Skip to main content

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 node_modules/.git defaults 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

Release files for dirsql 0.4.49

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for dirsql 0.4.49
File Size Uploaded
dirsql-0.4.49.tar.gz 484.5 kB Details

Built distributions (wheels)

Table of built distributions (wheels) for dirsql 0.4.49
File
dirsql-0.4.49-cp310-abi3-win_amd64.whl CPython 3.10 abi3 Windows x86-64 Details
dirsql-0.4.49-cp310-abi3-manylinux_2_28_x86_64.whl CPython 3.10 abi3 Linux glibc 2.28+ x86-64 Details
dirsql-0.4.49-cp310-abi3-manylinux_2_28_aarch64.whl CPython 3.10 abi3 Linux glibc 2.28+ ARM64 Details
dirsql-0.4.49-cp310-abi3-macosx_11_0_arm64.whl CPython 3.10 abi3 macOS 11.0+ ARM64 Details
dirsql-0.4.49-cp310-abi3-macosx_10_12_x86_64.whl CPython 3.10 abi3 macOS 10.12+ x86-64 Details

Total release size: 15.9 MB

Release files / dirsql-0.4.49.tar.gz

Download URL dirsql-0.4.49.tar.gz
Size 484.5 kB
Tags Source
SHA-256 checksum
How to use checksums
e99fed7d9ca197c911e562c12bcb28de64badfe1cd08481452efdc7e75fe00ee
BLAKE2b-256 checksum
How to use checksums
0190a734498ff55b3f6d43a5bfad90c39c5a338e3df77e2715e861a0bef1a6e1
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Sep 18, 2026.

Transparency log

Release files / dirsql-0.4.49-cp310-abi3-win_amd64.whl

Download URL dirsql-0.4.49-cp310-abi3-win_amd64.whl
Size 3.2 MB
Tags CPython 3.10 Windows x86-64 abi3
SHA-256 checksum
How to use checksums
121bdfc1d5ef3aa1d47a2cd46ca9f1f08496c710246027ac86532e74301ff198
BLAKE2b-256 checksum
How to use checksums
1013fad9ac58dce9e63e7f2087d6a79241ecf9d45f937a3fbc5c897ac5face0f
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Sep 18, 2026.

Transparency log

Release files / dirsql-0.4.49-cp310-abi3-manylinux_2_28_x86_64.whl

Download URL dirsql-0.4.49-cp310-abi3-manylinux_2_28_x86_64.whl
Size 3.2 MB
Tags CPython 3.10 Linux glibc 2.28+ x86-64 abi3
SHA-256 checksum
How to use checksums
14d6a7a6a5236a7df4e6ed881eae5f984df3f848ea3c517b2eb67a64c13eb5bf
BLAKE2b-256 checksum
How to use checksums
2a3a641ec8a91f81ea42510d6d268dda241114df4b7a1b454a129c97b1b2adff
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Sep 18, 2026.

Transparency log

Release files / dirsql-0.4.49-cp310-abi3-manylinux_2_28_aarch64.whl

Download URL dirsql-0.4.49-cp310-abi3-manylinux_2_28_aarch64.whl
Size 3.1 MB
Tags CPython 3.10 Linux glibc 2.28+ ARM64 abi3
SHA-256 checksum
How to use checksums
250221a7fe4cd83296fbeff2062bc084546b0c0d1ba8a8293fe22c39019db814
BLAKE2b-256 checksum
How to use checksums
c0d72a01fb96dbc37773293358e232ad70565e78243ccd854294886e2f606ad1
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Sep 18, 2026.

Transparency log

Release files / dirsql-0.4.49-cp310-abi3-macosx_11_0_arm64.whl

Download URL dirsql-0.4.49-cp310-abi3-macosx_11_0_arm64.whl
Size 2.9 MB
Tags CPython 3.10 abi3 macOS 11.0+ ARM64
SHA-256 checksum
How to use checksums
f102460eed090ade60866fd92b86173ebcb97c309827d460365db5743204f495
BLAKE2b-256 checksum
How to use checksums
970d37abc518edf27bd6a781493adb6e9afa0da97a02da118d3af167c83cf1e6
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Sep 18, 2026.

Transparency log

Release files / dirsql-0.4.49-cp310-abi3-macosx_10_12_x86_64.whl

Download URL dirsql-0.4.49-cp310-abi3-macosx_10_12_x86_64.whl
Size 3.1 MB
Tags CPython 3.10 abi3 macOS 10.12+ x86-64
SHA-256 checksum
How to use checksums
980cc1def5d646ac0bd5659d942e1b12fb9057ae85b6577a75e3939da96bd65c
BLAKE2b-256 checksum
How to use checksums
b43cf41f624fc253e585b0eb2bf776fb8dcffb449431d28d0ddae4bcbc4d165a
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.14

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Sep 18, 2026.

Transparency log

Release history Release notifications | RSS feed

0.4.56

6 release files

0.4.55

6 release files

0.4.54

6 release files

0.4.53

6 release files

0.4.52

6 release files

0.4.51

6 release files

0.4.50

6 release files

This release

0.4.49 This release

6 release files

0.4.48

6 release files

0.4.47

6 release files

0.4.46

6 release files

0.4.45

6 release files

0.4.44

6 release files

0.4.43

6 release files

0.4.42

6 release files

0.4.41

6 release files

0.4.40

6 release files

0.4.30

6 release files

0.4.29

6 release files

0.4.28

6 release files

0.4.27

6 release files

0.4.26

6 release files

0.4.25

6 release files

0.4.24

6 release files

0.4.23

6 release files

0.4.22

6 release files

0.4.21

6 release files

0.4.20

6 release files

0.4.19

6 release files

0.4.18

6 release files

0.4.17

6 release files

0.4.16

6 release files

0.4.9

6 release files

0.4.8

6 release files

0.4.7

6 release files

0.4.6

6 release files

0.4.5

6 release files

0.4.4

6 release files

0.4.3

6 release files

0.4.2

6 release files

0.4.1

6 release files

0.4.0

6 release files

0.3.99

6 release files

0.3.98

6 release files

0.3.97

6 release files

0.3.96

6 release files

0.3.95

6 release files

0.3.94

6 release files

0.3.93

6 release files

0.3.92

6 release files

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page