terraform/command/plan.go

package command

import (
	"fmt"
	"strings"

	"github.com/hashicorp/terraform/backend"
	"github.com/hashicorp/terraform/command/arguments"
	"github.com/hashicorp/terraform/command/views"
	"github.com/hashicorp/terraform/tfdiags"
)

// PlanCommand is a Command implementation that compares a Terraform
// configuration to an actual infrastructure and shows the differences.
type PlanCommand struct {
	Meta
}

func (c *PlanCommand) Run(rawArgs []string) int {
	// Parse and apply global view arguments
	common, rawArgs := arguments.ParseView(rawArgs)
	c.View.Configure(common)

	// Propagate -no-color for the remote backend's legacy use of Ui. This
	// should be removed when the remote backend is migrated to views.
	c.Meta.color = !common.NoColor
	c.Meta.Color = c.Meta.color

	// Parse and validate flags
	args, diags := arguments.ParsePlan(rawArgs)

	// Instantiate the view, even if there are flag errors, so that we render
	// diagnostics according to the desired view
	view := views.NewPlan(args.ViewType, c.View)

	if diags.HasErrors() {
		view.Diagnostics(diags)
		view.HelpPrompt()
		return 1
	}

	// Check for user-supplied plugin path
	var err error
	if c.pluginPath, err = c.loadPluginPath(); err != nil {
		diags = diags.Append(err)
		view.Diagnostics(diags)
		return 1
	}

	// FIXME: the -input flag value is needed to initialize the backend and the
	// operation, but there is no clear path to pass this value down, so we
	// continue to mutate the Meta object state for now.
	c.Meta.input = args.InputEnabled

	// FIXME: the -parallelism flag is used to control the concurrency of
	// Terraform operations. At the moment, this value is used both to
	// initialize the backend via the ContextOpts field inside CLIOpts, and to
	// set a largely unused field on the Operation request. Again, there is no
	// clear path to pass this value down, so we continue to mutate the Meta
	// object state for now.
	c.Meta.parallelism = args.Operation.Parallelism

	diags = diags.Append(c.providerDevOverrideRuntimeWarnings())

	// Prepare the backend with the backend-specific arguments
	be, beDiags := c.PrepareBackend(args.State)
	diags = diags.Append(beDiags)
	if diags.HasErrors() {
		view.Diagnostics(diags)
		return 1
	}

	// Build the operation request
	opReq, opDiags := c.OperationRequest(be, view, args.Operation, args.OutPath)
	diags = diags.Append(opDiags)
	if diags.HasErrors() {
		view.Diagnostics(diags)
		return 1
	}

	// Collect variable value and add them to the operation request
	diags = diags.Append(c.GatherVariables(opReq, args.Vars))
	if diags.HasErrors() {
		view.Diagnostics(diags)
		return 1
	}

	// Before we delegate to the backend, we'll print any warning diagnostics
	// we've accumulated here, since the backend will start fresh with its own
	// diagnostics.
	view.Diagnostics(diags)
	diags = nil

	// Perform the operation
	op, err := c.RunOperation(be, opReq)
	if err != nil {
		diags = diags.Append(err)
		view.Diagnostics(diags)
		return 1
	}

	if op.Result != backend.OperationSuccess {
		return op.Result.ExitStatus()
	}
	if args.DetailedExitCode && !op.PlanEmpty {
		return 2
	}

	return op.Result.ExitStatus()
}

func (c *PlanCommand) PrepareBackend(args *arguments.State) (backend.Enhanced, tfdiags.Diagnostics) {
	// FIXME: we need to apply the state arguments to the meta object here
	// because they are later used when initializing the backend. Carving a
	// path to pass these arguments to the functions that need them is
	// difficult but would make their use easier to understand.
	c.Meta.applyStateArguments(args)

	backendConfig, diags := c.loadBackendConfig(".")
	if diags.HasErrors() {
		return nil, diags
	}

	// Load the backend
	be, beDiags := c.Backend(&BackendOpts{
		Config: backendConfig,
	})
	diags = diags.Append(beDiags)
	if beDiags.HasErrors() {
		return nil, diags
	}

	return be, diags
}

func (c *PlanCommand) OperationRequest(
	be backend.Enhanced,
	view views.Plan,
	args *arguments.Operation,
	planOutPath string,
) (*backend.Operation, tfdiags.Diagnostics) {
	var diags tfdiags.Diagnostics

	// Build the operation
	opReq := c.Operation(be)
	opReq.ConfigDir = "."
	opReq.PlanMode = args.PlanMode
	opReq.Hooks = view.Hooks()
	opReq.PlanRefresh = args.Refresh
	opReq.PlanOutPath = planOutPath
	opReq.Targets = args.Targets
	opReq.ForceReplace = args.ForceReplace
	opReq.Type = backend.OperationTypePlan
	opReq.View = view.Operation()

	var err error
	opReq.ConfigLoader, err = c.initConfigLoader()
	if err != nil {
		diags = diags.Append(fmt.Errorf("Failed to initialize config loader: %s", err))
		return nil, diags
	}

	return opReq, diags
}

func (c *PlanCommand) GatherVariables(opReq *backend.Operation, args *arguments.Vars) tfdiags.Diagnostics {
	var diags tfdiags.Diagnostics

	// FIXME the arguments package currently trivially gathers variable related
	// arguments in a heterogenous slice, in order to minimize the number of
	// code paths gathering variables during the transition to this structure.
	// Once all commands that gather variables have been converted to this
	// structure, we could move the variable gathering code to the arguments
	// package directly, removing this shim layer.

	varArgs := args.All()
	items := make([]rawFlag, len(varArgs))
	for i := range varArgs {
		items[i].Name = varArgs[i].Name
		items[i].Value = varArgs[i].Value
	}
	c.Meta.variableArgs = rawFlags{items: &items}
	opReq.Variables, diags = c.collectVariableValues()

	return diags
}

func (c *PlanCommand) Help() string {
	helpText := `
Usage: terraform [global options] plan [options]

  Generates a speculative execution plan, showing what actions Terraform
  would take to apply the current configuration. This command will not
  actually perform the planned actions.

  You can optionally save the plan to a file, which you can then pass to
  the "apply" command to perform exactly the actions described in the plan.

Plan Customization Options:

  The following options customize how Terraform will produce its plan. You
  can also use these options when you run "terraform apply" without passing
  it a saved plan, in order to plan and apply in a single command.

  -destroy            If set, a plan will be generated to destroy all resources
                      managed by the given configuration and state.

  -refresh=false      Skip checking for changes to remote objects while
                      creating the plan. This can potentially make planning
                      faster, but at the expense of possibly planning against
                      a stale record of the remote system state.

  -replace=resource   Force replacement of a particular resource instance using
                      its resource address. If the plan would've normally
                      produced an update or no-op action for this instance,
                      Terraform will plan to replace it instead.

  -target=resource    Limit the planning operation to only the given module,
                      resource, or resource instance and all of its
                      dependencies. You can use this option multiple times to
                      include more than one object. This is for exceptional
                      use only.

  -var 'foo=bar'      Set a variable in the Terraform configuration. This
                      flag can be set multiple times.

  -var-file=foo       Set variables in the Terraform configuration from
                      a file. If "terraform.tfvars" or any ".auto.tfvars"
                      files are present, they will be automatically loaded.

Other Options:

  -compact-warnings   If Terraform produces any warnings that are not
                      accompanied by errors, show them in a more compact form
                      that includes only the summary messages.

  -detailed-exitcode  Return detailed exit codes when the command exits. This
                      will change the meaning of exit codes to:
                      0 - Succeeded, diff is empty (no changes)
                      1 - Errored
                      2 - Succeeded, there is a diff

  -input=true         Ask for input for variables if not directly set.

  -lock=false         Don't hold a state lock during the operation. This is
                      dangerous if others might concurrently run commands
                      against the same workspace.

  -lock-timeout=0s    Duration to retry a state lock.

  -no-color           If specified, output won't contain any color.

  -out=path           Write a plan file to the given path. This can be used as
                      input to the "apply" command.

  -parallelism=n      Limit the number of concurrent operations. Defaults to 10.

  -state=statefile    A legacy option used for the local backend only. See the
                      local backend's documentation for more information.
`
	return strings.TrimSpace(helpText)
}

func (c *PlanCommand) Synopsis() string {
	return "Show changes required by the current configuration"
}