> For the complete documentation index, see [llms.txt](https://docs.zama.org/tfhe-rs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.zama.org/tfhe-rs/fhe-computation/types/kv-store.md).

# KVStore

This document describes the `KVStore` type provided by the High-level API.

The KVStore is a storage type that associates keys to values, similar to a hash table. In the KVStore, the keys are **clear** numbers, and values are encrypted numbers such as `FheUint` or `FheInt`.

## Clear-key operations

The KVStore supports operations where the queried key is **clear**. These operations are inexpensive and efficient to perform.

* `insert_with_clear_key` - insert or replace a key-value pair
* `update_with_clear_key` - update the value associated to an existing key
* `remove_with_clear_key` - remove an existing key-value pair
* `get_with_clear_key` - get the value associated to a key

## Encrypted-key operations

The KVStore also supports doing queries using an **encrypted** key.

* `get` - get the value associated to a key
* `update` - update the value associated to an already existing key
* `map` - update the value associated to an already existing key, by computing a function on it. This is faster than doing `get`, then `update`.

Encrypted-key operations do not support inserting or removing key-value pairs.

## Serialization

To serialize a `KVStore`, it must first be compressed.

```toml
# Cargo.toml

[dependencies]
tfhe = { version = "~1.8.2", features = ["integer"] }
```

```rust
use tfhe::shortint::parameters::{PARAM_MESSAGE_2_CARRY_2, COMP_PARAM_MESSAGE_2_CARRY_2};
use tfhe::{ConfigBuilder, generate_keys, set_server_key, KVStore, FheUint32, FheUint8,CompressedKVStore};
use tfhe::prelude::*;

fn main() {
    let config = ConfigBuilder::with_custom_parameters(PARAM_MESSAGE_2_CARRY_2)
            .enable_compression(COMP_PARAM_MESSAGE_2_CARRY_2)
            .build();
    let (cks, sks) = generate_keys(config);

    set_server_key(sks);

    let mut kv_store = KVStore::new();
    for i in 0..5u8 {
        let value = FheUint32::encrypt(u32::MAX - u32::from(i), &cks);
        let old = kv_store.insert_with_clear_key(i, value);
        assert!(old.is_none());
    }


    // Example of serialization
    // First, compress, then serialize
    let compressed = kv_store.compress().unwrap();
    let mut data = vec![];
    tfhe::safe_serialization::safe_serialize(&compressed, &mut data, 1 << 30)
        .unwrap();

    // Deserialization: deserialize, then decompress
    let compressed = 
        tfhe::safe_serialization::safe_deserialize::<CompressedKVStore<u8, FheUint32>,>(data.as_slice(), 1 << 30)
        .unwrap();
    let mut kv_store = compressed.decompress().unwrap();


    for i in 0..5u8 {
        let encrypted_key = FheUint32::encrypt(i, &cks);
        let (value, is_some) = kv_store.get(&encrypted_key);
        let is_some = is_some.decrypt(&cks);
        let value: u32 = value.decrypt(&cks);
        assert!(is_some);
        assert_eq!(value, u32::MAX - u32::from(i));
    }

    let value = FheUint32::encrypt(9682734u32, &cks);
    let encrypted_key = FheUint8::encrypt(10u8, &cks);
    let was_updated = kv_store.update(&encrypted_key, &value);
    assert!(!was_updated.decrypt(&cks));

    let encrypted_key = FheUint8::encrypt(1u8, &cks);
    let (old, new, check) = kv_store.map(&encrypted_key, |value| value * 32);
    let old: u32 = old.decrypt(&cks);
    let new: u32 = new.decrypt(&cks);
    let check = check.decrypt(&cks);
    // check encrypts true, if a matching key was found
    assert!(check);
    assert_eq!(old, u32::MAX - 1);
    assert_eq!(new, (u32::MAX - 1).wrapping_mul(32));
}
```


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.zama.org/tfhe-rs/fhe-computation/types/kv-store.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.