# TODO:
-# - warnings at compile stage about pointer size on amd64 - needs check
-# - build dynamic library, not the static one
+# - tensorflow support
#
-%define lang_version 2.00
+# Conditional build:
+%bcond_without openmp # OpenMP support
+%bcond_with opencl # OpenCL support
Summary: Tesseract Open Source OCR Engine
Summary(pl.UTF-8): Tesseract - silnik OCR o otwartych źródłach
Name: tesseract
-Version: 2.04
+Version: 5.3.4
Release: 1
License: Apache v2.0
Group: Applications/Graphics
-Source0: http://tesseract-ocr.googlecode.com/files/%{name}-%{version}.tar.gz
-# Source0-md5: b44eba1a9f4892ac62e484c807fe0533
-Source1: http://tesseract-ocr.googlecode.com/files/%{name}-%{lang_version}.eng.tar.gz
-# Source1-md5: b8291d6b3a63ce7879d688e845e341a9
-Source2: http://tesseract-ocr.googlecode.com/files/%{name}-%{lang_version}.fra.tar.gz
-# Source2-md5: 64896b462e62572a3708bb461820126c
-Source3: http://tesseract-ocr.googlecode.com/files/%{name}-%{lang_version}.ita.tar.gz
-# Source3-md5: 2759e1dae91a989a43490ff4c2253a4b
-Source4: http://tesseract-ocr.googlecode.com/files/%{name}-%{lang_version}.deu.tar.gz
-# Source4-md5: 609d91b1ae3759a756b819b5d8403653
-Source5: http://tesseract-ocr.googlecode.com/files/%{name}-%{lang_version}.spa.tar.gz
-# Source5-md5: bc26a777b2384613895677cb8e61ca75
-Source6: http://tesseract-ocr.googlecode.com/files/%{name}-%{lang_version}.nld.tar.gz
-# Source6-md5: b2f6ede182cea4bbfffd3b040533ce58
-Patch0: %{name}-no-java.patch
-Patch1: %{name}-gcc44.patch
-URL: http://code.google.com/p/tesseract-ocr/
+#Source0Download: https://github.com/tesseract-ocr/tesseract/releases
+Source0: https://github.com/tesseract-ocr/tesseract/archive/%{version}/%{name}-%{version}.tar.gz
+# Source0-md5: 53de961804ccbcb83b73122f721d8f83
+URL: https://github.com/tesseract-ocr/
+%{?with_opencl:BuildRequires: OpenCL-devel}
+BuildRequires: asciidoc
+BuildRequires: autoconf >= 2.69
BuildRequires: automake
+BuildRequires: cairo-devel
+BuildRequires: curl-devel
+BuildRequires: leptonlib-devel >= 1.74
+BuildRequires: libarchive-devel
+%{?with_openmp:BuildRequires: libgomp-devel}
+BuildRequires: libicu-devel >= 52.1
+BuildRequires: libstdc++-devel >= 6:7
BuildRequires: libtiff-devel
+BuildRequires: libtool >= 2:2
+BuildRequires: libxslt-progs
+BuildRequires: pango-devel >= 1:1.38.0
+Requires: leptonlib >= 1.74
+Suggests: tesseract-data >= 3
BuildRoot: %{tmpdir}/%{name}-%{version}-root-%(id -u -n)
%description
latach 1985-1995. W 1995 roku był jednym z 3 najlepszych wg UNLV.
Źródła zostały uwolnione przez HP i UNLV w 2005 roku.
-%package lang-de
-Summary: German language data for Tesseract
-Summary(pl.UTF-8): Dane języka niemieckiego dla Tesseracta
+%package training
+Summary: Tesseract training tools
+Summary(pl.UTF-8): Narzędzia treningowe Tesseracta
Group: Applications/Graphics
Requires: %{name} = %{version}-%{release}
-Obsoletes: tesseract-deu
+Requires: pango >= 1:1.38.0
-%description lang-de
-This package contains the data files required to recognize German
-language.
+%description training
+This package contains the Tesseract training tools.
-%description lang-de -l pl.UTF-8
-Ten pakiet zawiera pliki danych potrzebne do rozpoznawania języka
-niemieckiego.
+%description training -l pl.UTF-8
+Ten pakiet zawiera programy do trenowania Tesseracta.
-%package lang-en
-Summary: English language data for Tesseract
-Summary(pl.UTF-8): Dane języka angielskiego dla Tesseracta
-Group: Applications/Graphics
-Requires: %{name} = %{version}-%{release}
-Obsoletes: tesseract-eng
-
-%description lang-en
-This package contains the data files required to recognize English
-language.
-
-%description lang-en -l pl.UTF-8
-Ten pakiet zawiera pliki danych potrzebne do rozpoznawania języka
-angielskiego.
-
-%package lang-es
-Summary: Spanish language data for Tesseract
-Summary(pl.UTF-8): Dane języka hiszpańskiego dla Tesseracta
-Group: Applications/Graphics
-Requires: %{name} = %{version}-%{release}
-Obsoletes: tesseract-spa
-
-%description lang-es
-This package contains the data files required to recognize Spanish
-language.
-
-%description lang-es -l pl.UTF-8
-Ten pakiet zawiera pliki danych potrzebne do rozpoznawania języka
-hiszpańskiego.
-
-%package lang-fr
-Summary: French language data for Tesseract
-Summary(pl.UTF-8): Dane języka francuskiego dla Tesseracta
-Group: Applications/Graphics
-Requires: %{name} = %{version}-%{release}
-Obsoletes: tesseract-fra
-
-%description lang-fr
-This package contains the data files required to recognize French
-language.
-
-%description lang-fr -l pl.UTF-8
-Ten pakiet zawiera pliki danych potrzebne do rozpoznawania języka
-francuskiego.
-
-%package lang-it
-Summary: Italian language data for Tesseract
-Summary(pl.UTF-8): Dane języka włoskiego dla Tesseracta
-Group: Applications/Graphics
+%package devel
+Summary: Header files for Tesseract libraries
+Summary(pl.UTF-8): Pliki nagłówkowe bibliotek Tesseracta
+Group: Development/Libraries
Requires: %{name} = %{version}-%{release}
-Obsoletes: tesseract-ita
-
-%description lang-it
-This package contains the data files required to recognize Italian
-language.
+Requires: curl-devel
+Requires: leptonlib-devel >= 1.74
+Requires: libarchive-devel
+Requires: libstdc++-devel >= 6:7
-%description lang-it -l pl.UTF-8
-Ten pakiet zawiera pliki danych potrzebne do rozpoznawania języka
-włoskiego.
-
-%package lang-nl
-Summary: Dutch language data for Tesseract
-Summary(pl.UTF-8): Dane języka holenderskiego dla Tesseracta
-Group: Applications/Graphics
-Requires: %{name} = %{version}-%{release}
-Obsoletes: tesseract-nl
+%description devel
+This package contains the development header files necessary to
+develop applications using Tesseract API.
-%description lang-nl
-This package contains the data files required to recognize Dutch
-language.
+%description devel -l pl.UTF-8
+Ten pakiet zawiera pliki nagłówkowe potrzebne do tworzenia programów
+wykorzystujących API Tesseracta.
-%description lang-nl -l pl.UTF-8
-Ten pakiet zawiera pliki danych potrzebne do rozpoznawania języka
-holenderskiego.
+%package static
+Summary: Static Tesseract libraries
+Summary(pl.UTF-8): Statyczne biblioteki Tesseracta
+Group: Development/Libraries
+Requires: %{name}-devel = %{version}-%{release}
-%package devel
-Summary: Tesseract - Development header files and libraries
-Summary(pl.UTF-8): Tesseract - Pliki nagłówkowe i biblioteki dla programistów
-Group: Development/Libraries
+%description static
+Static Tesseract libraries.
-%description devel
-This package contains the development header files and libraries
-necessary to develop applications using Tesseract.
+%description static -l pl.UTF-8
+Statyczne biblioteki Tesseracta.
%prep
%setup -q
-tar xzf %{SOURCE1}
-tar xzf %{SOURCE2}
-tar xzf %{SOURCE3}
-tar xzf %{SOURCE4}
-tar xzf %{SOURCE5}
-tar xzf %{SOURCE6}
-%patch0 -p1
-%patch1 -p1
%build
-cp -f /usr/share/automake/config.sub config
-%configure
+%{__libtoolize}
+%{__aclocal}
+%{__autoconf}
+%{__autoheader}
+%{__automake}
+%configure \
+ %{?with_opencl:--enable-opencl} \
+ %{!?with_openmp:--disable-openmp} \
+ --disable-silent-rules
%{__make}
+%{__make} training
%install
rm -rf $RPM_BUILD_ROOT
+install -d $RPM_BUILD_ROOT%{_datadir}/tessdata/script
%{__make} install \
DESTDIR=$RPM_BUILD_ROOT
+%{__make} training-install \
+ DESTDIR=$RPM_BUILD_ROOT
+
+%{__rm} $RPM_BUILD_ROOT%{_libdir}/libtesseract.la
+
%clean
rm -rf $RPM_BUILD_ROOT
+%post -p /sbin/ldconfig
+%postun -p /sbin/ldconfig
+
%files
%defattr(644,root,root,755)
-%doc AUTHORS COPYING ChangeLog README
-%attr(755,root,root) %{_bindir}/cntraining
-%attr(755,root,root) %{_bindir}/mftraining
+%doc AUTHORS ChangeLog README.md
%attr(755,root,root) %{_bindir}/tesseract
-%attr(755,root,root) %{_bindir}/unicharset_extractor
-%attr(755,root,root) %{_bindir}/wordlist2dawg
+%attr(755,root,root) %{_libdir}/libtesseract.so.*.*.*
+%attr(755,root,root) %ghost %{_libdir}/libtesseract.so.5
%dir %{_datadir}/tessdata
-%{_datadir}/tessdata/confsets
-%dir %{_datadir}/tessdata/configs
-%{_datadir}/tessdata/configs/*
-%dir %{_datadir}/tessdata/tessconfigs
-%{_datadir}/tessdata/tessconfigs/*
+%{_datadir}/tessdata/configs
+%dir %{_datadir}/tessdata/script
+%{_datadir}/tessdata/tessconfigs
+%{_datadir}/tessdata/pdf.ttf
+%{_mandir}/man1/tesseract.1*
-%files lang-de
+%files training
%defattr(644,root,root,755)
-%{_datadir}/tessdata/deu.*
-
-%files lang-en
-%defattr(644,root,root,755)
-%{_datadir}/tessdata/eng.*
-
-%files lang-es
-%defattr(644,root,root,755)
-%{_datadir}/tessdata/spa.*
-
-%files lang-fr
-%defattr(644,root,root,755)
-%{_datadir}/tessdata/fra.*
-
-%files lang-it
-%defattr(644,root,root,755)
-%{_datadir}/tessdata/ita.*
+%attr(755,root,root) %{_bindir}/ambiguous_words
+%attr(755,root,root) %{_bindir}/classifier_tester
+%attr(755,root,root) %{_bindir}/cntraining
+%attr(755,root,root) %{_bindir}/combine_lang_model
+%attr(755,root,root) %{_bindir}/combine_tessdata
+%attr(755,root,root) %{_bindir}/dawg2wordlist
+%attr(755,root,root) %{_bindir}/lstmeval
+%attr(755,root,root) %{_bindir}/lstmtraining
+%attr(755,root,root) %{_bindir}/merge_unicharsets
+%attr(755,root,root) %{_bindir}/mftraining
+%attr(755,root,root) %{_bindir}/set_unicharset_properties
+%attr(755,root,root) %{_bindir}/shapeclustering
+%attr(755,root,root) %{_bindir}/text2image
+%attr(755,root,root) %{_bindir}/unicharset_extractor
+%attr(755,root,root) %{_bindir}/wordlist2dawg
+%{_mandir}/man1/ambiguous_words.1*
+%{_mandir}/man1/classifier_tester.1*
+%{_mandir}/man1/cntraining.1*
+%{_mandir}/man1/combine_lang_model.1*
+%{_mandir}/man1/combine_tessdata.1*
+%{_mandir}/man1/dawg2wordlist.1*
+%{_mandir}/man1/lstmeval.1*
+%{_mandir}/man1/lstmtraining.1*
+%{_mandir}/man1/merge_unicharsets.1*
+%{_mandir}/man1/mftraining.1*
+%{_mandir}/man1/set_unicharset_properties.1*
+%{_mandir}/man1/shapeclustering.1*
+%{_mandir}/man1/text2image.1*
+%{_mandir}/man1/unicharset_extractor.1*
+%{_mandir}/man1/wordlist2dawg.1*
-%files lang-nl
+%files devel
%defattr(644,root,root,755)
-%{_datadir}/tessdata/nld.*
+%attr(755,root,root) %{_libdir}/libtesseract.so
+%{_includedir}/%{name}
+%{_pkgconfigdir}/tesseract.pc
+%{_mandir}/man5/unicharambigs.5*
+%{_mandir}/man5/unicharset.5*
-%files devel
+%files static
%defattr(644,root,root,755)
-%dir %{_includedir}/%{name}
-%{_includedir}/%{name}/*.h
-%{_libdir}/*.a
+%{_libdir}/libtesseract.a
projects / packages / tesseract.git / blobdiff