diff --git a/tesseract/PKGBUILD b/tesseract/PKGBUILD index febedf07d..56d9d30cf 100644 --- a/tesseract/PKGBUILD +++ b/tesseract/PKGBUILD @@ -3,99 +3,121 @@ # # contributor (x86_64): Giuseppe CalĂ  -# include global config -source ../_buildscripts/${current_repo}-${_arch}-cfg.conf +# chinese requires two files +_langs=(afr ara aze bel ben bul cat ces chi_sim chi_tra chr dan deu ell +eng enm epo epo_alt 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) -pkgname=tesseract -pkgver=3.01 -pkgrel=1 +pkgbase=tesseract +pkgname=(tesseract $(for l in ${_langs[@]}; do echo tesseract-data-${l}; done)) +pkgver=3.02.02 +_pkgver=3.02 +pkgrel=2 pkgdesc="An OCR programm" -arch=(x86_64 i686) +arch=(i686 x86_64) url="http://code.google.com/p/tesseract-ocr" license=("APACHE") depends=(libpng libtiff libjpeg zlib giflib gcc-libs leptonica) -categories=('office') source=(http://tesseract-ocr.googlecode.com/files/$pkgname-$pkgver.tar.gz - http://tesseract-ocr.googlecode.com/files/chi_tra.traineddata.gz - http://tesseract-ocr.googlecode.com/files/chi_sim.traineddata.gz - http://tesseract-ocr.googlecode.com/files/ind.traineddata.gz - http://tesseract-ocr.googlecode.com/files/swe.traineddata.gz - http://tesseract-ocr.googlecode.com/files/ron.traineddata.gz - http://tesseract-ocr.googlecode.com/files/slv.traineddata.gz - http://tesseract-ocr.googlecode.com/files/srp.traineddata.gz - http://tesseract-ocr.googlecode.com/files/tgl.traineddata.gz - http://tesseract-ocr.googlecode.com/files/tur.traineddata.gz - http://tesseract-ocr.googlecode.com/files/hun.traineddata.gz - http://tesseract-ocr.googlecode.com/files/fin.traineddata.gz - http://tesseract-ocr.googlecode.com/files/ita.traineddata.gz - http://tesseract-ocr.googlecode.com/files/nld.traineddata.gz - http://tesseract-ocr.googlecode.com/files/jpn.traineddata.gz - http://tesseract-ocr.googlecode.com/files/vie.traineddata.gz - http://tesseract-ocr.googlecode.com/files/spa.traineddata.gz - http://tesseract-ocr.googlecode.com/files/ukr.traineddata.gz - http://tesseract-ocr.googlecode.com/files/fra.traineddata.gz - http://tesseract-ocr.googlecode.com/files/slk.traineddata.gz - http://tesseract-ocr.googlecode.com/files/kor.traineddata.gz - http://tesseract-ocr.googlecode.com/files/ell.traineddata.gz - http://tesseract-ocr.googlecode.com/files/rus.traineddata.gz - http://tesseract-ocr.googlecode.com/files/por.traineddata.gz - http://tesseract-ocr.googlecode.com/files/bul.traineddata.gz - http://tesseract-ocr.googlecode.com/files/lav.traineddata.gz - http://tesseract-ocr.googlecode.com/files/lit.traineddata.gz - http://tesseract-ocr.googlecode.com/files/pol.traineddata.gz - http://tesseract-ocr.googlecode.com/files/dan-frak.traineddata.gz - http://tesseract-ocr.googlecode.com/files/deu.traineddata.gz - http://tesseract-ocr.googlecode.com/files/dan.traineddata.gz - http://tesseract-ocr.googlecode.com/files/ces.traineddata.gz - http://tesseract-ocr.googlecode.com/files/cat.traineddata.gz - http://tesseract-ocr.googlecode.com/files/eng.traineddata.gz) -md5sums=('1ba496e51a42358fb9d3ffe781b2d20a' - '06fce5f5c0221286eab591819406d91e' - '127e0b742b615d81c3b97a555acd8285' - 'f4ae2e5668f693cc51c6e3c3f57a8f1d' - '3787ff0fc1a07517203466830f6f868e' - '7b50075692e7a07ac8f59e0cde51cfb9' - 'defc498fa09480ef0052fde79f4f84e5' - 'c1e0f0ef07d26a72063802370bbfe2c9' - 'af86cab0da52fff7e83f9c7668617a21' - 'b0f4798bf1f9ee541ae73aff49f08821' - '966fa3ebe24d7dfade951e0a4a9f52b5' - '33812ef03197dce20b8c888409d8541a' - '9a307661a010703cd2bb265a3da5eaff' - '89992f69802b9f4a36fec40efd95c503' - 'cad3dd7c9401fa0676bf195456ba0c14' - 'af75c7a696b18a1d2e4f7e21a222cf95' - 'dd9078211127189995504f034468fcd6' - '56a82ac3994f0552d54fd14ff5ed6817' - 'e6000550883684b5e93b7e0a21016287' - '2e7fbf80215b38f423dcf78e396b92d7' - '909b811739b2e7d51cc3ca109998353a' - '0d3a21c18f752eb058e7b854c37b7a69' - 'c477b7faddc044a4595135d0378be34c' - 'd0b59fcfd80f9e09be890647ae93f00d' - 'a7501f9a48bc7ef2fa50cba8b8f0c7b1' - '1f366f55e5e0725deb751ec847bf42dc' - '1b4460ab998151f1ed8e48928ded58be' - 'c3d6447245663138f1d3aa4567c72192' - 'cfa8a81df6a34940604fafa7353ffc43' - 'be81a761f61800f6d39393a31435fff3' - '7aaeb321dbf069d14d2601dc36d3dddc' - '2e97c952075064834f1cad6a133970fa' - '308099b2b95834916ec410191c86cd48' - 'd91041ad156cf2db36664e91ef799451') + $(for l in ${_langs[@]}; do + echo http://tesseract-ocr.googlecode.com/files/tesseract-ocr-${_pkgver}.${l}.tar.gz + done)) +md5sums=('3d57ee5777fa998632ad0693c13a0e9e' + '16a756bf353896e34386f4509582e487' + 'd7b3708f1f1f2898c258209923063f57' + '143044ff7b643b844c2f0b2a170ceda2' + '3bb6dab515166d641ac62e973c868ac5' + '0b662630eb4b2e0ebf40c38184068fba' + '91fc34ef041aafe32549a7a6232afad0' + '27b77ecc66f8c8d904cd6b87886aca22' + 'ae144b9d230267b5ae62804ad7a09dac' + '718779539d3202ae95ae5ff52a3ad5e8' + '391ed3fcf014bae777b33dc7a771a00d' + 'dd5b4fa247500d7713ab5556af62d9fa' + '2f6a93920b3ce55de24ca762a66a5de1' + '57bdb26ec7c767e126ff97776d8bfb10' + '3760cc3dee6489dcc865b31e7a36bd47' + '3562250fe6f4e76229a329166b8ae853' + 'd9dc4ad9cbcde78e5be59eb226865661' + '5dc0cf01d21ee43ed834ee947e1e810a' + '67815529386679223092ddf301bf57d4' + '94667836c651fc4fd43645efdc34e19a' + 'a067f5975cbe83fea205897187fa256b' + '4c2cee5bf3558f6384e5907aa02ae80c' + '71f8b343ed2010f775ff474faec5874f' + '2bca200b783fdba2f546063c68740c50' + 'dd734ac212a98a02945760891b657e70' + 'e0ca24ca45de83e67057f0e208233ffc' + 'ef7b8e13891e08f29b02011a5dd6b630' + 'ac493e5a49e456f2dd354f5b6449573e' + '7ae7f3384083c66f06d63b6c0f0a66c3' + '186f17b326a58313344c2ea987ea7785' + '0396a267d894f485b6b534476f3322dc' + 'eb1b3537cd664fec48bebac75fac5992' + '9f672ea7c90eecbbe35ec353ff45c764' + '3f05946a9e57a8ffabb3fccc8a80cfaa' + 'fbdba86f34b907f5a709051c01b53f71' + '989ca558eb05fcf6873dfdd8c692291d' + 'e1be2b08e667c3bb0ab0f18c9456bf2f' + '510016e85d7ea8e1f4ebf98741138909' + 'fc9125f6652fb661575d6e15c94f88f6' + '2095b56c9c6b0dba9b837fd226ed6d29' + '801021a1ad6e4aa3ef697f065ff6c798' + '1f5e8f022d3c2697075bdfc03935f128' + 'b8930c64befcaadbb47951cad4a2f075' + '90a41ac86375b230980b643b35b88011' + 'cb42ec9504eecdc4f974dbca4387ac34' + '6db5753b3b03296544bbb2c0f59a7178' + '6d979585abb07c13dda003820129461c' + '90b3f1f4113a9dccd0287e6b5b7512d6' + 'ad490a8091ee329190266f3fd057c511' + '9f498e01855a57473af8f083c125455a' + 'e8c785df8aaa49e61f3a95aa24066122' + '991e4771e16abd6a66caeaeeb2a4d4d2' + 'c7bea975ba663d13e03c29d6cf33b5a8' + 'fe1385ed203ecd9e0999a44b135069ec' + '7838da948e404bb760bef83a5c3d64dd' + '372e38617fcc7e25c77996c5499ae063' + 'd8350668fa790b5d4007348a55840bdf' + '6c68107a71f82fa42c858e4d6dc70aca' + '46d3a5d7159236edb33272dd8b1090f6' + 'c68b0fdcdf34dd431c25f323673ef1d6' + '349d203ef88864eab8d1201672ca84be' + 'd458f0719b99e4dcfd4fa2a06e5150a3' + '3dc5a5028e75da12eaba62669a80acc4' + 'a8934e0e6496997112cbef2961adef5e' + '0aa61f18463bb89234a6d06fcbf0c644' + '78a74432848b4c53cc2ccf2eb339c0b6') build() { cd $srcdir/$pkgname-$pkgver - [ -f Makefile.in ] || automake + sed -i '1,1i#include ' viewer/svutil.cpp + [ -f Makefile.in ] || { aclocal && automake; } [ -f Makefile ] || ./configure --prefix=/usr make } -package() { +package_tesseract() { + install=tesseract.install + optdepends=($(for l in ${_langs[@]}; do echo tesseract-data-${l}; done)) + cd $srcdir/$pkgname-$pkgver make DESTDIR=$pkgdir install mkdir -p $pkgdir/usr/share/tessdata - cp $srcdir/*.traineddata $pkgdir/usr/share/tessdata/ - find $pkgdir/usr/share/tessdata -type f -exec chmod 0644 {} \; } +# Declare the package functions for data +for l in ${_langs[@]}; do + eval " +package_tesseract-data-${l}(){ + pkgdesc=\"Tesseract OCR data ($l)\" + depends=('tesseract') + groups=('tesseract-data') + + mkdir -p \$pkgdir/usr/share/tessdata + cp \$srcdir/tesseract-ocr/tessdata/${l}.* \$pkgdir/usr/share/tessdata/ + find \$pkgdir/usr/share/tessdata -type f -exec chmod 0644 {} \; +} + " +done diff --git a/tesseract/tesseract.install b/tesseract/tesseract.install new file mode 100644 index 000000000..114b82821 --- /dev/null +++ b/tesseract/tesseract.install @@ -0,0 +1,7 @@ +post_install() { + echo "You must install one of tesseract-data-* packages or whole tesseract-data group" +} + +post_upgrade() { + post_install +}