fba.py

   1 import bs4
   2 import hashlib
   3 import re
   4 import reqto
   5 import json
   6 import sqlite3
   7 import sys
   8 import time
   9 import validators
  10
  11 with open("config.json") as f:
  12     config = json.loads(f.read())
  13
  14 # Don't check these, known trolls/flooders/testing/developing
  15 blacklist = [
  16     # Floods network with fake nodes as "research" project
  17     "activitypub-troll.cf",
  18     # Similar troll
  19     "gab.best",
  20     # Similar troll
  21     "4chan.icu",
  22     # Flooder (?)
  23     "social.shrimpcam.pw",
  24     # Flooder (?)
  25     "mastotroll.netz.org",
  26     # Testing/developing installations
  27     "ngrok.io",
  28 ]
  29
  30 # Array with pending errors needed to be written to database
  31 pending_errors = {
  32 }
  33
  34 # "rel" identifiers (no real URLs)
  35 nodeinfo_identifier = [
  36     "http://nodeinfo.diaspora.software/ns/schema/2.1",
  37     "http://nodeinfo.diaspora.software/ns/schema/2.0",
  38     "http://nodeinfo.diaspora.software/ns/schema/1.1",
  39     "http://nodeinfo.diaspora.software/ns/schema/1.0",
  40 ]
  41
  42 # HTTP headers for requests
  43 headers = {
  44     "user-agent": config["useragent"],
  45 }
  46
  47 # Found info from node, such as nodeinfo URL, detection mode that needs to be
  48 # written to database. Both arrays must be filled at the same time or else
  49 # update_nodeinfos() will fail
  50 nodeinfos = {
  51     # Detection mode: 'AUTO_DISCOVERY', 'STATIC_CHECKS' or 'GENERATOR'
  52     # NULL means all detection methods have failed (maybe still reachable instance)
  53     "detection_mode": {},
  54     # Found nodeinfo URL
  55     "nodeinfo_url": {},
  56     # Where to fetch peers (other instances)
  57     "get_peers_url": {},
  58 }
  59
  60 language_mapping = {
  61     # English -> English
  62     "Silenced instances"            : "Silenced servers",
  63     "Suspended instances"           : "Suspended servers",
  64     "Limited instances"             : "Limited servers",
  65     # Mappuing German -> English
  66     "Gesperrte Server"              : "Suspended servers",
  67     "Gefilterte Medien"             : "Filtered media",
  68     "Stummgeschaltete Server"       : "Silenced servers",
  69     # Japanese -> English
  70     "停止済みのサーバー"            : "Suspended servers",
  71     "制限中のサーバー"              : "Limited servers",
  72     "メディアを拒否しているサーバー": "Filtered media",
  73     "サイレンス済みのサーバー"      : "Silenced servers",
  74     # ??? -> English
  75     "שרתים מושעים"                  : "Suspended servers",
  76     "מדיה מסוננת"                   : "Filtered media",
  77     "שרתים מוגבלים"                 : "Silenced servers",
  78     # French -> English
  79     "Serveurs suspendus"            : "Suspended servers",
  80     "Médias filtrés"                : "Filtered media",
  81     "Serveurs limités"              : "Limited servers",
  82     "Serveurs modérés"              : "Limited servers",
  83 }
  84
  85 # URL for fetching peers
  86 get_peers_url = "/api/v1/instance/peers"
  87
  88 # Connect to database
  89 connection = sqlite3.connect("blocks.db")
  90 cursor = connection.cursor()
  91
  92 # Pattern instance for version numbers
  93 patterns = [
  94     # semantic version number (with v|V) prefix)
  95     re.compile("^(?P<version>v|V{0,1})(\.{0,1})(?P<major>0|[1-9]\d*)\.(?P<minor>0|[1-9]\d*)(\.(?P<patch>0|[1-9]\d*)(?:-(?P<prerelease>(?:0|[1-9]\d*|\d*[a-zA-Z-][0-9a-zA-Z-]*)(?:\.(?:0|[1-9]\d*|\d*[a-zA-Z-][0-9a-zA-Z-]*))*))?(?:\+(?P<buildmetadata>[0-9a-zA-Z-]+(?:\.[0-9a-zA-Z-]+)*))?)?$"),
  96     # non-sematic, e.g. 1.2.3.4
  97     re.compile("^(?P<version>v|V{0,1})(\.{0,1})(?P<major>0|[1-9]\d*)\.(?P<minor>0|[1-9]\d*)(\.(?P<patch>0|[1-9]\d*)(\.(?P<subpatch>0|[1-9]\d*))?)$"),
  98     # non-sematic, e.g. 2023-05
  99     re.compile("^(?P<year>[1-9]{1}[0-9]{3})\.(?P<month>[0-9]{2})$")
 100 ]
 101
 102 def remove_version(software: str) -> str:
 103     # NOISY-DEBUG: print(f"DEBUG: software='{software}' - CALLED!")
 104     if not "." in software:
 105         print(f"WARNING: software='{software}' does not contain a version number.")
 106         raise
 107
 108     version = None
 109     if " " in software:
 110         version = software.split(" ")[-1]
 111     elif "/" in software:
 112         version = software.split("/")[-1]
 113     elif "-" in software:
 114         version = software.split("-")[-1]
 115     else:
 116         # NOISY-DEBUG: print(f"DEBUG: Was not able to find common seperator, returning untouched software='{software}'")
 117         return software
 118
 119     matches = None
 120     # NOISY-DEBUG: print(f"DEBUG: Checking {len(patterns)} patterns ...")
 121     for pattern in patterns:
 122         # Run match()
 123         match = pattern.match(version)
 124
 125         # NOISY-DEBUG: print(f"DEBUG: match[]={type(match)}")
 126         if type(match) is re.Match:
 127             break
 128
 129     # NOISY-DEBUG: print(f"DEBUG: version[{type(version)}]='{version}',match='{match}'")
 130     if type(match) is not re.Match:
 131         print(f"WARNING: version='{version}' does not match regex, leaving software='{software}' untouched.")
 132         return software
 133
 134     # NOISY-DEBUG: print(f"DEBUG: Found valid version number: '{version}', removing it ...")
 135     end = len(software) - len(version)
 136
 137     # NOISY-DEBUG: print(f"DEBUG: end[{type(end)}]={end}")
 138     software = software[0:end].strip()
 139
 140     # NOISY-DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
 141     return software
 142
 143 def strip_powered_by(software: str) -> str:
 144     # NOISY-DEBUG: print(f"DEBUG: software='{software}' - CALLED!")
 145     if not "powered by" in software:
 146         print(f"WARNING: Cannot find 'powered by' in '{software}'!")
 147         return software
 148
 149     start = software.find("powered by ")
 150     # NOISY-DEBUG: print(f"DEBUG: start[{type(start)}]='{start}'")
 151
 152     software = software[start + 11:].strip()
 153     # NOISY-DEBUG: print(f"DEBUG: software='{software}'")
 154
 155     # Next, strip of ' - ' part
 156     end = software.find(" - ")
 157     # NOISY-DEBUG: print(f"DEBUG: end[{type(end)}]='{end}'")
 158     if end > 0:
 159         software = software[0:end].strip()
 160
 161     # NOISY-DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
 162     return software
 163
 164 def is_blacklisted(domain: str) -> bool:
 165     blacklisted = False
 166     for peer in blacklist:
 167         if peer in domain:
 168             blacklisted = True
 169
 170     return blacklisted
 171
 172 def remove_pending_error(domain: str):
 173     try:
 174         # Prevent updating any pending errors, nodeinfo was found
 175         del pending_errors[domain]
 176
 177     except:
 178         pass
 179
 180 def get_hash(domain: str) -> str:
 181     return hashlib.sha256(domain.encode("utf-8")).hexdigest()
 182
 183 def update_last_blocked(domain: str):
 184     # NOISY-DEBUG: print("DEBUG: Updating last_blocked for domain", domain)
 185     try:
 186         cursor.execute("UPDATE instances SET last_blocked = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
 187             time.time(),
 188             time.time(),
 189             domain
 190         ])
 191
 192         if cursor.rowcount == 0:
 193             print("WARNING: Did not update any rows:", domain)
 194
 195     except BaseException as e:
 196         print("ERROR: failed SQL query:", domain, e)
 197         sys.exit(255)
 198
 199     # NOISY-DEBUG: print("DEBUG: EXIT!")
 200
 201 def update_nodeinfos(domain: str):
 202     # NOISY-DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain)
 203     sql_string = ''
 204     fields = list()
 205     for key in nodeinfos:
 206         # NOISY-DEBUG: print("DEBUG: key:", key)
 207         if domain in nodeinfos[key]:
 208            # NOISY-DEBUG: print(f"DEBUG: Adding '{nodeinfos[key][domain]}' for key='{key}' ...")
 209            fields.append(nodeinfos[key][domain])
 210            sql_string += f" {key} = ?,"
 211
 212     fields.append(domain)
 213     # NOISY-DEBUG: print(f"DEBUG: sql_string='{sql_string}',fields()={len(fields)}")
 214
 215     sql = "UPDATE instances SET" + sql_string + " last_status_code = NULL, last_error_details = NULL WHERE domain = ? LIMIT 1"
 216     # NOISY-DEBUG: print("DEBUG: sql:", sql)
 217
 218     try:
 219         # NOISY-DEBUG: print("DEBUG: Executing SQL:", sql)
 220         cursor.execute(sql, fields)
 221         # NOISY-DEBUG: print(f"DEBUG: Success! (rowcount={cursor.rowcount })")
 222
 223         if cursor.rowcount == 0:
 224             print("WARNING: Did not update any rows:", domain)
 225
 226     except BaseException as e:
 227         print(f"ERROR: failed SQL query: domain='{domain}',sql='{sql}',exception:'{e}'")
 228         sys.exit(255)
 229
 230     # NOISY-DEBUG: print("DEBUG: Deleting nodeinfos for domain:", domain)
 231     for key in nodeinfos:
 232         try:
 233             # NOISY-DEBUG: print("DEBUG: Deleting key:", key)
 234             del nodeinfos[key][domain]
 235         except:
 236             pass
 237
 238     # NOISY-DEBUG: print("DEBUG: EXIT!")
 239
 240 def update_last_error(domain: str, res: any):
 241     # NOISY-DEBUG: print("DEBUG: domain,res[]:", domain, type(res))
 242     try:
 243         # NOISY-DEBUG: print("DEBUG: BEFORE res[]:", type(res))
 244         if isinstance(res, BaseException) or isinstance(res, json.JSONDecodeError):
 245             res = str(res)
 246
 247         # NOISY-DEBUG: print("DEBUG: AFTER res[]:", type(res))
 248         if type(res) is str:
 249             # NOISY-DEBUG: print(f"DEBUG: Setting last_error_details='{res}'");
 250             cursor.execute("UPDATE instances SET last_status_code = 999, last_error_details = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
 251                 res,
 252                 time.time(),
 253                 domain
 254             ])
 255         else:
 256             # NOISY-DEBUG: print(f"DEBUG: Setting last_error_details='{res.reason}'");
 257             cursor.execute("UPDATE instances SET last_status_code = ?, last_error_details = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
 258                 res.status_code,
 259                 res.reason,
 260                 time.time(),
 261                 domain
 262             ])
 263
 264         if cursor.rowcount == 0:
 265             # NOISY-DEBUG: print("DEBUG: Did not update any rows:", domain)
 266             pending_errors[domain] = res
 267
 268     except BaseException as e:
 269         print("ERROR: failed SQL query:", domain, e)
 270         sys.exit(255)
 271
 272     # NOISY-DEBUG: print("DEBUG: EXIT!")
 273
 274 def update_last_nodeinfo(domain: str):
 275     # NOISY-DEBUG: print("DEBUG: Updating last_nodeinfo for domain:", domain)
 276     try:
 277         cursor.execute("UPDATE instances SET last_nodeinfo = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
 278             time.time(),
 279             time.time(),
 280             domain
 281         ])
 282
 283         if cursor.rowcount == 0:
 284             print("WARNING: Did not update any rows:", domain)
 285
 286     except BaseException as e:
 287         print("ERROR: failed SQL query:", domain, e)
 288         sys.exit(255)
 289
 290     connection.commit()
 291     # NOISY-DEBUG: print("DEBUG: EXIT!")
 292
 293 def get_peers(domain: str, software: str) -> list:
 294     # NOISY-DEBUG: print("DEBUG: Getting peers for domain:", domain, software)
 295     peers = list()
 296
 297     if software == "lemmy":
 298         # NOISY-DEBUG: print(f"DEBUG: domain='{domain}' is Lemmy, fetching JSON ...")
 299         try:
 300             res = reqto.get(f"https://{domain}/api/v3/site", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
 301
 302             # NOISY-DEBUG: print(f"DEBUG: res.ok={res.ok},res.json[]={type(res.json())}")
 303             if res.ok and isinstance(res.json(), dict):
 304                 # NOISY-DEBUG: print("DEBUG: Success, res.json():", len(res.json()))
 305                 json = res.json()
 306
 307                 if "federated_instances" in json and "linked" in json["federated_instances"]:
 308                     # NOISY-DEBUG: print("DEBUG: Found federated_instances", domain)
 309                     peers = json["federated_instances"]["linked"] + json["federated_instances"]["allowed"] + json["federated_instances"]["blocked"]
 310
 311         except BaseException as e:
 312             print("WARNING: Exception during fetching JSON:", domain, e)
 313
 314         update_last_nodeinfo(domain)
 315
 316         # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers))
 317         return peers
 318     elif software == "peertube":
 319         # NOISY-DEBUG: print(f"DEBUG: domain='{domain}' is a PeerTube, fetching JSON ...")
 320
 321         start = 0
 322         for mode in ["followers", "following"]:
 323             # NOISY-DEBUG: print(f"DEBUG: domain='{domain}',mode='{mode}'")
 324             while True:
 325                 try:
 326                     res = reqto.get(f"https://{domain}/api/v1/server/{mode}?start={start}&count=100", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
 327
 328                     # NOISY-DEBUG: print(f"DEBUG: res.ok={res.ok},res.json[]={type(res.json())}")
 329                     if res.ok and isinstance(res.json(), dict):
 330                         # NOISY-DEBUG: print("DEBUG: Success, res.json():", len(res.json()))
 331                         json = res.json()
 332
 333                         if "data" in json:
 334                             # NOISY-DEBUG: print(f"DEBUG: Found {len(json['data'])} record(s).")
 335                             for record in json["data"]:
 336                                 # NOISY-DEBUG: print(f"DEBUG: record()={len(record)}")
 337                                 if mode in record and "host" in record[mode]:
 338                                     # NOISY-DEBUG: print(f"DEBUG: Found host={record[mode]['host']}, adding ...")
 339                                     peers.append(record[mode]["host"])
 340                                 else:
 341                                     print(f"WARNING: record from '{domain}' has no '{mode}' or 'host' record: {record}")
 342
 343                             if len(json["data"]) < 100:
 344                                 # NOISY-DEBUG: print("DEBUG: Reached end of JSON response:", domain)
 345                                 break
 346
 347                         # Continue with next row
 348                         start = start + 100
 349
 350                 except BaseException as e:
 351                     print("WARNING: Exception during fetching JSON:", domain, e)
 352
 353             update_last_nodeinfo(domain)
 354
 355             # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers))
 356             return peers
 357
 358     # NOISY-DEBUG: print(f"DEBUG: Fetching '{get_peers_url}' from '{domain}' ...")
 359     try:
 360         res = reqto.get(f"https://{domain}{get_peers_url}", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
 361
 362         # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json()))
 363         if not res.ok or res.status_code >= 400:
 364             res = reqto.get(f"https://{domain}/api/v3/site", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
 365
 366             if "federated_instances" in json and "linked" in json["federated_instances"]:
 367                 # NOISY-DEBUG: print("DEBUG: Found federated_instances", domain)
 368                 peers = json["federated_instances"]["linked"] + json["federated_instances"]["allowed"] + json["federated_instances"]["blocked"]
 369             else:
 370                 print("WARNING: Could not reach any JSON API:", domain)
 371                 update_last_error(domain, res)
 372         else:
 373             # NOISY-DEBUG: print("DEBUG:Querying API was successful:", domain, len(res.json()))
 374             peers = res.json()
 375             nodeinfos["get_peers_url"][domain] = get_peers_url
 376
 377     except BaseException as e:
 378         print("WARNING: Some error during get():", domain, e)
 379         update_last_error(domain, e)
 380
 381     update_last_nodeinfo(domain)
 382
 383     # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers))
 384     return peers
 385
 386 def post_json_api(domain: str, path: str, data: str) -> list:
 387     # NOISY-DEBUG: print("DEBUG: Sending POST to domain,path,data:", domain, path, data)
 388     json = {}
 389     try:
 390         res = reqto.post(f"https://{domain}{path}", data=data, headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
 391
 392         # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json()))
 393         if not res.ok or res.status_code >= 400:
 394             print("WARNING: Cannot query JSON API:", domain, path, data, res.status_code)
 395             update_last_error(domain, res)
 396             raise
 397
 398         update_last_nodeinfo(domain)
 399         json = res.json()
 400     except BaseException as e:
 401         print("WARNING: Some error during post():", domain, path, data, e)
 402
 403     # NOISY-DEBUG: print("DEBUG: Returning json():", len(json))
 404     return json
 405
 406 def fetch_nodeinfo(domain: str) -> list:
 407     # NOISY-DEBUG: print("DEBUG: Fetching nodeinfo from domain:", domain)
 408
 409     nodeinfo = fetch_wellknown_nodeinfo(domain)
 410     # NOISY-DEBUG: print("DEBUG:nodeinfo:", len(nodeinfo))
 411
 412     if len(nodeinfo) > 0:
 413         # NOISY-DEBUG: print("DEBUG: Returning auto-discovered nodeinfo:", len(nodeinfo))
 414         return nodeinfo
 415
 416     requests = [
 417        f"https://{domain}/nodeinfo/2.1.json",
 418        f"https://{domain}/nodeinfo/2.1",
 419        f"https://{domain}/nodeinfo/2.0.json",
 420        f"https://{domain}/nodeinfo/2.0",
 421        f"https://{domain}/nodeinfo/1.0",
 422        f"https://{domain}/api/v1/instance"
 423     ]
 424
 425     json = {}
 426     for request in requests:
 427         try:
 428             # NOISY-DEBUG: print("DEBUG: Fetching request:", request)
 429             res = reqto.get(request, headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
 430
 431             # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json()))
 432             if res.ok and isinstance(res.json(), dict):
 433                 # NOISY-DEBUG: print("DEBUG: Success:", request)
 434                 json = res.json()
 435                 nodeinfos["detection_mode"][domain] = "STATIC_CHECK"
 436                 nodeinfos["nodeinfo_url"][domain] = request
 437                 break
 438             elif not res.ok or res.status_code >= 400:
 439                 print("WARNING: Failed fetching nodeinfo from domain:", domain)
 440                 update_last_error(domain, res)
 441                 continue
 442
 443         except BaseException as e:
 444             # NOISY-DEBUG: print("DEBUG: Cannot fetch API request:", request)
 445             update_last_error(domain, e)
 446             pass
 447
 448     # NOISY-DEBUG: print("DEBUG: Returning json[]:", type(json))
 449     return json
 450
 451 def fetch_wellknown_nodeinfo(domain: str) -> list:
 452     # NOISY-DEBUG: print("DEBUG: Fetching .well-known info for domain:", domain)
 453     json = {}
 454
 455     try:
 456         res = reqto.get(f"https://{domain}/.well-known/nodeinfo", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
 457         # NOISY-DEBUG: print("DEBUG: domain,res.ok,res.json[]:", domain, res.ok, type(res.json()))
 458         if res.ok and isinstance(res.json(), dict):
 459             nodeinfo = res.json()
 460             # NOISY-DEBUG: print("DEBUG: Found entries:", len(nodeinfo), domain)
 461             if "links" in nodeinfo:
 462                 # NOISY-DEBUG: print("DEBUG: Found links in nodeinfo():", len(nodeinfo["links"]))
 463                 for link in nodeinfo["links"]:
 464                     # NOISY-DEBUG: print("DEBUG: rel,href:", link["rel"], link["href"])
 465                     if link["rel"] in nodeinfo_identifier:
 466                         # NOISY-DEBUG: print("DEBUG: Fetching nodeinfo from:", link["href"])
 467                         res = reqto.get(link["href"])
 468
 469                         # NOISY-DEBUG: print("DEBUG: href,res.ok,res.status_code:", link["href"], res.ok, res.status_code)
 470                         if res.ok and isinstance(res.json(), dict):
 471                             # NOISY-DEBUG: print("DEBUG: Found JSON nodeinfo():", len(res.json()))
 472                             json = res.json()
 473                             nodeinfos["detection_mode"][domain] = "AUTO_DISCOVERY"
 474                             nodeinfos["nodeinfo_url"][domain] = link["href"]
 475                             break
 476                     else:
 477                         print("WARNING: Unknown 'rel' value:", domain, link["rel"])
 478             else:
 479                 print("WARNING: nodeinfo does not contain 'links':", domain)
 480
 481     except BaseException as e:
 482         print("WARNING: Failed fetching .well-known info:", domain)
 483         update_last_error(domain, e)
 484         pass
 485
 486     # NOISY-DEBUG: print("DEBUG: Returning json[]:", type(json))
 487     return json
 488
 489 def fetch_generator_from_path(domain: str, path: str = "/") -> str:
 490     # NOISY-DEBUG: print(f"DEBUG: domain='{domain}',path='{path}' - CALLED!")
 491     software = None
 492
 493     try:
 494         # NOISY-DEBUG: print(f"DEBUG: Fetching '{path}' from '{domain}' ...")
 495         res = reqto.get(f"https://{domain}{path}", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
 496
 497         # NOISY-DEBUG: print("DEBUG: domain,res.ok,res.status_code,res.text[]:", domain, res.ok, res.status_code, type(res.text))
 498         if res.ok and res.status_code < 300 and len(res.text) > 0:
 499             # NOISY-DEBUG: print("DEBUG: Search for <meta name='generator'>:", domain)
 500             doc = bs4.BeautifulSoup(res.text, "html.parser")
 501
 502             # NOISY-DEBUG: print("DEBUG: doc[]:", type(doc))
 503             tag = doc.find("meta", {"name": "generator"})
 504
 505             # NOISY-DEBUG: print(f"DEBUG: tag[{type(tag)}: {tag}")
 506             if isinstance(tag, bs4.element.Tag):
 507                 # NOISY-DEBUG: print("DEBUG: Found generator meta tag: ", domain)
 508                 software = tidyup(tag.get("content"))
 509                 print(f"INFO: domain='{domain}' is generated by '{software}'")
 510                 nodeinfos["detection_mode"][domain] = "GENERATOR"
 511                 remove_pending_error(domain)
 512
 513     except BaseException as e:
 514         # NOISY-DEBUG: print(f"DEBUG: Cannot fetch / from '{domain}':", e)
 515         update_last_error(domain, e)
 516         pass
 517
 518     # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}")
 519     if type(software) is str and software == "":
 520         # NOISY-DEBUG: print(f"DEBUG: Corrected empty string to None for software of domain='{domain}'")
 521         software = None
 522     elif type(software) is str and "." in software:
 523         # NOISY-DEBUG: print(f"DEBUG: software='{software}' may contain a version number, domain='{domain}', removing it ...")
 524         software = remove_version(software)
 525
 526     # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}")
 527     if type(software) is str and "powered by" in software:
 528         # NOISY-DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it")
 529         software = remove_version(strip_powered_by(software))
 530
 531     # NOISY-DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
 532     return software
 533
 534 def determine_software(domain: str) -> str:
 535     # NOISY-DEBUG: print("DEBUG: Determining software for domain:", domain)
 536     software = None
 537
 538     # NOISY-DEBUG: print(f"DEBUG: Fetching nodeinfo from '{domain}' ...")
 539     json = fetch_nodeinfo(domain)
 540
 541     # NOISY-DEBUG: print("DEBUG: json[]:", type(json))
 542     if not isinstance(json, dict) or len(json) == 0:
 543         # NOISY-DEBUG: print("DEBUG: Could not determine software type:", domain)
 544         return fetch_generator_from_path(domain)
 545
 546     # NOISY-DEBUG: print("DEBUG: json():", len(json), json)
 547     if "status" in json and json["status"] == "error" and "message" in json:
 548         print("WARNING: JSON response is an error:", json["message"])
 549         update_last_error(domain, json["message"])
 550         return fetch_generator_from_path(domain)
 551     elif "software" not in json or "name" not in json["software"]:
 552         # NOISY-DEBUG: print(f"DEBUG: JSON response from {domain} does not include [software][name], fetching / ...")
 553         software = fetch_generator_from_path(domain)
 554
 555         # NOISY-DEBUG: print(f"DEBUG: Generator for domain='{domain}' is: {software}, EXIT!")
 556         return software
 557
 558     software = tidyup(json["software"]["name"])
 559
 560     # NOISY-DEBUG: print("DEBUG: sofware after tidyup():", software)
 561     if software in ["akkoma", "rebased"]:
 562         # NOISY-DEBUG: print("DEBUG: Setting pleroma:", domain, software)
 563         software = "pleroma"
 564     elif software in ["hometown", "ecko"]:
 565         # NOISY-DEBUG: print("DEBUG: Setting mastodon:", domain, software)
 566         software = "mastodon"
 567     elif software in ["calckey", "groundpolis", "foundkey", "cherrypick", "meisskey"]:
 568         # NOISY-DEBUG: print("DEBUG: Setting misskey:", domain, software)
 569         software = "misskey"
 570     elif software.find("/") > 0:
 571         print("WARNING: Spliting of slash:", software)
 572         software = software.split("/")[-1];
 573     elif software.find("|") > 0:
 574         print("WARNING: Spliting of pipe:", software)
 575         software = tidyup(software.split("|")[0]);
 576     elif "powered by" in software:
 577         print(f"DEBUG: software='{software}' has 'powered by' in it")
 578         software = strip_powered_by(software)
 579
 580     # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}")
 581     if software == "":
 582         print("WARNING: tidyup() left no software name behind:", domain)
 583         software = None
 584
 585     # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}")
 586     if str(software) == "":
 587         # NOISY-DEBUG: print(f"DEBUG: software for '{domain}' was not detected, trying generator ...")
 588         software = fetch_generator_from_path(domain)
 589     elif len(str(software)) > 0 and "." in software:
 590         # NOISY-DEBUG: print(f"DEBUG: software='{software}' may contain a version number, domain='{domain}', removing it ...")
 591         software = remove_version(software)
 592
 593     # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}")
 594     if type(software) is str and "powered by" in software:
 595         # NOISY-DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it")
 596         software = remove_version(strip_powered_by(software))
 597
 598     # NOISY-DEBUG: print("DEBUG: Returning domain,software:", domain, software)
 599     return software
 600
 601 def update_block_reason(reason: str, blocker: str, blocked: str, block_level: str):
 602     # NOISY-DEBUG: print("DEBUG: Updating block reason:", reason, blocker, blocked, block_level)
 603     try:
 604         cursor.execute(
 605             "UPDATE blocks SET reason = ?, last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ? AND reason = ''",
 606             (
 607                 reason,
 608                 time.time(),
 609                 blocker,
 610                 blocked,
 611                 block_level
 612             ),
 613         )
 614
 615         # NOISY-DEBUG: print(f"DEBUG: cursor.rowcount={cursor.rowcount}")
 616         if cursor.rowcount == 0:
 617             print("WARNING: Did not update any rows:", domain)
 618
 619     except BaseException as e:
 620         print("ERROR: failed SQL query:", reason, blocker, blocked, block_level, e)
 621         sys.exit(255)
 622
 623     # NOISY-DEBUG: print("DEBUG: EXIT!")
 624
 625 def update_last_seen(blocker: str, blocked: str, block_level: str):
 626     # NOISY-DEBUG: print("DEBUG: Updating last_seen for:", blocker, blocked, block_level)
 627     try:
 628         cursor.execute(
 629             "UPDATE blocks SET last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ?",
 630             (
 631                 time.time(),
 632                 blocker,
 633                 blocked,
 634                 block_level
 635             )
 636         )
 637
 638         if cursor.rowcount == 0:
 639             print("WARNING: Did not update any rows:", domain)
 640
 641     except BaseException as e:
 642         print("ERROR: failed SQL query:", last_seen, blocker, blocked, block_level, e)
 643         sys.exit(255)
 644
 645     # NOISY-DEBUG: print("DEBUG: EXIT!")
 646
 647 def block_instance(blocker: str, blocked: str, reason: str, block_level: str):
 648     # NOISY-DEBUG: print("DEBUG: blocker,blocked,reason,block_level:", blocker, blocked, reason, block_level)
 649     if not validators.domain(blocker):
 650         print("WARNING: Bad blocker:", blocker)
 651         raise
 652     elif not validators.domain(blocked):
 653         print("WARNING: Bad blocked:", blocked)
 654         raise
 655
 656     print("INFO: New block:", blocker, blocked, reason, block_level, first_added, last_seen)
 657     try:
 658         cursor.execute(
 659             "INSERT INTO blocks (blocker, blocked, reason, block_level, first_added, last_seen) VALUES(?, ?, ?, ?, ?, ?)",
 660              (
 661                  blocker,
 662                  blocked,
 663                  reason,
 664                  block_level,
 665                  time.time(),
 666                  time.time()
 667              ),
 668         )
 669
 670     except BaseException as e:
 671         print("ERROR: failed SQL query:", blocker, blocked, reason, block_level, first_added, last_seen, e)
 672         sys.exit(255)
 673
 674     # NOISY-DEBUG: print("DEBUG: EXIT!")
 675
 676 def add_instance(domain: str, origin: str, originator: str):
 677     # NOISY-DEBUG: print("DEBUG: domain,origin:", domain, origin, originator)
 678     if not validators.domain(domain):
 679         print("WARNING: Bad domain name:", domain)
 680         raise
 681     elif origin is not None and not validators.domain(origin):
 682         print("WARNING: Bad origin name:", origin)
 683         raise
 684
 685     software = determine_software(domain)
 686     # NOISY-DEBUG: print("DEBUG: Determined software:", software)
 687
 688     print(f"INFO: Adding instance {domain} (origin: {origin})")
 689     try:
 690         cursor.execute(
 691             "INSERT INTO instances (domain, origin, originator, hash, software, first_seen) VALUES (?, ?, ?, ?, ?, ?)",
 692             (
 693                domain,
 694                origin,
 695                originator,
 696                get_hash(domain),
 697                software,
 698                time.time()
 699             ),
 700         )
 701
 702         for key in nodeinfos:
 703             # NOISY-DEBUG: pprint(f"DEBUG: key='{key}',domain='{domain}',nodeinfos[key]={nodeinfos[key]}")
 704             if domain in nodeinfos[key]:
 705                 # NOISY-DEBUG: pprint(f"DEBUG: domain='{domain}' has pending nodeinfo being updated ...")
 706                 update_nodeinfos(domain)
 707                 remove_pending_error(domain)
 708                 break
 709
 710         if domain in pending_errors:
 711             # NOISY-DEBUG: print("DEBUG: domain has pending error being updated:", domain)
 712             update_last_error(domain, pending_errors[domain])
 713             remove_pending_error(domain)
 714
 715     except BaseException as e:
 716         print("ERROR: failed SQL query:", domain, e)
 717         sys.exit(255)
 718     else:
 719         # NOISY-DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain)
 720         update_last_nodeinfo(domain)
 721
 722     # NOISY-DEBUG: print("DEBUG: EXIT!")
 723
 724 def send_bot_post(instance: str, blocks: dict):
 725     message = instance + " has blocked the following instances:\n\n"
 726     truncated = False
 727
 728     if len(blocks) > 20:
 729         truncated = True
 730         blocks = blocks[0 : 19]
 731
 732     for block in blocks:
 733         if block["reason"] == None or block["reason"] == '':
 734             message = message + block["blocked"] + " with unspecified reason\n"
 735         else:
 736             if len(block["reason"]) > 420:
 737                 block["reason"] = block["reason"][0:419] + "[…]"
 738
 739             message = message + block["blocked"] + ' for "' + block["reason"].replace("@", "@\u200b") + '"\n'
 740
 741     if truncated:
 742         message = message + "(the list has been truncated to the first 20 entries)"
 743
 744     botheaders = {**headers, **{"Authorization": "Bearer " + config["bot_token"]}}
 745
 746     req = reqto.post(
 747         f"{config['bot_instance']}/api/v1/statuses",
 748         data={
 749             "status"      : message,
 750             "visibility"  : config['bot_visibility'],
 751             "content_type": "text/plain"
 752         },
 753         headers=botheaders,
 754         timeout=10
 755     ).json()
 756
 757     return True
 758
 759 def get_mastodon_blocks(domain: str) -> dict:
 760     # NOISY-DEBUG: print("DEBUG: Fetching mastodon blocks from domain:", domain)
 761     blocks = {
 762         "Suspended servers": [],
 763         "Filtered media"   : [],
 764         "Limited servers"  : [],
 765         "Silenced servers" : [],
 766     }
 767
 768     try:
 769         doc = bs4.BeautifulSoup(
 770             reqto.get(f"https://{domain}/about/more", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
 771             "html.parser",
 772         )
 773     except BaseException as e:
 774         print("ERROR: Cannot fetch from domain:", domain, e)
 775         update_last_error(domain, e)
 776         return {}
 777
 778     for header in doc.find_all("h3"):
 779         header_text = tidyup(header.text)
 780
 781         if header_text in language_mapping:
 782             # NOISY-DEBUG: print(f"DEBUG: header_text='{header_text}'")
 783             header_text = language_mapping[header_text]
 784
 785         if header_text in blocks or header_text.lower() in blocks:
 786             # replaced find_next_siblings with find_all_next to account for instances that e.g. hide lists in dropdown menu
 787             for line in header.find_all_next("table")[0].find_all("tr")[1:]:
 788                 blocks[header_text].append(
 789                     {
 790                         "domain": tidyup(line.find("span").text),
 791                         "hash"  : tidyup(line.find("span")["title"][9:]),
 792                         "reason": tidyup(line.find_all("td")[1].text),
 793                     }
 794                 )
 795
 796     # NOISY-DEBUG: print("DEBUG: Returning blocks for domain:", domain)
 797     return {
 798         "reject"        : blocks["Suspended servers"],
 799         "media_removal" : blocks["Filtered media"],
 800         "followers_only": blocks["Limited servers"] + blocks["Silenced servers"],
 801     }
 802
 803 def get_friendica_blocks(domain: str) -> dict:
 804     # NOISY-DEBUG: print("DEBUG: Fetching friendica blocks from domain:", domain)
 805     blocks = []
 806
 807     try:
 808         doc = bs4.BeautifulSoup(
 809             reqto.get(f"https://{domain}/friendica", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
 810             "html.parser",
 811         )
 812     except BaseException as e:
 813         print("WARNING: Failed to fetch /friendica from domain:", domain, e)
 814         update_last_error(domain, e)
 815         return {}
 816
 817     blocklist = doc.find(id="about_blocklist")
 818
 819     # Prevents exceptions:
 820     if blocklist is None:
 821         # NOISY-DEBUG: print("DEBUG:Instance has no block list:", domain)
 822         return {}
 823
 824     for line in blocklist.find("table").find_all("tr")[1:]:
 825         blocks.append({
 826             "domain": tidyup(line.find_all("td")[0].text),
 827             "reason": tidyup(line.find_all("td")[1].text)
 828         })
 829
 830     # NOISY-DEBUG: print("DEBUG: Returning blocks() for domain:", domain, len(blocks))
 831     return {
 832         "reject": blocks
 833     }
 834
 835 def get_misskey_blocks(domain: str) -> dict:
 836     # NOISY-DEBUG: print("DEBUG: Fetching misskey blocks from domain:", domain)
 837     blocks = {
 838         "suspended": [],
 839         "blocked"  : []
 840     }
 841
 842     counter = 0
 843     step = 99
 844     while True:
 845         # iterating through all "suspended" (follow-only in its terminology)
 846         # instances page-by-page, since that troonware doesn't support
 847         # sending them all at once
 848         try:
 849             if counter == 0:
 850                 # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 851                 doc = post_json_api(domain, "/api/federation/instances/", json.dumps({
 852                     "sort"     : "+caughtAt",
 853                     "host"     : None,
 854                     "suspended": True,
 855                     "limit"    : step
 856                 }))
 857             else:
 858                 # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 859                 doc = post_json_api(domain, "/api/federation/instances/", json.dumps({
 860                     "sort"     : "+caughtAt",
 861                     "host"     : None,
 862                     "suspended": True,
 863                     "limit"    : step,
 864                     "offset"   : counter-1
 865                 }))
 866
 867             # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
 868             if len(doc) == 0:
 869                 # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
 870                 break
 871
 872             for instance in doc:
 873                 # just in case
 874                 if instance["isSuspended"]:
 875                     blocks["suspended"].append(
 876                         {
 877                             "domain": tidyup(instance["host"]),
 878                             # no reason field, nothing
 879                             "reason": ""
 880                         }
 881                     )
 882
 883             if len(doc) < step:
 884                 # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
 885                 break
 886
 887             # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
 888             counter = counter + step
 889
 890         except BaseException as e:
 891             print("WARNING: Caught error, exiting loop:", domain, e)
 892             update_last_error(domain, e)
 893             counter = 0
 894             break
 895
 896     while True:
 897         # same shit, different asshole ("blocked" aka full suspend)
 898         try:
 899             if counter == 0:
 900                 # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 901                 doc = post_json_api(domain,"/api/federation/instances", json.dumps({
 902                     "sort"   : "+caughtAt",
 903                     "host"   : None,
 904                     "blocked": True,
 905                     "limit"  : step
 906                 }))
 907             else:
 908                 # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
 909                 doc = post_json_api(domain,"/api/federation/instances", json.dumps({
 910                     "sort"   : "+caughtAt",
 911                     "host"   : None,
 912                     "blocked": True,
 913                     "limit"  : step,
 914                     "offset" : counter-1
 915                 }))
 916
 917             # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
 918             if len(doc) == 0:
 919                 # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
 920                 break
 921
 922             for instance in doc:
 923                 if instance["isBlocked"]:
 924                     blocks["blocked"].append({
 925                         "domain": tidyup(instance["host"]),
 926                         "reason": ""
 927                     })
 928
 929             if len(doc) < step:
 930                 # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
 931                 break
 932
 933             # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
 934             counter = counter + step
 935
 936         except BaseException as e:
 937             print("ERROR: Exception during POST:", domain, e)
 938             update_last_error(domain, e)
 939             counter = 0
 940             break
 941
 942     # NOISY-DEBUG: print("DEBUG: Returning for domain,blocked(),suspended():", domain, len(blocks["blocked"]), len(blocks["suspended"]))
 943     return {
 944         "reject"        : blocks["blocked"],
 945         "followers_only": blocks["suspended"]
 946     }
 947
 948 def tidyup(string: str) -> str:
 949     # some retards put their blocks in variable case
 950     string = string.lower().strip()
 951
 952     # other retards put the port
 953     string = re.sub("\:\d+$", "", string)
 954
 955     # bigger retards put the schema in their blocklist, sometimes even without slashes
 956     string = re.sub("^https?\:(\/*)", "", string)
 957
 958     # and trailing slash
 959     string = re.sub("\/$", "", string)
 960
 961     # and the @
 962     string = re.sub("^\@", "", string)
 963
 964     # the biggest retards of them all try to block individual users
 965     string = re.sub("(.+)\@", "", string)
 966
 967     return string