De-duplicate images in final discovery output

This change modifies DiscoveredImage to store the name/type of one
or more containers which use the image, and it also adds a new
appendToManifest function, which acts similarly to the built-in
append except duplicate entries are merged into the existing
DiscoveredImage instance in the list.

Signed-off-by: Caleb Xu <caxu@redhat.com>

Author: caxu-rh

discovery/discovery.go

@@ -5,13 +5,42 @@ type Manifest struct {
 	DiscoveredImages []DiscoveredImage
 }
 
-// DiscoveredImage is a container image that' been discovered based on the
-// workload watchers.
+// DiscoveredImage is a container image which was discovered in one or more workloads.
 type DiscoveredImage struct {
-	// PodName is the pod.metadata.name value where the image was discovered.
-	PodName string
-	// ContainerName represents the container in the pod that had the image.
-	ContainerName string
 	// Image is a fully qualified container image name and tag or digest.
 	Image string
+
+	// Containers is a list of DiscoveredContainer objects which are using
+	// the discovered image.
+	Containers []DiscoveredContainer
+}
+
+// DiscoveredContainer is a container which was observed during the discovery process.
+type DiscoveredContainer struct {
+	// Name is the name of a container in a pod.
+	Name string
+
+	// Type is the ContainerType of the container in its pod.
+	Type ContainerType
+
+	// Pod is the DiscoveredPod which this container is a part of.
+	Pod DiscoveredPod
+}
+
+// ContainerType is the type of a container in a pod.
+type ContainerType = string
+
+const (
+	ContainerTypeStandard  ContainerType = "Container"
+	ContainerTypeInit      ContainerType = "InitContainer"
+	ContainerTypeEphemeral ContainerType = "EphemeralContainer"
+)
+
+// DiscoveredPod is a pod that contains a discovered image.
+type DiscoveredPod struct {
+	// Name is the pod.metadata.name value where the image was discovered.
+	Name string
+
+	// Namespace is the pod.metadata.namespace value of the pod.
+	Namespace string
 }

internal/discover/processor.go

@@ -6,6 +6,7 @@ import (
 	"fmt"
 	"io"
 	"log/slog"
+	"slices"
 
 	corev1 "k8s.io/api/core/v1"
 
@@ -28,7 +29,7 @@ func NewManifestJSONProcessorFn(out io.Writer) ProcessingFunction {
 					continueRunning = false
 					break
 				}
-				m.DiscoveredImages = append(m.DiscoveredImages, processContainers(p, logger)...)
+				m = appendToManifest(m, processContainers(p, logger)...)
 			case <-ctx.Done():
 				logger.Debug("processorFn completing because the context completed")
 				continueRunning = false
@@ -64,9 +65,17 @@ func processContainers(
 		found = append(
 			found,
 			discovery.DiscoveredImage{
-				PodName:       p.Name,
-				ContainerName: c.Name,
-				Image:         c.Image,
+				Image: c.Image,
+				Containers: []discovery.DiscoveredContainer{
+					{
+						Name: c.Name,
+						Type: discovery.ContainerTypeStandard,
+						Pod: discovery.DiscoveredPod{
+							Name:      p.Name,
+							Namespace: p.Namespace,
+						},
+					},
+				},
 			},
 		)
 	}
@@ -76,9 +85,17 @@ func processContainers(
 		found = append(
 			found,
 			discovery.DiscoveredImage{
-				PodName:       p.Name,
-				ContainerName: c.Name,
-				Image:         c.Image,
+				Image: c.Image,
+				Containers: []discovery.DiscoveredContainer{
+					{
+						Name: c.Name,
+						Type: discovery.ContainerTypeInit,
+						Pod: discovery.DiscoveredPod{
+							Name:      p.Name,
+							Namespace: p.Namespace,
+						},
+					},
+				},
 			},
 		)
 	}
@@ -88,12 +105,45 @@ func processContainers(
 		found = append(
 			found,
 			discovery.DiscoveredImage{
-				PodName:       p.Name,
-				ContainerName: c.Name,
-				Image:         c.Image,
+				Image: c.Image,
+				Containers: []discovery.DiscoveredContainer{
+					{
+						Name: c.Name,
+						Type: discovery.ContainerTypeEphemeral,
+						Pod: discovery.DiscoveredPod{
+							Name:      p.Name,
+							Namespace: p.Namespace,
+						},
+					},
+				},
 			},
 		)
 	}
 
 	return found
 }
+
+func appendToManifest(m discovery.Manifest, images ...discovery.DiscoveredImage) discovery.Manifest {
+	for _, image := range images {
+		idx := slices.IndexFunc(m.DiscoveredImages, func(i discovery.DiscoveredImage) bool {
+			return i.Image == image.Image
+		})
+
+		if idx == -1 {
+			m.DiscoveredImages = append(m.DiscoveredImages, image)
+			continue
+		}
+
+		for _, container := range image.Containers {
+			if !slices.Contains(m.DiscoveredImages[idx].Containers, container) {
+				m.DiscoveredImages[idx].Containers = append(m.DiscoveredImages[idx].Containers, container)
+			}
+		}
+	}
+
+	return m
+}
+
+func imagesEqual(i1, i2 discovery.DiscoveredImage) bool {
+	return i1.Image == i2.Image && slices.Equal(i1.Containers, i2.Containers)
+}