stringvalidator: Support for multibyte strings

[sugimotofp]

Terraform CLI and Framework Versions

❯ terraform version
Terraform v1.2.3
on darwin_arm64
...

❯ go list -m github.com/hashicorp/terraform-plugin-framework
github.com/hashicorp/terraform-plugin-framework v0.17.0

Terraform Plugin Framework Code

[]validator.String{
  stringvalidator.LengthBetween(1, 5),
},

Terraform Configuration

resource "examplecloud_thing" "test" {
  display_name = "あいうえお"
}

Expected Behavior

No errors are returned.

Actual Behavior

│ Error: Invalid Attribute Value Length
│
│ with examplecloud_thing.test,
│ on ../../example/test.tf line 2, in resource "examplecloud_thing" "test":
│ 2: display_name = "あいうえお"
│
│ Attribute display_name string length must be between 1 and 5, got: 15

Steps to Reproduce

1. terraform apply

Logs

No response

Additional Information

len(string) does not count by characters but by bytes, so if a multibyte string is specified for validation, it cannot be validated by characters.
In order to accurately determine the number of characters, utf8.RuneCountInString() should be used.
When developing services for Japan etc, It is convenient to handle the length of multibyte characters.

terraform-plugin-framework-validators/stringvalidator/length_between.go

Line 36 in d124347

if l := len(value); l < v.minLength || l > v.maxLength {

ref: https://pkg.go.dev/unicode/utf8#RuneCountInString

Code of Conduct

• I agree to follow this project's Code of Conduct

[bflad]

Hi @sugimotofp 👋 Thank you for raising this.

Given that Terraform and framework string handling is UTF-8 based, this seems like a valid use case that should be handled in this Go module. Whether the existing Length* implementations should be updated to use utf8.RuneCountInString() or separate UTF-8 specific length validator should be introduced may be something worth exploring. Some APIs may only anticipate ASCII-based strings and therefore the encoding and length differences between the two may matter. If we were to introduce new length validators, I would propose something along the lines of UTF8LengthAtLeast, etc.

One immediate change we can accept for this issue is updating the Length* Go documentation to match the existing implementation and denote that it operates on single-byte characters.

[bflad]

Submitted #87 with new stringvalidator.UTF8Length* validators.

[bflad]

terraform-plugin-framework-validators v0.9.0 has been released with new UTF-8 character compatible validators:

• stringvalidator.UTF8LengthAtLeast
• stringvalidator.UTF8LengthAtMost
• stringvalidator.UTF8LengthBetween

Please reach out if you need anything else. 😄

[sugimotofp]

Many thanks for your swift response - fantastic..!!