# Copyright 1999-2008 Gentoo Foundation
# Distributed under the terms of the GNU General Public License v2
# $Header: $

inherit eutils

DESCRIPTION="A commercial quality OCR engine developed at HP in the 80's and early 90's."
HOMEPAGE="http://code.google.com/p/tesseract-ocr/"
SRC_URI="http://tesseract-ocr.googlecode.com/files/${P}.tar.gz
	linguas_de? ( http://tesseract-ocr.googlecode.com/files/${PN}-2.00.deu.tar.gz )
	linguas_en? ( http://tesseract-ocr.googlecode.com/files/${PN}-2.00.eng.tar.gz )
	linguas_fr? ( http://tesseract-ocr.googlecode.com/files/${PN}-2.00.fra.tar.gz )
	linguas_it? ( http://tesseract-ocr.googlecode.com/files/${PN}-2.00.ita.tar.gz )
	linguas_nl? ( http://tesseract-ocr.googlecode.com/files/${PN}-2.00.nld.tar.gz )
	linguas_es? ( http://tesseract-ocr.googlecode.com/files/${PN}-2.00.spa.tar.gz )
	linguas_pt? ( http://tesseract-ocr.googlecode.com/files/${PN}-2.01.por.tar.gz )
	linguas_de_FR? ( http://tesseract-ocr.googlecode.com/files/${PN}-2.01.deu-f.tar.gz )
	linguas_vn? ( http://tesseract-ocr.googlecode.com/files/${PN}-2.01.vie.tar.gz )"

LICENSE="Apache-2.0"
SLOT="0"
KEYWORDS="~x86"
IUSE="tiff linguas_de linguas_de_FR linguas_en linguas_es linguas_fr linguas_it linguas_nl linguas_vn"

DEPEND="tiff? ( media-libs/tiff )"
RDEPEND="$DEPEND"

src_unpack() {
	unpack ${A}
	cd "${S}"
	epatch "${FILESDIR}/${P}-unicharset_extractor.patch"

	# Add language data (linguas_de...)
	if [ -e "${WORKDIR}/tessdata/" ]; then
		mv -f "${WORKDIR}/tessdata/"* "${S}/tessdata/"
	else
		ewarn "TIP:"
		ewarn "You need at least one of the language data files (linguas_??) or tesseract will not work."
		ewarn " "
	fi

	# ERROR -> Make: *** No rule to make target `install'.  Stop.
	# Look into this, but removing makefile seems to be a quick fix.
	rm -f "${S}/java/makefile"
}

src_compile() {
	econf $(use_with tiff libtiff ) || die "econf failed"
	emake || die
}

src_install() {
	emake DESTDIR="${D}" install || die "install failed"

	dodoc README ReleaseNotes AUTHORS
}