majuna/app/lists/mirror_mapping.py

76 lines
2.4 KiB
Python

import logging
from datetime import datetime, timedelta
from typing import Dict, List, Optional, TypedDict
from flask import current_app
from sqlalchemy import or_
from sqlalchemy.orm import selectinload
from tldextract import extract
from app.extensions import db
from app.models.base import Group, Pool
from app.models.mirrors import Proxy, Origin
class MirrorMappingMirror(TypedDict):
origin_domain: str
origin_domain_normalized: str
origin_domain_root: str
valid_from: str
valid_to: Optional[str]
countries: Dict[str, int]
country: Optional[str]
risk: int
class MirrorMapping(TypedDict):
version: str
mappings: Dict[str, MirrorMappingMirror]
s3_buckets: List[str]
def mirror_mapping(_: Optional[Pool]) -> MirrorMapping:
two_days_ago = datetime.utcnow() - timedelta(days=2)
proxies = (
db.session.query(Proxy)
.options(selectinload(Proxy.origin).selectinload(Origin.countries))
.filter(or_(Proxy.destroyed.is_(None), Proxy.destroyed > two_days_ago))
.filter(Proxy.url.is_not(None))
.all()
)
result: Dict[str, MirrorMappingMirror] = {}
for proxy in proxies:
if proxy.url is None:
logging.error("No URL for proxy %s", proxy)
continue
countries = proxy.origin.risk_level
if countries:
highest_risk_country_code, highest_risk_level = max(countries.items(), key=lambda x: x[1])
else:
highest_risk_country_code = "ZZ"
highest_risk_level = 0
result[proxy.url.lstrip("https://")] = {
"origin_domain": proxy.origin.domain_name,
"origin_domain_normalized": proxy.origin.domain_name.replace("www.", ""),
"origin_domain_root": extract(proxy.origin.domain_name).registered_domain,
"valid_from": proxy.added.isoformat(),
"valid_to": proxy.destroyed.isoformat() if proxy.destroyed else None,
"countries": countries,
"country": highest_risk_country_code,
"risk": highest_risk_level
}
groups = db.session.query(Group).options(selectinload(Group.pools))
s3_buckets = [
f"{current_app.config['GLOBAL_NAMESPACE']}-{g.group_name.lower()}-logs-cloudfront"
for g in groups.filter(Group.destroyed.is_(None)).all()
]
return {
"version": "1.2",
"mappings": result,
"s3_buckets": s3_buckets
}