skypydb 1.0.0

Open Source Reactive Database

Skypydb - Open Source Reactive and Vector Embeddings Database.
The better way to build Python logging system! And adding memory to an LLM.

pip install skypydb # python database
# or download from the source
# git clone https://github.com/Ahen-Studio/skypydb.git
# cd skypydb
# pip install -r requirements.txt

[!NOTE] While waiting for mem0 to merge the changes necessary for skypydb to function, you can still use mem0 with skypydb by downloading the mem0 folder and placing every file in the correct directory in the mem0 installation folder.

Features

• Simple: fully-documented and easy to debug with detailed error messages

• Table: create, delete, search data from tables

• Vector embeddings: create, search and delete vectors collections. It supports Ollama, OpenAI, Sentence-Transformers embeddings models (default model is mxbai-embed-large from Ollama).

• Memory: add memory to a LLM by using mem0 and our integration.

• Security, Input Validation: AES-256-GCM encryption for data at rest with selective field encryption, automatic protection against SQL injection attacks

• CLI: command line interface to initialize your database and launch the dashboard with one simple command

• Observable: Dashboard with real-time data, metrics, and query inspection

• Free & Open Source: MIT Licensed

• Cross-platform: Windows, Linux, MacOS

TODO

• Code a rust client

What's next!

• give us ideas!

Error Codes

• Skypydb uses standardized error codes to help you quickly identify and handle issues:

Code	Error	Description
SKY001	SkypydbError	Base exception for all Skypydb errors
SKY101	TableNotFoundError	Raised when attempting to access a table that doesn't exist
SKY102	TableAlreadyExistsError	Raised when trying to create a table that already exists
SKY103	DatabaseError	Raised when a database operation fails
SKY201	InvalidSearchError	Raised when search parameters are invalid
SKY301	SecurityError	Raised when a security operation fails
SKY302	ValidationError	Raised when input validation fails
SKY303	EncryptionError	Raised when encryption/decryption operations fail
SKY401	CollectionNotFoundError	Raised when attempting to access a vector collection that doesn't exist
SKY402	CollectionAlreadyExistsError	Raised when trying to create a collection that already exists
SKY403	EmbeddingError	Raised when embedding generation fails
SKY404	VectorSearchError	Raised when vector similarity search fails

Cli

• use the cli to initialize your database and launch the dashboard with one simple command

skypydb dev

• run this command in your terminal

API

• Use the API to interact with your database; before doing so, make sure to create a schema to define your tables.

"""
Schema definition for Skypydb database tables.
This file defines all tables, their columns, types, and indexes.
"""

from skypydb.schema import defineSchema, defineTable
from skypydb.schema.values import value

# Define the schema with all tables
schema = defineSchema({
    
    # Table for success logs
    "success": defineTable({
        "component": value.string(),
        "action": value.string(),
        "message": value.string(),
        "details": value.optional(value.string()),
        "user_id": value.optional(value.string()),
    })
    .index("by_component", ["component"])
    .index("by_action", ["action"])
    .index("by_user", ["user_id"])
    .index("by_component_and_action", ["component", "action"]),

    # Table for warning logs
    "warning": defineTable({
        "component": value.string(),
        "action": value.string(),
        "message": value.string(),
        "details": value.optional(value.string()),
        "user_id": value.optional(value.string()),
    })
    .index("by_component", ["component"])
    .index("by_action", ["action"])
    .index("by_user", ["user_id"])
    .index("by_component_and_action", ["component", "action"]),

    # Table for error logs
    "error": defineTable({
        "component": value.string(),
        "action": value.string(),
        "message": value.string(),
        "details": value.optional(value.string()),
        "user_id": value.optional(value.string()),
    })
    .index("by_component", ["component"])
    .index("by_action", ["action"])
    .index("by_user", ["user_id"])
    .index("by_component_and_action", ["component", "action"]),
})

• after creating the schema file containing the tables, you can add data to your database

import skypydb

# Create a client
client = skypydb.ReactiveClient()

# Create tables from the schema
# This reads the schema from db/schema.py and creates all tables
tables = client.get_or_create_table()
# if the tables already exists the programe get them instead

# Access your tables
success_table = tables["success"]
warning_table = tables["warning"]
error_table = tables["error"]

# Insert data
# Insert success logs
success_table.add(
    component="AuthService",
    action="login",
    message="User logged in successfully",
    user_id="user123"
)

# Insert warning logs
warning_table.add(
    component="AuthService",
    action="login_attempt",
    message="Multiple failed login attempts",
    user_id="user456",
    details="5 failed attempts in 5 minutes"
)

# Insert error logs
error_table.add(
    component="DatabaseService",
    action="connection",
    message="Connection timeout",
    user_id="system",
    details="Timeout after 30 seconds"
)

• after adding data to your database you can search specific data using the search method

# Search results by filter
user_success_logs = success_table.search(
    user_id="user123"
)

if not user_success_logs:
    print("No results found.")
else:
    for user_success_log in user_success_logs:
        print(user_success_log)

• you can also delete specific data from your database using the delete method

success_table.delete(
    component="AuthService",
    user_id="user123"
)

Vector

• Use the vector API to perform vector operations on your database, it is useful for adding memory to an LLM.

import skypydb

# Create a client
client = skypydb.VectorClient(
    embedding_provider="ollama",
    embedding_model_config={
        "model": "mxbai-embed-large",
        "base_url": "http://localhost:11434"
    }
)

# Create a collection
collection = client.get_or_create_collection("my-documents")

# Add documents
collection.add(
    documents=["This is document1", "This is document2"],
    metadatas=[{"source": "notion"}, {"source": "google-docs"}],
    ids=["doc1", "doc2"]
)

# Query for similar documents
results = collection.query(
    query_texts=["This is a query document"],
    n_results=2
)

# Access results
if not results:
    print("No results found.")
else:
    for i, doc_id in enumerate(results["ids"][0]):
        print(f"{doc_id}, {results['documents'][0][i]}, {results['distances'][0][i]}")

• Use the vector API with OpenAI

import skypydb

# Create a client
client = skypydb.VectorClient(
    embedding_provider="openai",
    embedding_model_config={
        "api_key": "your-openai-api-key",
        "model": "text-embedding-3-small"
    }
)

# Create a collection
collection = client.get_or_create_collection("my-documents")

# Add documents
collection.add(
    documents=["This is document1", "This is document2"],
    metadatas=[{"source": "notion"}, {"source": "google-docs"}],
    ids=["doc1", "doc2"]
)

# Query for similar documents
results = collection.query(
    query_texts=["This is a query document"],
    n_results=2
)

# Access results
if not results:
    print("No results found.")
else:
    for i, doc_id in enumerate(results["ids"][0]):
        print(f"{doc_id}, {results['documents'][0][i]}, {results['distances'][0][i]}")

• Use the vector API with Sentence transformers

import skypydb

# Create a client
client = skypydb.VectorClient(
    embedding_provider="sentence-transformers",
    embedding_model_config={
        "model": "all-MiniLM-L6-v2"
    }
)

# Create a collection
collection = client.get_or_create_collection("my-documents")

# Add documents
collection.add(
    documents=["This is document1", "This is document2"],
    metadatas=[{"source": "notion"}, {"source": "google-docs"}],
    ids=["doc1", "doc2"]
)

# Query for similar documents
results = collection.query(
    query_texts=["This is a query document"],
    n_results=2
)

# Access results
if not results:
    print("No results found.")
else:
    for i, doc_id in enumerate(results["ids"][0]):
        print(f"{doc_id}, {results['documents'][0][i]}, {results['distances'][0][i]}")

Mem0

• use this command to install skypydb and mem0

pip install skypydb[mem0]

from mem0 import Memory

# Local mem0 config
config = {
    "vector_store": {
        "provider": "skypydb",
        "config": {
            "collection_name": "memory",
            "path": "db/_generated/mem0_vector.db"
        }
    },
    "llm": {
        "provider": "ollama",
        "config": {
            "model": "llama3.1:latest",
            "temperature": 0.3,
            "max_tokens": 1024,
            "ollama_base_url": "http://localhost:11434",
        },
    },
    "embedder": {
        "provider": "ollama",
        "config": {
            "model": "mxbai-embed-large"
        }
    }
}

m = Memory.from_config(config)

# Add memories
m.add("I love Python programming", user_id="user1")
m.add("My favorite color is blue", user_id="user1")

# Search memories
results = m.search("What programming language do I like?", user_id="user1")

print(results)

Secure Implementation

• first create an encryption key and a salt key and make them available in the .env.local file don't show those keys to anyone, you can use the Cli to generate those keys

# you can generate a secure encryption key and salt using the cli
# or generate a secure encryption key and salt using the this example code

from skypydb.security import EncryptionManager

# Generate a secure encryption key
encryption_key = EncryptionManager.generate_key()
salt = EncryptionManager.generate_salt()
print(encryption_key) # don't show this key to anyone
print(salt) # don't show this salt to anyone

• Use the encryption key to encrypt sensitive data

import os
import skypydb
from dotenv import load_dotenv

# Load environment variables from .env file
load_dotenv(".env.local")

# Load encryption key from environment
encryption_key = os.getenv("ENCRYPTION_KEY") # create a encryption key and make it available in .env file before using it, don't show this key to anyone
salt_key = os.getenv("SALT_KEY") # create a salt key and make it available in .env file before using it, don't show this salt to anyone

# transform salt key to bytes
if salt_key is None:
    raise ValueError("SALT_KEY missing")
salt_bytes = salt_key.encode("utf-8")

# Create encrypted database
client = skypydb.ReactiveClient(
    encryption_key=encryption_key,
    salt=salt_bytes,
    encrypted_fields=["message"]  # Optional: encrypt only sensitive fields
)

# All operations work the same - encryption is transparent!
tables = client.get_or_create_table()
# if the tables already exists the programe get them instead

# Access your tables
success_table = tables["success"]
warning_table = tables["warning"]
error_table = tables["error"]

# Automatically encrypted
success_table.add(
    component="AuthService",
    action="login",
    message="User logged in successfully", # only this field is encrypted if encrypted_fields is not None
    user_id="user123"
)

# Data is automatically decrypted when retrieved
user_success_logs = success_table.search(
    user_id="user123"
)

if not user_success_logs:
    print("No results found.")
else:
    for user_success_log in user_success_logs:
        print(user_success_log)

Learn more on our Docs

All Thanks To Our Contributors:

License

MIT