#!/usr/bin/env bats
# Seam: the Anubis bot-challenge proxy (dankclaw/anubis). Anubis sits between
# Caddy's public listeners and the guarded vhosts: clients hit Caddy (TLS),
# Caddy proxies everything through Anubis, Anubis verifies (DENY known AI
# scrapers, CHALLENGE everyone else per policy.json) and forwards accepted
# traffic back to Caddy's internal listener at :8016, where per-vhost routing
# resumes. Static assertions only — the container runs on dankclaw.

load helpers

setup() {
    UNIT="$REPO_ROOT/dankclaw/anubis/quadlet/anubis.container"
    POLICY="$REPO_ROOT/dankclaw/anubis/policy.json"
    CADDY="$REPO_ROOT/dankclaw/Caddyfile"
}

@test "the anubis quadlet unit exists and is boot-persistent by [Install]" {
    test -f "$UNIT"
    grep -q 'WantedBy=default.target' "$UNIT"
    # Generated units cannot be enabled on this host (v1 quadlet generator);
    # boot persistence comes from [Install], not from systemctl enable.
    ! grep -q '^Enable=true' "$UNIT"
}

@test "the image is pinned to an exact upstream tag and never pulled at boot" {
    # The unit runs a localhost/ alias tag of the canonical upstream ref
    # (podman 5.7 quadlet quirk: registry-prefixed refs fail container
    # create on this host). The pin is still an exact upstream tag — the
    # deploy handler pulls the canonical ghcr.io ref and re-tags the alias;
    # nothing is pulled at boot (Pull=never).
    grep -qE '^Image=localhost/techarohq/anubis:v[0-9]+\.[0-9]+\.[0-9]+$' "$UNIT"
    grep -q '^Pull=never$' "$UNIT"
    grep -q 'podman pull ghcr\.io/techarohq/anubis:v[0-9]*\.[0-9]*\.[0-9]*' "$REPO_ROOT/scripts/deploy-remote.sh"
}

@test "anubis publishes no host port — it is internal-network-only" {
    # Metrics (:9090) are unauthenticated; the protection strategy is that
    # only dankclaw_default members can reach the container at all. Caddy
    # reaches it by container DNS name, so no published port may exist.
    ! grep -q '^PublishPort=' "$UNIT"
}

@test "the unit points TARGET back at caddy's internal listener" {
    grep -q '^Environment=TARGET=http://caddy:8016$' "$UNIT"
    grep -q '^Environment=POLICY_FNAME=/policy/policy.json$' "$UNIT"
    # The policy file is mounted read-only from the repo checkout on the host.
    grep -q 'policy.json:/policy/policy.json:ro' "$UNIT"
}

@test "the policy parses as JSON and matches v1.13.0 schema (bots array, action rules)" {
    # At the pinned tag the parser is JSON-only with a minimal schema:
    # {"bots":[{"name","user_agent_regex"|"path_regex","action"}],"dnsbl":bool}
    # — imports/CEL/expression keys do not exist at this tag and would crash
    # the container at parse time. Assert the shape statically via python.
    python3 - "$POLICY" <<'PY'
import json, sys
with open(sys.argv[1]) as f:
    p = json.load(f)
assert set(p.keys()) <= {"bots", "dnsbl"}, p.keys()
actions = [b["action"] for b in p["bots"]]
assert any(a == "DENY" for a in actions), actions
assert actions[-1] == "CHALLENGE", actions
for b in p["bots"]:
    keys = set(b) ^ {"name", "action"}
    assert len(keys) == 1, f"{b.get('name')}: exactly one matcher required, got {keys}"
    assert (keys ^ {"user_agent_regex"}) != (keys ^ {"path_regex"}), b
PY
}

@test "guarded vhosts proxy through anubis with the real client IP forwarded" {
    # Anubis reads X-Real-Ip for its decisions; without the header every
    # request looks like it comes from the Caddy container. Every guarded
    # public listener must set it.
    grep -q 'header_up X-Real-Ip {remote_host}' "$CADDY"
}

@test "exempt vhosts never route through anubis (auth, plex, own-auth apps)" {
    # The IdP login flow and git smart-HTTP clients cannot solve challenges;
    # these vhosts bypass the chain by construction.
    local block
    block=$(awk '/^auth\.franksdank\.cloud/,/^}/' "$CADDY")
    echo "$block" | grep -q 'reverse_proxy authentik-server:9000'
    ! echo "$block" | grep -q anubis
    block=$(awk '/^plex\.franksdank\.cloud/,/^}/' "$CADDY")
    echo "$block" | grep -q '192\.168\.0\.235:32400'
    ! echo "$block" | grep -q anubis
}