diskhash: Disk-based hash table

[ data, library, mit ] [ Propose Tags ]

Disk-based hash table

[Skip to Readme]
Versions [faq],,,,,,,,
Change log ChangeLog
Dependencies base (>4.8 && <5), bytestring (==0.10.*) [details]
License MIT
Author Luis Pedro Coelho
Maintainer Luis Pedro Coelho
Category Data
Bug tracker https://github.com/luispedro/diskhash/issues
Source repo head: git clone https://github.com/luispedro/diskhash
Uploaded by luispedro at 2017-11-09T14:29:54Z
Distributions NixOS:
Downloads 5158 total (23 in the last 30 days)
Rating 2.0 (votes: 1) [estimated by Bayesian average]
Your Rating
  • λ
  • λ
  • λ
Status Hackage Matrix CI
Docs available [build log]
Last success reported on 2017-11-09 [all 1 reports]




Maintainer's Corner

For package maintainers and hackage trustees

Readme for diskhash-

[back to package description]

Disk-based hashtable

Travis License: MIT

A simple disk-based hash table (i.e., persistent hash table).

It is a hashtable implemented on memory-mapped disk, so that it can be loaded with a single mmap() system call and used in memory directly (being as fast as an in-memory hashtable once it is loaded from disk).

The code is in C, wrappers are provided for Python, Haskell, and C++. The wrappers follow similar APIs with variations to accommodate the language specificity. They all use the same underlying code, so you can open a hashtable created in C from Haskell, modify it within your Haskell code, and later open the result in Python (although Python's version currently only deals with integers, stored as longs).

Cross-language functionality will work best for very simple types so that you can control their binary representation (64-bit integers, for example).

Reading does not touch the disk representation at all and, thus, can be done on top of read-only files or using multiple threads (and different processes will share the memory: the operating system does that for you). Writing or modifying values is, however, not thread-safe.


The following examples all create a hashtable to store longs (int64_t), then set the value associated with the key "key" to 9. In the current API, the maximum size of the keys needs to be pre-specified, which is the value 15 below.

Raw C

#include <stdio.h>
#include <inttypes.h>
#include "diskhash.h"

int main(void) {
    HashTableOpts opts;
    opts.key_maxlen = 15;
    opts.object_datalen = sizeof(int64_t);
    char* err = NULL;
    HashTable* ht = dht_open("testing.dht", opts, O_RDWR|O_CREAT, &err);
    if (!ht) {
        if (!err) err = "Unknown error";
        fprintf(stderr, "Failed opening hash table: %s.\n", err);
        return 1;
    long i = 9;
    dht_insert(ht, "key", &i);
    long* val = (long*) dht_lookup(ht, "key");
    printf("Looked up value: %l\n", *val);

    return 0;


In Haskell, you have different types/functions for read-write and read-only hashtables.

Read write example:

import Data.DiskHash
import Data.Int
main = do
    ht <- htOpenRW "testing.dht" 15
    htInsertRW ht "key" (9 :: Int64)
    val <- htLookupRW "key" ht
    print val

Read only example (htLookupRO is pure in this case):

import Data.DiskHash
import Data.Int
main = do
    ht <- htOpenRO "testing.dht" 15
    let val :: Int64
        val = htLookupRO "key" ht
    print val


Python's interface is based on the struct module. For example, 'll' refers to a pair of 64-bit ints (longs):

import diskhash
tb = diskhash.StructHash("testing.dht", 15, 'll', 'rw')
tb.insert("key", 1, 2)

The Python interface is currently Python 3 only. Patches to extend it to 2.7 are welcome, but it's not a priority.


In C++, a simple wrapper is defined, which provides a modicum of type-safety. You use the DiskHash<T> template. Additionally, errors are reported through exceptions (both std::bad_alloc and std::runtime_error can be thrown) and not return codes.

#include <iostream>
#include <string>

#include <diskhash.hpp>

int main() {
    const int key_maxlen = 15;
    dht::DiskHash<uint64_t> ht("testing.dht", key_maxlen, dht::DHOpenRW);
    std::string line;
    uint64_t ix = 0;
    while (std::getline(std::cine, line)) {
        if (line.length() > key_maxlen) {
            std::cerr << "Key too long: '" << line << "'. Aborting.\n";
            return 2;
        const bool inserted = ht.insert(line.c_str(), ix);
        if (!inserted) {
            std::cerr  << "Found repeated key '" << line << "' (ignored).\n";
    return 0;


This is beta software. It is good enough that I am using it, but the API can change in the future with little warning. The binary format is versioned (the magic string encodes its version, so changes can be detected and you will get an error message in the future rather than some silent misbehaviour.

Automated unit testing ensures that basic mistakes will not go uncaught.


  • You must specify the maximum key size. This can be worked around either by pre-hashing the keys (with a strong hash) or using multiple hash tables for different key sizes. Neither is currently implemented in diskhash.

  • You cannot delete objects. This was not a necessity for my uses, so it was not implemented. A simple implementation could be done by marking objects as "deleted" in place and recompacting when the hash table size changes or with an explicit dht_gc() call. It may also be important to add functionality to shrink hashtables so as to not waste disk space.

  • The algorithm is a rather na├»ve implementation of linear addression. It would not be hard to switch to robin hood hashing and this may indeed happen in the near future.

License: MIT