-# Fedi API Block - An aggregator for fetching blocking data from fediverse nodes
# Copyright (C) 2023 Free Software Foundation
#
# This program is free software: you can redistribute it and/or modify
# You should have received a copy of the GNU Affero General Public License
# along with this program. If not, see <https://www.gnu.org/licenses/>.
-import bs4
import hashlib
-import re
-import reqto
-import requests
-import json
import sqlite3
-import sys
-import time
-import validators
from urllib.parse import urlparse
-from fba import blacklist
-from fba import cache
-from fba import config
-from fba import instances
-
-from fba.federation import lemmy
-from fba.federation import misskey
-from fba.federation import peertube
-
-# Array with pending errors needed to be written to database
-pending_errors = {
-}
-
-# "rel" identifiers (no real URLs)
-nodeinfo_identifier = [
- "https://nodeinfo.diaspora.software/ns/schema/2.1",
- "https://nodeinfo.diaspora.software/ns/schema/2.0",
- "https://nodeinfo.diaspora.software/ns/schema/1.1",
- "https://nodeinfo.diaspora.software/ns/schema/1.0",
- "http://nodeinfo.diaspora.software/ns/schema/2.1",
- "http://nodeinfo.diaspora.software/ns/schema/2.0",
- "http://nodeinfo.diaspora.software/ns/schema/1.1",
- "http://nodeinfo.diaspora.software/ns/schema/1.0",
-]
-
-# HTTP headers for non-API requests
-headers = {
- "User-Agent": config.get("useragent"),
-}
+import requests
+import validators
-# HTTP headers for API requests
-api_headers = {
- "User-Agent": config.get("useragent"),
- "Content-Type": "application/json",
-}
+from fba import blacklist
+from fba import federation
+from fba import network
-# URL for fetching peers
-get_peers_url = "/api/v1/instance/peers"
+from fba.models import instances
# Connect to database
connection = sqlite3.connect("blocks.db")
cursor = connection.cursor()
-# Pattern instance for version numbers
-patterns = [
- # semantic version number (with v|V) prefix)
- 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-]+)*))?)?$"),
- # non-sematic, e.g. 1.2.3.4
- re.compile("^(?P<version>v|V{0,1})(\.{0,1})(?P<major>0|[1-9]\d*)\.(?P<minor>0+|[1-9]\d*)(\.(?P<patch>0+|[1-9]\d*)(\.(?P<subpatch>0|[1-9]\d*))?)$"),
- # non-sematic, e.g. 2023-05[-dev]
- re.compile("^(?P<year>[1-9]{1}[0-9]{3})\.(?P<month>[0-9]{2})(-dev){0,1}$"),
- # non-semantic, e.g. abcdef0
- re.compile("^[a-f0-9]{7}$"),
-]
-
##### Other functions #####
def is_primitive(var: any) -> bool:
# DEBUG: print(f"DEBUG: var[]='{type(var)}' - CALLED!")
- return type(var) in {int, str, float, bool} or var == None
-
-def fetch_instances(domain: str, origin: str, software: str, script: str, path: str = None):
- # DEBUG: print(f"DEBUG: domain='{domain}',origin='{origin}',software='{software}',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(origin) != str and origin != None:
- raise ValueError(f"Parameter origin[]={type(origin)} is not 'str'")
- elif type(script) != str:
- raise ValueError(f"Parameter script[]={type(script)} is not 'str'")
- elif domain == "":
- raise ValueError(f"Parameter 'domain' is empty")
-
- if not is_instance_registered(domain):
- # DEBUG: print("DEBUG: Adding new domain:", domain, origin)
- add_instance(domain, origin, script, path)
-
- # DEBUG: print("DEBUG: Fetching instances for domain:", domain, software)
- peerlist = get_peers(domain, software)
-
- if (peerlist is None):
- print("ERROR: Cannot fetch peers:", domain)
- return
- elif instances.has_pending_instance_data(domain):
- # DEBUG: print(f"DEBUG: domain='{domain}' has pending nodeinfo data, flushing ...")
- instances.update_instance_data(domain)
-
- print(f"INFO: Checking {len(peerlist)} instances from {domain} ...")
- for instance in peerlist:
- if instance == None:
- # Skip "None" types as tidup() cannot parse them
- continue
-
- # DEBUG: print(f"DEBUG: instance='{instance}' - BEFORE")
- instance = tidyup_domain(instance)
- # DEBUG: print(f"DEBUG: instance='{instance}' - AFTER")
-
- if instance == "":
- print("WARNING: Empty instance after tidyup_domain(), domain:", domain)
- continue
- elif not validators.domain(instance.split("/")[0]):
- print(f"WARNING: Bad instance='{instance}' from domain='{domain}',origin='{origin}',software='{software}'")
- continue
- elif blacklist.is_blacklisted(instance):
- # DEBUG: print("DEBUG: instance is blacklisted:", instance)
- continue
-
- # DEBUG: print("DEBUG: Handling instance:", instance)
- try:
- if not is_instance_registered(instance):
- # DEBUG: print("DEBUG: Adding new instance:", instance, domain)
- add_instance(instance, domain, script)
- except BaseException as e:
- print(f"ERROR: instance='{instance}',exception[{type(e)}]:'{str(e)}'")
- continue
-
- # DEBUG: print("DEBUG: EXIT!")
-
-def add_peers(rows: dict) -> list:
- # DEBUG: print(f"DEBUG: rows()={len(rows)} - CALLED!")
- peers = list()
- for key in ["linked", "allowed", "blocked"]:
- # DEBUG: print(f"DEBUG: Checking key='{key}'")
- if key in rows and rows[key] != None:
- # DEBUG: print(f"DEBUG: Adding {len(rows[key])} peer(s) to peers list ...")
- for peer in rows[key]:
- # DEBUG: print(f"DEBUG: peer='{peer}' - BEFORE!")
- peer = tidyup_domain(peer)
-
- # DEBUG: print(f"DEBUG: peer='{peer}' - AFTER!")
- if blacklist.is_blacklisted(peer):
- # DEBUG: print(f"DEBUG: peer='{peer}' is blacklisted, skipped!")
- continue
-
- # DEBUG: print(f"DEBUG: Adding peer='{peer}' ...")
- peers.append(peer)
-
- # DEBUG: print(f"DEBUG: peers()={len(peers)} - EXIT!")
- return peers
-
-def remove_version(software: str) -> str:
- # DEBUG: print(f"DEBUG: software='{software}' - CALLED!")
- if not "." in software and " " not in software:
- print(f"WARNING: software='{software}' does not contain a version number.")
- return software
-
- temp = software
- if ";" in software:
- temp = software.split(";")[0]
- elif "," in software:
- temp = software.split(",")[0]
- elif " - " in software:
- temp = software.split(" - ")[0]
-
- # DEBUG: print(f"DEBUG: software='{software}'")
- version = None
- if " " in software:
- version = temp.split(" ")[-1]
- elif "/" in software:
- version = temp.split("/")[-1]
- elif "-" in software:
- version = temp.split("-")[-1]
- else:
- # DEBUG: print(f"DEBUG: Was not able to find common seperator, returning untouched software='{software}'")
- return software
-
- matches = None
- match = None
- # DEBUG: print(f"DEBUG: Checking {len(patterns)} patterns ...")
- for pattern in patterns:
- # Run match()
- match = pattern.match(version)
-
- # DEBUG: print(f"DEBUG: match[]={type(match)}")
- if type(match) is re.Match:
- break
-
- # DEBUG: print(f"DEBUG: version[{type(version)}]='{version}',match='{match}'")
- if type(match) is not re.Match:
- print(f"WARNING: version='{version}' does not match regex, leaving software='{software}' untouched.")
- return software
-
- # DEBUG: print(f"DEBUG: Found valid version number: '{version}', removing it ...")
- end = len(temp) - len(version) - 1
-
- # DEBUG: print(f"DEBUG: end[{type(end)}]={end}")
- software = temp[0:end].strip()
- if " version" in software:
- # DEBUG: print(f"DEBUG: software='{software}' contains word ' version'")
- software = strip_until(software, " version")
-
- # DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
- return software
-
-def strip_powered_by(software: str) -> str:
- # DEBUG: print(f"DEBUG: software='{software}' - CALLED!")
- if software == "":
- print(f"ERROR: Bad method call, 'software' is empty")
- raise Exception("Parameter 'software' is empty")
- elif not "powered by" in software:
- print(f"WARNING: Cannot find 'powered by' in '{software}'!")
- return software
-
- start = software.find("powered by ")
- # DEBUG: print(f"DEBUG: start[{type(start)}]='{start}'")
-
- software = software[start + 11:].strip()
- # DEBUG: print(f"DEBUG: software='{software}'")
-
- software = strip_until(software, " - ")
-
- # DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
- return software
-
-def strip_hosted_on(software: str) -> str:
- # DEBUG: print(f"DEBUG: software='{software}' - CALLED!")
- if software == "":
- print(f"ERROR: Bad method call, 'software' is empty")
- raise Exception("Parameter 'software' is empty")
- elif not "hosted on" in software:
- print(f"WARNING: Cannot find 'hosted on' in '{software}'!")
- return software
-
- end = software.find("hosted on ")
- # DEBUG: print(f"DEBUG: end[{type(end)}]='{end}'")
-
- software = software[0, start].strip()
- # DEBUG: print(f"DEBUG: software='{software}'")
-
- software = strip_until(software, " - ")
-
- # DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
- return software
-
-def strip_until(software: str, until: str) -> str:
- # DEBUG: print(f"DEBUG: software='{software}',until='{until}' - CALLED!")
- if software == "":
- print(f"ERROR: Bad method call, 'software' is empty")
- raise Exception("Parameter 'software' is empty")
- elif until == "":
- print(f"ERROR: Bad method call, 'until' is empty")
- raise Exception("Parameter 'until' is empty")
- elif not until in software:
- print(f"WARNING: Cannot find '{until}' in '{software}'!")
- return software
-
- # Next, strip until part
- end = software.find(until)
-
- # DEBUG: print(f"DEBUG: end[{type(end)}]='{end}'")
- if end > 0:
- software = software[0:end].strip()
-
- # DEBUG: print(f"DEBUG: software='{software}' - EXIT!")
- return software
-
-def remove_pending_error(domain: str):
- if type(domain) != str:
- raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
- elif domain == "":
- raise ValueError(f"Parameter 'domain' is empty")
-
- try:
- # Prevent updating any pending errors, nodeinfo was found
- del pending_errors[domain]
-
- except:
- pass
-
- # DEBUG: print("DEBUG: EXIT!")
+ return type(var) in {int, str, float, bool} or var is None
def get_hash(domain: str) -> str:
- if type(domain) != str:
- raise ValueError(f"Parameter domain[]={type(domain)} is not 'str'")
- elif domain == "":
- raise ValueError(f"Parameter 'domain' is empty")
-
- return hashlib.sha256(domain.encode("utf-8")).hexdigest()
-
-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 update_last_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")
-
- # DEBUG: print("DEBUG: BEFORE response[]:", type(response))
- if isinstance(response, BaseException) or isinstance(response, json.decoder.JSONDecodeError):
- response = f"{type}:str(response)"
-
- # DEBUG: print("DEBUG: AFTER response[]:", type(response))
- if type(response) is str:
- # DEBUG: print(f"DEBUG: Setting last_error_details='{response}'");
- instances.set("last_status_code" , domain, 999)
- instances.set("last_error_details", domain, response)
- else:
- # DEBUG: print(f"DEBUG: Setting last_error_details='{response.reason}'");
- instances.set("last_status_code" , domain, response.status_code)
- instances.set("last_error_details", domain, response.reason)
-
- # Running pending updated
- # DEBUG: print(f"DEBUG: Invoking instances.update_instance_data({domain}) ...")
- instances.update_instance_data(domain)
-
- log_error(domain, response)
-
- # DEBUG: print("DEBUG: EXIT!")
-
-def update_last_nodeinfo(domain: str):
- # 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: Updating last_nodeinfo for domain:", domain)
- instances.set("last_nodeinfo", domain, time.time())
- instances.set("last_updated" , domain, time.time())
-
- # Running pending updated
- # DEBUG: print(f"DEBUG: Invoking instances.update_instance_data({domain}) ...")
- instances.update_instance_data(domain)
-
- # DEBUG: print("DEBUG: EXIT!")
-
-def get_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.get_peers({domain}) ...")
- return misskey.get_peers(domain)
- elif software == "lemmy":
- # DEBUG: print(f"DEBUG: Invoking lemmy.get_peers({domain}) ...")
- return lemmy.get_peers(domain)
- elif software == "peertube":
- # DEBUG: print(f"DEBUG: Invoking peertube.get_peers({domain}) ...")
- return peertube.get_peers(domain)
-
- # DEBUG: print(f"DEBUG: Fetching get_peers_url='{get_peers_url}' from '{domain}',software='{software}' ...")
- peers = list()
- try:
- response = get_response(domain, get_peers_url, 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 '{get_peers_url}', trying alternative ...")
- response = get_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)
- 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)
- 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)
- 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 post_json_api(domain: str, path: str, parameter: str, extra_headers: dict = {}) -> dict:
- # DEBUG: print(f"DEBUG: domain='{domain}',path='{path}',parameter='{parameter}',extra_headers()={len(extra_headers)} - 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("Parameter 'path' cannot be empty")
- elif type(parameter) != str:
- raise ValueError(f"parameter[]={type(parameter)} is not 'str'")
-
- # DEBUG: print("DEBUG: Sending POST to domain,path,parameter:", domain, path, parameter, extra_headers)
- data = {}
- try:
- response = reqto.post(
- f"https://{domain}{path}",
- data=parameter,
- headers={**api_headers, **extra_headers},
- timeout=(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(f"WARNING: Cannot query JSON API: domain='{domain}',path='{path}',parameter()={len(parameter)},response.status_code='{response.status_code}',data[]='{type(data)}'")
- update_last_error(domain, response)
-
- except BaseException as e:
- print(f"WARNING: Some error during post(): domain='{domain}',path='{path}',parameter()={len(parameter)},exception[{type(e)}]:'{str(e)}'")
-
- # DEBUG: print(f"DEBUG: Returning data({len(data)})=[]:{type(data)}")
- return data
-
-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 = get_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)
- update_last_error(domain, response)
- continue
-
- except BaseException as e:
- # DEBUG: print("DEBUG: Cannot fetch API request:", request)
- 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 = get_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 = get_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)
- 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 = get_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)
- 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"])
- 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"])
- 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 is_instance_registered(domain: str) -> bool:
- # 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(f"DEBUG: domain='{domain}' - CALLED!")
- if not cache.key_exists("is_registered"):
- # DEBUG: print(f"DEBUG: Cache for 'is_registered' not initialized, fetching all rows ...")
- try:
- cursor.execute("SELECT domain FROM instances")
-
- # Check Set all
- cache.set_all("is_registered", cursor.fetchall(), True)
- except BaseException as e:
- print(f"ERROR: failed SQL query: domain='{domain}',exception[{type(e)}]:'{str(e)}'")
- sys.exit(255)
-
- # Is cache found?
- registered = cache.sub_key_exists("is_registered", domain)
-
- # DEBUG: print(f"DEBUG: registered='{registered}' - EXIT!")
- return registered
-
-def add_instance(domain: str, origin: str, originator: str, path: str = None):
- # DEBUG: print(f"DEBUG: domain='{domain}',origin='{origin}',originator='{originator}',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(origin) != str and origin != None:
- raise ValueError(f"origin[]={type(origin)} is not 'str'")
- elif type(originator) != str:
- raise ValueError(f"originator[]={type(originator)} is not 'str'")
- elif originator == "":
- raise ValueError(f"originator cannot be empty")
- elif not validators.domain(domain.split("/")[0]):
- raise ValueError(f"Bad domain name='{domain}'")
- elif origin is not None and not validators.domain(origin.split("/")[0]):
- raise ValueError(f"Bad origin name='{origin}'")
- elif blacklist.is_blacklisted(domain):
- raise Exception(f"domain='{domain}' is blacklisted, but method invoked")
-
- # DEBUG: print("DEBUG: domain,origin,originator,path:", domain, origin, originator, path)
- software = determine_software(domain, path)
- # DEBUG: print("DEBUG: Determined software:", software)
-
- print(f"INFO: Adding instance domain='{domain}' (origin='{origin}',software='{software}')")
- try:
- cursor.execute(
- "INSERT INTO instances (domain, origin, originator, hash, software, first_seen) VALUES (?, ?, ?, ?, ?, ?)",
- (
- domain,
- origin,
- originator,
- get_hash(domain),
- software,
- time.time()
- ),
- )
-
- cache.set_sub_key("is_registered", domain, True)
-
- if instances.has_pending_instance_data(domain):
- # DEBUG: print(f"DEBUG: domain='{domain}' has pending nodeinfo being updated ...")
- instances.set("last_status_code" , domain, None)
- instances.set("last_error_details", domain, None)
- instances.update_instance_data(domain)
- remove_pending_error(domain)
-
- if domain in pending_errors:
- # DEBUG: print("DEBUG: domain has pending error being updated:", domain)
- update_last_error(domain, pending_errors[domain])
- remove_pending_error(domain)
-
- except BaseException as e:
- print(f"ERROR: failed SQL query: domain='{domain}',exception[{type(e)}]:'{str(e)}'")
- sys.exit(255)
- else:
- # DEBUG: print("DEBUG: Updating nodeinfo for domain:", domain)
- update_last_nodeinfo(domain)
-
- # DEBUG: print("DEBUG: EXIT!")
-
-def send_bot_post(instance: str, blocklist: dict):
- # DEBUG: print(f"DEBUG: instance={instance},blocklist()={len(blocklist)} - CALLED!")
- if type(domain) != str:
+ if not isinstance(domain, str):
raise ValueError(f"Parameter domain[]='{type(domain)}' is not 'str'")
elif domain == "":
raise ValueError("Parameter 'domain' is empty")
- elif type(blocklist) != dict:
- raise ValueError(f"Parameter blocklist[]='{type(blocklist)}' is not 'dict'")
-
- message = instance + " has blocked the following instances:\n\n"
- truncated = False
-
- if len(blocklist) > 20:
- truncated = True
- blocklist = blocklist[0 : 19]
-
- # DEBUG: print(f"DEBUG: blocklist()={len(blocklist)}")
- for block in blocklist:
- # DEBUG: print(f"DEBUG: block['{type(block)}']={block}")
- if block["reason"] == None or block["reason"] == '':
- message = message + block["blocked"] + " with unspecified reason\n"
- else:
- if len(block["reason"]) > 420:
- block["reason"] = block["reason"][0:419] + "[…]"
-
- message = message + block["blocked"] + ' for "' + block["reason"].replace("@", "@\u200b") + '"\n'
-
- if truncated:
- message = message + "(the list has been truncated to the first 20 entries)"
-
- botheaders = {**api_headers, **{"Authorization": "Bearer " + config.get("bot_token")}}
-
- req = reqto.post(
- f"{config.get('bot_instance')}/api/v1/statuses",
- data={
- "status" : message,
- "visibility" : config.get('bot_visibility'),
- "content_type": "text/plain"
- },
- headers=botheaders,
- timeout=10
- ).json()
-
- return True
-
-def fetch_friendica_blocks(domain: str) -> dict:
- # 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 friendica blocks from domain:", domain)
- blocklist = []
-
- try:
- doc = bs4.BeautifulSoup(
- get_response(domain, "/friendica", headers, (config.get("connection_timeout"), config.get("read_timeout"))).text,
- "html.parser",
- )
- except BaseException as e:
- print("WARNING: Failed to fetch /friendica from domain:", domain, e)
- update_last_error(domain, e)
- return {}
-
- blocklist = doc.find(id="about_blocklist")
-
- # Prevents exceptions:
- if blocklist is None:
- # DEBUG: print("DEBUG: Instance has no block list:", domain)
- return {}
- for line in blocklist.find("table").find_all("tr")[1:]:
- # DEBUG: print(f"DEBUG: line='{line}'")
- blocklist.append({
- "domain": tidyup_domain(line.find_all("td")[0].text),
- "reason": tidyup_domain(line.find_all("td")[1].text)
- })
-
- # DEBUG: print("DEBUG: Returning blocklist() for domain:", domain, len(blocklist))
- return {
- "reject": blocklist
- }
-
-def fetch_misskey_blocks(domain: str) -> dict:
- # 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 misskey blocks from domain:", domain)
- blocklist = {
- "suspended": [],
- "blocked" : []
- }
-
- offset = 0
- step = config.get("misskey_limit")
- while True:
- # iterating through all "suspended" (follow-only in its terminology)
- # instances page-by-page, since that troonware doesn't support
- # sending them all at once
- try:
- # DEBUG: print(f"DEBUG: Fetching offset='{offset}' from '{domain}' ...")
- if offset == 0:
- # DEBUG: print("DEBUG: Sending JSON API request to domain,step,offset:", domain, step, offset)
- fetched = post_json_api(domain, "/api/federation/instances", json.dumps({
- "sort" : "+pubAt",
- "host" : None,
- "suspended": True,
- "limit" : step
- }), {
- "Origin": domain
- })
- else:
- # DEBUG: print("DEBUG: Sending JSON API request to domain,step,offset:", domain, step, offset)
- fetched = post_json_api(domain, "/api/federation/instances", json.dumps({
- "sort" : "+pubAt",
- "host" : None,
- "suspended": True,
- "limit" : step,
- "offset" : offset - 1
- }), {
- "Origin": domain
- })
-
- # DEBUG: print("DEBUG: fetched():", len(fetched))
- if len(fetched) == 0:
- # DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
- break
- elif len(fetched) != config.get("misskey_limit"):
- # DEBUG: print(f"DEBUG: Fetched '{len(fetched)}' row(s) but expected: '{config.get('misskey_limit')}'")
- offset = offset + (config.get("misskey_limit") - len(fetched))
- else:
- # DEBUG: print("DEBUG: Raising offset by step:", step)
- offset = offset + step
-
- count = 0
- for instance in fetched:
- # Is it there?
- if instance["isSuspended"] and not has_key(blocklist["suspended"], "domain", instance):
- count = count + 1
- blocklist["suspended"].append(
- {
- "domain": tidyup_domain(instance["host"]),
- # no reason field, nothing
- "reason": None
- }
- )
-
- # DEBUG: print(f"DEBUG: count={count}")
- if count == 0:
- # DEBUG: print(f"DEBUG: API is no more returning new instances, aborting loop!")
- break
-
- except BaseException as e:
- print("WARNING: Caught error, exiting loop:", domain, e)
- update_last_error(domain, e)
- offset = 0
- break
-
- while True:
- # same shit, different asshole ("blocked" aka full suspend)
- try:
- if offset == 0:
- # DEBUG: print("DEBUG: Sending JSON API request to domain,step,offset:", domain, step, offset)
- fetched = post_json_api(domain,"/api/federation/instances", json.dumps({
- "sort" : "+pubAt",
- "host" : None,
- "blocked": True,
- "limit" : step
- }), {
- "Origin": domain
- })
- else:
- # DEBUG: print("DEBUG: Sending JSON API request to domain,step,offset:", domain, step, offset)
- fetched = post_json_api(domain,"/api/federation/instances", json.dumps({
- "sort" : "+pubAt",
- "host" : None,
- "blocked": True,
- "limit" : step,
- "offset" : offset - 1
- }), {
- "Origin": domain
- })
-
- # DEBUG: print("DEBUG: fetched():", len(fetched))
- if len(fetched) == 0:
- # DEBUG: print("DEBUG: Returned zero bytes, exiting loop:", domain)
- break
- elif len(fetched) != config.get("misskey_limit"):
- # DEBUG: print(f"DEBUG: Fetched '{len(fetched)}' row(s) but expected: '{config.get('misskey_limit')}'")
- offset = offset + (config.get("misskey_limit") - len(fetched))
- else:
- # DEBUG: print("DEBUG: Raising offset by step:", step)
- offset = offset + step
-
- count = 0
- for instance in fetched:
- # Is it there?
- if instance["isBlocked"] and not has_key(blocklist["blocked"], "domain", instance):
- count = count + 1
- blocklist["blocked"].append({
- "domain": tidyup_domain(instance["host"]),
- "reason": None
- })
-
- # DEBUG: print(f"DEBUG: count={count}")
- if count == 0:
- # DEBUG: print(f"DEBUG: API is no more returning new instances, aborting loop!")
- break
-
- except BaseException as e:
- print("ERROR: Exception during POST:", domain, e)
- update_last_error(domain, e)
- offset = 0
- break
-
- # DEBUG: print(f"DEBUG: Updating last_instance_fetch for domain='{domain}' ...")
- instances.update_last_instance_fetch(domain)
-
- # DEBUG: print("DEBUG: Returning for domain,blocked(),suspended():", domain, len(blocklist["blocked"]), len(blocklist["suspended"]))
- return {
- "reject" : blocklist["blocked"],
- "followers_only": blocklist["suspended"]
- }
-
-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 expected")
-
- # 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 expected")
-
- # 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)
-
- # 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 get_response(domain: str, path: str, headers: dict, timeout: list) -> requests.models.Response:
- # DEBUG: print(f"DEBUG: domain='{domain}',path='{path}',headers()={len(headers)},timeout={timeout} - CALLED!")
- if type(domain) != str:
- raise ValueError(f"Parameter domain[]='{type(domain)}' is not 'str'")
- elif domain == "":
- raise ValueError("Parameter 'domain' is empty")
- elif type(path) != str:
- raise ValueError(f"Parameter path[]='{type(path)}' is not 'str'")
- elif path == "":
- raise ValueError("Parameter 'path' is empty")
-
- try:
- # DEBUG: print(f"DEBUG: Sending request to '{domain}{path}' ...")
- response = reqto.get(
- f"https://{domain}{path}",
- headers=headers,
- timeout=timeout
- );
- except requests.exceptions.ConnectionError as e:
- # DEBUG: print(f"DEBUG: Fetching '{path}' from '{domain}' failed. exception[{type(e)}]='{str(e)}'")
- update_last_error(domain, e)
- raise e
-
- # DEBUG: print(f"DEBUG: response[]='{type(response)}' - EXXIT!")
- return response
-
-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
+ return hashlib.sha256(domain.encode("utf-8")).hexdigest()
-def get_url(url: str, headers: dict, timeout: list) -> requests.models.Response:
+def fetch_url(url: str, headers: dict, timeout: tuple) -> requests.models.Response:
# DEBUG: print(f"DEBUG: url='{url}',headers()={len(headers)},timeout={timeout} - CALLED!")
- if type(url) != str:
+ if not isinstance(url, str):
raise ValueError(f"Parameter url[]='{type(url)}' is not 'str'")
elif url == "":
raise ValueError("Parameter 'url' is empty")
+ elif not isinstance(headers, dict):
+ raise ValueError(f"Parameter headers[]='{type(headers)}' is not 'dict'")
+ elif not isinstance(timeout, tuple):
+ raise ValueError(f"Parameter timeout[]='{type(timeout)}' is not 'tuple'")
# DEBUG: print(f"DEBUG: Parsing url='{url}'")
components = urlparse(url)
# Invoke other function, avoid trailing ?
# DEBUG: print(f"DEBUG: components[{type(components)}]={components}")
if components.query != "":
- response = get_response(components.hostname, f"{components.path}?{components.query}", headers, timeout)
+ response = network.fetch_response(components.netloc, f"{components.path}?{components.query}", headers, timeout)
else:
- response = get_response(components.hostname, f"{components.path}", headers, timeout)
+ response = network.fetch_response(components.netloc, f"{components.path}", headers, timeout)
# DEBUG: print(f"DEBUG: response[]='{type(response)}' - EXXIT!")
return response
+
+def process_domain(domain: str, blocker: str, command: str) -> bool:
+ # DEBUG: print(f"DEBUG: domain='{domain}',blocker='{blocker}',command='{command}' - CALLED!")
+ if not isinstance(domain, str):
+ raise ValueError(f"Parameter domain[]='{type(domain)}' is not 'str'")
+ elif domain == "":
+ raise ValueError("Parameter 'domain' is empty")
+ elif not isinstance(blocker, str):
+ raise ValueError(f"Parameter blocker[]='{type(blocker)}' is not 'str'")
+ elif blocker == "":
+ raise ValueError("Parameter 'blocker' is empty")
+ elif not isinstance(command, str):
+ raise ValueError(f"Parameter command[]='{type(command)}' is not 'str'")
+ elif command == "":
+ raise ValueError("Parameter 'command' is empty")
+
+ if domain.find("*") > 0:
+ # Try to de-obscure it
+ row = instances.deobscure("*", domain)
+
+ # DEBUG: print(f"DEBUG: row[{type(row)}]='{row}'")
+ if row is None:
+ print(f"WARNING: Cannot de-obfucate domain='{domain}' - SKIPPED!")
+ return False
+
+ # DEBUG: print(f"DEBUG: domain='{domain}' de-obscured to '{row[0]}'")
+ domain = row[0]
+ elif domain.find("?") > 0:
+ # Try to de-obscure it
+ row = instances.deobscure("?", domain)
+
+ # DEBUG: print(f"DEBUG: row[{type(row)}]='{row}'")
+ if row is None:
+ print(f"WARNING: Cannot de-obfucate domain='{domain}' - SKIPPED!")
+ return False
+
+ # DEBUG: print(f"DEBUG: domain='{domain}' de-obscured to '{row[0]}'")
+ domain = row[0]
+
+ if not validators.domain(domain):
+ print(f"WARNING: domain='{domain}' is not a valid domain - SKIPPED!")
+ return False
+ elif domain.endswith(".arpa"):
+ print(f"WARNING: domain='{domain}' is a reversed .arpa domain and should not be used generally.")
+ return False
+ elif blacklist.is_blacklisted(domain):
+ # DEBUG: print(f"DEBUG: domain='{domain}' is blacklisted - SKIPPED!")
+ return False
+ elif instances.is_recent(domain):
+ # DEBUG: print(f"DEBUG: domain='{domain}' has been recently checked - SKIPPED!")
+ return False
+
+ processed = False
+ try:
+ print(f"INFO: Fetching instances for instane='{domain}',blocker='{blocker}',command='{command}' ...")
+ federation.fetch_instances(domain, blocker, None, command)
+ processed = True
+ except network.exceptions as exception:
+ print(f"WARNING: Exception '{type(exception)}' during fetching instances (fetch_oliphant) from domain='{domain}'")
+ instances.set_last_error(domain, exception)
+
+ # DEBUG: print(f"DEBUG: processed='{processed}' - EXIT!")
+ return processed
projects / fba.git / blobdiff