pyproject.toml

[project]
name = "camelot-py"
version = "2.0.0rc1"
description = "PDF Table Extraction for Humans."
authors = [
    {name = "Vinayak Mehta", email = "vmehta94@gmail.com"},
]
license = {text = "MIT"}
readme = "README.md"
classifiers = [
    "Development Status :: 5 - Production/Stable",
    "Programming Language :: Python :: 3",
    "Programming Language :: Python :: 3.10",
    "Programming Language :: Python :: 3.11",
    "Programming Language :: Python :: 3.12",
    "Programming Language :: Python :: 3.13",
    "Programming Language :: Python :: 3.14",
]
requires-python = ">=3.10"
dependencies = [
    "click>=8.0.1",
    "numpy>=1.26.1",
    "openpyxl>=3.1.0",
    "pandas>=2.2.2",
    "tabulate>=0.9.0",
    "typing-extensions>=4.12.2; python_version < '3.11'",
    "opencv-python-headless>=4.7.0.68",
    "pypdfium2>=4",
    "pillow>=10.4.0",
    "playa-pdf>=0.8.1",
]

[build-system]
requires = ["setuptools"]
build-backend = "setuptools.build_meta"

[tool.setuptools.packages.find]
where = ["."]
include = ["camelot*"]
exclude = ["activate"]

[project.urls]
Homepage = "https://github.com/camelot-dev/camelot"
Repository = "https://github.com/camelot-dev/camelot"
Documentation = "https://camelot-py.readthedocs.io/"
Changelog = "https://github.com/camelot-dev/camelot/releases"

[project.optional-dependencies]
plot = [
    "matplotlib>=3.7.5; python_version < '3.12'",
    "matplotlib>=3.8.0; python_version >= '3.12'",
]
# Optional neural backend for flavor='ml' (Table Transformer / TATR via
# HuggingFace transformers). Heavy (pulls torch); opt in explicitly with
# `pip install 'camelot-py[ml]'`. Imported lazily so the core install and
# the other flavors never load torch.
ml = [
    "torch>=2.0",
    # <5: transformers v5 strict-validates configs and rejects the TATR
    # checkpoints (published for v4, e.g. `dilation: None`). The whole TATR
    # ecosystem pins to v4; revisit when the checkpoints are v5-clean.
    "transformers>=4.40,<5",
    "timm>=0.9",
    "pillow>=10.4.0",
]
# Optional OCR text source for flavor='ml' on scanned / image-only PDFs
# (structure still from TATR; text from OCR instead of the PDF text layer).
# Pip-only, no system tesseract binary. `pip install 'camelot-py[ocr]'`.
ocr = [
    "rapidocr-onnxruntime>=1.3",
]
ghostscript = [
    "ghostscript>=0.7",
]
dev = [
    "Pygments>=2.10.0",
    "coverage[toml]>=6.2",
    "ruff>=0.12.10",
    "sphinx-book-theme>=1.0.1",
    "mypy>=0.930",
    "pre-commit>=2.16.0",
    "pre-commit-hooks>=4.1.0",
    "pytest>=6.2.5",
    "pytest-mpl>=0.17.0",
    "safety>=2.2.3",
    "sphinx>=4.3.2",
    "sphinx-autobuild>=2021.3.14",
    "sphinx-click>=3.0.2",
    "sphinx-copybutton>=0.5.0",
    "sphinx-prompt>=1.5.0",
    "typeguard>=2.13.3",
    "xdoctest[colors]>=0.15.10",
    "myst-parser>=2.0.0",
    "nox>=2024.10.9",
    "twine",
    "build",
]

[project.scripts]
camelot = "camelot.__main__:main"
# Alias matching the PyPI package name so `uvx camelot-py …` works without
# the `--from camelot-py` prefix. See #639.
camelot-py = "camelot.__main__:main"

[tool.coverage.paths]
source = ["camelot", "*\\camelot", "*/site-packages"]
tests = ["tests", "*/tests"]

[tool.coverage.run]
branch = true
source = ["camelot", "tests"]

[tool.coverage.report]
show_missing = true
fail_under = 90

[tool.ruff]
# Ruff replaces Black (formatter) and isort (import sorting). Match
# `requires-python = ">=3.10"` so the formatter targets the floor version
# (also keeps formatting stable across the 3.10–3.15 CI matrix).
target-version = "py310"
line-length = 88  # Black's default, which the codebase was formatted to.
# Notebooks were never linted/formatted by the old black/flake8 (types: [python]);
# keep that behaviour. Vendored Flask theme is third-party.
extend-exclude = ["*.ipynb", "docs/_themes/*"]
# pre-commit passes file paths explicitly; force-exclude makes Ruff honour
# the excludes above even then (otherwise excluded files still get linted).
force-exclude = true

[tool.ruff.lint]
# Ruff replaces flake8 + its plugins + isort + pyupgrade. Mapping of the
# old `.flake8` select (B,B9,C,D,DAR,E,F,N,RST,S,W):
#   B   -> flake8-bugbear        S   -> flake8-bandit
#   C   -> mccabe (C90)          W/E -> pycodestyle
#   D   -> pydocstyle            F   -> pyflakes
#   N   -> pep8-naming           I   -> isort (was a separate tool)
#   UP  -> pyupgrade (was a separate hook)
# Not ported: RST (flake8-rst-docstrings) and DAR (darglint) have no Ruff
# equivalent. DAR was inert (darglint wasn't installed); RST docstring
# linting is dropped — the one capability lost in the move.
select = ["B", "C90", "D", "E", "F", "I", "N", "S", "UP", "W"]
ignore = [
    "D401", # imperative-mood first line (was ignored under flake8)
    "E203", # whitespace before ':' — Ruff-format/Black compatibility
    "E501", # line too long — the formatter owns wrapping
]

[tool.ruff.lint.pydocstyle]
convention = "numpy"

[tool.ruff.lint.mccabe]
max-complexity = 10

[tool.ruff.lint.isort]
# Mirror the previous `isort` config (profile = "black", one import per line).
force-single-line = true

[tool.ruff.lint.per-file-ignores]
"tests/*" = [
    "D100",
    "D101",
    "D102",
    "D103",
    "D104",
    "D200",
    "D401",
    "S101",
    "S106",
    "F403",
    "F405",
    "F841",
]
"__init__.py" = ["D100", "D103", "D104", "F401"]
"__version__.py" = ["D100"]
"__main__.py" = ["D100", "D103"]
# Benchmark scripts are dev-only (not shipped); docstrings optional and some
# deliberately copy legacy complex code for apples-to-apples timing.
"bench/*" = ["C901", "D100", "D101", "D102", "D103", "S101"]

[tool.mypy]
warn_unreachable = true
pretty = true
show_column_numbers = true
show_error_codes = true
show_error_context = true
ignore_missing_imports = true


# strict = true
# Strict is too agressive for now and not passing.
# should uncomment the settings below one by one and make them pass.
# Then strict can be enabled. (Values taken from mypy docs. section existing_code)

# Start off with these
warn_unused_configs = true
warn_redundant_casts = true
# warn_unused_ignores = true

# Getting these passing should be easy
strict_equality = true
# extra_checks = true

# Strongly recommend enabling this one as soon as you can
# check_untyped_defs = true

# These shouldn't be too much additional work, but may be tricky to
# get passing if you use a lot of untyped libraries
disallow_subclassing_any = true
disallow_untyped_decorators = true
disallow_any_generics = true

# These next few are various gradations of forcing use of type annotations
# disallow_untyped_calls = true
# disallow_incomplete_defs = true
# disallow_untyped_defs = true

# This one isn't too hard to get passing, but return on investment is lower
# no_implicit_reexport = true

# This one can be tricky to get passing if you use a lot of untyped libraries
# warn_return_any = true