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
93 pattern = 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 def remove_version(software: str) -> str:
96 # NOISY-DEBUG: print(f"DEBUG: software='{software}' - CALLED!")
97 if not "." in software:
98 print(f"WARNING: software='{software}' does not contain a version number.")
103 version = software.split(" ")[-1]
104 elif "/" in software:
105 version = software.split("/")[-1]
106 elif "-" in software:
107 version = software.split("-")[-1]
109 # NOISY-DEBUG: print(f"DEBUG: Was not able to find common seperator, returning untouched software='{software}'")
113 matches = pattern.match(version)
115 # NOISY-DEBUG: print(f"DEBUG: version[{type(version)}]='{version}',matches='{matches}'")
116 if type(matches) is not re.Match:
117 print(f"WARNING: version='{version}' does not match regex, leaving software='{software}' untouched.")
120 # NOISY-DEBUG: print(f"DEBUG: Found valid version number: '{version}', removing it ...")
121 end = len(software) - len(version)
123 # NOISY-DEBUG: print(f"DEBUG: end[{type(end)}]={end}")
124 software = software[0:end].strip()
126 # NOISY-DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
129 def is_blacklisted(domain: str) -> bool:
131 for peer in blacklist:
137 def remove_pending_error(domain: str):
139 # Prevent updating any pending errors, nodeinfo was found
140 del pending_errors[domain]
145 def get_hash(domain: str) -> str:
146 return hashlib.sha256(domain.encode("utf-8")).hexdigest()
148 def update_last_blocked(domain: str):
149 # NOISY-DEBUG: print("DEBUG: Updating last_blocked for domain", domain)
151 cursor.execute("UPDATE instances SET last_blocked = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
157 if cursor.rowcount == 0:
158 print("WARNING: Did not update any rows:", domain)
160 except BaseException as e:
161 print("ERROR: failed SQL query:", domain, e)
164 # NOISY-DEBUG: print("DEBUG: EXIT!")
166 def update_nodeinfos(domain: str):
167 # NOISY-DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain)
170 for key in nodeinfos:
171 # NOISY-DEBUG: print("DEBUG: key:", key)
172 if domain in nodeinfos[key]:
173 # NOISY-DEBUG: print(f"DEBUG: Adding '{nodeinfos[key][domain]}' for key='{key}' ...")
174 fields.append(nodeinfos[key][domain])
175 sql_string += f" {key} = ?,"
177 fields.append(domain)
178 # NOISY-DEBUG: print(f"DEBUG: sql_string='{sql_string}',fields()={len(fields)}")
180 sql = "UPDATE instances SET" + sql_string + " last_status_code = NULL, last_error_details = NULL WHERE domain = ? LIMIT 1"
181 # NOISY-DEBUG: print("DEBUG: sql:", sql)
184 # NOISY-DEBUG: print("DEBUG: Executing SQL:", sql)
185 cursor.execute(sql, fields)
186 # NOISY-DEBUG: print(f"DEBUG: Success! (rowcount={cursor.rowcount })")
188 if cursor.rowcount == 0:
189 print("WARNING: Did not update any rows:", domain)
191 except BaseException as e:
192 print(f"ERROR: failed SQL query: domain='{domain}',sql='{sql}',exception:'{e}'")
195 # NOISY-DEBUG: print("DEBUG: Deleting nodeinfos for domain:", domain)
196 for key in nodeinfos:
198 # NOISY-DEBUG: print("DEBUG: Deleting key:", key)
199 del nodeinfos[key][domain]
203 # NOISY-DEBUG: print("DEBUG: EXIT!")
205 def update_last_error(domain: str, res: any):
206 # NOISY-DEBUG: print("DEBUG: domain,res[]:", domain, type(res))
208 # NOISY-DEBUG: print("DEBUG: BEFORE res[]:", type(res))
209 if isinstance(res, BaseException) or isinstance(res, json.JSONDecodeError):
212 # NOISY-DEBUG: print("DEBUG: AFTER res[]:", type(res))
214 # NOISY-DEBUG: print(f"DEBUG: Setting last_error_details='{res}'");
215 cursor.execute("UPDATE instances SET last_status_code = 999, last_error_details = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
221 # NOISY-DEBUG: print(f"DEBUG: Setting last_error_details='{res.reason}'");
222 cursor.execute("UPDATE instances SET last_status_code = ?, last_error_details = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
229 if cursor.rowcount == 0:
230 # NOISY-DEBUG: print("DEBUG: Did not update any rows:", domain)
231 pending_errors[domain] = res
233 except BaseException as e:
234 print("ERROR: failed SQL query:", domain, e)
237 # NOISY-DEBUG: print("DEBUG: EXIT!")
239 def update_last_nodeinfo(domain: str):
240 # NOISY-DEBUG: print("DEBUG: Updating last_nodeinfo for domain:", domain)
242 cursor.execute("UPDATE instances SET last_nodeinfo = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
248 if cursor.rowcount == 0:
249 print("WARNING: Did not update any rows:", domain)
251 except BaseException as e:
252 print("ERROR: failed SQL query:", domain, e)
256 # NOISY-DEBUG: print("DEBUG: EXIT!")
258 def get_peers(domain: str, software: str) -> list:
259 # NOISY-DEBUG: print("DEBUG: Getting peers for domain:", domain, software)
262 if software == "lemmy":
263 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}' is Lemmy, fetching JSON ...")
265 res = reqto.get(f"https://{domain}/api/v3/site", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
267 # NOISY-DEBUG: print(f"DEBUG: res.ok={res.ok},res.json[]={type(res.json())}")
268 if res.ok and isinstance(res.json(), dict):
269 # NOISY-DEBUG: print("DEBUG: Success, res.json():", len(res.json()))
272 if "federated_instances" in json and "linked" in json["federated_instances"]:
273 # NOISY-DEBUG: print("DEBUG: Found federated_instances", domain)
274 peers = json["federated_instances"]["linked"] + json["federated_instances"]["allowed"] + json["federated_instances"]["blocked"]
276 except BaseException as e:
277 print("WARNING: Exception during fetching JSON:", domain, e)
279 update_last_nodeinfo(domain)
281 # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers))
283 elif software == "peertube":
284 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}' is a PeerTube, fetching JSON ...")
287 for mode in ["followers", "following"]:
288 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}',mode='{mode}'")
291 res = reqto.get(f"https://{domain}/api/v1/server/{mode}?start={start}&count=100", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
293 # NOISY-DEBUG: print(f"DEBUG: res.ok={res.ok},res.json[]={type(res.json())}")
294 if res.ok and isinstance(res.json(), dict):
295 # NOISY-DEBUG: print("DEBUG: Success, res.json():", len(res.json()))
299 # NOISY-DEBUG: print(f"DEBUG: Found {len(json['data'])} record(s).")
300 for record in json["data"]:
301 # NOISY-DEBUG: print(f"DEBUG: record()={len(record)}")
302 if mode in record and "host" in record[mode]:
303 # NOISY-DEBUG: print(f"DEBUG: Found host={record[mode]['host']}, adding ...")
304 peers.append(record[mode]["host"])
306 print(f"WARNING: record from '{domain}' has no '{mode}' or 'host' record: {record}")
308 if len(json["data"]) < 100:
309 # NOISY-DEBUG: print("DEBUG: Reached end of JSON response:", domain)
312 # Continue with next row
315 except BaseException as e:
316 print("WARNING: Exception during fetching JSON:", domain, e)
318 update_last_nodeinfo(domain)
320 # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers))
323 # NOISY-DEBUG: print(f"DEBUG: Fetching '{get_peers_url}' from '{domain}' ...")
325 res = reqto.get(f"https://{domain}{get_peers_url}", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
327 # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json()))
328 if not res.ok or res.status_code >= 400:
329 res = reqto.get(f"https://{domain}/api/v3/site", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
331 if "federated_instances" in json and "linked" in json["federated_instances"]:
332 # NOISY-DEBUG: print("DEBUG: Found federated_instances", domain)
333 peers = json["federated_instances"]["linked"] + json["federated_instances"]["allowed"] + json["federated_instances"]["blocked"]
335 print("WARNING: Could not reach any JSON API:", domain)
336 update_last_error(domain, res)
338 # NOISY-DEBUG: print("DEBUG:Querying API was successful:", domain, len(res.json()))
340 nodeinfos["get_peers_url"][domain] = get_peers_url
342 except BaseException as e:
343 print("WARNING: Some error during get():", domain, e)
344 update_last_error(domain, e)
346 update_last_nodeinfo(domain)
348 # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers))
351 def post_json_api(domain: str, path: str, data: str) -> list:
352 # NOISY-DEBUG: print("DEBUG: Sending POST to domain,path,data:", domain, path, data)
355 res = reqto.post(f"https://{domain}{path}", data=data, headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
357 # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json()))
358 if not res.ok or res.status_code >= 400:
359 print("WARNING: Cannot query JSON API:", domain, path, data, res.status_code)
360 update_last_error(domain, res)
363 update_last_nodeinfo(domain)
365 except BaseException as e:
366 print("WARNING: Some error during post():", domain, path, data, e)
368 # NOISY-DEBUG: print("DEBUG: Returning json():", len(json))
371 def fetch_nodeinfo(domain: str) -> list:
372 # NOISY-DEBUG: print("DEBUG: Fetching nodeinfo from domain:", domain)
374 nodeinfo = fetch_wellknown_nodeinfo(domain)
375 # NOISY-DEBUG: print("DEBUG:nodeinfo:", len(nodeinfo))
377 if len(nodeinfo) > 0:
378 # NOISY-DEBUG: print("DEBUG: Returning auto-discovered nodeinfo:", len(nodeinfo))
382 f"https://{domain}/nodeinfo/2.1.json",
383 f"https://{domain}/nodeinfo/2.1",
384 f"https://{domain}/nodeinfo/2.0.json",
385 f"https://{domain}/nodeinfo/2.0",
386 f"https://{domain}/nodeinfo/1.0",
387 f"https://{domain}/api/v1/instance"
391 for request in requests:
393 # NOISY-DEBUG: print("DEBUG: Fetching request:", request)
394 res = reqto.get(request, headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
396 # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json()))
397 if res.ok and isinstance(res.json(), dict):
398 # NOISY-DEBUG: print("DEBUG: Success:", request)
400 nodeinfos["detection_mode"][domain] = "STATIC_CHECK"
401 nodeinfos["nodeinfo_url"][domain] = request
403 elif not res.ok or res.status_code >= 400:
404 print("WARNING: Failed fetching nodeinfo from domain:", domain)
405 update_last_error(domain, res)
408 except BaseException as e:
409 # NOISY-DEBUG: print("DEBUG: Cannot fetch API request:", request)
410 update_last_error(domain, e)
413 # NOISY-DEBUG: print("DEBUG: Returning json[]:", type(json))
416 def fetch_wellknown_nodeinfo(domain: str) -> list:
417 # NOISY-DEBUG: print("DEBUG: Fetching .well-known info for domain:", domain)
421 res = reqto.get(f"https://{domain}/.well-known/nodeinfo", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
422 # NOISY-DEBUG: print("DEBUG: domain,res.ok,res.json[]:", domain, res.ok, type(res.json()))
423 if res.ok and isinstance(res.json(), dict):
424 nodeinfo = res.json()
425 # NOISY-DEBUG: print("DEBUG: Found entries:", len(nodeinfo), domain)
426 if "links" in nodeinfo:
427 # NOISY-DEBUG: print("DEBUG: Found links in nodeinfo():", len(nodeinfo["links"]))
428 for link in nodeinfo["links"]:
429 # NOISY-DEBUG: print("DEBUG: rel,href:", link["rel"], link["href"])
430 if link["rel"] in nodeinfo_identifier:
431 # NOISY-DEBUG: print("DEBUG: Fetching nodeinfo from:", link["href"])
432 res = reqto.get(link["href"])
434 # NOISY-DEBUG: print("DEBUG: href,res.ok,res.status_code:", link["href"], res.ok, res.status_code)
435 if res.ok and isinstance(res.json(), dict):
436 # NOISY-DEBUG: print("DEBUG: Found JSON nodeinfo():", len(res.json()))
438 nodeinfos["detection_mode"][domain] = "AUTO_DISCOVERY"
439 nodeinfos["nodeinfo_url"][domain] = link["href"]
442 print("WARNING: Unknown 'rel' value:", domain, link["rel"])
444 print("WARNING: nodeinfo does not contain 'links':", domain)
446 except BaseException as e:
447 print("WARNING: Failed fetching .well-known info:", domain)
448 update_last_error(domain, e)
451 # NOISY-DEBUG: print("DEBUG: Returning json[]:", type(json))
454 def fetch_generator_from_path(domain: str, path: str = "/") -> str:
455 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}',path='{path}' - CALLED!")
459 # NOISY-DEBUG: print(f"DEBUG: Fetching '{path}' from '{domain}' ...")
460 res = reqto.get(f"https://{domain}{path}", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
462 # NOISY-DEBUG: print("DEBUG: domain,res.ok,res.status_code,res.text[]:", domain, res.ok, res.status_code, type(res.text))
463 if res.ok and res.status_code < 300 and len(res.text) > 0:
464 # NOISY-DEBUG: print("DEBUG: Search for <meta name='generator'>:", domain)
465 doc = bs4.BeautifulSoup(res.text, "html.parser")
467 # NOISY-DEBUG: print("DEBUG: doc[]:", type(doc))
468 tag = doc.find("meta", {"name": "generator"})
470 # NOISY-DEBUG: print(f"DEBUG: tag[{type(tag)}: {tag}")
471 if isinstance(tag, bs4.element.Tag):
472 # NOISY-DEBUG: print("DEBUG: Found generator meta tag: ", domain)
473 software = tidyup(tag.get("content"))
474 print(f"INFO: domain='{domain}' is generated by '{software}'")
475 nodeinfos["detection_mode"][domain] = "GENERATOR"
476 remove_pending_error(domain)
478 except BaseException as e:
479 # NOISY-DEBUG: print(f"DEBUG: Cannot fetch / from '{domain}':", e)
480 update_last_error(domain, e)
483 # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}")
484 if type(software) is str and software == "":
485 print(f"DEBUG: Corrected empty string to None for software of domain='{domain}'")
487 elif type(software) is str and "." in software:
488 # NOISY-DEBUG: print(f"DEBUG: software='{software}' may contain a version number, domain='{domain}', removing it ...")
489 software = remove_version(software)
491 # NOISY-DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
495 def determine_software(domain: str) -> str:
496 # NOISY-DEBUG: print("DEBUG: Determining software for domain:", domain)
499 # NOISY-DEBUG: print(f"DEBUG: Fetching nodeinfo from '{domain}' ...")
500 json = fetch_nodeinfo(domain)
502 # NOISY-DEBUG: print("DEBUG: json[]:", type(json))
503 if not isinstance(json, dict) or len(json) == 0:
504 # NOISY-DEBUG: print("DEBUG: Could not determine software type:", domain)
505 return fetch_generator_from_path(domain)
507 # NOISY-DEBUG: print("DEBUG: json():", len(json), json)
508 if "status" in json and json["status"] == "error" and "message" in json:
509 print("WARNING: JSON response is an error:", json["message"])
510 update_last_error(domain, json["message"])
511 return fetch_generator_from_path(domain)
512 elif "software" not in json or "name" not in json["software"]:
513 # NOISY-DEBUG: print(f"DEBUG: JSON response from {domain} does not include [software][name], fetching / ...")
514 software = fetch_generator_from_path(domain)
516 # NOISY-DEBUG: print(f"DEBUG: Generator for domain='{domain}' is: {software}, EXIT!")
519 software = tidyup(json["software"]["name"])
521 # NOISY-DEBUG: print("DEBUG: sofware after tidyup():", software)
522 if software in ["akkoma", "rebased"]:
523 # NOISY-DEBUG: print("DEBUG: Setting pleroma:", domain, software)
525 elif software in ["hometown", "ecko"]:
526 # NOISY-DEBUG: print("DEBUG: Setting mastodon:", domain, software)
527 software = "mastodon"
528 elif software in ["calckey", "groundpolis", "foundkey", "cherrypick", "meisskey"]:
529 # NOISY-DEBUG: print("DEBUG: Setting misskey:", domain, software)
531 elif software.find("/") > 0:
532 print("WARNING: Spliting of slash:", software)
533 software = software.split("/")[-1];
534 elif software.find("|") > 0:
535 print("WARNING: Spliting of pipe:", software)
536 software = tidyup(software.split("|")[0]);
538 # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}")
540 print("WARNING: tidyup() left no software name behind:", domain)
543 # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}")
544 if str(software) == "":
545 # NOISY-DEBUG: print(f"DEBUG: software for '{domain}' was not detected, trying generator ...")
546 software = fetch_generator_from_path(domain)
547 elif len(str(software)) > 0 and "." in software:
548 # NOISY-DEBUG: print(f"DEBUG: software='{software}' may contain a version number, domain='{domain}', removing it ...")
549 software = remove_version(software)
551 # NOISY-DEBUG: print("DEBUG: Returning domain,software:", domain, software)
554 def update_block_reason(reason: str, blocker: str, blocked: str, block_level: str):
555 # NOISY-DEBUG: print("DEBUG: Updating block reason:", reason, blocker, blocked, block_level)
558 "UPDATE blocks SET reason = ?, last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ? AND reason = ''",
568 # NOISY-DEBUG: print(f"DEBUG: cursor.rowcount={cursor.rowcount}")
569 if cursor.rowcount == 0:
570 print("WARNING: Did not update any rows:", domain)
572 except BaseException as e:
573 print("ERROR: failed SQL query:", reason, blocker, blocked, block_level, e)
576 # NOISY-DEBUG: print("DEBUG: EXIT!")
578 def update_last_seen(blocker: str, blocked: str, block_level: str):
579 # NOISY-DEBUG: print("DEBUG: Updating last_seen for:", blocker, blocked, block_level)
582 "UPDATE blocks SET last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ?",
591 if cursor.rowcount == 0:
592 print("WARNING: Did not update any rows:", domain)
594 except BaseException as e:
595 print("ERROR: failed SQL query:", last_seen, blocker, blocked, block_level, e)
598 # NOISY-DEBUG: print("DEBUG: EXIT!")
600 def block_instance(blocker: str, blocked: str, reason: str, block_level: str):
601 # NOISY-DEBUG: print("DEBUG: blocker,blocked,reason,block_level:", blocker, blocked, reason, block_level)
602 if not validators.domain(blocker):
603 print("WARNING: Bad blocker:", blocker)
605 elif not validators.domain(blocked):
606 print("WARNING: Bad blocked:", blocked)
609 print("INFO: New block:", blocker, blocked, reason, block_level, first_added, last_seen)
612 "INSERT INTO blocks (blocker, blocked, reason, block_level, first_added, last_seen) VALUES(?, ?, ?, ?, ?, ?)",
623 except BaseException as e:
624 print("ERROR: failed SQL query:", blocker, blocked, reason, block_level, first_added, last_seen, e)
627 # NOISY-DEBUG: print("DEBUG: EXIT!")
629 def add_instance(domain: str, origin: str, originator: str):
630 # NOISY-DEBUG: print("DEBUG: domain,origin:", domain, origin, originator)
631 if not validators.domain(domain):
632 print("WARNING: Bad domain name:", domain)
634 elif origin is not None and not validators.domain(origin):
635 print("WARNING: Bad origin name:", origin)
638 software = determine_software(domain)
639 # NOISY-DEBUG: print("DEBUG: Determined software:", software)
641 print(f"INFO: Adding instance {domain} (origin: {origin})")
644 "INSERT INTO instances (domain, origin, originator, hash, software, first_seen) VALUES (?, ?, ?, ?, ?, ?)",
655 for key in nodeinfos:
656 # NOISY-DEBUG: pprint(f"DEBUG: key='{key}',domain='{domain}',nodeinfos[key]={nodeinfos[key]}")
657 if domain in nodeinfos[key]:
658 # NOISY-DEBUG: pprint(f"DEBUG: domain='{domain}' has pending nodeinfo being updated ...")
659 update_nodeinfos(domain)
660 remove_pending_error(domain)
663 if domain in pending_errors:
664 # NOISY-DEBUG: print("DEBUG: domain has pending error being updated:", domain)
665 update_last_error(domain, pending_errors[domain])
666 remove_pending_error(domain)
668 except BaseException as e:
669 print("ERROR: failed SQL query:", domain, e)
672 # NOISY-DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain)
673 update_last_nodeinfo(domain)
675 # NOISY-DEBUG: print("DEBUG: EXIT!")
677 def send_bot_post(instance: str, blocks: dict):
678 message = instance + " has blocked the following instances:\n\n"
683 blocks = blocks[0 : 19]
686 if block["reason"] == None or block["reason"] == '':
687 message = message + block["blocked"] + " with unspecified reason\n"
689 if len(block["reason"]) > 420:
690 block["reason"] = block["reason"][0:419] + "[…]"
692 message = message + block["blocked"] + ' for "' + block["reason"].replace("@", "@\u200b") + '"\n'
695 message = message + "(the list has been truncated to the first 20 entries)"
697 botheaders = {**headers, **{"Authorization": "Bearer " + config["bot_token"]}}
700 f"{config['bot_instance']}/api/v1/statuses",
703 "visibility" : config['bot_visibility'],
704 "content_type": "text/plain"
712 def get_mastodon_blocks(domain: str) -> dict:
713 # NOISY-DEBUG: print("DEBUG: Fetching mastodon blocks from domain:", domain)
715 "Suspended servers": [],
716 "Filtered media" : [],
717 "Limited servers" : [],
718 "Silenced servers" : [],
722 doc = bs4.BeautifulSoup(
723 reqto.get(f"https://{domain}/about/more", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
726 except BaseException as e:
727 print("ERROR: Cannot fetch from domain:", domain, e)
728 update_last_error(domain, e)
731 for header in doc.find_all("h3"):
732 header_text = tidyup(header.text)
734 if header_text in language_mapping:
735 # NOISY-DEBUG: print(f"DEBUG: header_text='{header_text}'")
736 header_text = language_mapping[header_text]
738 if header_text in blocks or header_text.lower() in blocks:
739 # replaced find_next_siblings with find_all_next to account for instances that e.g. hide lists in dropdown menu
740 for line in header.find_all_next("table")[0].find_all("tr")[1:]:
741 blocks[header_text].append(
743 "domain": tidyup(line.find("span").text),
744 "hash" : tidyup(line.find("span")["title"][9:]),
745 "reason": tidyup(line.find_all("td")[1].text),
749 # NOISY-DEBUG: print("DEBUG: Returning blocks for domain:", domain)
751 "reject" : blocks["Suspended servers"],
752 "media_removal" : blocks["Filtered media"],
753 "followers_only": blocks["Limited servers"] + blocks["Silenced servers"],
756 def get_friendica_blocks(domain: str) -> dict:
757 # NOISY-DEBUG: print("DEBUG: Fetching friendica blocks from domain:", domain)
761 doc = bs4.BeautifulSoup(
762 reqto.get(f"https://{domain}/friendica", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
765 except BaseException as e:
766 print("WARNING: Failed to fetch /friendica from domain:", domain, e)
767 update_last_error(domain, e)
770 blocklist = doc.find(id="about_blocklist")
772 # Prevents exceptions:
773 if blocklist is None:
774 # NOISY-DEBUG: print("DEBUG:Instance has no block list:", domain)
777 for line in blocklist.find("table").find_all("tr")[1:]:
779 "domain": tidyup(line.find_all("td")[0].text),
780 "reason": tidyup(line.find_all("td")[1].text)
783 # NOISY-DEBUG: print("DEBUG: Returning blocks() for domain:", domain, len(blocks))
788 def get_misskey_blocks(domain: str) -> dict:
789 # NOISY-DEBUG: print("DEBUG: Fetching misskey blocks from domain:", domain)
798 # iterating through all "suspended" (follow-only in its terminology)
799 # instances page-by-page, since that troonware doesn't support
800 # sending them all at once
803 # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
804 doc = post_json_api(domain, "/api/federation/instances/", json.dumps({
805 "sort" : "+caughtAt",
811 # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
812 doc = post_json_api(domain, "/api/federation/instances/", json.dumps({
813 "sort" : "+caughtAt",
820 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
822 # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
827 if instance["isSuspended"]:
828 blocks["suspended"].append(
830 "domain": tidyup(instance["host"]),
831 # no reason field, nothing
837 # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
840 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
841 counter = counter + step
843 except BaseException as e:
844 print("WARNING: Caught error, exiting loop:", domain, e)
845 update_last_error(domain, e)
850 # same shit, different asshole ("blocked" aka full suspend)
853 # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
854 doc = post_json_api(domain,"/api/federation/instances", json.dumps({
855 "sort" : "+caughtAt",
861 # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
862 doc = post_json_api(domain,"/api/federation/instances", json.dumps({
863 "sort" : "+caughtAt",
870 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
872 # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
876 if instance["isBlocked"]:
877 blocks["blocked"].append({
878 "domain": tidyup(instance["host"]),
883 # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
886 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
887 counter = counter + step
889 except BaseException as e:
890 print("ERROR: Exception during POST:", domain, e)
891 update_last_error(domain, e)
895 # NOISY-DEBUG: print("DEBUG: Returning for domain,blocked(),suspended():", domain, len(blocks["blocked"]), len(blocks["suspended"]))
897 "reject" : blocks["blocked"],
898 "followers_only": blocks["suspended"]
901 def tidyup(string: str) -> str:
902 # some retards put their blocks in variable case
903 string = string.lower().strip()
905 # other retards put the port
906 string = re.sub("\:\d+$", "", string)
908 # bigger retards put the schema in their blocklist, sometimes even without slashes
909 string = re.sub("^https?\:(\/*)", "", string)
912 string = re.sub("\/$", "", string)
915 string = re.sub("^\@", "", string)
917 # the biggest retards of them all try to block individual users
918 string = re.sub("(.+)\@", "", string)