# Contributor: Carlo Landmeter <clandmeter@alpinelinux.org>
# Maintainer: Francesco Colista <fcolista@alpinelinux.org>
pkgname=tesseract-ocr
pkgver=5.3.0
_tdver=4.1.0
pkgrel=0
pkgdesc="open source OCR engine"
url="https://github.com/tesseract-ocr/tesseract/releases"
arch="all"
license="Apache-2.0"
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
	"
_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 kat 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
	GIT_DIR=. ./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="
fc0cfda980059390b7d6d3d8e475b4ecc9dcfee1abc3b16fc9ffc8904db0a0847a5f6f1b32bd2e1f6f5a714a8c4df0fb3dc7c6aa25ece6162bf52fe86b4d0f34  tesseract-ocr-5.3.0.tar.gz
c0b55fb5542d25ebd4b56d25155cc8254027c9503af298641686388886403ee26ebf9ef47d21d530d372deeba6a01f6eb17705e1ab46bb02f9f7bf4f63805cd0  tessdata-4.1.0.tar.gz
"