From 6e306b0e8640d28024178b711eaa17d9ed7e40d0 Mon Sep 17 00:00:00 2001
From: Arik Hadas <ahadas@redhat.com>
Date: Wed, 3 Apr 2024 13:25:36 +0300
Subject: [PATCH] Sanitize datastore identifiers from ESXi providers

when we get datastores from the ESXi SDK, their identifier may be in the
form of '10.11.12.13:/vol/virtv2v/function' which leads to invalid
endpoints in the inventory so we sanitize such identifiers

Signed-off-by: Arik Hadas <ahadas@redhat.com>
---
 .../provider/container/vsphere/BUILD.bazel      |  1 +
 .../provider/container/vsphere/collector.go     | 12 ++++++++++--
 .../provider/container/vsphere/model.go         |  9 ++++++---
 .../provider/container/vsphere/utils.go         | 17 +++++++++++++++++
 4 files changed, 34 insertions(+), 5 deletions(-)
 create mode 100644 pkg/controller/provider/container/vsphere/utils.go

diff --git a/pkg/controller/provider/container/vsphere/BUILD.bazel b/pkg/controller/provider/container/vsphere/BUILD.bazel
index 98552dea5..e6de53c8b 100644
--- a/pkg/controller/provider/container/vsphere/BUILD.bazel
+++ b/pkg/controller/provider/container/vsphere/BUILD.bazel
@@ -6,6 +6,7 @@ go_library(
         "collector.go",
         "doc.go",
         "model.go",
+        "utils.go",
         "watch.go",
     ],
     importpath = "github.com/konveyor/forklift-controller/pkg/controller/provider/container/vsphere",
diff --git a/pkg/controller/provider/container/vsphere/collector.go b/pkg/controller/provider/container/vsphere/collector.go
index f7077d7c7..899e559ea 100644
--- a/pkg/controller/provider/container/vsphere/collector.go
+++ b/pkg/controller/provider/container/vsphere/collector.go
@@ -838,10 +838,17 @@ func (r *Collector) selectAdapter(u types.ObjectUpdate) (Adapter, bool) {
 			},
 		}
 	case Datastore:
+		// when we get datastores from the ESXi SDK, their identifier may be in the
+		// form of '10.11.12.13:/vol/virtv2v/function' which leads to invalid
+		// endpoints in the inventory so we sanitize such identifiers
+		datastoreId, changed := sanitize(u.Obj.Value)
+		if changed {
+			r.log.Info("sanitized datastore ID", "reported", u.Obj.Value, "sanitized", datastoreId)
+		}
 		adapter = &DatastoreAdapter{
 			model: model.Datastore{
 				Base: model.Base{
-					ID: u.Obj.Value,
+					ID: datastoreId,
 				},
 			},
 		}
@@ -932,9 +939,10 @@ func (r Collector) applyLeave(tx *libmodel.Tx, u types.ObjectUpdate) error {
 			},
 		}
 	case Datastore:
+		datastoreId, _ := sanitize(u.Obj.Value)
 		deleted = &model.Datastore{
 			Base: model.Base{
-				ID: u.Obj.Value,
+				ID: datastoreId,
 			},
 		}
 	case VirtualMachine:
diff --git a/pkg/controller/provider/container/vsphere/model.go b/pkg/controller/provider/container/vsphere/model.go
index bcead2414..45ed63eac 100644
--- a/pkg/controller/provider/container/vsphere/model.go
+++ b/pkg/controller/provider/container/vsphere/model.go
@@ -688,18 +688,20 @@ func (v *VmAdapter) updateDisks(devArray *types.ArrayOfVirtualDevice) {
 			switch disk.Backing.(type) {
 			case *types.VirtualDiskFlatVer1BackingInfo:
 				backing := disk.Backing.(*types.VirtualDiskFlatVer1BackingInfo)
+				datastoreId, _ := sanitize(backing.Datastore.Value)
 				md := model.Disk{
 					Key:      disk.Key,
 					File:     backing.FileName,
 					Capacity: disk.CapacityInBytes,
 					Datastore: model.Ref{
 						Kind: model.DsKind,
-						ID:   backing.Datastore.Value,
+						ID:   datastoreId,
 					},
 				}
 				disks = append(disks, md)
 			case *types.VirtualDiskFlatVer2BackingInfo:
 				backing := disk.Backing.(*types.VirtualDiskFlatVer2BackingInfo)
+				datastoreId, _ := sanitize(backing.Datastore.Value)
 				md := model.Disk{
 					Key:      disk.Key,
 					File:     backing.FileName,
@@ -707,12 +709,13 @@ func (v *VmAdapter) updateDisks(devArray *types.ArrayOfVirtualDevice) {
 					Shared:   backing.Sharing != "sharingNone",
 					Datastore: model.Ref{
 						Kind: model.DsKind,
-						ID:   backing.Datastore.Value,
+						ID:   datastoreId,
 					},
 				}
 				disks = append(disks, md)
 			case *types.VirtualDiskRawDiskMappingVer1BackingInfo:
 				backing := disk.Backing.(*types.VirtualDiskRawDiskMappingVer1BackingInfo)
+				datastoreId, _ := sanitize(backing.Datastore.Value)
 				md := model.Disk{
 					Key:      disk.Key,
 					File:     backing.FileName,
@@ -720,7 +723,7 @@ func (v *VmAdapter) updateDisks(devArray *types.ArrayOfVirtualDevice) {
 					Shared:   backing.Sharing != "sharingNone",
 					Datastore: model.Ref{
 						Kind: model.DsKind,
-						ID:   backing.Datastore.Value,
+						ID:   datastoreId,
 					},
 					RDM: true,
 				}
diff --git a/pkg/controller/provider/container/vsphere/utils.go b/pkg/controller/provider/container/vsphere/utils.go
new file mode 100644
index 000000000..4e1bf53e6
--- /dev/null
+++ b/pkg/controller/provider/container/vsphere/utils.go
@@ -0,0 +1,17 @@
+package vsphere
+
+import (
+	"crypto/md5"
+	"encoding/hex"
+	"net/url"
+)
+
+func sanitize(datastoreId string) (sanitizedId string, changed bool) {
+	sanitizedId = url.PathEscape(datastoreId)
+	if sanitizedId != datastoreId {
+		sum := md5.Sum([]byte(datastoreId))
+		sanitizedId = hex.EncodeToString(sum[:])
+		changed = true
+	}
+	return
+}