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_nodeinfos() 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'")
324 for peer in blacklist:
330 def remove_pending_error(domain: str):
331 if type(domain) != str:
332 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
335 # Prevent updating any pending errors, nodeinfo was found
336 del pending_errors[domain]
341 def get_hash(domain: str) -> str:
342 if type(domain) != str:
343 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
345 return hashlib.sha256(domain.encode("utf-8")).hexdigest()
347 def update_last_blocked(domain: str):
348 if type(domain) != str:
349 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
351 # DEBUG: print("DEBUG: Updating last_blocked for domain", domain)
352 nodeinfos["last_blocked"] = time.time()
353 nodeinfos["last_updated"] = time.time()
355 # Running pending updated
356 # DEBUG: print(f"DEBUG: Invoking update_nodeinfos({domain}) ...")
357 update_nodeinfos(domain)
359 # DEBUG: print("DEBUG: EXIT!")
361 def has_pending_nodeinfos(domain: str) -> bool:
362 # DEBUG: print(f"DEBUG: domain='{domain}' - CALLED!")
363 if type(domain) != str:
364 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
367 for key in nodeinfos:
368 # DEBUG: print(f"DEBUG: key='{key}',domain='{domain}',nodeinfos[key]='{nodeinfos[key]}'")
369 if domain in nodeinfos[key]:
373 # DEBUG: print(f"DEBUG: has_pending='{has_pending}' - EXIT!")
376 def update_nodeinfos(domain: str):
377 if type(domain) != str:
378 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
380 # DEBUG: print(f"DEBUG: Updating nodeinfo for domain='{domain}' ...")
383 for key in nodeinfos:
384 # DEBUG: print("DEBUG: key:", key)
385 if domain in nodeinfos[key]:
386 # DEBUG: print(f"DEBUG: Adding '{nodeinfos[key][domain]}' for key='{key}' ...")
387 fields.append(nodeinfos[key][domain])
388 sql_string += f" {key} = ?,"
390 fields.append(domain)
393 raise ValueError(f"WARNING: No fields have been set, but method invoked, domain='{domain}'")
395 # DEBUG: print(f"DEBUG: sql_string='{sql_string}',fields()={len(fields)}")
396 sql = "UPDATE instances SET" + sql_string + " last_status_code = NULL, last_error_details = NULL WHERE domain = ? LIMIT 1"
397 # DEBUG: print("DEBUG: sql:", sql)
400 # DEBUG: print("DEBUG: Executing SQL:", sql)
401 cursor.execute(sql, fields)
402 # DEBUG: print(f"DEBUG: Success! (rowcount={cursor.rowcount })")
404 if cursor.rowcount == 0:
405 print("WARNING: Did not update any rows:", domain)
409 except BaseException as e:
410 print(f"ERROR: failed SQL query: domain='{domain}',sql='{sql}',exception[{type(e)}]:'{str(e)}'")
413 # DEBUG: print("DEBUG: Deleting nodeinfos for domain:", domain)
414 for key in nodeinfos:
416 # DEBUG: print("DEBUG: Deleting key:", key)
417 del nodeinfos[key][domain]
421 # DEBUG: print("DEBUG: EXIT!")
423 def log_error(domain: str, res: any):
424 # DEBUG: print("DEBUG: domain,res[]:", domain, type(res))
425 if type(domain) != str:
426 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
429 # DEBUG: print("DEBUG: BEFORE res[]:", type(res))
430 if isinstance(res, BaseException) or isinstance(res, json.JSONDecodeError):
433 # DEBUG: print("DEBUG: AFTER res[]:", type(res))
435 cursor.execute("INSERT INTO error_log (domain, error_code, error_message, created) VALUES (?, 999, ?, ?)",[
441 cursor.execute("INSERT INTO error_log (domain, error_code, error_message, created) VALUES (?, ?, ?, ?)",[
448 # Cleanup old entries
449 # DEBUG: print(f"DEBUG: Purging old records (distance: {config['error_log_cleanup']})")
450 cursor.execute("DELETE FROM error_log WHERE created < ?", [time.time() - config["error_log_cleanup"]])
451 except BaseException as e:
452 print(f"ERROR: failed SQL query: domain='{domain}',exception[{type(e)}]:'{str(e)}'")
455 # DEBUG: print("DEBUG: EXIT!")
457 def update_last_error(domain: str, res: any):
458 # DEBUG: print("DEBUG: domain,res[]:", domain, type(res))
459 if type(domain) != str:
460 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
462 # DEBUG: print("DEBUG: BEFORE res[]:", type(res))
463 if isinstance(res, BaseException) or isinstance(res, json.JSONDecodeError):
466 # DEBUG: print("DEBUG: AFTER res[]:", type(res))
467 nodeinfos["last_updated"][domain] = time.time()
469 # DEBUG: print(f"DEBUG: Setting last_error_details='{res}'");
470 nodeinfos["last_status_code"][domain] = 999
471 nodeinfos["last_error_details"][domain] = res
473 # DEBUG: print(f"DEBUG: Setting last_error_details='{res.reason}'");
474 nodeinfos["last_status_code"][domain] = res.status_code
475 nodeinfos["last_error_details"][domain] = res.reason
477 # Running pending updated
478 # DEBUG: print(f"DEBUG: Invoking update_nodeinfos({domain}) ...")
479 update_nodeinfos(domain)
481 log_error(domain, res)
483 # DEBUG: print("DEBUG: EXIT!")
485 def update_last_instance_fetch(domain: str):
486 if type(domain) != str:
487 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
489 # DEBUG: print("DEBUG: Updating last_instance_fetch for domain:", domain)
490 nodeinfos["last_instance_fetch"][domain] = time.time()
491 nodeinfos["last_updated"][domain] = time.time()
493 # Running pending updated
494 # DEBUG: print(f"DEBUG: Invoking update_nodeinfos({domain}) ...")
495 update_nodeinfos(domain)
497 # DEBUG: print("DEBUG: EXIT!")
499 def update_last_nodeinfo(domain: str):
500 if type(domain) != str:
501 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
503 # DEBUG: print("DEBUG: Updating last_nodeinfo for domain:", domain)
504 nodeinfos["last_nodeinfo"][domain] = time.time()
505 nodeinfos["last_updated"][domain] = time.time()
507 # Running pending updated
508 # DEBUG: print(f"DEBUG: Invoking update_nodeinfos({domain}) ...")
509 update_nodeinfos(domain)
511 # DEBUG: print("DEBUG: EXIT!")
513 def get_peers(domain: str, software: str) -> list:
514 if type(domain) != str:
515 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
516 elif type(software) != str and software != None:
517 raise ValueError(f"WARNING: software[]={type(software)} is not 'str'")
519 # DEBUG: print(f"DEBUG: domain='{domain}',software='{software}' - CALLED!")
522 if software == "misskey":
523 # DEBUG: print(f"DEBUG: domain='{domain}' is misskey, sending API POST request ...")
525 step = config["misskey_offset"]
527 # iterating through all "suspended" (follow-only in its terminology)
528 # instances page-by-page, since that troonware doesn't support
529 # sending them all at once
531 # DEBUG: print(f"DEBUG: Fetching offset='{offset}' from '{domain}' ...")
533 fetched = post_json_api(domain, "/api/federation/instances", json.dumps({
537 }), {"Origin": domain})
539 fetched = post_json_api(domain, "/api/federation/instances", json.dumps({
544 }), {"Origin": domain})
546 # DEBUG: print("DEBUG: fetched():", len(fetched))
547 if len(fetched) == 0:
548 # DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
550 elif len(fetched) != config["misskey_offset"]:
551 # DEBUG: print(f"DEBUG: Fetched '{len(fetched)}' row(s) but expected: '{config['misskey_offset']}'")
552 offset = offset + (config["misskey_offset"] - len(fetched))
554 # DEBUG: print("DEBUG: Raising offset by step:", step)
555 offset = offset + step
558 # DEBUG: print(f"DEBUG: fetched({len(fetched)})[]={type(fetched)}")
559 if isinstance(fetched, dict) and "error" in fetched and "message" in fetched["error"]:
560 print(f"WARNING: post_json_api() returned error: {fetched['error']['message']}")
561 update_last_error(domain, fetched["error"]["message"])
565 # DEBUG: print(f"DEBUG: row():{len(row)}")
566 if not "host" in row:
567 print(f"WARNING: row()={len(row)} does not contain element 'host': {row},domain='{domain}'")
569 elif type(row["host"]) != str:
570 print(f"WARNING: row[host][]={type(row['host'])} is not 'str'")
572 elif is_blacklisted(row["host"]):
573 # DEBUG: print(f"DEBUG: row[host]='{row['host']}' is blacklisted. domain='{domain}'")
576 # DEBUG: print(f"DEBUG: Adding peer: '{row['host']}'")
577 peers.append(row["host"])
579 # DEBUG: print(f"DEBUG: Adding '{len(peers)}' for domain='{domain}'")
580 nodeinfos["total_peers"][domain] = len(peers)
582 # DEBUG: print(f"DEBUG: Updating last_instance_fetch for domain='{domain}' ...")
583 update_last_instance_fetch(domain)
585 # DEBUG: print("DEBUG: Returning peers[]:", type(peers))
587 elif software == "lemmy":
588 # DEBUG: print(f"DEBUG: domain='{domain}' is Lemmy, fetching JSON ...")
590 res = reqto.get(f"https://{domain}/api/v3/site", headers=api_headers, timeout=(config["connection_timeout"], config["read_timeout"]))
593 # DEBUG: print(f"DEBUG: res.ok={res.ok},res.status_code='{res.status_code}',data[]='{type(data)}'")
594 if not res.ok or res.status_code >= 400:
595 print("WARNING: Could not reach any JSON API:", domain)
596 update_last_error(domain, res)
597 elif res.ok and isinstance(data, list):
598 # DEBUG: print(f"DEBUG: domain='{domain}' returned a list: '{data}'")
600 elif "federated_instances" in data:
601 # DEBUG: print(f"DEBUG: Found federated_instances for domain='{domain}'")
602 peers = peers + add_peers(data["federated_instances"])
603 # DEBUG: print("DEBUG: Added instance(s) to peers")
605 print("WARNING: JSON response does not contain 'federated_instances':", domain)
606 update_last_error(domain, res)
608 except BaseException as e:
609 print(f"WARNING: Exception during fetching JSON: domain='{domain}',exception[{type(e)}]:'{str(e)}'")
611 # DEBUG: print(f"DEBUG: Adding '{len(peers)}' for domain='{domain}'")
612 nodeinfos["total_peers"][domain] = len(peers)
614 # DEBUG: print(f"DEBUG: Updating last_instance_fetch for domain='{domain}' ...")
615 update_last_instance_fetch(domain)
617 # DEBUG: print("DEBUG: Returning peers[]:", type(peers))
619 elif software == "peertube":
620 # DEBUG: print(f"DEBUG: domain='{domain}' is a PeerTube, fetching JSON ...")
623 for mode in ["followers", "following"]:
624 # DEBUG: print(f"DEBUG: domain='{domain}',mode='{mode}'")
627 res = reqto.get(f"https://{domain}/api/v1/server/{mode}?start={start}&count=100", headers=api_headers, timeout=(config["connection_timeout"], config["read_timeout"]))
630 # DEBUG: print(f"DEBUG: res.ok={res.ok},res.status_code='{res.status_code}',data[]='{type(data)}'")
631 if res.ok and isinstance(data, dict):
632 # DEBUG: print("DEBUG: Success, data:", len(data))
634 # DEBUG: print(f"DEBUG: Found {len(data['data'])} record(s).")
635 for record in data["data"]:
636 # DEBUG: print(f"DEBUG: record()={len(record)}")
637 if mode in record and "host" in record[mode]:
638 # DEBUG: print(f"DEBUG: Found host={record[mode]['host']}, adding ...")
639 peers.append(record[mode]["host"])
641 print(f"WARNING: record from '{domain}' has no '{mode}' or 'host' record: {record}")
643 if len(data["data"]) < 100:
644 # DEBUG: print("DEBUG: Reached end of JSON response:", domain)
647 # Continue with next row
650 except BaseException as e:
651 print(f"WARNING: Exception during fetching JSON: domain='{domain}',exception[{type(e)}]:'{str(e)}'")
653 # DEBUG: print(f"DEBUG: Adding '{len(peers)}' for domain='{domain}'")
654 nodeinfos["total_peers"][domain] = len(peers)
656 # DEBUG: print(f"DEBUG: Updating last_instance_fetch for domain='{domain}' ...")
657 update_last_instance_fetch(domain)
659 # DEBUG: print("DEBUG: Returning peers[]:", type(peers))
662 # DEBUG: print(f"DEBUG: Fetching get_peers_url='{get_peers_url}' from '{domain}' ...")
664 res = reqto.get(f"https://{domain}{get_peers_url}", headers=api_headers, timeout=(config["connection_timeout"], config["read_timeout"]))
667 # DEBUG: print(f"DEBUG: res.ok={res.ok},res.status_code={res.status_code},data[]='{type(data)}'")
668 if not res.ok or res.status_code >= 400:
669 # DEBUG: print(f"DEBUG: Was not able to fetch '{get_peers_url}', trying alternative ...")
670 res = reqto.get(f"https://{domain}/api/v3/site", headers=api_headers, timeout=(config["connection_timeout"], config["read_timeout"]))
673 # DEBUG: print(f"DEBUG: res.ok={res.ok},res.status_code={res.status_code},data[]='{type(data)}'")
674 if not res.ok or res.status_code >= 400:
675 print("WARNING: Could not reach any JSON API:", domain)
676 update_last_error(domain, res)
677 elif res.ok and isinstance(data, list):
678 # DEBUG: print(f"DEBUG: domain='{domain}' returned a list: '{data}'")
680 elif "federated_instances" in data:
681 # DEBUG: print(f"DEBUG: Found federated_instances for domain='{domain}'")
682 peers = peers + add_peers(data["federated_instances"])
683 # DEBUG: print("DEBUG: Added instance(s) to peers")
685 print("WARNING: JSON response does not contain 'federated_instances':", domain)
686 update_last_error(domain, res)
688 # DEBUG: print("DEBUG: Querying API was successful:", domain, len(data))
691 except BaseException as e:
692 print("WARNING: Some error during get():", domain, e)
693 update_last_error(domain, e)
695 # DEBUG: print(f"DEBUG: Adding '{len(peers)}' for domain='{domain}'")
696 nodeinfos["total_peers"][domain] = len(peers)
698 # DEBUG: print(f"DEBUG: Updating last_instance_fetch for domain='{domain}' ...")
699 update_last_instance_fetch(domain)
701 # DEBUG: print("DEBUG: Returning peers[]:", type(peers))
704 def post_json_api(domain: str, path: str, parameter: str, extra_headers: dict = {}) -> dict:
705 if type(domain) != str:
706 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
707 elif type(path) != str:
708 raise ValueError(f"WARNING: path[]={type(path)} is not 'str'")
709 elif type(parameter) != str:
710 raise ValueError(f"WARNING: parameter[]={type(parameter)} is not 'str'")
712 # DEBUG: print("DEBUG: Sending POST to domain,path,parameter:", domain, path, parameter, extra_headers)
715 res = reqto.post(f"https://{domain}{path}", data=parameter, headers={**api_headers, **extra_headers}, timeout=(config["connection_timeout"], config["read_timeout"]))
718 # DEBUG: print(f"DEBUG: res.ok={res.ok},res.status_code={res.status_code},data[]='{type(data)}'")
719 if not res.ok or res.status_code >= 400:
720 print(f"WARNING: Cannot query JSON API: domain='{domain}',path='{path}',parameter()={len(parameter)},res.status_code='{res.status_code}',data[]='{type(data)}'")
721 update_last_error(domain, res)
723 except BaseException as e:
724 print(f"WARNING: Some error during post(): domain='{domain}',path='{path}',parameter()={len(parameter)},exception[{type(e)}]:'{str(e)}'")
726 # DEBUG: print(f"DEBUG: Returning data({len(data)})=[]:{type(data)}")
729 def fetch_nodeinfo(domain: str, path: str = None) -> list:
730 if type(domain) != str:
731 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
733 # DEBUG: print("DEBUG: Fetching nodeinfo from domain,path:", domain, path)
735 nodeinfo = fetch_wellknown_nodeinfo(domain)
736 # DEBUG: print("DEBUG: nodeinfo:", nodeinfo)
738 if len(nodeinfo) > 0:
739 # DEBUG: print("DEBUG: Returning auto-discovered nodeinfo:", len(nodeinfo))
743 f"https://{domain}/nodeinfo/2.1.json",
744 f"https://{domain}/nodeinfo/2.1",
745 f"https://{domain}/nodeinfo/2.0.json",
746 f"https://{domain}/nodeinfo/2.0",
747 f"https://{domain}/nodeinfo/1.0",
748 f"https://{domain}/api/v1/instance"
752 for request in requests:
753 if path != None and path != "" and request != path:
754 # DEBUG: print(f"DEBUG: path='{path}' does not match request='{request}' - SKIPPED!")
758 # DEBUG: print("DEBUG: Fetching request:", request)
759 res = reqto.get(request, headers=api_headers, timeout=(config["connection_timeout"], config["read_timeout"]))
762 # DEBUG: print(f"DEBUG: res.ok={res.ok},res.status_code={res.status_code},data[]='{type(data)}'")
763 if res.ok and isinstance(data, dict):
764 # DEBUG: print("DEBUG: Success:", request)
765 nodeinfos["detection_mode"][domain] = "STATIC_CHECK"
766 nodeinfos["nodeinfo_url"][domain] = request
768 elif res.ok and isinstance(data, list):
769 # DEBUG: print(f"DEBUG: domain='{domain}' returned a list: '{data}'")
771 elif not res.ok or res.status_code >= 400:
772 print("WARNING: Failed fetching nodeinfo from domain:", domain)
773 update_last_error(domain, res)
776 except BaseException as e:
777 # DEBUG: print("DEBUG: Cannot fetch API request:", request)
778 update_last_error(domain, e)
781 # DEBUG: print("DEBUG: Returning data[]:", type(data))
784 def fetch_wellknown_nodeinfo(domain: str) -> list:
785 if type(domain) != str:
786 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
788 # DEBUG: print("DEBUG: Fetching .well-known info for domain:", domain)
792 res = reqto.get(f"https://{domain}/.well-known/nodeinfo", headers=api_headers, timeout=(config["connection_timeout"], config["read_timeout"]))
795 # DEBUG: print("DEBUG: domain,res.ok,data[]:", domain, res.ok, type(data))
796 if res.ok and isinstance(data, dict):
798 # DEBUG: print("DEBUG: Found entries:", len(nodeinfo), domain)
799 if "links" in nodeinfo:
800 # DEBUG: print("DEBUG: Found links in nodeinfo():", len(nodeinfo["links"]))
801 for link in nodeinfo["links"]:
802 # DEBUG: print("DEBUG: rel,href:", link["rel"], link["href"])
803 if link["rel"] in nodeinfo_identifier:
804 # DEBUG: print("DEBUG: Fetching nodeinfo from:", link["href"])
805 res = reqto.get(link["href"])
808 # DEBUG: print("DEBUG: href,res.ok,res.status_code:", link["href"], res.ok, res.status_code)
809 if res.ok and isinstance(data, dict):
810 # DEBUG: print("DEBUG: Found JSON nodeinfo():", len(data))
811 nodeinfos["detection_mode"][domain] = "AUTO_DISCOVERY"
812 nodeinfos["nodeinfo_url"][domain] = link["href"]
815 print("WARNING: Unknown 'rel' value:", domain, link["rel"])
817 print("WARNING: nodeinfo does not contain 'links':", domain)
819 except BaseException as e:
820 print("WARNING: Failed fetching .well-known info:", domain)
821 update_last_error(domain, e)
824 # DEBUG: print("DEBUG: Returning data[]:", type(data))
827 def fetch_generator_from_path(domain: str, path: str = "/") -> str:
828 if type(domain) != str:
829 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
830 elif type(path) != str:
831 raise ValueError(f"WARNING: path[]={type(path)} is not 'str'")
833 # DEBUG: print(f"DEBUG: domain='{domain}',path='{path}' - CALLED!")
837 # DEBUG: print(f"DEBUG: Fetching path='{path}' from '{domain}' ...")
838 res = reqto.get(f"https://{domain}{path}", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
840 # DEBUG: print("DEBUG: domain,res.ok,res.status_code,res.text[]:", domain, res.ok, res.status_code, type(res.text))
841 if res.ok and res.status_code < 300 and len(res.text) > 0:
842 # DEBUG: print("DEBUG: Search for <meta name='generator'>:", domain)
843 doc = bs4.BeautifulSoup(res.text, "html.parser")
845 # DEBUG: print("DEBUG: doc[]:", type(doc))
846 generator = doc.find("meta", {"name": "generator"})
847 site_name = doc.find("meta", {"property": "og:site_name"})
849 # DEBUG: print(f"DEBUG: generator='{generator}',site_name='{site_name}'")
850 if isinstance(generator, bs4.element.Tag):
851 # DEBUG: print("DEBUG: Found generator meta tag:", domain)
852 software = tidyup(generator.get("content"))
853 print(f"INFO: domain='{domain}' is generated by '{software}'")
854 nodeinfos["detection_mode"][domain] = "GENERATOR"
855 remove_pending_error(domain)
856 elif isinstance(site_name, bs4.element.Tag):
857 # DEBUG: print("DEBUG: Found property=og:site_name:", domain)
858 sofware = tidyup(site_name.get("content"))
859 print(f"INFO: domain='{domain}' has og:site_name='{software}'")
860 nodeinfos["detection_mode"][domain] = "SITE_NAME"
861 remove_pending_error(domain)
863 except BaseException as e:
864 # DEBUG: print(f"DEBUG: Cannot fetch / from '{domain}':", e)
865 update_last_error(domain, e)
868 # DEBUG: print(f"DEBUG: software[]={type(software)}")
869 if type(software) is str and software == "":
870 # DEBUG: print(f"DEBUG: Corrected empty string to None for software of domain='{domain}'")
872 elif type(software) is str and ("." in software or " " in software):
873 # DEBUG: print(f"DEBUG: software='{software}' may contain a version number, domain='{domain}', removing it ...")
874 software = remove_version(software)
876 # DEBUG: print(f"DEBUG: software[]={type(software)}")
877 if type(software) is str and " powered by " in software:
878 # DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it")
879 software = remove_version(strip_powered_by(software))
880 elif type(software) is str and " hosted on " in software:
881 # DEBUG: print(f"DEBUG: software='{software}' has 'hosted on' in it")
882 software = remove_version(strip_hosted_on(software))
883 elif type(software) is str and " by " in software:
884 # DEBUG: print(f"DEBUG: software='{software}' has ' by ' in it")
885 software = strip_until(software, " by ")
886 elif type(software) is str and " see " in software:
887 # DEBUG: print(f"DEBUG: software='{software}' has ' see ' in it")
888 software = strip_until(software, " see ")
890 # DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
893 def determine_software(domain: str, path: str = None) -> str:
894 if type(domain) != str:
895 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
897 # DEBUG: print("DEBUG: Determining software for domain,path:", domain, path)
900 # DEBUG: print(f"DEBUG: Fetching nodeinfo from '{domain}' ...")
901 data = fetch_nodeinfo(domain, path)
903 # DEBUG: print("DEBUG: data[]:", type(data))
904 if not isinstance(data, dict) or len(data) == 0:
905 # DEBUG: print("DEBUG: Could not determine software type:", domain)
906 return fetch_generator_from_path(domain)
908 # DEBUG: print("DEBUG: data():", len(data), data)
909 if "status" in data and data["status"] == "error" and "message" in data:
910 print("WARNING: JSON response is an error:", data["message"])
911 update_last_error(domain, data["message"])
912 return fetch_generator_from_path(domain)
913 elif "message" in data:
914 print("WARNING: JSON response contains only a message:", data["message"])
915 update_last_error(domain, data["message"])
916 return fetch_generator_from_path(domain)
917 elif "software" not in data or "name" not in data["software"]:
918 # DEBUG: print(f"DEBUG: JSON response from domain='{domain}' does not include [software][name], fetching / ...")
919 software = fetch_generator_from_path(domain)
921 # DEBUG: print(f"DEBUG: Generator for domain='{domain}' is: {software}, EXIT!")
924 software = tidyup(data["software"]["name"])
926 # DEBUG: print("DEBUG: sofware after tidyup():", software)
927 if software in ["akkoma", "rebased"]:
928 # DEBUG: print("DEBUG: Setting pleroma:", domain, software)
930 elif software in ["hometown", "ecko"]:
931 # DEBUG: print("DEBUG: Setting mastodon:", domain, software)
932 software = "mastodon"
933 elif software in ["calckey", "groundpolis", "foundkey", "cherrypick", "meisskey"]:
934 # DEBUG: print("DEBUG: Setting misskey:", domain, software)
936 elif software.find("/") > 0:
937 print("WARNING: Spliting of slash:", software)
938 software = software.split("/")[-1];
939 elif software.find("|") > 0:
940 print("WARNING: Spliting of pipe:", software)
941 software = tidyup(software.split("|")[0]);
942 elif "powered by" in software:
943 # DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it")
944 software = strip_powered_by(software)
945 elif type(software) is str and " by " in software:
946 # DEBUG: print(f"DEBUG: software='{software}' has ' by ' in it")
947 software = strip_until(software, " by ")
948 elif type(software) is str and " see " in software:
949 # DEBUG: print(f"DEBUG: software='{software}' has ' see ' in it")
950 software = strip_until(software, " see ")
952 # DEBUG: print(f"DEBUG: software[]={type(software)}")
954 print("WARNING: tidyup() left no software name behind:", domain)
957 # DEBUG: print(f"DEBUG: software[]={type(software)}")
958 if str(software) == "":
959 # DEBUG: print(f"DEBUG: software for '{domain}' was not detected, trying generator ...")
960 software = fetch_generator_from_path(domain)
961 elif len(str(software)) > 0 and ("." in software or " " in software):
962 # DEBUG: print(f"DEBUG: software='{software}' may contain a version number, domain='{domain}', removing it ...")
963 software = remove_version(software)
965 # DEBUG: print(f"DEBUG: software[]={type(software)}")
966 if type(software) is str and "powered by" in software:
967 # DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it")
968 software = remove_version(strip_powered_by(software))
970 # DEBUG: print("DEBUG: Returning domain,software:", domain, software)
973 def update_block_reason(reason: str, blocker: str, blocked: str, block_level: str):
974 # DEBUG: print("DEBUG: Updating block reason:", reason, blocker, blocked, block_level)
977 "UPDATE blocks SET reason = ?, last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ? AND reason = ''",
987 # DEBUG: print(f"DEBUG: cursor.rowcount={cursor.rowcount}")
988 if cursor.rowcount == 0:
989 print("WARNING: Did not update any rows:", domain)
991 except BaseException as e:
992 print(f"ERROR: failed SQL query: reason='{reason}',blocker='{blocker}',blocked='{blocked}',block_level='{block_level}',sql='{sql}',exception[{type(e)}]:'{str(e)}'")
995 # DEBUG: print("DEBUG: EXIT!")
997 def update_last_seen(blocker: str, blocked: str, block_level: str):
998 # DEBUG: print("DEBUG: Updating last_seen for:", blocker, blocked, block_level)
1001 "UPDATE blocks SET last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ?",
1010 if cursor.rowcount == 0:
1011 print("WARNING: Did not update any rows:", domain)
1013 except BaseException as e:
1014 print(f"ERROR: failed SQL query: last_seen='{last_seen}',blocker='{blocker}',blocked='{blocked}',block_level='{block_level}',exception[{type(e)}]:'{str(e)}'")
1017 # DEBUG: print("DEBUG: EXIT!")
1019 def block_instance(blocker: str, blocked: str, reason: str, block_level: str):
1020 # DEBUG: print("DEBUG: blocker,blocked,reason,block_level:", blocker, blocked, reason, block_level)
1021 if not validators.domain(blocker.split("/")[0]):
1022 print("WARNING: Bad blocker:", blocker)
1024 elif not validators.domain(blocked.split("/")[0]):
1025 print("WARNING: Bad blocked:", blocked)
1028 print("INFO: New block:", blocker, blocked, reason, block_level, first_seen, last_seen)
1031 "INSERT INTO blocks (blocker, blocked, reason, block_level, first_seen, last_seen) VALUES(?, ?, ?, ?, ?, ?)",
1042 except BaseException as e:
1043 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)}'")
1046 # DEBUG: print("DEBUG: EXIT!")
1048 def is_instance_registered(domain: str) -> bool:
1049 if type(domain) != str:
1050 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
1052 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}' - CALLED!")
1053 if not is_cache_initialized("is_registered"):
1054 # NOISY-DEBUG: print(f"DEBUG: Cache for 'is_registered' not initialized, fetching all rows ...")
1056 cursor.execute("SELECT domain FROM instances")
1059 set_all_cache_key("is_registered", cursor.fetchall(), True)
1060 except BaseException as e:
1061 print(f"ERROR: failed SQL query: domain='{domain}',exception[{type(e)}]:'{str(e)}'")
1065 registered = is_cache_key_set("is_registered", domain)
1067 # NOISY-DEBUG: print(f"DEBUG: registered='{registered}' - EXIT!")
1070 def add_instance(domain: str, origin: str, originator: str, path: str = None):
1071 if type(domain) != str:
1072 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
1073 elif type(origin) != str and origin != None:
1074 raise ValueError(f"WARNING: origin[]={type(origin)} is not 'str'")
1075 elif type(originator) != str:
1076 raise ValueError(f"WARNING: originator[]={type(originator)} is not 'str'")
1078 # DEBUG: print("DEBUG: domain,origin,originator,path:", domain, origin, originator, path)
1079 if not validators.domain(domain.split("/")[0]):
1080 print("WARNING: Bad domain name:", domain)
1082 elif origin is not None and not validators.domain(origin.split("/")[0]):
1083 print("WARNING: Bad origin name:", origin)
1086 software = determine_software(domain, path)
1087 # DEBUG: print("DEBUG: Determined software:", software)
1089 print(f"INFO: Adding instance domain='{domain}' (origin='{origin}',software='{software}')")
1092 "INSERT INTO instances (domain, origin, originator, hash, software, first_seen) VALUES (?, ?, ?, ?, ?, ?)",
1103 set_cache_key("is_registered", domain, True)
1105 if has_pending_nodeinfos(domain):
1106 # DEBUG: print(f"DEBUG: domain='{domain}' has pending nodeinfo being updated ...")
1107 update_nodeinfos(domain)
1108 remove_pending_error(domain)
1110 if domain in pending_errors:
1111 # DEBUG: print("DEBUG: domain has pending error being updated:", domain)
1112 update_last_error(domain, pending_errors[domain])
1113 remove_pending_error(domain)
1115 except BaseException as e:
1116 print(f"ERROR: failed SQL query: domain='{domain}',exception[{type(e)}]:'{str(e)}'")
1119 # DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain)
1120 update_last_nodeinfo(domain)
1122 # DEBUG: print("DEBUG: EXIT!")
1124 def send_bot_post(instance: str, blocks: dict):
1125 message = instance + " has blocked the following instances:\n\n"
1128 if len(blocks) > 20:
1130 blocks = blocks[0 : 19]
1132 for block in blocks:
1133 if block["reason"] == None or block["reason"] == '':
1134 message = message + block["blocked"] + " with unspecified reason\n"
1136 if len(block["reason"]) > 420:
1137 block["reason"] = block["reason"][0:419] + "[…]"
1139 message = message + block["blocked"] + ' for "' + block["reason"].replace("@", "@\u200b") + '"\n'
1142 message = message + "(the list has been truncated to the first 20 entries)"
1144 botheaders = {**api_headers, **{"Authorization": "Bearer " + config["bot_token"]}}
1147 f"{config['bot_instance']}/api/v1/statuses",
1150 "visibility" : config['bot_visibility'],
1151 "content_type": "text/plain"
1159 def get_mastodon_blocks(domain: str) -> dict:
1160 if type(domain) != str:
1161 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
1163 # DEBUG: print("DEBUG: Fetching mastodon blocks from domain:", domain)
1165 "Suspended servers": [],
1166 "Filtered media" : [],
1167 "Limited servers" : [],
1168 "Silenced servers" : [],
1172 doc = bs4.BeautifulSoup(
1173 reqto.get(f"https://{domain}/about/more", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
1176 except BaseException as e:
1177 print("ERROR: Cannot fetch from domain:", domain, e)
1178 update_last_error(domain, e)
1181 for header in doc.find_all("h3"):
1182 header_text = tidyup(header.text)
1184 if header_text in language_mapping:
1185 # DEBUG: print(f"DEBUG: header_text='{header_text}'")
1186 header_text = language_mapping[header_text]
1188 if header_text in blocks or header_text.lower() in blocks:
1189 # replaced find_next_siblings with find_all_next to account for instances that e.g. hide lists in dropdown menu
1190 for line in header.find_all_next("table")[0].find_all("tr")[1:]:
1191 blocks[header_text].append(
1193 "domain": tidyup(line.find("span").text),
1194 "hash" : tidyup(line.find("span")["title"][9:]),
1195 "reason": tidyup(line.find_all("td")[1].text),
1199 # DEBUG: print("DEBUG: Returning blocks for domain:", domain)
1201 "reject" : blocks["Suspended servers"],
1202 "media_removal" : blocks["Filtered media"],
1203 "followers_only": blocks["Limited servers"] + blocks["Silenced servers"],
1206 def get_friendica_blocks(domain: str) -> dict:
1207 if type(domain) != str:
1208 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
1210 # DEBUG: print("DEBUG: Fetching friendica blocks from domain:", domain)
1214 doc = bs4.BeautifulSoup(
1215 reqto.get(f"https://{domain}/friendica", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
1218 except BaseException as e:
1219 print("WARNING: Failed to fetch /friendica from domain:", domain, e)
1220 update_last_error(domain, e)
1223 blocklist = doc.find(id="about_blocklist")
1225 # Prevents exceptions:
1226 if blocklist is None:
1227 # DEBUG: print("DEBUG:Instance has no block list:", domain)
1230 for line in blocklist.find("table").find_all("tr")[1:]:
1231 # DEBUG: print(f"DEBUG: line='{line}'")
1233 "domain": tidyup(line.find_all("td")[0].text),
1234 "reason": tidyup(line.find_all("td")[1].text)
1237 # DEBUG: print("DEBUG: Returning blocks() for domain:", domain, len(blocks))
1242 def get_misskey_blocks(domain: str) -> dict:
1243 if type(domain) != str:
1244 raise ValueError(f"WARNING: domain[]={type(domain)} is not 'str'")
1246 # DEBUG: print("DEBUG: Fetching misskey blocks from domain:", domain)
1253 step = config["misskey_offset"]
1255 # iterating through all "suspended" (follow-only in its terminology)
1256 # instances page-by-page, since that troonware doesn't support
1257 # sending them all at once
1259 # DEBUG: print(f"DEBUG: Fetching offset='{offset}' from '{domain}' ...")
1261 # DEBUG: print("DEBUG: Sending JSON API request to domain,step,offset:", domain, step, offset)
1262 fetched = post_json_api(domain, "/api/federation/instances", json.dumps({
1267 }), {"Origin": domain})
1269 # DEBUG: print("DEBUG: Sending JSON API request to domain,step,offset:", domain, step, offset)
1270 fetched = post_json_api(domain, "/api/federation/instances", json.dumps({
1275 "offset" : offset - 1
1276 }), {"Origin": domain})
1278 # DEBUG: print("DEBUG: fetched():", len(fetched))
1279 if len(fetched) == 0:
1280 # DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
1282 elif len(fetched) != config["misskey_offset"]:
1283 # DEBUG: print(f"DEBUG: Fetched '{len(fetched)}' row(s) but expected: '{config['misskey_offset']}'")
1284 offset = offset + (config["misskey_offset"] - len(fetched))
1286 # DEBUG: print("DEBUG: Raising offset by step:", step)
1287 offset = offset + step
1289 for instance in fetched:
1291 if instance["isSuspended"]:
1292 blocks["suspended"].append(
1294 "domain": tidyup(instance["host"]),
1295 # no reason field, nothing
1300 except BaseException as e:
1301 print("WARNING: Caught error, exiting loop:", domain, e)
1302 update_last_error(domain, e)
1307 # same shit, different asshole ("blocked" aka full suspend)
1310 # DEBUG: print("DEBUG: Sending JSON API request to domain,step,offset:", domain, step, offset)
1311 fetched = post_json_api(domain,"/api/federation/instances", json.dumps({
1316 }), {"Origin": domain})
1318 # DEBUG: print("DEBUG: Sending JSON API request to domain,step,offset:", domain, step, offset)
1319 fetched = post_json_api(domain,"/api/federation/instances", json.dumps({
1325 }), {"Origin": domain})
1327 # DEBUG: print("DEBUG: fetched():", len(fetched))
1328 if len(fetched) == 0:
1329 # DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
1331 elif len(fetched) != config["misskey_offset"]:
1332 # DEBUG: print(f"DEBUG: Fetched '{len(fetched)}' row(s) but expected: '{config['misskey_offset']}'")
1333 offset = offset + (config["misskey_offset"] - len(fetched))
1335 # DEBUG: print("DEBUG: Raising offset by step:", step)
1336 offset = offset + step
1338 for instance in fetched:
1339 if instance["isBlocked"]:
1340 blocks["blocked"].append({
1341 "domain": tidyup(instance["host"]),
1345 except BaseException as e:
1346 print("ERROR: Exception during POST:", domain, e)
1347 update_last_error(domain, e)
1351 # DEBUG: print(f"DEBUG: Updating last_instance_fetch for domain='{domain}' ...")
1352 update_last_instance_fetch(domain)
1354 # DEBUG: print("DEBUG: Returning for domain,blocked(),suspended():", domain, len(blocks["blocked"]), len(blocks["suspended"]))
1356 "reject" : blocks["blocked"],
1357 "followers_only": blocks["suspended"]
1360 def tidyup(string: str) -> str:
1361 if type(string) != str:
1362 raise ValueError(f"WARNING: string[]={type(string)} is not expected")
1364 # some retards put their blocks in variable case
1365 string = string.lower().strip()
1367 # other retards put the port
1368 string = re.sub("\:\d+$", "", string)
1370 # bigger retards put the schema in their blocklist, sometimes even without slashes
1371 string = re.sub("^https?\:(\/*)", "", string)
1373 # and trailing slash
1374 string = re.sub("\/$", "", string)
1377 string = re.sub("^\@", "", string)
1379 # the biggest retards of them all try to block individual users
1380 string = re.sub("(.+)\@", "", string)