stitch-proj 0.1.1

Some tools for building a Translator BigKG. This software project is experimental and unfinished.

stitch-proj

Some tools for building a Translator BigKG.
This software project is experimental and under active development.

---

Installation

From PyPI

pip install stitch-proj

For development

pip install stitch-proj[dev]

From source

git clone https://github.com/Translator-CATRAX/stitch-proj.git
cd stitch-proj
pip install -e .[dev]

---

Overview

There are two primary intended users of stitch-proj:

1. Ingester A developer who wants to ingest the Babel concept identifier normalization database into a local SQLite database.

2. Querier A developer building an application (e.g., a BigKG build system) who wants to programmatically query a local Babel SQLite database.

---

Package Structure

This project uses a src/ layout:

src/
  stitch_proj/
    ingest_babel.py
    local_babel.py
    row_counts.py
    stitchutils.py

Import the package as:

import stitch_proj

---

Tools

• stitch_proj.ingest_babel Downloads and ingests the Babel database into a local SQLite database.

• stitch_proj.local_babel Provides functions for querying a local Babel SQLite database.

• stitch_proj.row_counts Prints table row counts for a local Babel SQLite database.

---

Running the Ingest

After installation, the console script is available:

ingest-babel --help

Or invoke via module:

python -m stitch_proj.ingest_babel --help

A full ingest requires:

• CPython 3.12
• At least 32 GiB RAM
• ~600 GiB temporary disk space
• ~200 GiB for the final SQLite database

A full ingest may take 30–40 hours depending on hardware.

---

Downloading a Pre-Built Babel Database

A pre-built SQLite file is available from S3:

https://rtx-kg2-public.s3.us-west-2.amazonaws.com/babel-20250331-p1.sqlite

Place it in a directory such as:

db/babel.sqlite

You can then use stitch_proj.local_babel to query it.

---

Running Tests

Ensure a valid babel.sqlite file exists locally, then run:

pytest -v

Some tests require internet connectivity.

---

Systems Tested

ingest_babel.py has been tested on:

• Ubuntu 24.04 (x86_64, Intel Xeon)
• Ubuntu 24.04 (ARM64, AWS Graviton3)
• macOS 14 (Apple Silicon)

The package is pure Python and platform-independent, but large ingests require substantial memory and storage.

---

Development Workflow

Run linting, typing, and tests with:

pytest
ruff check .
mypy src

Or install development dependencies:

pip install -e .[dev]

---

License

MIT License. See LICENSE.

---

Citation

Please see the Babel project's CITATION.cff:

https://github.com/TranslatorSRI/Babel/blob/master/CITATION.cff