test(idn-hostname): reject A-labels that decode to an invalid U-label

[vtushar06]

Following the methodology I used for ipv4 and uuid, I read RFC 5890 section 2.3.2.1 and found the current idn-hostname.json tests Punycode validity but never an A-label that decodes to an invalid U-label.

A string is only an A-label if it decodes to a valid U-label. A validator that accepts any well-formed xn-- without decoding and re-checking the result will accept labels whose U-label is invalid.

Changes

• Added 2 test cases across draft7, draft2019-09, draft2020-12, and v1.
• xn--7a decodes to U+00A1, which is DISALLOWED (RFC 5892 section 2.6) - invalid.
• xn--0ca24w decodes to a-grave + Hebrew aleph, which violates the Bidi rule (RFC 5893) - invalid.

Ecosystem Impact

1. python-jsonschema 4.x: PASSES both (rejects them). idna.encode decodes and re-validates.
2. Go x/net/idna, Node (WHATWG), PHP idn_to_ascii, java.net.IDN: FAIL xn--7a (accept it). Node also FAILS xn--0ca24w.

RFC References

• RFC 5890 section 2.3.2.1 (A-label and U-label definitions): https://www.rfc-editor.org/rfc/rfc5890#section-2.3.2.1
• RFC 5892 section 2.6 (DISALLOWED category): https://www.rfc-editor.org/rfc/rfc5892#section-2.6
• RFC 5893 (the Bidi rule): https://www.rfc-editor.org/rfc/rfc5893

Reproduction commands and the idn-hostname cross-implementation matrix are in my evidence repo: https://github.com/vtushar06/JSON-Schema-format-test-Evidence/blob/main/idn-hostname.md

Related: #965

[Copilot]

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

[jviotti]

Perfect. Looks valid!

[jdesrosiers]

👍

We cover this in hostname.json, but it's completely missing in idn-hostname.json.

[jdesrosiers]

Needs rebase.