94 lines
1.5 KiB
Nix
94 lines
1.5 KiB
Nix
{
|
|
lib,
|
|
buildPythonPackage,
|
|
fetchFromGitHub,
|
|
cmake,
|
|
pkg-config,
|
|
cxxopts,
|
|
poetry-core,
|
|
pybind11,
|
|
zlib,
|
|
nlohmann_json,
|
|
utf8cpp,
|
|
libjpeg,
|
|
qpdf,
|
|
loguru-cpp,
|
|
# python dependencies
|
|
tabulate,
|
|
pillow,
|
|
pydantic,
|
|
docling-core,
|
|
pytestCheckHook,
|
|
}:
|
|
|
|
buildPythonPackage rec {
|
|
pname = "docling-parse";
|
|
version = "4.0.5";
|
|
pyproject = true;
|
|
|
|
src = fetchFromGitHub {
|
|
owner = "docling-project";
|
|
repo = "docling-parse";
|
|
tag = "v${version}";
|
|
hash = "sha256-H8/T9gwQ6MeNsNcJ5I9cVnQVFEXHfmqYCxhkxszD8/w=";
|
|
};
|
|
|
|
dontUseCmakeConfigure = true;
|
|
|
|
nativeBuildInputs = [
|
|
cmake
|
|
pkg-config
|
|
];
|
|
|
|
build-system = [
|
|
poetry-core
|
|
];
|
|
|
|
env.NIX_CFLAGS_COMPILE = "-I${lib.getDev utf8cpp}/include/utf8cpp";
|
|
|
|
buildInputs = [
|
|
pybind11
|
|
cxxopts
|
|
libjpeg
|
|
loguru-cpp
|
|
nlohmann_json
|
|
qpdf
|
|
utf8cpp
|
|
zlib
|
|
];
|
|
|
|
env.USE_SYSTEM_DEPS = true;
|
|
|
|
cmakeFlags = [
|
|
"-DUSE_SYSTEM_DEPS=True"
|
|
];
|
|
|
|
dependencies = [
|
|
tabulate
|
|
pillow
|
|
pydantic
|
|
docling-core
|
|
];
|
|
|
|
pythonRelaxDeps = [
|
|
"pydantic"
|
|
"pillow"
|
|
];
|
|
|
|
pythonImportsCheck = [
|
|
"docling_parse"
|
|
];
|
|
|
|
nativeCheckInputs = [
|
|
pytestCheckHook
|
|
];
|
|
|
|
meta = {
|
|
changelog = "https://github.com/DS4SD/docling-parse/blob/${src.tag}/CHANGELOG.md";
|
|
description = "Simple package to extract text with coordinates from programmatic PDFs";
|
|
homepage = "https://github.com/DS4SD/docling-parse";
|
|
license = lib.licenses.mit;
|
|
maintainers = with lib.maintainers; [ drupol ];
|
|
};
|
|
}
|