phishingclub/backend/server/shared.go

package server

import (
	"context"
	"net/http"
	"strings"

	"github.com/google/uuid"
	"github.com/phishingclub/phishingclub/lure"
	"github.com/phishingclub/phishingclub/model"
	"github.com/phishingclub/phishingclub/repository"
)

// LastPathSegment returns the final segment of a request path, with any
// trailing slash trimmed. A mail client or message preview commonly appends
// one, so a link that arrives as /account/4H7K9QM2XR3T/ must resolve the same
// as one without.
//
// It refuses paths carrying traversal or an encoded slash rather than trying to
// clean them, because a segment that needs cleaning is never a lure code.
func LastPathSegment(path string) (string, bool) {
	if path == "" || strings.Contains(path, "..") {
		return "", false
	}
	if strings.Contains(path, "%2f") || strings.Contains(path, "%2F") {
		return "", false
	}
	trimmed := strings.TrimSuffix(path, "/")
	if trimmed == "" {
		return "", false
	}
	index := strings.LastIndex(trimmed, "/")
	if index < 0 {
		return trimmed, true
	}
	segment := trimmed[index+1:]
	if segment == "" {
		return "", false
	}
	return segment, true
}

// TrimLastPathSegment removes the final segment of a path, which is how a
// consumed lure code is taken back out of the URL before the request is
// forwarded onward.
func TrimLastPathSegment(path string) string {
	trimmed := strings.TrimSuffix(path, "/")
	index := strings.LastIndex(trimmed, "/")
	if index <= 0 {
		return "/"
	}
	return trimmed[:index]
}

// LureMatch records how a request identified its recipient, so a caller can
// undo whatever carried the identifier before passing the request on.
type LureMatch struct {
	// ParamName is the query parameter that matched. Empty when the identifier
	// came from the path.
	ParamName string
	// PathSegment is the trailing path segment that matched. Empty when the
	// identifier came from the query.
	PathSegment string
}

// GetCampaignRecipientFromURLParams resolves the campaign recipient a request
// belongs to.
//
// Three forms are accepted, in this order:
//
//   - a query parameter holding a campaign recipient UUID, the original form
//   - a query parameter holding a lure code, so an operator set code such as
//     special-42 works as ?id=special-42 too
//   - the last path segment as a lure code, https://example.com/acc/4H7K9QM2XR3T
//
// All three are accepted regardless of the campaign's configured mode. The mode
// only decides which form is emitted, so changing it never breaks a link that
// has already been delivered.
//
// The query forms take priority over the path. An identifier in the query
// states which recipient a request belongs to, while a path segment only looks
// like a code: a query mode template whose URL path ends in a code shaped
// segment, say /careers, collides with any custom code holding that string.
//
// allowPathLookup lets a caller turn off the path form. The proxy switches it
// off once a session cookie exists, because on a proxy domain every path
// mirrors the target site and the recipient is already known by then.
func GetCampaignRecipientFromURLParams(
	ctx context.Context,
	req *http.Request,
	identifierRepo *repository.Identifier,
	campaignRecipientRepo *repository.CampaignRecipient,
	allowPathLookup bool,
) (*model.CampaignRecipient, LureMatch, error) {
	// get all identifiers
	identifiers, err := identifierRepo.GetAll(ctx, &repository.IdentifierOption{})
	if err != nil {
		return nil, LureMatch{}, err
	}

	query := req.URL.Query()
	var matchingUUIDParams []struct {
		name string
		id   *uuid.UUID
	}
	var matchingCodeParams []struct {
		name  string
		value string
	}

	// collect all query parameters that match identifier names, splitting them
	// by whether they carry a campaign recipient UUID or a lure code
	for _, identifier := range identifiers.Rows {
		name := identifier.Name.MustGet()
		if !query.Has(name) {
			continue
		}
		value := query.Get(name)
		if id, err := uuid.Parse(value); err == nil {
			matchingUUIDParams = append(matchingUUIDParams, struct {
				name string
				id   *uuid.UUID
			}{name: name, id: &id})
			continue
		}
		if lure.IsCandidate(value) {
			matchingCodeParams = append(matchingCodeParams, struct {
				name  string
				value string
			}{name: name, value: value})
		}
	}

	// check each matching parameter to find a valid campaign recipient
	for _, param := range matchingUUIDParams {
		campaignRecipient, err := campaignRecipientRepo.GetByCampaignRecipientID(ctx, param.id)
		if err == nil && campaignRecipient != nil {
			return campaignRecipient, LureMatch{ParamName: param.name}, nil
		}
	}
	for _, param := range matchingCodeParams {
		campaignRecipient, err := campaignRecipientRepo.GetByLureCode(ctx, param.value)
		if err == nil && campaignRecipient != nil {
			return campaignRecipient, LureMatch{ParamName: param.name}, nil
		}
	}

	// nothing in the query identified a recipient, so fall back to reading the
	// last path segment as a code
	if allowPathLookup {
		if segment, ok := LastPathSegment(req.URL.Path); ok {
			if lure.IsCandidate(segment) {
				campaignRecipient, err := campaignRecipientRepo.GetByLureCode(ctx, segment)
				if err == nil && campaignRecipient != nil {
					return campaignRecipient, LureMatch{PathSegment: segment}, nil
				}
			}
		}
	}

	return nil, LureMatch{}, nil
}