sbol-db 0.1.0

Pure-Python client for the sbol-db HTTP API, with a PartShop-compatible facade.

sbol-db

A pure-Python client for the sbol-db HTTP API, plus a PartShop-compatible facade for code migrating from SynBioHub's pysbol2 PartShop.

The client depends only on httpx. SBOL documents cross the wire as RDF strings; structured records come back as typed dataclasses. There is no SBOL object model and no dependency on pysbol2/pysbol3 or sbol-rs; SBOL2 to SBOL3 conversion happens on the server via the version parameter.

Install

pip install sbol-db
# optional: rdflib, for navigating pulled RDF as a graph
pip install "sbol-db[rdf]"

The import name is sbol_db:

from sbol_db import SbolDbClient, PartShop

Broad client

from sbol_db import SbolDbClient

db = SbolDbClient("http://localhost:8888")

# import a document, search, pull, delete
db.create_graph(open("part.ttl").read(), format="turtle",
                document_iri="https://ex.org/part")
for obj in db.search("pLac", object_type="Component"):
    print(obj.display_id, obj.iri)

rdf = db.export_rdf("https://ex.org/pLac", version="sbol2")   # server converts
rows = db.sparql("SELECT ?s WHERE { ?s ?p ?o } LIMIT 10").bindings()
db.delete_graph_by_document_iri("https://ex.org/part")

PartShop facade

from sbol_db import PartShop

shop = PartShop("http://localhost:8888")
shop.login("dba", "dba")                       # HTTP Basic; getKey() is empty
rdf = shop.pull("https://ex.org/pLac")         # RDF text
shop.submit(rdf, collection="https://ex.org/myCollection", overwrite=1)
print(shop.searchCount("pLac"))

Attachments (attachFile, downloadAttachment) raise NotImplementedError: they are out of scope for sbol-db.

Development

The environment is managed by uv; the dev tools live in a dev dependency group that uv run installs automatically.

make test    # unit tests only (uv run pytest -m "not e2e")
make e2e     # build the sbol-db server, then run the full suite against it
make lint    # isort + black + flake8 + mypy

make e2e builds a fresh sbol-db binary and points the test fixtures at it, so the end-to-end tests never run against a stale build. They boot a real server on a throwaway SQLite database (no external services). To also run them against Postgres:

docker compose up -d postgres          # from the repo root
make e2e SBOL_DB_TEST_BACKENDS=sqlite,postgres

Without a discoverable binary the e2e tests skip, so uv run pytest still runs the unit suite anywhere.

Publishing

The package is built and published with uv. Bump version in pyproject.toml and in src/sbol_db/__init__.py, then:

make build           # sdist + wheel into dist/
make check           # twine check dist/*
make publish         # upload to PyPI (needs a token, see below)

make publish reads a PyPI API token from the environment:

export UV_PUBLISH_TOKEN=pypi-...
make publish

To dry-run against TestPyPI first:

uv publish --publish-url https://test.pypi.org/legacy/ dist/*