Files
lancedb/python/pyproject.toml
T
Xuanwo c72931b30f feat(python): add TypeSafe reranker (#4209)
Adds `TypeSafeReranker`, which reranks vector, FTS, and hybrid results
with the [TypeSafe System One
API](https://docs.typesafe.ai/introduction).

Each result is scored independently: TypeSafe reads `{"query",
"document"}` and answers a yes/no (noul) question, and the probability
of yes becomes `_relevance_score`. Unlike listwise LLM rerankers, the
score is an absolute probability, so it is comparable across queries and
can be thresholded. The question's `instructions` and `true`/`false`
`criteria` are configurable, since domain-specific criteria are what
make this kind of scoring work well ([TypeSafe's re-ranking
cookbook](https://docs.typesafe.ai/cookbooks/rerank_typesafe)).

The API takes one state per request, so the reranker sends one request
per result on a thread pool bounded by `max_concurrency`. It
deliberately does not use the background event loop: rerankers are
called synchronously from inside the async query APIs, where `LOOP.run`
would deadlock.

TypeSafe scores for the same pair vary slightly between calls, so
results with close scores can swap places when a search is repeated. The
shared reranker test helper now takes `deterministic=False` for this
case: it still checks result sizes and descending scores, but not that
two identical searches return the same order.

The SDK is imported only when the client is created and questions are
sent as plain dicts, so the new tests run in CI with a fake client and
without `typesafe-sdk` installed. The live-API test is skipped without
`TYPESAFE_API_KEY`. Ranking quality has not been compared with other API
rerankers.
2026-09-17 15:52:34 +08:00

152 lines
4.2 KiB
TOML

[project]
name = "lancedb"
# version in Cargo.toml
dynamic = ["version"]
dependencies = [
"deprecation>=2.1.0",
"numpy>=1.24.0",
"overrides>=0.7; python_version<'3.12'",
"packaging>=23.0",
"pyarrow>=16",
"pydantic>=2.7.4,<3",
"tqdm>=4.27.0",
"lance-namespace>=0.3.2"
]
description = "lancedb"
authors = [{ name = "LanceDB Devs", email = "dev@lancedb.com" }]
license = { file = "LICENSE" }
readme = "README.md"
requires-python = ">=3.10"
keywords = [
"data-format",
"data-science",
"machine-learning",
"arrow",
"data-analytics",
]
classifiers = [
"Development Status :: 3 - Alpha",
"Environment :: Console",
"Intended Audience :: Science/Research",
"License :: OSI Approved :: Apache Software License",
"Operating System :: OS Independent",
"Programming Language :: Python",
"Programming Language :: Python :: 3",
"Programming Language :: Python :: 3 :: Only",
"Programming Language :: Python :: 3.10",
"Programming Language :: Python :: 3.11",
"Programming Language :: Python :: 3.12",
"Programming Language :: Python :: 3.13",
"Topic :: Scientific/Engineering",
]
[project.urls]
repository = "https://github.com/lancedb/lancedb"
[project.optional-dependencies]
pylance = [
"pylance>=5.0.0b5",
]
# A library only needs the OpenTelemetry API; the application supplies and
# configures the SDK (the actual exporter/reader). See
# https://opentelemetry.io/docs/languages/python/instrumentation/
otel = ["opentelemetry-api"]
tests = [
"aiohttp>=3.9.0",
"boto3>=1.28.57",
"pandas>=1.4",
"pytest>=7.0",
"pytest-mock>=3.10",
"pytest-asyncio>=0.21",
"duckdb>=0.9.0",
"pytz>=2023.3",
"polars>=0.19, <=1.32.3",
"pyarrow<25",
"pyarrow-stubs>=16.0",
"pylance==9.0.0",
"requests>=2.31.0",
"datafusion>=54,<55",
"opentelemetry-sdk>=1.30.0",
]
dev = [
"ruff>=0.3.0",
"pre-commit>=3.5.0",
"pyright>=1.1.350",
'typing-extensions>=4.0.0; python_version < "3.11"',
]
docs = ["mkdocs", "mkdocs-jupyter", "mkdocs-material", "mkdocstrings-python"]
clip = ["torch", "pillow>=12.1.1", "open-clip-torch"]
siglip = ["torch", "pillow>=12.1.1", "transformers>=4.41.0","sentencepiece"]
embeddings = [
"requests>=2.31.0",
"openai>=1.6.1",
"sentence-transformers>=2.2.0",
"torch>=2.0.0",
"pillow>=12.1.1",
"open-clip-torch>=2.20.0",
"cohere>=4.0",
"colpali-engine>=0.3.10",
"huggingface_hub>=0.19.0",
"InstructorEmbedding>=1.0.1",
"google-genai>=1.0.0",
"boto3>=1.28.57",
"awscli>=1.44.38",
"botocore>=1.31.57",
'ibm-watsonx-ai>=1.1.2; python_version >= "3.10"',
"ollama>=0.3.0",
"sentencepiece>=0.1.99"
]
azure = ["adlfs>=2024.2.0"]
[tool.maturin]
python-source = "python"
module-name = "lancedb._lancedb"
# uv installs the project as an editable package before `uv run`, so keep that
# bootstrap build consistent with `maturin develop`.
editable-profile = "dev"
[build-system]
requires = ["maturin>=1.10"]
build-backend = "maturin"
[tool.ruff.lint]
select = ["F", "E", "W", "G", "PERF"]
[tool.pytest.ini_options]
addopts = "--strict-markers --ignore-glob=lancedb/embeddings/*.py"
markers = [
"slow: marks tests as slow (deselect with '-m \"not slow\"')",
"asyncio",
"s3_test",
]
[tool.pyright]
include = [
"python/lancedb/index.py",
"python/lancedb/rerankers/util.py",
"python/lancedb/rerankers/__init__.py",
"python/lancedb/rerankers/voyageai.py",
"python/lancedb/rerankers/typesafe.py",
"python/lancedb/rerankers/jinaai.py",
"python/lancedb/rerankers/openai.py",
"python/lancedb/rerankers/cross_encoder.py",
"python/lancedb/rerankers/colbert.py",
"python/lancedb/rerankers/answerdotai.py",
"python/lancedb/rerankers/cohere.py",
"python/lancedb/arrow.py",
"python/lancedb/__init__.py",
"python/lancedb/types.py",
"python/lancedb/integrations/__init__.py",
"python/lancedb/exceptions.py",
"python/lancedb/background_loop.py",
"python/lancedb/schema.py",
"python/lancedb/sql.py",
"python/lancedb/remote/__init__.py",
"python/lancedb/remote/errors.py",
"python/lancedb/embeddings/__init__.py",
"python/lancedb/_lancedb.pyi",
"python/type_tests/connect.py",
]
exclude = ["python/tests/"]
pythonVersion = "3.13"