1 # Fedi API Block - An aggregator for fetching blocking data from fediverse nodes
2 # Copyright (C) 2023 Free Software Foundation
4 # This program is free software: you can redistribute it and/or modify
5 # it under the terms of the GNU Affero General Public License as published
6 # by the Free Software Foundation, either version 3 of the License, or
7 # (at your option) any later version.
9 # This program is distributed in the hope that it will be useful,
10 # but WITHOUT ANY WARRANTY; without even the implied warranty of
11 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 # GNU Affero General Public License for more details.
14 # You should have received a copy of the GNU Affero General Public License
15 # along with this program. If not, see <https://www.gnu.org/licenses/>.
27 with open("config.json") as f:
28 config = json.loads(f.read())
30 # Don't check these, known trolls/flooders/testing/developing
32 # Floods network with fake nodes as "research" project
33 "activitypub-troll.cf",
39 "social.shrimpcam.pw",
41 "mastotroll.netz.org",
42 # Testing/developing installations
45 "misskeytest.chn.moe",
48 # Array with pending errors needed to be written to database
52 # "rel" identifiers (no real URLs)
53 nodeinfo_identifier = [
54 "https://nodeinfo.diaspora.software/ns/schema/2.1",
55 "https://nodeinfo.diaspora.software/ns/schema/2.0",
56 "https://nodeinfo.diaspora.software/ns/schema/1.1",
57 "https://nodeinfo.diaspora.software/ns/schema/1.0",
58 "http://nodeinfo.diaspora.software/ns/schema/2.1",
59 "http://nodeinfo.diaspora.software/ns/schema/2.0",
60 "http://nodeinfo.diaspora.software/ns/schema/1.1",
61 "http://nodeinfo.diaspora.software/ns/schema/1.0",
64 # HTTP headers for non-API requests
66 "User-Agent": config["useragent"],
68 # HTTP headers for API requests
70 "User-Agent": config["useragent"],
71 "Content-Type": "application/json",
74 # Found info from node, such as nodeinfo URL, detection mode that needs to be
75 # written to database. Both arrays must be filled at the same time or else
76 # update_instance_data() will fail
78 # Detection mode: 'AUTO_DISCOVERY', 'STATIC_CHECKS' or 'GENERATOR'
79 # NULL means all detection methods have failed (maybe still reachable instance)
80 "detection_mode" : {},
85 # Last fetched instances
86 "last_instance_fetch": {},
91 # Last nodeinfo (fetched)
94 "last_status_code" : {},
96 "last_error_details" : {},
101 "Silenced instances" : "Silenced servers",
102 "Suspended instances" : "Suspended servers",
103 "Limited instances" : "Limited servers",
104 # Mappuing German -> English
105 "Gesperrte Server" : "Suspended servers",
106 "Gefilterte Medien" : "Filtered media",
107 "Stummgeschaltete Server" : "Silenced servers",
108 # Japanese -> English
109 "停止済みのサーバー" : "Suspended servers",
110 "制限中のサーバー" : "Limited servers",
111 "メディアを拒否しているサーバー": "Filtered media",
112 "サイレンス済みのサーバー" : "Silenced servers",
114 "שרתים מושעים" : "Suspended servers",
115 "מדיה מסוננת" : "Filtered media",
116 "שרתים מוגבלים" : "Silenced servers",
118 "Serveurs suspendus" : "Suspended servers",
119 "Médias filtrés" : "Filtered media",
120 "Serveurs limités" : "Limited servers",
121 "Serveurs modérés" : "Limited servers",
124 # URL for fetching peers
125 get_peers_url = "/api/v1/instance/peers"
127 # Cache for redundant SQL queries
130 # Connect to database
131 connection = sqlite3.connect("blocks.db")
132 cursor = connection.cursor()
134 # Pattern instance for version numbers
136 # semantic version number (with v|V) prefix)
137 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-]+)*))?)?$"),
138 # non-sematic, e.g. 1.2.3.4
139 re.compile("^(?P<version>v|V{0,1})(\.{0,1})(?P<major>0|[1-9]\d*)\.(?P<minor>0+|[1-9]\d*)(\.(?P<patch>0+|[1-9]\d*)(\.(?P<subpatch>0|[1-9]\d*))?)$"),
140 # non-sematic, e.g. 2023-05[-dev]
141 re.compile("^(?P<year>[1-9]{1}[0-9]{3})\.(?P<month>[0-9]{2})(-dev){0,1}$"),
142 # non-semantic, e.g. abcdef0
143 re.compile("^[a-f0-9]{7}$"),
148 def is_cache_initialized(key: str) -> bool:
151 def set_all_cache_key(key: str, rows: list, value: any):
152 # NOISY-DEBUG: print(f"DEBUG: key='{key}',rows()={len(rows)},value[]={type(value)} - CALLED!")
153 if not is_cache_initialized(key):
154 # NOISY-DEBUG: print(f"DEBUG: Cache for key='{key}' not initialized.")
158 # NOISY-DEBUG: print(f"DEBUG: Setting key='{key}',sub[{type(sub)}]='{sub}'")
160 if isinstance(sub, tuple):
161 cache[key][sub[0]] = value
163 print(f"WARNING: Unsupported type row[]='{type(row)}'")
165 # NOISY-DEBUG: print("DEBUG: EXIT!")
167 def set_cache_key(key: str, sub: str, value: any):
168 if not is_cache_initialized(key):
169 print(f"WARNING: Bad method call, key='{key}' is not initialized yet.")
170 raise Exception(f"Cache for key='{key}' is not initialized, but function called")
172 cache[key][sub] = value
174 def is_cache_key_set(key: str, sub: str) -> bool:
175 if not is_cache_initialized(key):
176 print(f"WARNING: Bad method call, key='{key}' is not initialized yet.")
177 raise Exception(f"Cache for key='{key}' is not initialized, but function called")
179 return sub in cache[key]
181 def add_peers(rows: dict) -> list:
182 # DEBUG: print(f"DEBUG: rows()={len(rows)} - CALLED!")
184 for element in ["linked", "allowed", "blocked"]:
185 # DEBUG: print(f"DEBUG: Checking element='{element}'")
186 if element in rows and rows[element] != None:
187 # DEBUG: print(f"DEBUG: Adding {len(rows[element])} peer(s) to peers list ...")
188 for peer in rows[element]:
189 # DEBUG: print(f"DEBUG: peer='{peer}' - BEFORE!")
192 # DEBUG: print(f"DEBUG: peer='{peer}' - AFTER!")
193 if is_blacklisted(peer):
194 # DEBUG: print(f"DEBUG: peer='{peer}' is blacklisted, skipped!")
197 # DEBUG: print(f"DEBUG: Adding peer='{peer}' ...")
200 # DEBUG: print(f"DEBUG: peers()={len(peers)} - EXIT!")
203 def remove_version(software: str) -> str:
204 # DEBUG: print(f"DEBUG: software='{software}' - CALLED!")
205 if not "." in software and " " not in software:
206 print(f"WARNING: software='{software}' does not contain a version number.")
211 temp = software.split(";")[0]
212 elif "," in software:
213 temp = software.split(",")[0]
214 elif " - " in software:
215 temp = software.split(" - ")[0]
217 # DEBUG: print(f"DEBUG: software='{software}'")
220 version = temp.split(" ")[-1]
221 elif "/" in software:
222 version = temp.split("/")[-1]
223 elif "-" in software:
224 version = temp.split("-")[-1]
226 # DEBUG: print(f"DEBUG: Was not able to find common seperator, returning untouched software='{software}'")
231 # DEBUG: print(f"DEBUG: Checking {len(patterns)} patterns ...")
232 for pattern in patterns:
234 match = pattern.match(version)
236 # DEBUG: print(f"DEBUG: match[]={type(match)}")
237 if type(match) is re.Match:
240 # DEBUG: print(f"DEBUG: version[{type(version)}]='{version}',match='{match}'")
241 if type(match) is not re.Match:
242 print(f"WARNING: version='{version}' does not match regex, leaving software='{software}' untouched.")
245 # DEBUG: print(f"DEBUG: Found valid version number: '{version}', removing it ...")
246 end = len(temp) - len(version) - 1
248 # DEBUG: print(f"DEBUG: end[{type(end)}]={end}")
249 software = temp[0:end].strip()
250 if " version" in software:
251 # DEBUG: print(f"DEBUG: software='{software}' contains word ' version'")
252 software = strip_until(software, " version")
254 # DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
257 def strip_powered_by(software: str) -> str:
258 # DEBUG: print(f"DEBUG: software='{software}' - CALLED!")
260 print(f"ERROR: Bad method call, 'software' is empty")
261 raise Exception("Parameter 'software' is empty")
262 elif not "powered by" in software:
263 print(f"WARNING: Cannot find 'powered by' in '{software}'!")
266 start = software.find("powered by ")
267 # DEBUG: print(f"DEBUG: start[{type(start)}]='{start}'")
269 software = software[start + 11:].strip()
270 # DEBUG: print(f"DEBUG: software='{software}'")
272 software = strip_until(software, " - ")
274 # DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
277 def strip_hosted_on(software: str) -> str:
278 # DEBUG: print(f"DEBUG: software='{software}' - CALLED!")
280 print(f"ERROR: Bad method call, 'software' is empty")
281 raise Exception("Parameter 'software' is empty")
282 elif not "hosted on" in software:
283 print(f"WARNING: Cannot find 'hosted on' in '{software}'!")
286 end = software.find("hosted on ")
287 # DEBUG: print(f"DEBUG: end[{type(end)}]='{end}'")
289 software = software[0, start].strip()
290 # DEBUG: print(f"DEBUG: software='{software}'")
292 software = strip_until(software, " - ")
294 # DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
297 def strip_until(software: str, until: str) -> str:
298 # DEBUG: print(f"DEBUG: software='{software}',until='{until}' - CALLED!")
300 print(f"ERROR: Bad method call, 'software' is empty")
301 raise Exception("Parameter 'software' is empty")
303 print(f"ERROR: Bad method call, 'until' is empty")
304 raise Exception("Parameter 'until' is empty")
305 elif not until in software:
306 print(f"WARNING: Cannot find '{until}' in '{software}'!")
309 # Next, strip until part
310 end = software.find(until)
312 # DEBUG: print(f"DEBUG: end[{type(end)}]='{end}'")
314 software = software[0:end].strip()
316 # DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
319 def is_blacklisted(domain: str) -> bool:
320 if type(domain) != str:
321 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
323 raise ValueError(f"WARNING: domain cannot be empty")
326 for peer in blacklist:
332 def remove_pending_error(domain: str):
333 if type(domain) != str:
334 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
336 raise ValueError(f"WARNING: domain cannot be empty")
339 # Prevent updating any pending errors, nodeinfo was found
340 del pending_errors[domain]
345 def get_hash(domain: str) -> str:
346 if type(domain) != str:
347 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
349 raise ValueError(f"WARNING: domain cannot be empty")
351 return hashlib.sha256(domain.encode("utf-8")).hexdigest()
353 def update_last_blocked(domain: str):
354 if type(domain) != str:
355 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
357 raise ValueError(f"WARNING: domain cannot be empty")
359 # DEBUG: print("DEBUG: Updating last_blocked for domain", domain)
360 instance_data["last_blocked"][domain] = time.time()
362 # Running pending updated
363 # DEBUG: print(f"DEBUG: Invoking update_instance_data({domain}) ...")
364 update_instance_data(domain)
366 # DEBUG: print("DEBUG: EXIT!")
368 def has_pending_instance_data(domain: str) -> bool:
369 # DEBUG: print(f"DEBUG: domain='{domain}' - CALLED!")
370 if type(domain) != str:
371 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
373 raise ValueError(f"WARNING: domain cannot be empty")
376 for key in instance_data:
377 # DEBUG: print(f"DEBUG: key='{key}',domain='{domain}',instance_data[key]()='{len(instance_data[key])}'")
378 if domain in instance_data[key]:
382 # DEBUG: print(f"DEBUG: has_pending='{has_pending}' - EXIT!")
385 def update_instance_data(domain: str):
386 if type(domain) != str:
387 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
389 raise ValueError(f"WARNING: domain cannot be empty")
391 # DEBUG: print(f"DEBUG: Updating nodeinfo for domain='{domain}' ...")
394 for key in instance_data:
395 # DEBUG: print("DEBUG: key:", key)
396 if domain in instance_data[key]:
397 # DEBUG: print(f"DEBUG: Adding '{instance_data[key][domain]}' for key='{key}' ...")
398 fields.append(instance_data[key][domain])
399 sql_string += f" {key} = ?,"
401 fields.append(domain)
404 raise ValueError(f"WARNING: No fields have been set, but method invoked, domain='{domain}'")
406 # DEBUG: print(f"DEBUG: sql_string='{sql_string}',fields()={len(fields)}")
407 sql = "UPDATE instances SET" + sql_string + " last_status_code = NULL, last_error_details = NULL, last_updated = TIME() WHERE domain = ? LIMIT 1"
408 # DEBUG: print("DEBUG: sql:", sql)
411 # DEBUG: print("DEBUG: Executing SQL:", sql)
412 cursor.execute(sql, fields)
413 # DEBUG: print(f"DEBUG: Success! (rowcount={cursor.rowcount })")
415 if cursor.rowcount == 0:
416 print(f"WARNING: Did not update any rows: domain='{domain}',fields()={len(fields)} - EXIT!")
421 # DEBUG: print("DEBUG: Deleting instance_data for domain:", domain)
422 for key in instance_data:
424 # DEBUG: print("DEBUG: Deleting key:", key)
425 del instance_data[key][domain]
429 except BaseException as e:
430 print(f"ERROR: failed SQL query: domain='{domain}',sql='{sql}',exception[{type(e)}]:'{str(e)}'")
433 # DEBUG: print("DEBUG: EXIT!")
435 def log_error(domain: str, res: any):
436 # DEBUG: print("DEBUG: domain,res[]:", domain, type(res))
437 if type(domain) != str:
438 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
440 raise ValueError(f"WARNING: domain cannot be empty")
443 # DEBUG: print("DEBUG: BEFORE res[]:", type(res))
444 if isinstance(res, BaseException) or isinstance(res, json.JSONDecodeError):
447 # DEBUG: print("DEBUG: AFTER res[]:", type(res))
449 cursor.execute("INSERT INTO error_log (domain, error_code, error_message, created) VALUES (?, 999, ?, ?)",[
455 cursor.execute("INSERT INTO error_log (domain, error_code, error_message, created) VALUES (?, ?, ?, ?)",[
462 # Cleanup old entries
463 # DEBUG: print(f"DEBUG: Purging old records (distance: {config['error_log_cleanup']})")
464 cursor.execute("DELETE FROM error_log WHERE created < ?", [time.time() - config["error_log_cleanup"]])
465 except BaseException as e:
466 print(f"ERROR: failed SQL query: domain='{domain}',exception[{type(e)}]:'{str(e)}'")
469 # DEBUG: print("DEBUG: EXIT!")
471 def update_last_error(domain: str, res: any):
472 # DEBUG: print("DEBUG: domain,res[]:", domain, type(res))
473 if type(domain) != str:
474 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
476 raise ValueError(f"WARNING: domain cannot be empty")
478 # DEBUG: print("DEBUG: BEFORE res[]:", type(res))
479 if isinstance(res, BaseException) or isinstance(res, json.JSONDecodeError):
482 # DEBUG: print("DEBUG: AFTER res[]:", type(res))
484 # DEBUG: print(f"DEBUG: Setting last_error_details='{res}'");
485 instance_data["last_status_code"][domain] = 999
486 instance_data["last_error_details"][domain] = res
488 # DEBUG: print(f"DEBUG: Setting last_error_details='{res.reason}'");
489 instance_data["last_status_code"][domain] = res.status_code
490 instance_data["last_error_details"][domain] = res.reason
492 # Running pending updated
493 # DEBUG: print(f"DEBUG: Invoking update_instance_data({domain}) ...")
494 update_instance_data(domain)
496 log_error(domain, res)
498 # DEBUG: print("DEBUG: EXIT!")
500 def update_last_instance_fetch(domain: str):
501 if type(domain) != str:
502 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
504 raise ValueError(f"WARNING: domain cannot be empty")
506 # DEBUG: print("DEBUG: Updating last_instance_fetch for domain:", domain)
507 instance_data["last_instance_fetch"][domain] = time.time()
509 # Running pending updated
510 # DEBUG: print(f"DEBUG: Invoking update_instance_data({domain}) ...")
511 update_instance_data(domain)
513 # DEBUG: print("DEBUG: EXIT!")
515 def update_last_nodeinfo(domain: str):
516 if type(domain) != str:
517 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
519 raise ValueError(f"WARNING: domain cannot be empty")
521 # DEBUG: print("DEBUG: Updating last_nodeinfo for domain:", domain)
522 instance_data["last_nodeinfo"][domain] = time.time()
523 instance_data["last_updated"][domain] = time.time()
525 # Running pending updated
526 # DEBUG: print(f"DEBUG: Invoking update_instance_data({domain}) ...")
527 update_instance_data(domain)
529 # DEBUG: print("DEBUG: EXIT!")
531 def get_peers(domain: str, software: str) -> list:
532 if type(domain) != str:
533 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
535 raise ValueError(f"WARNING: domain cannot be empty")
536 elif type(software) != str and software != None:
537 raise ValueError(f"WARNING: software[]={type(software)} is not 'str'")
539 # DEBUG: print(f"DEBUG: domain='{domain}',software='{software}' - CALLED!")
542 if software == "misskey":
543 # DEBUG: print(f"DEBUG: domain='{domain}' is misskey, sending API POST request ...")
545 step = config["misskey_offset"]
547 # iterating through all "suspended" (follow-only in its terminology)
548 # instances page-by-page, since that troonware doesn't support
549 # sending them all at once
551 # DEBUG: print(f"DEBUG: Fetching offset='{offset}' from '{domain}' ...")
553 fetched = post_json_api(domain, "/api/federation/instances", json.dumps({
557 }), {"Origin": domain})
559 fetched = post_json_api(domain, "/api/federation/instances", json.dumps({
564 }), {"Origin": domain})
566 # DEBUG: print("DEBUG: fetched():", len(fetched))
567 if len(fetched) == 0:
568 # DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
570 elif len(fetched) != config["misskey_offset"]:
571 # DEBUG: print(f"DEBUG: Fetched '{len(fetched)}' row(s) but expected: '{config['misskey_offset']}'")
572 offset = offset + (config["misskey_offset"] - len(fetched))
574 # DEBUG: print("DEBUG: Raising offset by step:", step)
575 offset = offset + step
578 # DEBUG: print(f"DEBUG: fetched({len(fetched)})[]={type(fetched)}")
579 if isinstance(fetched, dict) and "error" in fetched and "message" in fetched["error"]:
580 print(f"WARNING: post_json_api() returned error: {fetched['error']['message']}")
581 update_last_error(domain, fetched["error"]["message"])
585 # DEBUG: print(f"DEBUG: row():{len(row)}")
586 if not "host" in row:
587 print(f"WARNING: row()={len(row)} does not contain element 'host': {row},domain='{domain}'")
589 elif type(row["host"]) != str:
590 print(f"WARNING: row[host][]={type(row['host'])} is not 'str'")
592 elif is_blacklisted(row["host"]):
593 # DEBUG: print(f"DEBUG: row[host]='{row['host']}' is blacklisted. domain='{domain}'")
596 # DEBUG: print(f"DEBUG: Adding peer: '{row['host']}'")
597 peers.append(row["host"])
599 # DEBUG: print(f"DEBUG: Adding '{len(peers)}' for domain='{domain}'")
600 instance_data["total_peers"][domain] = len(peers)
602 # DEBUG: print(f"DEBUG: Updating last_instance_fetch for domain='{domain}' ...")
603 update_last_instance_fetch(domain)
605 # DEBUG: print("DEBUG: Returning peers[]:", type(peers))
607 elif software == "lemmy":
608 # DEBUG: print(f"DEBUG: domain='{domain}' is Lemmy, fetching JSON ...")
610 res = reqto.get(f"https://{domain}/api/v3/site", headers=api_headers, timeout=(config["connection_timeout"], config["read_timeout"]))
613 # DEBUG: print(f"DEBUG: res.ok={res.ok},res.status_code='{res.status_code}',data[]='{type(data)}'")
614 if not res.ok or res.status_code >= 400:
615 print("WARNING: Could not reach any JSON API:", domain)
616 update_last_error(domain, res)
617 elif res.ok and isinstance(data, list):
618 # DEBUG: print(f"DEBUG: domain='{domain}' returned a list: '{data}'")
620 elif "federated_instances" in data:
621 # DEBUG: print(f"DEBUG: Found federated_instances for domain='{domain}'")
622 peers = peers + add_peers(data["federated_instances"])
623 # DEBUG: print("DEBUG: Added instance(s) to peers")
625 print("WARNING: JSON response does not contain 'federated_instances':", domain)
626 update_last_error(domain, res)
628 except BaseException as e:
629 print(f"WARNING: Exception during fetching JSON: domain='{domain}',exception[{type(e)}]:'{str(e)}'")
631 # DEBUG: print(f"DEBUG: Adding '{len(peers)}' for domain='{domain}'")
632 instance_data["total_peers"][domain] = len(peers)
634 # DEBUG: print(f"DEBUG: Updating last_instance_fetch for domain='{domain}' ...")
635 update_last_instance_fetch(domain)
637 # DEBUG: print("DEBUG: Returning peers[]:", type(peers))
639 elif software == "peertube":
640 # DEBUG: print(f"DEBUG: domain='{domain}' is a PeerTube, fetching JSON ...")
643 for mode in ["followers", "following"]:
644 # DEBUG: print(f"DEBUG: domain='{domain}',mode='{mode}'")
647 res = reqto.get(f"https://{domain}/api/v1/server/{mode}?start={start}&count=100", headers=api_headers, timeout=(config["connection_timeout"], config["read_timeout"]))
650 # DEBUG: print(f"DEBUG: res.ok={res.ok},res.status_code='{res.status_code}',data[]='{type(data)}'")
651 if res.ok and isinstance(data, dict):
652 # DEBUG: print("DEBUG: Success, data:", len(data))
654 # DEBUG: print(f"DEBUG: Found {len(data['data'])} record(s).")
655 for record in data["data"]:
656 # DEBUG: print(f"DEBUG: record()={len(record)}")
657 if mode in record and "host" in record[mode]:
658 # DEBUG: print(f"DEBUG: Found host={record[mode]['host']}, adding ...")
659 peers.append(record[mode]["host"])
661 print(f"WARNING: record from '{domain}' has no '{mode}' or 'host' record: {record}")
663 if len(data["data"]) < 100:
664 # DEBUG: print("DEBUG: Reached end of JSON response:", domain)
667 # Continue with next row
670 except BaseException as e:
671 print(f"WARNING: Exception during fetching JSON: domain='{domain}',exception[{type(e)}]:'{str(e)}'")
673 # DEBUG: print(f"DEBUG: Adding '{len(peers)}' for domain='{domain}'")
674 instance_data["total_peers"][domain] = len(peers)
676 # DEBUG: print(f"DEBUG: Updating last_instance_fetch for domain='{domain}' ...")
677 update_last_instance_fetch(domain)
679 # DEBUG: print("DEBUG: Returning peers[]:", type(peers))
682 # DEBUG: print(f"DEBUG: Fetching get_peers_url='{get_peers_url}' from '{domain}' ...")
684 res = reqto.get(f"https://{domain}{get_peers_url}", headers=api_headers, timeout=(config["connection_timeout"], config["read_timeout"]))
687 # DEBUG: print(f"DEBUG: res.ok={res.ok},res.status_code={res.status_code},data[]='{type(data)}'")
688 if not res.ok or res.status_code >= 400:
689 # DEBUG: print(f"DEBUG: Was not able to fetch '{get_peers_url}', trying alternative ...")
690 res = reqto.get(f"https://{domain}/api/v3/site", headers=api_headers, timeout=(config["connection_timeout"], config["read_timeout"]))
693 # DEBUG: print(f"DEBUG: res.ok={res.ok},res.status_code={res.status_code},data[]='{type(data)}'")
694 if not res.ok or res.status_code >= 400:
695 print("WARNING: Could not reach any JSON API:", domain)
696 update_last_error(domain, res)
697 elif res.ok and isinstance(data, list):
698 # DEBUG: print(f"DEBUG: domain='{domain}' returned a list: '{data}'")
700 elif "federated_instances" in data:
701 # DEBUG: print(f"DEBUG: Found federated_instances for domain='{domain}'")
702 peers = peers + add_peers(data["federated_instances"])
703 # DEBUG: print("DEBUG: Added instance(s) to peers")
705 print("WARNING: JSON response does not contain 'federated_instances':", domain)
706 update_last_error(domain, res)
708 # DEBUG: print("DEBUG: Querying API was successful:", domain, len(data))
711 except BaseException as e:
712 print("WARNING: Some error during get():", domain, e)
713 update_last_error(domain, e)
715 # DEBUG: print(f"DEBUG: Adding '{len(peers)}' for domain='{domain}'")
716 instance_data["total_peers"][domain] = len(peers)
718 # DEBUG: print(f"DEBUG: Updating last_instance_fetch for domain='{domain}' ...")
719 update_last_instance_fetch(domain)
721 # DEBUG: print("DEBUG: Returning peers[]:", type(peers))
724 def post_json_api(domain: str, path: str, parameter: str, extra_headers: dict = {}) -> dict:
725 if type(domain) != str:
726 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
728 raise ValueError(f"WARNING: domain cannot be empty")
729 elif type(path) != str:
730 raise ValueError(f"WARNING: path[]={type(path)} is not 'str'")
732 raise ValueError(f"WARNING: path cannot be empty")
733 elif type(parameter) != str:
734 raise ValueError(f"WARNING: parameter[]={type(parameter)} is not 'str'")
736 # DEBUG: print("DEBUG: Sending POST to domain,path,parameter:", domain, path, parameter, extra_headers)
739 res = reqto.post(f"https://{domain}{path}", data=parameter, headers={**api_headers, **extra_headers}, timeout=(config["connection_timeout"], config["read_timeout"]))
742 # DEBUG: print(f"DEBUG: res.ok={res.ok},res.status_code={res.status_code},data[]='{type(data)}'")
743 if not res.ok or res.status_code >= 400:
744 print(f"WARNING: Cannot query JSON API: domain='{domain}',path='{path}',parameter()={len(parameter)},res.status_code='{res.status_code}',data[]='{type(data)}'")
745 update_last_error(domain, res)
747 except BaseException as e:
748 print(f"WARNING: Some error during post(): domain='{domain}',path='{path}',parameter()={len(parameter)},exception[{type(e)}]:'{str(e)}'")
750 # DEBUG: print(f"DEBUG: Returning data({len(data)})=[]:{type(data)}")
753 def fetch_nodeinfo(domain: str, path: str = None) -> list:
754 if type(domain) != str:
755 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
757 raise ValueError(f"WARNING: domain cannot be empty")
759 # DEBUG: print("DEBUG: Fetching nodeinfo from domain,path:", domain, path)
761 nodeinfo = fetch_wellknown_nodeinfo(domain)
762 # DEBUG: print("DEBUG: nodeinfo:", nodeinfo)
764 if len(nodeinfo) > 0:
765 # DEBUG: print("DEBUG: Returning auto-discovered nodeinfo:", len(nodeinfo))
769 f"https://{domain}/nodeinfo/2.1.json",
770 f"https://{domain}/nodeinfo/2.1",
771 f"https://{domain}/nodeinfo/2.0.json",
772 f"https://{domain}/nodeinfo/2.0",
773 f"https://{domain}/nodeinfo/1.0",
774 f"https://{domain}/api/v1/instance"
778 for request in requests:
779 if path != None and path != "" and request != path:
780 # DEBUG: print(f"DEBUG: path='{path}' does not match request='{request}' - SKIPPED!")
784 # DEBUG: print("DEBUG: Fetching request:", request)
785 res = reqto.get(request, headers=api_headers, timeout=(config["connection_timeout"], config["read_timeout"]))
788 # DEBUG: print(f"DEBUG: res.ok={res.ok},res.status_code={res.status_code},data[]='{type(data)}'")
789 if res.ok and isinstance(data, dict):
790 # DEBUG: print("DEBUG: Success:", request)
791 instance_data["detection_mode"][domain] = "STATIC_CHECK"
792 instance_data["nodeinfo_url"][domain] = request
794 elif res.ok and isinstance(data, list):
795 # DEBUG: print(f"DEBUG: domain='{domain}' returned a list: '{data}'")
797 elif not res.ok or res.status_code >= 400:
798 print("WARNING: Failed fetching nodeinfo from domain:", domain)
799 update_last_error(domain, res)
802 except BaseException as e:
803 # DEBUG: print("DEBUG: Cannot fetch API request:", request)
804 update_last_error(domain, e)
807 # DEBUG: print("DEBUG: Returning data[]:", type(data))
810 def fetch_wellknown_nodeinfo(domain: str) -> list:
811 if type(domain) != str:
812 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
814 raise ValueError(f"WARNING: domain cannot be empty")
816 # DEBUG: print("DEBUG: Fetching .well-known info for domain:", domain)
820 res = reqto.get(f"https://{domain}/.well-known/nodeinfo", headers=api_headers, timeout=(config["connection_timeout"], config["read_timeout"]))
823 # DEBUG: print("DEBUG: domain,res.ok,data[]:", domain, res.ok, type(data))
824 if res.ok and isinstance(data, dict):
826 # DEBUG: print("DEBUG: Found entries:", len(nodeinfo), domain)
827 if "links" in nodeinfo:
828 # DEBUG: print("DEBUG: Found links in nodeinfo():", len(nodeinfo["links"]))
829 for link in nodeinfo["links"]:
830 # DEBUG: print("DEBUG: rel,href:", link["rel"], link["href"])
831 if link["rel"] in nodeinfo_identifier:
832 # DEBUG: print("DEBUG: Fetching nodeinfo from:", link["href"])
833 res = reqto.get(link["href"])
836 # DEBUG: print("DEBUG: href,res.ok,res.status_code:", link["href"], res.ok, res.status_code)
837 if res.ok and isinstance(data, dict):
838 # DEBUG: print("DEBUG: Found JSON nodeinfo():", len(data))
839 instance_data["detection_mode"][domain] = "AUTO_DISCOVERY"
840 instance_data["nodeinfo_url"][domain] = link["href"]
843 print("WARNING: Unknown 'rel' value:", domain, link["rel"])
845 print("WARNING: nodeinfo does not contain 'links':", domain)
847 except BaseException as e:
848 print("WARNING: Failed fetching .well-known info:", domain)
849 update_last_error(domain, e)
852 # DEBUG: print("DEBUG: Returning data[]:", type(data))
855 def fetch_generator_from_path(domain: str, path: str = "/") -> str:
856 if type(domain) != str:
857 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
859 raise ValueError(f"WARNING: domain cannot be empty")
860 elif type(path) != str:
861 raise ValueError(f"WARNING: path[]={type(path)} is not 'str'")
863 raise ValueError(f"WARNING: domain cannot be empty")
865 # DEBUG: print(f"DEBUG: domain='{domain}',path='{path}' - CALLED!")
869 # DEBUG: print(f"DEBUG: Fetching path='{path}' from '{domain}' ...")
870 res = reqto.get(f"https://{domain}{path}", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
872 # DEBUG: print("DEBUG: domain,res.ok,res.status_code,res.text[]:", domain, res.ok, res.status_code, type(res.text))
873 if res.ok and res.status_code < 300 and len(res.text) > 0:
874 # DEBUG: print("DEBUG: Search for <meta name='generator'>:", domain)
875 doc = bs4.BeautifulSoup(res.text, "html.parser")
877 # DEBUG: print("DEBUG: doc[]:", type(doc))
878 generator = doc.find("meta", {"name": "generator"})
879 site_name = doc.find("meta", {"property": "og:site_name"})
881 # DEBUG: print(f"DEBUG: generator='{generator}',site_name='{site_name}'")
882 if isinstance(generator, bs4.element.Tag):
883 # DEBUG: print("DEBUG: Found generator meta tag:", domain)
884 software = tidyup(generator.get("content"))
885 print(f"INFO: domain='{domain}' is generated by '{software}'")
886 instance_data["detection_mode"][domain] = "GENERATOR"
887 remove_pending_error(domain)
888 elif isinstance(site_name, bs4.element.Tag):
889 # DEBUG: print("DEBUG: Found property=og:site_name:", domain)
890 sofware = tidyup(site_name.get("content"))
891 print(f"INFO: domain='{domain}' has og:site_name='{software}'")
892 instance_data["detection_mode"][domain] = "SITE_NAME"
893 remove_pending_error(domain)
895 except BaseException as e:
896 # DEBUG: print(f"DEBUG: Cannot fetch / from '{domain}':", e)
897 update_last_error(domain, e)
900 # DEBUG: print(f"DEBUG: software[]={type(software)}")
901 if type(software) is str and software == "":
902 # DEBUG: print(f"DEBUG: Corrected empty string to None for software of domain='{domain}'")
904 elif type(software) is str and ("." in software or " " in software):
905 # DEBUG: print(f"DEBUG: software='{software}' may contain a version number, domain='{domain}', removing it ...")
906 software = remove_version(software)
908 # DEBUG: print(f"DEBUG: software[]={type(software)}")
909 if type(software) is str and " powered by " in software:
910 # DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it")
911 software = remove_version(strip_powered_by(software))
912 elif type(software) is str and " hosted on " in software:
913 # DEBUG: print(f"DEBUG: software='{software}' has 'hosted on' in it")
914 software = remove_version(strip_hosted_on(software))
915 elif type(software) is str and " by " in software:
916 # DEBUG: print(f"DEBUG: software='{software}' has ' by ' in it")
917 software = strip_until(software, " by ")
918 elif type(software) is str and " see " in software:
919 # DEBUG: print(f"DEBUG: software='{software}' has ' see ' in it")
920 software = strip_until(software, " see ")
922 # DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
925 def determine_software(domain: str, path: str = None) -> str:
926 if type(domain) != str:
927 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
929 raise ValueError(f"WARNING: domain cannot be empty")
931 # DEBUG: print("DEBUG: Determining software for domain,path:", domain, path)
934 # DEBUG: print(f"DEBUG: Fetching nodeinfo from '{domain}' ...")
935 data = fetch_nodeinfo(domain, path)
937 # DEBUG: print("DEBUG: data[]:", type(data))
938 if not isinstance(data, dict) or len(data) == 0:
939 # DEBUG: print("DEBUG: Could not determine software type:", domain)
940 return fetch_generator_from_path(domain)
942 # DEBUG: print("DEBUG: data():", len(data), data)
943 if "status" in data and data["status"] == "error" and "message" in data:
944 print("WARNING: JSON response is an error:", data["message"])
945 update_last_error(domain, data["message"])
946 return fetch_generator_from_path(domain)
947 elif "message" in data:
948 print("WARNING: JSON response contains only a message:", data["message"])
949 update_last_error(domain, data["message"])
950 return fetch_generator_from_path(domain)
951 elif "software" not in data or "name" not in data["software"]:
952 # DEBUG: print(f"DEBUG: JSON response from domain='{domain}' does not include [software][name], fetching / ...")
953 software = fetch_generator_from_path(domain)
955 # DEBUG: print(f"DEBUG: Generator for domain='{domain}' is: {software}, EXIT!")
958 software = tidyup(data["software"]["name"])
960 # DEBUG: print("DEBUG: sofware after tidyup():", software)
961 if software in ["akkoma", "rebased"]:
962 # DEBUG: print("DEBUG: Setting pleroma:", domain, software)
964 elif software in ["hometown", "ecko"]:
965 # DEBUG: print("DEBUG: Setting mastodon:", domain, software)
966 software = "mastodon"
967 elif software in ["calckey", "groundpolis", "foundkey", "cherrypick", "meisskey"]:
968 # DEBUG: print("DEBUG: Setting misskey:", domain, software)
970 elif software.find("/") > 0:
971 print("WARNING: Spliting of slash:", software)
972 software = software.split("/")[-1];
973 elif software.find("|") > 0:
974 print("WARNING: Spliting of pipe:", software)
975 software = tidyup(software.split("|")[0]);
976 elif "powered by" in software:
977 # DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it")
978 software = strip_powered_by(software)
979 elif type(software) is str and " by " in software:
980 # DEBUG: print(f"DEBUG: software='{software}' has ' by ' in it")
981 software = strip_until(software, " by ")
982 elif type(software) is str and " see " in software:
983 # DEBUG: print(f"DEBUG: software='{software}' has ' see ' in it")
984 software = strip_until(software, " see ")
986 # DEBUG: print(f"DEBUG: software[]={type(software)}")
988 print("WARNING: tidyup() left no software name behind:", domain)
991 # DEBUG: print(f"DEBUG: software[]={type(software)}")
992 if str(software) == "":
993 # DEBUG: print(f"DEBUG: software for '{domain}' was not detected, trying generator ...")
994 software = fetch_generator_from_path(domain)
995 elif len(str(software)) > 0 and ("." in software or " " in software):
996 # DEBUG: print(f"DEBUG: software='{software}' may contain a version number, domain='{domain}', removing it ...")
997 software = remove_version(software)
999 # DEBUG: print(f"DEBUG: software[]={type(software)}")
1000 if type(software) is str and "powered by" in software:
1001 # DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it")
1002 software = remove_version(strip_powered_by(software))
1004 # DEBUG: print("DEBUG: Returning domain,software:", domain, software)
1007 def update_block_reason(reason: str, blocker: str, blocked: str, block_level: str):
1008 # DEBUG: print("DEBUG: Updating block reason:", reason, blocker, blocked, block_level)
1011 "UPDATE blocks SET reason = ?, last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ? AND reason = ''",
1021 # DEBUG: print(f"DEBUG: cursor.rowcount={cursor.rowcount}")
1022 if cursor.rowcount == 0:
1023 print("WARNING: Did not update any rows:", domain)
1025 except BaseException as e:
1026 print(f"ERROR: failed SQL query: reason='{reason}',blocker='{blocker}',blocked='{blocked}',block_level='{block_level}',sql='{sql}',exception[{type(e)}]:'{str(e)}'")
1029 # DEBUG: print("DEBUG: EXIT!")
1031 def update_last_seen(blocker: str, blocked: str, block_level: str):
1032 # DEBUG: print("DEBUG: Updating last_seen for:", blocker, blocked, block_level)
1035 "UPDATE blocks SET last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ?",
1044 if cursor.rowcount == 0:
1045 print("WARNING: Did not update any rows:", domain)
1047 except BaseException as e:
1048 print(f"ERROR: failed SQL query: last_seen='{last_seen}',blocker='{blocker}',blocked='{blocked}',block_level='{block_level}',exception[{type(e)}]:'{str(e)}'")
1051 # DEBUG: print("DEBUG: EXIT!")
1053 def block_instance(blocker: str, blocked: str, reason: str, block_level: str):
1054 # DEBUG: print("DEBUG: blocker,blocked,reason,block_level:", blocker, blocked, reason, block_level)
1055 if not validators.domain(blocker.split("/")[0]):
1056 print("WARNING: Bad blocker:", blocker)
1058 elif not validators.domain(blocked.split("/")[0]):
1059 print("WARNING: Bad blocked:", blocked)
1062 print("INFO: New block:", blocker, blocked, reason, block_level, first_seen, last_seen)
1065 "INSERT INTO blocks (blocker, blocked, reason, block_level, first_seen, last_seen) VALUES(?, ?, ?, ?, ?, ?)",
1076 except BaseException as e:
1077 print(f"ERROR: failed SQL query: blocker='{blocker}',blocked='{blocked}',reason='{reason}',block_level='{block_level}',first_seen='{first_seen}',last_seen='{last_seen}',exception[{type(e)}]:'{str(e)}'")
1080 # DEBUG: print("DEBUG: EXIT!")
1082 def is_instance_registered(domain: str) -> bool:
1083 if type(domain) != str:
1084 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
1086 raise ValueError(f"WARNING: domain cannot be empty")
1088 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}' - CALLED!")
1089 if not is_cache_initialized("is_registered"):
1090 # NOISY-DEBUG: print(f"DEBUG: Cache for 'is_registered' not initialized, fetching all rows ...")
1092 cursor.execute("SELECT domain FROM instances")
1095 set_all_cache_key("is_registered", cursor.fetchall(), True)
1096 except BaseException as e:
1097 print(f"ERROR: failed SQL query: domain='{domain}',exception[{type(e)}]:'{str(e)}'")
1101 registered = is_cache_key_set("is_registered", domain)
1103 # NOISY-DEBUG: print(f"DEBUG: registered='{registered}' - EXIT!")
1106 def add_instance(domain: str, origin: str, originator: str, path: str = None):
1107 if type(domain) != str:
1108 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
1110 raise ValueError(f"WARNING: domain cannot be empty")
1111 elif type(origin) != str and origin != None:
1112 raise ValueError(f"WARNING: origin[]={type(origin)} is not 'str'")
1113 elif type(originator) != str:
1114 raise ValueError(f"WARNING: originator[]={type(originator)} is not 'str'")
1115 elif originator == "":
1116 raise ValueError(f"WARNING: originator cannot be empty")
1118 # DEBUG: print("DEBUG: domain,origin,originator,path:", domain, origin, originator, path)
1119 if not validators.domain(domain.split("/")[0]):
1120 print("WARNING: Bad domain name:", domain)
1122 elif origin is not None and not validators.domain(origin.split("/")[0]):
1123 print("WARNING: Bad origin name:", origin)
1126 software = determine_software(domain, path)
1127 # DEBUG: print("DEBUG: Determined software:", software)
1129 print(f"INFO: Adding instance domain='{domain}' (origin='{origin}',software='{software}')")
1132 "INSERT INTO instances (domain, origin, originator, hash, software, first_seen) VALUES (?, ?, ?, ?, ?, ?)",
1143 set_cache_key("is_registered", domain, True)
1145 if has_pending_instance_data(domain):
1146 # DEBUG: print(f"DEBUG: domain='{domain}' has pending nodeinfo being updated ...")
1147 update_instance_data(domain)
1148 remove_pending_error(domain)
1150 if domain in pending_errors:
1151 # DEBUG: print("DEBUG: domain has pending error being updated:", domain)
1152 update_last_error(domain, pending_errors[domain])
1153 remove_pending_error(domain)
1155 except BaseException as e:
1156 print(f"ERROR: failed SQL query: domain='{domain}',exception[{type(e)}]:'{str(e)}'")
1159 # DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain)
1160 update_last_nodeinfo(domain)
1162 # DEBUG: print("DEBUG: EXIT!")
1164 def send_bot_post(instance: str, blocks: dict):
1165 message = instance + " has blocked the following instances:\n\n"
1168 if len(blocks) > 20:
1170 blocks = blocks[0 : 19]
1172 for block in blocks:
1173 if block["reason"] == None or block["reason"] == '':
1174 message = message + block["blocked"] + " with unspecified reason\n"
1176 if len(block["reason"]) > 420:
1177 block["reason"] = block["reason"][0:419] + "[…]"
1179 message = message + block["blocked"] + ' for "' + block["reason"].replace("@", "@\u200b") + '"\n'
1182 message = message + "(the list has been truncated to the first 20 entries)"
1184 botheaders = {**api_headers, **{"Authorization": "Bearer " + config["bot_token"]}}
1187 f"{config['bot_instance']}/api/v1/statuses",
1190 "visibility" : config['bot_visibility'],
1191 "content_type": "text/plain"
1199 def get_mastodon_blocks(domain: str) -> dict:
1200 if type(domain) != str:
1201 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
1203 raise ValueError(f"WARNING: domain cannot be empty")
1205 # DEBUG: print("DEBUG: Fetching mastodon blocks from domain:", domain)
1207 "Suspended servers": [],
1208 "Filtered media" : [],
1209 "Limited servers" : [],
1210 "Silenced servers" : [],
1214 doc = bs4.BeautifulSoup(
1215 reqto.get(f"https://{domain}/about/more", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
1218 except BaseException as e:
1219 print("ERROR: Cannot fetch from domain:", domain, e)
1220 update_last_error(domain, e)
1223 for header in doc.find_all("h3"):
1224 header_text = tidyup(header.text)
1226 if header_text in language_mapping:
1227 # DEBUG: print(f"DEBUG: header_text='{header_text}'")
1228 header_text = language_mapping[header_text]
1230 if header_text in blocks or header_text.lower() in blocks:
1231 # replaced find_next_siblings with find_all_next to account for instances that e.g. hide lists in dropdown menu
1232 for line in header.find_all_next("table")[0].find_all("tr")[1:]:
1233 blocks[header_text].append(
1235 "domain": tidyup(line.find("span").text),
1236 "hash" : tidyup(line.find("span")["title"][9:]),
1237 "reason": tidyup(line.find_all("td")[1].text),
1241 # DEBUG: print("DEBUG: Returning blocks for domain:", domain)
1243 "reject" : blocks["Suspended servers"],
1244 "media_removal" : blocks["Filtered media"],
1245 "followers_only": blocks["Limited servers"] + blocks["Silenced servers"],
1248 def get_friendica_blocks(domain: str) -> dict:
1249 if type(domain) != str:
1250 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
1252 raise ValueError(f"WARNING: domain cannot be empty")
1254 # DEBUG: print("DEBUG: Fetching friendica blocks from domain:", domain)
1258 doc = bs4.BeautifulSoup(
1259 reqto.get(f"https://{domain}/friendica", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
1262 except BaseException as e:
1263 print("WARNING: Failed to fetch /friendica from domain:", domain, e)
1264 update_last_error(domain, e)
1267 blocklist = doc.find(id="about_blocklist")
1269 # Prevents exceptions:
1270 if blocklist is None:
1271 # DEBUG: print("DEBUG:Instance has no block list:", domain)
1274 for line in blocklist.find("table").find_all("tr")[1:]:
1275 # DEBUG: print(f"DEBUG: line='{line}'")
1277 "domain": tidyup(line.find_all("td")[0].text),
1278 "reason": tidyup(line.find_all("td")[1].text)
1281 # DEBUG: print("DEBUG: Returning blocks() for domain:", domain, len(blocks))
1286 def get_misskey_blocks(domain: str) -> dict:
1287 if type(domain) != str:
1288 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
1290 raise ValueError(f"WARNING: domain cannot be empty")
1292 # DEBUG: print("DEBUG: Fetching misskey blocks from domain:", domain)
1299 step = config["misskey_offset"]
1301 # iterating through all "suspended" (follow-only in its terminology)
1302 # instances page-by-page, since that troonware doesn't support
1303 # sending them all at once
1305 # DEBUG: print(f"DEBUG: Fetching offset='{offset}' from '{domain}' ...")
1307 # DEBUG: print("DEBUG: Sending JSON API request to domain,step,offset:", domain, step, offset)
1308 fetched = post_json_api(domain, "/api/federation/instances", json.dumps({
1313 }), {"Origin": domain})
1315 # DEBUG: print("DEBUG: Sending JSON API request to domain,step,offset:", domain, step, offset)
1316 fetched = post_json_api(domain, "/api/federation/instances", json.dumps({
1321 "offset" : offset - 1
1322 }), {"Origin": domain})
1324 # DEBUG: print("DEBUG: fetched():", len(fetched))
1325 if len(fetched) == 0:
1326 # DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
1328 elif len(fetched) != config["misskey_offset"]:
1329 # DEBUG: print(f"DEBUG: Fetched '{len(fetched)}' row(s) but expected: '{config['misskey_offset']}'")
1330 offset = offset + (config["misskey_offset"] - len(fetched))
1332 # DEBUG: print("DEBUG: Raising offset by step:", step)
1333 offset = offset + step
1335 for instance in fetched:
1337 if instance["isSuspended"]:
1338 blocks["suspended"].append(
1340 "domain": tidyup(instance["host"]),
1341 # no reason field, nothing
1346 except BaseException as e:
1347 print("WARNING: Caught error, exiting loop:", domain, e)
1348 update_last_error(domain, e)
1353 # same shit, different asshole ("blocked" aka full suspend)
1356 # DEBUG: print("DEBUG: Sending JSON API request to domain,step,offset:", domain, step, offset)
1357 fetched = post_json_api(domain,"/api/federation/instances", json.dumps({
1362 }), {"Origin": domain})
1364 # DEBUG: print("DEBUG: Sending JSON API request to domain,step,offset:", domain, step, offset)
1365 fetched = post_json_api(domain,"/api/federation/instances", json.dumps({
1371 }), {"Origin": domain})
1373 # DEBUG: print("DEBUG: fetched():", len(fetched))
1374 if len(fetched) == 0:
1375 # DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
1377 elif len(fetched) != config["misskey_offset"]:
1378 # DEBUG: print(f"DEBUG: Fetched '{len(fetched)}' row(s) but expected: '{config['misskey_offset']}'")
1379 offset = offset + (config["misskey_offset"] - len(fetched))
1381 # DEBUG: print("DEBUG: Raising offset by step:", step)
1382 offset = offset + step
1384 for instance in fetched:
1385 if instance["isBlocked"]:
1386 blocks["blocked"].append({
1387 "domain": tidyup(instance["host"]),
1391 except BaseException as e:
1392 print("ERROR: Exception during POST:", domain, e)
1393 update_last_error(domain, e)
1397 # DEBUG: print(f"DEBUG: Updating last_instance_fetch for domain='{domain}' ...")
1398 update_last_instance_fetch(domain)
1400 # DEBUG: print("DEBUG: Returning for domain,blocked(),suspended():", domain, len(blocks["blocked"]), len(blocks["suspended"]))
1402 "reject" : blocks["blocked"],
1403 "followers_only": blocks["suspended"]
1406 def tidyup(string: str) -> str:
1407 if type(string) != str:
1408 raise ValueError(f"WARNING: string[]={type(string)} is not expected")
1410 # some retards put their blocks in variable case
1411 string = string.lower().strip()
1413 # other retards put the port
1414 string = re.sub("\:\d+$", "", string)
1416 # bigger retards put the schema in their blocklist, sometimes even without slashes
1417 string = re.sub("^https?\:(\/*)", "", string)
1419 # and trailing slash
1420 string = re.sub("\/$", "", string)
1423 string = re.sub("^\@", "", string)
1425 # the biggest retards of them all try to block individual users
1426 string = re.sub("(.+)\@", "", string)