---

name: update-importer description: Update importer code when provider resource schemas change tags: [importer, schema, terraform, maintenance]

Update Importer Skill

When resource schemas in the provider are modified, the corresponding importer code must be updated to match. This skill guides you through the process.

When to Use This Skill

Use this skill when:

• A provider resource schema is modified (fields added, removed, or renamed)
• The API response structure changes
• You're reviewing changes that touch resource definitions in /provider/resource_*.go

Steps to Update Importer

1. Identify Changed Resource

Determine which resource(s) were modified. Common resources:

• resource_bulk_sync.go → /importer/bulk_syncs.go
• resource_sync.go → /importer/syncs.go
• resource_connection_*.go → /importer/connections.go
• resource_model.go → /importer/models.go

2. Check for Schema Validation

Each importer should use the schema validation pattern. Verify the importer has:

Copy// At the start of GenerateTerraformFiles
validator, err := NewSchemaValidator(ctx, provider.New{Resource}ResourceForSchemaIntrospection())
if err != nil {
    return fmt.Errorf("failed to create schema validator: %w", err)
}

If not present, add it following the pattern in /importer/bulk_syncs.go.

3. Update Field Mapping

In the buildFieldMapping() function, add any new fields:

Copy// Required fields
mapping := map[string]interface{}{
    "field_name": value,
    // ... existing fields
}

// Optional fields
if response.NewField != nil {
    mapping["new_field"] = response.NewField
}

Important: The mapping keys must exactly match the tfsdk tag names in the provider's resource data struct.

4. Update HCL Generation

In the GenerateTerraformFiles() function, add code to generate HCL for new fields:

Copy// For simple fields
if response.NewField != nil {
    resourceBlock.Body().SetAttributeValue("new_field",
        cty.StringVal(pointer.GetString(response.NewField)))
}

// For nested objects
if response.NewNested != nil {
    resourceBlock.Body().SetAttributeValue("new_nested",
        typeConverter(response.NewNested))
}

// For jsonencoded strings
tokens := wrapJSONEncode(response.NewConfig)
resourceBlock.Body().SetAttributeRaw("new_config", tokens)

5. Handle Field Renames

When fields are renamed in the schema:

1. Update the schema validator's suggestion map in /importer/schema_validator.go:

Copysuggestions := map[string]string{
    "old_field_name": "new_field_name",
}

1. Update both the mapping and HCL generation to use the new field name

6. Add Introspection Function

If the provider resource doesn't have an introspection function, add one:

Copy// In /provider/resource_{name}.go
func New{Resource}ResourceForSchemaIntrospection() resource.Resource {
    return &{resource}Resource{}
}

7. Verify with Tests

Build and test:

Copy# Build importer
go build ./importer/...

# Run roundtrip tests
POLYTOMIC_API_KEY=$API_KEY POLYTOMIC_DEPLOYMENT_URL=$URL \
  TF_ACC=1 go test ./tests/... -v -timeout 120m

Common Patterns

Nested Objects (SingleNestedAttribute)

Copy// Build tokens manually for proper HCL structure
tokens := hclwrite.Tokens{
    &hclwrite.Token{Bytes: []byte("{\n")},
    &hclwrite.Token{Bytes: []byte("    field = ")},
    &hclwrite.Token{Bytes: []byte(fmt.Sprintf(`"%s"`, value))},
    &hclwrite.Token{Bytes: []byte("\n  }")},
}
resourceBlock.Body().SetAttributeRaw("nested_field", tokens)

Set/List of Objects

Copy// Convert to array of maps
objects := make([]map[string]interface{}, 0, len(items))
for _, item := range items {
    objects = append(objects, map[string]interface{}{
        "id":      item.Id,
        "enabled": item.Enabled,
    })
}
resourceBlock.Body().SetAttributeValue("items", typeConverter(objects))

Optional Fields

Always check if optional fields are present before adding to mapping or HCL:

Copyif response.OptionalField != nil {
    // Add to mapping and generate HCL
}

Schema Drift Detection

The schema validator will fail with helpful errors if fields don't match:

Copyschema validation failed for resource 'my-resource':
field 'old_field_name' not found in schema, did you mean 'new_field_name'?

This prevents drift by failing fast during import rather than at terraform plan time.

Checklist

• Identified which importer file needs updates
• Added/verified schema validator is present
• Updated buildFieldMapping() with new/changed fields
• Updated GenerateTerraformFiles() HCL generation
• Handled renamed fields in validator suggestions
• Added introspection function to provider resource (if needed)
• Built importer successfully
• Verified with roundtrip tests (or documented why not possible)

Notes

• The schema validator acts as the "source of truth" - provider schema changes are automatically caught
• Always prefer using typeConverter() for complex types rather than manual token construction
• Keep field names in sync between mapping keys and HCL generation
• Optional fields should gracefully handle nil values