fetch_blocks.py

   1 import reqto
   2 import time
   3 import bs4
   4 import itertools
   5 import re
   6 import fba
   7
   8 fba.cursor.execute(
   9     "SELECT domain, software FROM instances WHERE software IN ('pleroma', 'mastodon', 'friendica', 'misskey', 'gotosocial', 'bookwyrm', 'takahe') AND (last_blocked IS NULL OR last_blocked < ?) ORDER BY rowid DESC", [time.time() - fba.config["recheck_block"]]
  10 )
  11
  12 rows = fba.cursor.fetchall()
  13 print(f"INFO: Checking {len(rows)} entries ...")
  14 for blocker, software in rows:
  15     # NOISY-DEBUG: print("DEBUG: BEFORE blocker,software:", blocker, software)
  16     blockdict = []
  17     blocker = fba.tidyup(blocker)
  18     # NOISY-DEBUG: print("DEBUG: AFTER blocker,software:", blocker, software)
  19
  20     if blocker == "":
  21         print("WARNING: blocker is now empty!")
  22         continue
  23     elif fba.is_blacklisted(blocker):
  24         print(f"WARNING: blocker='{blocker}' is blacklisted now!")
  25         continue
  26
  27     # NOISY-DEBUG: print(f"DEBUG: blocker='{blocker}'")
  28     fba.update_last_blocked(blocker)
  29
  30     if software == "pleroma":
  31         print("INFO: blocker:", blocker)
  32         try:
  33             # Blocks
  34             json = fba.fetch_nodeinfo(blocker)
  35             if json is None:
  36                 print("WARNING: Could not fetch nodeinfo from blocker:", blocker)
  37                 continue
  38
  39             print("DEBUG: Updating nodeinfo:", blocker)
  40             fba.update_last_nodeinfo(blocker)
  41
  42             federation = json["metadata"]["federation"]
  43
  44             if "enabled" in federation:
  45                 # NOISY-DEBUG: print("DEBUG: Instance has no block list to analyze:", blocker)
  46                 continue
  47
  48             if "mrf_simple" in federation:
  49                 for block_level, blocks in (
  50                     {**federation["mrf_simple"],
  51                     **{"quarantined_instances": federation["quarantined_instances"]}}
  52                 ).items():
  53                     # NOISY-DEBUG: print("DEBUG: block_level, blocks():", block_level, len(blocks))
  54                     block_level = fba.tidyup(block_level)
  55                     # NOISY-DEBUG: print("DEBUG: BEFORE block_level:", block_level)
  56
  57                     if block_level == "":
  58                         print("WARNING: block_level is now empty!")
  59                         continue
  60
  61                     for blocked in blocks:
  62                         # NOISY-DEBUG: print("DEBUG: BEFORE blocked:", blocked)
  63                         blocked = fba.tidyup(blocked)
  64                         # NOISY-DEBUG: print("DEBUG: AFTER blocked:", blocked)
  65
  66                         if blocked == "":
  67                             print("WARNING: blocked is empty after fba.tidyup():", blocker, block_level)
  68                             continue
  69
  70                         if blocked.count("*") > 1:
  71                             # -ACK!-oma also started obscuring domains without hash
  72                             fba.cursor.execute(
  73                                 "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", [blocked.replace("*", "_")]
  74                             )
  75                             searchres = fba.cursor.fetchone()
  76                             # NOISY-DEBUG: print("DEBUG: searchres[]:", type(searchres))
  77                             if searchres != None:
  78                                 blocked = searchres[0]
  79                                 # NOISY-DEBUG: print("DEBUG: Looked up domain:", blocked)
  80
  81                         # NOISY-DEBUG: print("DEBUG: Looking up instance by domain:", blocked)
  82                         fba.cursor.execute(
  83                             "SELECT domain FROM instances WHERE domain = ? LIMIT 1", [blocked]
  84                         )
  85
  86                         if fba.cursor.fetchone() == None:
  87                             # NOISY-DEBUG: print("DEBUG: Domain wasn't found, adding:", blocked, blocker)
  88                             fba.add_instance(blocked, blocker, argv[0])
  89
  90                         fba.cursor.execute(
  91                             "SELECT * FROM blocks WHERE blocker = ? AND blocked = ? AND block_level = ?",
  92                             (
  93                                blocker,
  94                                blocked,
  95                                block_level
  96                            ),
  97                         )
  98
  99                         if fba.cursor.fetchone() == None:
 100                             # NOISY-DEBUG: print("DEBUG: Blocking:", blocker, blocked, block_level)
 101                             fba.block_instance(blocker, blocked, "unknown", block_level)
 102
 103                             if block_level == "reject":
 104                                 # NOISY-DEBUG: print("DEBUG: Adding to blockdict:", blocked)
 105                                 blockdict.append(
 106                                     {
 107                                         "blocked": blocked,
 108                                         "reason": None
 109                                     })
 110                         else:
 111                             # NOISY-DEBUG: print("DEBUG: Updating last_seen:", blocker, blocked, block_level)
 112                             fba.update_last_seen(blocker, blocked, block_level)
 113
 114             fba.connection.commit()
 115
 116             # Reasons
 117             if "mrf_simple_info" in federation:
 118                 # NOISY-DEBUG: print("DEBUG: Found mrf_simple_info:", blocker)
 119                 for block_level, info in (
 120                     {**federation["mrf_simple_info"],
 121                     **(federation["quarantined_instances_info"]
 122                     if "quarantined_instances_info" in federation
 123                     else {})}
 124                 ).items():
 125                     # NOISY-DEBUG: print("DEBUG: block_level, info.items():", block_level, len(info.items()))
 126                     block_level = fba.tidyup(block_level)
 127                     # NOISY-DEBUG: print("DEBUG: BEFORE block_level:", block_level)
 128
 129                     if block_level == "":
 130                         print("WARNING: block_level is now empty!")
 131                         continue
 132
 133                     for blocked, reason in info.items():
 134                         # NOISY-DEBUG: print("DEBUG: BEFORE blocked:", blocked)
 135                         blocked = fba.tidyup(blocked)
 136                         # NOISY-DEBUG: print("DEBUG: AFTER blocked:", blocked)
 137
 138                         if blocked == "":
 139                             print("WARNING: blocked is empty after fba.tidyup():", blocker, block_level)
 140                             continue
 141                         elif blocked.count("*") > 1:
 142                             # same domain guess as above, but for reasons field
 143                             fba.cursor.execute(
 144                                 "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", [blocked.replace("*", "_")]
 145                             )
 146                             searchres = fba.cursor.fetchone()
 147
 148                             if searchres != None:
 149                                 blocked = searchres[0]
 150
 151                         # NOISY-DEBUG: print("DEBUG: Updating block reason:", blocker, blocked, reason["reason"])
 152                         fba.update_block_reason(reason["reason"], blocker, blocked, block_level)
 153
 154                         for entry in blockdict:
 155                             if entry["blocked"] == blocked:
 156                                 # NOISY-DEBUG: print("DEBUG: Updating entry reason:", blocked)
 157                                 entry["reason"] = reason["reason"]
 158
 159             fba.connection.commit()
 160         except Exception as e:
 161             print("error:", e, blocker, software)
 162     elif software == "mastodon":
 163         print("INFO: blocker:", blocker)
 164         try:
 165             # json endpoint for newer mastodongs
 166             try:
 167                 json = {
 168                     "reject": [],
 169                     "media_removal": [],
 170                     "followers_only": [],
 171                     "report_removal": []
 172                 }
 173
 174                 # handling CSRF, I've saw at least one server requiring it to access the endpoint
 175                 # NOISY-DEBUG: print("DEBUG: Fetching meta:", blocker)
 176                 meta = bs4.BeautifulSoup(
 177                     reqto.get(f"https://{blocker}/about", headers=fba.headers, timeout=(fba.config["connection_timeout"], config["read_timeout"])).text,
 178                     "html.parser",
 179                 )
 180                 try:
 181                     csrf = meta.find("meta", attrs={"name": "csrf-token"})["content"]
 182                     # NOISY-DEBUG: print("DEBUG: Adding CSRF token:", blocker, csrf)
 183                     reqheaders = {**fba.headers, **{"x-csrf-token": csrf}}
 184                 except:
 185                     # NOISY-DEBUG: print("DEBUG: No CSRF token found, using normal headers:", blocker)
 186                     reqheaders = fba.headers
 187
 188                 # NOISY-DEBUG: print("DEBUG: Quering API domain_blocks:", blocker)
 189                 blocks = reqto.get(f"https://{blocker}/api/v1/instance/domain_blocks", headers=reqheaders, timeout=(fba.config["connection_timeout"], config["read_timeout"])).json()
 190
 191                 # NOISY-DEBUG: print("DEBUG: blocks():", len(blocks))
 192                 for block in blocks:
 193                     entry = {
 194                         'domain': block['domain'],
 195                         'hash': block['digest'],
 196                         'reason': block['comment']
 197                     }
 198
 199                     # NOISY-DEBUG: print("DEBUG: severity,domain,hash,comment:", block['severity'], block['domain'], block['digest'], block['comment'])
 200                     if block['severity'] == 'suspend':
 201                         json['reject'].append(entry)
 202                     elif block['severity'] == 'silence':
 203                         json['followers_only'].append(entry)
 204                     elif block['severity'] == 'reject_media':
 205                         json['media_removal'].append(entry)
 206                     elif block['severity'] == 'reject_reports':
 207                         json['report_removal'].append(entry)
 208                     else:
 209                         print("WARNING: Unknown severity:", block['severity'], block['domain'])
 210             except:
 211                 # NOISY-DEBUG: print("DEBUG: Failed, Trying mastodon-specific fetches:", blocker)
 212                 json = fba.get_mastodon_blocks(blocker)
 213
 214             # NOISY-DEBUG: print("DEBUG: json.items():", blocker, len(json.items()))
 215             for block_level, blocks in json.items():
 216                 # NOISY-DEBUG: print("DEBUG: blocker,block_level,blocks():", blocker, block_level, len(blocks))
 217                 block_level = fba.tidyup(block_level)
 218                 # NOISY-DEBUG: print("DEBUG: AFTER-block_level:", block_level)
 219                 if block_level == "":
 220                     print("WARNING: block_level is empty, blocker:", blocker)
 221                     continue
 222
 223                 for instance in blocks:
 224                     blocked, blocked_hash, reason = instance.values()
 225                     # NOISY-DEBUG: print("DEBUG: blocked,hash,reason:", blocked, blocked_hash, reason)
 226                     blocked = fba.tidyup(blocked)
 227                     # NOISY-DEBUG: print("DEBUG: AFTER-blocked:", blocked)
 228
 229                     if blocked == "":
 230                         print("WARNING: blocked is empty:", blocker)
 231                         continue
 232                     elif blocked.count("*") < 1:
 233                         # No obsfucation for this instance
 234                         fba.cursor.execute(
 235                             "SELECT hash FROM instances WHERE domain = ? LIMIT 1", [blocked]
 236                         )
 237
 238                         if fba.cursor.fetchone() == None:
 239                             # NOISY-DEBUG: print("DEBUG: Hash wasn't found, adding:", blocked, blocker)
 240                             fba.add_instance(blocked, blocker, argv[0])
 241                     else:
 242                         # Doing the hash search for instance names as well to tidy up DB
 243                         fba.cursor.execute(
 244                             "SELECT domain FROM instances WHERE hash = ? LIMIT 1", [blocked_hash]
 245                         )
 246                         searchres = fba.cursor.fetchone()
 247
 248                         if searchres != None:
 249                             # NOISY-DEBUG: print("DEBUG: Updating domain: ", searchres[0])
 250                             blocked = searchres[0]
 251
 252                     fba.cursor.execute(
 253                         "SELECT * FROM blocks WHERE blocker = ? AND blocked = ? AND block_level = ?",
 254                         (
 255                             blocker,
 256                             blocked if blocked.count("*") <= 1 else blocked_hash,
 257                             block_level
 258                         ),
 259                     )
 260
 261                     if fba.cursor.fetchone() == None:
 262                         fba.block_instance(blocker, blocked if blocked.count("*") <= 1 else blocked_hash, reason, block_level)
 263
 264                         if block_level == "reject":
 265                             blockdict.append(
 266                                 {
 267                                     "blocked": blocked,
 268                                     "reason": reason
 269                                 })
 270                     else:
 271                         fba.update_last_seen(blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level)
 272
 273                     if reason != '':
 274                         # NOISY-DEBUG: print("DEBUG: Updating block reason:", blocker, blocked, reason)
 275                         fba.update_block_reason(reason, blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level)
 276
 277             fba.connection.commit()
 278         except Exception as e:
 279             print("error:", e, blocker, software)
 280     elif software == "friendica" or software == "misskey" or software == "bookwyrm" or software == "takahe":
 281         print("INFO: blocker:", blocker)
 282         try:
 283             if software == "friendica":
 284                 json = fba.get_friendica_blocks(blocker)
 285             elif software == "misskey":
 286                 json = fba.get_misskey_blocks(blocker)
 287             elif software == "bookwyrm":
 288                 print("WARNING: bookwyrm is not fully supported for fetching blacklist!", blocker)
 289                 #json = fba.get_bookwyrm_blocks(blocker)
 290             elif software == "takahe":
 291                 print("WARNING: takahe is not fully supported for fetching blacklist!", blocker)
 292                 #json = fba.get_takahe_blocks(blocker)
 293
 294             for block_level, blocks in json.items():
 295                 # NOISY-DEBUG: print("DEBUG: blocker,block_level,blocks():", blocker, block_level, len(blocks))
 296                 block_level = fba.tidyup(block_level)
 297                 # NOISY-DEBUG: print("DEBUG: AFTER-block_level:", block_level)
 298                 if block_level == "":
 299                     print("WARNING: block_level is empty, blocker:", blocker)
 300                     continue
 301
 302                 for instance in blocks:
 303                     blocked, reason = instance.values()
 304                     # NOISY-DEBUG: print("DEBUG: BEFORE blocked:", blocked)
 305                     blocked = fba.tidyup(blocked)
 306                     # NOISY-DEBUG: print("DEBUG: AFTER blocked:", blocked)
 307
 308                     if blocked == "":
 309                         print("WARNING: blocked is empty:", blocker)
 310                         continue
 311                     if blocked.count("*") > 0:
 312                         # Some friendica servers also obscure domains without hash
 313                         fba.cursor.execute(
 314                             "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", [blocked.replace("*", "_")]
 315                         )
 316                         searchres = fba.cursor.fetchone()
 317                         if searchres != None:
 318                             blocked = searchres[0]
 319
 320                     if blocked.count("?") > 0:
 321                         # Some obscure them with question marks, not sure if that's dependent on version or not
 322                         fba.cursor.execute(
 323                             "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", [blocked.replace("?", "_")]
 324                         )
 325                         searchres = fba.cursor.fetchone()
 326                         if searchres != None:
 327                             blocked = searchres[0]
 328
 329                     # NOISY-DEBUG: print("DEBUG: AFTER-blocked:", blocked)
 330                     fba.cursor.execute(
 331                         "SELECT domain FROM instances WHERE domain = ? LIMIT 1", [blocked]
 332                     )
 333
 334                     if fba.cursor.fetchone() == None:
 335                         # NOISY-DEBUG: print("DEBUG: Hash wasn't found, adding:", blocked, blocker)
 336                         fba.add_instance(blocked, blocker)
 337
 338                     fba.cursor.execute(
 339                         "SELECT * FROM blocks WHERE blocker = ? AND blocked = ?",
 340                         (blocker, blocked),
 341                     )
 342                     if fba.cursor.fetchone() == None:
 343                         fba.block_instance(blocker, blocked, reason, block_level)
 344
 345                         if block_level == "reject":
 346                             blockdict.append(
 347                                 {
 348                                     "blocked": blocked,
 349                                     "reason": reason
 350                                 })
 351                     else:
 352                         fba.update_last_seen(blocker, blocked, block_level)
 353
 354                     if reason != '':
 355                         # NOISY-DEBUG: print("DEBUG: Updating block reason:", blocker, blocked, reason)
 356                         fba.update_block_reason(reason, blocker, blocked, block_level)
 357
 358             fba.connection.commit()
 359         except Exception as e:
 360             print("error:", e, blocker, software)
 361     elif software == "gotosocial":
 362         print("INFO: blocker:", blocker)
 363         try:
 364             # Blocks
 365             federation = reqto.get(f"https://{blocker}{get_peers_url}?filter=suspended", headers=fba.headers, timeout=(fba.config["connection_timeout"], config["read_timeout"])).json()
 366
 367             if (federation == None):
 368                 print("WARNING: No valid response:", blocker);
 369             elif "error" in federation:
 370                 print("WARNING: API returned error:", federation["error"])
 371             else:
 372                 # NOISY-DEBUG: print("DEBUG: Checking fenderation():", len(federation))
 373                 for peer in federation:
 374                     blocked = peer["domain"].lower()
 375                     # NOISY-DEBUG: print("DEBUG: BEFORE blocked:", blocked)
 376                     blocked = fba.tidyup(blocked)
 377                     # NOISY-DEBUG: print("DEBUG: AFTER blocked:", blocked)
 378
 379                     if blocked == "":
 380                         print("WARNING: blocked is empty:", blocker)
 381                         continue
 382                     elif blocked.count("*") > 0:
 383                         # GTS does not have hashes for obscured domains, so we have to guess it
 384                         fba.cursor.execute(
 385                             "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", [blocked.replace("*", "_")]
 386                         )
 387                         searchres = fba.cursor.fetchone()
 388
 389                         if searchres != None:
 390                             blocked = searchres[0]
 391
 392                     fba.cursor.execute(
 393                         "SELECT domain FROM instances WHERE domain = ? LIMIT 1", [blocked]
 394                     )
 395
 396                     if fba.cursor.fetchone() == None:
 397                         # NOISY-DEBUG: print("DEBUG: Updating block reason:", blocker, blocked, reason)
 398                         fba.update_block_reason(reason, blocker, blocked, block_level)
 399
 400                     fba.cursor.execute(
 401                         "SELECT * FROM blocks WHERE blocker = ? AND blocked = ? AND block_level = ?",
 402                         (
 403                             blocker,
 404                             blocked,
 405                             "reject"
 406                         ),
 407                     )
 408
 409                     if fba.cursor.fetchone() == None:
 410                         fba.block_instance(blocker, blocked, "unknown", "reject")
 411
 412                         blockdict.append(
 413                             {
 414                                 "blocked": blocked,
 415                                 "reason": None
 416                             })
 417                     else:
 418                         fba.update_last_seen(blocker, blocked, "reject")
 419
 420                     if "public_comment" in peer:
 421                         # NOISY-DEBUG: print("DEBUG: Updating block reason:", blocker, blocked, peer["public_comment"])
 422                         fba.update_block_reason(peer["public_comment"], blocker, blocked, "reject")
 423
 424                         for entry in blockdict:
 425                             if entry["blocked"] == blocked:
 426                                 entry["reason"] = peer["public_comment"]
 427
 428                 fba.connection.commit()
 429         except Exception as e:
 430             print("error:", e, blocker, software)
 431     else:
 432         print("WARNING: Unknown software:", blocker, software)
 433
 434     if fba.config["bot_enabled"] and len(blockdict) > 0:
 435         send_bot_post(blocker, blockdict)
 436
 437     blockdict = []
 438
 439 fba.connection.close()