.update_last_seen(timestamp, blocker, blocked, block_level)
 105
 106             fba.conn.commit()
 107
 108             # Reasons
 109             if "mrf_simple_info" in federation:
 110                 # NOISY-DEBUG: print("DEBUG: Found mrf_simple_info:", blocker)
 111                 for block_level, info in (
 112                     {**federation["mrf_simple_info"],
 113                     **(federation["quarantined_instances_info"]
 114                     if "quarantined_instances_info" in federation
 115                     else {})}
 116                 ).items():
 117                     # NOISY-DEBUG: print("DEBUG: block_level, info.items():", block_level, len(info.items()))
 118                     block_level = fba.tidyup(block_level)
 119                     # NOISY-DEBUG: print("DEBUG: BEFORE block_level:", block_level)
 120
 121                     if block_level == "":
 122                         print("WARNING: block_level is now empty!")
 123                         continue
 124
 125                     for blocked, reason in info.items():
 126                         # NOISY-DEBUG: print("DEBUG: BEFORE blocked:", blocked)
 127                         blocked = fba.tidyup(blocked)
 128                         # NOISY-DEBUG: print("DEBUG: AFTER blocked:", blocked)
 129
 130                         if blocked == "":
 131                             print("WARNING: blocked is empty after fba.tidyup():", blocker, block_level)
 132                             continue
 133                         elif blocked.count("*") > 1:
 134                             # same domain guess as above, but for reasons field
 135                             fba.c.execute(
 136                                 "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", [blocked.replace("*", "_")]
 137                             )
 138                             searchres = fba.c.fetchone()
 139
 140                             if searchres != None:
 141                                 blocked = searchres[0]
 142
 143                         # NOISY-DEBUG: print("DEBUG: Updating block reason:", blocker, blocked, reason["reason"])
 144                         fba.update_block_reason(reason["reason"], blocker, blocked, block_level)
 145
 146                         for entry in blockdict:
 147                             if entry["blocked"] == blocked:
 148                                 # NOISY-DEBUG: print("DEBUG: Updating entry reason:", blocked)
 149                                 entry["reason"] = reason["reason"]
 150
 151             fba.conn.commit()
 152         except Exception as e:
 153             print("error:", e, blocker, software)
 154     elif software == "mastodon":
 155         print("INFO: blocker:", blocker)
 156         try:
 157             # json endpoint for newer mastodongs
 158             try:
 159                 json = {
 160                     "reject": [],
 161                     "media_removal": [],
 162                     "followers_only": [],
 163                     "report_removal": []
 164                 }
 165
 166                 # handling CSRF, I've saw at least one server requiring it to access the endpoint
 167                 # NOISY-DEBUG: print("DEBUG: Fetching meta:", blocker)
 168                 meta = bs4.BeautifulSoup(
 169                     reqto.get(f"https://{blocker}/about", headers=fba.headers, timeout=fba.config["timeout"]).text,
 170                     "html.parser",
 171                 )
 172                 try:
 173                     csrf = meta.find("meta", attrs={"name": "csrf-token"})["content"]
 174                     # NOISY-DEBUG: print("DEBUG: Adding CSRF token:", blocker, csrf)
 175                     reqheaders = {**fba.headers, **{"x-csrf-token": csrf}}
 176                 except:
 177                     # NOISY-DEBUG: print("DEBUG: No CSRF token found, using normal headers:", blocker)
 178                     reqheaders = fba.headers
 179
 180                 # NOISY-DEBUG: print("DEBUG: Quering API domain_blocks:", blocker)
 181                 blocks = reqto.get(f"https://{blocker}/api/v1/instance/domain_blocks", headers=reqheaders, timeout=fba.config["timeout"]).json()
 182
 183                 # NOISY-DEBUG: print("DEBUG: blocks():", len(blocks))
 184                 for block in blocks:
 185                     entry = {
 186                         'domain': block['domain'],
 187                         'hash': block['digest'],
 188                         'reason': block['comment']
 189                     }
 190
 191                     # NOISY-DEBUG: print("DEBUG: severity,domain,hash,comment:", block['severity'], block['domain'], block['digest'], block['comment'])
 192                     if block['severity'] == 'suspend':
 193                         json['reject'].append(entry)
 194                     elif block['severity'] == 'silence':
 195                         json['followers_only'].append(entry)
 196                     elif block['severity'] == 'reject_media':
 197                         json['media_removal'].append(entry)
 198                     elif block['severity'] == 'reject_reports':
 199                         json['report_removal'].append(entry)
 200                     else:
 201                         print("WARNING: Unknown severity:", block['severity'], block['domain'])
 202             except:
 203                 # NOISY-DEBUG: print("DEBUG: Failed, Trying mastodon-specific fetches:", blocker)
 204                 json = fba.get_mastodon_blocks(blocker)
 205
 206             # NOISY-DEBUG: print("DEBUG: json.items():", blocker, len(json.items()))
 207             for block_level, blocks in json.items():
 208                 # NOISY-DEBUG: print("DEBUG: blocker,block_level,blocks():", blocker, block_level, len(blocks))
 209                 block_level = fba.tidyup(block_level)
 210                 # NOISY-DEBUG: print("DEBUG: AFTER-block_level:", block_level)
 211                 if block_level == "":
 212                     print("WARNING: block_level is empty, blocker:", blocker)
 213                     continue
 214
 215                 for instance in blocks:
 216                     blocked, blocked_hash, reason = instance.values()
 217                     # NOISY-DEBUG: print("DEBUG: blocked,hash,reason:", blocked, blocked_hash, reason)
 218                     blocked = fba.tidyup(blocked)
 219                     # NOISY-DEBUG: print("DEBUG: AFTER-blocked:", blocked)
 220
 221                     if blocked == "":
 222                         print("WARNING: blocked is empty:", blocker)
 223                         continue
 224                     elif blocked.count("*") < 1:
 225                         # No obsfucation for this instance
 226                         fba.c.execute(
 227                             "SELECT hash FROM instances WHERE domain = ? LIMIT 1", [blocked]
 228                         )
 229
 230                         if fba.c.fetchone() == None:
 231                             # NOISY-DEBUG: print("DEBUG: Hash wasn't found, adding:", blocked, blocker)
 232                             fba.add_instance(blocked, blocker, argv[0])
 233                     else:
 234                         # Doing the hash search for instance names as well to tidy up DB
 235                         fba.c.execute(
 236                             "SELECT domain FROM instances WHERE hash = ? LIMIT 1", [blocked_hash]
 237                         )
 238                         searchres = fba.c.fetchone()
 239
 240                         if searchres != None:
 241                             # NOISY-DEBUG: print("DEBUG: Updating domain: ", searchres[0])
 242                             blocked = searchres[0]
 243
 244                     timestamp = int(time.time())
 245                     fba.c.execute(
 246                         "SELECT * FROM blocks WHERE blocker = ? AND blocked = ? AND block_level = ?",
 247                         (
 248                             blocker,
 249                             blocked if blocked.count("*") <= 1 else blocked_hash,
 250                             block_level
 251                         ),
 252                     )
 253
 254                     if fba.c.fetchone() == None:
 255                         fba.block_instance(blocker, blocked if blocked.count("*") <= 1 else blocked_hash, reason, block_level, timestamp, timestamp)
 256
 257                         if block_level == "reject":
 258                             blockdict.append(
 259                                 {
 260                                     "blocked": blocked,
 261                                     "reason": reason
 262                                 })
 263                     else:
 264                         fba.update_last_seen(timestamp, blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level)
 265
 266                     if reason != '':
 267                         # NOISY-DEBUG: print("DEBUG: Updating block reason:", blocker, blocked, reason)
 268                         fba.update_block_reason(reason, blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level)
 269
 270             fba.conn.commit()
 271         except Exception as e:
 272             print("error:", e, blocker, software)
 273     elif software == "friendica" or software == "misskey":
 274         print("INFO: blocker:", blocker)
 275         try:
 276             if software == "friendica":
 277                 json = fba.get_friendica_blocks(blocker)
 278             elif software == "misskey":
 279                 json = fba.get_misskey_blocks(blocker)
 280
 281             for block_level, blocks in json.items():
 282                 # NOISY-DEBUG: print("DEBUG: blocker,block_level,blocks():", blocker, block_level, len(blocks))
 283                 block_level = fba.tidyup(block_level)
 284                 # NOISY-DEBUG: print("DEBUG: AFTER-block_level:", block_level)
 285                 if block_level == "":
 286                     print("WARNING: block_level is empty, blocker:", blocker)
 287                     continue
 288
 289                 for instance in blocks:
 290                     blocked, reason = instance.values()
 291                     # NOISY-DEBUG: print("DEBUG: BEFORE-blocked:", blocked)
 292                     blocked = fba.tidyup(blocked)
 293                     # NOISY-DEBUG: print("DEBUG: AFTER-blocked:", blocked)
 294
 295                     if blocked == "":
 296                         print("WARNING: blocked is empty:", blocker)
 297                         continue
 298                     if blocked.count("*") > 0:
 299                         # Some friendica servers also obscure domains without hash
 300                         fba.c.execute(
 301                             "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", [blocked.replace("*", "_")]
 302                         )
 303                         searchres = fba.c.fetchone()
 304                         if searchres != None:
 305                             blocked = searchres[0]
 306
 307                     if blocked.count("?") > 0:
 308                         # Some obscure them with question marks, not sure if that's dependent on version or not
 309                         fba.c.execute(
 310                             "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", [blocked.replace("?", "_")]
 311                         )
 312                         searchres = fba.c.fetchone()
 313                         if searchres != None:
 314                             blocked = searchres[0]
 315
 316                     # NOISY-DEBUG: print("DEBUG: AFTER-blocked:", blocked)
 317                     fba.c.execute(
 318                         "SELECT domain FROM instances WHERE domain = ?", [blocked]
 319                     )
 320
 321                     if fba.c.fetchone() == None:
 322                         # NOISY-DEBUG: print("DEBUG: Hash wasn't found, adding:", blocked, blocker)
 323                         fba.add_instance(blocked, blocker)
 324
 325                     timestamp = int(time.time())
 326                     fba.c.execute(
 327                         "SELECT * FROM blocks WHERE blocker = ? AND blocked = ?",
 328                         (blocker, blocked),
 329                     )
 330                     if fba.c.fetchone() == None:
 331                         fba.block_instance(blocker, blocked, reason, block_level, timestamp, timestamp)
 332
 333                         if block_level == "reject":
 334                             blockdict.append(
 335                                 {
 336                                     "blocked": blocked,
 337                                     "reason": reason
 338                                 })
 339                     else:
 340                         fba.update_last_seen(timestamp, blocker, blocked, block_level)
 341
 342                     if reason != '':
 343                         # NOISY-DEBUG: print("DEBUG: Updating block reason:", blocker, blocked, reason)
 344                         fba.update_block_reason(reason, blocker, blocked, block_level)
 345
 346             fba.conn.commit()
 347         except Exception as e:
 348             print("error:", e, blocker, software)
 349     elif software == "gotosocial":
 350         print("INFO: blocker:", blocker)
 351         try:
 352             # Blocks
 353             federation = reqto.get(f"https://{blocker}/api/v1/instance/peers?filter=suspended", headers=fba.headers, timeout=fba.config["timeout"]).json()
 354
 355             if (federation == None):
 356                 print("WARNING: No valid response:", blocker);
 357             elif "error" in federation:
 358                 print("WARNING: API returned error:", federation["error"])
 359             else:
 360                 # NOISY-DEBUG: print("DEBUG: Checking fenderation():", len(federation))
 361                 for peer in federation:
 362                     blocked = peer["domain"].lower()
 363                     # NOISY-DEBUG: print("DEBUG: BEFORE-blocked:", blocked)
 364                     blocked = fba.tidyup(blocked)
 365                     # NOISY-DEBUG: print("DEBUG: AFTER-blocked:", blocked)
 366
 367                     if blocked == "":
 368                         print("WARNING: blocked is empty:", blocker)
 369                         continue
 370                     if blocked.count("*") > 0:
 371                         # GTS does not have hashes for obscured domains, so we have to guess it
 372                         fba.c.execute(
 373                             "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", [blocked.replace("*", "_")]
 374                         )
 375                         searchres = fba.c.fetchone()
 376
 377                         if searchres != None:
 378                             blocked = searchres[0]
 379
 380                     fba.c.execute(
 381                         "SELECT domain FROM instances WHERE domain = ?", [blocked]
 382                     )
 383
 384                     if fba.c.fetchone() == None:
 385                         # NOISY-DEBUG: print("DEBUG: Updating block reason:", blocker, blocked, reason)
 386                         fba.update_block_reason(reason, blocker, blocked, block_level)
 387
 388                     fba.c.execute(
 389                         "SELECT * FROM blocks WHERE blocker = ? AND blocked = ? AND block_level = ?",
 390                         (
 391                             blocker,
 392                             blocked,
 393                             "reject"
 394                         ),
 395                     )
 396                     timestamp = int(time.time())
 397
 398                     if fba.c.fetchone() == None:
 399                         fba.block_instance(blocker, blocked, "unknown", "reject", timestamp, timestamp)
 400
 401                         blockdict.append(
 402                             {
 403                                 "blocked": blocked,
 404                                 "reason": None
 405                             })
 406                     else:
 407                         fba.update_last_seen(timestamp, blocker, blocked, "reject")
 408
 409                     if "public_comment" in peer:
 410                         reason = peer["public_comment"]
 411                         # NOISY-DEBUG: print("DEBUG: Updating block reason:", blocker, blocked, reason)
 412                         fba.update_block_reason(reason, blocker, blocked, "reject")
 413
 414                         for entry in blockdict:
 415                             if entry["blocked"] == blocked:
 416                                 entry["reason"] = reason
 417
 418                 fba.conn.commit()
 419         except Exception as e:
 420             print("error:", e, blocker, software)
 421     else:
 422         print("WARNING: Unknown software:", software)
 423
 424     if fba.config["bot_enabled"] and len(blockdict) > 0:
 425         send_bot_post(blocker, blockdict)
 426
 427     blockdict = []
 428
 429 fba.conn.close()