-def log_error(domain: str, response: requests.models.Response):
- # DEBUG: print("DEBUG: domain,response[]:", domain, type(response))
- if type(domain) != str:
- raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
- elif domain == "":
- raise ValueError(f"Parameter 'domain' is empty")
-
- try:
- # DEBUG: print("DEBUG: BEFORE response[]:", type(response))
- if isinstance(response, BaseException) or isinstance(response, json.decoder.JSONDecodeError):
- response = str(response)
-
- # DEBUG: print("DEBUG: AFTER response[]:", type(response))
- if type(response) is str:
- cursor.execute("INSERT INTO error_log (domain, error_code, error_message, created) VALUES (?, 999, ?, ?)",[
- domain,
- response,
- time.time()
- ])
- else:
- cursor.execute("INSERT INTO error_log (domain, error_code, error_message, created) VALUES (?, ?, ?, ?)",[
- domain,
- response.status_code,
- response.reason,
- time.time()
- ])
-
- # Cleanup old entries
- # DEBUG: print(f"DEBUG: Purging old records (distance: {config.get('error_log_cleanup')})")
- cursor.execute("DELETE FROM error_log WHERE created < ?", [time.time() - config.get("error_log_cleanup")])
- except BaseException as e:
- print(f"ERROR: failed SQL query: domain='{domain}',exception[{type(e)}]:'{str(e)}'")
- sys.exit(255)
-
- # DEBUG: print("DEBUG: EXIT!")
-
-def fetch_peers(domain: str, software: str) -> list:
- # DEBUG: print(f"DEBUG: domain({len(domain)})={domain},software={software} - CALLED!")
- if type(domain) != str:
- raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
- elif domain == "":
- raise ValueError(f"Parameter 'domain' is empty")
- elif type(software) != str and software != None:
- raise ValueError(f"software[]={type(software)} is not 'str'")
-
- if software == "misskey":
- # DEBUG: print(f"DEBUG: Invoking misskey.fetch_peers({domain}) ...")
- return misskey.fetch_peers(domain)
- elif software == "lemmy":
- # DEBUG: print(f"DEBUG: Invoking lemmy.fetch_peers({domain}) ...")
- return lemmy.fetch_peers(domain)
- elif software == "peertube":
- # DEBUG: print(f"DEBUG: Invoking peertube.fetch_peers({domain}) ...")
- return peertube.fetch_peers(domain)
-
- # DEBUG: print(f"DEBUG: Fetching peers from '{domain}',software='{software}' ...")
- peers = list()
- try:
- response = network.fetch_response(domain, "/api/v1/instance/peers", api_headers, (config.get("connection_timeout"), config.get("read_timeout")))
-
- data = json_from_response(response)
-
- # DEBUG: print(f"DEBUG: response.ok={response.ok},response.status_code={response.status_code},data[]='{type(data)}'")
- if not response.ok or response.status_code >= 400:
- # DEBUG: print(f"DEBUG: Was not able to fetch peers, trying alternative ...")
- response = network.fetch_response(domain, "/api/v3/site", api_headers, (config.get("connection_timeout"), config.get("read_timeout")))
-
- data = json_from_response(response)
- # DEBUG: print(f"DEBUG: response.ok={response.ok},response.status_code={response.status_code},data[]='{type(data)}'")
- if not response.ok or response.status_code >= 400:
- print("WARNING: Could not reach any JSON API:", domain)
- instances.update_last_error(domain, response)
- elif response.ok and isinstance(data, list):
- # DEBUG: print(f"DEBUG: domain='{domain}' returned a list: '{data}'")
- sys.exit(255)
- elif "federated_instances" in data:
- # DEBUG: print(f"DEBUG: Found federated_instances for domain='{domain}'")
- peers = peers + add_peers(data["federated_instances"])
- # DEBUG: print("DEBUG: Added instance(s) to peers")
- else:
- print("WARNING: JSON response does not contain 'federated_instances':", domain)
- instances.update_last_error(domain, response)
- else:
- # DEBUG: print("DEBUG: Querying API was successful:", domain, len(data))
- peers = data
-
- except BaseException as e:
- print("WARNING: Some error during get():", domain, e)
- instances.update_last_error(domain, e)
-
- # DEBUG: print(f"DEBUG: Adding '{len(peers)}' for domain='{domain}'")
- instances.set("total_peers", domain, len(peers))
-
- # DEBUG: print(f"DEBUG: Updating last_instance_fetch for domain='{domain}' ...")
- instances.update_last_instance_fetch(domain)
-
- # DEBUG: print("DEBUG: Returning peers[]:", type(peers))
- return peers
-
-def fetch_nodeinfo(domain: str, path: str = None) -> list:
- # DEBUG: print(f"DEBUG: domain='{domain}',path={path} - CALLED!")
- if type(domain) != str:
- raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
- elif domain == "":
- raise ValueError(f"Parameter 'domain' is empty")
- elif type(path) != str and path != None:
- raise ValueError(f"Parameter path[]={type(path)} is not 'str'")
-
- # DEBUG: print(f"DEBUG: Fetching nodeinfo from domain='{domain}' ...")
- nodeinfo = fetch_wellknown_nodeinfo(domain)
-
- # DEBUG: print(f"DEBUG: nodeinfo({len(nodeinfo)})={nodeinfo}")
- if len(nodeinfo) > 0:
- # DEBUG: print("DEBUG: nodeinfo()={len(nodeinfo))} - EXIT!")
- return nodeinfo
-
- request_paths = [
- "/nodeinfo/2.1.json",
- "/nodeinfo/2.1",
- "/nodeinfo/2.0.json",
- "/nodeinfo/2.0",
- "/nodeinfo/1.0",
- "/api/v1/instance"
- ]
-
- data = {}
- for request in request_paths:
- if path != None and path != "" and path != f"https://{domain}{path}":
- # DEBUG: print(f"DEBUG: path='{path}' does not match request='{request}' - SKIPPED!")
- continue
-
- try:
- # DEBUG: print(f"DEBUG: Fetching request='{request}' from domain='{domain}' ...")
- response = network.fetch_response(domain, request, api_headers, (config.get("nodeinfo_connection_timeout"), config.get("nodeinfo_read_timeout")))
-
- data = json_from_response(response)
- # DEBUG: print(f"DEBUG: response.ok={response.ok},response.status_code={response.status_code},data[]='{type(data)}'")
- if response.ok and isinstance(data, dict):
- # DEBUG: print("DEBUG: Success:", request)
- instances.set("detection_mode", domain, "STATIC_CHECK")
- instances.set("nodeinfo_url" , domain, request)
- break
- elif response.ok and isinstance(data, list):
- print(f"UNSUPPORTED: domain='{domain}' returned a list: '{data}'")
- sys.exit(255)
- elif not response.ok or response.status_code >= 400:
- print("WARNING: Failed fetching nodeinfo from domain:", domain)
- instances.update_last_error(domain, response)
- continue
-
- except BaseException as e:
- # DEBUG: print("DEBUG: Cannot fetch API request:", request)
- instances.update_last_error(domain, e)
- pass
-
- # DEBUG: print(f"DEBUG: data()={len(data)} - EXIT!")
- return data
-
-def fetch_wellknown_nodeinfo(domain: str) -> list:
- # DEBUG: print(f"DEBUG: domain='{domain}' - CALLED!")
- if type(domain) != str:
- raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
- elif domain == "":
- raise ValueError(f"Parameter 'domain' is empty")
-
- # DEBUG: print("DEBUG: Fetching .well-known info for domain:", domain)
- data = {}
-
- try:
- response = network.fetch_response(domain, "/.well-known/nodeinfo", api_headers, (config.get("nodeinfo_connection_timeout"), config.get("nodeinfo_read_timeout")))
-
- data = json_from_response(response)
- # DEBUG: print("DEBUG: domain,response.ok,data[]:", domain, response.ok, type(data))
- if response.ok and isinstance(data, dict):
- nodeinfo = data
- # DEBUG: print("DEBUG: Found entries:", len(nodeinfo), domain)
- if "links" in nodeinfo:
- # DEBUG: print("DEBUG: Found links in nodeinfo():", len(nodeinfo["links"]))
- for link in nodeinfo["links"]:
- # DEBUG: print("DEBUG: rel,href:", link["rel"], link["href"])
- if link["rel"] in nodeinfo_identifier:
- # DEBUG: print("DEBUG: Fetching nodeinfo from:", link["href"])
- response = fetch_url(link["href"], api_headers, (config.get("connection_timeout"), config.get("read_timeout")))
-
- data = json_from_response(response)
- # DEBUG: print("DEBUG: href,response.ok,response.status_code:", link["href"], response.ok, response.status_code)
- if response.ok and isinstance(data, dict):
- # DEBUG: print("DEBUG: Found JSON nodeinfo():", len(data))
- instances.set("detection_mode", domain, "AUTO_DISCOVERY")
- instances.set("nodeinfo_url" , domain, link["href"])
- break
- else:
- print("WARNING: Unknown 'rel' value:", domain, link["rel"])
- else:
- print("WARNING: nodeinfo does not contain 'links':", domain)
-
- except BaseException as e:
- print("WARNING: Failed fetching .well-known info:", domain)
- instances.update_last_error(domain, e)
- pass
-
- # DEBUG: print("DEBUG: Returning data[]:", type(data))
- return data
-
-def fetch_generator_from_path(domain: str, path: str = "/") -> str:
- # DEBUG: print(f"DEBUG: domain({len(domain)})={domain},path={path} - CALLED!")
- if type(domain) != str:
- raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
- elif domain == "":
- raise ValueError(f"Parameter 'domain' is empty")
- elif type(path) != str:
- raise ValueError(f"path[]={type(path)} is not 'str'")
- elif path == "":
- raise ValueError(f"Parameter 'domain' is empty")
-
- # DEBUG: print(f"DEBUG: domain='{domain}',path='{path}' - CALLED!")
- software = None
-
- try:
- # DEBUG: print(f"DEBUG: Fetching path='{path}' from '{domain}' ...")
- response = network.fetch_response(domain, path, headers, (config.get("connection_timeout"), config.get("read_timeout")))
-
- # DEBUG: print("DEBUG: domain,response.ok,response.status_code,response.text[]:", domain, response.ok, response.status_code, type(response.text))
- if response.ok and response.status_code < 300 and len(response.text) > 0:
- # DEBUG: print("DEBUG: Search for <meta name='generator'>:", domain)
- doc = bs4.BeautifulSoup(response.text, "html.parser")
-
- # DEBUG: print("DEBUG: doc[]:", type(doc))
- generator = doc.find("meta", {"name": "generator"})
- site_name = doc.find("meta", {"property": "og:site_name"})
-
- # DEBUG: print(f"DEBUG: generator='{generator}',site_name='{site_name}'")
- if isinstance(generator, bs4.element.Tag):
- # DEBUG: print("DEBUG: Found generator meta tag:", domain)
- software = tidyup_domain(generator.get("content"))
- print(f"INFO: domain='{domain}' is generated by '{software}'")
- instances.set("detection_mode", domain, "GENERATOR")
- remove_pending_error(domain)
- elif isinstance(site_name, bs4.element.Tag):
- # DEBUG: print("DEBUG: Found property=og:site_name:", domain)
- sofware = tidyup_domain(site_name.get("content"))
- print(f"INFO: domain='{domain}' has og:site_name='{software}'")
- instances.set("detection_mode", domain, "SITE_NAME")
- remove_pending_error(domain)
-
- except BaseException as e:
- # DEBUG: print(f"DEBUG: Cannot fetch / from '{domain}':", e)
- instances.update_last_error(domain, e)
- pass
-
- # DEBUG: print(f"DEBUG: software[]={type(software)}")
- if type(software) is str and software == "":
- # DEBUG: print(f"DEBUG: Corrected empty string to None for software of domain='{domain}'")
- software = None
- elif type(software) is str and ("." in software or " " in software):
- # DEBUG: print(f"DEBUG: software='{software}' may contain a version number, domain='{domain}', removing it ...")
- software = remove_version(software)
-
- # DEBUG: print(f"DEBUG: software[]={type(software)}")
- if type(software) is str and " powered by " in software:
- # DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it")
- software = remove_version(strip_powered_by(software))
- elif type(software) is str and " hosted on " in software:
- # DEBUG: print(f"DEBUG: software='{software}' has 'hosted on' in it")
- software = remove_version(strip_hosted_on(software))
- elif type(software) is str and " by " in software:
- # DEBUG: print(f"DEBUG: software='{software}' has ' by ' in it")
- software = strip_until(software, " by ")
- elif type(software) is str and " see " in software:
- # DEBUG: print(f"DEBUG: software='{software}' has ' see ' in it")
- software = strip_until(software, " see ")
-
- # DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
- return software
-
-def determine_software(domain: str, path: str = None) -> str:
- # DEBUG: print(f"DEBUG: domain({len(domain)})={domain},path={path} - CALLED!")
- if type(domain) != str:
- raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
- elif domain == "":
- raise ValueError(f"Parameter 'domain' is empty")
- elif type(path) != str and path != None:
- raise ValueError(f"Parameter path[]={type(path)} is not 'str'")
-
- # DEBUG: print("DEBUG: Determining software for domain,path:", domain, path)
- software = None
-
- # DEBUG: print(f"DEBUG: Fetching nodeinfo from '{domain}' ...")
- data = fetch_nodeinfo(domain, path)
-
- # DEBUG: print("DEBUG: data[]:", type(data))
- if not isinstance(data, dict) or len(data) == 0:
- # DEBUG: print("DEBUG: Could not determine software type:", domain)
- return fetch_generator_from_path(domain)
-
- # DEBUG: print("DEBUG: data():", len(data), data)
- if "status" in data and data["status"] == "error" and "message" in data:
- print("WARNING: JSON response is an error:", data["message"])
- instances.update_last_error(domain, data["message"])
- return fetch_generator_from_path(domain)
- elif "message" in data:
- print("WARNING: JSON response contains only a message:", data["message"])
- instances.update_last_error(domain, data["message"])
- return fetch_generator_from_path(domain)
- elif "software" not in data or "name" not in data["software"]:
- # DEBUG: print(f"DEBUG: JSON response from domain='{domain}' does not include [software][name], fetching / ...")
- software = fetch_generator_from_path(domain)
-
- # DEBUG: print(f"DEBUG: Generator for domain='{domain}' is: {software}, EXIT!")
- return software
-
- software = tidyup_domain(data["software"]["name"])
-
- # DEBUG: print("DEBUG: sofware after tidyup_domain():", software)
- if software in ["akkoma", "rebased"]:
- # DEBUG: print("DEBUG: Setting pleroma:", domain, software)
- software = "pleroma"
- elif software in ["hometown", "ecko"]:
- # DEBUG: print("DEBUG: Setting mastodon:", domain, software)
- software = "mastodon"
- elif software in ["calckey", "groundpolis", "foundkey", "cherrypick", "meisskey"]:
- # DEBUG: print("DEBUG: Setting misskey:", domain, software)
- software = "misskey"
- elif software.find("/") > 0:
- print("WARNING: Spliting of slash:", software)
- software = tidup_domain(software.split("/")[-1]);
- elif software.find("|") > 0:
- print("WARNING: Spliting of pipe:", software)
- software = tidyup_domain(software.split("|")[0]);
- elif "powered by" in software:
- # DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it")
- software = strip_powered_by(software)
- elif type(software) is str and " by " in software:
- # DEBUG: print(f"DEBUG: software='{software}' has ' by ' in it")
- software = strip_until(software, " by ")
- elif type(software) is str and " see " in software:
- # DEBUG: print(f"DEBUG: software='{software}' has ' see ' in it")
- software = strip_until(software, " see ")
-
- # DEBUG: print(f"DEBUG: software[]={type(software)}")
- if software == "":
- print("WARNING: tidyup_domain() left no software name behind:", domain)
- software = None
-
- # DEBUG: print(f"DEBUG: software[]={type(software)}")
- if str(software) == "":
- # DEBUG: print(f"DEBUG: software for '{domain}' was not detected, trying generator ...")
- software = fetch_generator_from_path(domain)
- elif len(str(software)) > 0 and ("." in software or " " in software):
- # DEBUG: print(f"DEBUG: software='{software}' may contain a version number, domain='{domain}', removing it ...")
- software = remove_version(software)
-
- # DEBUG: print(f"DEBUG: software[]={type(software)}")
- if type(software) is str and "powered by" in software:
- # DEBUG: print(f"DEBUG: software='{software}' has 'powered by' in it")
- software = remove_version(strip_powered_by(software))
-
- # DEBUG: print("DEBUG: Returning domain,software:", domain, software)
- return software
-
-def tidyup_reason(reason: str) -> str:
- # DEBUG: print(f"DEBUG: reason='{reason}' - CALLED!")
- if type(reason) != str:
- raise ValueError(f"Parameter reason[]={type(reason)} is not 'str'")
-
- # Strip string
- reason = reason.strip()
-
- # Replace â with "
- reason = re.sub("â", "\"", reason)
-
- # DEBUG: print(f"DEBUG: reason='{reason}' - EXIT!")
- return reason
-
-def tidyup_domain(domain: str) -> str:
- # DEBUG: print(f"DEBUG: domain='{domain}' - CALLED!")
- if type(domain) != str:
- raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
-
- # All lower-case and strip spaces out + last dot
- domain = domain.lower().strip().rstrip(".")
-
- # No port number
- domain = re.sub("\:\d+$", "", domain)
-
- # No protocol, sometimes without the slashes
- domain = re.sub("^https?\:(\/*)", "", domain)
-
- # No trailing slash
- domain = re.sub("\/$", "", domain)
-
- # No @ sign
- domain = re.sub("^\@", "", domain)
-
- # No individual users in block lists
- domain = re.sub("(.+)\@", "", domain)
- if domain.find("/profile/"):
- domain = domain.split("/profile/")[0]
- elif domain.find("/users/"):
- domain = domain.split("/users/")[0]
-
- # DEBUG: print(f"DEBUG: domain='{domain}' - EXIT!")
- return domain
-
-def json_from_response(response: requests.models.Response) -> list:
- # DEBUG: print(f"DEBUG: response[]={type(response)} - CALLED!")
- if not isinstance(response, requests.models.Response):
- raise ValueError(f"Parameter response[]='{type(response)}' is not type of 'Response'")
-
- data = list()
- if response.text.strip() != "":
- # DEBUG: print(f"DEBUG: response.text()={len(response.text)} is not empty, invoking response.json() ...")
- try:
- data = response.json()
- except json.decoder.JSONDecodeError:
- pass
-
- # DEBUG: print(f"DEBUG: data[]={type(data)} - EXIT!")
- return data
-
-def has_key(keys: list, search: str, value: any) -> bool:
- # DEBUG: print(f"DEBUG: keys()={len(keys)},search='{search}',value[]='{type(value)}' - CALLED!")
- if type(keys) != list:
- raise ValueError(f"Parameter keys[]='{type(keys)}' is not 'list'")
- elif type(search) != str:
- raise ValueError(f"Parameter search[]='{type(search)}' is not 'str'")
- elif search == "":
- raise ValueError("Parameter 'search' is empty")
-
- has = False
- # DEBUG: print(f"DEBUG: Checking keys()={len(keys)} ...")
- for key in keys:
- # DEBUG: print(f"DEBUG: key['{type(key)}']={key}")
- if type(key) != dict:
- raise ValueError(f"key[]='{type(key)}' is not 'dict'")
- elif not search in key:
- raise KeyError(f"Cannot find search='{search}'")
- elif key[search] == value:
- has = True
- break
-
- # DEBUG: print(f"DEBUG: has={has} - EXIT!")
- return has
-
-def find_domains(tag: bs4.element.Tag) -> list:
- # DEBUG: print(f"DEBUG: tag[]={type(tag)} - CALLED!")
- if not isinstance(tag, bs4.element.Tag):
- raise ValueError(f"Parameter tag[]={type(tag)} is not type of bs4.element.Tag")
- elif not isinstance(tag, bs4.element.Tag):
- raise KeyError("Cannot find table with instances!")
- elif len(tag.select("tr")) == 0:
- raise KeyError("No table rows found in table!")
-
- domains = list()
- for element in tag.select("tr"):
- # DEBUG: print(f"DEBUG: element[]={type(element)}")
- if not element.find("td"):
- # DEBUG: print("DEBUG: Skipping element, no <td> found")
- continue
-
- domain = tidyup_domain(element.find("td").text)
- reason = tidyup_reason(element.findAll("td")[1].text)
-
- # DEBUG: print(f"DEBUG: domain='{domain}',reason='{reason}'")
-
- if blacklist.is_blacklisted(domain):
- print(f"WARNING: domain='{domain}' is blacklisted - skipped!")
- continue
- elif domain == "gab.com/.ai, develop.gab.com":
- # DEBUG: print(f"DEBUG: Multiple domains detected in one row")
- domains.append({
- "domain": "gab.com",
- "reason": reason,
- })
- domains.append({
- "domain": "gab.ai",
- "reason": reason,
- })
- domains.append({
- "domain": "develop.gab.com",
- "reason": reason,
- })
- continue
- elif not validators.domain(domain):
- print(f"WARNING: domain='{domain}' is not a valid domain - skipped!")
- continue
-
- # DEBUG: print(f"DEBUG: Adding domain='{domain}' ...")
- domains.append({
- "domain": domain,
- "reason": reason,
- })
-
- # DEBUG: print(f"DEBUG: domains()={len(domains)} - EXIT!")
- return domains
-
-def fetch_url(url: str, headers: dict, timeout: list) -> requests.models.Response: