[pbs-devel] [PATCH proxmox-backup v2 2/4] datastore: test DataBlob encode/decode roundtrip

[Dominik Csapak]

On 7/31/24 11:47, Lukas Wagner wrote:
> On  2024-07-31 11:36, Dominik Csapak wrote:
>> so that we can be sure we can decode an encoded blob again
>>
>> Signed-off-by: Dominik Csapak <d.csapak at proxmox.com>
>> ---
>> new in v2
>>   pbs-datastore/src/data_blob.rs | 66 ++++++++++++++++++++++++++++++++++
>>   1 file changed, 66 insertions(+)
>>
>> diff --git a/pbs-datastore/src/data_blob.rs b/pbs-datastore/src/data_blob.rs
>> index a7a55fb7..8715afef 100644
>> --- a/pbs-datastore/src/data_blob.rs
>> +++ b/pbs-datastore/src/data_blob.rs
>> @@ -562,3 +562,69 @@ impl<'a, 'b> DataChunkBuilder<'a, 'b> {
>>           chunk_builder.build()
>>       }
>>   }
>> +
>> +#[cfg(test)]
>> +mod test {
>> +    use pbs_tools::crypt_config::CryptConfig;
>> +
>> +    use super::DataChunkBuilder;
>> +
>> +    const TEST_DATA_LEN: usize = 50;
>> +
>> +    #[test]
>> +    fn test_data_blob_builder() {
>> +        let mut data = Vec::with_capacity(TEST_DATA_LEN);
>> +        for i in 0..TEST_DATA_LEN / 10 {
>> +            for _ in 0..10 {
>> +                data.push(i as u8);
>> +            }
>> +        }
>> +
>> +        // unencrypted, uncompressed
>> +        let (chunk, digest) = DataChunkBuilder::new(&data)
>> +            .compress(false)
>> +            .build()
>> +            .expect("could not create unencrypted, uncompressed chunk");
>> +
>> +        let data_decoded = chunk
>> +            .decode(None, Some(&digest))
>> +            .expect("cannot decode unencrypted, uncompressed chunk");
>> +        assert_eq!(data, data_decoded);
>> +
>> +        // unencrypted, compressed
>> +        let (chunk, digest) = DataChunkBuilder::new(&data)
>> +            .compress(true)
>> +            .build()
>> +            .expect("could not create unencrypted, compressed chunk");
>> +
>> +        let data_decoded = chunk
>> +            .decode(None, Some(&digest))
>> +            .expect("cannot decode unencrypted, compressed chunk");
>> +        assert_eq!(data, data_decoded);
>> +
>> +        // encrypted, uncompressed
>> +        let crypt_config = CryptConfig::new([9; 32]).expect("could not create crypt config");
>> +        let (chunk, digest) = DataChunkBuilder::new(&data)
>> +            .compress(false)
>> +            .crypt_config(&crypt_config)
>> +            .build()
>> +            .expect("could not create encrypted, uncompressed chunk");
>> +
>> +        let data_decoded = chunk
>> +            .decode(Some(&crypt_config), Some(&digest))
>> +            .expect("cannot decode encrypted, uncompressed chunk");
>> +        assert_eq!(data, data_decoded);
>> +
>> +        // encrypted, compressed
>> +        let (chunk, digest) = DataChunkBuilder::new(&data)
>> +            .compress(true)
>> +            .crypt_config(&crypt_config)
>> +            .build()
>> +            .expect("could not create encrypted, compressed chunk");
>> +
>> +        let data_decoded = chunk
>> +            .decode(Some(&crypt_config), Some(&digest))
>> +            .expect("cannot decode encrypted, compressed chunk");
>> +        assert_eq!(data, data_decoded);
>> +    }
>> +}
> 
> IMO it would be nicer to move the (sub) testcases
> (e.g. 'encrypted, compressed', 'encrypted, uncompressed') to individual test cases,
> transforming the info from the comments into the test name, for instance:
> 
>    #[test]
>    fn test_encrypted_uncompressed() { ... }
> 
> (the fact that you test the data_blob module is clear from the full test name, which includes
> the module prefix - so I think you can drop it from the test name)
> 
> IMO that is quite nice when at some point we end up with a failing assertion, because
> you see exactly from the name of the failing test what scenario did not work.
> 
> What's your take on this?
> 

sure makes sense

i'll send a v3, but i'll wait for a review on the other patches first