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