Skip to content
DocumentDrivenDXPublic

About

A universal UMF-powered SQL dialect with pluggable backends: Ashlar on Databricks and Truss on PostgreSQL.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Latest commit

 

History

1,038 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Weft

Explore Novirhyza

Weft compiles a request (a UMF document, a storage binding and a query) into SQL for a chosen backend, with bound parameters, result types, decoding instructions and diagnostics. If a selected meaning cannot be preserved, compilation returns an explicit refusal instead of guessing. Weft never connects to a database or executes SQL: execution, authorization and result publication belong to the host.

Weft is a Rust compiler with Python bindings (umf-weft, import weft) and a WebAssembly build. It consumes UMF and lowers to two backends: Truss on PostgreSQL and Ashlar on Databricks.

Status: experimental. The public surface is the JSON request and response contracts in docs/helix/02-design/contracts/.

What CI tests

  • The Rust workspace test suite (cargo test --workspace) on Linux x86_64.
  • The Python wheel, built from this source and installed into a fresh virtualenv: the quickstart below plus the installed-wheel checks listed in .github/workflows/ci.yml.
  • Nightly: backend feature matrices, WebAssembly parity in Chromium, and native PostgreSQL 18.6 table-operation tests (see .github/workflows/ci.yml).
  • Releases build abi3 wheels (Python 3.9 or later) for manylinux x86_64, manylinux aarch64 and macOS arm64, an sdist, and two WebAssembly tarballs. Each wheel is installed into a fresh virtualenv and runs the quickstart before anything is published.

The Truss/PostgreSQL and Ashlar/Databricks backends are compile-only in this repository. Nothing here is tested against a running Databricks warehouse.

Install

pip install umf-weft

Quickstart

git clone https://github.com/DocumentDrivenDX/weft && cd weft
python examples/quickstart.py

examples/quickstart.py loads examples/plain-field.request.json and compiles it:

import json, pathlib, weft

request = json.loads(pathlib.Path(__file__).with_name("plain-field.request.json").read_text())
result = json.loads(weft.compile_truss_runtime_json(json.dumps(request)))
assert result["status"] == "compiled", result
print(result["sql"])

The request is chosen by the host: a UMF document, a Truss runtime binding and the query to compile. The output is one PostgreSQL statement, which begins:

SELECT (("s0".props->'1'->>'text') COLLATE pg_catalog."C")::pg_catalog.text AS "A-label" FROM "truss"."query_objects"($1::pg_catalog.jsonb,1::pg_catalog.int4) AS "s0" LIMIT 101

CI runs this script against a freshly built wheel on every pull request.

Build from source

rustup show                                   # installs the pinned toolchain
cargo test --workspace --locked
pip install maturin==1.9.6
maturin develop --manifest-path crates/weft-python/Cargo.toml
bun install --frozen-lockfile && bun run specs:check && bun run modules:check

More

License

Dual-licensed under MIT OR Apache-2.0, at your option. See LICENSE.md. Third-party materials retain their original notices.

About

A universal UMF-powered SQL dialect with pluggable backends: Ashlar on Databricks and Truss on PostgreSQL.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages