Crypto.Hash

Represent a context for a given hash algorithm.

This type is an instance of ByteArrayAccess for debugging purpose. Internal layout is architecture dependent, may contain uninitialized data fragments, and change in future versions. The bytearray should not be used as input to cryptographic algorithms.

A context is not erased when it is finished with. A hash algorithm buffers its input a block at a time, and finalizing does not clear what is left there. How much survives depends on where the message ended relative to the block: with SHA-256, a 32-byte message is still in the context in full afterwards, and a 100-byte one leaves its last 36 bytes. hashFinalize works on a copy, so the caller's own context keeps what it had as well. Nothing clears either of them: this is Bytes rather than ScrubbedBytes, and the C clears nothing. They go to the garbage collector as they are, and a core file, a crash dump or a swapped page can carry them away afterwards.

That is a deliberate trade rather than an oversight, and there is no way to ask for the other side of it: no operation here clears a context. Scrubbing them all was measured at about 70% of a 32-byte hash and a third of an incremental one, because the allocation is most of the work when the message is short -- and short hashes are the common case, in HMAC, in HKDF, and anywhere a key or an identifier is hashed. A 64 KB hash does not notice it. Anything that must not be left in memory this way is better not hashed through this interface at all.

Represent a digest for a given hash algorithm.

This type is an instance of ByteArrayAccess from package ram. Module Data.ByteArray provides many primitives to work with those values including conversion to other types.

Creating a digest from a bytearray is also possible with function digestFromByteString.

Try to transform a bytearray into a Digest of specific algorithm.

If the digest is not the right size for the algorithm specified, then Nothing is returned.

Initialize a new context for a specified hash algorithm

Run the hash function but takes an explicit hash algorithm parameter

Run the hashPrefix function but takes an explicit hash algorithm parameter

Initialize a new context for this hash algorithm

Update the context with a list of strict bytestring, and return a new context with the updates.

run hashUpdates on one single bytestring and return the updated context.

Finalize a context and return a digest.

Update the context with the first N bytes of a bytestring and return the digest. The code path is independent from N but much slower than a normal hashUpdate. The function can be called for the last bytes of a message, in order to exclude a variable padding, without leaking the padding length. The begining of the message, never impacted by the padding, should preferably go through hashUpdate for better performance.

Get the block size of a hash algorithm

Get the digest size of a hash algorithm

Hash a strict bytestring into a digest.

Hash the first N bytes of a bytestring, with code path independent from N.

Hash a lazy bytestring into a digest.