opentofu/terraform/plan.go

265 lines
7.4 KiB
Go
Raw Normal View History

package terraform
import (
2014-06-25 20:22:42 -05:00
"bytes"
"encoding/gob"
"errors"
"fmt"
"io"
"log"
"sync"
terraform: ugly huge change to weave in new HCL2-oriented types Due to how deeply the configuration types go into Terraform Core, there isn't a great way to switch out to HCL2 gradually. As a consequence, this huge commit gets us from the old state to a _compilable_ new state, but does not yet attempt to fix any tests and has a number of known missing parts and bugs. We will continue to iterate on this in forthcoming commits, heading back towards passing tests and making Terraform fully-functional again. The three main goals here are: - Use the configuration models from the "configs" package instead of the older models in the "config" package, which is now deprecated and preserved only to help us write our migration tool. - Do expression inspection and evaluation using the functionality of the new "lang" package, instead of the Interpolator type and related functionality in the main "terraform" package. - Represent addresses of various objects using types in the addrs package, rather than hand-constructed strings. This is not critical to support the above, but was a big help during the implementation of these other points since it made it much more explicit what kind of address is expected in each context. Since our new packages are built to accommodate some future planned features that are not yet implemented (e.g. the "for_each" argument on resources, "count"/"for_each" on modules), and since there's still a fair amount of functionality still using old-style APIs, there is a moderate amount of shimming here to connect new assumptions with old, hopefully in a way that makes it easier to find and eliminate these shims later. I apologize in advance to the person who inevitably just found this huge commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
"github.com/hashicorp/hcl2/hcl"
"github.com/hashicorp/hcl2/hcl/hclsyntax"
"github.com/zclconf/go-cty/cty"
"github.com/hashicorp/terraform/addrs"
"github.com/hashicorp/terraform/configs"
"github.com/hashicorp/terraform/tfdiags"
"github.com/hashicorp/terraform/version"
)
func init() {
2014-07-02 18:17:26 -05:00
gob.Register(make([]interface{}, 0))
gob.Register(make([]map[string]interface{}, 0))
2014-08-05 00:08:55 -05:00
gob.Register(make(map[string]interface{}))
gob.Register(make(map[string]string))
}
// Plan represents a single Terraform execution plan, which contains
// all the information necessary to make an infrastructure change.
//
// A plan has to contain basically the entire state of the world
// necessary to make a change: the state, diff, config, backend config, etc.
// This is so that it can run alone without any other data.
type Plan struct {
// Diff describes the resource actions that must be taken when this
// plan is applied.
Diff *Diff
terraform: ugly huge change to weave in new HCL2-oriented types Due to how deeply the configuration types go into Terraform Core, there isn't a great way to switch out to HCL2 gradually. As a consequence, this huge commit gets us from the old state to a _compilable_ new state, but does not yet attempt to fix any tests and has a number of known missing parts and bugs. We will continue to iterate on this in forthcoming commits, heading back towards passing tests and making Terraform fully-functional again. The three main goals here are: - Use the configuration models from the "configs" package instead of the older models in the "config" package, which is now deprecated and preserved only to help us write our migration tool. - Do expression inspection and evaluation using the functionality of the new "lang" package, instead of the Interpolator type and related functionality in the main "terraform" package. - Represent addresses of various objects using types in the addrs package, rather than hand-constructed strings. This is not critical to support the above, but was a big help during the implementation of these other points since it made it much more explicit what kind of address is expected in each context. Since our new packages are built to accommodate some future planned features that are not yet implemented (e.g. the "for_each" argument on resources, "count"/"for_each" on modules), and since there's still a fair amount of functionality still using old-style APIs, there is a moderate amount of shimming here to connect new assumptions with old, hopefully in a way that makes it easier to find and eliminate these shims later. I apologize in advance to the person who inevitably just found this huge commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
// Config represents the entire configuration that was present when this
// plan was created.
terraform: ugly huge change to weave in new HCL2-oriented types Due to how deeply the configuration types go into Terraform Core, there isn't a great way to switch out to HCL2 gradually. As a consequence, this huge commit gets us from the old state to a _compilable_ new state, but does not yet attempt to fix any tests and has a number of known missing parts and bugs. We will continue to iterate on this in forthcoming commits, heading back towards passing tests and making Terraform fully-functional again. The three main goals here are: - Use the configuration models from the "configs" package instead of the older models in the "config" package, which is now deprecated and preserved only to help us write our migration tool. - Do expression inspection and evaluation using the functionality of the new "lang" package, instead of the Interpolator type and related functionality in the main "terraform" package. - Represent addresses of various objects using types in the addrs package, rather than hand-constructed strings. This is not critical to support the above, but was a big help during the implementation of these other points since it made it much more explicit what kind of address is expected in each context. Since our new packages are built to accommodate some future planned features that are not yet implemented (e.g. the "for_each" argument on resources, "count"/"for_each" on modules), and since there's still a fair amount of functionality still using old-style APIs, there is a moderate amount of shimming here to connect new assumptions with old, hopefully in a way that makes it easier to find and eliminate these shims later. I apologize in advance to the person who inevitably just found this huge commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
Config *configs.Config
// State is the Terraform state that was current when this plan was
// created.
//
// It is not allowed to apply a plan that has a stale state, since its
// diff could be outdated.
State *State
// Vars retains the variables that were set when creating the plan, so
// that the same variables can be applied during apply.
terraform: ugly huge change to weave in new HCL2-oriented types Due to how deeply the configuration types go into Terraform Core, there isn't a great way to switch out to HCL2 gradually. As a consequence, this huge commit gets us from the old state to a _compilable_ new state, but does not yet attempt to fix any tests and has a number of known missing parts and bugs. We will continue to iterate on this in forthcoming commits, heading back towards passing tests and making Terraform fully-functional again. The three main goals here are: - Use the configuration models from the "configs" package instead of the older models in the "config" package, which is now deprecated and preserved only to help us write our migration tool. - Do expression inspection and evaluation using the functionality of the new "lang" package, instead of the Interpolator type and related functionality in the main "terraform" package. - Represent addresses of various objects using types in the addrs package, rather than hand-constructed strings. This is not critical to support the above, but was a big help during the implementation of these other points since it made it much more explicit what kind of address is expected in each context. Since our new packages are built to accommodate some future planned features that are not yet implemented (e.g. the "for_each" argument on resources, "count"/"for_each" on modules), and since there's still a fair amount of functionality still using old-style APIs, there is a moderate amount of shimming here to connect new assumptions with old, hopefully in a way that makes it easier to find and eliminate these shims later. I apologize in advance to the person who inevitably just found this huge commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
Vars map[string]cty.Value
// Targets, if non-empty, contains a set of resource address strings that
// identify graph nodes that were selected as targets for plan.
//
// When targets are set, any graph node that is not directly targeted or
// indirectly targeted via dependencies is excluded from the graph.
Targets []string
// TerraformVersion is the version of Terraform that was used to create
// this plan.
//
// It is not allowed to apply a plan created with a different version of
// Terraform, since the other fields of this structure may be interpreted
// in different ways between versions.
TerraformVersion string
// ProviderSHA256s is a map giving the SHA256 hashes of the exact binaries
// used as plugins for each provider during plan.
//
// These must match between plan and apply to ensure that the diff is
// correctly interpreted, since different provider versions may have
// different attributes or attribute value constraints.
ProviderSHA256s map[string][]byte
// Backend is the backend that this plan should use and store data with.
Backend *BackendState
// Destroy indicates that this plan was created for a full destroy operation
Destroy bool
once sync.Once
}
2014-07-03 13:32:44 -05:00
// Context returns a Context with the data encapsulated in this plan.
//
// The following fields in opts are overridden by the plan: Config,
// Diff, Variables.
//
// If State is not provided, it is set from the plan. If it _is_ provided,
// it must be Equal to the state stored in plan, but may have a newer
// serial.
terraform: ugly huge change to weave in new HCL2-oriented types Due to how deeply the configuration types go into Terraform Core, there isn't a great way to switch out to HCL2 gradually. As a consequence, this huge commit gets us from the old state to a _compilable_ new state, but does not yet attempt to fix any tests and has a number of known missing parts and bugs. We will continue to iterate on this in forthcoming commits, heading back towards passing tests and making Terraform fully-functional again. The three main goals here are: - Use the configuration models from the "configs" package instead of the older models in the "config" package, which is now deprecated and preserved only to help us write our migration tool. - Do expression inspection and evaluation using the functionality of the new "lang" package, instead of the Interpolator type and related functionality in the main "terraform" package. - Represent addresses of various objects using types in the addrs package, rather than hand-constructed strings. This is not critical to support the above, but was a big help during the implementation of these other points since it made it much more explicit what kind of address is expected in each context. Since our new packages are built to accommodate some future planned features that are not yet implemented (e.g. the "for_each" argument on resources, "count"/"for_each" on modules), and since there's still a fair amount of functionality still using old-style APIs, there is a moderate amount of shimming here to connect new assumptions with old, hopefully in a way that makes it easier to find and eliminate these shims later. I apologize in advance to the person who inevitably just found this huge commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
func (p *Plan) Context(opts *ContextOpts) (*Context, tfdiags.Diagnostics) {
var err error
opts, err = p.contextOpts(opts)
if err != nil {
terraform: ugly huge change to weave in new HCL2-oriented types Due to how deeply the configuration types go into Terraform Core, there isn't a great way to switch out to HCL2 gradually. As a consequence, this huge commit gets us from the old state to a _compilable_ new state, but does not yet attempt to fix any tests and has a number of known missing parts and bugs. We will continue to iterate on this in forthcoming commits, heading back towards passing tests and making Terraform fully-functional again. The three main goals here are: - Use the configuration models from the "configs" package instead of the older models in the "config" package, which is now deprecated and preserved only to help us write our migration tool. - Do expression inspection and evaluation using the functionality of the new "lang" package, instead of the Interpolator type and related functionality in the main "terraform" package. - Represent addresses of various objects using types in the addrs package, rather than hand-constructed strings. This is not critical to support the above, but was a big help during the implementation of these other points since it made it much more explicit what kind of address is expected in each context. Since our new packages are built to accommodate some future planned features that are not yet implemented (e.g. the "for_each" argument on resources, "count"/"for_each" on modules), and since there's still a fair amount of functionality still using old-style APIs, there is a moderate amount of shimming here to connect new assumptions with old, hopefully in a way that makes it easier to find and eliminate these shims later. I apologize in advance to the person who inevitably just found this huge commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
var diags tfdiags.Diagnostics
diags = diags.Append(err)
return nil, diags
}
return NewContext(opts)
}
// contextOpts mutates the given base ContextOpts in place to use input
// objects obtained from the receiving plan.
func (p *Plan) contextOpts(base *ContextOpts) (*ContextOpts, error) {
opts := base
2014-07-03 13:32:44 -05:00
opts.Diff = p.Diff
terraform: ugly huge change to weave in new HCL2-oriented types Due to how deeply the configuration types go into Terraform Core, there isn't a great way to switch out to HCL2 gradually. As a consequence, this huge commit gets us from the old state to a _compilable_ new state, but does not yet attempt to fix any tests and has a number of known missing parts and bugs. We will continue to iterate on this in forthcoming commits, heading back towards passing tests and making Terraform fully-functional again. The three main goals here are: - Use the configuration models from the "configs" package instead of the older models in the "config" package, which is now deprecated and preserved only to help us write our migration tool. - Do expression inspection and evaluation using the functionality of the new "lang" package, instead of the Interpolator type and related functionality in the main "terraform" package. - Represent addresses of various objects using types in the addrs package, rather than hand-constructed strings. This is not critical to support the above, but was a big help during the implementation of these other points since it made it much more explicit what kind of address is expected in each context. Since our new packages are built to accommodate some future planned features that are not yet implemented (e.g. the "for_each" argument on resources, "count"/"for_each" on modules), and since there's still a fair amount of functionality still using old-style APIs, there is a moderate amount of shimming here to connect new assumptions with old, hopefully in a way that makes it easier to find and eliminate these shims later. I apologize in advance to the person who inevitably just found this huge commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
opts.Config = p.Config
opts.ProviderSHA256s = p.ProviderSHA256s
opts.Destroy = p.Destroy
terraform: ugly huge change to weave in new HCL2-oriented types Due to how deeply the configuration types go into Terraform Core, there isn't a great way to switch out to HCL2 gradually. As a consequence, this huge commit gets us from the old state to a _compilable_ new state, but does not yet attempt to fix any tests and has a number of known missing parts and bugs. We will continue to iterate on this in forthcoming commits, heading back towards passing tests and making Terraform fully-functional again. The three main goals here are: - Use the configuration models from the "configs" package instead of the older models in the "config" package, which is now deprecated and preserved only to help us write our migration tool. - Do expression inspection and evaluation using the functionality of the new "lang" package, instead of the Interpolator type and related functionality in the main "terraform" package. - Represent addresses of various objects using types in the addrs package, rather than hand-constructed strings. This is not critical to support the above, but was a big help during the implementation of these other points since it made it much more explicit what kind of address is expected in each context. Since our new packages are built to accommodate some future planned features that are not yet implemented (e.g. the "for_each" argument on resources, "count"/"for_each" on modules), and since there's still a fair amount of functionality still using old-style APIs, there is a moderate amount of shimming here to connect new assumptions with old, hopefully in a way that makes it easier to find and eliminate these shims later. I apologize in advance to the person who inevitably just found this huge commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
if len(p.Targets) != 0 {
// We're still using target strings in the Plan struct, so we need to
// convert to our address representation here.
// FIXME: Change the Plan struct to use addrs.Targetable itself, and
// then handle these conversions when we read/write plans on disk.
targets := make([]addrs.Targetable, len(p.Targets))
for i, targetStr := range p.Targets {
traversal, travDiags := hclsyntax.ParseTraversalAbs([]byte(targetStr), "", hcl.Pos{})
if travDiags.HasErrors() {
return nil, travDiags
}
target, targDiags := addrs.ParseTarget(traversal)
if targDiags.HasErrors() {
return nil, targDiags.Err()
}
targets[i] = target.Subject
}
opts.Targets = targets
}
if opts.State == nil {
opts.State = p.State
} else if !opts.State.Equal(p.State) {
// Even if we're overriding the state, it should be logically equal
// to what's in plan. The only valid change to have made by the time
// we get here is to have incremented the serial.
//
// Due to the fact that serialization may change the representation of
// the state, there is little chance that these aren't actually equal.
// Log the error condition for reference, but continue with the state
// we have.
log.Println("[WARN] Plan state and ContextOpts state are not equal")
}
thisVersion := version.String()
if p.TerraformVersion != "" && p.TerraformVersion != thisVersion {
return nil, fmt.Errorf(
"plan was created with a different version of Terraform (created with %s, but running %s)",
p.TerraformVersion, thisVersion,
)
}
terraform: ugly huge change to weave in new HCL2-oriented types Due to how deeply the configuration types go into Terraform Core, there isn't a great way to switch out to HCL2 gradually. As a consequence, this huge commit gets us from the old state to a _compilable_ new state, but does not yet attempt to fix any tests and has a number of known missing parts and bugs. We will continue to iterate on this in forthcoming commits, heading back towards passing tests and making Terraform fully-functional again. The three main goals here are: - Use the configuration models from the "configs" package instead of the older models in the "config" package, which is now deprecated and preserved only to help us write our migration tool. - Do expression inspection and evaluation using the functionality of the new "lang" package, instead of the Interpolator type and related functionality in the main "terraform" package. - Represent addresses of various objects using types in the addrs package, rather than hand-constructed strings. This is not critical to support the above, but was a big help during the implementation of these other points since it made it much more explicit what kind of address is expected in each context. Since our new packages are built to accommodate some future planned features that are not yet implemented (e.g. the "for_each" argument on resources, "count"/"for_each" on modules), and since there's still a fair amount of functionality still using old-style APIs, there is a moderate amount of shimming here to connect new assumptions with old, hopefully in a way that makes it easier to find and eliminate these shims later. I apologize in advance to the person who inevitably just found this huge commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
opts.Variables = make(InputValues)
for k, v := range p.Vars {
terraform: ugly huge change to weave in new HCL2-oriented types Due to how deeply the configuration types go into Terraform Core, there isn't a great way to switch out to HCL2 gradually. As a consequence, this huge commit gets us from the old state to a _compilable_ new state, but does not yet attempt to fix any tests and has a number of known missing parts and bugs. We will continue to iterate on this in forthcoming commits, heading back towards passing tests and making Terraform fully-functional again. The three main goals here are: - Use the configuration models from the "configs" package instead of the older models in the "config" package, which is now deprecated and preserved only to help us write our migration tool. - Do expression inspection and evaluation using the functionality of the new "lang" package, instead of the Interpolator type and related functionality in the main "terraform" package. - Represent addresses of various objects using types in the addrs package, rather than hand-constructed strings. This is not critical to support the above, but was a big help during the implementation of these other points since it made it much more explicit what kind of address is expected in each context. Since our new packages are built to accommodate some future planned features that are not yet implemented (e.g. the "for_each" argument on resources, "count"/"for_each" on modules), and since there's still a fair amount of functionality still using old-style APIs, there is a moderate amount of shimming here to connect new assumptions with old, hopefully in a way that makes it easier to find and eliminate these shims later. I apologize in advance to the person who inevitably just found this huge commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
opts.Variables[k] = &InputValue{
Value: v,
SourceType: ValueFromPlan,
}
}
return opts, nil
2014-07-03 13:32:44 -05:00
}
func (p *Plan) String() string {
2014-06-25 20:22:42 -05:00
buf := new(bytes.Buffer)
buf.WriteString("DIFF:\n\n")
buf.WriteString(p.Diff.String())
buf.WriteString("\n\nSTATE:\n\n")
2014-06-25 20:22:42 -05:00
buf.WriteString(p.State.String())
return buf.String()
}
func (p *Plan) init() {
p.once.Do(func() {
if p.Diff == nil {
p.Diff = new(Diff)
p.Diff.init()
}
if p.State == nil {
p.State = new(State)
p.State.init()
}
if p.Vars == nil {
terraform: ugly huge change to weave in new HCL2-oriented types Due to how deeply the configuration types go into Terraform Core, there isn't a great way to switch out to HCL2 gradually. As a consequence, this huge commit gets us from the old state to a _compilable_ new state, but does not yet attempt to fix any tests and has a number of known missing parts and bugs. We will continue to iterate on this in forthcoming commits, heading back towards passing tests and making Terraform fully-functional again. The three main goals here are: - Use the configuration models from the "configs" package instead of the older models in the "config" package, which is now deprecated and preserved only to help us write our migration tool. - Do expression inspection and evaluation using the functionality of the new "lang" package, instead of the Interpolator type and related functionality in the main "terraform" package. - Represent addresses of various objects using types in the addrs package, rather than hand-constructed strings. This is not critical to support the above, but was a big help during the implementation of these other points since it made it much more explicit what kind of address is expected in each context. Since our new packages are built to accommodate some future planned features that are not yet implemented (e.g. the "for_each" argument on resources, "count"/"for_each" on modules), and since there's still a fair amount of functionality still using old-style APIs, there is a moderate amount of shimming here to connect new assumptions with old, hopefully in a way that makes it easier to find and eliminate these shims later. I apologize in advance to the person who inevitably just found this huge commit while spelunking through the commit history.
2018-04-30 12:33:53 -05:00
p.Vars = make(map[string]cty.Value)
}
})
}
// The format byte is prefixed into the plan file format so that we have
// the ability in the future to change the file format if we want for any
// reason.
const planFormatMagic = "tfplan"
const planFormatVersion byte = 2
// ReadPlan reads a plan structure out of a reader in the format that
// was written by WritePlan.
func ReadPlan(src io.Reader) (*Plan, error) {
var result *Plan
var err error
n := 0
// Verify the magic bytes
magic := make([]byte, len(planFormatMagic))
for n < len(magic) {
n, err = src.Read(magic[n:])
if err != nil {
return nil, fmt.Errorf("error while reading magic bytes: %s", err)
}
}
if string(magic) != planFormatMagic {
return nil, fmt.Errorf("not a valid plan file")
}
// Verify the version is something we can read
var formatByte [1]byte
n, err = src.Read(formatByte[:])
if err != nil {
return nil, err
}
if n != len(formatByte) {
return nil, errors.New("failed to read plan version byte")
}
if formatByte[0] != planFormatVersion {
return nil, fmt.Errorf("unknown plan file version: %d", formatByte[0])
}
dec := gob.NewDecoder(src)
if err := dec.Decode(&result); err != nil {
return nil, err
}
return result, nil
}
// WritePlan writes a plan somewhere in a binary format.
func WritePlan(d *Plan, dst io.Writer) error {
// Write the magic bytes so we can determine the file format later
n, err := dst.Write([]byte(planFormatMagic))
if err != nil {
return err
}
if n != len(planFormatMagic) {
return errors.New("failed to write plan format magic bytes")
}
// Write a version byte so we can iterate on version at some point
n, err = dst.Write([]byte{planFormatVersion})
if err != nil {
return err
}
if n != 1 {
return errors.New("failed to write plan version byte")
}
return gob.NewEncoder(dst).Encode(d)
}