// This file is part of the happyDomain (R) project.
// Copyright (c) 2020-2026 happyDomain
// Authors: Pierre-Olivier Mercier, et al.

package checker

import (
	"context"
	"fmt"
	"net/http"
)

// ObservationKeyWellKnown is the Extensions[] key under which
// wellknownCollector publishes its observation.
const ObservationKeyWellKnown = "wellknown"

// WellKnownData captures whether each well-known URI returned a usable
// document. It is intentionally narrow: per-URI presence and HTTP status
// are enough for the current rule set; deeper parsing (e.g. PGP-signed
// security.txt fields) is left to dedicated collectors when the need
// arises.
type WellKnownData struct {
	URIs map[string]WellKnownProbe `json:"uris"`
}

// WellKnownProbe is a single (URI → outcome) entry.
type WellKnownProbe = PathProbe

// wellknownCollector probes a small, fixed set of standardised URIs
// served at the apex of the host. Today it covers:
//
//   - /.well-known/security.txt (RFC 9116) — security disclosure contact
//   - /robots.txt (RFC 9309)               — crawler directives
//
// It uses the first IP only because these documents are expected to be
// host-uniform: there is nothing to learn from probing every backend.
type wellknownCollector struct{}

func (wellknownCollector) Key() string { return ObservationKeyWellKnown }

func (wellknownCollector) Collect(ctx context.Context, t Target) (any, error) {
	if len(t.IPs) == 0 {
		return nil, fmt.Errorf("no IPs to probe")
	}
	transport, cleanup := newPinnedHTTPSTransport(t.IPs[0], t.Host, t.Timeout)
	defer cleanup()
	client := &http.Client{Transport: transport}

	uris := []string{"/.well-known/security.txt", "/robots.txt"}
	out := WellKnownData{URIs: make(map[string]WellKnownProbe, len(uris))}
	for _, path := range uris {
		out.URIs[path] = fetchHTTPSPath(ctx, client, t.Host, path, t.UserAgent, 64<<10)
	}
	return &out, nil
}

func init() { RegisterCollector(wellknownCollector{}) }