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/>.
29 from fba import config
30 from fba import instances
32 # Don't check these, known trolls/flooders/testing/developing
34 # Floods network with fake nodes as "research" project
35 "activitypub-troll.cf",
41 "social.shrimpcam.pw",
43 "mastotroll.netz.org",
44 # Testing/developing installations
47 "misskeytest.chn.moe",
50 # Array with pending errors needed to be written to database
54 # "rel" identifiers (no real URLs)
55 nodeinfo_identifier = [
56 "https://nodeinfo.diaspora.software/ns/schema/2.1",
57 "https://nodeinfo.diaspora.software/ns/schema/2.0",
58 "https://nodeinfo.diaspora.software/ns/schema/1.1",
59 "https://nodeinfo.diaspora.software/ns/schema/1.0",
60 "http://nodeinfo.diaspora.software/ns/schema/2.1",
61 "http://nodeinfo.diaspora.software/ns/schema/2.0",
62 "http://nodeinfo.diaspora.software/ns/schema/1.1",
63 "http://nodeinfo.diaspora.software/ns/schema/1.0",
66 # HTTP headers for non-API requests
68 "User-Agent": config.get("useragent"),
71 # HTTP headers for API requests
73 "User-Agent": config.get("useragent"),
74 "Content-Type": "application/json",
79 "Silenced instances" : "Silenced servers",
80 "Suspended instances" : "Suspended servers",
81 "Limited instances" : "Limited servers",
82 # Mappuing German -> English
83 "Gesperrte Server" : "Suspended servers",
84 "Gefilterte Medien" : "Filtered media",
85 "Stummgeschaltete Server" : "Silenced servers",
87 "停止済みのサーバー" : "Suspended servers",
88 "制限中のサーバー" : "Limited servers",
89 "メディアを拒否しているサーバー": "Filtered media",
90 "サイレンス済みのサーバー" : "Silenced servers",
92 "שרתים מושעים" : "Suspended servers",
93 "מדיה מסוננת" : "Filtered media",
94 "שרתים מוגבלים" : "Silenced servers",
96 "Serveurs suspendus" : "Suspended servers",
97 "Médias filtrés" : "Filtered media",
98 "Serveurs limités" : "Limited servers",
99 "Serveurs modérés" : "Limited servers",
102 # URL for fetching peers
103 get_peers_url = "/api/v1/instance/peers"
105 # Connect to database
106 connection = sqlite3.connect("blocks.db")
107 cursor = connection.cursor()
109 # Pattern instance for version numbers
111 # semantic version number (with v|V) prefix)
112 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-]+)*))?)?$"),
113 # non-sematic, e.g. 1.2.3.4
114 re.compile("^(?P<version>v|V{0,1})(\.{0,1})(?P<major>0|[1-9]\d*)\.(?P<minor>0+|[1-9]\d*)(\.(?P<patch>0+|[1-9]\d*)(\.(?P<subpatch>0|[1-9]\d*))?)$"),
115 # non-sematic, e.g. 2023-05[-dev]
116 re.compile("^(?P<year>[1-9]{1}[0-9]{3})\.(?P<month>[0-9]{2})(-dev){0,1}$"),
117 # non-semantic, e.g. abcdef0
118 re.compile("^[a-f0-9]{7}$"),
121 ##### Other functions #####
123 def is_primitive(var: any) -> bool:
124 # NOISY-DEBUG: print(f"DEBUG: var[]='{type(var)}' - CALLED!")
125 return type(var) in {int, str, float, bool} or var == None
127 def fetch_instances(domain: str, origin: str, software: str, script: str, path: str = None):
128 # DEBUG: print(f"DEBUG: domain={domain},origin={origin},software={software},path={path} - CALLED!")
129 if type(domain) != str:
130 raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
132 raise ValueError(f"Parameter 'domain' cannot be empty")
133 elif type(origin) != str and origin != None:
134 raise ValueError(f"Parameter origin[]={type(origin)} is not 'str'")
135 elif type(script) != str:
136 raise ValueError(f"Parameter script[]={type(script)} is not 'str'")
138 raise ValueError(f"Parameter 'domain' cannot be empty")
140 if not is_instance_registered(domain):
141 # DEBUG: print("DEBUG: Adding new domain:", domain, origin)
142 add_instance(domain, origin, script, path)
144 # DEBUG: print("DEBUG: Fetching instances for domain:", domain, software)
145 peerlist = get_peers(domain, software)
147 if (peerlist is None):
148 print("ERROR: Cannot fetch peers:", domain)
150 elif instances.has_pending_instance_data(domain):
151 # DEBUG: print(f"DEBUG: domain='{domain}' has pending nodeinfo data, flushing ...")
152 instances.update_instance_data(domain)
154 print(f"INFO: Checking {len(peerlist)} instances from {domain} ...")
155 for instance in peerlist:
157 # Skip "None" types as tidup() cannot parse them
160 # DEBUG: print(f"DEBUG: instance='{instance}' - BEFORE")
161 instance = tidyup_domain(instance)
162 # DEBUG: print(f"DEBUG: instance='{instance}' - AFTER")
165 print("WARNING: Empty instance after tidyup_domain(), domain:", domain)
167 elif not validators.domain(instance.split("/")[0]):
168 print(f"WARNING: Bad instance='{instance}' from domain='{domain}',origin='{origin}',software='{software}'")
170 elif is_blacklisted(instance):
171 # DEBUG: print("DEBUG: instance is blacklisted:", instance)
174 # DEBUG: print("DEBUG: Handling instance:", instance)
176 if not is_instance_registered(instance):
177 # DEBUG: print("DEBUG: Adding new instance:", instance, domain)
178 add_instance(instance, domain, sys.argv[0])
179 except BaseException as e:
180 print(f"ERROR: instance='{instance}',exception[{type(e)}]:'{str(e)}'")
183 # DEBUG: print("DEBUG: EXIT!")
185 def add_peers(rows: dict) -> list:
186 # DEBUG: print(f"DEBUG: rows()={len(rows)} - CALLED!")
188 for element in ["linked", "allowed", "blocked"]:
189 # DEBUG: print(f"DEBUG: Checking element='{element}'")
190 if element in rows and rows[element] != None:
191 # DEBUG: print(f"DEBUG: Adding {len(rows[element])} peer(s) to peers list ...")
192 for peer in rows[element]:
193 # DEBUG: print(f"DEBUG: peer='{peer}' - BEFORE!")
194 peer = tidyup_domain(peer)
196 # DEBUG: print(f"DEBUG: peer='{peer}' - AFTER!")
197 if is_blacklisted(peer):
198 # DEBUG: print(f"DEBUG: peer='{peer}' is blacklisted, skipped!")
201 # DEBUG: print(f"DEBUG: Adding peer='{peer}' ...")
204 # DEBUG: print(f"DEBUG: peers()={len(peers)} - EXIT!")
207 def remove_version(software: str) -> str:
208 # DEBUG: print(f"DEBUG: software='{software}' - CALLED!")
209 if not "." in software and " " not in software:
210 print(f"WARNING: software='{software}' does not contain a version number.")
215 temp = software.split(";")[0]
216 elif "," in software:
217 temp = software.split(",")[0]
218 elif " - " in software:
219 temp = software.split(" - ")[0]
221 # DEBUG: print(f"DEBUG: software='{software}'")
224 version = temp.split(" ")[-1]
225 elif "/" in software:
226 version = temp.split("/")[-1]
227 elif "-" in software:
228 version = temp.split("-")[-1]
230 # DEBUG: print(f"DEBUG: Was not able to find common seperator, returning untouched software='{software}'")
235 # DEBUG: print(f"DEBUG: Checking {len(patterns)} patterns ...")
236 for pattern in patterns:
238 match = pattern.match(version)
240 # DEBUG: print(f"DEBUG: match[]={type(match)}")
241 if type(match) is re.Match:
244 # DEBUG: print(f"DEBUG: version[{type(version)}]='{version}',match='{match}'")
245 if type(match) is not re.Match:
246 print(f"WARNING: version='{version}' does not match regex, leaving software='{software}' untouched.")
249 # DEBUG: print(f"DEBUG: Found valid version number: '{version}', removing it ...")
250 end = len(temp) - len(version) - 1
252 # DEBUG: print(f"DEBUG: end[{type(end)}]={end}")
253 software = temp[0:end].strip()
254 if " version" in software:
255 # DEBUG: print(f"DEBUG: software='{software}' contains word ' version'")
256 software = strip_until(software, " version")
258 # DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
261 def strip_powered_by(software: str) -> str:
262 # DEBUG: print(f"DEBUG: software='{software}' - CALLED!")
264 print(f"ERROR: Bad method call, 'software' is empty")
265 raise Exception("Parameter 'software' is empty")
266 elif not "powered by" in software:
267 print(f"WARNING: Cannot find 'powered by' in '{software}'!")
270 start = software.find("powered by ")
271 # DEBUG: print(f"DEBUG: start[{type(start)}]='{start}'")
273 software = software[start + 11:].strip()
274 # DEBUG: print(f"DEBUG: software='{software}'")
276 software = strip_until(software, " - ")
278 # DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
281 def strip_hosted_on(software: str) -> str:
282 # DEBUG: print(f"DEBUG: software='{software}' - CALLED!")
284 print(f"ERROR: Bad method call, 'software' is empty")
285 raise Exception("Parameter 'software' is empty")
286 elif not "hosted on" in software:
287 print(f"WARNING: Cannot find 'hosted on' in '{software}'!")
290 end = software.find("hosted on ")
291 # DEBUG: print(f"DEBUG: end[{type(end)}]='{end}'")
293 software = software[0, start].strip()
294 # DEBUG: print(f"DEBUG: software='{software}'")
296 software = strip_until(software, " - ")
298 # DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
301 def strip_until(software: str, until: str) -> str:
302 # DEBUG: print(f"DEBUG: software='{software}',until='{until}' - CALLED!")
304 print(f"ERROR: Bad method call, 'software' is empty")
305 raise Exception("Parameter 'software' is empty")
307 print(f"ERROR: Bad method call, 'until' is empty")
308 raise Exception("Parameter 'until' is empty")
309 elif not until in software:
310 print(f"WARNING: Cannot find '{until}' in '{software}'!")
313 # Next, strip until part
314 end = software.find(until)
316 # DEBUG: print(f"DEBUG: end[{type(end)}]='{end}'")
318 software = software[0:end].strip()
320 # DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
323 def is_blacklisted(domain: str) -> bool:
324 if type(domain) != str:
325 raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
327 raise ValueError(f"Parameter 'domain' cannot be empty")
330 for peer in blacklist:
336 def remove_pending_error(domain: str):
337 if type(domain) != str:
338 raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
340 raise ValueError(f"Parameter 'domain' cannot be empty")
343 # Prevent updating any pending errors, nodeinfo was found
344 del pending_errors[domain]
349 # DEBUG: print("DEBUG: EXIT!")
351 def get_hash(domain: str) -> str:
352 if type(domain) != str:
353 raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
355 raise ValueError(f"Parameter 'domain' cannot be empty")
357 return hashlib.sha256(domain.encode("utf-8")).hexdigest()
359 def update_last_blocked(domain: str):
360 if type(domain) != str:
361 raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
363 raise ValueError(f"Parameter 'domain' cannot be empty")
365 # DEBUG: print("DEBUG: Updating last_blocked for domain", domain)
366 instances.set("last_blocked", domain, time.time())
368 # Running pending updated
369 # DEBUG: print(f"DEBUG: Invoking instances.update_instance_data({domain}) ...")
370 instances.update_instance_data(domain)
372 # DEBUG: print("DEBUG: EXIT!")
374 def log_error(domain: str, response: requests.models.Response):
375 # DEBUG: print("DEBUG: domain,response[]:", domain, type(response))
376 if type(domain) != str:
377 raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
379 raise ValueError(f"Parameter 'domain' cannot be empty")
382 # DEBUG: print("DEBUG: BEFORE response[]:", type(response))
383 if isinstance(response, BaseException) or isinstance(response, json.decoder.JSONDecodeError):
384 response = str(response)
386 # DEBUG: print("DEBUG: AFTER response[]:", type(response))
387 if type(response) is str:
388 cursor.execute("INSERT INTO error_log (domain, error_code, error_message, created) VALUES (?, 999, ?, ?)",[
394 cursor.execute("INSERT INTO error_log (domain, error_code, error_message, created) VALUES (?, ?, ?, ?)",[
396 response.status_code,
401 # Cleanup old entries
402 # DEBUG: print(f"DEBUG: Purging old records (distance: {config.get('error_log_cleanup')})")
403 cursor.execute("DELETE FROM error_log WHERE created < ?", [time.time() - config.get("error_log_cleanup")])
404 except BaseException as e:
405 print(f"ERROR: failed SQL query: domain='{domain}',exception[{type(e)}]:'{str(e)}'")
408 # DEBUG: print("DEBUG: EXIT!")
410 def update_last_error(domain: str, response: requests.models.Response):
411 # DEBUG: print("DEBUG: domain,response[]:", domain, type(response))
412 if type(domain) != str:
413 raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
415 raise ValueError(f"Parameter 'domain' cannot be empty")
417 # DEBUG: print("DEBUG: BEFORE response[]:", type(response))
418 if isinstance(response, BaseException) or isinstance(response, json.decoder.JSONDecodeError):
419 response = f"{type}:str(response)"
421 # DEBUG: print("DEBUG: AFTER response[]:", type(response))
422 if type(response) is str:
423 # DEBUG: print(f"DEBUG: Setting last_error_details='{response}'");
424 instances.set("last_status_code" , domain, 999)
425 instances.set("last_error_details", domain, response)
427 # DEBUG: print(f"DEBUG: Setting last_error_details='{response.reason}'");
428 instances.set("last_status_code" , domain, response.status_code)
429 instances.set("last_error_details", domain, response.reason)
431 # Running pending updated
432 # DEBUG: print(f"DEBUG: Invoking instances.update_instance_data({domain}) ...")
433 instances.update_instance_data(domain)
435 log_error(domain, response)
437 # DEBUG: print("DEBUG: EXIT!")
439 def update_last_instance_fetch(domain: str):
440 # DEBUG: print(f"DEBUG: domain={domain} - CALLED!")
441 if type(domain) != str:
442 raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
444 raise ValueError(f"Parameter 'domain' cannot be empty")
446 # DEBUG: print("DEBUG: Updating last_instance_fetch for domain:", domain)
447 instances.set("last_instance_fetch", domain, time.time())
449 # Running pending updated
450 # DEBUG: print(f"DEBUG: Invoking instances.update_instance_data({domain}) ...")
451 instances.update_instance_data(domain)
453 # DEBUG: print("DEBUG: EXIT!")
455 def update_last_nodeinfo(domain: str):
456 # DEBUG: print(f"DEBUG: domain={domain} - CALLED!")
457 if type(domain) != str:
458 raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
460 raise ValueError(f"Parameter 'domain' cannot be empty")
462 # DEBUG: print("DEBUG: Updating last_nodeinfo for domain:", domain)
463 instances.set("last_nodeinfo", domain, time.time())
464 instances.set("last_updated" , domain, time.time())
466 # Running pending updated
467 # DEBUG: print(f"DEBUG: Invoking instances.update_instance_data({domain}) ...")
468 instances.update_instance_data(domain)
470 # DEBUG: print("DEBUG: EXIT!")
472 def get_peers(domain: str, software: str) -> list:
473 # DEBUG: print(f"DEBUG: domain({len(domain)})={domain},software={software} - CALLED!")
474 if type(domain) != str:
475 raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
477 raise ValueError(f"Parameter 'domain' cannot be empty")
478 elif type(software) != str and software != None:
479 raise ValueError(f"software[]={type(software)} is not 'str'")
483 if software == "misskey":
484 # DEBUG: print(f"DEBUG: domain='{domain}' is misskey, sending API POST request ...")
486 step = config.get("misskey_limit")
488 # iterating through all "suspended" (follow-only in its terminology)
489 # instances page-by-page, since that troonware doesn't support
490 # sending them all at once
492 # DEBUG: print(f"DEBUG: Fetching offset='{offset}' from '{domain}' ...")
494 fetched = post_json_api(domain, "/api/federation/instances", json.dumps({
502 fetched = post_json_api(domain, "/api/federation/instances", json.dumps({
511 # DEBUG: print(f"DEBUG: fetched()={len(fetched)}")
512 if len(fetched) == 0:
513 # DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
515 elif len(fetched) != config.get("misskey_limit"):
516 # DEBUG: print(f"DEBUG: Fetched '{len(fetched)}' row(s) but expected: '{config.get('misskey_limit')}'")
517 offset = offset + (config.get("misskey_limit") - len(fetched))
519 # DEBUG: print("DEBUG: Raising offset by step:", step)
520 offset = offset + step
523 # DEBUG: print(f"DEBUG: fetched({len(fetched)})[]={type(fetched)}")
524 if isinstance(fetched, dict) and "error" in fetched and "message" in fetched["error"]:
525 print(f"WARNING: post_json_api() returned error: {fetched['error']['message']}")
526 update_last_error(domain, fetched["error"]["message"])
531 # DEBUG: print(f"DEBUG: row():{len(row)}")
532 if not "host" in row:
533 print(f"WARNING: row()={len(row)} does not contain element 'host': {row},domain='{domain}'")
535 elif type(row["host"]) != str:
536 print(f"WARNING: row[host][]={type(row['host'])} is not 'str'")
538 elif is_blacklisted(row["host"]):
539 # DEBUG: print(f"DEBUG: row[host]='{row['host']}' is blacklisted. domain='{domain}'")
541 elif row["host"] in peers:
542 # DEBUG: print(f"DEBUG: Not adding row[host]='{row['host']}', already found.")
543 already = already + 1
546 # DEBUG: print(f"DEBUG: Adding peer: '{row['host']}'")
547 peers.append(row["host"])
549 if already == len(fetched):
550 print(f"WARNING: Host returned same set of '{already}' instances, aborting loop!")
553 # DEBUG: print(f"DEBUG: Adding '{len(peers)}' for domain='{domain}'")
554 instances.set("total_peers", domain, len(peers))
556 # DEBUG: print(f"DEBUG: Updating last_instance_fetch for domain='{domain}' ...")
557 update_last_instance_fetch(domain)
559 # DEBUG: print("DEBUG: Returning peers[]:", type(peers))
561 elif software == "lemmy":
562 # DEBUG: print(f"DEBUG: domain='{domain}' is Lemmy, fetching JSON ...")
564 response = get_response(domain, "/api/v3/site", api_headers, (config.get("connection_timeout"), config.get("read_timeout")))
566 data = json_from_response(response)
568 # DEBUG: print(f"DEBUG: response.ok={response.ok},response.status_code='{response.status_code}',data[]='{type(data)}'")
569 if not response.ok or response.status_code >= 400:
570 print("WARNING: Could not reach any JSON API:", domain)
571 update_last_error(domain, response)
572 elif response.ok and isinstance(data, list):
573 # DEBUG: print(f"DEBUG: domain='{domain}' returned a list: '{data}'")
575 elif "federated_instances" in data:
576 # DEBUG: print(f"DEBUG: Found federated_instances for domain='{domain}'")
577 peers = peers + add_peers(data["federated_instances"])
578 # DEBUG: print("DEBUG: Added instance(s) to peers")
580 print("WARNING: JSON response does not contain 'federated_instances':", domain)
581 update_last_error(domain, response)
583 except BaseException as e:
584 print(f"WARNING: Exception during fetching JSON: domain='{domain}',exception[{type(e)}]:'{str(e)}'")
586 # DEBUG: print(f"DEBUG: Adding '{len(peers)}' for domain='{domain}'")
587 instances.set("total_peers", domain, len(peers))
589 # DEBUG: print(f"DEBUG: Updating last_instance_fetch for domain='{domain}' ...")
590 update_last_instance_fetch(domain)
592 # DEBUG: print("DEBUG: Returning peers[]:", type(peers))
594 elif software == "peertube":
595 # DEBUG: print(f"DEBUG: domain='{domain}' is a PeerTube, fetching JSON ...")
598 for mode in ["followers", "following"]:
599 # DEBUG: print(f"DEBUG: domain='{domain}',mode='{mode}'")
602 response = get_response(domain, "/api/v1/server/{mode}?start={start}&count=100", headers, (config.get("connection_timeout"), config.get("read_timeout")))
604 data = json_from_response(response)
605 # DEBUG: print(f"DEBUG: response.ok={response.ok},response.status_code='{response.status_code}',data[]='{type(data)}'")
606 if response.ok and isinstance(data, dict):
607 # DEBUG: print("DEBUG: Success, data:", len(data))
609 # DEBUG: print(f"DEBUG: Found {len(data['data'])} record(s).")
610 for record in data["data"]:
611 # DEBUG: print(f"DEBUG: record()={len(record)}")
612 if mode in record and "host" in record[mode]:
613 # DEBUG: print(f"DEBUG: Found host={record[mode]['host']}, adding ...")
614 peers.append(record[mode]["host"])
616 print(f"WARNING: record from '{domain}' has no '{mode}' or 'host' record: {record}")
618 if len(data["data"]) < 100:
619 # DEBUG: print("DEBUG: Reached end of JSON response:", domain)
622 # Continue with next row
625 except BaseException as e:
626 print(f"WARNING: Exception during fetching JSON: domain='{domain}',exception[{type(e)}]:'{str(e)}'")
628 # DEBUG: print(f"DEBUG: Adding '{len(peers)}' for domain='{domain}'")
629 instances.set("total_peers", domain, len(peers))
631 # DEBUG: print(f"DEBUG: Updating last_instance_fetch for domain='{domain}' ...")
632 update_last_instance_fetch(domain)
634 # DEBUG: print("DEBUG: Returning peers[]:", type(peers))
637 # DEBUG: print(f"DEBUG: Fetching get_peers_url='{get_peers_url}' from '{domain}' ...")
639 response = get_response(domain, get_peers_url, api_headers, (config.get("connection_timeout"), config.get("read_timeout")))
641 data = json_from_response(response)
643 # DEBUG: print(f"DEBUG: response.ok={response.ok},response.status_code={response.status_code},data[]='{type(data)}'")
644 if not response.ok or response.status_code >= 400:
645 # DEBUG: print(f"DEBUG: Was not able to fetch '{get_peers_url}', trying alternative ...")
646 response = get_response(domain, "/api/v3/site", api_headers, (config.get("connection_timeout"), config.get("read_timeout")))
648 data = json_from_response(response)
649 # DEBUG: print(f"DEBUG: response.ok={response.ok},response.status_code={response.status_code},data[]='{type(data)}'")
650 if not response.ok or response.status_code >= 400:
651 print("WARNING: Could not reach any JSON API:", domain)
652 update_last_error(domain, response)
653 elif response.ok and isinstance(data, list):
654 # DEBUG: print(f"DEBUG: domain='{domain}' returned a list: '{data}'")
656 elif "federated_instances" in data:
657 # DEBUG: print(f"DEBUG: Found federated_instances for domain='{domain}'")
658 peers = peers + add_peers(data["federated_instances"])
659 # DEBUG: print("DEBUG: Added instance(s) to peers")
661 print("WARNING: JSON response does not contain 'federated_instances':", domain)
662 update_last_error(domain, response)
664 # DEBUG: print("DEBUG: Querying API was successful:", domain, len(data))
667 except BaseException as e:
668 print("WARNING: Some error during get():", domain, e)
669 update_last_error(domain, e)
671 # DEBUG: print(f"DEBUG: Adding '{len(peers)}' for domain='{domain}'")
672 instances.set("total_peers", domain, len(peers))
674 # DEBUG: print(f"DEBUG: Updating last_instance_fetch for domain='{domain}' ...")
675 update_last_instance_fetch(domain)
677 # DEBUG: print("DEBUG: Returning peers[]:", type(peers))
680 def post_json_api(domain: str, path: str, parameter: str, extra_headers: dict = {}) -> dict:
681 if type(domain) != str:
682 raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
684 raise ValueError(f"Parameter 'domain' cannot be empty")
685 elif type(path) != str:
686 raise ValueError(f"path[]={type(path)} is not 'str'")
688 raise ValueError("Parameter 'path' cannot be empty")
689 elif type(parameter) != str:
690 raise ValueError(f"parameter[]={type(parameter)} is not 'str'")
692 # DEBUG: print("DEBUG: Sending POST to domain,path,parameter:", domain, path, parameter, extra_headers)
695 response = reqto.post(f"https://{domain}{path}", data=parameter, headers={**api_headers, **extra_headers}, timeout=(config.get("connection_timeout"), config.get("read_timeout")))
697 data = json_from_response(response)
698 # DEBUG: print(f"DEBUG: response.ok={response.ok},response.status_code={response.status_code},data[]='{type(data)}'")
699 if not response.ok or response.status_code >= 400:
700 print(f"WARNING: Cannot query JSON API: domain='{domain}',path='{path}',parameter()={len(parameter)},response.status_code='{response.status_code}',data[]='{type(data)}'")
701 update_last_error(domain, response)
703 except BaseException as e:
704 print(f"WARNING: Some error during post(): domain='{domain}',path='{path}',parameter()={len(parameter)},exception[{type(e)}]:'{str(e)}'")
706 # DEBUG: print(f"DEBUG: Returning data({len(data)})=[]:{type(data)}")
709 def fetch_nodeinfo(domain: str, path: str = None) -> list:
710 # DEBUG: print(f"DEBUG: domain='{domain}',path={path} - CALLED!")
711 if type(domain) != str:
712 raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
714 raise ValueError(f"Parameter 'domain' cannot be empty")
715 elif type(path) != str and path != None:
716 raise ValueError(f"Parameter path[]={type(path)} is not 'str'")
718 # DEBUG: print("DEBUG: Fetching nodeinfo from domain,path:", domain, path)
719 nodeinfo = fetch_wellknown_nodeinfo(domain)
721 # DEBUG: print(f"DEBUG: nodeinfo({len(nodeinfo)})={nodeinfo}")
722 if len(nodeinfo) > 0:
723 # DEBUG: print("DEBUG: nodeinfo()={len(nodeinfo))} - EXIT!")
727 f"https://{domain}/nodeinfo/2.1.json",
728 f"https://{domain}/nodeinfo/2.1",
729 f"https://{domain}/nodeinfo/2.0.json",
730 f"https://{domain}/nodeinfo/2.0",
731 f"https://{domain}/nodeinfo/1.0",
732 f"https://{domain}/api/v1/instance"
736 for request in requests:
737 if path != None and path != "" and request != path:
738 # DEBUG: print(f"DEBUG: path='{path}' does not match request='{request}' - SKIPPED!")
742 # DEBUG: print("DEBUG: Fetching request:", request)
743 response = reqto.get(request, headers=api_headers, timeout=(config.get("nodeinfo_connection_timeout"), config.get("nodeinfo_read_timeout")))
745 data = json_from_response(response)
746 # DEBUG: print(f"DEBUG: response.ok={response.ok},response.status_code={response.status_code},data[]='{type(data)}'")
747 if response.ok and isinstance(data, dict):
748 # DEBUG: print("DEBUG: Success:", request)
749 instances.set("detection_mode", domain, "STATIC_CHECK")
750 instances.set("nodeinfo_url" , domain, request)
752 elif response.ok and isinstance(data, list):
753 # DEBUG: print(f"DEBUG: domain='{domain}' returned a list: '{data}'")
755 elif not response.ok or response.status_code >= 400:
756 print("WARNING: Failed fetching nodeinfo from domain:", domain)
757 update_last_error(domain, response)
760 except BaseException as e:
761 # DEBUG: print("DEBUG: Cannot fetch API request:", request)
762 update_last_error(domain, e)
765 # DEBUG: print(f"DEBUG: data()={len(data)} - EXIT!")
768 def fetch_wellknown_nodeinfo(domain: str) -> list:
769 # DEBUG: print(f"DEBUG: domain={domain} - CALLED!")
770 if type(domain) != str:
771 raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
773 raise ValueError(f"Parameter 'domain' cannot be empty")
775 # DEBUG: print("DEBUG: Fetching .well-known info for domain:", domain)
779 response = get_response(domain, "/.well-known/nodeinfo", api_headers, (config.get("nodeinfo_connection_timeout"), config.get("nodeinfo_read_timeout")))
781 data = json_from_response(response)
782 # DEBUG: print("DEBUG: domain,response.ok,data[]:", domain, response.ok, type(data))
783 if response.ok and isinstance(data, dict):
785 # DEBUG: print("DEBUG: Found entries:", len(nodeinfo), domain)
786 if "links" in nodeinfo:
787 # DEBUG: print("DEBUG: Found links in nodeinfo():", len(nodeinfo["links"]))
788 for link in nodeinfo["links"]:
789 # DEBUG: print("DEBUG: rel,href:", link["rel"], link["href"])
790 if link["rel"] in nodeinfo_identifier:
791 # DEBUG: print("DEBUG: Fetching nodeinfo from:", link["href"])
792 response = reqto.get(link["href"])
794 data = json_from_response(response)
795 # DEBUG: print("DEBUG: href,response.ok,response.status_code:", link["href"], response.ok, response.status_code)
796 if response.ok and isinstance(data, dict):
797 # DEBUG: print("DEBUG: Found JSON nodeinfo():", len(data))
798 instances.set("detection_mode", domain, "AUTO_DISCOVERY")
799 instances.set("nodeinfo_url" , domain, link["href"])
802 print("WARNING: Unknown 'rel' value:", domain, link["rel"])
804 print("WARNING: nodeinfo does not contain 'links':", domain)
806 except BaseException as e:
807 print("WARNING: Failed fetching .well-known info:", domain)
808 update_last_error(domain, e)
811 # DEBUG: print("DEBUG: Returning data[]:", type(data))
814 def fetch_generator_from_path(domain: str, path: str = "/") -> str:
815 # DEBUG: print(f"DEBUG: domain({len(domain)})={domain},path={path} - CALLED!")
816 if type(domain) != str:
817 raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
819 raise ValueError(f"Parameter 'domain' cannot be empty")
820 elif type(path) != str:
821 raise ValueError(f"path[]={type(path)} is not 'str'")
823 raise ValueError(f"Parameter 'domain' cannot be empty")
825 # DEBUG: print(f"DEBUG: domain='{domain}',path='{path}' - CALLED!")
829 # DEBUG: print(f"DEBUG: Fetching path='{path}' from '{domain}' ...")
830 response = get_response(domain, path, headers, (config.get("connection_timeout"), config.get("read_timeout")))
832 # DEBUG: print("DEBUG: domain,response.ok,response.status_code,response.text[]:", domain, response.ok, response.status_code, type(response.text))
833 if response.ok and response.status_code < 300 and len(response.text) > 0:
834 # DEBUG: print("DEBUG: Search for <meta name='generator'>:", domain)
835 doc = bs4.BeautifulSoup(response.text, "html.parser")
837 # DEBUG: print("DEBUG: doc[]:", type(doc))
838 generator = doc.find("meta", {"name": "generator"})
839 site_name = doc.find("meta", {"property": "og:site_name"})
841 # DEBUG: print(f"DEBUG: generator='{generator}',site_name='{site_name}'")
842 if isinstance(generator, bs4.element.Tag):
843 # DEBUG: print("DEBUG: Found generator meta tag:", domain)
844 software = tidyup_domain(generator.get("content"))
845 print(f"INFO: domain='{domain}' is generated by '{software}'")
846 instances.set("detection_mode", domain, "GENERATOR")
847 remove_pending_error(domain)
848 elif isinstance(site_name, bs4.element.Tag):
849 # DEBUG: print("DEBUG: Found property=og:site_name:", domain)
850 sofware = tidyup_domain(site_name.get("content"))
851 print(f"INFO: domain='{domain}' has og:site_name='{software}'")
852 instances.set("detection_mode", domain, "SITE_NAME")
853 remove_pending_error(domain)
855 except BaseException as e:
856 # DEBUG: print(f"DEBUG: Cannot fetch / from '{domain}':", e)
857 update_last_error(domain, e)
860 # DEBUG: print(f"DEBUG: software[]={type(software)}")
861 if type(software) is str and software == "":
862 # DEBUG: print(f"DEBUG: Corrected empty string to None for software of domain='{domain}'")
864 elif type(software) is str and ("." in software or " " in software):
865 # DEBUG: print(f"DEBUG: software='{software}' may contain a version number, domain='{domain}', removing it ...")
866 software = remove_version(software)
868 # DEBUG: print(f"DEBUG: software[]={type(software)}")
869 if type(software) is str and " powered by " in software:
870 # DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it")
871 software = remove_version(strip_powered_by(software))
872 elif type(software) is str and " hosted on " in software:
873 # DEBUG: print(f"DEBUG: software='{software}' has 'hosted on' in it")
874 software = remove_version(strip_hosted_on(software))
875 elif type(software) is str and " by " in software:
876 # DEBUG: print(f"DEBUG: software='{software}' has ' by ' in it")
877 software = strip_until(software, " by ")
878 elif type(software) is str and " see " in software:
879 # DEBUG: print(f"DEBUG: software='{software}' has ' see ' in it")
880 software = strip_until(software, " see ")
882 # DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
885 def determine_software(domain: str, path: str = None) -> str:
886 # DEBUG: print(f"DEBUG: domain({len(domain)})={domain},path={path} - CALLED!")
887 if type(domain) != str:
888 raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
890 raise ValueError(f"Parameter 'domain' cannot be empty")
891 elif type(path) != str and path != None:
892 raise ValueError(f"Parameter path[]={type(path)} is not 'str'")
894 # DEBUG: print("DEBUG: Determining software for domain,path:", domain, path)
897 # DEBUG: print(f"DEBUG: Fetching nodeinfo from '{domain}' ...")
898 data = fetch_nodeinfo(domain, path)
900 # DEBUG: print("DEBUG: data[]:", type(data))
901 if not isinstance(data, dict) or len(data) == 0:
902 # DEBUG: print("DEBUG: Could not determine software type:", domain)
903 return fetch_generator_from_path(domain)
905 # DEBUG: print("DEBUG: data():", len(data), data)
906 if "status" in data and data["status"] == "error" and "message" in data:
907 print("WARNING: JSON response is an error:", data["message"])
908 update_last_error(domain, data["message"])
909 return fetch_generator_from_path(domain)
910 elif "message" in data:
911 print("WARNING: JSON response contains only a message:", data["message"])
912 update_last_error(domain, data["message"])
913 return fetch_generator_from_path(domain)
914 elif "software" not in data or "name" not in data["software"]:
915 # DEBUG: print(f"DEBUG: JSON response from domain='{domain}' does not include [software][name], fetching / ...")
916 software = fetch_generator_from_path(domain)
918 # DEBUG: print(f"DEBUG: Generator for domain='{domain}' is: {software}, EXIT!")
921 software = tidyup_domain(data["software"]["name"])
923 # DEBUG: print("DEBUG: sofware after tidyup_domain():", software)
924 if software in ["akkoma", "rebased"]:
925 # DEBUG: print("DEBUG: Setting pleroma:", domain, software)
927 elif software in ["hometown", "ecko"]:
928 # DEBUG: print("DEBUG: Setting mastodon:", domain, software)
929 software = "mastodon"
930 elif software in ["calckey", "groundpolis", "foundkey", "cherrypick", "meisskey"]:
931 # DEBUG: print("DEBUG: Setting misskey:", domain, software)
933 elif software.find("/") > 0:
934 print("WARNING: Spliting of slash:", software)
935 software = software.split("/")[-1];
936 elif software.find("|") > 0:
937 print("WARNING: Spliting of pipe:", software)
938 software = tidyup_domain(software.split("|")[0]);
939 elif "powered by" in software:
940 # DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it")
941 software = strip_powered_by(software)
942 elif type(software) is str and " by " in software:
943 # DEBUG: print(f"DEBUG: software='{software}' has ' by ' in it")
944 software = strip_until(software, " by ")
945 elif type(software) is str and " see " in software:
946 # DEBUG: print(f"DEBUG: software='{software}' has ' see ' in it")
947 software = strip_until(software, " see ")
949 # DEBUG: print(f"DEBUG: software[]={type(software)}")
951 print("WARNING: tidyup_domain() left no software name behind:", domain)
954 # DEBUG: print(f"DEBUG: software[]={type(software)}")
955 if str(software) == "":
956 # DEBUG: print(f"DEBUG: software for '{domain}' was not detected, trying generator ...")
957 software = fetch_generator_from_path(domain)
958 elif len(str(software)) > 0 and ("." in software or " " in software):
959 # DEBUG: print(f"DEBUG: software='{software}' may contain a version number, domain='{domain}', removing it ...")
960 software = remove_version(software)
962 # DEBUG: print(f"DEBUG: software[]={type(software)}")
963 if type(software) is str and "powered by" in software:
964 # DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it")
965 software = remove_version(strip_powered_by(software))
967 # DEBUG: print("DEBUG: Returning domain,software:", domain, software)
970 def update_block_reason(reason: str, blocker: str, blocked: str, block_level: str):
971 # DEBUG: print(f"DEBUG: reason='{reason}',blocker={blocker},blocked={blocked},block_level={block_level} - CALLED!")
972 if type(reason) != str and reason != None:
973 raise ValueError(f"Parameter reason[]='{type(reason)}' is not 'str'")
974 elif type(blocker) != str:
975 raise ValueError(f"Parameter blocker[]='{type(blocker)}' is not 'str'")
976 elif type(blocked) != str:
977 raise ValueError(f"Parameter blocked[]='{type(blocked)}' is not 'str'")
978 elif type(block_level) != str:
979 raise ValueError(f"Parameter block_level[]='{type(block_level)}' is not 'str'")
981 # DEBUG: print("DEBUG: Updating block reason:", reason, blocker, blocked, block_level)
984 "UPDATE blocks SET reason = ?, last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ? AND reason IN ('','unknown') LIMIT 1",
994 # DEBUG: print(f"DEBUG: cursor.rowcount={cursor.rowcount}")
995 if cursor.rowcount == 0:
996 # DEBUG: print(f"DEBUG: Did not update any rows: blocker='{blocker}',blocked='{blocked}',block_level='{block_level}',reason='{reason}' - EXIT!")
999 except BaseException as e:
1000 print(f"ERROR: failed SQL query: reason='{reason}',blocker='{blocker}',blocked='{blocked}',block_level='{block_level}',exception[{type(e)}]:'{str(e)}'")
1003 # DEBUG: print("DEBUG: EXIT!")
1005 def update_last_seen(blocker: str, blocked: str, block_level: str):
1006 # DEBUG: print("DEBUG: Updating last_seen for:", blocker, blocked, block_level)
1009 "UPDATE blocks SET last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ? LIMIT 1",
1018 # DEBUG: print(f"DEBUG: cursor.rowcount={cursor.rowcount}")
1019 if cursor.rowcount == 0:
1020 # DEBUG: print(f"DEBUG: Did not update any rows: blocker='{blocker}',blocked='{blocked}',block_level='{block_level}' - EXIT!")
1023 except BaseException as e:
1024 print(f"ERROR: failed SQL query: blocker='{blocker}',blocked='{blocked}',block_level='{block_level}',exception[{type(e)}]:'{str(e)}'")
1027 # DEBUG: print("DEBUG: EXIT!")
1029 def is_instance_blocked(blocker: str, blocked: str, block_level: str) -> bool:
1030 # DEBUG: print(f"DEBUG: blocker={blocker},blocked={blocked},block_level={block_level} - CALLED!")
1031 if type(blocker) != str:
1032 raise ValueError(f"Parameter blocker[]={type(blocker)} is not of type 'str'")
1034 raise ValueError("Parameter 'blocker' cannot be empty")
1035 elif type(blocked) != str:
1036 raise ValueError(f"Parameter blocked[]={type(blocked)} is not of type 'str'")
1038 raise ValueError("Parameter 'blocked' cannot be empty")
1039 elif type(block_level) != str:
1040 raise ValueError(f"Parameter block_level[]={type(block_level)} is not of type 'str'")
1041 elif block_level == "":
1042 raise ValueError("Parameter 'block_level' cannot be empty")
1045 "SELECT * FROM blocks WHERE blocker = ? AND blocked = ? AND block_level = ? LIMIT 1",
1053 is_blocked = cursor.fetchone() != None
1055 # DEBUG: print(f"DEBUG: is_blocked='{is_blocked}' - EXIT!")
1058 def block_instance(blocker: str, blocked: str, reason: str, block_level: str):
1059 # DEBUG: print("DEBUG: blocker,blocked,reason,block_level:", blocker, blocked, reason, block_level)
1060 if type(blocker) != str:
1061 raise ValueError(f"Parameter blocker[]={type(blocker)} is not 'str'")
1063 raise ValueError(f"Parameter 'blocker' cannot be empty")
1064 elif not validators.domain(blocker.split("/")[0]):
1065 raise ValueError(f"Bad blocker='{blocker}'")
1066 elif type(blocked) != str:
1067 raise ValueError(f"Parameter blocked[]={type(blocked)} is not 'str'")
1069 raise ValueError(f"Parameter 'blocked' cannot be empty")
1070 elif not validators.domain(blocked.split("/")[0]):
1071 raise ValueError(f"Bad blocked='{blocked}'")
1072 elif is_blacklisted(blocker):
1073 raise Exception(f"blocker='{blocker}' is blacklisted but function invoked")
1074 elif is_blacklisted(blocked):
1075 raise Exception(f"blocked='{blocked}' is blacklisted but function invoked")
1078 # Maybe needs cleaning
1079 reason = tidyup_reason(reason)
1081 print(f"INFO: New block: blocker='{blocker}',blocked='{blocked}', reason='{reason}', block_level='{block_level}'")
1084 "INSERT INTO blocks (blocker, blocked, reason, block_level, first_seen, last_seen) VALUES(?, ?, ?, ?, ?, ?)",
1094 except BaseException as e:
1095 print(f"ERROR: failed SQL query: blocker='{blocker}',blocked='{blocked}',reason='{reason}',block_level='{block_level}',exception[{type(e)}]:'{str(e)}'")
1098 # DEBUG: print("DEBUG: EXIT!")
1100 def is_instance_registered(domain: str) -> bool:
1101 # DEBUG: print(f"DEBUG: domain={domain} - CALLED!")
1102 if type(domain) != str:
1103 raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
1105 raise ValueError(f"Parameter 'domain' cannot be empty")
1107 # DEBUG: print(f"DEBUG: domain='{domain}' - CALLED!")
1108 if not cache.key_exists("is_registered"):
1109 # DEBUG: print(f"DEBUG: Cache for 'is_registered' not initialized, fetching all rows ...")
1111 cursor.execute("SELECT domain FROM instances")
1114 cache.set_all("is_registered", cursor.fetchall(), True)
1115 except BaseException as e:
1116 print(f"ERROR: failed SQL query: domain='{domain}',exception[{type(e)}]:'{str(e)}'")
1120 registered = cache.sub_key_exists("is_registered", domain)
1122 # DEBUG: print(f"DEBUG: registered='{registered}' - EXIT!")
1125 def add_instance(domain: str, origin: str, originator: str, path: str = None):
1126 # DEBUG: print(f"DEBUG: domain={domain},origin={origin},originator={originator},path={path} - CALLED!")
1127 if type(domain) != str:
1128 raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
1130 raise ValueError(f"Parameter 'domain' cannot be empty")
1131 elif type(origin) != str and origin != None:
1132 raise ValueError(f"origin[]={type(origin)} is not 'str'")
1133 elif type(originator) != str:
1134 raise ValueError(f"originator[]={type(originator)} is not 'str'")
1135 elif originator == "":
1136 raise ValueError(f"originator cannot be empty")
1137 elif not validators.domain(domain.split("/")[0]):
1138 raise ValueError(f"Bad domain name='{domain}'")
1139 elif origin is not None and not validators.domain(origin.split("/")[0]):
1140 raise ValueError(f"Bad origin name='{origin}'")
1141 elif is_blacklisted(domain):
1142 raise Exception(f"domain='{domain}' is blacklisted, but method invoked")
1144 # DEBUG: print("DEBUG: domain,origin,originator,path:", domain, origin, originator, path)
1145 software = determine_software(domain, path)
1146 # DEBUG: print("DEBUG: Determined software:", software)
1148 print(f"INFO: Adding instance domain='{domain}' (origin='{origin}',software='{software}')")
1151 "INSERT INTO instances (domain, origin, originator, hash, software, first_seen) VALUES (?, ?, ?, ?, ?, ?)",
1162 cache.set_sub_key("is_registered", domain, True)
1164 if instances.has_pending_instance_data(domain):
1165 # DEBUG: print(f"DEBUG: domain='{domain}' has pending nodeinfo being updated ...")
1166 instances.set("last_status_code" , domain, None)
1167 instances.set("last_error_details", domain, None)
1168 instances.update_instance_data(domain)
1169 remove_pending_error(domain)
1171 if domain in pending_errors:
1172 # DEBUG: print("DEBUG: domain has pending error being updated:", domain)
1173 update_last_error(domain, pending_errors[domain])
1174 remove_pending_error(domain)
1176 except BaseException as e:
1177 print(f"ERROR: failed SQL query: domain='{domain}',exception[{type(e)}]:'{str(e)}'")
1180 # DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain)
1181 update_last_nodeinfo(domain)
1183 # DEBUG: print("DEBUG: EXIT!")
1185 def send_bot_post(instance: str, blocks: dict):
1186 # DEBUG: print(f"DEBUG: instance={instance},blocks()={len(blocks)} - CALLED!")
1187 message = instance + " has blocked the following instances:\n\n"
1190 if len(blocks) > 20:
1192 blocks = blocks[0 : 19]
1194 for block in blocks:
1195 if block["reason"] == None or block["reason"] == '':
1196 message = message + block["blocked"] + " with unspecified reason\n"
1198 if len(block["reason"]) > 420:
1199 block["reason"] = block["reason"][0:419] + "[…]"
1201 message = message + block["blocked"] + ' for "' + block["reason"].replace("@", "@\u200b") + '"\n'
1204 message = message + "(the list has been truncated to the first 20 entries)"
1206 botheaders = {**api_headers, **{"Authorization": "Bearer " + config.get("bot_token")}}
1209 f"{config.get('bot_instance')}/api/v1/statuses",
1212 "visibility" : config.get('bot_visibility'),
1213 "content_type": "text/plain"
1221 def get_mastodon_blocks(domain: str) -> dict:
1222 # DEBUG: print(f"DEBUG: domain={domain} - CALLED!")
1223 if type(domain) != str:
1224 raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
1226 raise ValueError(f"Parameter 'domain' cannot be empty")
1228 # DEBUG: print("DEBUG: Fetching mastodon blocks from domain:", domain)
1230 "Suspended servers": [],
1231 "Filtered media" : [],
1232 "Limited servers" : [],
1233 "Silenced servers" : [],
1237 doc = bs4.BeautifulSoup(
1238 get_response(domain, "/about", headers, (config.get("connection_timeout"), config.get("read_timeout"))).text,
1241 except BaseException as e:
1242 print("ERROR: Cannot fetch from domain:", domain, e)
1243 update_last_error(domain, e)
1246 for header in doc.find_all("h3"):
1247 header_text = tidyup_domain(header.text)
1249 if header_text in language_mapping:
1250 # DEBUG: print(f"DEBUG: header_text='{header_text}'")
1251 header_text = language_mapping[header_text]
1253 if header_text in blocks or header_text.lower() in blocks:
1254 # replaced find_next_siblings with find_all_next to account for instances that e.g. hide lists in dropdown menu
1255 for line in header.find_all_next("table")[0].find_all("tr")[1:]:
1256 blocks[header_text].append(
1258 "domain": tidyup_domain(line.find("span").text),
1259 "hash" : tidyup_domain(line.find("span")["title"][9:]),
1260 "reason": tidyup_domain(line.find_all("td")[1].text),
1264 # DEBUG: print("DEBUG: Returning blocks for domain:", domain)
1266 "reject" : blocks["Suspended servers"],
1267 "media_removal" : blocks["Filtered media"],
1268 "followers_only": blocks["Limited servers"] + blocks["Silenced servers"],
1271 def get_friendica_blocks(domain: str) -> dict:
1272 # DEBUG: print(f"DEBUG: domain={domain} - CALLED!")
1273 if type(domain) != str:
1274 raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
1276 raise ValueError(f"Parameter 'domain' cannot be empty")
1278 # DEBUG: print("DEBUG: Fetching friendica blocks from domain:", domain)
1282 doc = bs4.BeautifulSoup(
1283 get_response(domain, "/friendica", headers, (config.get("connection_timeout"), config.get("read_timeout"))).text,
1286 except BaseException as e:
1287 print("WARNING: Failed to fetch /friendica from domain:", domain, e)
1288 update_last_error(domain, e)
1291 blocklist = doc.find(id="about_blocklist")
1293 # Prevents exceptions:
1294 if blocklist is None:
1295 # DEBUG: print("DEBUG: Instance has no block list:", domain)
1298 for line in blocklist.find("table").find_all("tr")[1:]:
1299 # DEBUG: print(f"DEBUG: line='{line}'")
1301 "domain": tidyup_domain(line.find_all("td")[0].text),
1302 "reason": tidyup_domain(line.find_all("td")[1].text)
1305 # DEBUG: print("DEBUG: Returning blocks() for domain:", domain, len(blocks))
1310 def get_misskey_blocks(domain: str) -> dict:
1311 # DEBUG: print(f"DEBUG: domain={domain} - CALLED!")
1312 if type(domain) != str:
1313 raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
1315 raise ValueError(f"Parameter 'domain' cannot be empty")
1317 # DEBUG: print("DEBUG: Fetching misskey blocks from domain:", domain)
1324 step = config.get("misskey_limit")
1326 # iterating through all "suspended" (follow-only in its terminology)
1327 # instances page-by-page, since that troonware doesn't support
1328 # sending them all at once
1330 # DEBUG: print(f"DEBUG: Fetching offset='{offset}' from '{domain}' ...")
1332 # DEBUG: print("DEBUG: Sending JSON API request to domain,step,offset:", domain, step, offset)
1333 fetched = post_json_api(domain, "/api/federation/instances", json.dumps({
1342 # DEBUG: print("DEBUG: Sending JSON API request to domain,step,offset:", domain, step, offset)
1343 fetched = post_json_api(domain, "/api/federation/instances", json.dumps({
1348 "offset" : offset - 1
1353 # DEBUG: print("DEBUG: fetched():", len(fetched))
1354 if len(fetched) == 0:
1355 # DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
1357 elif len(fetched) != config.get("misskey_limit"):
1358 # DEBUG: print(f"DEBUG: Fetched '{len(fetched)}' row(s) but expected: '{config.get('misskey_limit')}'")
1359 offset = offset + (config.get("misskey_limit") - len(fetched))
1361 # DEBUG: print("DEBUG: Raising offset by step:", step)
1362 offset = offset + step
1364 for instance in fetched:
1366 if instance["isSuspended"] and not has_element(blocks["suspended"], "domain", instance):
1367 blocks["suspended"].append(
1369 "domain": tidyup_domain(instance["host"]),
1370 # no reason field, nothing
1375 except BaseException as e:
1376 print("WARNING: Caught error, exiting loop:", domain, e)
1377 update_last_error(domain, e)
1382 # same shit, different asshole ("blocked" aka full suspend)
1385 # DEBUG: print("DEBUG: Sending JSON API request to domain,step,offset:", domain, step, offset)
1386 fetched = post_json_api(domain,"/api/federation/instances", json.dumps({
1395 # DEBUG: print("DEBUG: Sending JSON API request to domain,step,offset:", domain, step, offset)
1396 fetched = post_json_api(domain,"/api/federation/instances", json.dumps({
1401 "offset" : offset - 1
1406 # DEBUG: print("DEBUG: fetched():", len(fetched))
1407 if len(fetched) == 0:
1408 # DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
1410 elif len(fetched) != config.get("misskey_limit"):
1411 # DEBUG: print(f"DEBUG: Fetched '{len(fetched)}' row(s) but expected: '{config.get('misskey_limit')}'")
1412 offset = offset + (config.get("misskey_limit") - len(fetched))
1414 # DEBUG: print("DEBUG: Raising offset by step:", step)
1415 offset = offset + step
1417 for instance in fetched:
1418 if instance["isBlocked"] and not has_element(blocks["blocked"], "domain", instance):
1419 blocks["blocked"].append({
1420 "domain": tidyup_domain(instance["host"]),
1424 except BaseException as e:
1425 print("ERROR: Exception during POST:", domain, e)
1426 update_last_error(domain, e)
1430 # DEBUG: print(f"DEBUG: Updating last_instance_fetch for domain='{domain}' ...")
1431 update_last_instance_fetch(domain)
1433 # DEBUG: print("DEBUG: Returning for domain,blocked(),suspended():", domain, len(blocks["blocked"]), len(blocks["suspended"]))
1435 "reject" : blocks["blocked"],
1436 "followers_only": blocks["suspended"]
1439 def tidyup_reason(reason: str) -> str:
1440 # DEBUG: print(f"DEBUG: reason='{reason}' - CALLED!")
1441 if type(reason) != str:
1442 raise ValueError(f"Parameter reason[]={type(reason)} is not expected")
1445 reason = reason.strip()
1448 reason = re.sub("â", "\"", reason)
1450 #print(f"DEBUG: reason='{reason}' - EXIT!")
1453 def tidyup_domain(domain: str) -> str:
1454 # DEBUG: print(f"DEBUG: domain='{domain}' - CALLED!")
1455 if type(domain) != str:
1456 raise ValueError(f"Parameter domain[]={type(domain)} is not expected")
1458 # All lower-case and strip spaces out + last dot
1459 domain = domain.lower().strip().rstrip(".")
1462 domain = re.sub("\:\d+$", "", domain)
1464 # No protocol, sometimes without the slashes
1465 domain = re.sub("^https?\:(\/*)", "", domain)
1468 domain = re.sub("\/$", "", domain)
1471 domain = re.sub("^\@", "", domain)
1473 # No individual users in block lists
1474 domain = re.sub("(.+)\@", "", domain)
1476 # DEBUG: print(f"DEBUG: domain='{domain}' - EXIT!")
1479 def json_from_response(response: requests.models.Response) -> list:
1480 # DEBUG: print(f"DEBUG: response[]={type(response)} - CALLED!")
1481 if not isinstance(response, requests.models.Response):
1482 raise ValueError(f"Parameter response[]='{type(response)}' is not type of 'Response'")
1485 if response.text.strip() != "":
1486 # DEBUG: print(f"DEBUG: response.text()={len(response.text)} is not empty, invoking response.json() ...")
1488 data = response.json()
1489 except json.decoder.JSONDecodeError:
1492 # DEBUG: print(f"DEBUG: data[]={type(data)} - EXIT!")
1495 def get_response(domain: str, path: str, headers: dict, timeout: list) -> requests.models.Response:
1496 # DEBUG: print(f"DEBUG: domain='{domain}',path='{path}',headers()={len(headers)},timeout={timeout} - CALLED!")
1497 if type(domain) != str:
1498 raise ValueError(f"Parameter domain[]='{type(domain)}' is not 'str'")
1500 raise ValueError("Parameter 'domain' cannot be empty")
1501 elif type(path) != str:
1502 raise ValueError(f"Parameter path[]='{type(path)}' is not 'str'")
1504 raise ValueError("Parameter 'path' cannot be empty")
1507 # DEBUG: print(f"DEBUG: Sending request to '{domain}{path}' ...")
1508 response = reqto.get(f"https://{domain}{path}", headers=headers, timeout=timeout);
1509 except requests.exceptions.ConnectionError as e:
1510 # DEBUG: print(f"DEBUG: Fetching '{path}' from '{domain}' failed. exception[{type(e)}]='{str(e)}'")
1511 update_last_error(domain, e)
1514 # DEBUG: print(f"DEBUG: response[]='{type(response)}' - EXXIT!")
1517 def has_element(elements: list, key: str, value: any) -> bool:
1518 # DEBUG: print(f"DEBUG: element()={len(element)},key='{key}',value[]='{type(value)}' - CALLED!")
1519 if type(key) != str:
1520 raise ValueError(f"Parameter key[]='{type(key)}' is not 'str'")
1522 raise ValueError("Parameter 'key' cannot be empty")
1525 # DEBUG: print(f"DEBUG: Checking elements()={len(elements)} ...")
1526 for element in elements:
1527 # DEBUG: print(f"DEBUG: element[]='{type(element)}'")
1528 if type(element) != dict:
1529 raise ValueError(f"element[]='{type(element)}' is not 'dict'")
1530 elif not key in element:
1531 raise KeyError(f"Cannot find key='{key}'")
1532 elif element[key] == value:
1536 # DEBUG: print(f"DEBUG: has={has} - EXIT!")
1539 def find_domains(tag: bs4.element.Tag) -> list:
1540 # DEBUG: print(f"DEBUG: tag[]={type(tag)} - CALLED!")
1541 if not isinstance(tag, bs4.element.Tag):
1542 raise ValueError(f"Parameter tag[]={type(tag)} is not type of bs4.element.Tag")
1543 elif not isinstance(tag, bs4.element.Tag):
1544 raise KeyError("Cannot find table with instances!")
1545 elif len(tag.select("tr")) == 0:
1546 raise KeyError("No table rows found in table!")
1549 for element in tag.select("tr"):
1550 # DEBUG: print(f"DEBUG: element[]={type(element)}")
1551 if not element.find("td"):
1552 # DEBUG: print("DEBUG: Skipping element, no <td> found")
1555 domain = tidyup_domain(element.find("td").text)
1556 reason = tidyup_reason(element.findAll("td")[1].text)
1558 # DEBUG: print(f"DEBUG: domain='{domain}',reason='{reason}'")
1560 if is_blacklisted(domain):
1561 print(f"WARNING: domain='{domain}' is blacklisted - skipped!")
1563 elif domain == "gab.com/.ai, develop.gab.com":
1564 print(f"DEBUG: Multiple domains detected in one row")
1566 "domain": "gab.com",
1574 "domain": "develop.gab.com",
1578 elif not validators.domain(domain):
1579 print(f"WARNING: domain='{domain}' is not a valid domain - skipped!")
1582 # DEBUG: print(f"DEBUG: Adding domain='{domain}' ...")
1588 # DEBUG: print(f"DEBUG: domains()={len(domains)} - EXIT!")