terraform/lang/eval.go

471 lines
15 KiB
Go
Raw Normal View History

package lang
import (
"fmt"
"log"
"strconv"
"github.com/hashicorp/terraform/addrs"
"github.com/hashicorp/hcl2/ext/dynblock"
"github.com/hashicorp/hcl2/hcl"
"github.com/hashicorp/hcl2/hcldec"
"github.com/hashicorp/terraform/configs/configschema"
"github.com/hashicorp/terraform/tfdiags"
"github.com/zclconf/go-cty/cty"
"github.com/zclconf/go-cty/cty/convert"
)
// ExpandBlock expands any "dynamic" blocks present in the given body. The
// result is a body with those blocks expanded, ready to be evaluated with
// EvalBlock.
//
// If the returned diagnostics contains errors then the result may be
// incomplete or invalid.
func (s *Scope) ExpandBlock(body hcl.Body, schema *configschema.Block) (hcl.Body, tfdiags.Diagnostics) {
spec := schema.DecoderSpec()
traversals := dynblock.ExpandVariablesHCLDec(body, spec)
refs, diags := References(traversals)
ctx, ctxDiags := s.EvalContext(refs)
diags = diags.Append(ctxDiags)
return dynblock.Expand(body, ctx), diags
}
// EvalBlock evaluates the given body using the given block schema and returns
// a cty object value representing its contents. The type of the result conforms
// to the implied type of the given schema.
//
// This function does not automatically expand "dynamic" blocks within the
// body. If that is desired, first call the ExpandBlock method to obtain
// an expanded body to pass to this method.
//
// If the returned diagnostics contains errors then the result may be
// incomplete or invalid.
func (s *Scope) EvalBlock(body hcl.Body, schema *configschema.Block) (cty.Value, tfdiags.Diagnostics) {
spec := schema.DecoderSpec()
traversals := hcldec.Variables(body, spec)
refs, diags := References(traversals)
ctx, ctxDiags := s.EvalContext(refs)
diags = diags.Append(ctxDiags)
core: Static-validate resource references against schemas In the initial move to HCL2 we started relying only on full expression evaluation to catch attribute errors, but that's not sufficient for resource attributes in practice because during validation we can't know yet whether a resource reference evaluates to a single object or to a list of objects (if count is set). To address this, here we reinstate some static validation of resource references by analyzing directly the reference objects, disregarding any instance index if present, and produce errors if the remaining subsequent traversal steps do not correspond to items within the resource type schema. This also allows us to produce some more specialized error messages for certain situations. In particular, we can recognize a reference like aws_instance.foo.count, which in 0.11 and prior was a weird special case for determining the count value of a resource block, and offer a helpful error showing the new length(aws_instance.foo) usage pattern. This eventually delegates to the static traversal validation logic that was added to the configschema package in a previous commit, which also includes some specialized error messages that distinguish between attributes and block types in the schema so that the errors relate more directly to constructs the user can see in the configuration. In future we could potentially move more of the checks from the dynamic schema construction step to the static validation step, but resources are the reference type that most needs this immediately due to the ambiguity caused by the instance indexing syntax. We can safely refactor other reference types to be statically validated in later releases. This is verified by two pre-existing context validate tests which we temporarily disabled during earlier work (now re-enabled) and also by a new validate test aimed specifically at the special case for the "count" attribute.
2018-11-21 02:25:05 +01:00
if diags.HasErrors() {
// We'll stop early if we found problems in the references, because
// it's likely evaluation will produce redundant copies of the same errors.
return cty.UnknownVal(schema.ImpliedType()), diags
}
val, evalDiags := hcldec.Decode(body, spec, ctx)
diags = diags.Append(evalDiags)
return val, diags
}
// EvalExpr evaluates a single expression in the receiving context and returns
// the resulting value. The value will be converted to the given type before
// it is returned if possible, or else an error diagnostic will be produced
// describing the conversion error.
//
// Pass an expected type of cty.DynamicPseudoType to skip automatic conversion
// and just obtain the returned value directly.
//
// If the returned diagnostics contains errors then the result may be
// incomplete, but will always be of the requested type.
func (s *Scope) EvalExpr(expr hcl.Expression, wantType cty.Type) (cty.Value, tfdiags.Diagnostics) {
refs, diags := ReferencesInExpr(expr)
ctx, ctxDiags := s.EvalContext(refs)
diags = diags.Append(ctxDiags)
core: Static-validate resource references against schemas In the initial move to HCL2 we started relying only on full expression evaluation to catch attribute errors, but that's not sufficient for resource attributes in practice because during validation we can't know yet whether a resource reference evaluates to a single object or to a list of objects (if count is set). To address this, here we reinstate some static validation of resource references by analyzing directly the reference objects, disregarding any instance index if present, and produce errors if the remaining subsequent traversal steps do not correspond to items within the resource type schema. This also allows us to produce some more specialized error messages for certain situations. In particular, we can recognize a reference like aws_instance.foo.count, which in 0.11 and prior was a weird special case for determining the count value of a resource block, and offer a helpful error showing the new length(aws_instance.foo) usage pattern. This eventually delegates to the static traversal validation logic that was added to the configschema package in a previous commit, which also includes some specialized error messages that distinguish between attributes and block types in the schema so that the errors relate more directly to constructs the user can see in the configuration. In future we could potentially move more of the checks from the dynamic schema construction step to the static validation step, but resources are the reference type that most needs this immediately due to the ambiguity caused by the instance indexing syntax. We can safely refactor other reference types to be statically validated in later releases. This is verified by two pre-existing context validate tests which we temporarily disabled during earlier work (now re-enabled) and also by a new validate test aimed specifically at the special case for the "count" attribute.
2018-11-21 02:25:05 +01:00
if diags.HasErrors() {
// We'll stop early if we found problems in the references, because
// it's likely evaluation will produce redundant copies of the same errors.
return cty.UnknownVal(wantType), diags
}
val, evalDiags := expr.Value(ctx)
diags = diags.Append(evalDiags)
if wantType != cty.DynamicPseudoType {
var convErr error
val, convErr = convert.Convert(val, wantType)
if convErr != nil {
val = cty.UnknownVal(wantType)
diags = diags.Append(&hcl.Diagnostic{
Severity: hcl.DiagError,
Summary: "Incorrect value type",
Detail: fmt.Sprintf("Invalid expression value: %s.", tfdiags.FormatError(convErr)),
Subject: expr.Range().Ptr(),
})
}
}
return val, diags
}
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 19:33:53 +02:00
// EvalReference evaluates the given reference in the receiving scope and
// returns the resulting value. The value will be converted to the given type before
// it is returned if possible, or else an error diagnostic will be produced
// describing the conversion error.
//
// Pass an expected type of cty.DynamicPseudoType to skip automatic conversion
// and just obtain the returned value directly.
//
// If the returned diagnostics contains errors then the result may be
// incomplete, but will always be of the requested type.
func (s *Scope) EvalReference(ref *addrs.Reference, wantType cty.Type) (cty.Value, tfdiags.Diagnostics) {
var diags tfdiags.Diagnostics
// We cheat a bit here and just build an EvalContext for our requested
// reference with the "self" address overridden, and then pull the "self"
// result out of it to return.
ctx, ctxDiags := s.evalContext([]*addrs.Reference{ref}, ref.Subject)
diags = diags.Append(ctxDiags)
val := ctx.Variables["self"]
if val == cty.NilVal {
val = cty.DynamicVal
}
var convErr error
val, convErr = convert.Convert(val, wantType)
if convErr != nil {
val = cty.UnknownVal(wantType)
diags = diags.Append(&hcl.Diagnostic{
Severity: hcl.DiagError,
Summary: "Incorrect value type",
Detail: fmt.Sprintf("Invalid expression value: %s.", tfdiags.FormatError(convErr)),
Subject: ref.SourceRange.ToHCL().Ptr(),
})
}
return val, diags
}
// EvalContext constructs a HCL expression evaluation context whose variable
// scope contains sufficient values to satisfy the given set of references.
//
// Most callers should prefer to use the evaluation helper methods that
// this type offers, but this is here for less common situations where the
// caller will handle the evaluation calls itself.
func (s *Scope) EvalContext(refs []*addrs.Reference) (*hcl.EvalContext, tfdiags.Diagnostics) {
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 19:33:53 +02:00
return s.evalContext(refs, s.SelfAddr)
}
func (s *Scope) evalContext(refs []*addrs.Reference, selfAddr addrs.Referenceable) (*hcl.EvalContext, tfdiags.Diagnostics) {
if s == nil {
panic("attempt to construct EvalContext for nil Scope")
}
var diags tfdiags.Diagnostics
vals := make(map[string]cty.Value)
funcs := s.Functions()
ctx := &hcl.EvalContext{
Variables: vals,
Functions: funcs,
}
if len(refs) == 0 {
// Easy path for common case where there are no references at all.
return ctx, diags
}
core: Static-validate resource references against schemas In the initial move to HCL2 we started relying only on full expression evaluation to catch attribute errors, but that's not sufficient for resource attributes in practice because during validation we can't know yet whether a resource reference evaluates to a single object or to a list of objects (if count is set). To address this, here we reinstate some static validation of resource references by analyzing directly the reference objects, disregarding any instance index if present, and produce errors if the remaining subsequent traversal steps do not correspond to items within the resource type schema. This also allows us to produce some more specialized error messages for certain situations. In particular, we can recognize a reference like aws_instance.foo.count, which in 0.11 and prior was a weird special case for determining the count value of a resource block, and offer a helpful error showing the new length(aws_instance.foo) usage pattern. This eventually delegates to the static traversal validation logic that was added to the configschema package in a previous commit, which also includes some specialized error messages that distinguish between attributes and block types in the schema so that the errors relate more directly to constructs the user can see in the configuration. In future we could potentially move more of the checks from the dynamic schema construction step to the static validation step, but resources are the reference type that most needs this immediately due to the ambiguity caused by the instance indexing syntax. We can safely refactor other reference types to be statically validated in later releases. This is verified by two pre-existing context validate tests which we temporarily disabled during earlier work (now re-enabled) and also by a new validate test aimed specifically at the special case for the "count" attribute.
2018-11-21 02:25:05 +01:00
// First we'll do static validation of the references. This catches things
// early that might otherwise not get caught due to unknown values being
// present in the scope during planning.
if staticDiags := s.Data.StaticValidateReferences(refs, selfAddr); staticDiags.HasErrors() {
diags = diags.Append(staticDiags)
return ctx, diags
}
// The reference set we are given has not been de-duped, and so there can
// be redundant requests in it for two reasons:
// - The same item is referenced multiple times
// - Both an item and that item's container are separately referenced.
// We will still visit every reference here and ask our data source for
// it, since that allows us to gather a full set of any errors and
// warnings, but once we've gathered all the data we'll then skip anything
// that's redundant in the process of populating our values map.
dataResources := map[string]map[string]map[addrs.InstanceKey]cty.Value{}
managedResources := map[string]map[string]map[addrs.InstanceKey]cty.Value{}
wholeModules := map[string]map[addrs.InstanceKey]cty.Value{}
moduleOutputs := map[string]map[addrs.InstanceKey]map[string]cty.Value{}
inputVariables := map[string]cty.Value{}
localValues := map[string]cty.Value{}
pathAttrs := map[string]cty.Value{}
terraformAttrs := map[string]cty.Value{}
countAttrs := map[string]cty.Value{}
var self cty.Value
for _, ref := range refs {
rng := ref.SourceRange
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 19:33:53 +02:00
isSelf := false
rawSubj := ref.Subject
if rawSubj == addrs.Self {
if selfAddr == nil {
diags = diags.Append(&hcl.Diagnostic{
Severity: hcl.DiagError,
Summary: `Invalid "self" reference`,
// This detail message mentions some current practice that
// this codepath doesn't really "know about". If the "self"
// object starts being supported in more contexts later then
// we'll need to adjust this message.
Detail: `The "self" object is not available in this context. This object can be used only in resource provisioner and connection blocks.`,
Subject: ref.SourceRange.ToHCL().Ptr(),
})
continue
}
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 19:33:53 +02:00
// Treat "self" as an alias for the configured self address.
rawSubj = selfAddr
isSelf = true
if rawSubj == addrs.Self {
// Programming error: the self address cannot alias itself.
panic("scope SelfAddr attempting to alias itself")
}
}
// This type switch must cover all of the "Referenceable" implementations
// in package addrs.
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 19:33:53 +02:00
switch subj := rawSubj.(type) {
case addrs.ResourceInstance:
var into map[string]map[string]map[addrs.InstanceKey]cty.Value
switch subj.Resource.Mode {
case addrs.ManagedResourceMode:
into = managedResources
case addrs.DataResourceMode:
into = dataResources
default:
panic(fmt.Errorf("unsupported ResourceMode %s", subj.Resource.Mode))
}
val, valDiags := normalizeRefValue(s.Data.GetResourceInstance(subj, rng))
diags = diags.Append(valDiags)
r := subj.Resource
if into[r.Type] == nil {
into[r.Type] = make(map[string]map[addrs.InstanceKey]cty.Value)
}
if into[r.Type][r.Name] == nil {
into[r.Type][r.Name] = make(map[addrs.InstanceKey]cty.Value)
}
into[r.Type][r.Name][subj.Key] = val
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 19:33:53 +02:00
if isSelf {
self = val
}
case addrs.ModuleCallInstance:
val, valDiags := normalizeRefValue(s.Data.GetModuleInstance(subj, rng))
diags = diags.Append(valDiags)
if wholeModules[subj.Call.Name] == nil {
wholeModules[subj.Call.Name] = make(map[addrs.InstanceKey]cty.Value)
}
wholeModules[subj.Call.Name][subj.Key] = val
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 19:33:53 +02:00
if isSelf {
self = val
}
case addrs.ModuleCallOutput:
val, valDiags := normalizeRefValue(s.Data.GetModuleInstanceOutput(subj, rng))
diags = diags.Append(valDiags)
callName := subj.Call.Call.Name
callKey := subj.Call.Key
if moduleOutputs[callName] == nil {
moduleOutputs[callName] = make(map[addrs.InstanceKey]map[string]cty.Value)
}
if moduleOutputs[callName][callKey] == nil {
moduleOutputs[callName][callKey] = make(map[string]cty.Value)
}
moduleOutputs[callName][callKey][subj.Name] = val
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 19:33:53 +02:00
if isSelf {
self = val
}
case addrs.InputVariable:
val, valDiags := normalizeRefValue(s.Data.GetInputVariable(subj, rng))
diags = diags.Append(valDiags)
inputVariables[subj.Name] = val
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 19:33:53 +02:00
if isSelf {
self = val
}
case addrs.LocalValue:
val, valDiags := normalizeRefValue(s.Data.GetLocalValue(subj, rng))
diags = diags.Append(valDiags)
localValues[subj.Name] = val
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 19:33:53 +02:00
if isSelf {
self = val
}
case addrs.PathAttr:
val, valDiags := normalizeRefValue(s.Data.GetPathAttr(subj, rng))
diags = diags.Append(valDiags)
pathAttrs[subj.Name] = val
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 19:33:53 +02:00
if isSelf {
self = val
}
case addrs.TerraformAttr:
val, valDiags := normalizeRefValue(s.Data.GetTerraformAttr(subj, rng))
diags = diags.Append(valDiags)
terraformAttrs[subj.Name] = val
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 19:33:53 +02:00
if isSelf {
self = val
}
case addrs.CountAttr:
val, valDiags := normalizeRefValue(s.Data.GetCountAttr(subj, rng))
diags = diags.Append(valDiags)
countAttrs[subj.Name] = val
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 19:33:53 +02:00
if isSelf {
self = val
}
default:
// Should never happen
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 19:33:53 +02:00
panic(fmt.Errorf("Scope.buildEvalContext cannot handle address type %T", rawSubj))
}
}
for k, v := range buildResourceObjects(managedResources) {
vals[k] = v
}
vals["data"] = cty.ObjectVal(buildResourceObjects(dataResources))
vals["module"] = cty.ObjectVal(buildModuleObjects(wholeModules, moduleOutputs))
vals["var"] = cty.ObjectVal(inputVariables)
vals["local"] = cty.ObjectVal(localValues)
vals["path"] = cty.ObjectVal(pathAttrs)
vals["terraform"] = cty.ObjectVal(terraformAttrs)
vals["count"] = cty.ObjectVal(countAttrs)
if self != cty.NilVal {
vals["self"] = self
}
return ctx, diags
}
func buildResourceObjects(resources map[string]map[string]map[addrs.InstanceKey]cty.Value) map[string]cty.Value {
vals := make(map[string]cty.Value)
for typeName, names := range resources {
nameVals := make(map[string]cty.Value)
for name, keys := range names {
nameVals[name] = buildInstanceObjects(keys)
}
vals[typeName] = cty.ObjectVal(nameVals)
}
return vals
}
func buildModuleObjects(wholeModules map[string]map[addrs.InstanceKey]cty.Value, moduleOutputs map[string]map[addrs.InstanceKey]map[string]cty.Value) map[string]cty.Value {
vals := make(map[string]cty.Value)
for name, keys := range wholeModules {
vals[name] = buildInstanceObjects(keys)
}
for name, keys := range moduleOutputs {
if _, exists := wholeModules[name]; exists {
// If we also have a whole module value for this name then we'll
// skip this since the individual outputs are embedded in that result.
continue
}
// The shape of this collection isn't compatible with buildInstanceObjects,
// but rather than replicating most of the buildInstanceObjects logic
// here we'll instead first transform the structure to be what that
// function expects and then use it. This is a little wasteful, but
// we do not expect this these maps to be large and so the extra work
// here should not hurt too much.
flattened := make(map[addrs.InstanceKey]cty.Value, len(keys))
for k, vals := range keys {
flattened[k] = cty.ObjectVal(vals)
}
vals[name] = buildInstanceObjects(flattened)
}
return vals
}
func buildInstanceObjects(keys map[addrs.InstanceKey]cty.Value) cty.Value {
if val, exists := keys[addrs.NoKey]; exists {
// If present, a "no key" value supersedes all other values,
// since they should be embedded inside it.
return val
}
// If we only have individual values then we need to construct
// either a list or a map, depending on what sort of keys we
// have.
haveInt := false
haveString := false
maxInt := 0
for k := range keys {
switch tk := k.(type) {
case addrs.IntKey:
haveInt = true
if int(tk) > maxInt {
maxInt = int(tk)
}
case addrs.StringKey:
haveString = true
}
}
// We should either have ints or strings and not both, but
// if we have both then we'll prefer strings and let the
// language interpreter try to convert the int keys into
// strings in a map.
switch {
case haveString:
vals := make(map[string]cty.Value)
for k, v := range keys {
switch tk := k.(type) {
case addrs.StringKey:
vals[string(tk)] = v
case addrs.IntKey:
sk := strconv.Itoa(int(tk))
vals[sk] = v
}
}
return cty.ObjectVal(vals)
case haveInt:
// We'll make a tuple that is long enough for our maximum
// index value. It doesn't matter if we end up shorter than
// the number of instances because if length(...) were
// being evaluated we would've got a NoKey reference and
// thus not ended up in this codepath at all.
vals := make([]cty.Value, maxInt+1)
for i := range vals {
if v, exists := keys[addrs.IntKey(i)]; exists {
vals[i] = v
} else {
// Just a placeholder, since nothing will access this anyway
vals[i] = cty.DynamicVal
}
}
return cty.TupleVal(vals)
default:
// Should never happen because there are no other key types.
log.Printf("[ERROR] strange makeInstanceObjects call with no supported key types")
return cty.EmptyObjectVal
}
}
func normalizeRefValue(val cty.Value, diags tfdiags.Diagnostics) (cty.Value, tfdiags.Diagnostics) {
if diags.HasErrors() {
// If there are errors then we will force an unknown result so that
// we can still evaluate and catch type errors but we'll avoid
// producing redundant re-statements of the same errors we've already
// dealt with here.
return cty.UnknownVal(val.Type()), diags
}
return val, diags
}