pyproject.toml

[build-system]
requires = ["setuptools>=68", "wheel"]
build-backend = "setuptools.build_meta"

[project]
name = "scidk"
version = "0.0.1"
description = "SciDK - Science Data Kit (MVP)"
authors = [{name = "SciDK Team"}]
readme = "README.md"
requires-python = ">=3.12"
dependencies = [
  "Flask>=3.0",
  # scidk/app.py imports this at module scope, before anything else — every
  # test module that imports create_app fails collection without it. CI
  # installs `-e .[dev]`, which reads this list and not requirements.txt, so
  # declaring it in the latter alone left CI unable to import the app at all.
  "python-dotenv>=1.0",
  "ijson>=3.2",
  "openpyxl>=3.1",
  "PyYAML>=6.0",
  "neo4j>=5.14",
  "psutil>=5.9",
  "python-dateutil>=2.8",
  "cryptography>=41.0",
  "jsonpath-ng>=1.6",
  "pandas>=2.0",
  "rapidfuzz>=3.0",
  "bcrypt>=4.0",
  "APScheduler>=3.10",
  # APScheduler's SQLAlchemyJobStore: the persistent jobstore that lets a
  # schedule change made in a worker reach the scheduler in the gunicorn master.
  "SQLAlchemy>=1.4",
  # Validates Pipeline mapping configs against scidk/pipeline/mapping_schema.json.
  "jsonschema>=4.0",
  "flasgger>=0.9.7",
  # Parses incoming RO-Crates in scidk/rocrate_bridge.py:ingest_crate(). Reading
  # only; the export path there is hand-rolled in-memory JSON.
  "rocrate>=0.15.0",
]

[project.optional-dependencies]
# Dev/test dependencies used locally and in CI
dev = [
  "pytest>=7.4",
  "pytest-playwright==0.4.3",
  "playwright==1.40.0",
  "requests>=2.32",
  "beautifulsoup4>=4.12",
  "coverage>=7.4",
]

# Optional LLM provider (only needed if you enable Graphrag LLM features)
llm = [
  "ollama>=0.3",
]

# Whole-slide image metadata (svs_interpreter). Only the first page's
# ImageDescription tag is read, never pixel data. Absent, SVSInterpreter
# returns a stub with a warning rather than failing the enrichment run — so
# this is genuinely optional, but no HistologySlide nodes appear without it.
imaging = [
  "tifffile>=2024.1",
]

[project.scripts]
scidk-serve = "scidk.app:main"
# Concept Graph seeding and clean re-initialisation (Cycle 8 Task C). Also
# reachable as `python -m scidk.concept_graph`.
scidk-concept-graph = "scidk.concept_graph.cli:main"

[tool.setuptools]
package-dir = {"" = "."}

[tool.setuptools.package-data]
# intents.yaml is what `scidk-concept-graph seed` reads, so it has to be in the
# wheel. Until Cycle 8 scidk/concept_graph/ had no __init__.py and was not
# collected as a package at all. Note that this declares data for this package
# only — templates, static JS and interpreter profiles are still undeclared.
"scidk.concept_graph" = ["*.yaml", "*.cypher"]

[tool.setuptools.packages.find]
include = ["scidk*"]
exclude = ["dev*", "data*", "singularity*", "scripts*", "neo4j*", "tests*", "*.dist-info*", "*.egg-info*"]

[tool.pytest.ini_options]
testpaths = [
  "tests",
]
addopts = "-ra"
markers = [
  "unit: fast, pure unit tests",
  "integration: tests that touch DB/files/HTTP without browser",
  "e2e: end-to-end Playwright/browser tests",
  "slow: slow tests",
  "smoke: short, high-signal tests",
]