diff options
Diffstat (limited to 'app-text/tesseract/tesseract-9999.ebuild')
| -rw-r--r-- | app-text/tesseract/tesseract-9999.ebuild | 100 |
1 files changed, 33 insertions, 67 deletions
diff --git a/app-text/tesseract/tesseract-9999.ebuild b/app-text/tesseract/tesseract-9999.ebuild index 6bba4a49da59..f44c6f3d8fc8 100644 --- a/app-text/tesseract/tesseract-9999.ebuild +++ b/app-text/tesseract/tesseract-9999.ebuild @@ -3,56 +3,21 @@ EAPI=7 -MY_PN="tesseract-ocr" -LANGPACKV="4.00" -URI_PREFIX="https://github.com/${MY_PN}/tessdata/raw/${LANGPACKV}/" - -inherit autotools git-r3 toolchain-funcs +inherit autotools git-r3 multilib-minimal toolchain-funcs DESCRIPTION="An OCR Engine, orginally developed at HP, now open source." HOMEPAGE="https://github.com/tesseract-ocr" -SRC_URI="${URI_PREFIX}eng.traineddata -> eng.traineddata-${LANGPACKV} - math? ( ${URI_PREFIX}equ.traineddata -> equ.traineddata-${LANGPACKV} ) - osd? ( ${URI_PREFIX}osd.traineddata -> osd.traineddata-${LANGPACKV} )" EGIT_REPO_URI="https://github.com/${MY_PN}/${PN}.git" LICENSE="Apache-2.0" SLOT="0" KEYWORDS="" -IUSE="doc jpeg math opencl openmp osd png static-libs tiff training webp" - -# List of supported Gentoo linguas and their upstream mapping -# https://github.com/tesseract-ocr/tesseract/wiki/Data-Files -# "old" variants were regrouped in the matching modern locale -LANGUAGES="af:afr am:amh ar:ara as:asm az:aze,aze_cyrl be:bel bn:ben bo:bod bs:bos bg:bul ca:cat cs:ces zh:chi_sim,chi_tra cy:cym da:dan de:deu,frk dz:dzo el:ell,grc en:enm eo:epo et:est eu:eus fa:fas fi:fin fr:fra,frm ga:gle gl:glg gu:guj he:heb hi:hin hr:hrv hu:hun id:ind is:isl it:ita,ita_old ja:jpn kn:kan ka:kat,kat_old kk:kaz km:khm ky:kir ko:kor ku:kur lo:lao la:lat lv:lav lt:lit ml:mal mr:mar mk:mkd ms:msa my:mya ne:nep nl:nld no:nor or:ori pa:pan pl:pol pt:por ro:ron ru:rus sa:san si:sin sk:slk sl:slv es:spa,spa_old sq:sqi sr:srp,srp_latn sw:swa sv:swe syc:syr ta:tam te:tel tg:tgk tl:tgl th:tha tr:tur ug:uig uk:ukr uz:uzb,uzb_cyrl vi:vie" -# Missing matches: -# ceb Cebuano -# chr Cherokee -# hat Haitian; Haitian Creole -# iku Inuktitut -# jav Javanese -# mlt Maltese -# pus Pushto; Pashto -# tir Tigrinya -# urd Urdu -# yid Yiddish -# l10n_en provides the additional data: -# enm English, Middle (1100-1500) - -for lang in ${LANGUAGES}; do - gentoo_lang=${lang%:*} - tess_langs=${lang#*:} - for tess_lang in ${tess_langs//,/ }; do - SRC_URI+=" l10n_${gentoo_lang}? ( ${URI_PREFIX}${tess_lang}.traineddata -> ${tess_lang}.traineddata-${LANGPACKV} )" - done - IUSE+=" l10n_${gentoo_lang}" -done +IUSE="doc jpeg opencl openmp png static-libs tiff training webp" -# With opencl USE=tiff is necessary in leptonica -RDEPEND=">=media-libs/leptonica-1.74:=[zlib,tiff?,jpeg?,png?,webp?] +COMMON_DEPEND=">=media-libs/leptonica-1.74:=[${MULTILIB_USEDEP},zlib,tiff?,jpeg?,png?,webp?] opencl? ( - virtual/opencl - media-libs/tiff:0= + virtual/opencl[${MULTILIB_USEDEP}] + media-libs/tiff:0=[${MULTILIB_USEDEP}] media-libs/leptonica:=[tiff] ) training? ( @@ -60,15 +25,18 @@ RDEPEND=">=media-libs/leptonica-1.74:=[zlib,tiff?,jpeg?,png?,webp?] x11-libs/pango:= x11-libs/cairo:= )" - -DEPEND="${RDEPEND} +RDEPEND="${COMMON_DEPEND} + || ( + >=app-text/tessdata_fast-4.0.0 + >=app-text/tessdata_best-4.0.0 + >=app-text/tessdata_legacy-4.0.0 + )" +DEPEND="${COMMON_DEPEND} app-text/asciidoc app-text/docbook-xsl-stylesheets dev-libs/libxslt doc? ( app-doc/doxygen )" -DOCS=( AUTHORS ChangeLog README.md ) - pkg_pretend() { [[ ${MERGE_TYPE} != binary ]] && use openmp && tc-check-openmp } @@ -77,21 +45,12 @@ pkg_setup() { [[ ${MERGE_TYPE} != binary ]] && use openmp && tc-check-openmp } -src_unpack() { - git-r3_src_unpack - for file in ${A}; do - if [[ "${file}" == *traineddata* ]]; then - cp "${DISTDIR}/${file}" "${S}/tessdata/${file%-*}" || die - fi - done -} - src_prepare() { default eautoreconf } -src_configure() { +multilib_src_configure() { # scrollview disabled for now, see bug #686944 local myeconfargs=( --enable-shared @@ -101,25 +60,32 @@ src_configure() { $(use_enable static-libs static) ) - econf "${myeconfargs[@]}" + ECONF_SOURCE="${S}" econf "${myeconfargs[@]}" } -src_compile() { +multilib_src_compile() { default - use doc && emake doc - use training && emake training + if multilib_is_native_abi; then + use doc && emake doc + use training && emake training + fi } -src_install() { - use doc && HTML_DOCS=( doc/html/. ) - default - - find "${D}" -name '*.la' -type f -delete || die +multilib_src_install() { + if multilib_is_native_abi; then + DOCS=( "${S}"/AUTHORS "${S}"/ChangeLog "${S}"/README.md ) + if use doc; then + HTML_DOCS=( doc/html/. ) + fi + einstalldocs - if use training; then - emake DESTDIR="${D}" training-install + if use training; then + emake DESTDIR="${D}" training-install + fi fi + emake DESTDIR="${D}" install +} - insinto /usr/share/tessdata - doins tessdata/*traineddata* # language files +multilib_src_install_all() { + find "${D}" -name '*.la' -type f -delete || die } |