pyproject.toml

[project]
name = "kazu"
dynamic = ["version"]
description = "Biomedical Named Entity Recognition and Entity Linking for Enterprise use cases"
readme = "README.md"
requires-python = ">=3.9"
authors = [{ name = "AstraZeneca AI and Korea University" }]
classifiers = [
  "Programming Language :: Python :: 3 :: Only",
  "License :: OSI Approved :: Apache Software License",
  "Typing :: Typed",
  "Operating System :: OS Independent",
  "Intended Audience :: Healthcare Industry",
  "Natural Language :: English",
  "Topic :: Scientific/Engineering :: Medical Science Apps.",
]
dependencies = [
  # pinning spacy because larger versions cause dependcy issues with thinc and various packages on 7/10/2024
  # Also needs to be at least 3.4.2 on python 3.11 (see https://github.com/explosion/spaCy/issues/12697)
  "spacy==3.7.0",
  # note to kazu devs: if the torch version changes in future,
  # we need to change for CI jobs where we specify this install
  # elsewhere because of the CPU wheel needing a different pypi index.
  # search the project for --index-url https://download.pytorch.org/whl/cpu
  # and/or torch>=2.0.0
  "torch>=2.0.0",
  "transformers>=4.0.0",
  "rdflib>=6.0.0",
  "requests>=2.20.0",
  "hydra-core>=1.3.0",
  "pandas",
  "pyarrow",
  "pyahocorasick",
  "pymongo>=4.3.3",
  "rapidfuzz>=1.0.0",
  "scikit-learn>=0.24.0",
  # scipy 1.12.0 introduced many changes to the sparse matrices api. https://docs.scipy.org/doc/scipy/reference/sparse.html#module-scipy.sparse
  # This is causing our acceptance tests to fail. Pinning to <1.12.0 until it's confirmed other libraries (e.g. sk-learn) don't have issues.
  "scipy<1.12.0",
  "regex>=2020.1.7",
  "psutil>=5.3.0",
  "cachetools>=5.2.0",
  "diskcache>=5.4.0",
  "cattrs>=23.2.0",
  "tqdm",
]

[project.urls]
documentation = "https://astrazeneca.github.io/KAZU/index.html"
repository = "https://github.com/AstraZeneca/KAZU"
changelog = "https://astrazeneca.github.io/KAZU/changelog.html"

[project.optional-dependencies]
webserver = [
  # the behaviour of declaring 'examples' for a FastAPI parameter changed in 0.99.0
  # in a way that broke our docs.
  # A new way was added in 0.103.0 , and we can handle both versions older than 0.99.0 and 0.103.0 or newer.
  # However, releases in between don't work.
  #  "fastapi!=0.99.0,!=0.99.1,!=0.100.0b1,!=0.100.0b2,!=0.100.0b3,!=0.100.0,!=0.100.1,!=0.101.0,!=0.101.1,!=0.102.0",
  # UPDATE: 0.109.0 also no longer works. It seems like there's some incompatibility with ray serve here, so pinning to
  # last known good version
  "fastapi==0.108.0",
  "pydantic<2.0",      # for the pinned spacy version pydantic should be > 1.8 as well.
  "ray[serve]>=2.0.0",
  "PyJWT>=2.0.0",
]
llm = [
  "google-cloud-aiplatform", # Required for vertex API
  "openai",                  # Required for OpenAI API
]
typed = [
  # version 2.31.0.3 introduced
  # overly strict typing of requests.HttpError
  # which was fixed in 2.31.0.6
  "types-requests",
  "types-cachetools",
  "types-regex",
  "types-psutil",
  "types-docutils",
  "pandas-stubs>=2.0.0",
  "types-tqdm",
]
all-steps = [
  "py4j>=0.10.9",
  "rdkit>=2023.3.1",
  "stanza>=1.0.0,<1.6.0", # We're seeing IT test failures on linux with stanza 1.6.0
  "gliner==0.1.7",        # GLINER is under active development, so pinning version until API stabilises
]
dev = [
  "kazu[webserver,typed,all-steps,llm]",
  "black~=24.0",
  "blacken-docs",
  "flake8",
  "mypy",
  "vulture",
  "bump2version",
  # docformatter breaks pre-commit on version 4.0. Need to pin this until this issue is resolved: https://github.com/PyCQA/docformatter/issues/289
  "pre-commit<4.0.0",
  # docformatter breaks pre-commit on version 4.0. Need to pin this until this issue is resolved: https://github.com/PyCQA/docformatter/issues/289
  "pre-commit<4.0.0",
  "pytest",
  "pytest-mock",
  "pytest-cov",
  "pytest-timeout",
  "hypothesis",
  "sphinx>=7.2,<8.0", # 8.0.0 leads to strange SSL errors, both locally and on the runner",
  "myst_parser",
  "furo>=2023.08.17",
  # to allow profiling
  # of the steps.
  "tensorboard",
  # towncrier versions older than this are broken for
  # importlib-resources version 6 or newer
  "towncrier>=23.10.0",
  "ray>=1.10.0",
  # required for parsing wikimedia data for disambiguation
  'mwparserfromhell',
  # required for krt
  'streamlit',
]

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

[tool.hatch.build]
include = ["/kazu"]
exclude = ["/kazu/tests"]

[tool.hatch.version]
path = "kazu/__init__.py"

[tool.black]
line-length = 100
target-version = ['py38']
include = '\.pyi?$'
exclude = '''

(
  /(
      \.eggs         # exclude a few common directories in the
    | \.git          # root of the project
    | \.hg
    | \.mypy_cache
    | \.tox
    | \.venv
    | _build
    | buck-out
    | build
    | dist
    | .idea
    | .pytest_cache
  )/
)
'''

[tool.docformatter]
black = true
recursive = true
in-place = true

[tool.pytest.ini_options]
testpaths = "kazu/tests"
filterwarnings = ["ignore::DeprecationWarning", "ignore::SyntaxWarning"]

[tool.mypy]
strict = true
warn_unreachable = true
# this is painful to have on because if there is an existing function that is 'untyped', any new
# call site of it would raise a new error. We only really care about our own functions that are 'untyped',
# which would raise an error anyway for having no/incomplete typing if one was introduced in a new file.
disallow_untyped_calls = false
# turning this on is a pain, as it means __init__.py files like kazu/steps/__init__.py
# need additional work to 'reexport' symbols - probably the simplest is a redundant 'from x import y as y' (adding the as y)
# but this seems more work than is justified by the benefit.
no_implicit_reexport = false

[[tool.mypy.overrides]]
module = [
  "seqeval.*",
  "transformers.*",
  "tokenizers.*",
  "stanza.*",
  "sklearn.*",
  "srsly.*",
  "py4j.*",
  "diskcache.*",
  "rdkit.*",
  "ahocorasick.*",
  "scipy.*",
  "mwparserfromhell.*",
  "gliner.*",
  "vertexai.*",
]
ignore_missing_imports = true

[[tool.mypy.overrides]]
module = ["kazu.tests.*", "conftest"]
# in particular, return types for pytest test
# functions are always None, we don't want to have to annotate that
disallow_incomplete_defs = false
disallow_untyped_defs = false

[[tool.mypy.overrides]]
# the number in the comment after each line is the number of errors of this type for that file
# the numbers can be re-calculated with:
# mypy kazu docs conftest.py | cut -f 1,2 -d ':' | sed '$d' | sort | uniq | cut -f 1 -d ':' | uniq -c | sort --reverse | awk '{file_without_dot_py = substr($2, 0, length($2)-3); gsub("/", ".", file_without_dot_py); print "  \""file_without_dot_py"\",  # "$1}'
module = [
  "kazu.utils.spacy_pipeline", # 2
]
# we had a bunch of these in the codebase before we moved to a 'strict' mypy config, and it was too many
# to fix at that time for the payoff. Having overrides for the modules that would error rather than
# everywhere means that new files get the full 'strict' treatment (and also existing files that are already ok),
# and the lines above (combined with the counts) should make it easier to 'chip away' at
# these existing untyped functions file-by-file incrementally.
disallow_untyped_defs = false


[[tool.mypy.overrides]]
# as above - number is errors for that file
module = [
  "kazu.data",                                                         # 13
  "kazu.annotation.label_studio",                                      # 9
  "kazu.steps.linking.post_processing.xref_manager",                   # 3
  "kazu.steps.linking.post_processing.mapping_strategies.strategies",  # 3
  "kazu.web.server",                                                   # 2
  "kazu.web.jwtauth",                                                  # 2
  "kazu.utils.link_index",                                             # 2
  "kazu.utils.build_and_test_model_packs",                             # 2
  "kazu.steps.ner.hf_token_classification",                            # 2
  "kazu.steps.linking.post_processing.disambiguation.strategies",      # 2
  "kazu.annotation.acceptance_test",                                   # 2
  "kazu.utils.string_normalizer",                                      # 1
  "kazu.steps.linking.post_processing.disambiguation.context_scoring", # 1
  "kazu.steps.linking.entity_class_disambiguation",                    # 1
  "kazu.pipeline",                                                     # 1
  "kazu.ontology_matching.ontology_matcher",                           # 1
]
# as above, we want this to be true for files it can be, especially new files.
disallow_any_generics = false

[tool.towncrier]
package = "kazu"
directory = "docs/_changelog.d"
package-dir = "kazu"
filename = "CHANGELOG.md"
title_format = "## {version} - {project_date}"
underlines = ["", "", ""]
start_string = "<!-- towncrier release notes start -->\n"

[tool.vulture]
exclude = ["kazu/tests/*"]
make_whitelist = false
# we may need to change this to 80 or 100 if we find it too noisy
min_confidence = 60
paths = ["kazu", ".vulture_whitelist.py"]

[tool.docsig]
check-class-constructor = true
check-dunders = true
check-overridden = true
check-protected = true
ignore-no-params = true
check-property-returns = true
disable = [
  "E113", # function is missing a docstring
  "E114", # class is missing a docstring
  # docsig considers it a syntax error if there's no actual description
  # of the parameter. This is the main syntax error this catches, so turning
  # it off doesn't do much harm
  "E115", # syntax error in description
  # this isn't a problem, it just means there's a blank 'returns' section
  # of the documentation. We could go back and clean this up.
  "E104", # return statement documented for None
]