core: Annotate for_each errors with expression info

Our diagnostics model allows for optionally annotating an error or warning
with information about the expression and eval context it was generated
from, which the diagnostic renderer for the UI will then use to give the
user some additional hints about what values may have contributed to the
error.

We previously didn't have those annotations on the results of evaluating
for_each expressions though, because in that case we were using the helper
function to evaluate an expression in one shot and thus we didn't ever
have a reference to the EvalContext in order to include it in the
diagnostic values.

Now, at the expense of having to handle the evaluation at a slightly lower
level of abstraction, we'll annotate all of the for_each error messages
with source expression information. This is valuable because we see users
often confused as to how their complex for_each expressions ended up being
invalid, and hopefully giving some information about what the inputs were
will allow more users to self-solve.

lang/eval.go

@@ -133,6 +133,8 @@ func (s *Scope) EvalExpr(expr hcl.Expression, wantType cty.Type) (cty.Value, tfd
 				Summary:     "Incorrect value type",
 				Detail:      fmt.Sprintf("Invalid expression value: %s.", tfdiags.FormatError(convErr)),
 				Subject:     expr.Range().Ptr(),
+				Expression:  expr,
+				EvalContext: ctx,
 			})
 		}
 	}

terraform/eval_for_each.go

@@ -4,6 +4,7 @@ import (
 	"fmt"
 
 	"github.com/hashicorp/hcl/v2"
+	"github.com/hashicorp/terraform/lang"
 	"github.com/hashicorp/terraform/tfdiags"
 	"github.com/zclconf/go-cty/cty"
 )
@@ -17,16 +18,9 @@ import (
 // returning an error if the count value is not known, and converting the
 // cty.Value to a map[string]cty.Value for compatibility with other calls.
 func evaluateForEachExpression(expr hcl.Expression, ctx EvalContext) (forEach map[string]cty.Value, diags tfdiags.Diagnostics) {
-	forEachVal, diags := evaluateForEachExpressionValue(expr, ctx)
-	if !forEachVal.IsKnown() {
-		// Attach a diag as we do with count, with the same downsides
-		diags = diags.Append(&hcl.Diagnostic{
-			Severity: hcl.DiagError,
-			Summary:  "Invalid for_each argument",
-			Detail:   `The "for_each" value depends on resource attributes that cannot be determined until apply, so Terraform cannot predict how many instances will be created. To work around this, use the -target argument to first apply only the resources that the for_each depends on.`,
-			Subject:  expr.Range().Ptr(),
-		})
-	}
+	forEachVal, diags := evaluateForEachExpressionValue(expr, ctx, false)
+	// forEachVal might be unknown, but if it is then there should already
+	// be an error about it in diags, which we'll return below.
 
 	if forEachVal.IsNull() || !forEachVal.IsKnown() || forEachVal.LengthInt() == 0 {
 		// we check length, because an empty set return a nil map
@@ -38,7 +32,7 @@ func evaluateForEachExpression(expr hcl.Expression, ctx EvalContext) (forEach ma
 
 // evaluateForEachExpressionValue is like evaluateForEachExpression
 // except that it returns a cty.Value map or set which can be unknown.
-func evaluateForEachExpressionValue(expr hcl.Expression, ctx EvalContext) (cty.Value, tfdiags.Diagnostics) {
+func evaluateForEachExpressionValue(expr hcl.Expression, ctx EvalContext, allowUnknown bool) (cty.Value, tfdiags.Diagnostics) {
 	var diags tfdiags.Diagnostics
 	nullMap := cty.NullVal(cty.Map(cty.DynamicPseudoType))
 
@@ -46,7 +40,23 @@ func evaluateForEachExpressionValue(expr hcl.Expression, ctx EvalContext) (cty.V
 		return nullMap, diags
 	}
 
-	forEachVal, forEachDiags := ctx.EvaluateExpr(expr, cty.DynamicPseudoType, nil)
+	refs, moreDiags := lang.ReferencesInExpr(expr)
+	diags = diags.Append(moreDiags)
+	scope := ctx.EvaluationScope(nil, EvalDataForNoInstanceKey)
+	var hclCtx *hcl.EvalContext
+	if scope != nil {
+		hclCtx, moreDiags = scope.EvalContext(refs)
+	} else {
+		// This shouldn't happen in real code, but it can unfortunately arise
+		// in unit tests due to incompletely-implemented mocks. :(
+		hclCtx = &hcl.EvalContext{}
+	}
+	diags = diags.Append(moreDiags)
+	if diags.HasErrors() { // Can't continue if we don't even have a valid scope
+		return nullMap, diags
+	}
+
+	forEachVal, forEachDiags := expr.Value(hclCtx)
 	diags = diags.Append(forEachDiags)
 	if forEachVal.ContainsMarked() {
 		diags = diags.Append(&hcl.Diagnostic{
@@ -54,6 +64,8 @@ func evaluateForEachExpressionValue(expr hcl.Expression, ctx EvalContext) (cty.V
 			Summary:     "Invalid for_each argument",
 			Detail:      "Sensitive variable, or values derived from sensitive variables, cannot be used as for_each arguments. If used, the sensitive value could be exposed as a resource instance key.",
 			Subject:     expr.Range().Ptr(),
+			Expression:  expr,
+			EvalContext: hclCtx,
 		})
 	}
 	if diags.HasErrors() {
@@ -68,9 +80,21 @@ func evaluateForEachExpressionValue(expr hcl.Expression, ctx EvalContext) (cty.V
 			Summary:     "Invalid for_each argument",
 			Detail:      `The given "for_each" argument value is unsuitable: the given "for_each" argument value is null. A map, or set of strings is allowed.`,
 			Subject:     expr.Range().Ptr(),
+			Expression:  expr,
+			EvalContext: hclCtx,
 		})
 		return nullMap, diags
 	case !forEachVal.IsKnown():
+		if !allowUnknown {
+			diags = diags.Append(&hcl.Diagnostic{
+				Severity:    hcl.DiagError,
+				Summary:     "Invalid for_each argument",
+				Detail:      errInvalidForEachUnknownDetail,
+				Subject:     expr.Range().Ptr(),
+				Expression:  expr,
+				EvalContext: hclCtx,
+			})
+		}
 		// ensure that we have a map, and not a DynamicValue
 		return cty.UnknownVal(cty.Map(cty.DynamicPseudoType)), diags
 
@@ -80,6 +104,8 @@ func evaluateForEachExpressionValue(expr hcl.Expression, ctx EvalContext) (cty.V
 			Summary:     "Invalid for_each argument",
 			Detail:      fmt.Sprintf(`The given "for_each" argument value is unsuitable: the "for_each" argument must be a map, or set of strings, and you have provided a value of type %s.`, ty.FriendlyName()),
 			Subject:     expr.Range().Ptr(),
+			Expression:  expr,
+			EvalContext: hclCtx,
 		})
 		return nullMap, diags
 
@@ -94,6 +120,16 @@ func evaluateForEachExpressionValue(expr hcl.Expression, ctx EvalContext) (cty.V
 		// since we can't use a set values that are unknown, we treat the
 		// entire set as unknown
 		if !forEachVal.IsWhollyKnown() {
+			if !allowUnknown {
+				diags = diags.Append(&hcl.Diagnostic{
+					Severity:    hcl.DiagError,
+					Summary:     "Invalid for_each argument",
+					Detail:      errInvalidForEachUnknownDetail,
+					Subject:     expr.Range().Ptr(),
+					Expression:  expr,
+					EvalContext: hclCtx,
+				})
+			}
 			return cty.UnknownVal(ty), diags
 		}
 
@@ -103,6 +139,8 @@ func evaluateForEachExpressionValue(expr hcl.Expression, ctx EvalContext) (cty.V
 				Summary:     "Invalid for_each set argument",
 				Detail:      fmt.Sprintf(`The given "for_each" argument value is unsuitable: "for_each" supports maps and sets of strings, but you have provided a set containing type %s.`, forEachVal.Type().ElementType().FriendlyName()),
 				Subject:     expr.Range().Ptr(),
+				Expression:  expr,
+				EvalContext: hclCtx,
 			})
 			return cty.NullVal(ty), diags
 		}
@@ -118,6 +156,8 @@ func evaluateForEachExpressionValue(expr hcl.Expression, ctx EvalContext) (cty.V
 					Summary:     "Invalid for_each set argument",
 					Detail:      fmt.Sprintf(`The given "for_each" argument value is unsuitable: "for_each" sets must not contain null values.`),
 					Subject:     expr.Range().Ptr(),
+					Expression:  expr,
+					EvalContext: hclCtx,
 				})
 				return cty.NullVal(ty), diags
 			}
@@ -126,3 +166,5 @@ func evaluateForEachExpressionValue(expr hcl.Expression, ctx EvalContext) (cty.V
 
 	return forEachVal, nil
 }
+
+const errInvalidForEachUnknownDetail = `The "for_each" value depends on resource attributes that cannot be determined until apply, so Terraform cannot predict how many instances will be created. To work around this, use the -target argument to first apply only the resources that the for_each depends on.`

terraform/eval_for_each_test.go

@@ -150,6 +150,16 @@ func TestEvaluateForEachExpression_errors(t *testing.T) {
 			if got, want := diags[0].Description().Detail, test.DetailSubstring; !strings.Contains(got, want) {
 				t.Errorf("wrong diagnostic detail %#v; want %#v", got, want)
 			}
+			if fromExpr := diags[0].FromExpr(); fromExpr != nil {
+				if fromExpr.Expression == nil {
+					t.Errorf("diagnostic does not refer to an expression")
+				}
+				if fromExpr.EvalContext == nil {
+					t.Errorf("diagnostic does not refer to an EvalContext")
+				}
+			} else {
+				t.Errorf("diagnostic does not support FromExpr\ngot: %s", spew.Sdump(diags[0]))
+			}
 		})
 	}
 }
@@ -164,7 +174,7 @@ func TestEvaluateForEachExpressionKnown(t *testing.T) {
 		t.Run(name, func(t *testing.T) {
 			ctx := &MockEvalContext{}
 			ctx.installSimpleEval()
-			forEachVal, diags := evaluateForEachExpressionValue(expr, ctx)
+			forEachVal, diags := evaluateForEachExpressionValue(expr, ctx, true)
 
 			if len(diags) != 0 {
 				t.Errorf("unexpected diagnostics %s", spew.Sdump(diags))

terraform/eval_validate.go

@@ -519,7 +519,7 @@ func (n *EvalValidateResource) validateCount(ctx EvalContext, expr hcl.Expressio
 }
 
 func (n *EvalValidateResource) validateForEach(ctx EvalContext, expr hcl.Expression) (diags tfdiags.Diagnostics) {
-	val, forEachDiags := evaluateForEachExpressionValue(expr, ctx)
+	val, forEachDiags := evaluateForEachExpressionValue(expr, ctx, true)
 	// If the value isn't known then that's the best we can do for now, but
 	// we'll check more thoroughly during the plan walk
 	if !val.IsKnown() {

terraform/node_module_expand.go

@@ -232,7 +232,7 @@ func (n *nodeValidateModule) Execute(ctx EvalContext, op walkOperation) (diags t
 			diags = diags.Append(countDiags)
 
 		case n.ModuleCall.ForEach != nil:
-			_, forEachDiags := evaluateForEachExpressionValue(n.ModuleCall.ForEach, ctx)
+			_, forEachDiags := evaluateForEachExpressionValue(n.ModuleCall.ForEach, ctx, true)
 			diags = diags.Append(forEachDiags)
 		}