httpsedge_controller.go

/*
MIT License

Copyright (c) 2022 ngrok, Inc.

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.
*/

package controllers

import (
	"context"
	"fmt"
	"reflect"

	v1 "k8s.io/api/core/v1"
	apierrors "k8s.io/apimachinery/pkg/api/errors"
	"k8s.io/apimachinery/pkg/runtime"
	"k8s.io/apimachinery/pkg/types"
	"k8s.io/client-go/tools/record"
	"k8s.io/utils/pointer"
	ctrl "sigs.k8s.io/controller-runtime"
	"sigs.k8s.io/controller-runtime/pkg/client"
	"sigs.k8s.io/controller-runtime/pkg/reconcile"

	"github.com/go-logr/logr"
	ingressv1alpha1 "github.com/ngrok/kubernetes-ingress-controller/api/v1alpha1"
	"github.com/ngrok/kubernetes-ingress-controller/internal/errors"
	"github.com/ngrok/kubernetes-ingress-controller/internal/ngrokapi"
	"github.com/ngrok/ngrok-api-go/v5"
	"github.com/ngrok/ngrok-api-go/v5/backends/tunnel_group"
)

type routeModuleComparision string

const (
	routeModuleComparisonBothNil        routeModuleComparision = "both nil"
	routeModuleComparisonBothNilOrEmpty routeModuleComparision = "both nil or empty"
	routeModuleComparisonDeepEqual      routeModuleComparision = "deep equal"
)

// HTTPSEdgeReconciler reconciles a HTTPSEdge object
type HTTPSEdgeReconciler struct {
	client.Client

	Log      logr.Logger
	Scheme   *runtime.Scheme
	Recorder record.EventRecorder

	NgrokClientset ngrokapi.Clientset
}

// SetupWithManager sets up the controller with the Manager.
func (r *HTTPSEdgeReconciler) SetupWithManager(mgr ctrl.Manager) error {
	return ctrl.NewControllerManagedBy(mgr).
		For(&ingressv1alpha1.HTTPSEdge{}).
		WithEventFilter(commonPredicateFilters).
		Complete(r)
}

//+kubebuilder:rbac:groups=ingress.k8s.ngrok.com,resources=httpsedges,verbs=get;list;watch;create;update;patch;delete
//+kubebuilder:rbac:groups=ingress.k8s.ngrok.com,resources=httpsedges/status,verbs=get;update;patch
//+kubebuilder:rbac:groups=ingress.k8s.ngrok.com,resources=httpsedges/finalizers,verbs=update

// Reconcile is part of the main kubernetes reconciliation loop which aims to
// move the current state of the cluster closer to the desired state.
//
// For more details, check Reconcile and its Result here:
// - https://pkg.go.dev/sigs.k8s.io/controller-runtime@v0.13.1/pkg/reconcile
func (r *HTTPSEdgeReconciler) Reconcile(ctx context.Context, req ctrl.Request) (ctrl.Result, error) {
	log := r.Log.WithValues("V1Alpha1HTTPSEdge", req.NamespacedName)
	ctx = ctrl.LoggerInto(ctx, log)

	edge := new(ingressv1alpha1.HTTPSEdge)
	if err := r.Get(ctx, req.NamespacedName, edge); err != nil {
		log.Error(err, "unable to fetch Edge")
		return ctrl.Result{}, client.IgnoreNotFound(err)
	}

	if edge == nil {
		return ctrl.Result{}, nil
	}

	if edge.ObjectMeta.DeletionTimestamp.IsZero() {
		if err := registerAndSyncFinalizer(ctx, r.Client, edge); err != nil {
			return ctrl.Result{}, err
		}
	} else {
		// The object is being deleted
		if hasFinalizer(edge) {
			if edge.Status.ID != "" {
				r.Recorder.Event(edge, v1.EventTypeNormal, "Deleting", fmt.Sprintf("Deleting Edge %s", edge.Name))
				if err := r.NgrokClientset.HTTPSEdges().Delete(ctx, edge.Status.ID); err != nil {
					if !ngrok.IsNotFound(err) {
						r.Recorder.Event(edge, v1.EventTypeWarning, "FailedDelete", fmt.Sprintf("Failed to delete Edge %s: %s", edge.Name, err.Error()))
						return ctrl.Result{}, err
					}
				}
				edge.Status.ID = ""
			}

			if err := removeAndSyncFinalizer(ctx, r.Client, edge); err != nil {
				return ctrl.Result{}, err
			}
		}

		r.Recorder.Event(edge, v1.EventTypeNormal, "Deleted", fmt.Sprintf("Deleted Edge %s", edge.Name))

		// Stop reconciliation as the item is being deleted
		return ctrl.Result{}, nil
	}

	err := r.reconcileEdge(ctx, edge)
	if err != nil {
		log.Error(err, "error reconciling Edge")
	}
	if errors.IsErrorReconcilable(err) {
		return ctrl.Result{}, err
	} else {
		return ctrl.Result{}, nil
	}
}

func (r *HTTPSEdgeReconciler) reconcileEdge(ctx context.Context, edge *ingressv1alpha1.HTTPSEdge) error {
	var remoteEdge *ngrok.HTTPSEdge
	var err error

	logger := ctrl.LoggerFrom(ctx)

	if edge.Status.ID != "" {
		logger = logger.WithValues("ngrok.edge.id", edge.Status.ID)
		// We already have an ID, so we can just update the resource
		logger.V(1).Info("Getting existing edge")
		remoteEdge, err = r.NgrokClientset.HTTPSEdges().Get(ctx, edge.Status.ID)
		if err != nil {
			return err
		}
		logger.V(1).Info("Found existing edge")

		if !edge.Equal(remoteEdge) {
			logger.Info("Updating edge")
			remoteEdge, err = r.NgrokClientset.HTTPSEdges().Update(ctx, &ngrok.HTTPSEdgeUpdate{
				ID:          edge.Status.ID,
				Metadata:    &edge.Spec.Metadata,
				Description: &edge.Spec.Description,
				Hostports:   edge.Spec.Hostports,
			})
			if err != nil {
				return err
			}
			logger.Info("Updated edge")
		}
	} else {
		logger.Info("Searching for existing edge by hostports", "hostports", edge.Spec.Hostports)
		remoteEdge, err = r.findEdgeByHostports(ctx, edge.Spec.Hostports)
		if err != nil {
			return err
		}

		// Not found, so create it
		if remoteEdge == nil {
			logger.Info("No existing edge found. Creating new edge")
			remoteEdge, err = r.NgrokClientset.HTTPSEdges().Create(ctx, &ngrok.HTTPSEdgeCreate{
				Metadata:    edge.Spec.Metadata,
				Description: edge.Spec.Description,
				Hostports:   edge.Spec.Hostports,
			})
			if err != nil {
				return err
			}
			logger.Info("Created new edge", "ngrok.edge.id", remoteEdge.ID)
		} else {
			logger.Info("Found existing edge", "ngrok.edge.id", remoteEdge.ID)
		}
		logger = logger.WithValues("ngrok.edge.id", remoteEdge.ID)
	}

	ctx = ctrl.LoggerInto(ctx, logger)

	if err = r.updateStatus(ctx, edge, remoteEdge); err != nil {
		return err
	}

	if err = r.reconcileRoutes(ctx, edge, remoteEdge); err != nil {
		return err
	}

	return r.setEdgeTLSTermination(ctx, remoteEdge, edge.Spec.TLSTermination)
}

// TODO: This is going to be a bit messy right now, come back and make this cleaner
func (r *HTTPSEdgeReconciler) reconcileRoutes(ctx context.Context, edge *ingressv1alpha1.HTTPSEdge, remoteEdge *ngrok.HTTPSEdge) error {
	log := ctrl.LoggerFrom(ctx)

	routeStatuses := make([]ingressv1alpha1.HTTPSEdgeRouteStatus, len(edge.Spec.Routes))
	tunnelGroupReconciler, err := newTunnelGroupBackendReconciler(r.NgrokClientset.TunnelGroupBackends())
	if err != nil {
		return err
	}

	routeModuleUpdater := &edgeRouteModuleUpdater{
		edge:             edge,
		clientset:        r.NgrokClientset.EdgeModules().HTTPS().Routes(),
		ipPolicyResolver: ipPolicyResolver{r.Client},
		secretResolver:   secretResolver{r.Client},
	}

	edgeRoutes := r.NgrokClientset.HTTPSEdgeRoutes()

	// TODO: clean this up. This is way too much nesting
	for i, routeSpec := range edge.Spec.Routes {

		routeLog := log.WithValues("route.match", routeSpec.Match, "route.match_type", routeSpec.MatchType)

		if routeSpec.IPRestriction != nil {
			if err := routeModuleUpdater.ipPolicyResolver.validateIPPolicyNames(ctx, edge.Namespace, routeSpec.IPRestriction.IPPolicies); err != nil {
				if apierrors.IsNotFound(err) {
					r.Recorder.Eventf(edge, v1.EventTypeWarning, "FailedValidate", "Could not validate ip restriction: %v", err)
					continue
				}
				return err
			}
		}

		match := r.getMatchingRouteFromEdgeStatus(edge, routeSpec)
		var route *ngrok.HTTPSEdgeRoute
		// Now we go ahead and create the route if it doesn't exist.
		// It's important to note here that we are intentionally ommiting the `route.Backend` for new routes.
		//  The success or failure of applying a route's modules is then strongly linked the state of its backend.
		//  Thus, any route with a backend is considered properly configured.
		//  See https://github.com/ngrok/kubernetes-ingress-controller/issues/208 for additional context.
		if match == nil {
			routeLog.Info("Creating new route")
			req := &ngrok.HTTPSEdgeRouteCreate{
				EdgeID:    edge.Status.ID,
				Match:     routeSpec.Match,
				MatchType: routeSpec.MatchType,
			}
			route, err = edgeRoutes.Create(ctx, req)
			if err != nil {
				return err
			}
			routeLog.Info("Created new route", "ngrok.route.id", route.ID)
		} else {
			req := &ngrok.EdgeRouteItem{
				ID:     match.ID,
				EdgeID: edge.Status.ID,
			}
			route, err = edgeRoutes.Get(ctx, req)
			if err != nil {
				return err
			}
			routeLog.Info("Got existing route", "ngrok.route.id", route.ID)
		}

		routeLog = routeLog.WithValues("ngrok.route.id", route.ID)
		routeCtx := ctrl.LoggerInto(ctx, routeLog)

		if isMigratingAuthProviders(route, &routeSpec) {
			routeLog.Info("Route is migrating auth types. Taking offline before updating")
			if err := r.takeOfflineWithoutAuth(routeCtx, route); err != nil {
				r.Recorder.Event(edge, v1.EventTypeWarning, "RouteTakeOfflineFailed", err.Error())
				return err
			}
		}

		// Update status for newly created route
		routeStatuses[i] = ingressv1alpha1.HTTPSEdgeRouteStatus{
			ID:        route.ID,
			URI:       route.URI,
			Match:     route.Match,
			MatchType: route.MatchType,
		}

		// With the route properly staged, we now attempt to apply its module updates
		// TODO: Check if there are no updates to apply here to skip any unnecessary disruption
		routeLog.Info("Applying route modules")
		if err := routeModuleUpdater.updateModulesForRoute(routeCtx, route, &routeSpec); err != nil {
			r.Recorder.Event(edge, v1.EventTypeWarning, "RouteModuleUpdateFailed", err.Error())
			return err
		}

		// The route modules were successfully applied, so now we update the route with its specified backend
		backend, err := tunnelGroupReconciler.findOrCreate(routeCtx, routeSpec.Backend)
		if err != nil {
			return err
		}
		routeLog.Info("Updating route", "ngrok.backend.id", backend.ID)

		// TODO: Do an entropy check here to avoid unnecessary updates
		req := &ngrok.HTTPSEdgeRouteUpdate{
			EdgeID:    edge.Status.ID,
			ID:        route.ID,
			Match:     routeSpec.Match,
			MatchType: routeSpec.MatchType,
			Backend: &ngrok.EndpointBackendMutate{
				BackendID: backend.ID,
			},
		}
		route, err = edgeRoutes.Update(routeCtx, req)
		if err != nil {
			return err
		}
		routeLog.Info("Updated route")

		// With the route modules successfully applied and the edge updated, we now update the route's backend status
		if route.Backend != nil {
			routeStatuses[i].Backend = ingressv1alpha1.TunnelGroupBackendStatus{
				ID: route.Backend.Backend.ID,
			}
		}
	}

	log.V(1).Info("Deleting routes that are no longer in the spec")
	for _, remoteRoute := range remoteEdge.Routes {
		found := false
		for _, routeStatus := range routeStatuses {
			if routeStatus.ID == remoteRoute.ID {
				found = true
				break
			}
		}
		if !found {
			routeLog := log.WithValues("ngrok.route.id", remoteRoute.ID)
			routeLog.Info("Deleting route")
			if err := edgeRoutes.Delete(ctx, &ngrok.EdgeRouteItem{EdgeID: edge.Status.ID, ID: remoteRoute.ID}); err != nil {
				return err
			}
			routeLog.Info("Deleted route")
		}
	}

	edge.Status.Routes = routeStatuses

	return r.Status().Update(ctx, edge)
}

func (r *HTTPSEdgeReconciler) setEdgeTLSTermination(ctx context.Context, edge *ngrok.HTTPSEdge, tlsTermination *ingressv1alpha1.EndpointTLSTerminationAtEdge) error {
	log := ctrl.LoggerFrom(ctx)

	client := r.NgrokClientset.EdgeModules().HTTPS().TLSTermination()
	if tlsTermination == nil {
		if edge.TlsTermination == nil {
			log.V(1).Info("Edge TLS termination matches spec")
			return nil
		}

		log.Info("Deleting Edge TLS termination")
		return client.Delete(ctx, edge.ID)
	}

	_, err := client.Replace(ctx, &ngrok.EdgeTLSTerminationAtEdgeReplace{
		ID: edge.ID,
		Module: ngrok.EndpointTLSTerminationAtEdge{
			MinVersion: pointer.String(tlsTermination.MinVersion),
		},
	})
	return err
}

func (r *HTTPSEdgeReconciler) findEdgeByHostports(ctx context.Context, hostports []string) (*ngrok.HTTPSEdge, error) {
	iter := r.NgrokClientset.HTTPSEdges().List(&ngrok.Paging{})
	for iter.Next(ctx) {
		edge := iter.Item()

		// If the number of hostports doesn't match, then we can't match this edge
		if len(edge.Hostports) != len(hostports) {
			continue
		}

		// if the edge has all hostports, then it is the one we want. It might have
		// additional hostports.
		if r.edgeHasAllHostports(edge, hostports) {
			return edge, nil
		}
	}

	return nil, iter.Err()
}

func (r *HTTPSEdgeReconciler) edgeHasAllHostports(edge *ngrok.HTTPSEdge, hostports []string) bool {
	edgeHostportMap := make(map[string]bool)
	for _, hostport := range edge.Hostports {
		edgeHostportMap[hostport] = true
	}

	for _, hostport := range hostports {
		if _, ok := edgeHostportMap[hostport]; !ok {
			return false
		}
	}

	return true
}

func (r *HTTPSEdgeReconciler) updateStatus(ctx context.Context, edge *ingressv1alpha1.HTTPSEdge, remoteEdge *ngrok.HTTPSEdge) error {
	edge.Status.ID = remoteEdge.ID
	edge.Status.URI = remoteEdge.URI
	edge.Status.Routes = make([]ingressv1alpha1.HTTPSEdgeRouteStatus, len(remoteEdge.Routes))
	for i, route := range remoteEdge.Routes {
		edge.Status.Routes[i] = ingressv1alpha1.HTTPSEdgeRouteStatus{
			ID:        route.ID,
			URI:       route.URI,
			Match:     route.Match,
			MatchType: route.MatchType,
		}

		if route.Backend != nil {
			edge.Status.Routes[i].Backend = ingressv1alpha1.TunnelGroupBackendStatus{
				ID: route.Backend.Backend.ID,
			}
		}
	}

	return r.Status().Update(ctx, edge)
}

// getMatchingRouteFromEdgeStatus returns the route status for the given ingressv1alpha1.HTTPSEdgeRouteSpec. If there is
// no match in the ingressv1alpha1.HTTPSEdge.Status.Routes, then nil is returned. In the Ingress Spec, we can have both
// a Prefix and Exact match for the same path. In ngrok, Route match expressions must be unique across all routes for the
// edge. So we match on just the Match field and ignore the MatchType field.
func (r *HTTPSEdgeReconciler) getMatchingRouteFromEdgeStatus(edge *ingressv1alpha1.HTTPSEdge, route ingressv1alpha1.HTTPSEdgeRouteSpec) *ingressv1alpha1.HTTPSEdgeRouteStatus {
	for _, routeStatus := range edge.Status.Routes {
		if route.Match == routeStatus.Match {
			return &routeStatus
		}
	}
	return nil
}

//nolint:unused
func (r *HTTPSEdgeReconciler) listHTTPSEdgesForIPPolicy(obj client.Object) []reconcile.Request {
	r.Log.Info("Listing HTTPSEdges for ip policy to determine if they need to be reconciled")
	policy, ok := obj.(*ingressv1alpha1.IPPolicy)
	if !ok {
		r.Log.Error(nil, "failed to convert object to IPPolicy", "object", obj)
		return []reconcile.Request{}
	}

	edges := &ingressv1alpha1.HTTPSEdgeList{}
	if err := r.Client.List(context.Background(), edges); err != nil {
		r.Log.Error(err, "failed to list HTTPSEdges for ippolicy", "name", policy.Name, "namespace", policy.Namespace)
		return []reconcile.Request{}
	}

	recs := []reconcile.Request{}

	for _, edge := range edges.Items {
		for _, route := range edge.Spec.Routes {
			if route.IPRestriction == nil {
				continue
			}

			for _, edgePolicyID := range route.IPRestriction.IPPolicies {
				if edgePolicyID == policy.Name || edgePolicyID == policy.Status.ID {
					recs = append(recs, reconcile.Request{
						NamespacedName: types.NamespacedName{
							Name:      edge.GetName(),
							Namespace: edge.GetNamespace(),
						},
					})
					break
				}
			}
		}
	}

	r.Log.Info("IPPolicy change triggered HTTPSEdge reconciliation", "count", len(recs), "policy", policy.Name, "namespace", policy.Namespace)
	return recs
}

// Tunnel Group Backend planner
type tunnelGroupBackendReconciler struct {
	client   *tunnel_group.Client
	backends []*ngrok.TunnelGroupBackend
}

func newTunnelGroupBackendReconciler(client *tunnel_group.Client) (*tunnelGroupBackendReconciler, error) {
	backends := make([]*ngrok.TunnelGroupBackend, 0)
	iter := client.List(&ngrok.Paging{})
	for iter.Next(context.Background()) {
		backends = append(backends, iter.Item())
	}

	return &tunnelGroupBackendReconciler{
		client:   client,
		backends: backends,
	}, iter.Err()
}

func (r *tunnelGroupBackendReconciler) findOrCreate(ctx context.Context, backend ingressv1alpha1.TunnelGroupBackend) (*ngrok.TunnelGroupBackend, error) {
	log := ctrl.LoggerFrom(ctx).WithValues("backend.labels", backend.Labels)
	log.V(3).Info("Searching for tunnel group backend with matching labels")
	for _, b := range r.backends {
		// The labels match, so we can use this backend
		if reflect.DeepEqual(b.Labels, backend.Labels) {
			log.V(3).Info("Found matching tunnel group backend", "id", b.ID)
			return b, nil
		}
	}

	log.V(3).Info("No matching tunnel group backend found, creating a new one")
	be, err := r.client.Create(ctx, &ngrok.TunnelGroupBackendCreate{
		Description: backend.Description,
		Metadata:    backend.Metadata,
		Labels:      backend.Labels,
	})
	if err != nil {
		return nil, err
	}
	log.V(3).Info("Created new tunnel group backend", "id", be.ID)
	r.backends = append(r.backends, be)
	return be, nil
}

type edgeRouteModuleUpdater struct {
	edge *ingressv1alpha1.HTTPSEdge

	clientset ngrokapi.HTTPSEdgeRouteModulesClientset

	ipPolicyResolver ipPolicyResolver
	secretResolver   secretResolver
}

func (u *edgeRouteModuleUpdater) updateModulesForRoute(ctx context.Context, route *ngrok.HTTPSEdgeRoute, routeSpec *ingressv1alpha1.HTTPSEdgeRouteSpec) error {
	funcs := []func(context.Context, *ngrok.HTTPSEdgeRoute, *ingressv1alpha1.HTTPSEdgeRouteSpec) error{
		u.setEdgeRouteCircuitBreaker,
		u.setEdgeRouteCompression,
		u.setEdgeRouteIPRestriction,
		u.setEdgeRouteRequestHeaders,
		u.setEdgeRouteResponseHeaders,
		u.setEdgeRouteOAuth,
		u.setEdgeRouteOIDC,
		u.setEdgeRouteSAML,
		u.setEdgeRouteWebhookVerification,
	}

	for _, f := range funcs {
		if err := f(ctx, route, routeSpec); err != nil {
			return err
		}
	}
	return nil
}

func edgeRouteItem(route *ngrok.HTTPSEdgeRoute) *ngrok.EdgeRouteItem {
	return &ngrok.EdgeRouteItem{
		EdgeID: route.EdgeID,
		ID:     route.ID,
	}
}

func (u *edgeRouteModuleUpdater) logMatches(log logr.Logger, module string, checkType routeModuleComparision) {
	log.V(1).Info("Module matches desired state, skipping update", "module", module, "comparison", checkType)
}

func (u *edgeRouteModuleUpdater) setEdgeRouteCircuitBreaker(ctx context.Context, route *ngrok.HTTPSEdgeRoute, routeSpec *ingressv1alpha1.HTTPSEdgeRouteSpec) error {
	log := ctrl.LoggerFrom(ctx)
	circuitBreaker := routeSpec.CircuitBreaker

	client := u.clientset.CircuitBreaker()

	// Early return if nothing to be done
	if circuitBreaker == nil {
		if route.CircuitBreaker == nil {
			u.logMatches(log, "CircuitBreaker", routeModuleComparisonBothNil)
			return nil
		}

		log.Info("Deleting CircuitBreaker module")
		return client.Delete(ctx, edgeRouteItem(route))
	}

	module := ngrok.EndpointCircuitBreaker{
		TrippedDuration:          uint32(circuitBreaker.TrippedDuration.Seconds()),
		RollingWindow:            uint32(circuitBreaker.RollingWindow.Seconds()),
		NumBuckets:               circuitBreaker.NumBuckets,
		VolumeThreshold:          circuitBreaker.VolumeThreshold,
		ErrorThresholdPercentage: circuitBreaker.ErrorThresholdPercentage.AsApproximateFloat64(),
	}

	if reflect.DeepEqual(module, route.CircuitBreaker) {
		u.logMatches(log, "CircuitBreaker", routeModuleComparisonDeepEqual)
		return nil
	}

	log.Info("Updating CircuitBreaker", "module", module)
	_, err := client.Replace(ctx, &ngrok.EdgeRouteCircuitBreakerReplace{
		EdgeID: route.EdgeID,
		ID:     route.ID,
		Module: module,
	})
	return err
}

func (u *edgeRouteModuleUpdater) setEdgeRouteCompression(ctx context.Context, route *ngrok.HTTPSEdgeRoute, routeSpec *ingressv1alpha1.HTTPSEdgeRouteSpec) error {
	log := ctrl.LoggerFrom(ctx)
	compression := routeSpec.Compression

	client := u.clientset.Compression()

	// Early return if nothing to be done
	if compression == nil {
		if route.Compression == nil {
			u.logMatches(log, "Compression", routeModuleComparisonBothNil)
			return nil
		}

		log.Info("Deleting Compression module")
		return client.Delete(ctx, edgeRouteItem(route))
	}

	log.Info("Updating Compression", "module", compression)
	_, err := client.Replace(ctx, &ngrok.EdgeRouteCompressionReplace{
		EdgeID: route.EdgeID,
		ID:     route.ID,
		Module: ngrok.EndpointCompression{
			Enabled: pointer.Bool(routeSpec.Compression.Enabled),
		},
	})
	return err
}

func (u *edgeRouteModuleUpdater) setEdgeRouteIPRestriction(ctx context.Context, route *ngrok.HTTPSEdgeRoute, routeSpec *ingressv1alpha1.HTTPSEdgeRouteSpec) error {
	log := ctrl.LoggerFrom(ctx)
	ipRestriction := routeSpec.IPRestriction
	client := u.clientset.IPRestriction()

	if ipRestriction == nil || len(ipRestriction.IPPolicies) == 0 {
		if route.IpRestriction == nil || len(route.IpRestriction.IPPolicies) == 0 {
			u.logMatches(log, "IP Restriction", routeModuleComparisonBothNilOrEmpty)
			return nil
		}

		log.Info("Deleting IP Restriction module")
		return client.Delete(ctx, edgeRouteItem(route))
	}

	policyIds, err := u.ipPolicyResolver.resolveIPPolicyNamesorIds(ctx, u.edge.Namespace, ipRestriction.IPPolicies)
	if err != nil {
		return err
	}
	log.V(1).Info("Resolved IP Policy NamesOrIDs to IDs", "NamesOrIds", ipRestriction.IPPolicies, "policyIds", policyIds)

	var remoteIPPolicies []string
	if route.IpRestriction != nil && len(route.IpRestriction.IPPolicies) > 0 {
		remoteIPPolicies = make([]string, 0, len(route.IpRestriction.IPPolicies))
		for _, policy := range route.IpRestriction.IPPolicies {
			remoteIPPolicies = append(remoteIPPolicies, policy.ID)
		}
	}

	if reflect.DeepEqual(remoteIPPolicies, policyIds) {
		u.logMatches(log, "IP Restriction", routeModuleComparisonDeepEqual)
		return nil
	}

	log.Info("Updating IP Restriction", "policyIDs", policyIds)
	_, err = client.Replace(ctx, &ngrok.EdgeRouteIPRestrictionReplace{
		EdgeID: route.EdgeID,
		ID:     route.ID,
		Module: ngrok.EndpointIPPolicyMutate{
			IPPolicyIDs: policyIds,
		},
	})
	return err
}

func (u *edgeRouteModuleUpdater) setEdgeRouteRequestHeaders(ctx context.Context, route *ngrok.HTTPSEdgeRoute, routeSpec *ingressv1alpha1.HTTPSEdgeRouteSpec) error {
	log := ctrl.LoggerFrom(ctx)
	var requestHeaders *ingressv1alpha1.EndpointRequestHeaders
	if routeSpec.Headers != nil {
		requestHeaders = routeSpec.Headers.Request
	}

	client := u.clientset.RequestHeaders()

	if requestHeaders == nil {
		if route.RequestHeaders == nil {
			u.logMatches(log, "Request Headers", routeModuleComparisonBothNil)
			return nil
		}

		log.Info("Deleting Request Headers module")
		return client.Delete(ctx, edgeRouteItem(route))
	}

	module := ngrok.EndpointRequestHeaders{}
	if len(requestHeaders.Add) > 0 {
		module.Add = requestHeaders.Add
	}
	if len(requestHeaders.Remove) > 0 {
		module.Remove = requestHeaders.Remove
	}

	if reflect.DeepEqual(&module, route.RequestHeaders) {
		u.logMatches(log, "Request Headers", routeModuleComparisonDeepEqual)
		return nil
	}

	log.Info("Updating Request Headers", "module", module)
	_, err := client.Replace(ctx, &ngrok.EdgeRouteRequestHeadersReplace{
		EdgeID: route.EdgeID,
		ID:     route.ID,
		Module: module,
	})
	return err
}

func (u *edgeRouteModuleUpdater) setEdgeRouteResponseHeaders(ctx context.Context, route *ngrok.HTTPSEdgeRoute, routeSpec *ingressv1alpha1.HTTPSEdgeRouteSpec) error {
	log := ctrl.LoggerFrom(ctx)
	var responseHeaders *ingressv1alpha1.EndpointResponseHeaders
	if routeSpec.Headers != nil {
		responseHeaders = routeSpec.Headers.Response
	}

	client := u.clientset.ResponseHeaders()
	if responseHeaders == nil {
		if route.ResponseHeaders == nil {
			u.logMatches(log, "Response Headers", routeModuleComparisonBothNil)
			return nil
		}

		log.Info("Deleting Response Headers module")
		return client.Delete(ctx, edgeRouteItem(route))
	}

	module := ngrok.EndpointResponseHeaders{}
	if len(responseHeaders.Add) > 0 {
		module.Add = responseHeaders.Add
	}
	if len(responseHeaders.Remove) > 0 {
		module.Remove = responseHeaders.Remove
	}

	if reflect.DeepEqual(&module, route.ResponseHeaders) {
		u.logMatches(log, "Response Headers", routeModuleComparisonDeepEqual)
		return nil
	}

	log.Info("Updating Response Headers", "module", module)
	_, err := client.Replace(ctx, &ngrok.EdgeRouteResponseHeadersReplace{
		EdgeID: route.EdgeID,
		ID:     route.ID,
		Module: module,
	})
	return err
}

func (u *edgeRouteModuleUpdater) setEdgeRouteOAuth(ctx context.Context, route *ngrok.HTTPSEdgeRoute, routeSpec *ingressv1alpha1.HTTPSEdgeRouteSpec) error {
	log := ctrl.LoggerFrom(ctx)
	oauth := routeSpec.OAuth
	oauthClient := u.clientset.OAuth()

	if oauth == nil {
		if route.OAuth == nil {
			u.logMatches(log, "OAuth", routeModuleComparisonBothNil)
			return nil
		}

		log.Info("Deleting OAuth module")
		return oauthClient.Delete(ctx, edgeRouteItem(route))
	}

	var module *ngrok.EndpointOAuth
	var err error

	providers := []OAuthProvider{
		oauth.Google,
		oauth.Github,
		oauth.Gitlab,
		oauth.Amazon,
		oauth.Facebook,
		oauth.Microsoft,
		oauth.Twitch,
		oauth.Linkedin,
	}

	for _, p := range providers {
		if p == nil {
			continue
		}

		var secret *string
		secretKeyRef := p.ClientSecretKeyRef()

		// Look up the client secret key if its specified,
		// otherwise default to nil
		if secretKeyRef != nil {
			secret, err = u.getSecret(ctx, *secretKeyRef)
			if err != nil {
				return err
			}
		}

		module = p.ToNgrok(secret)
		break
	}

	if module == nil {
		return errors.NewErrInvalidConfiguration(fmt.Errorf("no OAuth provider configured"))
	}

	if reflect.DeepEqual(module, route.OAuth) {
		u.logMatches(log, "OAuth", routeModuleComparisonDeepEqual)
		return nil
	}

	log.Info("Updating OAuth module")
	_, err = oauthClient.Replace(ctx, &ngrok.EdgeRouteOAuthReplace{
		EdgeID: route.EdgeID,
		ID:     route.ID,
		Module: *module,
	})
	return err
}

func (u *edgeRouteModuleUpdater) setEdgeRouteOIDC(ctx context.Context, route *ngrok.HTTPSEdgeRoute, routeSpec *ingressv1alpha1.HTTPSEdgeRouteSpec) error {
	log := ctrl.LoggerFrom(ctx)
	oidc := routeSpec.OIDC
	client := u.clientset.OIDC()

	if oidc == nil {
		if route.OIDC == nil {
			u.logMatches(log, "OIDC", routeModuleComparisonBothNil)
			return nil
		}

		log.Info("Deleting OIDC module")
		return client.Delete(ctx, edgeRouteItem(route))
	}

	clientSecret, err := u.getSecret(ctx, oidc.ClientSecret)
	if err != nil {
		return err
	}
	if clientSecret == nil {
		return errors.NewErrMissingRequiredSecret("missing clientSecret for OIDC")
	}

	module := ngrok.EndpointOIDC{
		OptionsPassthrough: oidc.OptionsPassthrough,
		CookiePrefix:       oidc.CookiePrefix,
		InactivityTimeout:  uint32(oidc.InactivityTimeout.Seconds()),
		MaximumDuration:    uint32(oidc.MaximumDuration.Seconds()),
		Issuer:             oidc.Issuer,
		ClientID:           oidc.ClientID,
		ClientSecret:       *clientSecret,
		Scopes:             oidc.Scopes,
	}

	if reflect.DeepEqual(&module, route.OIDC) {
		u.logMatches(log, "OIDC", routeModuleComparisonDeepEqual)
		return nil
	}

	log.Info("Updating OIDC module")
	_, err = client.Replace(ctx, &ngrok.EdgeRouteOIDCReplace{
		EdgeID: route.EdgeID,
		ID:     route.ID,
		Module: module,
	})
	return err
}

func (u *edgeRouteModuleUpdater) setEdgeRouteSAML(ctx context.Context, route *ngrok.HTTPSEdgeRoute, routeSpec *ingressv1alpha1.HTTPSEdgeRouteSpec) error {
	log := ctrl.LoggerFrom(ctx)
	saml := routeSpec.SAML
	client := u.clientset.SAML()

	if saml == nil {
		if route.SAML == nil {
			u.logMatches(log, "SAML", routeModuleComparisonBothNil)
			return nil
		}

		log.Info("Deleting SAML module")
		return client.Delete(ctx, edgeRouteItem(route))
	}

	module := ngrok.EndpointSAMLMutate{
		OptionsPassthrough: saml.OptionsPassthrough,
		CookiePrefix:       saml.CookiePrefix,
		InactivityTimeout:  uint32(saml.InactivityTimeout.Seconds()),
		MaximumDuration:    uint32(saml.MaximumDuration.Seconds()),
		IdPMetadata:        saml.IdPMetadata,
		ForceAuthn:         saml.ForceAuthn,
		AllowIdPInitiated:  saml.AllowIdPInitiated,
		AuthorizedGroups:   saml.AuthorizedGroups,
		NameIDFormat:       saml.NameIDFormat,
	}

	if reflect.DeepEqual(&module, route.SAML) {
		u.logMatches(log, "SAML", routeModuleComparisonDeepEqual)
		return nil
	}

	log.Info("Updating SAML module")
	_, err := client.Replace(ctx, &ngrok.EdgeRouteSAMLReplace{
		EdgeID: route.EdgeID,
		ID:     route.ID,
		Module: module,
	})
	return err
}

func (u *edgeRouteModuleUpdater) setEdgeRouteWebhookVerification(ctx context.Context, route *ngrok.HTTPSEdgeRoute, routeSpec *ingressv1alpha1.HTTPSEdgeRouteSpec) error {
	log := ctrl.LoggerFrom(ctx)
	webhookVerification := routeSpec.WebhookVerification

	client := u.clientset.WebhookVerification()

	if webhookVerification == nil {
		if route.WebhookVerification == nil {
			u.logMatches(log, "Webhook Verification", routeModuleComparisonBothNil)
			return nil
		}

		log.Info("Deleting Webhook Verification module")
		return client.Delete(ctx, edgeRouteItem(route))
	}

	// Some WebhookVerification providers don't require a secret,
	// so default to an empty string.
	var webhookSecret = ""

	if webhookVerification.SecretRef != nil {
		s, err := u.getSecret(ctx, *webhookVerification.SecretRef)
		if err != nil {
			return err
		}
		webhookSecret = *s
	}

	module := ngrok.EndpointWebhookValidation{
		Provider: webhookVerification.Provider,
		Secret:   webhookSecret,
	}

	if reflect.DeepEqual(&module, route.WebhookVerification) {
		u.logMatches(log, "Webhook Verification", routeModuleComparisonDeepEqual)
		return nil
	}

	log.Info("Updating Webhook Verification module")
	_, err := client.Replace(ctx, &ngrok.EdgeRouteWebhookVerificationReplace{
		EdgeID: route.EdgeID,
		ID:     route.ID,
		Module: module,
	})
	return err
}

func (u *edgeRouteModuleUpdater) getSecret(ctx context.Context, secretRef ingressv1alpha1.SecretKeyRef) (*string, error) {
	secret, err := u.secretResolver.getSecret(ctx,
		u.edge.Namespace,
		secretRef.Name,
		secretRef.Key,
	)
	return &secret, err
}

type OAuthProvider interface {
	ClientSecretKeyRef() *ingressv1alpha1.SecretKeyRef
	ToNgrok(*string) *ngrok.EndpointOAuth
}

// isMigratingAuthProviders returns true if the auth provider is changing
// It takes in the current ngrok.HTTPSEdgeRoute and the desired ingressv1alpha1.HTTPSEdgeRouteSpec
// if the current and desired have different auth types (OAuth, OIDC, SAML), it returns true
func isMigratingAuthProviders(current *ngrok.HTTPSEdgeRoute, desired *ingressv1alpha1.HTTPSEdgeRouteSpec) bool {
	modifiedAuthTypes := 0
	if (current.OAuth == nil && desired.OAuth != nil) || (current.OAuth != nil && desired.OAuth == nil) {
		modifiedAuthTypes += 1
	}
	if (current.OIDC == nil && desired.OIDC != nil) || (current.OIDC != nil && desired.OIDC == nil) {
		modifiedAuthTypes += 1
	}
	if (current.SAML == nil && desired.SAML != nil) || (current.SAML != nil && desired.SAML == nil) {
		modifiedAuthTypes += 1
	}

	// Each check above tells if that auth type is being added or removed in some way.
	// If it happens 0 times, no modifications happened. If it happens only once, then its
	// just being added or removed, so there is no chance of conflict. But if multiple are triggered
	// then it must be moving from 1 type to another, so we can return true.
	return modifiedAuthTypes > 1
}

// takeOfflineWithoutAuth takes an ngrok.HTTPSEdgeRoute and will remove the backed first.
// It will save the route without the backend so its offline. Then for each of the auth types (OAuth, OIDC, SAML)
// it will try to remove them if nil. If removed, it will set the route to nil for that auth type.
func (r *HTTPSEdgeReconciler) takeOfflineWithoutAuth(ctx context.Context, route *ngrok.HTTPSEdgeRoute) error {
	log := ctrl.LoggerFrom(ctx)

	routeUpdate := &ngrok.HTTPSEdgeRouteUpdate{
		EdgeID:    route.EdgeID,
		ID:        route.ID,
		Match:     route.Match,
		MatchType: route.MatchType,
		Backend:   nil,
	}
	routeClientSet := r.NgrokClientset.EdgeModules().HTTPS().Routes()

	log.V(1).Info("Setting route backend to nil to take offline")
	route, err := r.NgrokClientset.HTTPSEdgeRoutes().Update(ctx, routeUpdate)
	if err != nil {
		return err
	}
	log.V(1).Info("Successfully set route backend to nil")

	if route.OAuth != nil {
		log.V(1).Info("Removing OAuth from route")
		if err := routeClientSet.OAuth().Delete(ctx, edgeRouteItem(route)); err != nil {
			return err
		}
		route.OAuth = nil
		log.V(1).Info("Successfully removed OAuth from route")
	}

	if route.OIDC != nil {
		log.V(1).Info("Removing OIDC from route")
		if err := routeClientSet.OIDC().Delete(ctx, edgeRouteItem(route)); err != nil {
			return err
		}
		route.OIDC = nil
		log.V(1).Info("Successfully removed OIDC from route")
	}

	if route.SAML != nil {
		log.V(1).Info("Removing SAML from route")
		if err := routeClientSet.SAML().Delete(ctx, edgeRouteItem(route)); err != nil {
			return err
		}
		route.SAML = nil
		log.V(1).Info("Successfully removed SAML from route")
	}

	return nil
}