[Dropped] Adding encrypt_with_per_column_encryption.parquet.encrypted to test Parquet Modular Encryption (PME) per-column encryption

[argmarco-tkd]

What
My team is developing an extension to the C++ Parquet Modular Encryption (PME). In this work we are developing two new features:

• allowing new encryptors to be used
• enabling independent per-column encryption algorithms to be used

In this PR, we add a new encrypted test file to data/ that specifically exercises per-column encryption. This test validates that readers can handle files where different columns are encrypted independently (potentially with different algorithms/configurations), complementing existing PME test vectors.

File Description

• Data: Synthetic, deterministic test table with 1,000 rows and columns orderId, productId, price, vat, customer_name.
• File is encrypted.
  • Footer and default column algorithm: AES_GCM_V1 (via ExternalEncryptionConfiguration.encryption_algorithm).
  • Per-column overrides:
    • orderId: (using test, XOR encryption)
    • productId: AES_GCM_V1 via column_keys mapping to key id productid_key.
• File size: Size: 26,763 bytes
• File SHA-256: 15627b579d994aee66ef55d9e2a9dd719752e7c169d7ea8245a63f5edd94104d

[ggershinsky]

enabling independent per-column encryption algorithms to be used

I think this is not compatible with the parquet spec, https://parquet.apache.org/docs/file-format/data-pages/encryption/ .
The encryption algorithm is stored per-file, not per-column.

[argmarco-tkd]

enabling independent per-column encryption algorithms to be used

I think this is not compatible with the parquet spec, https://parquet.apache.org/docs/file-format/data-pages/encryption/ . The encryption algorithm is stored per-file, not per-column.

Hi - our work will propose an extension to allow algorithms to be specified at column-level. We currently have a fork of apache/arrow where we're performing this work (as mentioned in the description of this PR). We would like to avoid also having to have a fork of parquet-testing also, thus this PR.

[argmarco-tkd]

Dropping this ticket. Once the extensions for per-column algorithm have been discussed with the community, we can bring the discussion of adding the specific tests to parquet-testing.