Pin Alpine image digest and add SHA256 verification for TF provider

Dockerfile

@@ -1,4 +1,4 @@
-FROM alpine:3.22 as builder
+FROM alpine:3.22@sha256:55ae5d250caebc548793f321534bc6a8ef1d116f334f18f4ada1b2daad3251b2 as builder
 
 RUN ["apk", "add", "jq"]
 RUN ["apk", "add", "bash"]
@@ -13,7 +13,7 @@ ARG ARCH
 RUN /build/docker/setup.sh
 
 # Start from a fresh base image, to remove any build artifacts and scripts.
-FROM alpine:3.22
+FROM alpine:3.22@sha256:55ae5d250caebc548793f321534bc6a8ef1d116f334f18f4ada1b2daad3251b2
 
 ENV DATABRICKS_TF_EXEC_PATH "/app/bin/terraform"
 ENV DATABRICKS_TF_CLI_CONFIG_FILE "/app/config/config.tfrc"

bundle/deploy/terraform/pkg.go

@@ -77,11 +77,12 @@ type Checksum struct {
 }
 
 type TerraformMetadata struct {
-	Version         string   `json:"version"`
-	Checksum        Checksum `json:"checksum"`
-	ProviderHost    string   `json:"providerHost"`
-	ProviderSource  string   `json:"providerSource"`
-	ProviderVersion string   `json:"providerVersion"`
+	Version          string   `json:"version"`
+	Checksum         Checksum `json:"checksum"`
+	ProviderHost     string   `json:"providerHost"`
+	ProviderSource   string   `json:"providerSource"`
+	ProviderVersion  string   `json:"providerVersion"`
+	ProviderChecksum Checksum `json:"providerChecksum"`
 }
 
 func NewTerraformMetadata(ctx context.Context) (*TerraformMetadata, error) {
@@ -98,6 +99,10 @@ func NewTerraformMetadata(ctx context.Context) (*TerraformMetadata, error) {
 		ProviderHost:    schema.ProviderHost,
 		ProviderSource:  schema.ProviderSource,
 		ProviderVersion: schema.ProviderVersion,
+		ProviderChecksum: Checksum{
+			LinuxAmd64: schema.ProviderChecksumLinuxAmd64,
+			LinuxArm64: schema.ProviderChecksumLinuxArm64,
+		},
 	}, nil
 }
 

bundle/deploy/terraform/pkg_test.go

@@ -54,6 +54,17 @@ func TestTerraformArchiveChecksums(t *testing.T) {
 	downloadAndChecksum(t, armUrl, tv.ChecksumLinuxArm64)
 }
 
+func TestTerraformProviderArchiveChecksums(t *testing.T) {
+	metadata, err := NewTerraformMetadata(t.Context())
+	require.NoError(t, err)
+
+	amdUrl := fmt.Sprintf("https://github.com/databricks/terraform-provider-databricks/releases/download/v%s/terraform-provider-databricks_%s_linux_amd64.zip", metadata.ProviderVersion, metadata.ProviderVersion)
+	armUrl := fmt.Sprintf("https://github.com/databricks/terraform-provider-databricks/releases/download/v%s/terraform-provider-databricks_%s_linux_arm64.zip", metadata.ProviderVersion, metadata.ProviderVersion)
+
+	downloadAndChecksum(t, amdUrl, metadata.ProviderChecksum.LinuxAmd64)
+	downloadAndChecksum(t, armUrl, metadata.ProviderChecksum.LinuxArm64)
+}
+
 func TestGetTerraformVersionDefault(t *testing.T) {
 	// Verify that the default version is used
 	tv, isDefault, err := GetTerraformVersion(t.Context())

bundle/internal/tf/codegen/README.md

@@ -7,6 +7,7 @@ The entry point for this tool is `.`.
 It uses `./tmp` a temporary data directory and `../schema` as output directory.
 
 It automatically installs the Terraform binary as well as the Databricks Terraform provider.
+It also fetches SHA256 checksums for the provider archive from GitHub releases.
 
 Run with:
 

bundle/internal/tf/codegen/generator/generator.go

@@ -35,8 +35,10 @@ func (c *collection) Generate(path string) error {
 }
 
 type root struct {
-	OutputFile      string
-	ProviderVersion string
+	OutputFile                 string
+	ProviderVersion            string
+	ProviderChecksumLinuxAmd64 string
+	ProviderChecksumLinuxArm64 string
 }
 
 func (r *root) Generate(path string) error {
@@ -51,7 +53,7 @@ func (r *root) Generate(path string) error {
 	return tmpl.Execute(f, r)
 }
 
-func Run(ctx context.Context, schema *tfjson.ProviderSchema, path string) error {
+func Run(ctx context.Context, schema *tfjson.ProviderSchema, checksums *schemapkg.ProviderChecksums, path string) error {
 	// Generate types for resources
 	var resources []*namedBlock
 	for _, k := range sortKeys(schema.ResourceSchemas) {
@@ -147,8 +149,10 @@ func Run(ctx context.Context, schema *tfjson.ProviderSchema, path string) error
 	// Generate root.go
 	{
 		r := &root{
-			OutputFile:      "root.go",
-			ProviderVersion: schemapkg.ProviderVersion,
+			OutputFile:                 "root.go",
+			ProviderVersion:            schemapkg.ProviderVersion,
+			ProviderChecksumLinuxAmd64: checksums.LinuxAmd64,
+			ProviderChecksumLinuxArm64: checksums.LinuxArm64,
 		}
 		err := r.Generate(path)
 		if err != nil {

bundle/internal/tf/codegen/main.go

@@ -11,12 +11,20 @@ import (
 func main() {
 	ctx := context.Background()
 
-	schema, err := schema.Load(ctx)
+	s, err := schema.Load(ctx)
 	if err != nil {
 		log.Fatal(err)
 	}
 
-	err = generator.Run(ctx, schema, "../schema")
+	log.Printf("fetching provider checksums for v%s", schema.ProviderVersion)
+	checksums, err := schema.FetchProviderChecksums(schema.ProviderVersion)
+	if err != nil {
+		log.Fatal(err)
+	}
+	log.Printf("  linux_amd64: %s", checksums.LinuxAmd64)
+	log.Printf("  linux_arm64: %s", checksums.LinuxArm64)
+
+	err = generator.Run(ctx, s, checksums, "../schema")
 	if err != nil {
 		log.Fatal(err)
 	}

bundle/internal/tf/codegen/schema/checksum.go

@@ -0,0 +1,67 @@
+package schema
+
+import (
+	"bufio"
+	"fmt"
+	"net/http"
+	"strings"
+)
+
+// ProviderChecksums holds the SHA256 checksums for the Databricks Terraform
+// provider archive for supported Linux architectures.
+type ProviderChecksums struct {
+	LinuxAmd64 string
+	LinuxArm64 string
+}
+
+// FetchProviderChecksums downloads the SHA256SUMS file from the GitHub release
+// for the given provider version and extracts checksums for the linux_amd64 and
+// linux_arm64 archives.
+// https://github.com/databricks/terraform-provider-databricks/releases
+func FetchProviderChecksums(version string) (*ProviderChecksums, error) {
+	url := fmt.Sprintf(
+		"https://github.com/databricks/terraform-provider-databricks/releases/download/v%s/terraform-provider-databricks_%s_SHA256SUMS",
+		version, version,
+	)
+
+	resp, err := http.Get(url)
+	if err != nil {
+		return nil, fmt.Errorf("downloading SHA256SUMS for provider v%s: %w", version, err)
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode != http.StatusOK {
+		return nil, fmt.Errorf("downloading SHA256SUMS for provider v%s: HTTP %s", version, resp.Status)
+	}
+
+	checksums := &ProviderChecksums{}
+	amd64Suffix := fmt.Sprintf("terraform-provider-databricks_%s_linux_amd64.zip", version)
+	arm64Suffix := fmt.Sprintf("terraform-provider-databricks_%s_linux_arm64.zip", version)
+
+	scanner := bufio.NewScanner(resp.Body)
+	for scanner.Scan() {
+		line := scanner.Text()
+		parts := strings.Fields(line)
+		if len(parts) != 2 {
+			continue
+		}
+		switch parts[1] {
+		case amd64Suffix:
+			checksums.LinuxAmd64 = parts[0]
+		case arm64Suffix:
+			checksums.LinuxArm64 = parts[0]
+		}
+	}
+	if err := scanner.Err(); err != nil {
+		return nil, fmt.Errorf("reading SHA256SUMS for provider v%s: %w", version, err)
+	}
+
+	if checksums.LinuxAmd64 == "" {
+		return nil, fmt.Errorf("checksum not found for %s in SHA256SUMS", amd64Suffix)
+	}
+	if checksums.LinuxArm64 == "" {
+		return nil, fmt.Errorf("checksum not found for %s in SHA256SUMS", arm64Suffix)
+	}
+
+	return checksums, nil
+}

bundle/internal/tf/codegen/templates/root.go.tmpl

@@ -22,6 +22,8 @@ type Root struct {
 const ProviderHost = "registry.terraform.io"
 const ProviderSource = "databricks/databricks"
 const ProviderVersion = "{{ .ProviderVersion }}"
+const ProviderChecksumLinuxAmd64 = "{{ .ProviderChecksumLinuxAmd64 }}"
+const ProviderChecksumLinuxArm64 = "{{ .ProviderChecksumLinuxArm64 }}"
 
 func NewRoot() *Root {
 	return &Root{

docker/setup.sh

@@ -30,3 +30,11 @@ mv zip/terraform/terraform /app/bin/terraform
 TF_PROVIDER_NAME=terraform-provider-databricks_${DATABRICKS_TF_PROVIDER_VERSION}_linux_${ARCH}.zip
 mkdir -p /app/providers/registry.terraform.io/databricks/databricks
 wget https://github.com/databricks/terraform-provider-databricks/releases/download/v${DATABRICKS_TF_PROVIDER_VERSION}/${TF_PROVIDER_NAME} -O /app/providers/registry.terraform.io/databricks/databricks/${TF_PROVIDER_NAME}
+
+# Verify the provider checksum.
+EXPECTED_PROVIDER_CHECKSUM="$(/app/databricks bundle debug terraform --output json | jq -r .terraform.providerChecksum.linux_$ARCH)"
+COMPUTED_PROVIDER_CHECKSUM=$(sha256sum /app/providers/registry.terraform.io/databricks/databricks/${TF_PROVIDER_NAME} | awk '{ print $1 }')
+if [ "$COMPUTED_PROVIDER_CHECKSUM" != "$EXPECTED_PROVIDER_CHECKSUM" ]; then
+    echo "Checksum mismatch for Terraform provider. Version: $DATABRICKS_TF_PROVIDER_VERSION, Arch: $ARCH, Expected checksum: $EXPECTED_PROVIDER_CHECKSUM, Computed checksum: $COMPUTED_PROVIDER_CHECKSUM."
+    exit 1
+fi