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 or STATIC_CHECKS
55 # Where to fetch peers (other instances)
61 "Silenced instances" : "Silenced servers",
62 "Suspended instances" : "Suspended servers",
63 "Limited instances" : "Limited servers",
64 # Mappuing German -> English
65 "Gesperrte Server" : "Suspended servers",
66 "Gefilterte Medien" : "Filtered media",
67 "Stummgeschaltete Server" : "Silenced servers",
69 "停止済みのサーバー" : "Suspended servers",
70 "制限中のサーバー" : "Limited servers",
71 "メディアを拒否しているサーバー": "Filtered media",
72 "サイレンス済みのサーバー" : "Silenced servers",
74 "שרתים מושעים" : "Suspended servers",
75 "מדיה מסוננת" : "Filtered media",
76 "שרתים מוגבלים" : "Silenced servers",
78 "Serveurs suspendus" : "Suspended servers",
79 "Médias filtrés" : "Filtered media",
80 "Serveurs limités" : "Limited servers",
81 "Serveurs modérés" : "Limited servers",
84 # URL for fetching peers
85 get_peers_url = "/api/v1/instance/peers"
88 connection = sqlite3.connect("blocks.db")
89 cursor = connection.cursor()
91 def is_blacklisted(domain: str) -> bool:
93 for peer in blacklist:
99 def remove_pending_error(domain: str):
101 # Prevent updating any pending errors, nodeinfo was found
102 del pending_errors[domain]
107 def get_hash(domain: str) -> str:
108 return hashlib.sha256(domain.encode("utf-8")).hexdigest()
110 def update_last_blocked(domain: str):
111 # NOISY-DEBUG: print("DEBUG: Updating last_blocked for domain", domain)
113 cursor.execute("UPDATE instances SET last_blocked = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
119 if cursor.rowcount == 0:
120 print("WARNING: Did not update any rows:", domain)
122 except BaseException as e:
123 print("ERROR: failed SQL query:", domain, e)
126 # NOISY-DEBUG: print("DEBUG: EXIT!")
128 def update_nodeinfos(domain: str):
129 # NOISY-DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain)
132 for key in nodeinfos:
133 # NOISY-DEBUG: print("DEBUG: key:", key)
134 if domain in nodeinfos[key]:
135 # NOISY-DEBUG: print(f"DEBUG: Adding '{nodeinfos[key][domain]}' for key='{key}' ...")
136 fields.append(nodeinfos[key][domain])
137 sql_string += f" {key} = ?,"
139 fields.append(domain)
140 # NOISY-DEBUG: print(f"DEBUG: sql_string='{sql_string}',fields()={len(fields)}")
142 sql = "UPDATE instances SET" + sql_string + " last_status_code = NULL, last_error_details = NULL WHERE domain = ? LIMIT 1"
143 # NOISY-DEBUG: print("DEBUG: sql:", sql)
146 # NOISY-DEBUG: print("DEBUG: Executing SQL:", sql)
147 cursor.execute(sql, fields)
148 # NOISY-DEBUG: print(f"DEBUG: Success! (rowcount={cursor.rowcount })")
150 if cursor.rowcount == 0:
151 print("WARNING: Did not update any rows:", domain)
153 except BaseException as e:
154 print(f"ERROR: failed SQL query: domain='{domain}',sql='{sql}',exception='{e}'")
157 # NOISY-DEBUG: print("DEBUG: Deleting nodeinfos for domain:", domain)
158 for key in nodeinfos:
160 # NOISY-DEBUG: print("DEBUG: Deleting key:", key)
161 del nodeinfos[key][domain]
165 # NOISY-DEBUG: print("DEBUG: EXIT!")
167 def update_last_error(domain: str, res: any):
168 # NOISY-DEBUG: print("DEBUG: domain,res[]:", domain, type(res))
170 # NOISY-DEBUG: print("DEBUG: BEFORE res[]:", type(res))
171 if isinstance(res, BaseException) or isinstance(res, json.JSONDecodeError):
174 # NOISY-DEBUG: print("DEBUG: AFTER res[]:", type(res))
176 # NOISY-DEBUG: print(f"DEBUG: Setting last_error_details='{res}'");
177 cursor.execute("UPDATE instances SET last_status_code = 999, last_error_details = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
183 # NOISY-DEBUG: print(f"DEBUG: Setting last_error_details='{res.reason}'");
184 cursor.execute("UPDATE instances SET last_status_code = ?, last_error_details = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
191 if cursor.rowcount == 0:
192 # NOISY-DEBUG: print("DEBUG: Did not update any rows:", domain)
193 pending_errors[domain] = res
195 except BaseException as e:
196 print("ERROR: failed SQL query:", domain, e)
199 # NOISY-DEBUG: print("DEBUG: EXIT!")
201 def update_last_nodeinfo(domain: str):
202 # NOISY-DEBUG: print("DEBUG: Updating last_nodeinfo for domain:", domain)
204 cursor.execute("UPDATE instances SET last_nodeinfo = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
210 if cursor.rowcount == 0:
211 print("WARNING: Did not update any rows:", domain)
213 except BaseException as e:
214 print("ERROR: failed SQL query:", domain, e)
218 # NOISY-DEBUG: print("DEBUG: EXIT!")
220 def get_peers(domain: str, software: str) -> list:
221 # NOISY-DEBUG: print("DEBUG: Getting peers for domain:", domain, software)
224 if software == "lemmy":
225 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}' is Lemmy, fetching JSON ...")
227 res = reqto.get(f"https://{domain}/api/v3/site", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
229 # NOISY-DEBUG: print(f"DEBUG: res.ok={res.ok},res.json[]={type(res.json())}")
230 if res.ok and isinstance(res.json(), dict):
231 # NOISY-DEBUG: print("DEBUG: Success, res.json():", len(res.json()))
234 if "federated_instances" in json and "linked" in json["federated_instances"]:
235 # NOISY-DEBUG: print("DEBUG: Found federated_instances", domain)
236 peers = json["federated_instances"]["linked"] + json["federated_instances"]["allowed"] + json["federated_instances"]["blocked"]
238 except BaseException as e:
239 print("WARNING: Exception during fetching JSON:", domain, e)
241 update_last_nodeinfo(domain)
243 # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers))
245 elif software == "peertube":
246 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}' is a PeerTube, fetching JSON ...")
249 for mode in ["followers", "following"]:
250 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}',mode='{mode}'")
253 res = reqto.get(f"https://{domain}/api/v1/server/{mode}?start={start}&count=100", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
255 # NOISY-DEBUG: print(f"DEBUG: res.ok={res.ok},res.json[]={type(res.json())}")
256 if res.ok and isinstance(res.json(), dict):
257 # NOISY-DEBUG: print("DEBUG: Success, res.json():", len(res.json()))
261 # NOISY-DEBUG: print(f"DEBUG: Found {len(json['data'])} record(s).")
262 for record in json["data"]:
263 # NOISY-DEBUG: print(f"DEBUG: record()={len(record)}")
264 if mode in record and "host" in record[mode]:
265 # NOISY-DEBUG: print(f"DEBUG: Found host={record[mode]['host']}, adding ...")
266 peers.append(record[mode]["host"])
268 print(f"WARNING: record from '{domain}' has no '{mode}' or 'host' record: {record}")
270 if len(json["data"]) < 100:
271 # NOISY-DEBUG: print("DEBUG: Reached end of JSON response:", domain)
274 # Continue with next row
277 except BaseException as e:
278 print("WARNING: Exception during fetching JSON:", domain, e)
280 update_last_nodeinfo(domain)
282 # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers))
285 # NOISY-DEBUG: print(f"DEBUG: Fetching '{get_peers_url}' from '{domain}' ...")
287 res = reqto.get(f"https://{domain}{get_peers_url}", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
289 # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json()))
290 if not res.ok or res.status_code >= 400:
291 res = reqto.get(f"https://{domain}/api/v3/site", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
293 if "federated_instances" in json and "linked" in json["federated_instances"]:
294 # NOISY-DEBUG: print("DEBUG: Found federated_instances", domain)
295 peers = json["federated_instances"]["linked"] + json["federated_instances"]["allowed"] + json["federated_instances"]["blocked"]
297 print("WARNING: Could not reach any JSON API:", domain)
298 update_last_error(domain, res)
300 # NOISY-DEBUG: print("DEBUG:Querying API was successful:", domain, len(res.json()))
302 nodeinfos["get_peers_url"][domain] = get_peers_url
304 except BaseException as e:
305 print("WARNING: Some error during get():", domain, e)
306 update_last_error(domain, e)
308 update_last_nodeinfo(domain)
310 # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers))
313 def post_json_api(domain: str, path: str, data: str) -> list:
314 # NOISY-DEBUG: print("DEBUG: Sending POST to domain,path,data:", domain, path, data)
317 res = reqto.post(f"https://{domain}{path}", data=data, headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
319 # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json()))
320 if not res.ok or res.status_code >= 400:
321 print("WARNING: Cannot query JSON API:", domain, path, data, res.status_code)
322 update_last_error(domain, res)
325 update_last_nodeinfo(domain)
327 except BaseException as e:
328 print("WARNING: Some error during post():", domain, path, data, e)
330 # NOISY-DEBUG: print("DEBUG: Returning json():", len(json))
333 def fetch_nodeinfo(domain: str) -> list:
334 # NOISY-DEBUG: print("DEBUG: Fetching nodeinfo from domain:", domain)
336 nodeinfo = fetch_wellknown_nodeinfo(domain)
337 # NOISY-DEBUG: print("DEBUG:nodeinfo:", len(nodeinfo))
339 if len(nodeinfo) > 0:
340 # NOISY-DEBUG: print("DEBUG: Returning auto-discovered nodeinfo:", len(nodeinfo))
344 f"https://{domain}/nodeinfo/2.1.json",
345 f"https://{domain}/nodeinfo/2.1",
346 f"https://{domain}/nodeinfo/2.0.json",
347 f"https://{domain}/nodeinfo/2.0",
348 f"https://{domain}/nodeinfo/1.0",
349 f"https://{domain}/api/v1/instance"
353 for request in requests:
355 # NOISY-DEBUG: print("DEBUG: Fetching request:", request)
356 res = reqto.get(request, headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
358 # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json()))
359 if res.ok and isinstance(res.json(), dict):
360 # NOISY-DEBUG: print("DEBUG: Success:", request)
362 nodeinfos["detection_mode"][domain] = "STATIC_CHECK"
363 nodeinfos["nodeinfo_url"][domain] = request
365 elif not res.ok or res.status_code >= 400:
366 print("WARNING: Failed fetching nodeinfo from domain:", domain)
367 update_last_error(domain, res)
370 except BaseException as e:
371 # NOISY-DEBUG: print("DEBUG: Cannot fetch API request:", request)
372 update_last_error(domain, e)
375 # NOISY-DEBUG: print("DEBUG: Returning json[]:", type(json))
378 def fetch_wellknown_nodeinfo(domain: str) -> list:
379 # NOISY-DEBUG: print("DEBUG: Fetching .well-known info for domain:", domain)
383 res = reqto.get(f"https://{domain}/.well-known/nodeinfo", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
384 # NOISY-DEBUG: print("DEBUG: domain,res.ok,res.json[]:", domain, res.ok, type(res.json()))
385 if res.ok and isinstance(res.json(), dict):
386 nodeinfo = res.json()
387 # NOISY-DEBUG: print("DEBUG: Found entries:", len(nodeinfo), domain)
388 if "links" in nodeinfo:
389 # NOISY-DEBUG: print("DEBUG: Found links in nodeinfo():", len(nodeinfo["links"]))
390 for link in nodeinfo["links"]:
391 # NOISY-DEBUG: print("DEBUG: rel,href:", link["rel"], link["href"])
392 if link["rel"] in nodeinfo_identifier:
393 # NOISY-DEBUG: print("DEBUG: Fetching nodeinfo from:", link["href"])
394 res = reqto.get(link["href"])
396 # NOISY-DEBUG: print("DEBUG: href,res.ok,res.status_code:", link["href"], res.ok, res.status_code)
397 if res.ok and isinstance(res.json(), dict):
398 # NOISY-DEBUG: print("DEBUG: Found JSON nodeinfo():", len(res.json()))
400 nodeinfos["detection_mode"][domain] = "AUTO_DISCOVERY"
401 nodeinfos["nodeinfo_url"][domain] = link["href"]
404 print("WARNING: Unknown 'rel' value:", domain, link["rel"])
406 print("WARNING: nodeinfo does not contain 'links':", domain)
408 except BaseException as e:
409 print("WARNING: Failed fetching .well-known info:", domain)
410 update_last_error(domain, e)
413 # NOISY-DEBUG: print("DEBUG: Returning json[]:", type(json))
416 def fetch_generator_from_path(domain: str, path: str = "/") -> str:
417 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}',path='{path}' - CALLED!")
421 # NOISY-DEBUG: print(f"DEBUG: Fetching '{path}' from '{domain}' ...")
422 res = reqto.get(f"https://{domain}{path}", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
424 # NOISY-DEBUG: print("DEBUG: domain,res.ok,res.status_code:", domain, res.ok, res.status_code)
425 if res.ok and res.status_code < 300 and len(res.text) > 0:
426 # NOISY-DEBUG: print("DEBUG: Search for <meta name='generator'>:", domain)
427 doc = bs4.BeautifulSoup(res.text, "html.parser")
429 # NOISY-DEBUG: print("DEBUG: doc[]:", type(doc))
430 tag = doc.find("meta", {"name": "generator"})
432 # NOISY-DEBUG: print(f"DEBUG: tag[{type(tag)}: {tag}")
433 if isinstance(tag, bs4.element.Tag):
434 # NOISY-DEBUG: print("DEBUG: Found generator meta tag: ", domain)
435 software = tidyup(tag.get("content"))
436 # NOISY-DEBUG: print(f"DEBUG: software='{software}'")
437 remove_pending_error(domain)
439 except BaseException as e:
440 print(f"WARNING: Cannot fetch / from '{domain}':", e)
441 update_last_error(domain, e)
444 # NOISY-DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
447 def determine_software(domain: str) -> str:
448 # NOISY-DEBUG: print("DEBUG: Determining software for domain:", domain)
451 # NOISY-DEBUG: print(f"DEBUG: Fetching nodeinfo from '{domain}' ...")
452 json = fetch_nodeinfo(domain)
454 # NOISY-DEBUG: print("DEBUG: json[]:", type(json))
455 if not isinstance(json, dict) or len(json) == 0:
456 # NOISY-DEBUG: print("DEBUG: Could not determine software type:", domain)
457 return fetch_generator_from_path(domain)
459 # NOISY-DEBUG: print("DEBUG: json():", len(json), json)
460 if "status" in json and json["status"] == "error" and "message" in json:
461 print("WARNING: JSON response is an error:", json["message"])
462 update_last_error(domain, json["message"])
463 return fetch_generator_from_path(domain)
464 elif "software" not in json or "name" not in json["software"]:
465 # NOISY-DEBUG: print(f"DEBUG: JSON response from {domain} does not include [software][name], fetching / ...")
466 software = fetch_generator_from_path(domain)
468 # NOISY-DEBUG: print(f"DEBUG: Generator for domain='{domain}' is: {software}, EXIT!")
471 software = tidyup(json["software"]["name"])
473 # NOISY-DEBUG: print("DEBUG: sofware after tidyup():", software)
474 if software in ["akkoma", "rebased"]:
475 # NOISY-DEBUG: print("DEBUG: Setting pleroma:", domain, software)
477 elif software in ["hometown", "ecko"]:
478 # NOISY-DEBUG: print("DEBUG: Setting mastodon:", domain, software)
479 software = "mastodon"
480 elif software in ["calckey", "groundpolis", "foundkey", "cherrypick", "meisskey"]:
481 # NOISY-DEBUG: print("DEBUG: Setting misskey:", domain, software)
483 elif software.find("/") > 0:
484 print("WARNING: Spliting of slash:", software)
485 software = software.split("/")[-1];
486 elif software.find("|") > 0:
487 print("WARNING: Spliting of pipe:", software)
488 software = tidyup(software.split("|")[0]);
490 # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}")
492 print("WARNING: tidyup() left no software name behind:", domain)
495 # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}")
496 if str(software) == "":
497 # NOISY-DEBUG: print(f"DEBUG: software for '{domain}' was not detected, trying generator ...")
498 software = fetch_generator_from_path(domain)
500 # NOISY-DEBUG: print("DEBUG: Returning domain,software:", domain, software)
503 def update_block_reason(reason: str, blocker: str, blocked: str, block_level: str):
504 # NOISY-DEBUG: print("DEBUG: Updating block reason:", reason, blocker, blocked, block_level)
507 "UPDATE blocks SET reason = ?, last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ? AND reason = ''",
517 # NOISY-DEBUG: print(f"DEBUG: cursor.rowcount={cursor.rowcount}")
518 if cursor.rowcount == 0:
519 print("WARNING: Did not update any rows:", domain)
521 except BaseException as e:
522 print("ERROR: failed SQL query:", reason, blocker, blocked, block_level, e)
525 # NOISY-DEBUG: print("DEBUG: EXIT!")
527 def update_last_seen(blocker: str, blocked: str, block_level: str):
528 # NOISY-DEBUG: print("DEBUG: Updating last_seen for:", blocker, blocked, block_level)
531 "UPDATE blocks SET last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ?",
540 if cursor.rowcount == 0:
541 print("WARNING: Did not update any rows:", domain)
543 except BaseException as e:
544 print("ERROR: failed SQL query:", last_seen, blocker, blocked, block_level, e)
547 # NOISY-DEBUG: print("DEBUG: EXIT!")
549 def block_instance(blocker: str, blocked: str, reason: str, block_level: str):
550 # NOISY-DEBUG: print("DEBUG: blocker,blocked,reason,block_level:", blocker, blocked, reason, block_level)
551 if not validators.domain(blocker):
552 print("WARNING: Bad blocker:", blocker)
554 elif not validators.domain(blocked):
555 print("WARNING: Bad blocked:", blocked)
558 print("INFO: New block:", blocker, blocked, reason, block_level, first_added, last_seen)
561 "INSERT INTO blocks (blocker, blocked, reason, block_level, first_added, last_seen) VALUES(?, ?, ?, ?, ?, ?)",
572 except BaseException as e:
573 print("ERROR: failed SQL query:", blocker, blocked, reason, block_level, first_added, last_seen, e)
576 # NOISY-DEBUG: print("DEBUG: EXIT!")
578 def add_instance(domain: str, origin: str, originator: str):
579 # NOISY-DEBUG: print("DEBUG: domain,origin:", domain, origin, originator)
580 if not validators.domain(domain):
581 print("WARNING: Bad domain name:", domain)
583 elif origin is not None and not validators.domain(origin):
584 print("WARNING: Bad origin name:", origin)
587 software = determine_software(domain)
588 # NOISY-DEBUG: print("DEBUG: Determined software:", software)
590 print(f"INFO: Adding new instance {domain} (origin: {origin})")
593 "INSERT INTO instances (domain, origin, originator, hash, software, first_seen) VALUES (?, ?, ?, ?, ?, ?)",
604 if domain in nodeinfos["nodeinfo_url"]:
605 # NOISY-DEBUG # NOISY-DEBUG: print("DEBUG: domain has pending nodeinfo being updated:", domain)
606 update_nodeinfos(domain)
607 remove_pending_error(domain)
608 elif domain in pending_errors:
609 # NOISY-DEBUG: print("DEBUG: domain has pending error being updated:", domain)
610 update_last_error(domain, pending_errors[domain])
611 remove_pending_error(domain)
613 except BaseException as e:
614 print("ERROR: failed SQL query:", domain, e)
617 # NOISY-DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain)
618 update_last_nodeinfo(domain)
620 # NOISY-DEBUG: print("DEBUG: EXIT!")
622 def send_bot_post(instance: str, blocks: dict):
623 message = instance + " has blocked the following instances:\n\n"
628 blocks = blocks[0 : 19]
631 if block["reason"] == None or block["reason"] == '':
632 message = message + block["blocked"] + " with unspecified reason\n"
634 if len(block["reason"]) > 420:
635 block["reason"] = block["reason"][0:419] + "[…]"
637 message = message + block["blocked"] + ' for "' + block["reason"].replace("@", "@\u200b") + '"\n'
640 message = message + "(the list has been truncated to the first 20 entries)"
642 botheaders = {**headers, **{"Authorization": "Bearer " + config["bot_token"]}}
645 f"{config['bot_instance']}/api/v1/statuses",
648 "visibility" : config['bot_visibility'],
649 "content_type": "text/plain"
657 def get_mastodon_blocks(domain: str) -> dict:
658 # NOISY-DEBUG: print("DEBUG: Fetching mastodon blocks from domain:", domain)
660 "Suspended servers": [],
661 "Filtered media" : [],
662 "Limited servers" : [],
663 "Silenced servers" : [],
667 doc = bs4.BeautifulSoup(
668 reqto.get(f"https://{domain}/about/more", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
671 except BaseException as e:
672 print("ERROR: Cannot fetch from domain:", domain, e)
673 update_last_error(domain, e)
676 for header in doc.find_all("h3"):
677 header_text = tidyup(header.text)
679 if header_text in language_mapping:
680 # NOISY-DEBUG: print(f"DEBUG: header_text='{header_text}'")
681 header_text = language_mapping[header_text]
683 if header_text in blocks or header_text.lower() in blocks:
684 # replaced find_next_siblings with find_all_next to account for instances that e.g. hide lists in dropdown menu
685 for line in header.find_all_next("table")[0].find_all("tr")[1:]:
686 blocks[header_text].append(
688 "domain": tidyup(line.find("span").text),
689 "hash" : tidyup(line.find("span")["title"][9:]),
690 "reason": tidyup(line.find_all("td")[1].text),
694 # NOISY-DEBUG: print("DEBUG: Returning blocks for domain:", domain)
696 "reject" : blocks["Suspended servers"],
697 "media_removal" : blocks["Filtered media"],
698 "followers_only": blocks["Limited servers"] + blocks["Silenced servers"],
701 def get_friendica_blocks(domain: str) -> dict:
702 # NOISY-DEBUG: print("DEBUG: Fetching friendica blocks from domain:", domain)
706 doc = bs4.BeautifulSoup(
707 reqto.get(f"https://{domain}/friendica", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
710 except BaseException as e:
711 print("WARNING: Failed to fetch /friendica from domain:", domain, e)
712 update_last_error(domain, e)
715 blocklist = doc.find(id="about_blocklist")
717 # Prevents exceptions:
718 if blocklist is None:
719 # NOISY-DEBUG: print("DEBUG:Instance has no block list:", domain)
722 for line in blocklist.find("table").find_all("tr")[1:]:
724 "domain": tidyup(line.find_all("td")[0].text),
725 "reason": tidyup(line.find_all("td")[1].text)
728 # NOISY-DEBUG: print("DEBUG: Returning blocks() for domain:", domain, len(blocks))
733 def get_misskey_blocks(domain: str) -> dict:
734 # NOISY-DEBUG: print("DEBUG: Fetching misskey blocks from domain:", domain)
743 # iterating through all "suspended" (follow-only in its terminology)
744 # instances page-by-page, since that troonware doesn't support
745 # sending them all at once
748 # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
749 doc = post_json_api(domain, "/api/federation/instances/", json.dumps({
750 "sort" : "+caughtAt",
756 # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
757 doc = post_json_api(domain, "/api/federation/instances/", json.dumps({
758 "sort" : "+caughtAt",
765 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
767 # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
772 if instance["isSuspended"]:
773 blocks["suspended"].append(
775 "domain": tidyup(instance["host"]),
776 # no reason field, nothing
782 # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
785 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
786 counter = counter + step
788 except BaseException as e:
789 print("WARNING: Caught error, exiting loop:", domain, e)
790 update_last_error(domain, e)
795 # same shit, different asshole ("blocked" aka full suspend)
798 # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
799 doc = post_json_api(domain,"/api/federation/instances", json.dumps({
800 "sort" : "+caughtAt",
806 # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
807 doc = post_json_api(domain,"/api/federation/instances", json.dumps({
808 "sort" : "+caughtAt",
815 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
817 # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
821 if instance["isBlocked"]:
822 blocks["blocked"].append({
823 "domain": tidyup(instance["host"]),
828 # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
831 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
832 counter = counter + step
834 except BaseException as e:
835 print("ERROR: Exception during POST:", domain, e)
836 update_last_error(domain, e)
840 # NOISY-DEBUG: print("DEBUG: Returning for domain,blocked(),suspended():", domain, len(blocks["blocked"]), len(blocks["suspended"]))
842 "reject" : blocks["blocked"],
843 "followers_only": blocks["suspended"]
846 def tidyup(string: str) -> str:
847 # some retards put their blocks in variable case
848 string = string.lower().strip()
850 # other retards put the port
851 string = re.sub("\:\d+$", "", string)
853 # bigger retards put the schema in their blocklist, sometimes even without slashes
854 string = re.sub("^https?\:(\/*)", "", string)
857 string = re.sub("\/$", "", string)
860 string = re.sub("^\@", "", string)
862 # the biggest retards of them all try to block individual users
863 string = re.sub("(.+)\@", "", string)