[tool.poetry]
name = "wordfreq"
version = "3.1.0"
description = "Look up the frequencies of words in many languages, based on many sources of data."
authors = ["Robyn Speer <rspeer@arborelia.net>"]
license = "Apache-2.0"
readme = "README.md"
homepage = "https://github.com/rspeer/wordfreq/"

[tool.poetry.dependencies]
python = ">= 3.8, < 4"
msgpack = "^1.0.7"
langcodes = ">= 3.0"
regex = ">= 2023.10.3"
ftfy = ">= 6.1"
mecab-python3 = { version = "^1.0.5", optional = true }
ipadic = { version = "^1.0.0", optional = true }
mecab-ko-dic = { version = "^1.0.0", optional = true }
jieba = { version = ">=0.42", optional = true }
locate = "^1.1.1"

[tool.poetry.group.dev.dependencies]
pytest = "^7.2.0"
mecab-python3 = "^1.0.5"
jieba = "^0.42.1"
ipadic = "^1.0.0"
mecab-ko-dic = "^1.0.0"
ipython = ">=7"
mypy = "^1.7.0"
ruff = "^0.1.6"
setuptools = "^69.0.2"   # implicit dependency in jieba
pytest-profiling = "^1.7.0"

[tool.poetry.extras]
cjk = ["mecab-python3", "ipadic", "mecab-ko-dic", "jieba"]
mecab = ["mecab-python3", "ipadic", "mecab-ko-dic"]
jieba = ["jieba"]

[tool.ruff]
# ruff is a Python linter and formatter, which reimplements black, flake8, pylint, and more
line-length = 99
show-fixes = true
src = ["src"]
target-version = "py38"

[tool.ruff.lint]
extend-select = [
    "D",     # docstring style
    "I",     # isort
    "UP",    # pyupgrade -- catches obsolete code patterns
    "ANN",   # type annotations
    "ASYNC", # async code checks
    "B",     # flake8-bugbear -- catches bug-prone usage
    "C4",    # list comprehensions
    "FA",    # correct use of `from __future__ import annotations`
    "INP",   # checks for presence of __init__.py
    "T20",   # print statements
    "TID",   # tidy imports
    "PTH",   # use pathlib instead of os.path
    "PLE",   # pylint errors
]
ignore = [
    "D100",   # modules without top-of-module docstrings are okay
    "D104",   # __init__.py doesn't need a docstring
    "D107",   # __init__ method doesn't need a docstring
    "D2",     # don't check whitespace in docstrings
    "D4",     # don't check grammar and style in docstrings
    "E501",   # long lines after auto-formatting, such as long strings, are okay
    "ANN002", # we don't demand a type on *args, particularly because it's difficult to specify
    "ANN003", # we don't demand a type on **kwargs, particularly because it's difficult to specify
    "ANN101", # self does not need a type
    "ANN102", # `cls` in classmethod does not need a type
    "PTH123", # we don't need to construct a Path just to open a file by filename
]

[tool.ruff.lint.per-file-ignores]
# We are less strict about test code
"tests/**" = ["D", "ANN", "T20", "INP"]

[build-system]
requires = ["poetry-core>=1.0.0"]
build-backend = "poetry.core.masonry.api"