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')"
  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             federation = reqto.get(
  27                 f"https://{blocker}/nodeinfo/2.1.json", headers=fba.headers, timeout=5
  28             ).json()["metadata"]["federation"]
  29             if "mrf_simple" in federation:
  30                 for block_level, blocks in (
  31                     {**federation["mrf_simple"],
  32                     **{"quarantined_instances": federation["quarantined_instances"]}}
  33                 ).items():
  34                     # NOISY-DEBUG: print("DEBUG: block_level, blocks.items():", block_level, len(blocks.items()))
  35                     block_level = fba.tidyup(block_level)
  36                     # NOISY-DEBUG: print("DEBUG: BEFORE block_level:", block_level)
  37
  38                     if block_level == "":
  39                         print("WARNING: block_level is now empty!")
  40                         continue
  41
  42                     for blocked in blocks:
  43                         # NOISY-DEBUG: print("DEBUG: BEFORE blocked:", blocked)
  44                         blocked = fba.tidyup(blocked)
  45                         # NOISY-DEBUG: print("DEBUG: AFTER blocked:", blocked)
  46
  47                         if blocked == "":
  48                             print("WARNING: blocked is empty after fba.tidyup():", blocker, block_level)
  49                             continue
  50
  51                         if blocked.count("*") > 1:
  52                             # -ACK!-oma also started obscuring domains without hash
  53                             fba.c.execute(
  54                                 "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", (blocked.replace("*", "_"),)
  55                             )
  56                             searchres = fba.c.fetchone()
  57                             # NOISY-DEBUG: print("DEBUG: searchres[]:", type(searchres))
  58                             if searchres != None:
  59                                 blocked = searchres[0]
  60                                 # NOISY-DEBUG: print("DEBUG: Looked up domain:", blocked)
  61
  62                         # NOISY-DEBUG: print("DEBUG: Looking up instance by domain:", blocked)
  63                         fba.c.execute(
  64                             "SELECT domain FROM instances WHERE domain = ?", (blocked,)
  65                         )
  66
  67                         if fba.c.fetchone() == None:
  68                             # NOISY-DEBUG: print("DEBUG: Domain wasn't found, adding:", blocked)
  69                             fba.add_instance(blocked)
  70
  71                         timestamp = int(time.time())
  72                         fba.c.execute(
  73                             "SELECT * FROM blocks WHERE blocker = ? AND blocked = ? AND block_level = ?",
  74                             (
  75                                blocker,
  76                                blocked,
  77                                block_level
  78                            ),
  79                         )
  80
  81                         if fba.c.fetchone() == None:
  82                             # NOISY-DEBUG: print("DEBUG: Blocking:", blocker, blocked, block_level)
  83                             fba.block_instance(blocker, blocked, "unknown", block_level, timestamp, timestamp)
  84
  85                             if block_level == "reject":
  86                                 # NOISY-DEBUG: print("DEBUG: Adding to blockdict:", blocked)
  87                                 blockdict.append(
  88                                     {
  89                                         "blocked": blocked,
  90                                         "reason": None
  91                                     })
  92                         else:
  93                             # NOISY-DEBUG: print("DEBUG: Updating last_seen:", blocker, blocked, block_level)
  94                             fba.update_last_seen(timestamp, blocker, blocked, block_level)
  95
  96             fba.conn.commit()
  97
  98             # Reasons
  99             if "mrf_simple_info" in federation:
 100                 # NOISY-DEBUG: print("DEBUG: Found mrf_simple_info:", blocker)
 101                 for block_level, info in (
 102                     {**federation["mrf_simple_info"],
 103                     **(federation["quarantined_instances_info"]
 104                     if "quarantined_instances_info" in federation
 105                     else {})}
 106                 ).items():
 107                     # NOISY-DEBUG: print("DEBUG: block_level, info.items():", block_level, len(info.items()))
 108                     block_level = fba.tidyup(block_level)
 109                     # NOISY-DEBUG: print("DEBUG: BEFORE block_level:", block_level)
 110
 111                     if block_level == "":
 112                         print("WARNING: block_level is now empty!")
 113                         continue
 114
 115                     for blocked, reason in info.items():
 116                         # NOISY-DEBUG: print("DEBUG: BEFORE blocked:", blocked)
 117                         blocked = fba.tidyup(blocked)
 118                         # NOISY-DEBUG: print("DEBUG: AFTER blocked:", blocked)
 119
 120                         if blocked == "":
 121                             print("WARNING: blocked is empty after fba.tidyup():", blocker, block_level)
 122                             continue
 123                         elif blocked.count("*") > 1:
 124                             # same domain guess as above, but for reasons field
 125                             fba.c.execute(
 126                                 "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", (blocked.replace("*", "_"),)
 127                             )
 128                             searchres = fba.c.fetchone()
 129
 130                             if searchres != None:
 131                                 blocked = searchres[0]
 132
 133                         # NOISY-DEBUG: print("DEBUG: Updating block reason:", blocker, blocked, reason["reason"])
 134                         fba.update_block_reason(reason["reason"], blocker, blocked, block_level)
 135
 136                         for entry in blockdict:
 137                             if entry["blocked"] == blocked:
 138                                 # NOISY-DEBUG: print("DEBUG: Updating entry reason:", blocked)
 139                                 entry["reason"] = reason["reason"]
 140
 141             fba.conn.commit()
 142         except Exception as e:
 143             print("error:", e, blocker, software)
 144     elif software == "mastodon":
 145         print("INFO: blocker:", blocker)
 146         try:
 147             # json endpoint for newer mastodongs
 148             try:
 149                 json = {
 150                     "reject": [],
 151                     "media_removal": [],
 152                     "followers_only": [],
 153                     "report_removal": []
 154                 }
 155
 156                 # handling CSRF, I've saw at least one server requiring it to access the endpoint
 157                 # NOISY-DEBUG: print("DEBUG: Fetching meta:", blocker)
 158                 meta = bs4.BeautifulSoup(
 159                     reqto.get(f"https://{blocker}/about", headers=fba.headers, timeout=5).text,
 160                     "html.parser",
 161                 )
 162                 try:
 163                     csrf = meta.find("meta", attrs={"name": "csrf-token"})["content"]
 164                     # NOISY-DEBUG: print("DEBUG: Adding CSRF token:", blocker, csrf)
 165                     reqheaders = {**fba.headers, **{"x-csrf-token": csrf}}
 166                 except:
 167                     # NOISY-DEBUG: print("DEBUG: No CSRF token found, using normal headers:", blocker)
 168                     reqheaders = fba.headers
 169
 170                 # NOISY-DEBUG: print("DEBUG: Quering API domain_blocks:", blocker)
 171                 blocks = reqto.get(
 172                     f"https://{blocker}/api/v1/instance/domain_blocks", headers=reqheaders, timeout=5
 173                 ).json()
 174
 175                 # NOISY-DEBUG: print("DEBUG: blocks():", len(blocks))
 176                 for block in blocks:
 177                     entry = {
 178                         'domain': block['domain'],
 179                         'hash': block['digest'],
 180                         'reason': block['comment']
 181                     }
 182
 183                     # NOISY-DEBUG: print("DEBUG: severity,domain,hash,comment:", block['severity'], block['domain'], block['digest'], block['comment'])
 184                     if block['severity'] == 'suspend':
 185                         json['reject'].append(entry)
 186                     elif block['severity'] == 'silence':
 187                         json['followers_only'].append(entry)
 188                     elif block['severity'] == 'reject_media':
 189                         json['media_removal'].append(entry)
 190                     elif block['severity'] == 'reject_reports':
 191                         json['report_removal'].append(entry)
 192                     else:
 193                         print("WARNING: Unknown severity:", block['severity'], block['domain'])
 194             except:
 195                 # NOISY-DEBUG: print("DEBUG: Failed, Trying mastodon-specific fetches:", blocker)
 196                 json = fba.get_mastodon_blocks(blocker)
 197
 198             # NOISY-DEBUG: print("DEBUG: json.items():", blocker, len(json.items()))
 199             for block_level, blocks in json.items():
 200                 # NOISY-DEBUG: print("DEBUG: blocker,block_level,blocks():", blocker, block_level, len(blocks))
 201                 block_level = fba.tidyup(block_level)
 202                 # NOISY-DEBUG: print("DEBUG: AFTER-block_level:", block_level)
 203                 if block_level == "":
 204                     print("WARNING: block_level is empty, blocker:", blocker)
 205                     continue
 206
 207                 for instance in blocks:
 208                     blocked, blocked_hash, reason = instance.values()
 209                     # NOISY-DEBUG: print("DEBUG: blocked,hash,reason:", blocked, blocked_hash, reason)
 210                     blocked = fba.tidyup(blocked)
 211                     # NOISY-DEBUG: print("DEBUG: AFTER-blocked:", blocked)
 212
 213                     if blocked == "":
 214                         print("WARNING: blocked is empty:", blocker)
 215                         continue
 216                     elif blocked.count("*") < 1:
 217                         # No obsfucation for this instance
 218                         fba.c.execute(
 219                             "SELECT hash FROM instances WHERE domain = ? LIMIT 1", (blocked,)
 220                         )
 221
 222                         if fba.c.fetchone() == None:
 223                             # NOISY-DEBUG: print("DEBUG: Hash wasn't found, adding:", blocked)
 224                             fba.add_instance(blocked)
 225                     else:
 226                         # Doing the hash search for instance names as well to tidy up DB
 227                         fba.c.execute(
 228                             "SELECT domain FROM instances WHERE hash = ? LIMIT 1", (blocked_hash,)
 229                         )
 230                         searchres = fba.c.fetchone()
 231
 232                         if searchres != None:
 233                             # NOISY-DEBUG: print("DEBUG: Updating domain: ", searchres[0])
 234                             blocked = searchres[0]
 235
 236                     timestamp = int(time.time())
 237                     fba.c.execute(
 238                         "SELECT * FROM blocks WHERE blocker = ? AND blocked = ? AND block_level = ?",
 239                         (
 240                             blocker,
 241                             blocked if blocked.count("*") <= 1 else blocked_hash,
 242                             block_level
 243                         ),
 244                     )
 245
 246                     if fba.c.fetchone() == None:
 247                         fba.block_instance(blocker, blocked if blocked.count("*") <= 1 else blocked_hash, reason, block_level, timestamp, timestamp)
 248
 249                         if block_level == "reject":
 250                             blockdict.append(
 251                                 {
 252                                     "blocked": blocked,
 253                                     "reason": reason
 254                                 })
 255                     else:
 256                         fba.update_last_seen(timestamp, blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level)
 257
 258                     if reason != '':
 259                         # NOISY-DEBUG: print("DEBUG: Updating block reason:", blocker, blocked, reason)
 260                         fba.update_block_reason(reason, blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level)
 261
 262             fba.conn.commit()
 263         except Exception as e:
 264             print("error:", e, blocker, software)
 265     elif software == "friendica" or software == "misskey":
 266         print("INFO: blocker:", blocker)
 267         try:
 268             if software == "friendica":
 269                 json = fba.get_friendica_blocks(blocker)
 270             elif software == "misskey":
 271                 json = fba.get_misskey_blocks(blocker)
 272
 273             for block_level, blocks in json.items():
 274                 # NOISY-DEBUG: print("DEBUG: blocker,block_level,blocks():", blocker, block_level, len(blocks))
 275                 block_level = fba.tidyup(block_level)
 276                 # NOISY-DEBUG: print("DEBUG: AFTER-block_level:", block_level)
 277                 if block_level == "":
 278                     print("WARNING: block_level is empty, blocker:", blocker)
 279                     continue
 280
 281                 for instance in blocks:
 282                     blocked, reason = instance.values()
 283                     # NOISY-DEBUG: print("DEBUG: BEFORE-blocked:", blocked)
 284                     blocked = fba.tidyup(blocked)
 285                     # NOISY-DEBUG: print("DEBUG: AFTER-blocked:", blocked)
 286
 287                     if blocked == "":
 288                         print("WARNING: blocked is empty:", blocker)
 289                         continue
 290                     if blocked.count("*") > 0:
 291                         # Some friendica servers also obscure domains without hash
 292                         fba.c.execute(
 293                             "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", (blocked.replace("*", "_"),)
 294                         )
 295                         searchres = fba.c.fetchone()
 296                         if searchres != None:
 297                             blocked = searchres[0]
 298
 299                     if blocked.count("?") > 0:
 300                         # Some obscure them with question marks, not sure if that's dependent on version or not
 301                         fba.c.execute(
 302                             "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", (blocked.replace("?", "_"),)
 303                         )
 304                         searchres = fba.c.fetchone()
 305                         if searchres != None:
 306                             blocked = searchres[0]
 307
 308                     # NOISY-DEBUG: print("DEBUG: AFTER-blocked:", blocked)
 309                     fba.c.execute(
 310                         "SELECT domain FROM instances WHERE domain = ?", (blocked,)
 311                     )
 312
 313                     if fba.c.fetchone() == None:
 314                         # NOISY-DEBUG: print("DEBUG: Hash wasn't found, adding:", blocked)
 315                         fba.add_instance(blocked)
 316
 317                     timestamp = int(time.time())
 318                     fba.c.execute(
 319                         "SELECT * FROM blocks WHERE blocker = ? AND blocked = ?",
 320                         (blocker, blocked),
 321                     )
 322                     if fba.c.fetchone() == None:
 323                         fba.block_instance(blocker, blocked, reason, block_level, timestamp, timestamp)
 324
 325                         if block_level == "reject":
 326                             blockdict.append(
 327                                 {
 328                                     "blocked": blocked,
 329                                     "reason": reason
 330                                 })
 331                     else:
 332                         fba.update_last_seen(timestamp, blocker, blocked, block_level)
 333
 334                     if reason != '':
 335                         # NOISY-DEBUG: print("DEBUG: Updating block reason:", blocker, blocked, reason)
 336                         fba.update_block_reason(reason, blocker, blocked, block_level)
 337
 338             fba.conn.commit()
 339         except Exception as e:
 340             print("error:", e, blocker, software)
 341     elif software == "gotosocial":
 342         print("INFO: blocker:", blocker)
 343         try:
 344             # Blocks
 345             federation = reqto.get(
 346                 f"https://{blocker}/api/v1/instance/peers?filter=suspended", headers=fba.headers, timeout=5
 347             ).json()
 348
 349             if (federation == None):
 350                 print("WARNING: No valid response:", blocker);
 351             else:
 352                 for peer in federation:
 353                     # NOISY-DEBUG: print("DEBUG: peer(),[]:", len(peer), type(peer))
 354                     if (isinstance(peer, str) and peer == "error"):
 355                         print("WARNING: Cannot continue, maybe authentication required?", blocker)
 356                         break
 357
 358                     blocked = peer["domain"].lower()
 359                     # NOISY-DEBUG: print("DEBUG: BEFORE-blocked:", blocked)
 360                     blocked = fba.tidyup(blocked)
 361                     # NOISY-DEBUG: print("DEBUG: AFTER-blocked:", blocked)
 362
 363                     if blocked == "":
 364                         print("WARNING: blocked is empty:", blocker)
 365                         continue
 366                     if blocked.count("*") > 0:
 367                         # GTS does not have hashes for obscured domains, so we have to guess it
 368                         fba.c.execute(
 369                             "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", (blocked.replace("*", "_"),)
 370                         )
 371                         searchres = fba.c.fetchone()
 372
 373                         if searchres != None:
 374                             blocked = searchres[0]
 375
 376                     fba.c.execute(
 377                         "SELECT domain FROM instances WHERE domain = ?", (blocked,)
 378                     )
 379
 380                     if fba.c.fetchone() == None:
 381                         # NOISY-DEBUG: print("DEBUG: Hash wasn't found, adding:", blocked)
 382                         fba.add_instance(blocked)
 383
 384                     fba.c.execute(
 385                         "SELECT * FROM blocks WHERE blocker = ? AND blocked = ? AND block_level = ?",
 386                         (
 387                             blocker,
 388                             blocked,
 389                             "reject"
 390                         ),
 391                     )
 392                     timestamp = int(time.time())
 393
 394                     if fba.c.fetchone() == None:
 395                         fba.block_instance(blocker, blocked, "unknown", "reject", timestamp, timestamp)
 396
 397                         blockdict.append(
 398                             {
 399                                 "blocked": blocked,
 400                                 "reason": None
 401                             })
 402                     else:
 403                         fba.update_last_seen(timestamp, blocker, blocked, "reject")
 404
 405                     if "public_comment" in peer:
 406                         reason = peer["public_comment"]
 407                         # NOISY-DEBUG: print("DEBUG: Updating block reason:", blocker, blocked, reason)
 408                         fba.update_block_reason(reason, blocker, blocked, "reject")
 409
 410                         for entry in blockdict:
 411                             if entry["blocked"] == blocked:
 412                                 entry["reason"] = reason
 413
 414                 fba.conn.commit()
 415         except Exception as e:
 416             print("error:", e, blocker, software)
 417     else:
 418         print("WARNING: Unknown software:", software)
 419
 420     if fba.config["bot_enabled"] and len(blockdict) > 0:
 421         send_bot_post(blocker, blockdict)
 422
 423     blockdict = []
 424
 425 fba.conn.close()