fetch_blocks.py

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