pyproject.toml

[project]
name = "raincloud"
dynamic = ["version"]
description = "Client-reproducible pipeline for building a curated catalog of public datasets as Parquet + Vortex files."
readme = "README.md"
requires-python = ">=3.11"
authors = [{ name = "Raincloud Maintainers", email = "raincloud@spiraldb.com" }]
license = "Apache-2.0"
license-files = ["LICENSE"]
keywords = ["datasets", "parquet", "vortex", "arrow", "benchmark", "data-catalog", "columnar"]
classifiers = [
    "Development Status :: 4 - Beta",
    "Intended Audience :: Developers",
    "Intended Audience :: Science/Research",
    "Operating System :: POSIX :: Linux",
    "Programming Language :: Python :: 3",
    "Programming Language :: Python :: 3 :: Only",
    "Programming Language :: Python :: 3.11",
    "Topic :: Database",
    "Topic :: Scientific/Engineering",
]

dependencies = [
    "pyarrow>=23.0",
    "numpy>=2.0",
    "fsspec>=2024.0",
    "platformdirs>=4.0",
]

[project.urls]
Homepage = "https://github.com/spiraldb/raincloud"
Repository = "https://github.com/spiraldb/raincloud"
Issues = "https://github.com/spiraldb/raincloud/issues"
Changelog = "https://github.com/spiraldb/raincloud/blob/develop/CHANGELOG.md"

[project.scripts]
raincloud = "raincloud.cli:main"

[project.optional-dependencies]
# Exact: the Vortex file format is still versioned by its writer, and the
# snapshot's checksums are recorded against this one.
vortex = ["vortex-data==0.86.1"]
s3 = ["s3fs>=2024.0"]
http = ["aiohttp>=3.9"]
pandas = ["pandas>=2.0"]
# `raincloud.duckdb_connect()`, the one way raincloud code and examples open DuckDB.
duckdb = ["duckdb>=1.5.0"]
# The build pipeline core; local builds are explicitly requested. A handler's
# format-specific dependency is its own extra, imported only when that handler
# runs; a build missing one names the extra.
build = ["raincloud[vortex,duckdb]", "zstandard>=0.25.0", "jsonschema>=4.0"]
osm = ["osmium>=4.3"]
sas = ["pyreadstat>=1.3"]
excel = ["openpyxl>=3.1"]
archives = ["py7zr>=1.1.0", "unlzw3>=0.2.0"]
# Exact: generated datasets are recorded by checksum, and a generator's output
# changes with its version (DuckDB's tpch/tpcds extensions, tpchgen-cli).
generated = ["raincloud[build]", "duckdb==1.5.5", "tpchgen-cli==3.0.0"]
kaggle = ["kaggle>=2.0"]
huggingface = ["huggingface-hub>=0.25"]
tui = ["textual>=0.80"]
dev = ["pytest>=8.0", "ruff>=0.13"]
all = [
    "raincloud[s3,http,pandas,build,osm,sas,excel,archives,generated,kaggle,huggingface,tui]",
]

[tool.ruff]
line-length = 120
# The vendored submodule carries its own tooling and lint rules.
extend-exclude = ["sidecars/java/parquet-arrow-java"]

[tool.ruff.lint]
# Conservative starter set: pyflakes (F), pycodestyle errors + warnings
# (E, W), and isort (I) for import ordering. Add B/UP/SIM later once the
# baseline is clean.
select = ["E", "F", "W", "I"]
# E501 line-too-long: `line-length = 120` gives headroom; rare comment-heavy
# lines shouldn't fail CI.
# E701 / E702 multiple-statements-on-one-line: deliberate column-aligned
# `if cond: return ...` lookup tables (see e.g. browse.py:219-222) read
# better as one-liners than split across multiple lines.
ignore = ["E501", "E701", "E702"]

[tool.pytest.ini_options]
markers = [
    "wheel: builds the wheel + spins venvs (slow; gated by --run-wheel)",
    "network: hits real upstream sources (flaky; gated by --run-network)",
]

[build-system]
requires = ["hatchling"]
build-backend = "hatchling.build"

# The release version has ONE literal, in `raincloud/__init__.py`. This table
# and the Java client's `version` read it; `clients/rust/Cargo.toml` and
# `CITATION.cff` are hand-bumped copies, checked by
# tests/test_loader_package.py::test_version_mirrors_agree.
[tool.hatch.version]
path = "raincloud/__init__.py"

[tool.hatch.build.targets.wheel]
packages = ["raincloud"]

[tool.hatch.build.targets.wheel.force-include]
# Package the current schema_version's snapshot — the loader derives its
# catalog version from the snapshot it reads, so a wheel install (no checkout)
# must ship the v2 snapshot to expose the canonical arrow format + request the
# v2 mirror keys `publish` actually writes.
"docs/v2/snapshot.json" = "raincloud/_data/snapshot.json"
"sources.json" = "raincloud/_data/sources.json"
"sources.schema.json" = "raincloud/_data/sources.schema.json"

# Python source releases contain code and operational docs only. Private worktree notes,
# local environments, artifacts, native build trees and caches never enter them.
[tool.hatch.build.targets.sdist]
include = [
    "/raincloud", "/examples", "/templates",
    "/docs/v2/snapshot.json", "/sources.json", "/sources.schema.json", "/sources.schema.md",
    "/README.md", "/LICENSE", "/DISCLAIMER.md", "/CHANGELOG.md", "/pyproject.toml",
]
exclude = ["**/build/**", "**/target/**", "**/.gradle/**", "**/__pycache__/**", "**/*.pyc"]