# Contributor: Carlo Landmeter <clandmeter@gmail.com>
# Maintainer: Francesco Colista <fcolista@alpinelinux.org>
pkgname=tesseract-ocr
pkgver=4.1.0
_tdver=4.0.0
pkgrel=0
pkgdesc="open source OCR engine"
url="https://github.com/tesseract-ocr/tesseract/releases"
arch="all"
license="Apache"
makedepends="automake autoconf libtool leptonica-dev pango-dev icu-dev
	cairo-dev"
subpackages="$pkgname-dev"
source="$pkgname-$pkgver.tar.gz::https://github.com/tesseract-ocr/tesseract/archive/$pkgver.tar.gz
	https://github.com/tesseract-ocr/tessdata/archive/${_tdver}/tessdata-${_tdver}.tar.gz
	include-time-h.patch
	"
_langs="afr ara aze bel ben bul cat ces chi_sim chi_tra chr dan deu ell
enm epo equ est eus fin fra frk frm glg grc heb hin hrv hun ind isl ita
ita_old jpn kan kor lav lit mal mkd mlt msa nld nor pol por ron rus slk
slv spa spa_old sqi srp swa swe tam tel tgl tha tur ukr vie"

for _lang in $_langs; do
	subpackages="$subpackages $pkgname-data-$_lang:_lang_data:noarch"
done

builddir="$srcdir"/${pkgname/-*}-$pkgver

prepare() {
	default_prepare
	./autogen.sh
}

build() {
	./configure \
		--build=$CBUILD \
		--host=$CHOST \
		--prefix=/usr \
		--sysconfdir=/etc \
		--mandir=/usr/share/man \
		--infodir=/usr/share/info \
		--localstatedir=/var \
		--disable-static
	make
	make training
}

package() {
	make DESTDIR="$pkgdir" install
	make DESTDIR="$pkgdir" training-install
	install -D "$srcdir"/tessdata-$_tdver/eng.* \
		"$srcdir"/tessdata-$_tdver/osd.* \
		"$srcdir"/tessdata-$_tdver/equ.* \
		"$pkgdir"/usr/share/tessdata/
}

_lang_data() {
	local lang="${subpkgname#$pkgname-data-}"
	pkgdesc="Tesseract language data for $lang"
	depends="$pkgname"
	mkdir -p "$subpkgdir"/usr/share/tessdata
	mv "$srcdir"/tessdata-$_tdver/$lang.* \
		"$subpkgdir"/usr/share/tessdata/
}

sha512sums="d617f5c5b826640b2871dbe3d7973bcc5e66fafd837921a20e009d683806ed50f0f258aa455019d99fc54f5cb65c2fa0380e3a3c92b39ab0684b8799c730b09d  tesseract-ocr-4.1.0.tar.gz
cd71bb99d44eefb53b359ba64b472c509fff773b2737a8d51e10d5d52d9a3a7ff870d470b1c72a7c78be3263b5ecfbb58a6eab13cf7128d8599681676cdcef6b  tessdata-4.0.0.tar.gz
3cc100bfa4eccac4123311c90626e747576b9d4b8b998f5e12bf6b2b9b0ffd8d9ea0160d11747e09d00283c4f782147c9b97c7b5c31679fb44784c2314de3cab  include-time-h.patch"