Struct merk::Merk

pub struct Merk { /* fields omitted */ }

A handle to a Merkle key/value store backed by RocksDB.

Implementations

impl Merk

pub fn open<P: AsRef<Path>>(path: P) -> Result<Merk>

Opens a store with the specified file path. If no store exists at that path, one will be created.

pub fn open_opt<P>(path: P, db_opts: Options) -> Result<Merk> where
    P: AsRef<Path>, 

Opens a store with the specified file path and the given options. If no store exists at that path, one will be created.

pub fn default_db_opts() -> Options

pub fn get_aux(&self, key: &[u8]) -> Result<Option<Vec<u8>>>

Gets an auxiliary value.

pub fn get(&self, key: &[u8]) -> Result<Option<Vec<u8>>>

Gets a value for the given key. If the key is not found, None is returned.

Note that this is essentially the same as a normal RocksDB get, so should be a fast operation and has almost no tree overhead.

pub fn root_hash(&self) -> Hash

Returns the root hash of the tree (a digest for the entire store which proofs can be checked against). If the tree is empty, returns the null hash (zero-filled).

pub fn apply(&mut self, batch: &Batch, aux: &Batch) -> Result<()>

Applies a batch of operations (puts and deletes) to the tree.

This will fail if the keys in batch are not sorted and unique. This check creates some overhead, so if you are sure your batch is sorted and unique you can use the unsafe apply_unchecked for a small performance gain.

Example

use merk::Op;

let batch = &[
    (vec![1, 2, 3], Op::Put(vec![4, 5, 6])), // puts value [4,5,6] to key [1,2,3]
    (vec![4, 5, 6], Op::Delete) // deletes key [4,5,6]
];
store.apply(batch, &[]).unwrap();

pub unsafe fn apply_unchecked(
    &mut self,
    batch: &Batch,
    aux: &Batch
) -> Result<()>

Applies a batch of operations (puts and deletes) to the tree.

This is unsafe because the keys in batch must be sorted and unique - if they are not, there will be undefined behavior. For a safe version of this method which checks to ensure the batch is sorted and unique, see apply.

Example

use merk::Op;

let batch = &[
    (vec![1, 2, 3], Op::Put(vec![4, 5, 6])), // puts value [4,5,6] to key [1,2,3]
    (vec![4, 5, 6], Op::Delete) // deletes key [4,5,6]
];
unsafe { store.apply_unchecked(batch, &[]).unwrap() };

pub fn destroy(self) -> Result<()>

Closes the store and deletes all data from disk.

pub fn prove(&self, query: &[Vec<u8>]) -> Result<Vec<u8>>

Creates a Merkle proof for the list of queried keys. For each key in the query, if the key is found in the store then the value will be proven to be in the tree. For each key in the query that does not exist in the tree, its absence will be proven by including boundary keys.

The proof returned is in an encoded format which can be verified with merk::verify.

This will fail if the keys in query are not sorted and unique. This check adds some overhead, so if you are sure your batch is sorted and unique you can use the unsafe prove_unchecked for a small performance gain.

pub unsafe fn prove_unchecked(&self, query: &[Vec<u8>]) -> Result<Vec<u8>>

Creates a Merkle proof for the list of queried keys. For each key in the query, if the key is found in the store then the value will be proven to be in the tree. For each key in the query that does not exist in the tree, its absence will be proven by including boundary keys.

The proof returned is in an encoded format which can be verified with merk::verify.

This is unsafe because the keys in query must be sorted and unique - if they are not, there will be undefined behavior. For a safe version of this method which checks to ensure the batch is sorted and unique, see prove.

pub fn flush(&self) -> Result<()>

pub fn commit(
    &mut self,
    deleted_keys: LinkedList<Vec<u8>>,
    aux: &Batch
) -> Result<()>

pub fn walk<T>(
    &self,
    f: impl FnOnce(Option<RefWalker<'_, MerkSource<'_>>>) -> T
) -> T

pub fn raw_iter(&self) -> DBRawIterator<'_>

Trait Implementations

impl Drop for Merk

fn drop(&mut self)

Executes the destructor for this type.