tesseract/tesseract.spec

%define majver %(echo %version | cut -d. -f1-2)
Name:          tesseract
Version:       5.5.0
Release:       1mamba
Summary:       An OCR Engine that was developed at HP Labs between 1985 and 1995
Group:         Applications/Accessibility
Vendor:        openmamba
Distribution:  openmamba
Packager:      Silvan Calarco <silvan.calarco@mambasoft.it>
URL:           https://github.com/tesseract-ocr/
Source:        https://github.com/tesseract-ocr/tesseract.git/%{version}/tesseract-%{version}.tar.bz2
Source1:       https://github.com/tesseract-ocr/langdata.git/main/langdata-%{version}.tar.bz2
#Source1:       http://tesseract-ocr.googlecode.com/files/tesseract-ocr-%{majver}.eng.tar.gz
#Source2:       http://tesseract-ocr.googlecode.com/files/tesseract-ocr-%{majver}.hin.tar.gz
#Source3:       http://tesseract-ocr.googlecode.com/files/tesseract-ocr-%{majver}.ara.tar.gz
#Source4:       http://tesseract-ocr.googlecode.com/files/tesseract-ocr-3.01.osd.tar.gz
#Source5:       http://tesseract-ocr.googlecode.com/files/tesseract-ocr-%{majver}.tha.tar.gz
#Source6:       http://tesseract-ocr.googlecode.com/files/tesseract-ocr-3.01.slk-frak.tar.gz
#Source7:       http://tesseract-ocr.googlecode.com/files/tesseract-ocr-3.01.heb-com.tar.gz
#Source8:       http://tesseract-ocr.googlecode.com/files/tesseract-ocr-%{majver}.heb.tar.gz
Source9:       https://tesseract-ocr.googlecode.com/files/tesseract-ocr-3.02.ita.tar.gz
Source10:      http://tesseract-ocr.googlecode.com/files/chi_tra.traineddata.gz
Source11:      http://tesseract-ocr.googlecode.com/files/chi_sim.traineddata.gz
Source12:      http://tesseract-ocr.googlecode.com/files/ind.traineddata.gz
Source13:      http://tesseract-ocr.googlecode.com/files/swe.traineddata.gz
Source14:      http://tesseract-ocr.googlecode.com/files/ron.traineddata.gz
Source15:      http://tesseract-ocr.googlecode.com/files/slv.traineddata.gz
Source16:      http://tesseract-ocr.googlecode.com/files/srp.traineddata.gz
Source17:      http://tesseract-ocr.googlecode.com/files/tgl.traineddata.gz
Source18:      http://tesseract-ocr.googlecode.com/files/tur.traineddata.gz
Source19:      http://tesseract-ocr.googlecode.com/files/hun.traineddata.gz
Source20:      http://tesseract-ocr.googlecode.com/files/fin.traineddata.gz
Source22:      http://tesseract-ocr.googlecode.com/files/nld.traineddata.gz
Source23:      http://tesseract-ocr.googlecode.com/files/nor.traineddata.gz
Source24:      http://tesseract-ocr.googlecode.com/files/jpn.traineddata.gz
Source25:      http://tesseract-ocr.googlecode.com/files/vie.traineddata.gz
Source26:      http://tesseract-ocr.googlecode.com/files/spa.traineddata.gz
Source27:      http://tesseract-ocr.googlecode.com/files/ukr.traineddata.gz
Source28:      http://tesseract-ocr.googlecode.com/files/fra.traineddata.gz
Source29:      http://tesseract-ocr.googlecode.com/files/slk.traineddata.gz
Source30:      http://tesseract-ocr.googlecode.com/files/kor.traineddata.gz
Source31:      http://tesseract-ocr.googlecode.com/files/ell.traineddata.gz
Source32:      http://tesseract-ocr.googlecode.com/files/rus.traineddata.gz
Source33:      http://tesseract-ocr.googlecode.com/files/por.traineddata.gz
Source34:      http://tesseract-ocr.googlecode.com/files/bul.traineddata.gz
Source35:      http://tesseract-ocr.googlecode.com/files/lav.traineddata.gz
Source36:      http://tesseract-ocr.googlecode.com/files/lit.traineddata.gz
Source37:      http://tesseract-ocr.googlecode.com/files/pol.traineddata.gz
Source38:      http://tesseract-ocr.googlecode.com/files/dan-frak.traineddata.gz
Source39:      http://tesseract-ocr.googlecode.com/files/deu.traineddata.gz
Source40:      http://tesseract-ocr.googlecode.com/files/dan.traineddata.gz
Source41:      http://tesseract-ocr.googlecode.com/files/ces.traineddata.gz
Source42:      http://tesseract-ocr.googlecode.com/files/cat.traineddata.gz
Source43:       http://tesseract-ocr.googlecode.com/files/chr.traineddata.gz
Source44:       http://tesseract-ocr.googlecode.com/files/deu-frak.traineddata.gz
Source45:       http://tesseract-ocr.googlecode.com/files/swe-frak.traineddata.gz
Patch0:        %{name}-3.01-gcc-4.7.patch
License:       Apache License 2.0
## AUTOBUILDREQ-BEGIN
BuildRequires: glibc-devel
BuildRequires: libarchive-devel
BuildRequires: libcurl-devel
BuildRequires: libgcc
BuildRequires: libgomp-devel
BuildRequires: libleptonica-devel
BuildRequires: libstdc++6-devel
## AUTOBUILDREQ-END
BuildRequires: libleptonica-devel >= 1.83.0
Requires:      lib%{name} = %{?epoch:%epoch:}%{version}-%{release}

%description
Tesseract is an OCR Engine that was developed at HP Labs between 1985 and 1995.

%package -n lib%{name}
Group:         System/Libraries
Summary:       Shared libraries for %{name}

%description -n lib%{name}
This package contains shared libraries for %{name}.

%package -n lib%{name}-devel
Group:         Development/Libraries
Summary:       Development files for %{name}
Requires:      lib%{name} = %{?epoch:%epoch:}%{version}-%{release}
## note: you can add this requirement if .pc files are provided by this package
#Requires:      pkg-config

%description -n lib%{name}-devel
This package contains libraries and header files for developing applications that use %{name}.

%debug_package

%prep
%setup -q -a1
#-D -T
#:<< _EOF
#-a2 -a3 -a4 -a5 -a6 -a7 -a8 -a9
#%patch0 -p1

cd tessdata
cp %{S:9} %{S:10} %{S:11} %{S:12} %{S:13} %{S:14} %{S:15} %{S:16} %{S:17} %{S:18} %{S:19} .
cp %{S:20} %{S:22} %{S:23} %{S:24} %{S:25} %{S:26} %{S:27} %{S:28} %{S:29} .
cp %{S:30} %{S:31} %{S:32} %{S:33} %{S:34} %{S:35} %{S:36} %{S:37} %{S:38} %{S:39} .
cp %{S:40} %{S:41} %{S:42} %{S:43} %{S:44} %{S:45} .
gunzip *.gz
cd ..

#sed -i "s|AM_CONFIG_HEADER|AC_CONFIG_HEADERS|" configure.ac

./autogen.sh

%build
#:<< _EOF
%configure
%make

%install
[ "%{buildroot}" != / ] && rm -rf "%{buildroot}"
%makeinstall

cp -a tessdata/*.traineddata %{buildroot}%{_datadir}/tessdata/

%clean
[ "%{buildroot}" != / ] && rm -rf "%{buildroot}"


%files
%defattr(-,root,root)
#%{_bindir}/ambiguous_words
#%{_bindir}/classifier_tester
#%{_bindir}/dawg2wordlist
#%{_bindir}/shapeclustering
#%{_bindir}/cntraining
#%{_bindir}/combine_tessdata
#%{_bindir}/mftraining
%{_bindir}/tesseract
#%{_bindir}/unicharset_extractor
#%{_bindir}/wordlist2dawg
%{_datadir}/tessdata/pdf.ttf
%{_mandir}/man1/*.1*
%{_mandir}/man5/unicharambigs.5*
%{_mandir}/man5/unicharset.5*

%files -n lib%{name}
%defattr(-,root,root)
%dir %{_datadir}/tessdata
#%{_datadir}/tessdata/eng.*
#%{_datadir}/tessdata/osd.traineddata
%dir %{_datadir}/tessdata/configs
%{_datadir}/tessdata/configs/*
%dir %{_datadir}/tessdata/tessconfigs
%{_datadir}/tessdata/tessconfigs/*
%{_libdir}/libtesseract.so.*
#%doc AUTHORS COPYING

%files -n lib%{name}-devel
%defattr(-,root,root)
%dir %{_includedir}/tesseract
%{_includedir}/tesseract/*.h
%{_libdir}/libtesseract.a
%{_libdir}/libtesseract.so
%{_libdir}/pkgconfig/tesseract.pc
#%doc ChangeLog NEWS README

%package i18n-ar
Summary:       Arabic language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-ar
Arabic language support for Tesseract.

#%files i18n-ar
#%defattr(-,root,root)
#%{_datadir}/tessdata/ara.*

%package i18n-bg
Summary:       Bulgarian language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-bg
Bulgarian language data for Tesseract.

%files i18n-bg
%defattr(-,root,root)
%{_datadir}/tessdata/bul.traineddata

%package i18n-ca
Summary:       Catalan language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-ca
Catalan language data for Tesseract.

%files i18n-ca
%defattr(-,root,root)
%{_datadir}/tessdata/cat.traineddata

%package i18n-cs
Summary:       Czech language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-cs
Czech language data for Tesseract.

%files i18n-cs
%defattr(-,root,root)
%{_datadir}/tessdata/ces.traineddata

%package i18n-da
Summary:       Danish language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-da
Danish language data for Tesseract.

%files i18n-da
%defattr(-,root,root)
%{_datadir}/tessdata/dan-frak.traineddata
%{_datadir}/tessdata/dan.traineddata

%package i18n-de
Summary:       German language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-de
German language data for Tesseract.

%files i18n-de
%defattr(-,root,root)
%{_datadir}/tessdata/deu-frak.traineddata
%{_datadir}/tessdata/deu.traineddata

%package i18n-el
Summary:       Greek language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-el
Greek language data for Tesseract.

%files i18n-el
%defattr(-,root,root)
%{_datadir}/tessdata/ell.traineddata


%package i18n-es
Summary:       Spanish language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-es
Spanish language data for Tesseract.

%files i18n-es
%defattr(-,root,root)
%{_datadir}/tessdata/spa.*

%package i18n-fi
Summary:       Finnish language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-fi
Finnish language data for Tesseract.

%files i18n-fi
%defattr(-,root,root)
%{_datadir}/tessdata/fin.traineddata

%package i18n-fr
Summary:       French language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-fr
French language data for Tesseract.

%files i18n-fr
%defattr(-,root,root)
%{_datadir}/tessdata/fra.*

%package i18n-hu
Summary:       Hungarian language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-hu
Hungarian language data for Tesseract.

%files i18n-hu
%defattr(-,root,root)
%{_datadir}/tessdata/hun.traineddata

%package i18n-he
Summary:       Hebrew language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-he
Hebrew language support for Tesseract.

#%files i18n-he
#%defattr(-,root,root)
#%{_datadir}/tessdata/heb*

%package i18n-hi-IN
Summary:       Hindi language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-hi-IN
Hindi language support for Tesseract.

#%files i18n-hi-IN
#%defattr(-,root,root)
#%{_datadir}/tessdata/hin.*

%package i18n-id
Summary:       Indonesian language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-id
Indonesian language data for Tesseract.

%files i18n-id
%defattr(-,root,root)
%{_datadir}/tessdata/ind.traineddata

%package i18n-it
Summary:       Italian language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-it
Italian language data for Tesseract.

#%files i18n-it
#%defattr(-,root,root)
#%{_datadir}/tessdata/ita.*
#%{_datadir}/tessdata/ita_old.traineddata

%package i18n-ja
Summary:       Japanese language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-ja
Japanese language data for Tesseract.

%files i18n-ja
%defattr(-,root,root)
%{_datadir}/tessdata/jpn.traineddata

%package i18n-ko
Summary:       Korean language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-ko
Korean language data for Tesseract.

%files i18n-ko
%defattr(-,root,root)
%{_datadir}/tessdata/kor.traineddata

%package i18n-nl
Summary:       Dutch language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-nl
Dutch language data for Tesseract.

%files i18n-nl
%defattr(-,root,root)
%{_datadir}/tessdata/nld.traineddata

%package i18n-nn
Summary:       Norwegian language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-nn
Norwegian language data for Tesseract.

%files i18n-nn
%defattr(-,root,root)
%{_datadir}/tessdata/nor.traineddata

%package i18n-pl
Summary:       Polish language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-pl
Polish language data for Tesseract.

%files i18n-pl
%defattr(-,root,root)
%{_datadir}/tessdata/pol.traineddata


%package i18n-pt-BR
Summary:       Portuguese (Brazilian) language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-pt-BR
Portuguese (Brazilian) language data for Tesseract.

%files i18n-pt-BR
%defattr(-,root,root)
%{_datadir}/tessdata/por.traineddata

%package i18n-ro
Summary:       Romanian language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-ro
Romanian language data for Tesseract.

%files i18n-ro
%defattr(-,root,root)
%{_datadir}/tessdata/ron.traineddata

%package i18n-ru
Summary:       Russian language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-ru
Russian language data for Tesseract.

%files i18n-ru
%defattr(-,root,root)
%{_datadir}/tessdata/rus.*

%package i18n-sk
Summary:       Slovak language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-sk
Slovak language data for Tesseract.

%files i18n-sk
%defattr(-,root,root)
#%{_datadir}/tessdata/slk-frak.traineddata
%{_datadir}/tessdata/slk.traineddata

%package i18n-sl
Summary:       Slovenian language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-sl
Slovenian language data for Tesseract.

%files i18n-sl
%defattr(-,root,root)
%{_datadir}/tessdata/slv.traineddata

%package i18n-zh-CN
Summary:       Chinese (Simplified) language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-zh-CN
Chinese (Simplified) language data for Tesseract.

%files i18n-zh-CN
%defattr(-,root,root)
%{_datadir}/tessdata/chi_sim.traineddata

%package i18n-zh-TW
Summary:       Chinese (Traditional) language data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-zh-TW
Chinese (Traditional) language data for Tesseract.

%files i18n-zh-TW
%defattr(-,root,root)
%{_datadir}/tessdata/chi_tra.traineddata

%package i18n-other
Summary:       Other languages data for Tesseract
Group:         System/Internationalization
Requires:      %{name} = %{?epoch:%epoch:}%{version}-%{release}
Provides:      tesseractlang

%description i18n-other
Othert languages data for Tesseract.

# TODO: next time
%files i18n-other
%defattr(-,root,root)
%{_datadir}/tessdata/chr.traineddata
%{_datadir}/tessdata/lav.traineddata
%{_datadir}/tessdata/lit.traineddata
%{_datadir}/tessdata/srp.traineddata
%{_datadir}/tessdata/swe-frak.traineddata
%{_datadir}/tessdata/swe.traineddata
%{_datadir}/tessdata/tgl.traineddata
#%{_datadir}/tessdata/tha.traineddata
%{_datadir}/tessdata/tur.traineddata
%{_datadir}/tessdata/ukr.traineddata
%{_datadir}/tessdata/vie.traineddata

%changelog
* Tue Nov 12 2024 Automatic Build System <autodist@openmamba.org> 5.5.0-1mamba
- automatic version update by autodist

* Wed Jun 12 2024 Automatic Build System <autodist@openmamba.org> 5.4.1-1mamba
- automatic version update by autodist

* Sat Jun 08 2024 Automatic Build System <autodist@openmamba.org> 5.4.0-1mamba
- automatic version update by autodist

* Thu Jan 18 2024 Automatic Build System <autodist@openmamba.org> 5.3.4-1mamba
- automatic version update by autodist

* Mon Oct 09 2023 Automatic Build System <autodist@mambasoft.it> 5.3.3-1mamba
- automatic version update by autodist

* Wed Jul 12 2023 Automatic Build System <autodist@mambasoft.it> 5.3.2-1mamba
- automatic version update by autodist

* Sun Apr 02 2023 Automatic Build System <autodist@mambasoft.it> 5.3.1-1mamba
- automatic version update by autodist

* Sun Jan 08 2023 Silvan Calarco <silvan.calarco@mambasoft.it> 5.3.0-1mamba
- update to 5.3.0

* Fri Jan 31 2020 Automatic Build System <autodist@mambasoft.it> 4.1.1-1mamba
- automatic version update by autodist

* Mon Jul 08 2019 Automatic Build System <autodist@mambasoft.it> 4.1.0-1mamba
- automatic version update by autodist

* Fri Dec 28 2018 Automatic Build System <autodist@mambasoft.it> 4.0.0-1mamba
- automatic version update by autodist

* Thu Aug 10 2017 Automatic Build System <autodist@mambasoft.it> 3.05.01-1mamba
- automatic version update by autodist

* Fri May 13 2016 Automatic Build System <autodist@mambasoft.it> 3.04.01-1mamba
- automatic version update by autodist

* Fri Mar 11 2016 Silvan Calarco <silvan.calarco@mambasoft.it> 3.04.00-2mamba
- rebuilt by autoport with build requirements: libleptonica-devel>=1.73-1mamba

* Wed Sep 16 2015 Silvan Calarco <silvan.calarco@mambasoft.it> 3.04.00-1mamba
- update to 3.04.00

* Tue Aug 18 2015 Silvan Calarco <silvan.calarco@mambasoft.it> 3.02.02-2mamba
- rebuilt by autoport with build requirements: libleptonica-devel>=1.71-1mamba

* Tue Dec 11 2012 Automatic Build System <autodist@mambasoft.it> 3.02.02-1mamba
- update to 3.02.02

* Sat Jul 14 2012 Silvan Calarco <silvan.calarco@mambasoft.it> 3.01-1mamba
- package created by autospec