Overview

Python wrapper to DataWarehouse Connect.

• Free software: ISC license

Installation

pip install dwclib (add the dwcmeta extra to search patients in DWCmeta, the dask extra for dask support). Python 3.13 or newer.

The MCP server that used to ship here has moved to laribdb_mcp.

Development

The project is managed with uv (uv_build backend). Linting and formatting moved from nox, flake8 and darglint to ruff; type checking uses ty.

uv sync
uv run ruff check && uv run ruff format --check
uv run ty check
uv run pytest
uv run --group docs sphinx-build -W docs docs/_build
uvx pip-audit

Changelog

• Unreleased

  • pandas 3 (no dask by default; dask is an optional extra), Python 3.13+
  • read_patients filters and returns the NDA (nda=, encounterid)
  • Config is read at call time, so update_config() takes effect at once
  • Move the MCP server to laribdb_mcp
  • Replace nox, flake8 and darglint with ruff and ty; build with uv_build
  • SQLAlchemy below 2.0.42 (newer MSSQL reflection drops DWC columns)

• 2026.07.11

  • Update documentation, add MCP server, update dependencies, move from poetry to uv

• 2024.12.27

  • Support updating the config file through the API

• 2024.9.17

  • Support reading waves in binary chunks without conversion

• 2024.5.10

  • Support searching patients directly in DWC without DWCmeta

• 2024.4.4

  • Support querying enumerations
  • Update dependencies

• 2023.6.21

  • Remove all occurences of naive datetime since dask now support tz-aware
  • New config file syntax to be compatible with other libraries
  • Fix a bug when Pleth waveform return all NaN

• 2023.6.7

  • Support for querying alerts using the read_alerts call
  • Use the platformdirs package for config file location

• 2022.9.14

  • Support numeric labels and sublabels in read_patients and read_numerics
  • Support to query for multiple patients at once in read_numerics

• 2022.6.23

  • Convert packaging from flit to poetry
  • Add linting and testing with nox, flake8 and safety
  • Create scaffolding for future Sphinx documentation
  • Fix a number of bugs in corner cases (division by zero, ...)
  • Add a generic Dask wrapper to run custom DWC queries with Dask

• 2022.3.22

  • Convert packaging from old-style setup.py to flit
  • Refactor: extract common code between dask and pandas version
  • No longer relies on user defined function in the database
  • Patients: add read_patient function to fetch a single patient
  • Numerics: read_numerics patientids can be a list or a str. When it is a list, a MultiIndex is returned