Simple and fast implementation of a static on disk kv, in python

leveldb, rocksdb and lmdb all have issues for a static collections of key and values:

• slow to build (many hours) : 3h for rocksdb compared to 1h for this lib (for a 5B collections for 1 long and 2 float16)
• uses more space than necessary (100GB for rocksdb unlike 60GB)
• as fast as this much simpler lib: about 5k sample/s on nvme drive

What this lib does not support:

• non static collection
• variable length values and keys

pip install static_ondisk_kv

Checkout these examples:

• synthetic_example.py
• end_to_end_example.py
• prebuilt_example.py

from static_ondisk_kv import OnDiskKV
from tqdm import tqdm
import random

kv = OnDiskKV(file='/media/nvme/mybigfile', key_format="q", value_format="ee")
print("length", kv.length)
k = kv.get_key(100)
v = kv.get_value(100)
print(k)
print(v)
print(kv[k])

Creates an ondisk kv from file using key_format and value_format for decoding.

Returns the key at position i.

Returns the value at position i.

Returns the value for the key k

sort parquet files of collection input_collection by key_column and writes to output_folder

read parquet of sorted input_collection and writes to output_file the key and values using format key_format and value_format

Either locally, or in gitpod (do export PIP_USER=false there)

Setup a virtualenv:

python3 -m venv .env
source .env/bin/activate
pip install -e .

to run tests:

pip install -r requirements-test.txt

then

make lint
make test

You can use make black to reformat the code

python -m pytest -x -s -v tests -k "dummy" to run a specific test