stampdb 1.0.0

A tiny C++ Time Series Database library designed for compatibility with the PyData Ecosystem.

StampDB

StampDB is a performant time series database inspired by tinyflux, with a focus on maximizing compatibility with the PyData ecosystem. It is designed to work natively with NumPy and Pythons datetime module.

Key Features

C++ Core

• Efficient CSV Parsing (csv2 based).
• In-Memory Indexing for fast lookups.
• Append-Only Writes for data integrity.
• Simple and fast Range Queries.
• Atmoic Writes.

Python Frontend

• Seamless conversion from C++ CSV objects to NumPy structured arrays.
• Relational algebra operations like joins, summations, and more using NumPy on structured arrays.
• Use Native Datetime objects for I/O.

You should not use StampDB if you need advanced database features like:

• Access from multiple processes or threads
• An HTTP server
• Management of relationships between tables
• Access control and users
• ACID guarantees
• High performance as the size of your dataset grows

Use Cases

• IOT and Sensor Data
• Scientific and Research Data Acquisition
• Single Node Data Processing
• Private Data Storage

Installation

Supported Python versions:

> 3.6 && <= 3.13

OS	3.7	3.8	3.9	3.10	3.11	3.12	3.13	PyPy
Windows	✅	✅	✅	✅	✅	✅	✅	🚫
Linux	✅	✅	✅	✅	✅	✅	✅	🚫
MacOS	✅	✅	✅	✅	✅	✅	✅	🚫

i686 ISA not supported.

Using pip

pip install stampdb

Build from source

Clone the repository.

git clone --recursive https://github.com/you/stampdb.git
# If csv2 C++ library is not cloned, you might have to explicitly clone it at `libs/csv2`.

Build the Python API.

python -m build

Running tests

After going to the tests/ folder, run:

python -m pytest -s

Quick Start

I/O using StampDB.

from stampdb import *

# This will create a csv store with time, temp, humidity columns.
db = StampDB("test.csv", schema={"temp": "float", "humidity": "float"})

# Appending a point.
p = Point(time=1, data=[22.5, "moderate"])
db.append_point(p)

# Doing append only writes to the disk.
db.checkpoint()

# Doing in memory deletion.
db.delete_point(time=1)

# Forcing actual disk deletion.
db.compact() # If not done explicitly, it happens on close.

# Closing the database.
db.close()

Relational Algebra using StampDB.

from stampdb.relational import *

# Given the db is loaded and running using the `Quick Start` section.

out = db.read_range(0, 10)
assert isinstance(out, np.ndarray)

s = Selection("temp > 24", out)
assert s.do().size == 1

p = Projection(["temp"], out)
assert p.do().size == 2

plus = Summation("temp", out)
assert plus.do() == 48

orderby = OrderBy(["temp"], out)
assert orderby.do().size == 2
assert orderby.do()["temp"][0] == 23.5

Joins using StampDB.

from stampdb.relational import *

db = StampDB("test.csv", schema={"temp": "float", "humidity": "float"})
for i in range(100):
    time = i
    temp = random.randint(0, 50)
    humidity = random.choice(["low", "moderate", "high"])
    p = Point(time=time, data=[temp, humidity])
    db.append_point(p)

# Written to disk.
db.compact()

db2 = StampDB("test2.csv", schema={"weather": "string", "temp": "float"})
for i in range(100):
    time = i
    temp = random.randint(0, 50)
    weather = random.choice(["sunny", "rainy", "cloudy"])
    p = Point(time=time, data=[weather, temp])
    db2.append_point(p)

# Written to disk.
db2.compact()

data = db.read_range(0, 100)
assert data.size == 100

ij = InnerJoin(data, db2.read_range(0, 100), "temp", "temp")
assert ij.do().size > 0

oj = OuterJoin(data, db2.read_range(0, 100), "temp", "temp")
assert oj.do().size > 0

loj = LeftOuterJoin(data, db2.read_range(0, 100), "temp", "temp")
assert loj.do().size > 0

db.close()
db2.close()

Runtime Comparison.

Though high performance is not the primary goal of StampDB, it performs significantly better than native Python libraries like tinyflux.

Runtime Comparison with tinyflux

Operation	Speedup
Writes	2×
Queries	50×
Reads	30×

Steps to Reproduce

1. Install tinyflux and StampDB.
2. Navigate to the directory containing benchmarks.py.
3. Run the benchmark:

python benchmarks.py

Contributing Guidelines

• To get started on a pull request, fork the repository on GitHub, create a new branch, and make updates.
• Write unit tests, ensure the code is 100% covered, update documentation where necessary, and format and style the code correctly.
• Send a pull request.