fetch_blocks.py

   1 import time
   2 import bs4
   3 import fba
   4 import itertools
   5 import re
   6
   7 fba.c.execute(
   8     "SELECT domain, software FROM instances WHERE software IN ('pleroma', 'mastodon', 'friendica', 'misskey', 'gotosocial')"
   9 )
  10
  11 for blocker, software in fba.c.fetchall():
  12     print("DEBUG: blocker,software:", blocker, software)
  13     blockdict = []
  14     blocker = fba.tidyup(blocker)
  15     if software == "pleroma":
  16         print("DEBUG: blocker:", blocker)
  17         try:
  18             # Blocks
  19             federation = reqto.get(
  20                 f"https://{blocker}/nodeinfo/2.1.json", headers=headers, timeout=5
  21             ).json()["metadata"]["federation"]
  22             if "mrf_simple" in federation:
  23                 for block_level, blocks in (
  24                     {**federation["mrf_simple"],
  25                     **{"quarantined_instances": federation["quarantined_instances"]}}
  26                 ).items():
  27                     for blocked in blocks:
  28                         blocked = fba.tidyup(blocked)
  29
  30                         if blocked == "":
  31                             print("WARNING: blocked is empty after fba.tidyup():", blocker, block_level)
  32                             continue
  33
  34                         if blocked.count("*") > 1:
  35                             # -ACK!-oma also started obscuring domains without hash
  36                             fba.c.execute(
  37                                 "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", (blocked.replace("*", "_"),)
  38                             )
  39                             searchres = fba.c.fetchone()
  40                             if searchres != None:
  41                                 blocked = searchres[0]
  42
  43                         fba.c.execute(
  44                             "SELECT domain FROM instances WHERE domain = ?", (blocked)
  45                         )
  46
  47                         if fba.c.fetchone() == None:
  48                             print("DEBUG: Hash wasn't found, adding:", blocked)
  49                             fba.add_instance(blocked)
  50
  51                         timestamp = int(time.time())
  52                         fba.c.execute(
  53                             "SELECT * FROM blocks WHERE blocker = ? AND blocked = ? AND block_level = ?",
  54                             (
  55                                blocker,
  56                                blocked,
  57                                block_level
  58                            ),
  59                         )
  60                         if fba.c.fetchone() == None:
  61                             fba.block_instance(blocker, blocked, reason, block_level, timestamp, timestamp)
  62
  63                             if block_level == "reject":
  64                                 blockdict.append(
  65                                     {
  66                                         "blocked": blocked,
  67                                         "reason": None
  68                                     })
  69                         else:
  70                             fba.update_last_seen(timestamp, blocker, blocked, block_level)
  71
  72             fba.conn.commit()
  73             # Reasons
  74             if "mrf_simple_info" in federation:
  75                 for block_level, info in (
  76                     {**federation["mrf_simple_info"],
  77                     **(federation["quarantined_instances_info"]
  78                     if "quarantined_instances_info" in federation
  79                     else {})}
  80                 ).items():
  81                     for blocked, reason in info.items():
  82                         blocked = fba.tidyup(blocked)
  83
  84                         if blocked == "":
  85                             print("WARNING: blocked is empty after fba.tidyup():", blocker, block_level)
  86                             continue
  87
  88                         if blocked.count("*") > 1:
  89                             # same domain guess as above, but for reasons field
  90                             fba.c.execute(
  91                                 "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", (blocked.replace("*", "_"),)
  92                             )
  93                             searchres = fba.c.fetchone()
  94
  95                             if searchres != None:
  96                                 blocked = searchres[0]
  97
  98                         fba.update_block_reason(reason["reason"], blocker, blocked, block_level)
  99
 100                         for entry in blockdict:
 101                             if entry["blocked"] == blocked:
 102                                 entry["reason"] = reason["reason"]
 103
 104             fba.conn.commit()
 105         except Exception as e:
 106             print("error:", e, blocker, software)
 107     elif software == "mastodon":
 108         print("DEBUG: blocker:", blocker)
 109         try:
 110             # json endpoint for newer mastodongs
 111             try:
 112                 json = {
 113                     "reject": [],
 114                     "media_removal": [],
 115                     "followers_only": [],
 116                     "report_removal": []
 117                 }
 118
 119                 # handling CSRF, I've saw at least one server requiring it to access the endpoint
 120                 meta = bs4.BeautifulSoup(
 121                     reqto.get(f"https://{blocker}/about", headers=headers, timeout=5).text,
 122                     "html.parser",
 123                 )
 124                 try:
 125                     csrf = meta.find("meta", attrs={"name": "csrf-token"})["content"]
 126                     reqheaders = {**headers, **{"x-csrf-token": csrf}}
 127                 except:
 128                     reqheaders = headers
 129
 130                 blocks = reqto.get(
 131                     f"https://{blocker}/api/v1/instance/domain_blocks", headers=reqheaders, timeout=5
 132                 ).json()
 133
 134                 print("DEBUG: blocks():", len(blocks))
 135                 for block in blocks:
 136                     entry = {'domain': block['domain'], 'hash': block['digest'], 'reason': block['comment']}
 137
 138                     print("DEBUG: severity,domain,hash,comment:", block['severity'], block['domain'], block['digest'], block['comment'])
 139                     if block['severity'] == 'suspend':
 140                         json['reject'].append(entry)
 141                     elif block['severity'] == 'silence':
 142                         json['followers_only'].append(entry)
 143                     elif block['severity'] == 'reject_media':
 144                         json['media_removal'].append(entry)
 145                     elif block['severity'] == 'reject_reports':
 146                         json['report_removal'].append(entry)
 147                     else:
 148                         print("WARNING: Unknown severity:", block['severity'], block['domain'])
 149             except:
 150                 json = fba.get_mastodon_blocks(blocker)
 151
 152             for block_level, blocks in json.items():
 153                 for instance in blocks:
 154                     blocked, blocked_hash, reason = instance.values()
 155                     print("DEBUG: blocked,hash,reason:", blocked, blocked_hash, reason)
 156
 157                     blocked = fba.tidyup(blocked)
 158                     print("DEBUG: blocked:", blocked)
 159
 160                     if blocked.count("*") < 1:
 161                         # No obsfucation for this instance
 162                         fba.c.execute(
 163                             "SELECT hash FROM instances WHERE domain = ? LIMIT 1", (blocked,)
 164                         )
 165
 166                         if fba.c.fetchone() == None:
 167                             print("DEBUG: Hash wasn't found, adding:", blocked)
 168                             fba.add_instance(blocked)
 169                     else:
 170                         # Doing the hash search for instance names as well to tidy up DB
 171                         fba.c.execute(
 172                             "SELECT domain FROM instances WHERE hash = ? LIMIT 1", (blocked_hash,)
 173                         )
 174                         searchres = fba.c.fetchone()
 175
 176                         if searchres != None:
 177                             print("DEBUG: Updating domain: ", searchres[0])
 178                             blocked = searchres[0]
 179
 180                     timestamp = int(time.time())
 181                     fba.c.execute(
 182                         "SELECT * FROM blocks WHERE blocker = ? AND blocked = ? AND block_level = ?",
 183                         (
 184                             blocker,
 185                             blocked if blocked.count("*") <= 1 else blocked_hash,
 186                             block_level
 187                         ),
 188                     )
 189
 190                     if fba.c.fetchone() == None:
 191                         fba.block_instance(blocker, blocked if blocked.count("*") <= 1 else blocked_hash, reason, block_level, timestamp, timestamp)
 192
 193                         if block_level == "reject":
 194                             blockdict.append(
 195                                 {
 196                                     "blocked": blocked,
 197                                     "reason": reason
 198                                 })
 199                     else:
 200                         fba.update_last_seen(timestamp, blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level)
 201
 202                     if reason != '':
 203                         fba.update_block_reason(reason, blocker, blocked if blocked.count("*") <= 1 else blocked_hash, block_level)
 204
 205             fba.conn.commit()
 206         except Exception as e:
 207             print("error:", e, blocker, software)
 208     elif software == "friendica" or software == "misskey":
 209         print("DEBUG: blocker:", blocker)
 210         try:
 211             if software == "friendica":
 212                 json = fba.get_friendica_blocks(blocker)
 213             elif software == "misskey":
 214                 json = fba.get_misskey_blocks(blocker)
 215             for block_level, blocks in json.items():
 216                 for instance in blocks:
 217                     blocked, reason = instance.values()
 218                     blocked = fba.tidyup(blocked)
 219
 220                     print("BEFORE-blocked:", blocked)
 221                     if blocked.count("*") > 0:
 222                         # Some friendica servers also obscure domains without hash
 223                         fba.c.execute(
 224                             "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", (blocked.replace("*", "_"),)
 225                         )
 226                         searchres = fba.c.fetchone()
 227                         if searchres != None:
 228                             blocked = searchres[0]
 229
 230                     if blocked.count("?") > 0:
 231                         # Some obscure them with question marks, not sure if that's dependent on version or not
 232                         fba.c.execute(
 233                             "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", (blocked.replace("?", "_"),)
 234                         )
 235                         searchres = fba.c.fetchone()
 236                         if searchres != None:
 237                             blocked = searchres[0]
 238
 239                     print("AFTER-blocked:", blocked)
 240                     fba.c.execute(
 241                         "SELECT domain FROM instances WHERE domain = ?", (blocked,)
 242                     )
 243
 244                     if fba.c.fetchone() == None:
 245                         print("DEBUG: Hash wasn't found, adding:", blocked)
 246                         fba.add_instance(blocked)
 247
 248                     timestamp = int(time.time())
 249                     fba.c.execute(
 250                         "SELECT * FROM blocks WHERE blocker = ? AND blocked = ?",
 251                         (blocker, blocked),
 252                     )
 253                     if fba.c.fetchone() == None:
 254                         fba.block_instance(blocker, blocked, reason, block_level, timestamp, timestamp)
 255
 256                         if block_level == "reject":
 257                             blockdict.append(
 258                                 {
 259                                     "blocked": blocked,
 260                                     "reason": reason
 261                                 })
 262                     else:
 263                         fba.update_last_seen(timestamp, blocker, blocked, block_level)
 264
 265                     if reason != '':
 266                         fba.update_block_reason(reason, blocker, blocked, block_level)
 267
 268             fba.conn.commit()
 269         except Exception as e:
 270             print("error:", e, blocker, software)
 271     elif software == "gotosocial":
 272         print("DEBUG: blocker:", blocker)
 273         try:
 274             # Blocks
 275             federation = reqto.get(
 276                 f"https://{blocker}/api/v1/instance/peers?filter=suspended", headers=headers, timeout=5
 277             ).json()
 278
 279             if (federation == None):
 280                 print("WARNING: No valid response:", blocker);
 281             else:
 282                 for peer in federation:
 283                     blocked = peer["domain"].lower()
 284
 285                     if blocked.count("*") > 0:
 286                         # GTS does not have hashes for obscured domains, so we have to guess it
 287                         fba.c.execute(
 288                             "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", (blocked.replace("*", "_"),)
 289                         )
 290                         searchres = fba.c.fetchone()
 291
 292                         if searchres != None:
 293                             blocked = searchres[0]
 294
 295                     fba.c.execute(
 296                         "SELECT domain FROM instances WHERE domain = ?", (blocked,)
 297                     )
 298
 299                     if fba.c.fetchone() == None:
 300                         print("DEBUG: Hash wasn't found, adding:", blocked)
 301                         fba.add_instance(blocked)
 302
 303                     fba.c.execute(
 304                         "SELECT * FROM blocks WHERE blocker = ? AND blocked = ? AND block_level = ?",
 305                         (
 306                             blocker,
 307                             blocked,
 308                             "reject"
 309                         ),
 310                     )
 311                     timestamp = int(time.time())
 312
 313                     if fba.c.fetchone() == None:
 314                         fba.block_instance(blocker, blocked, "", "reject", timestamp, timestamp)
 315
 316                         blockdict.append(
 317                             {
 318                                 "blocked": blocked,
 319                                 "reason": None
 320                             })
 321                     else:
 322                         fba.update_last_seen(timestamp, blocker, blocked, "reject")
 323
 324                     if "public_comment" in peer:
 325                         reason = peer["public_comment"]
 326                         fba.update_block_reason(reason, blocker, blocked, "reject")
 327
 328                         for entry in blockdict:
 329                             if entry["blocked"] == blocked:
 330                                 entry["reason"] = reason
 331                 fba.conn.commit()
 332         except Exception as e:
 333             print("error:", e, blocker, software)
 334     else:
 335         print("WARNING: Unknown software:", software)
 336
 337     if fba.config["bot_enabled"] and len(blockdict) > 0:
 338         send_bot_post(blocker, blockdict)
 339
 340     blockdict = []
 341
 342 fba.conn.close()