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