SQELDB 0.1.1

A simple ELDB engine for Python

ELDB — Embedded Lightweight DB Engine

Simple file-backed fixed-width record database for small projects and experimentation.

ELDB is a minimal, easy-to-use embedded database implemented in Python. It stores tables as binary files with a small text header and fixed-width records. This project is intended for learning, prototyping, and small single-user applications — not for production use.

Features

• Tiny footprint: each table is a single .eldb file under a database folder
• Fixed-size fields: supports int and fixed-length str columns
• Simple API: create/load tables, insert/select/update/delete rows, add/delete columns
• Primary key index: builds an in-memory index for a declared primary key on load

Installation

No external dependencies — just Python 3.6+.

Clone or copy the repository files to your project folder. The engine writes table files to a folder (default mydb).

Basic Usage

Example session using the ELDB class from ELDB_engine.py:

from SQELDB import ELDB

# Create engine (defaults to folder 'mydb')
db = ELDB(db_folder='ELDB_Data')

# Define table schema: column -> {type, size, nullable}
columns = {
    'id': {'type': 'int', 'size': 4, 'nullable': False},
    'name': {'type': 'str', 'size': 50, 'nullable': False},
    'age': {'type': 'int', 'size': 4, 'nullable': True}
}

# Create table with a primary key
db.create_table('users', columns, primary_key='id', if_not_exists=True)

# Insert a row (note: columns with no value must be nullable or provide default)
db.insert('users', {'id': 1, 'name': 'Alice', 'age': 30})
db.insert('users', {'id': 2, 'name': 'Bob'})

# Select rows (where uses dict: column -> (operator, value))
rows = db.select('users', where={'age': ('>=', 18)}, order_by=('age', False), limit=10)
print(rows)

# Update rows
db.update('users', updates={'age': 31}, where={'id': ('=', 1)})

# Delete rows
db.delete('users', where={'id': ('=', 2)})

# Add a new column
db.add_column('users', 'email', column_type='str', size=100, nullable=True, default='')

# Delete table
# db.delete_table('users')

API Reference

• ELDB(db_folder='mydb')

  • Create engine instance. db_folder is the folder where .eldb files live.

• create_table(name, columns, primary_key=None, if_not_exists=False)

  • columns is a dict mapping column name to options: {'type': 'int'|'str', 'size': <int>, 'nullable': True|False}.
  • primary_key (optional) column name to build an in-memory index on load.

• load_table(table_name)

  • Loads an existing .eldb table and builds the primary key index if present.

• insert(table_name, row)

  • row is a dict of column -> value. Type checks are performed.

• select(table_name=None, where=None, order_by=None, limit=None, columns=None)

  • table_name: choose specific table or None to select across all tables.
  • where: dict of column -> (operator, value), operators: =, !=, >, <, >=, <=.
  • order_by: tuple (column, reverse) where reverse is True for descending.
  • limit: maximum number of rows to return.
  • columns: list of column names to return.

• update(table_name, updates, where=None)

  • updates is a dict of column -> new value.

• delete(table_name, where=None)

  • Remove rows matching where — omit where to remove all rows.

• add_column(table_name, column_name, column_type='str', size=50, nullable=True, default=None)

  • Adds a column to the schema and writes the default value to existing rows.

• delete_column(table_name, column_name)

  • Removes a column from the schema and from existing rows.

• clear_table(table_name)

  • Remove all rows from a table (keeps file and schema).

• delete_table(table_name) and drop_all_tables()

  • Remove files from disk.

Table File Format

Each table is stored as a .eldb file under the database folder. The file begins with a single text header line (a Python dictionary representation) that contains the columns schema, primary_key, and row_count. Following the header, records are stored as fixed-size binary blobs. String columns are fixed-length and padded with NUL bytes; integers are 4-byte little-endian signed integers using Python's struct format i.

Note: The header is written using str() and read with eval() in the current implementation. This is simple but unsafe for untrusted inputs; do not open files from unknown sources.

Limitations & Notes

• Binary fixed-width format: you must choose adequate size for str columns.
• No concurrency control: this engine is not safe for concurrent writers.
• Header uses eval() and thus can execute arbitrary code if the file is tampered with — treat files as untrusted.
• Index is in-memory and rebuilt on load; large tables will use more memory.

Suggested Improvements

• Replace eval() with a safe serializer (JSON) for the header.
• Add transaction/locking and append-mode writes for better concurrency.
• Support variable-length fields or a simple heap for strings to save space.
• Add basic tests and a small CLI for table inspection.

Development

To experiment interactively, run python in the project folder and import the ELDB class.

Contributing

Feel free to open issues or create pull requests with improvements, tests, or bug fixes.

License

This project is provided as-is for learning and prototyping. No explicit license file is included.