# automatically generated by g-sorcery
# please do not edit this file

EAPI=8

REALNAME="${PN}"
LITERALNAME="${PN}"
REALVERSION="${PV}"
DIGEST_SOURCES="yes"
PYTHON_COMPAT=( python{3_11,3_12,3_13,3_14} )
DISTUTILS_USE_PEP517=standalone

inherit python-r1 gs-pypi

DESCRIPTION="website article / adobe pdf file discovery & extraction"

HOMEPAGE="https://github.com/Stimson-Center/stimson-web-scraper"
LICENSE="MIT"
SRC_URI="https://files.pythonhosted.org/packages/source/${REALNAME::1}/${REALNAME}/${REALNAME}-${REALVERSION}.tar.gz"
SOURCEFILE="${REALNAME}-${REALVERSION}.tar.gz"
RESTRICT="test"

SLOT="0"
KEYWORDS="~amd64 ~x86"

IUSE=""
DEPENDENCIES=">=dev-python/beautifulsoup4-4.9.1[${PYTHON_USEDEP}]
	>=dev-python/click-7.1.2[${PYTHON_USEDEP}]
	>=dev-python/cssselect-1.1.0[${PYTHON_USEDEP}]
	dev-python/date-extractor[${PYTHON_USEDEP}]
	dev-python/fake-useragent[${PYTHON_USEDEP}]
	dev-python/feedfinder2[${PYTHON_USEDEP}]
	>=dev-python/feedparser-5.2.1[${PYTHON_USEDEP}]
	dev-python/fpdf2[${PYTHON_USEDEP}]
	dev-python/jieba3k[${PYTHON_USEDEP}]
	>=dev-python/lxml-4.5.2[${PYTHON_USEDEP}]
	>=dev-python/numpy-1.19.1[${PYTHON_USEDEP}]
	>=dev-python/pandas-1.1.0[${PYTHON_USEDEP}]
	dev-python/pdftotext[${PYTHON_USEDEP}]
	~dev-python/pillow-6.2.2[${PYTHON_USEDEP}]
	>=dev-python/psutil-5.7.2[${PYTHON_USEDEP}]
	dev-python/PyArabic[${PYTHON_USEDEP}]
	dev-python/pytextrank[${PYTHON_USEDEP}]
	dev-python/pythainlp[${PYTHON_USEDEP}]
	>=dev-python/python-dateutil-2.8.1[${PYTHON_USEDEP}]
	dev-python/dateparser[${PYTHON_USEDEP}]
	dev-python/PyPDF4[${PYTHON_USEDEP}]
	>=dev-python/pyyaml-5.3.1[${PYTHON_USEDEP}]
	>=dev-python/requests-2.24.0[${PYTHON_USEDEP}]
	>=dev-python/requests-toolbelt-0.9.1[${PYTHON_USEDEP}]
	dev-python/spacy[${PYTHON_USEDEP}]
	>=dev-python/tldextract-2.2.2[${PYTHON_USEDEP}]
	dev-python/tinysegmenter[${PYTHON_USEDEP}]
	>=dev-python/waitress-1.4.4[${PYTHON_USEDEP}]"
BDEPEND="${DEPENDENCIES}"
RDEPEND="${DEPENDENCIES}"