dirsql (Python SDK)

Ephemeral SQL index over a local directory. dirsql watches a filesystem, ingests structured files into an in-memory SQLite database, and exposes a SQL query interface -- the filesystem is always the source of truth.

Documentation

Also available as dirsql on crates.io and dirsql on npm.

Installation

pip install dirsql

Requires Python >= 3.12. Ships as a native extension (Rust via PyO3); prebuilt binary wheels are provided for common platforms.

Quick start

DirSQL is async by default: the constructor returns immediately, scanning runs in a background thread, and you await db.ready() before querying. Each table is a (ddl, glob, extract) triple: the DDL defines the SQLite schema, the glob selects files (relative to the root), and extract turns a matched file into a list of row dicts. dirsql does not read file contents -- if extract needs the file body it reads path itself; return an empty list to skip a file.

import asyncio
import json
from dirsql import DirSQL, Table

async def main():
    db = DirSQL(
        "./my-blog",
        tables=[
            Table(
                ddl="CREATE TABLE posts (title TEXT, author TEXT)",
                glob="posts/*.json",
                extract=lambda path: [json.loads(open(path, encoding="utf-8").read())],
            ),
        ],
    )
    await db.ready()

    posts = await db.query("SELECT * FROM posts WHERE author = 'alice'")
    print(posts)

asyncio.run(main())

Multiple tables and joins

db = DirSQL(
    "./my-blog",
    tables=[
        Table(
            ddl="CREATE TABLE posts (title TEXT, author_id TEXT)",
            glob="posts/*.json",
            extract=lambda path: [json.loads(open(path, encoding="utf-8").read())],
        ),
        Table(
            ddl="CREATE TABLE authors (id TEXT, name TEXT)",
            glob="authors/*.json",
            extract=lambda path: [json.loads(open(path, encoding="utf-8").read())],
        ),
    ],
)
await db.ready()

results = await db.query("""
    SELECT posts.title, authors.name
    FROM posts JOIN authors ON posts.author_id = authors.id
""")

Ignoring files

Pass ignore patterns to skip files during scanning and watching:

db = DirSQL(
    "./my-blog",
    ignore=["**/drafts/**", "**/.git/**"],
    tables=[...],
)

Path-table scans (SELECT ... FROM './') respect .gitignore files by default. Pass no_ignore=True to restore the full walk; the built-in .git default and any ignore patterns still apply:

db = DirSQL("./my-blog", no_ignore=True)

Loading SQLite extensions

Pass extensions to load SQLite extension shared libraries onto the connection at startup (before any CREATE TABLE). Each entry is a dict with a path and an optional entrypoint init-symbol override:

db = DirSQL(
    "./my-blog",
    tables=[...],
    extensions=[
        {"path": "./ext/vec0.dylib", "entrypoint": "sqlite3_vec_init"},
        {"path": "./ext/myext.so"},  # entrypoint derived from the filename
    ],
)
await db.ready()

# The extension's functions are now callable in queries:
rows = await db.query("SELECT vec_version() AS v")

dirsql enables extension loading only while loading the configured libraries, then disables it again, so the SQL load_extension() function is never exposed to your queries. Programmatic entries load first, followed by any [[dirsql.extension]] entries declared in a config file. See the config reference.

Watching for changes

db.watch() returns an async iterator of row-level change events as files change on disk:

async for event in db.watch():
    print(f"{event.action} on {event.table}: {event.row}")
    if event.action == "error":
        print(f"  error: {event.error}")

Each event has .action ("insert", "update", "delete", or "error"), .table, .row (the new row, or the deleted row on delete), .old_row (the previous row, on update), .file_path, and .error (on error).

CLI

pip install dirsql also installs a dirsql console script. dirsql "<sql>" (or uvx dirsql "<sql>") runs one query and prints the rows as JSON — the default. dirsql server starts an HTTP server exposing the SDK over HTTP: POST /query for SQL and GET /events for a Server-Sent Events change stream. See the CLI reference.

License

MIT