NemesisDB is an in-memory database using JSON over websockets. There are APIs for:

• Key values
• Arrays
• Lists

• A JSON key is mapped to its value
• Keys cannot expire, they must be deleted by command
• A future update will add key expiry

These are fixed sized arrays, implemented in contigious memory, with versions for:

• Unsorted arrays for integers, strings and JSON objects
• Sorted arrays for integers and strings

A node based linked list for JSON objects:

• Common features plus splicing
• More features will be added

• Nemesis docs.
• Python API docs

There is an early version of a Python API with docs here.

KV:

from ndb.client import NdbClient
from ndb.kv import KV

client = NdbClient()
await client.open('ws://127.0.0.1:1987/')

kv = KV(client)
await kv.set({'username':'billy', 'password':'billy_password'})

username = await kv.get(key='username')
print (username) # billy

values = await kv.get(keys=('username','password'))
print (values) # {'password':'billy_password', 'username':'billy'}

Sorted integer array:

from ndb.client import NdbClient
from ndb.arrays import SortedIntArrays


client = NdbClient()
await client.open('ws://127.0.0.1:1987/')

sortedInts = SortedIntArrays(client)

await sortedInts.create('array1', capacity=5)
await sortedInts.create('array2', capacity=6)

await sortedInts.set_rng('array1', [25,10,50,100,80])
await sortedInts.set_rng('array2', [10,25,100,120,200,5])

intersected = await sortedInts.intersect('array1', 'array2')
print(intersected) # [10,25,100]

Object list:

from ndb.client import NdbClient
from ndb.lists import ObjLists

client = NdbClient()
await client.open('ws://127.0.0.1:1987/')

lists = ObjLists(client)

await lists.create('names')

# add 3 items
await lists.add('names', [{'name':'James'}, {'name':'Jane'}, {'name':'John'}])
# add new head (Jack, James, Jane, John)
await lists.add_head('names', {'name':'Jack'}) 
# overwrite Jane and John (Jack, James, Brian, Bryan)
await lists.set('names', [{'name':'Brian'},{'name':'Bryan'}], start=2)
# splice Brian and Bryan to a newly created list
await lists.splice(destName='other_names', srcName='names', srcStart=2, srcEnd=4)

print(await lists.get_rng('names', start=0))        # [{'name': 'Jack'}, {'name': 'James'}]
print(await lists.get_rng('other_names', start=0))  #[{'name': 'Brian'}, {'name': 'Bryan'}]

NemesisDB is available as a Debian package and Docker image:

• Package: Releases
• Docker: Docker Hub

You can compile for Linux, instructions below.

Key values can be saved to file and restored at either runtime or at startup in the command line.

1. Clone via SSH with submodules: git clone --recursive git@github.com:nemesisdb/nemesisdb.git
2. Prepare and grab vcpkg libs: cd nemesisdb && ./prepare_vcpkg.sh
3. With VS Code (assuming you have C/C++ and CMake extensions):
   • code .
   • Select kit (tested with GCC 12.3 and GCC 13.1)
   • Select Release variant
   • Select target as nemesisdb
   • Build
4. Binary is in server/Release/bin

After installing via dpkg:

• cd /usr/local/bin/nemesisdb
• Then cd <version>

After build:

• cd <clone_location>/server/Release/bin

In both cases, start server listening on 127.0.0.1:1987 and persistence disabled:

./nemesisdb --config=default.jsonc

ctrl+c to exit.

Externals are either GitHub submodules or managed by vcpkg.

Server:

• uWebsockets : WebSocket server
• jsoncons : json
• ankerl::unordered_dense for map and segmented map
• plog : logging
• uuid_v4 : create UUIDs with SIMD
• Boost Program Options : argv options

Tests Client (will be replaced with Python):

• nlohmann json
• Boost Beast (WebSocket client)
• Google test