terraform_empty_list_equality: improve expression detection

[bmbferreira]

Fixes partially #1424 to be able to also detect empty lists on more complex conditional expressions such as var.enabled && var.proxy_subnets != [] ? 1 : 0.

As @bendrucker referred here, the evaluation of complex attribute values is a deeper issue (#1257) and this PR does not fix that part.

[bmbferreira]

Hi @bendrucker , thanks for the swift review. 🙇
Addressed all your comments, please let me know if there's anything else to fix/improve.

Have a great weekend!

[bendrucker]

Nice, getting closer!

[bendrucker]

Thanks, will be away for a few days, can give this another review in about a week.

[bmbferreira]

Thanks, will be away for a few days, can give this another review in about a week.

Ok! Thank you! 🙇

[wata727]

While working on another issue, added a change to walk expressions inside expressions in WalkExpressions #1432
Perhaps this issue has also been resolved. Can you confirm it?

[wata727]

I checked it again on the latest master branch, but it didn't fix the issue. This fix is still needed.

[wata727]

Perhaps this search logic can be rewritten fairly easily by #1432.

[bmbferreira]

Hi @wata727 ! Thanks for the suggestion and sorry for being silent in the last couple of weeks. Would you mind to ellaborate on your suggestion please? 😕 I'm calling WalkExpressions here but as you noticed the latest change didn't fix the issue, therefore I think I still need to process the expression recursively here.

[wata727]

I think the pattern that this rule essentially wants to detect is any == []. In other words, it might be better to check against BinaryOpExpr while walking through all the expressions, rather than checking for ConditionalExpr first.

Imagine something like:

runner.WalkExpression(func(expr hcl.Expression) error {
  if binaryOpExpr, ok := expr.(*hclsyntax.BinaryOpExpr); ok && binaryOpExpr.Op.Type == cty.Bool {
    if tupleConsExpr, ok := binaryOpExpr.RHS.(*hclsyntax.TupleConsExpr); ok && len(tupleConsExpr.Exprs) == 0 {
      runner.EmitIssue(...)
    }

    if tupleConsExpr, ok := binaryOpExpr.LHS.(*hclsyntax.TupleConsExpr); ok && len(tupleConsExpr.Exprs) == 0 {
      runner.EmitIssue(...)
    }
  }
})

What do you think?

[bmbferreira]

ahh! got it 🙇 Just tried it and it is much better now! I could even remove the map to avoid duplicated ranges 👌 thank you!

[bendrucker]

One small style change, would apply myself but I don't seem to have edit access to push to your PR branch

diff --git a/rules/terraformrules/terraform_empty_list_equality.go b/rules/terraformrules/terraform_empty_list_equality.go
index 25baf37a..83bbfd7a 100644
--- a/rules/terraformrules/terraform_empty_list_equality.go
+++ b/rules/terraformrules/terraform_empty_list_equality.go
@@ -58,17 +58,17 @@ func (r *TerraformEmptyListEqualityRule) checkEmptyList(runner *tflint.Runner) e
 	return runner.WalkExpressions(func(expr hcl.Expression) error {
 		if binaryOpExpr, ok := expr.(*hclsyntax.BinaryOpExpr); ok && binaryOpExpr.Op.Type == cty.Bool {
 			if tupleConsExpr, ok := binaryOpExpr.LHS.(*hclsyntax.TupleConsExpr); ok && len(tupleConsExpr.Exprs) == 0 {
-				r.emitEmptyListEqualityIssue(binaryOpExpr.Range(), runner)
+				r.emitIssue(binaryOpExpr.Range(), runner)
 			} else if tupleConsExpr, ok := binaryOpExpr.RHS.(*hclsyntax.TupleConsExpr); ok && len(tupleConsExpr.Exprs) == 0 {
-				r.emitEmptyListEqualityIssue(binaryOpExpr.Range(), runner)
+				r.emitIssue(binaryOpExpr.Range(), runner)
 			}
 		}
 		return nil
 	})
 }
 
-// emitEmptyListEqualityIssue emits issue for comparison with static empty list
-func (r *TerraformEmptyListEqualityRule) emitEmptyListEqualityIssue(exprRange hcl.Range, runner *tflint.Runner) {
+// emitIssue emits issue for comparison with static empty list
+func (r *TerraformEmptyListEqualityRule) emitIssue(exprRange hcl.Range, runner *tflint.Runner) {
 	runner.EmitIssue(
 		r,
 		"Comparing a collection with an empty list is invalid. To detect an empty collection, check its length.",

You can copy the above and pipe to git apply.

Otherwise LGTM

[bmbferreira]

One small style change, would apply myself but I don't seem to have edit access to push to your PR branch

diff --git a/rules/terraformrules/terraform_empty_list_equality.go b/rules/terraformrules/terraform_empty_list_equality.go
index 25baf37a..83bbfd7a 100644
--- a/rules/terraformrules/terraform_empty_list_equality.go
+++ b/rules/terraformrules/terraform_empty_list_equality.go
@@ -58,17 +58,17 @@ func (r *TerraformEmptyListEqualityRule) checkEmptyList(runner *tflint.Runner) e
 	return runner.WalkExpressions(func(expr hcl.Expression) error {
 		if binaryOpExpr, ok := expr.(*hclsyntax.BinaryOpExpr); ok && binaryOpExpr.Op.Type == cty.Bool {
 			if tupleConsExpr, ok := binaryOpExpr.LHS.(*hclsyntax.TupleConsExpr); ok && len(tupleConsExpr.Exprs) == 0 {
-				r.emitEmptyListEqualityIssue(binaryOpExpr.Range(), runner)
+				r.emitIssue(binaryOpExpr.Range(), runner)
 			} else if tupleConsExpr, ok := binaryOpExpr.RHS.(*hclsyntax.TupleConsExpr); ok && len(tupleConsExpr.Exprs) == 0 {
-				r.emitEmptyListEqualityIssue(binaryOpExpr.Range(), runner)
+				r.emitIssue(binaryOpExpr.Range(), runner)
 			}
 		}
 		return nil
 	})
 }
 
-// emitEmptyListEqualityIssue emits issue for comparison with static empty list
-func (r *TerraformEmptyListEqualityRule) emitEmptyListEqualityIssue(exprRange hcl.Range, runner *tflint.Runner) {
+// emitIssue emits issue for comparison with static empty list
+func (r *TerraformEmptyListEqualityRule) emitIssue(exprRange hcl.Range, runner *tflint.Runner) {
 	runner.EmitIssue(
 		r,
 		"Comparing a collection with an empty list is invalid. To detect an empty collection, check its length.",

You can copy the above and pipe to git apply.

Otherwise LGTM

thanks again for the review @bendrucker. just applied the suggested change!

[bendrucker]

Getting some 500s right now from GitHub for merge status. I'll revisit this and merge tomorrow.

[bmbferreira]

Getting some 500s right now from GitHub for merge status. I'll revisit this and merge tomorrow.

yup, np. Seems they're having some issues: https://www.githubstatus.com/