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 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-]+)*))?)?$"),
95 re.compile("^(?P<year>[1-9]{1}[0-9]{3})\.(?P<month>[0-9]{2})$")
98 def remove_version(software: str) -> str:
99 # NOISY-DEBUG: print(f"DEBUG: software='{software}' - CALLED!")
100 if not "." in software:
101 print(f"WARNING: software='{software}' does not contain a version number.")
106 version = software.split(" ")[-1]
107 elif "/" in software:
108 version = software.split("/")[-1]
109 elif "-" in software:
110 version = software.split("-")[-1]
112 # NOISY-DEBUG: print(f"DEBUG: Was not able to find common seperator, returning untouched software='{software}'")
116 # NOISY-DEBUG: print(f"DEBUG: Checking {len(patterns)} patterns ...")
117 for pattern in patterns:
119 match = pattern.match(version)
120 # NOISY-DEBUG: print(f"DEBUG: match[]={type(match)}")
121 if type(match) is re.Match:
124 # NOISY-DEBUG: print(f"DEBUG: version[{type(version)}]='{version}',match='{match}'")
125 if type(match) is not re.Match:
126 print(f"WARNING: version='{version}' does not match regex, leaving software='{software}' untouched.")
129 # NOISY-DEBUG: print(f"DEBUG: Found valid version number: '{version}', removing it ...")
130 end = len(software) - len(version)
132 # NOISY-DEBUG: print(f"DEBUG: end[{type(end)}]={end}")
133 software = software[0:end].strip()
135 # NOISY-DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
138 def is_blacklisted(domain: str) -> bool:
140 for peer in blacklist:
146 def remove_pending_error(domain: str):
148 # Prevent updating any pending errors, nodeinfo was found
149 del pending_errors[domain]
154 def get_hash(domain: str) -> str:
155 return hashlib.sha256(domain.encode("utf-8")).hexdigest()
157 def update_last_blocked(domain: str):
158 # NOISY-DEBUG: print("DEBUG: Updating last_blocked for domain", domain)
160 cursor.execute("UPDATE instances SET last_blocked = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
166 if cursor.rowcount == 0:
167 print("WARNING: Did not update any rows:", domain)
169 except BaseException as e:
170 print("ERROR: failed SQL query:", domain, e)
173 # NOISY-DEBUG: print("DEBUG: EXIT!")
175 def update_nodeinfos(domain: str):
176 # NOISY-DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain)
179 for key in nodeinfos:
180 # NOISY-DEBUG: print("DEBUG: key:", key)
181 if domain in nodeinfos[key]:
182 # NOISY-DEBUG: print(f"DEBUG: Adding '{nodeinfos[key][domain]}' for key='{key}' ...")
183 fields.append(nodeinfos[key][domain])
184 sql_string += f" {key} = ?,"
186 fields.append(domain)
187 # NOISY-DEBUG: print(f"DEBUG: sql_string='{sql_string}',fields()={len(fields)}")
189 sql = "UPDATE instances SET" + sql_string + " last_status_code = NULL, last_error_details = NULL WHERE domain = ? LIMIT 1"
190 # NOISY-DEBUG: print("DEBUG: sql:", sql)
193 # NOISY-DEBUG: print("DEBUG: Executing SQL:", sql)
194 cursor.execute(sql, fields)
195 # NOISY-DEBUG: print(f"DEBUG: Success! (rowcount={cursor.rowcount })")
197 if cursor.rowcount == 0:
198 print("WARNING: Did not update any rows:", domain)
200 except BaseException as e:
201 print(f"ERROR: failed SQL query: domain='{domain}',sql='{sql}',exception:'{e}'")
204 # NOISY-DEBUG: print("DEBUG: Deleting nodeinfos for domain:", domain)
205 for key in nodeinfos:
207 # NOISY-DEBUG: print("DEBUG: Deleting key:", key)
208 del nodeinfos[key][domain]
212 # NOISY-DEBUG: print("DEBUG: EXIT!")
214 def update_last_error(domain: str, res: any):
215 # NOISY-DEBUG: print("DEBUG: domain,res[]:", domain, type(res))
217 # NOISY-DEBUG: print("DEBUG: BEFORE res[]:", type(res))
218 if isinstance(res, BaseException) or isinstance(res, json.JSONDecodeError):
221 # NOISY-DEBUG: print("DEBUG: AFTER res[]:", type(res))
223 # NOISY-DEBUG: print(f"DEBUG: Setting last_error_details='{res}'");
224 cursor.execute("UPDATE instances SET last_status_code = 999, last_error_details = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
230 # NOISY-DEBUG: print(f"DEBUG: Setting last_error_details='{res.reason}'");
231 cursor.execute("UPDATE instances SET last_status_code = ?, last_error_details = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
238 if cursor.rowcount == 0:
239 # NOISY-DEBUG: print("DEBUG: Did not update any rows:", domain)
240 pending_errors[domain] = res
242 except BaseException as e:
243 print("ERROR: failed SQL query:", domain, e)
246 # NOISY-DEBUG: print("DEBUG: EXIT!")
248 def update_last_nodeinfo(domain: str):
249 # NOISY-DEBUG: print("DEBUG: Updating last_nodeinfo for domain:", domain)
251 cursor.execute("UPDATE instances SET last_nodeinfo = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
257 if cursor.rowcount == 0:
258 print("WARNING: Did not update any rows:", domain)
260 except BaseException as e:
261 print("ERROR: failed SQL query:", domain, e)
265 # NOISY-DEBUG: print("DEBUG: EXIT!")
267 def get_peers(domain: str, software: str) -> list:
268 # NOISY-DEBUG: print("DEBUG: Getting peers for domain:", domain, software)
271 if software == "lemmy":
272 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}' is Lemmy, fetching JSON ...")
274 res = reqto.get(f"https://{domain}/api/v3/site", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
276 # NOISY-DEBUG: print(f"DEBUG: res.ok={res.ok},res.json[]={type(res.json())}")
277 if res.ok and isinstance(res.json(), dict):
278 # NOISY-DEBUG: print("DEBUG: Success, res.json():", len(res.json()))
281 if "federated_instances" in json and "linked" in json["federated_instances"]:
282 # NOISY-DEBUG: print("DEBUG: Found federated_instances", domain)
283 peers = json["federated_instances"]["linked"] + json["federated_instances"]["allowed"] + json["federated_instances"]["blocked"]
285 except BaseException as e:
286 print("WARNING: Exception during fetching JSON:", domain, e)
288 update_last_nodeinfo(domain)
290 # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers))
292 elif software == "peertube":
293 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}' is a PeerTube, fetching JSON ...")
296 for mode in ["followers", "following"]:
297 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}',mode='{mode}'")
300 res = reqto.get(f"https://{domain}/api/v1/server/{mode}?start={start}&count=100", 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()))
308 # NOISY-DEBUG: print(f"DEBUG: Found {len(json['data'])} record(s).")
309 for record in json["data"]:
310 # NOISY-DEBUG: print(f"DEBUG: record()={len(record)}")
311 if mode in record and "host" in record[mode]:
312 # NOISY-DEBUG: print(f"DEBUG: Found host={record[mode]['host']}, adding ...")
313 peers.append(record[mode]["host"])
315 print(f"WARNING: record from '{domain}' has no '{mode}' or 'host' record: {record}")
317 if len(json["data"]) < 100:
318 # NOISY-DEBUG: print("DEBUG: Reached end of JSON response:", domain)
321 # Continue with next row
324 except BaseException as e:
325 print("WARNING: Exception during fetching JSON:", domain, e)
327 update_last_nodeinfo(domain)
329 # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers))
332 # NOISY-DEBUG: print(f"DEBUG: Fetching '{get_peers_url}' from '{domain}' ...")
334 res = reqto.get(f"https://{domain}{get_peers_url}", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
336 # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json()))
337 if not res.ok or res.status_code >= 400:
338 res = reqto.get(f"https://{domain}/api/v3/site", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
340 if "federated_instances" in json and "linked" in json["federated_instances"]:
341 # NOISY-DEBUG: print("DEBUG: Found federated_instances", domain)
342 peers = json["federated_instances"]["linked"] + json["federated_instances"]["allowed"] + json["federated_instances"]["blocked"]
344 print("WARNING: Could not reach any JSON API:", domain)
345 update_last_error(domain, res)
347 # NOISY-DEBUG: print("DEBUG:Querying API was successful:", domain, len(res.json()))
349 nodeinfos["get_peers_url"][domain] = get_peers_url
351 except BaseException as e:
352 print("WARNING: Some error during get():", domain, e)
353 update_last_error(domain, e)
355 update_last_nodeinfo(domain)
357 # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers))
360 def post_json_api(domain: str, path: str, data: str) -> list:
361 # NOISY-DEBUG: print("DEBUG: Sending POST to domain,path,data:", domain, path, data)
364 res = reqto.post(f"https://{domain}{path}", data=data, headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
366 # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json()))
367 if not res.ok or res.status_code >= 400:
368 print("WARNING: Cannot query JSON API:", domain, path, data, res.status_code)
369 update_last_error(domain, res)
372 update_last_nodeinfo(domain)
374 except BaseException as e:
375 print("WARNING: Some error during post():", domain, path, data, e)
377 # NOISY-DEBUG: print("DEBUG: Returning json():", len(json))
380 def fetch_nodeinfo(domain: str) -> list:
381 # NOISY-DEBUG: print("DEBUG: Fetching nodeinfo from domain:", domain)
383 nodeinfo = fetch_wellknown_nodeinfo(domain)
384 # NOISY-DEBUG: print("DEBUG:nodeinfo:", len(nodeinfo))
386 if len(nodeinfo) > 0:
387 # NOISY-DEBUG: print("DEBUG: Returning auto-discovered nodeinfo:", len(nodeinfo))
391 f"https://{domain}/nodeinfo/2.1.json",
392 f"https://{domain}/nodeinfo/2.1",
393 f"https://{domain}/nodeinfo/2.0.json",
394 f"https://{domain}/nodeinfo/2.0",
395 f"https://{domain}/nodeinfo/1.0",
396 f"https://{domain}/api/v1/instance"
400 for request in requests:
402 # NOISY-DEBUG: print("DEBUG: Fetching request:", request)
403 res = reqto.get(request, headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
405 # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json()))
406 if res.ok and isinstance(res.json(), dict):
407 # NOISY-DEBUG: print("DEBUG: Success:", request)
409 nodeinfos["detection_mode"][domain] = "STATIC_CHECK"
410 nodeinfos["nodeinfo_url"][domain] = request
412 elif not res.ok or res.status_code >= 400:
413 print("WARNING: Failed fetching nodeinfo from domain:", domain)
414 update_last_error(domain, res)
417 except BaseException as e:
418 # NOISY-DEBUG: print("DEBUG: Cannot fetch API request:", request)
419 update_last_error(domain, e)
422 # NOISY-DEBUG: print("DEBUG: Returning json[]:", type(json))
425 def fetch_wellknown_nodeinfo(domain: str) -> list:
426 # NOISY-DEBUG: print("DEBUG: Fetching .well-known info for domain:", domain)
430 res = reqto.get(f"https://{domain}/.well-known/nodeinfo", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
431 # NOISY-DEBUG: print("DEBUG: domain,res.ok,res.json[]:", domain, res.ok, type(res.json()))
432 if res.ok and isinstance(res.json(), dict):
433 nodeinfo = res.json()
434 # NOISY-DEBUG: print("DEBUG: Found entries:", len(nodeinfo), domain)
435 if "links" in nodeinfo:
436 # NOISY-DEBUG: print("DEBUG: Found links in nodeinfo():", len(nodeinfo["links"]))
437 for link in nodeinfo["links"]:
438 # NOISY-DEBUG: print("DEBUG: rel,href:", link["rel"], link["href"])
439 if link["rel"] in nodeinfo_identifier:
440 # NOISY-DEBUG: print("DEBUG: Fetching nodeinfo from:", link["href"])
441 res = reqto.get(link["href"])
443 # NOISY-DEBUG: print("DEBUG: href,res.ok,res.status_code:", link["href"], res.ok, res.status_code)
444 if res.ok and isinstance(res.json(), dict):
445 # NOISY-DEBUG: print("DEBUG: Found JSON nodeinfo():", len(res.json()))
447 nodeinfos["detection_mode"][domain] = "AUTO_DISCOVERY"
448 nodeinfos["nodeinfo_url"][domain] = link["href"]
451 print("WARNING: Unknown 'rel' value:", domain, link["rel"])
453 print("WARNING: nodeinfo does not contain 'links':", domain)
455 except BaseException as e:
456 print("WARNING: Failed fetching .well-known info:", domain)
457 update_last_error(domain, e)
460 # NOISY-DEBUG: print("DEBUG: Returning json[]:", type(json))
463 def fetch_generator_from_path(domain: str, path: str = "/") -> str:
464 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}',path='{path}' - CALLED!")
468 # NOISY-DEBUG: print(f"DEBUG: Fetching '{path}' from '{domain}' ...")
469 res = reqto.get(f"https://{domain}{path}", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
471 # NOISY-DEBUG: print("DEBUG: domain,res.ok,res.status_code,res.text[]:", domain, res.ok, res.status_code, type(res.text))
472 if res.ok and res.status_code < 300 and len(res.text) > 0:
473 # NOISY-DEBUG: print("DEBUG: Search for <meta name='generator'>:", domain)
474 doc = bs4.BeautifulSoup(res.text, "html.parser")
476 # NOISY-DEBUG: print("DEBUG: doc[]:", type(doc))
477 tag = doc.find("meta", {"name": "generator"})
479 # NOISY-DEBUG: print(f"DEBUG: tag[{type(tag)}: {tag}")
480 if isinstance(tag, bs4.element.Tag):
481 # NOISY-DEBUG: print("DEBUG: Found generator meta tag: ", domain)
482 software = tidyup(tag.get("content"))
483 print(f"INFO: domain='{domain}' is generated by '{software}'")
484 nodeinfos["detection_mode"][domain] = "GENERATOR"
485 remove_pending_error(domain)
487 except BaseException as e:
488 # NOISY-DEBUG: print(f"DEBUG: Cannot fetch / from '{domain}':", e)
489 update_last_error(domain, e)
492 # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}")
493 if type(software) is str and software == "":
494 print(f"DEBUG: Corrected empty string to None for software of domain='{domain}'")
496 elif type(software) is str and "." in software:
497 # NOISY-DEBUG: print(f"DEBUG: software='{software}' may contain a version number, domain='{domain}', removing it ...")
498 software = remove_version(software)
500 # NOISY-DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
504 def determine_software(domain: str) -> str:
505 # NOISY-DEBUG: print("DEBUG: Determining software for domain:", domain)
508 # NOISY-DEBUG: print(f"DEBUG: Fetching nodeinfo from '{domain}' ...")
509 json = fetch_nodeinfo(domain)
511 # NOISY-DEBUG: print("DEBUG: json[]:", type(json))
512 if not isinstance(json, dict) or len(json) == 0:
513 # NOISY-DEBUG: print("DEBUG: Could not determine software type:", domain)
514 return fetch_generator_from_path(domain)
516 # NOISY-DEBUG: print("DEBUG: json():", len(json), json)
517 if "status" in json and json["status"] == "error" and "message" in json:
518 print("WARNING: JSON response is an error:", json["message"])
519 update_last_error(domain, json["message"])
520 return fetch_generator_from_path(domain)
521 elif "software" not in json or "name" not in json["software"]:
522 # NOISY-DEBUG: print(f"DEBUG: JSON response from {domain} does not include [software][name], fetching / ...")
523 software = fetch_generator_from_path(domain)
525 # NOISY-DEBUG: print(f"DEBUG: Generator for domain='{domain}' is: {software}, EXIT!")
528 software = tidyup(json["software"]["name"])
530 # NOISY-DEBUG: print("DEBUG: sofware after tidyup():", software)
531 if software in ["akkoma", "rebased"]:
532 # NOISY-DEBUG: print("DEBUG: Setting pleroma:", domain, software)
534 elif software in ["hometown", "ecko"]:
535 # NOISY-DEBUG: print("DEBUG: Setting mastodon:", domain, software)
536 software = "mastodon"
537 elif software in ["calckey", "groundpolis", "foundkey", "cherrypick", "meisskey"]:
538 # NOISY-DEBUG: print("DEBUG: Setting misskey:", domain, software)
540 elif software.find("/") > 0:
541 print("WARNING: Spliting of slash:", software)
542 software = software.split("/")[-1];
543 elif software.find("|") > 0:
544 print("WARNING: Spliting of pipe:", software)
545 software = tidyup(software.split("|")[0]);
547 # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}")
549 print("WARNING: tidyup() left no software name behind:", domain)
552 # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}")
553 if str(software) == "":
554 # NOISY-DEBUG: print(f"DEBUG: software for '{domain}' was not detected, trying generator ...")
555 software = fetch_generator_from_path(domain)
556 elif len(str(software)) > 0 and "." in software:
557 # NOISY-DEBUG: print(f"DEBUG: software='{software}' may contain a version number, domain='{domain}', removing it ...")
558 software = remove_version(software)
560 # NOISY-DEBUG: print("DEBUG: Returning domain,software:", domain, software)
563 def update_block_reason(reason: str, blocker: str, blocked: str, block_level: str):
564 # NOISY-DEBUG: print("DEBUG: Updating block reason:", reason, blocker, blocked, block_level)
567 "UPDATE blocks SET reason = ?, last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ? AND reason = ''",
577 # NOISY-DEBUG: print(f"DEBUG: cursor.rowcount={cursor.rowcount}")
578 if cursor.rowcount == 0:
579 print("WARNING: Did not update any rows:", domain)
581 except BaseException as e:
582 print("ERROR: failed SQL query:", reason, blocker, blocked, block_level, e)
585 # NOISY-DEBUG: print("DEBUG: EXIT!")
587 def update_last_seen(blocker: str, blocked: str, block_level: str):
588 # NOISY-DEBUG: print("DEBUG: Updating last_seen for:", blocker, blocked, block_level)
591 "UPDATE blocks SET last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ?",
600 if cursor.rowcount == 0:
601 print("WARNING: Did not update any rows:", domain)
603 except BaseException as e:
604 print("ERROR: failed SQL query:", last_seen, blocker, blocked, block_level, e)
607 # NOISY-DEBUG: print("DEBUG: EXIT!")
609 def block_instance(blocker: str, blocked: str, reason: str, block_level: str):
610 # NOISY-DEBUG: print("DEBUG: blocker,blocked,reason,block_level:", blocker, blocked, reason, block_level)
611 if not validators.domain(blocker):
612 print("WARNING: Bad blocker:", blocker)
614 elif not validators.domain(blocked):
615 print("WARNING: Bad blocked:", blocked)
618 print("INFO: New block:", blocker, blocked, reason, block_level, first_added, last_seen)
621 "INSERT INTO blocks (blocker, blocked, reason, block_level, first_added, last_seen) VALUES(?, ?, ?, ?, ?, ?)",
632 except BaseException as e:
633 print("ERROR: failed SQL query:", blocker, blocked, reason, block_level, first_added, last_seen, e)
636 # NOISY-DEBUG: print("DEBUG: EXIT!")
638 def add_instance(domain: str, origin: str, originator: str):
639 # NOISY-DEBUG: print("DEBUG: domain,origin:", domain, origin, originator)
640 if not validators.domain(domain):
641 print("WARNING: Bad domain name:", domain)
643 elif origin is not None and not validators.domain(origin):
644 print("WARNING: Bad origin name:", origin)
647 software = determine_software(domain)
648 # NOISY-DEBUG: print("DEBUG: Determined software:", software)
650 print(f"INFO: Adding instance {domain} (origin: {origin})")
653 "INSERT INTO instances (domain, origin, originator, hash, software, first_seen) VALUES (?, ?, ?, ?, ?, ?)",
664 for key in nodeinfos:
665 # NOISY-DEBUG: pprint(f"DEBUG: key='{key}',domain='{domain}',nodeinfos[key]={nodeinfos[key]}")
666 if domain in nodeinfos[key]:
667 # NOISY-DEBUG: pprint(f"DEBUG: domain='{domain}' has pending nodeinfo being updated ...")
668 update_nodeinfos(domain)
669 remove_pending_error(domain)
672 if domain in pending_errors:
673 # NOISY-DEBUG: print("DEBUG: domain has pending error being updated:", domain)
674 update_last_error(domain, pending_errors[domain])
675 remove_pending_error(domain)
677 except BaseException as e:
678 print("ERROR: failed SQL query:", domain, e)
681 # NOISY-DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain)
682 update_last_nodeinfo(domain)
684 # NOISY-DEBUG: print("DEBUG: EXIT!")
686 def send_bot_post(instance: str, blocks: dict):
687 message = instance + " has blocked the following instances:\n\n"
692 blocks = blocks[0 : 19]
695 if block["reason"] == None or block["reason"] == '':
696 message = message + block["blocked"] + " with unspecified reason\n"
698 if len(block["reason"]) > 420:
699 block["reason"] = block["reason"][0:419] + "[…]"
701 message = message + block["blocked"] + ' for "' + block["reason"].replace("@", "@\u200b") + '"\n'
704 message = message + "(the list has been truncated to the first 20 entries)"
706 botheaders = {**headers, **{"Authorization": "Bearer " + config["bot_token"]}}
709 f"{config['bot_instance']}/api/v1/statuses",
712 "visibility" : config['bot_visibility'],
713 "content_type": "text/plain"
721 def get_mastodon_blocks(domain: str) -> dict:
722 # NOISY-DEBUG: print("DEBUG: Fetching mastodon blocks from domain:", domain)
724 "Suspended servers": [],
725 "Filtered media" : [],
726 "Limited servers" : [],
727 "Silenced servers" : [],
731 doc = bs4.BeautifulSoup(
732 reqto.get(f"https://{domain}/about/more", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
735 except BaseException as e:
736 print("ERROR: Cannot fetch from domain:", domain, e)
737 update_last_error(domain, e)
740 for header in doc.find_all("h3"):
741 header_text = tidyup(header.text)
743 if header_text in language_mapping:
744 # NOISY-DEBUG: print(f"DEBUG: header_text='{header_text}'")
745 header_text = language_mapping[header_text]
747 if header_text in blocks or header_text.lower() in blocks:
748 # replaced find_next_siblings with find_all_next to account for instances that e.g. hide lists in dropdown menu
749 for line in header.find_all_next("table")[0].find_all("tr")[1:]:
750 blocks[header_text].append(
752 "domain": tidyup(line.find("span").text),
753 "hash" : tidyup(line.find("span")["title"][9:]),
754 "reason": tidyup(line.find_all("td")[1].text),
758 # NOISY-DEBUG: print("DEBUG: Returning blocks for domain:", domain)
760 "reject" : blocks["Suspended servers"],
761 "media_removal" : blocks["Filtered media"],
762 "followers_only": blocks["Limited servers"] + blocks["Silenced servers"],
765 def get_friendica_blocks(domain: str) -> dict:
766 # NOISY-DEBUG: print("DEBUG: Fetching friendica blocks from domain:", domain)
770 doc = bs4.BeautifulSoup(
771 reqto.get(f"https://{domain}/friendica", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
774 except BaseException as e:
775 print("WARNING: Failed to fetch /friendica from domain:", domain, e)
776 update_last_error(domain, e)
779 blocklist = doc.find(id="about_blocklist")
781 # Prevents exceptions:
782 if blocklist is None:
783 # NOISY-DEBUG: print("DEBUG:Instance has no block list:", domain)
786 for line in blocklist.find("table").find_all("tr")[1:]:
788 "domain": tidyup(line.find_all("td")[0].text),
789 "reason": tidyup(line.find_all("td")[1].text)
792 # NOISY-DEBUG: print("DEBUG: Returning blocks() for domain:", domain, len(blocks))
797 def get_misskey_blocks(domain: str) -> dict:
798 # NOISY-DEBUG: print("DEBUG: Fetching misskey blocks from domain:", domain)
807 # iterating through all "suspended" (follow-only in its terminology)
808 # instances page-by-page, since that troonware doesn't support
809 # sending them all at once
812 # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
813 doc = post_json_api(domain, "/api/federation/instances/", json.dumps({
814 "sort" : "+caughtAt",
820 # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
821 doc = post_json_api(domain, "/api/federation/instances/", json.dumps({
822 "sort" : "+caughtAt",
829 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
831 # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
836 if instance["isSuspended"]:
837 blocks["suspended"].append(
839 "domain": tidyup(instance["host"]),
840 # no reason field, nothing
846 # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
849 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
850 counter = counter + step
852 except BaseException as e:
853 print("WARNING: Caught error, exiting loop:", domain, e)
854 update_last_error(domain, e)
859 # same shit, different asshole ("blocked" aka full suspend)
862 # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
863 doc = post_json_api(domain,"/api/federation/instances", json.dumps({
864 "sort" : "+caughtAt",
870 # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
871 doc = post_json_api(domain,"/api/federation/instances", json.dumps({
872 "sort" : "+caughtAt",
879 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
881 # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
885 if instance["isBlocked"]:
886 blocks["blocked"].append({
887 "domain": tidyup(instance["host"]),
892 # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
895 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
896 counter = counter + step
898 except BaseException as e:
899 print("ERROR: Exception during POST:", domain, e)
900 update_last_error(domain, e)
904 # NOISY-DEBUG: print("DEBUG: Returning for domain,blocked(),suspended():", domain, len(blocks["blocked"]), len(blocks["suspended"]))
906 "reject" : blocks["blocked"],
907 "followers_only": blocks["suspended"]
910 def tidyup(string: str) -> str:
911 # some retards put their blocks in variable case
912 string = string.lower().strip()
914 # other retards put the port
915 string = re.sub("\:\d+$", "", string)
917 # bigger retards put the schema in their blocklist, sometimes even without slashes
918 string = re.sub("^https?\:(\/*)", "", string)
921 string = re.sub("\/$", "", string)
924 string = re.sub("^\@", "", string)
926 # the biggest retards of them all try to block individual users
927 string = re.sub("(.+)\@", "", string)