11 with open("config.json") as f:
12 config = json.loads(f.read())
14 # Don't check these, known trolls/flooders/testing/developing
16 # Floods network with fake nodes as "research" project
17 "activitypub-troll.cf",
23 "social.shrimpcam.pw",
25 "mastotroll.netz.org",
26 # Testing/developing installations
30 # Array with pending errors needed to be written to database
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",
42 # HTTP headers for requests
44 "user-agent": config["useragent"],
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
51 # Detection mode: 'AUTO_DISCOVERY', 'STATIC_CHECKS' or 'GENERATOR'
52 # NULL means all detection methods have failed (maybe still reachable instance)
56 # Where to fetch peers (other instances)
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",
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",
75 "שרתים מושעים" : "Suspended servers",
76 "מדיה מסוננת" : "Filtered media",
77 "שרתים מוגבלים" : "Silenced servers",
79 "Serveurs suspendus" : "Suspended servers",
80 "Médias filtrés" : "Filtered media",
81 "Serveurs limités" : "Limited servers",
82 "Serveurs modérés" : "Limited servers",
85 # URL for fetching peers
86 get_peers_url = "/api/v1/instance/peers"
89 connection = sqlite3.connect("blocks.db")
90 cursor = connection.cursor()
92 # Pattern instance for version numbers
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})$")
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.")
110 version = software.split(" ")[-1]
111 elif "/" in software:
112 version = software.split("/")[-1]
113 elif "-" in software:
114 version = software.split("-")[-1]
116 # NOISY-DEBUG: print(f"DEBUG: Was not able to find common seperator, returning untouched software='{software}'")
120 # NOISY-DEBUG: print(f"DEBUG: Checking {len(patterns)} patterns ...")
121 for pattern in patterns:
123 match = pattern.match(version)
125 # NOISY-DEBUG: print(f"DEBUG: match[]={type(match)}")
126 if type(match) is re.Match:
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.")
134 # NOISY-DEBUG: print(f"DEBUG: Found valid version number: '{version}', removing it ...")
135 end = len(software) - len(version)
137 # NOISY-DEBUG: print(f"DEBUG: end[{type(end)}]={end}")
138 software = software[0:end].strip()
140 # NOISY-DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
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}'!")
149 start = software.find("powered by ")
150 # NOISY-DEBUG: print(f"DEBUG: start[{type(start)}]='{start}'")
152 software = software[start + 11:].strip()
153 # NOISY-DEBUG: print(f"DEBUG: software='{software}'")
155 # Next, strip of ' - ' part
156 end = software.find(" - ")
157 # NOISY-DEBUG: print(f"DEBUG: end[{type(end)}]='{end}'")
159 software = software[0:end].strip()
161 # NOISY-DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
164 def is_blacklisted(domain: str) -> bool:
166 for peer in blacklist:
172 def remove_pending_error(domain: str):
174 # Prevent updating any pending errors, nodeinfo was found
175 del pending_errors[domain]
180 def get_hash(domain: str) -> str:
181 return hashlib.sha256(domain.encode("utf-8")).hexdigest()
183 def update_last_blocked(domain: str):
184 # NOISY-DEBUG: print("DEBUG: Updating last_blocked for domain", domain)
186 cursor.execute("UPDATE instances SET last_blocked = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
192 if cursor.rowcount == 0:
193 print("WARNING: Did not update any rows:", domain)
195 except BaseException as e:
196 print("ERROR: failed SQL query:", domain, e)
199 # NOISY-DEBUG: print("DEBUG: EXIT!")
201 def update_nodeinfos(domain: str):
202 # NOISY-DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain)
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} = ?,"
212 fields.append(domain)
213 # NOISY-DEBUG: print(f"DEBUG: sql_string='{sql_string}',fields()={len(fields)}")
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)
219 # NOISY-DEBUG: print("DEBUG: Executing SQL:", sql)
220 cursor.execute(sql, fields)
221 # NOISY-DEBUG: print(f"DEBUG: Success! (rowcount={cursor.rowcount })")
223 if cursor.rowcount == 0:
224 print("WARNING: Did not update any rows:", domain)
226 except BaseException as e:
227 print(f"ERROR: failed SQL query: domain='{domain}',sql='{sql}',exception:'{e}'")
230 # NOISY-DEBUG: print("DEBUG: Deleting nodeinfos for domain:", domain)
231 for key in nodeinfos:
233 # NOISY-DEBUG: print("DEBUG: Deleting key:", key)
234 del nodeinfos[key][domain]
238 # NOISY-DEBUG: print("DEBUG: EXIT!")
240 def update_last_error(domain: str, res: any):
241 # NOISY-DEBUG: print("DEBUG: domain,res[]:", domain, type(res))
243 # NOISY-DEBUG: print("DEBUG: BEFORE res[]:", type(res))
244 if isinstance(res, BaseException) or isinstance(res, json.JSONDecodeError):
247 # NOISY-DEBUG: print("DEBUG: AFTER res[]:", type(res))
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", [
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", [
264 if cursor.rowcount == 0:
265 # NOISY-DEBUG: print("DEBUG: Did not update any rows:", domain)
266 pending_errors[domain] = res
268 except BaseException as e:
269 print("ERROR: failed SQL query:", domain, e)
272 # NOISY-DEBUG: print("DEBUG: EXIT!")
274 def update_last_nodeinfo(domain: str):
275 # NOISY-DEBUG: print("DEBUG: Updating last_nodeinfo for domain:", domain)
277 cursor.execute("UPDATE instances SET last_nodeinfo = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
283 if cursor.rowcount == 0:
284 print("WARNING: Did not update any rows:", domain)
286 except BaseException as e:
287 print("ERROR: failed SQL query:", domain, e)
291 # NOISY-DEBUG: print("DEBUG: EXIT!")
293 def get_peers(domain: str, software: str) -> list:
294 # NOISY-DEBUG: print("DEBUG: Getting peers for domain:", domain, software)
297 if software == "lemmy":
298 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}' is Lemmy, fetching JSON ...")
300 res = reqto.get(f"https://{domain}/api/v3/site", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
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()))
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"]
311 except BaseException as e:
312 print("WARNING: Exception during fetching JSON:", domain, e)
314 update_last_nodeinfo(domain)
316 # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers))
318 elif software == "peertube":
319 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}' is a PeerTube, fetching JSON ...")
322 for mode in ["followers", "following"]:
323 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}',mode='{mode}'")
326 res = reqto.get(f"https://{domain}/api/v1/server/{mode}?start={start}&count=100", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
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()))
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"])
341 print(f"WARNING: record from '{domain}' has no '{mode}' or 'host' record: {record}")
343 if len(json["data"]) < 100:
344 # NOISY-DEBUG: print("DEBUG: Reached end of JSON response:", domain)
347 # Continue with next row
350 except BaseException as e:
351 print("WARNING: Exception during fetching JSON:", domain, e)
353 update_last_nodeinfo(domain)
355 # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers))
358 # NOISY-DEBUG: print(f"DEBUG: Fetching '{get_peers_url}' from '{domain}' ...")
360 res = reqto.get(f"https://{domain}{get_peers_url}", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
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"]))
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"]
370 print("WARNING: Could not reach any JSON API:", domain)
371 update_last_error(domain, res)
373 # NOISY-DEBUG: print("DEBUG:Querying API was successful:", domain, len(res.json()))
375 nodeinfos["get_peers_url"][domain] = get_peers_url
377 except BaseException as e:
378 print("WARNING: Some error during get():", domain, e)
379 update_last_error(domain, e)
381 update_last_nodeinfo(domain)
383 # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers))
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)
390 res = reqto.post(f"https://{domain}{path}", data=data, headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
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)
398 update_last_nodeinfo(domain)
400 except BaseException as e:
401 print("WARNING: Some error during post():", domain, path, data, e)
403 # NOISY-DEBUG: print("DEBUG: Returning json():", len(json))
406 def fetch_nodeinfo(domain: str) -> list:
407 # NOISY-DEBUG: print("DEBUG: Fetching nodeinfo from domain:", domain)
409 nodeinfo = fetch_wellknown_nodeinfo(domain)
410 # NOISY-DEBUG: print("DEBUG:nodeinfo:", len(nodeinfo))
412 if len(nodeinfo) > 0:
413 # NOISY-DEBUG: print("DEBUG: Returning auto-discovered nodeinfo:", len(nodeinfo))
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"
426 for request in requests:
428 # NOISY-DEBUG: print("DEBUG: Fetching request:", request)
429 res = reqto.get(request, headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
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)
435 nodeinfos["detection_mode"][domain] = "STATIC_CHECK"
436 nodeinfos["nodeinfo_url"][domain] = request
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)
443 except BaseException as e:
444 # NOISY-DEBUG: print("DEBUG: Cannot fetch API request:", request)
445 update_last_error(domain, e)
448 # NOISY-DEBUG: print("DEBUG: Returning json[]:", type(json))
451 def fetch_wellknown_nodeinfo(domain: str) -> list:
452 # NOISY-DEBUG: print("DEBUG: Fetching .well-known info for domain:", domain)
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"])
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()))
473 nodeinfos["detection_mode"][domain] = "AUTO_DISCOVERY"
474 nodeinfos["nodeinfo_url"][domain] = link["href"]
477 print("WARNING: Unknown 'rel' value:", domain, link["rel"])
479 print("WARNING: nodeinfo does not contain 'links':", domain)
481 except BaseException as e:
482 print("WARNING: Failed fetching .well-known info:", domain)
483 update_last_error(domain, e)
486 # NOISY-DEBUG: print("DEBUG: Returning json[]:", type(json))
489 def fetch_generator_from_path(domain: str, path: str = "/") -> str:
490 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}',path='{path}' - CALLED!")
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"]))
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")
502 # NOISY-DEBUG: print("DEBUG: doc[]:", type(doc))
503 tag = doc.find("meta", {"name": "generator"})
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)
513 except BaseException as e:
514 # NOISY-DEBUG: print(f"DEBUG: Cannot fetch / from '{domain}':", e)
515 update_last_error(domain, e)
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}'")
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)
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))
531 # NOISY-DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
534 def determine_software(domain: str) -> str:
535 # NOISY-DEBUG: print("DEBUG: Determining software for domain:", domain)
538 # NOISY-DEBUG: print(f"DEBUG: Fetching nodeinfo from '{domain}' ...")
539 json = fetch_nodeinfo(domain)
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)
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)
555 # NOISY-DEBUG: print(f"DEBUG: Generator for domain='{domain}' is: {software}, EXIT!")
558 software = tidyup(json["software"]["name"])
560 # NOISY-DEBUG: print("DEBUG: sofware after tidyup():", software)
561 if software in ["akkoma", "rebased"]:
562 # NOISY-DEBUG: print("DEBUG: Setting pleroma:", domain, software)
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)
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)
580 # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}")
582 print("WARNING: tidyup() left no software name behind:", domain)
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)
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))
598 # NOISY-DEBUG: print("DEBUG: Returning domain,software:", domain, software)
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)
605 "UPDATE blocks SET reason = ?, last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ? AND reason = ''",
615 # NOISY-DEBUG: print(f"DEBUG: cursor.rowcount={cursor.rowcount}")
616 if cursor.rowcount == 0:
617 print("WARNING: Did not update any rows:", domain)
619 except BaseException as e:
620 print("ERROR: failed SQL query:", reason, blocker, blocked, block_level, e)
623 # NOISY-DEBUG: print("DEBUG: EXIT!")
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)
629 "UPDATE blocks SET last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ?",
638 if cursor.rowcount == 0:
639 print("WARNING: Did not update any rows:", domain)
641 except BaseException as e:
642 print("ERROR: failed SQL query:", last_seen, blocker, blocked, block_level, e)
645 # NOISY-DEBUG: print("DEBUG: EXIT!")
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)
652 elif not validators.domain(blocked):
653 print("WARNING: Bad blocked:", blocked)
656 print("INFO: New block:", blocker, blocked, reason, block_level, first_added, last_seen)
659 "INSERT INTO blocks (blocker, blocked, reason, block_level, first_added, last_seen) VALUES(?, ?, ?, ?, ?, ?)",
670 except BaseException as e:
671 print("ERROR: failed SQL query:", blocker, blocked, reason, block_level, first_added, last_seen, e)
674 # NOISY-DEBUG: print("DEBUG: EXIT!")
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)
681 elif origin is not None and not validators.domain(origin):
682 print("WARNING: Bad origin name:", origin)
685 software = determine_software(domain)
686 # NOISY-DEBUG: print("DEBUG: Determined software:", software)
688 print(f"INFO: Adding instance {domain} (origin: {origin})")
691 "INSERT INTO instances (domain, origin, originator, hash, software, first_seen) VALUES (?, ?, ?, ?, ?, ?)",
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)
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)
715 except BaseException as e:
716 print("ERROR: failed SQL query:", domain, e)
719 # NOISY-DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain)
720 update_last_nodeinfo(domain)
722 # NOISY-DEBUG: print("DEBUG: EXIT!")
724 def send_bot_post(instance: str, blocks: dict):
725 message = instance + " has blocked the following instances:\n\n"
730 blocks = blocks[0 : 19]
733 if block["reason"] == None or block["reason"] == '':
734 message = message + block["blocked"] + " with unspecified reason\n"
736 if len(block["reason"]) > 420:
737 block["reason"] = block["reason"][0:419] + "[…]"
739 message = message + block["blocked"] + ' for "' + block["reason"].replace("@", "@\u200b") + '"\n'
742 message = message + "(the list has been truncated to the first 20 entries)"
744 botheaders = {**headers, **{"Authorization": "Bearer " + config["bot_token"]}}
747 f"{config['bot_instance']}/api/v1/statuses",
750 "visibility" : config['bot_visibility'],
751 "content_type": "text/plain"
759 def get_mastodon_blocks(domain: str) -> dict:
760 # NOISY-DEBUG: print("DEBUG: Fetching mastodon blocks from domain:", domain)
762 "Suspended servers": [],
763 "Filtered media" : [],
764 "Limited servers" : [],
765 "Silenced servers" : [],
769 doc = bs4.BeautifulSoup(
770 reqto.get(f"https://{domain}/about/more", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
773 except BaseException as e:
774 print("ERROR: Cannot fetch from domain:", domain, e)
775 update_last_error(domain, e)
778 for header in doc.find_all("h3"):
779 header_text = tidyup(header.text)
781 if header_text in language_mapping:
782 # NOISY-DEBUG: print(f"DEBUG: header_text='{header_text}'")
783 header_text = language_mapping[header_text]
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(
790 "domain": tidyup(line.find("span").text),
791 "hash" : tidyup(line.find("span")["title"][9:]),
792 "reason": tidyup(line.find_all("td")[1].text),
796 # NOISY-DEBUG: print("DEBUG: Returning blocks for domain:", domain)
798 "reject" : blocks["Suspended servers"],
799 "media_removal" : blocks["Filtered media"],
800 "followers_only": blocks["Limited servers"] + blocks["Silenced servers"],
803 def get_friendica_blocks(domain: str) -> dict:
804 # NOISY-DEBUG: print("DEBUG: Fetching friendica blocks from domain:", domain)
808 doc = bs4.BeautifulSoup(
809 reqto.get(f"https://{domain}/friendica", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
812 except BaseException as e:
813 print("WARNING: Failed to fetch /friendica from domain:", domain, e)
814 update_last_error(domain, e)
817 blocklist = doc.find(id="about_blocklist")
819 # Prevents exceptions:
820 if blocklist is None:
821 # NOISY-DEBUG: print("DEBUG:Instance has no block list:", domain)
824 for line in blocklist.find("table").find_all("tr")[1:]:
826 "domain": tidyup(line.find_all("td")[0].text),
827 "reason": tidyup(line.find_all("td")[1].text)
830 # NOISY-DEBUG: print("DEBUG: Returning blocks() for domain:", domain, len(blocks))
835 def get_misskey_blocks(domain: str) -> dict:
836 # NOISY-DEBUG: print("DEBUG: Fetching misskey blocks from domain:", domain)
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
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",
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",
867 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
869 # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
874 if instance["isSuspended"]:
875 blocks["suspended"].append(
877 "domain": tidyup(instance["host"]),
878 # no reason field, nothing
884 # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
887 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
888 counter = counter + step
890 except BaseException as e:
891 print("WARNING: Caught error, exiting loop:", domain, e)
892 update_last_error(domain, e)
897 # same shit, different asshole ("blocked" aka full suspend)
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",
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",
917 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
919 # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
923 if instance["isBlocked"]:
924 blocks["blocked"].append({
925 "domain": tidyup(instance["host"]),
930 # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
933 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
934 counter = counter + step
936 except BaseException as e:
937 print("ERROR: Exception during POST:", domain, e)
938 update_last_error(domain, e)
942 # NOISY-DEBUG: print("DEBUG: Returning for domain,blocked(),suspended():", domain, len(blocks["blocked"]), len(blocks["suspended"]))
944 "reject" : blocks["blocked"],
945 "followers_only": blocks["suspended"]
948 def tidyup(string: str) -> str:
949 # some retards put their blocks in variable case
950 string = string.lower().strip()
952 # other retards put the port
953 string = re.sub("\:\d+$", "", string)
955 # bigger retards put the schema in their blocklist, sometimes even without slashes
956 string = re.sub("^https?\:(\/*)", "", string)
959 string = re.sub("\/$", "", string)
962 string = re.sub("^\@", "", string)
964 # the biggest retards of them all try to block individual users
965 string = re.sub("(.+)\@", "", string)