11 with open("config.json") as f:
12 config = json.loads(f.read())
14 # Don't check these, known trolls/flooders/testing/developing
16 # Floods network with fake nodes as "research" project
17 "activitypub-troll.cf",
23 "social.shrimpcam.pw",
25 "mastotroll.netz.org",
26 # Testing/developing installations
30 # Array with pending errors needed to be written to database
34 # "rel" identifiers (no real URLs)
35 nodeinfo_identifier = [
36 "http://nodeinfo.diaspora.software/ns/schema/2.1",
37 "http://nodeinfo.diaspora.software/ns/schema/2.0",
38 "http://nodeinfo.diaspora.software/ns/schema/1.1",
39 "http://nodeinfo.diaspora.software/ns/schema/1.0",
42 # HTTP headers for requests
44 "user-agent": config["useragent"],
47 # Found info from node, such as nodeinfo URL, detection mode that needs to be
48 # written to database. Both arrays must be filled at the same time or else
49 # update_nodeinfos() will fail
51 # Detection mode: 'AUTO_DISCOVERY', 'STATIC_CHECKS' or 'GENERATOR'
52 # NULL means all detection methods have failed (maybe still reachable instance)
56 # Where to fetch peers (other instances)
62 "Silenced instances" : "Silenced servers",
63 "Suspended instances" : "Suspended servers",
64 "Limited instances" : "Limited servers",
65 # Mappuing German -> English
66 "Gesperrte Server" : "Suspended servers",
67 "Gefilterte Medien" : "Filtered media",
68 "Stummgeschaltete Server" : "Silenced servers",
70 "停止済みのサーバー" : "Suspended servers",
71 "制限中のサーバー" : "Limited servers",
72 "メディアを拒否しているサーバー": "Filtered media",
73 "サイレンス済みのサーバー" : "Silenced servers",
75 "שרתים מושעים" : "Suspended servers",
76 "מדיה מסוננת" : "Filtered media",
77 "שרתים מוגבלים" : "Silenced servers",
79 "Serveurs suspendus" : "Suspended servers",
80 "Médias filtrés" : "Filtered media",
81 "Serveurs limités" : "Limited servers",
82 "Serveurs modérés" : "Limited servers",
85 # URL for fetching peers
86 get_peers_url = "/api/v1/instance/peers"
89 connection = sqlite3.connect("blocks.db")
90 cursor = connection.cursor()
92 # Pattern instance for version numbers
94 # semantic version number (with v|V) prefix)
95 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-]+)*))?)?$"),
96 # non-sematic, e.g. 1.2.3.4
97 re.compile("^(?P<version>v|V{0,1})(\.{0,1})(?P<major>0|[1-9]\d*)\.(?P<minor>0+|[1-9]\d*)(\.(?P<patch>0+|[1-9]\d*)(\.(?P<subpatch>0|[1-9]\d*))?)$"),
98 # non-sematic, e.g. 2023-05
99 re.compile("^(?P<year>[1-9]{1}[0-9]{3})\.(?P<month>[0-9]{2})$")
102 def remove_version(software: str) -> str:
103 # NOISY-DEBUG: print(f"DEBUG: software='{software}' - CALLED!")
104 if not "." in software:
105 print(f"WARNING: software='{software}' does not contain a version number.")
110 version = software.split(" ")[-1]
111 elif "/" in software:
112 version = software.split("/")[-1]
113 elif "-" in software:
114 version = software.split("-")[-1]
116 # NOISY-DEBUG: print(f"DEBUG: Was not able to find common seperator, returning untouched software='{software}'")
120 # NOISY-DEBUG: print(f"DEBUG: Checking {len(patterns)} patterns ...")
121 for pattern in patterns:
123 match = pattern.match(version)
125 # NOISY-DEBUG: print(f"DEBUG: match[]={type(match)}")
126 if type(match) is re.Match:
129 # NOISY-DEBUG: print(f"DEBUG: version[{type(version)}]='{version}',match='{match}'")
130 if type(match) is not re.Match:
131 print(f"WARNING: version='{version}' does not match regex, leaving software='{software}' untouched.")
134 # NOISY-DEBUG: print(f"DEBUG: Found valid version number: '{version}', removing it ...")
135 end = len(software) - len(version) - 1
137 # NOISY-DEBUG: print(f"DEBUG: end[{type(end)}]={end}")
138 software = software[0:end].strip()
139 if " version" in software:
140 # NOISY-DEBUG: print(f"DEBUG: software='{software}' contains word ' version'")
141 software = strip_until(software, " version")
143 # NOISY-DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
146 def strip_powered_by(software: str) -> str:
147 # NOISY-DEBUG: print(f"DEBUG: software='{software}' - CALLED!")
149 print(f"ERROR: Bad method call, 'software' is empty")
150 raise Exception("Parameter 'software' is empty")
151 elif not "powered by" in software:
152 print(f"WARNING: Cannot find 'powered by' in '{software}'!")
155 start = software.find("powered by ")
156 # NOISY-DEBUG: print(f"DEBUG: start[{type(start)}]='{start}'")
158 software = software[start + 11:].strip()
159 # NOISY-DEBUG: print(f"DEBUG: software='{software}'")
161 software = strip_until(software, " - ")
163 # NOISY-DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
166 def strip_until(software: str, until: str) -> str:
167 # NOISY-DEBUG: print(f"DEBUG: software='{software}',until='{until}' - CALLED!")
169 print(f"ERROR: Bad method call, 'software' is empty")
170 raise Exception("Parameter 'software' is empty")
172 print(f"ERROR: Bad method call, 'until' is empty")
173 raise Exception("Parameter 'until' is empty")
174 elif not until in software:
175 print(f"WARNING: Cannot find 'powered by' in '{software}'!")
178 # Next, strip until part
179 end = software.find(until)
181 # NOISY-DEBUG: print(f"DEBUG: end[{type(end)}]='{end}'")
183 software = software[0:end].strip()
185 # NOISY-DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
188 def is_blacklisted(domain: str) -> bool:
190 for peer in blacklist:
196 def remove_pending_error(domain: str):
198 # Prevent updating any pending errors, nodeinfo was found
199 del pending_errors[domain]
204 def get_hash(domain: str) -> str:
205 return hashlib.sha256(domain.encode("utf-8")).hexdigest()
207 def update_last_blocked(domain: str):
208 # NOISY-DEBUG: print("DEBUG: Updating last_blocked for domain", domain)
210 cursor.execute("UPDATE instances SET last_blocked = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
216 if cursor.rowcount == 0:
217 print("WARNING: Did not update any rows:", domain)
219 except BaseException as e:
220 print("ERROR: failed SQL query:", domain, e)
223 # NOISY-DEBUG: print("DEBUG: EXIT!")
225 def update_nodeinfos(domain: str):
226 # NOISY-DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain)
229 for key in nodeinfos:
230 # NOISY-DEBUG: print("DEBUG: key:", key)
231 if domain in nodeinfos[key]:
232 # NOISY-DEBUG: print(f"DEBUG: Adding '{nodeinfos[key][domain]}' for key='{key}' ...")
233 fields.append(nodeinfos[key][domain])
234 sql_string += f" {key} = ?,"
236 fields.append(domain)
237 # NOISY-DEBUG: print(f"DEBUG: sql_string='{sql_string}',fields()={len(fields)}")
239 sql = "UPDATE instances SET" + sql_string + " last_status_code = NULL, last_error_details = NULL WHERE domain = ? LIMIT 1"
240 # NOISY-DEBUG: print("DEBUG: sql:", sql)
243 # NOISY-DEBUG: print("DEBUG: Executing SQL:", sql)
244 cursor.execute(sql, fields)
245 # NOISY-DEBUG: print(f"DEBUG: Success! (rowcount={cursor.rowcount })")
247 if cursor.rowcount == 0:
248 print("WARNING: Did not update any rows:", domain)
250 except BaseException as e:
251 print(f"ERROR: failed SQL query: domain='{domain}',sql='{sql}',exception:'{e}'")
254 # NOISY-DEBUG: print("DEBUG: Deleting nodeinfos for domain:", domain)
255 for key in nodeinfos:
257 # NOISY-DEBUG: print("DEBUG: Deleting key:", key)
258 del nodeinfos[key][domain]
262 # NOISY-DEBUG: print("DEBUG: EXIT!")
264 def update_last_error(domain: str, res: any):
265 # NOISY-DEBUG: print("DEBUG: domain,res[]:", domain, type(res))
267 # NOISY-DEBUG: print("DEBUG: BEFORE res[]:", type(res))
268 if isinstance(res, BaseException) or isinstance(res, json.JSONDecodeError):
271 # NOISY-DEBUG: print("DEBUG: AFTER res[]:", type(res))
273 # NOISY-DEBUG: print(f"DEBUG: Setting last_error_details='{res}'");
274 cursor.execute("UPDATE instances SET last_status_code = 999, last_error_details = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
280 # NOISY-DEBUG: print(f"DEBUG: Setting last_error_details='{res.reason}'");
281 cursor.execute("UPDATE instances SET last_status_code = ?, last_error_details = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
288 if cursor.rowcount == 0:
289 # NOISY-DEBUG: print("DEBUG: Did not update any rows:", domain)
290 pending_errors[domain] = res
292 except BaseException as e:
293 print("ERROR: failed SQL query:", domain, e)
296 # NOISY-DEBUG: print("DEBUG: EXIT!")
298 def update_last_nodeinfo(domain: str):
299 # NOISY-DEBUG: print("DEBUG: Updating last_nodeinfo for domain:", domain)
301 cursor.execute("UPDATE instances SET last_nodeinfo = ?, last_updated = ? WHERE domain = ? LIMIT 1", [
307 if cursor.rowcount == 0:
308 print("WARNING: Did not update any rows:", domain)
310 except BaseException as e:
311 print("ERROR: failed SQL query:", domain, e)
315 # NOISY-DEBUG: print("DEBUG: EXIT!")
317 def get_peers(domain: str, software: str) -> list:
318 # NOISY-DEBUG: print("DEBUG: Getting peers for domain:", domain, software)
321 if software == "lemmy":
322 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}' is Lemmy, fetching JSON ...")
324 res = reqto.get(f"https://{domain}/api/v3/site", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
326 # NOISY-DEBUG: print(f"DEBUG: res.ok={res.ok},res.json[]={type(res.json())}")
327 if res.ok and isinstance(res.json(), dict):
328 # NOISY-DEBUG: print("DEBUG: Success, res.json():", len(res.json()))
331 if "federated_instances" in json and "linked" in json["federated_instances"]:
332 # NOISY-DEBUG: print("DEBUG: Found federated_instances", domain)
333 peers = json["federated_instances"]["linked"] + json["federated_instances"]["allowed"] + json["federated_instances"]["blocked"]
335 except BaseException as e:
336 print("WARNING: Exception during fetching JSON:", domain, e)
338 update_last_nodeinfo(domain)
340 # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers))
342 elif software == "peertube":
343 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}' is a PeerTube, fetching JSON ...")
346 for mode in ["followers", "following"]:
347 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}',mode='{mode}'")
350 res = reqto.get(f"https://{domain}/api/v1/server/{mode}?start={start}&count=100", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
352 # NOISY-DEBUG: print(f"DEBUG: res.ok={res.ok},res.json[]={type(res.json())}")
353 if res.ok and isinstance(res.json(), dict):
354 # NOISY-DEBUG: print("DEBUG: Success, res.json():", len(res.json()))
358 # NOISY-DEBUG: print(f"DEBUG: Found {len(json['data'])} record(s).")
359 for record in json["data"]:
360 # NOISY-DEBUG: print(f"DEBUG: record()={len(record)}")
361 if mode in record and "host" in record[mode]:
362 # NOISY-DEBUG: print(f"DEBUG: Found host={record[mode]['host']}, adding ...")
363 peers.append(record[mode]["host"])
365 print(f"WARNING: record from '{domain}' has no '{mode}' or 'host' record: {record}")
367 if len(json["data"]) < 100:
368 # NOISY-DEBUG: print("DEBUG: Reached end of JSON response:", domain)
371 # Continue with next row
374 except BaseException as e:
375 print("WARNING: Exception during fetching JSON:", domain, e)
377 update_last_nodeinfo(domain)
379 # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers))
382 # NOISY-DEBUG: print(f"DEBUG: Fetching '{get_peers_url}' from '{domain}' ...")
384 res = reqto.get(f"https://{domain}{get_peers_url}", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
386 # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json()))
387 if not res.ok or res.status_code >= 400:
388 res = reqto.get(f"https://{domain}/api/v3/site", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
390 if "federated_instances" in json and "linked" in json["federated_instances"]:
391 # NOISY-DEBUG: print("DEBUG: Found federated_instances", domain)
392 peers = json["federated_instances"]["linked"] + json["federated_instances"]["allowed"] + json["federated_instances"]["blocked"]
394 print("WARNING: Could not reach any JSON API:", domain)
395 update_last_error(domain, res)
397 # NOISY-DEBUG: print("DEBUG:Querying API was successful:", domain, len(res.json()))
399 nodeinfos["get_peers_url"][domain] = get_peers_url
401 except BaseException as e:
402 print("WARNING: Some error during get():", domain, e)
403 update_last_error(domain, e)
405 update_last_nodeinfo(domain)
407 # NOISY-DEBUG: print("DEBUG: Returning peers[]:", type(peers))
410 def post_json_api(domain: str, path: str, data: str) -> list:
411 # NOISY-DEBUG: print("DEBUG: Sending POST to domain,path,data:", domain, path, data)
414 res = reqto.post(f"https://{domain}{path}", data=data, headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
416 # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json()))
417 if not res.ok or res.status_code >= 400:
418 print("WARNING: Cannot query JSON API:", domain, path, data, res.status_code)
419 update_last_error(domain, res)
421 update_last_nodeinfo(domain)
424 except BaseException as e:
425 print("WARNING: Some error during post():", domain, path, data, e)
427 # NOISY-DEBUG: print("DEBUG: Returning json():", len(json))
430 def fetch_nodeinfo(domain: str) -> list:
431 # NOISY-DEBUG: print("DEBUG: Fetching nodeinfo from domain:", domain)
433 nodeinfo = fetch_wellknown_nodeinfo(domain)
434 # NOISY-DEBUG: print("DEBUG:nodeinfo:", len(nodeinfo))
436 if len(nodeinfo) > 0:
437 # NOISY-DEBUG: print("DEBUG: Returning auto-discovered nodeinfo:", len(nodeinfo))
441 f"https://{domain}/nodeinfo/2.1.json",
442 f"https://{domain}/nodeinfo/2.1",
443 f"https://{domain}/nodeinfo/2.0.json",
444 f"https://{domain}/nodeinfo/2.0",
445 f"https://{domain}/nodeinfo/1.0",
446 f"https://{domain}/api/v1/instance"
450 for request in requests:
452 # NOISY-DEBUG: print("DEBUG: Fetching request:", request)
453 res = reqto.get(request, headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
455 # NOISY-DEBUG: print("DEBUG: res.ok,res.json[]:", res.ok, type(res.json()))
456 if res.ok and isinstance(res.json(), dict):
457 # NOISY-DEBUG: print("DEBUG: Success:", request)
459 nodeinfos["detection_mode"][domain] = "STATIC_CHECK"
460 nodeinfos["nodeinfo_url"][domain] = request
462 elif not res.ok or res.status_code >= 400:
463 print("WARNING: Failed fetching nodeinfo from domain:", domain)
464 update_last_error(domain, res)
467 except BaseException as e:
468 # NOISY-DEBUG: print("DEBUG: Cannot fetch API request:", request)
469 update_last_error(domain, e)
472 # NOISY-DEBUG: print("DEBUG: Returning json[]:", type(json))
475 def fetch_wellknown_nodeinfo(domain: str) -> list:
476 # NOISY-DEBUG: print("DEBUG: Fetching .well-known info for domain:", domain)
480 res = reqto.get(f"https://{domain}/.well-known/nodeinfo", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
481 # NOISY-DEBUG: print("DEBUG: domain,res.ok,res.json[]:", domain, res.ok, type(res.json()))
482 if res.ok and isinstance(res.json(), dict):
483 nodeinfo = res.json()
484 # NOISY-DEBUG: print("DEBUG: Found entries:", len(nodeinfo), domain)
485 if "links" in nodeinfo:
486 # NOISY-DEBUG: print("DEBUG: Found links in nodeinfo():", len(nodeinfo["links"]))
487 for link in nodeinfo["links"]:
488 # NOISY-DEBUG: print("DEBUG: rel,href:", link["rel"], link["href"])
489 if link["rel"] in nodeinfo_identifier:
490 # NOISY-DEBUG: print("DEBUG: Fetching nodeinfo from:", link["href"])
491 res = reqto.get(link["href"])
493 # NOISY-DEBUG: print("DEBUG: href,res.ok,res.status_code:", link["href"], res.ok, res.status_code)
494 if res.ok and isinstance(res.json(), dict):
495 # NOISY-DEBUG: print("DEBUG: Found JSON nodeinfo():", len(res.json()))
497 nodeinfos["detection_mode"][domain] = "AUTO_DISCOVERY"
498 nodeinfos["nodeinfo_url"][domain] = link["href"]
501 print("WARNING: Unknown 'rel' value:", domain, link["rel"])
503 print("WARNING: nodeinfo does not contain 'links':", domain)
505 except BaseException as e:
506 print("WARNING: Failed fetching .well-known info:", domain)
507 update_last_error(domain, e)
510 # NOISY-DEBUG: print("DEBUG: Returning json[]:", type(json))
513 def fetch_generator_from_path(domain: str, path: str = "/") -> str:
514 # NOISY-DEBUG: print(f"DEBUG: domain='{domain}',path='{path}' - CALLED!")
518 # NOISY-DEBUG: print(f"DEBUG: Fetching '{path}' from '{domain}' ...")
519 res = reqto.get(f"https://{domain}{path}", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
521 # NOISY-DEBUG: print("DEBUG: domain,res.ok,res.status_code,res.text[]:", domain, res.ok, res.status_code, type(res.text))
522 if res.ok and res.status_code < 300 and len(res.text) > 0:
523 # NOISY-DEBUG: print("DEBUG: Search for <meta name='generator'>:", domain)
524 doc = bs4.BeautifulSoup(res.text, "html.parser")
526 # NOISY-DEBUG: print("DEBUG: doc[]:", type(doc))
527 tag = doc.find("meta", {"name": "generator"})
529 # NOISY-DEBUG: print(f"DEBUG: tag[{type(tag)}: {tag}")
530 if isinstance(tag, bs4.element.Tag):
531 # NOISY-DEBUG: print("DEBUG: Found generator meta tag: ", domain)
532 software = tidyup(tag.get("content"))
533 print(f"INFO: domain='{domain}' is generated by '{software}'")
534 nodeinfos["detection_mode"][domain] = "GENERATOR"
535 remove_pending_error(domain)
537 except BaseException as e:
538 # NOISY-DEBUG: print(f"DEBUG: Cannot fetch / from '{domain}':", e)
539 update_last_error(domain, e)
542 # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}")
543 if type(software) is str and software == "":
544 # NOISY-DEBUG: print(f"DEBUG: Corrected empty string to None for software of domain='{domain}'")
546 elif type(software) is str and "." in software:
547 # NOISY-DEBUG: print(f"DEBUG: software='{software}' may contain a version number, domain='{domain}', removing it ...")
548 software = remove_version(software)
550 # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}")
551 if type(software) is str and "powered by" in software:
552 # NOISY-DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it")
553 software = remove_version(strip_powered_by(software))
554 elif type(software) is str and " by " in software:
555 # NOISY-DEBUG: print(f"DEBUG: software='{software}' has ' by ' in it")
556 software = strip_until(software, " by ")
557 elif type(software) is str and " see " in software:
558 # NOISY-DEBUG: print(f"DEBUG: software='{software}' has ' see ' in it")
559 software = strip_until(software, " see ")
561 # NOISY-DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
564 def determine_software(domain: str) -> str:
565 # NOISY-DEBUG: print("DEBUG: Determining software for domain:", domain)
568 # NOISY-DEBUG: print(f"DEBUG: Fetching nodeinfo from '{domain}' ...")
569 json = fetch_nodeinfo(domain)
571 # NOISY-DEBUG: print("DEBUG: json[]:", type(json))
572 if not isinstance(json, dict) or len(json) == 0:
573 # NOISY-DEBUG: print("DEBUG: Could not determine software type:", domain)
574 return fetch_generator_from_path(domain)
576 # NOISY-DEBUG: print("DEBUG: json():", len(json), json)
577 if "status" in json and json["status"] == "error" and "message" in json:
578 print("WARNING: JSON response is an error:", json["message"])
579 update_last_error(domain, json["message"])
580 return fetch_generator_from_path(domain)
581 elif "software" not in json or "name" not in json["software"]:
582 # NOISY-DEBUG: print(f"DEBUG: JSON response from {domain} does not include [software][name], fetching / ...")
583 software = fetch_generator_from_path(domain)
585 # NOISY-DEBUG: print(f"DEBUG: Generator for domain='{domain}' is: {software}, EXIT!")
588 software = tidyup(json["software"]["name"])
590 # NOISY-DEBUG: print("DEBUG: sofware after tidyup():", software)
591 if software in ["akkoma", "rebased"]:
592 # NOISY-DEBUG: print("DEBUG: Setting pleroma:", domain, software)
594 elif software in ["hometown", "ecko"]:
595 # NOISY-DEBUG: print("DEBUG: Setting mastodon:", domain, software)
596 software = "mastodon"
597 elif software in ["calckey", "groundpolis", "foundkey", "cherrypick", "meisskey"]:
598 # NOISY-DEBUG: print("DEBUG: Setting misskey:", domain, software)
600 elif software.find("/") > 0:
601 print("WARNING: Spliting of slash:", software)
602 software = software.split("/")[-1];
603 elif software.find("|") > 0:
604 print("WARNING: Spliting of pipe:", software)
605 software = tidyup(software.split("|")[0]);
606 elif "powered by" in software:
607 # NOISY-DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it")
608 software = strip_powered_by(software)
609 elif type(software) is str and " by " in software:
610 # NOISY-DEBUG: print(f"DEBUG: software='{software}' has ' by ' in it")
611 software = strip_until(software, " by ")
612 elif type(software) is str and " see " in software:
613 # NOISY-DEBUG: print(f"DEBUG: software='{software}' has ' see ' in it")
614 software = strip_until(software, " see ")
616 # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}")
618 print("WARNING: tidyup() left no software name behind:", domain)
621 # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}")
622 if str(software) == "":
623 # NOISY-DEBUG: print(f"DEBUG: software for '{domain}' was not detected, trying generator ...")
624 software = fetch_generator_from_path(domain)
625 elif len(str(software)) > 0 and "." in software:
626 # NOISY-DEBUG: print(f"DEBUG: software='{software}' may contain a version number, domain='{domain}', removing it ...")
627 software = remove_version(software)
629 # NOISY-DEBUG: print(f"DEBUG: software[]={type(software)}")
630 if type(software) is str and "powered by" in software:
631 # NOISY-DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it")
632 software = remove_version(strip_powered_by(software))
634 # NOISY-DEBUG: print("DEBUG: Returning domain,software:", domain, software)
637 def update_block_reason(reason: str, blocker: str, blocked: str, block_level: str):
638 # NOISY-DEBUG: print("DEBUG: Updating block reason:", reason, blocker, blocked, block_level)
641 "UPDATE blocks SET reason = ?, last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ? AND reason = ''",
651 # NOISY-DEBUG: print(f"DEBUG: cursor.rowcount={cursor.rowcount}")
652 if cursor.rowcount == 0:
653 print("WARNING: Did not update any rows:", domain)
655 except BaseException as e:
656 print("ERROR: failed SQL query:", reason, blocker, blocked, block_level, e)
659 # NOISY-DEBUG: print("DEBUG: EXIT!")
661 def update_last_seen(blocker: str, blocked: str, block_level: str):
662 # NOISY-DEBUG: print("DEBUG: Updating last_seen for:", blocker, blocked, block_level)
665 "UPDATE blocks SET last_seen = ? WHERE blocker = ? AND blocked = ? AND block_level = ?",
674 if cursor.rowcount == 0:
675 print("WARNING: Did not update any rows:", domain)
677 except BaseException as e:
678 print("ERROR: failed SQL query:", last_seen, blocker, blocked, block_level, e)
681 # NOISY-DEBUG: print("DEBUG: EXIT!")
683 def block_instance(blocker: str, blocked: str, reason: str, block_level: str):
684 # NOISY-DEBUG: print("DEBUG: blocker,blocked,reason,block_level:", blocker, blocked, reason, block_level)
685 if not validators.domain(blocker):
686 print("WARNING: Bad blocker:", blocker)
688 elif not validators.domain(blocked):
689 print("WARNING: Bad blocked:", blocked)
692 print("INFO: New block:", blocker, blocked, reason, block_level, first_added, last_seen)
695 "INSERT INTO blocks (blocker, blocked, reason, block_level, first_added, last_seen) VALUES(?, ?, ?, ?, ?, ?)",
706 except BaseException as e:
707 print("ERROR: failed SQL query:", blocker, blocked, reason, block_level, first_added, last_seen, e)
710 # NOISY-DEBUG: print("DEBUG: EXIT!")
712 def add_instance(domain: str, origin: str, originator: str):
713 # NOISY-DEBUG: print("DEBUG: domain,origin:", domain, origin, originator)
714 if not validators.domain(domain):
715 print("WARNING: Bad domain name:", domain)
717 elif origin is not None and not validators.domain(origin):
718 print("WARNING: Bad origin name:", origin)
721 software = determine_software(domain)
722 # NOISY-DEBUG: print("DEBUG: Determined software:", software)
724 print(f"INFO: Adding instance {domain} (origin: {origin})")
727 "INSERT INTO instances (domain, origin, originator, hash, software, first_seen) VALUES (?, ?, ?, ?, ?, ?)",
738 for key in nodeinfos:
739 # NOISY-DEBUG: pprint(f"DEBUG: key='{key}',domain='{domain}',nodeinfos[key]={nodeinfos[key]}")
740 if domain in nodeinfos[key]:
741 # NOISY-DEBUG: pprint(f"DEBUG: domain='{domain}' has pending nodeinfo being updated ...")
742 update_nodeinfos(domain)
743 remove_pending_error(domain)
746 if domain in pending_errors:
747 # NOISY-DEBUG: print("DEBUG: domain has pending error being updated:", domain)
748 update_last_error(domain, pending_errors[domain])
749 remove_pending_error(domain)
751 except BaseException as e:
752 print("ERROR: failed SQL query:", domain, e)
755 # NOISY-DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain)
756 update_last_nodeinfo(domain)
758 # NOISY-DEBUG: print("DEBUG: EXIT!")
760 def send_bot_post(instance: str, blocks: dict):
761 message = instance + " has blocked the following instances:\n\n"
766 blocks = blocks[0 : 19]
769 if block["reason"] == None or block["reason"] == '':
770 message = message + block["blocked"] + " with unspecified reason\n"
772 if len(block["reason"]) > 420:
773 block["reason"] = block["reason"][0:419] + "[…]"
775 message = message + block["blocked"] + ' for "' + block["reason"].replace("@", "@\u200b") + '"\n'
778 message = message + "(the list has been truncated to the first 20 entries)"
780 botheaders = {**headers, **{"Authorization": "Bearer " + config["bot_token"]}}
783 f"{config['bot_instance']}/api/v1/statuses",
786 "visibility" : config['bot_visibility'],
787 "content_type": "text/plain"
795 def get_mastodon_blocks(domain: str) -> dict:
796 # NOISY-DEBUG: print("DEBUG: Fetching mastodon blocks from domain:", domain)
798 "Suspended servers": [],
799 "Filtered media" : [],
800 "Limited servers" : [],
801 "Silenced servers" : [],
805 doc = bs4.BeautifulSoup(
806 reqto.get(f"https://{domain}/about/more", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
809 except BaseException as e:
810 print("ERROR: Cannot fetch from domain:", domain, e)
811 update_last_error(domain, e)
814 for header in doc.find_all("h3"):
815 header_text = tidyup(header.text)
817 if header_text in language_mapping:
818 # NOISY-DEBUG: print(f"DEBUG: header_text='{header_text}'")
819 header_text = language_mapping[header_text]
821 if header_text in blocks or header_text.lower() in blocks:
822 # replaced find_next_siblings with find_all_next to account for instances that e.g. hide lists in dropdown menu
823 for line in header.find_all_next("table")[0].find_all("tr")[1:]:
824 blocks[header_text].append(
826 "domain": tidyup(line.find("span").text),
827 "hash" : tidyup(line.find("span")["title"][9:]),
828 "reason": tidyup(line.find_all("td")[1].text),
832 # NOISY-DEBUG: print("DEBUG: Returning blocks for domain:", domain)
834 "reject" : blocks["Suspended servers"],
835 "media_removal" : blocks["Filtered media"],
836 "followers_only": blocks["Limited servers"] + blocks["Silenced servers"],
839 def get_friendica_blocks(domain: str) -> dict:
840 # NOISY-DEBUG: print("DEBUG: Fetching friendica blocks from domain:", domain)
844 doc = bs4.BeautifulSoup(
845 reqto.get(f"https://{domain}/friendica", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"])).text,
848 except BaseException as e:
849 print("WARNING: Failed to fetch /friendica from domain:", domain, e)
850 update_last_error(domain, e)
853 blocklist = doc.find(id="about_blocklist")
855 # Prevents exceptions:
856 if blocklist is None:
857 # NOISY-DEBUG: print("DEBUG:Instance has no block list:", domain)
860 for line in blocklist.find("table").find_all("tr")[1:]:
862 "domain": tidyup(line.find_all("td")[0].text),
863 "reason": tidyup(line.find_all("td")[1].text)
866 # NOISY-DEBUG: print("DEBUG: Returning blocks() for domain:", domain, len(blocks))
871 def get_misskey_blocks(domain: str) -> dict:
872 # NOISY-DEBUG: print("DEBUG: Fetching misskey blocks from domain:", domain)
881 # iterating through all "suspended" (follow-only in its terminology)
882 # instances page-by-page, since that troonware doesn't support
883 # sending them all at once
886 # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
887 doc = post_json_api(domain, "/api/federation/instances/", json.dumps({
888 "sort" : "+caughtAt",
894 # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
895 doc = post_json_api(domain, "/api/federation/instances/", json.dumps({
896 "sort" : "+caughtAt",
903 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
905 # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
910 if instance["isSuspended"]:
911 blocks["suspended"].append(
913 "domain": tidyup(instance["host"]),
914 # no reason field, nothing
920 # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
923 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
924 counter = counter + step
926 except BaseException as e:
927 print("WARNING: Caught error, exiting loop:", domain, e)
928 update_last_error(domain, e)
933 # same shit, different asshole ("blocked" aka full suspend)
936 # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
937 doc = post_json_api(domain,"/api/federation/instances", json.dumps({
938 "sort" : "+caughtAt",
944 # NOISY-DEBUG: print("DEBUG: Sending JSON API request to domain,step,counter:", domain, step, counter)
945 doc = post_json_api(domain,"/api/federation/instances", json.dumps({
946 "sort" : "+caughtAt",
953 # NOISY-DEBUG: print("DEBUG: doc():", len(doc))
955 # NOISY-DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
959 if instance["isBlocked"]:
960 blocks["blocked"].append({
961 "domain": tidyup(instance["host"]),
966 # NOISY-DEBUG: print("DEBUG: End of request:", len(doc), step)
969 # NOISY-DEBUG: print("DEBUG: Raising counter by step:", step)
970 counter = counter + step
972 except BaseException as e:
973 print("ERROR: Exception during POST:", domain, e)
974 update_last_error(domain, e)
978 # NOISY-DEBUG: print("DEBUG: Returning for domain,blocked(),suspended():", domain, len(blocks["blocked"]), len(blocks["suspended"]))
980 "reject" : blocks["blocked"],
981 "followers_only": blocks["suspended"]
984 def tidyup(string: str) -> str:
985 # some retards put their blocks in variable case
986 string = string.lower().strip()
988 # other retards put the port
989 string = re.sub("\:\d+$", "", string)
991 # bigger retards put the schema in their blocklist, sometimes even without slashes
992 string = re.sub("^https?\:(\/*)", "", string)
995 string = re.sub("\/$", "", string)
998 string = re.sub("^\@", "", string)
1000 # the biggest retards of them all try to block individual users
1001 string = re.sub("(.+)\@", "", string)