%if 0%{?fedora}
%global with_python3 1
%endif # fedora

%define unicode_version 6.2.0
%define srcdir emptypage-uniseg-python-8fed7c16392f

Name:           python-uniseg
Version:        0.7.0
Release:        1%{?snapshot}%{?dist}
Summary:        A pure Python module to determine Unicode text segmentation

Group:          Applications/System
License:        MIT
URL:            https://bitbucket.org/emptypage/uniseg-python
Source0:        https://bitbucket.org/emptypage/uniseg-python/get/rel-%{version}.tar.gz

# Unicode files from http://www.unicode.org/Public/6.2.0/ucd/auxiliary/
Source1001:     GraphemeBreakProperty.txt
Source1002:     GraphemeBreakTest.txt
Source1003:     LineBreak.txt
Source1004:     LineBreakTest.txt
Source1005:     SentenceBreakProperty.txt
Source1006:     SentenceBreakTest.txt
Source1007:     WordBreakProperty.txt
Source1008:     WordBreakTest.txt

BuildArch:      noarch
BuildRequires:  curl
BuildRequires:  sqlite
BuildRequires:  python-setuptools
BuildRequires:  python-devel

%global pkgdesc Uniseg is a Python module that provides functions to wrap words based on \
Unicode Character Database (UCD) version %{unicode_version}.

%description
%{pkgdesc}

%if 0%{?with_python3}
%package -n python3-uniseg
Summary:        A pure Python module to determine Unicode text segmentation
BuildRequires:  python3-devel

%description -n python3-uniseg
%{pkgdesc}
%endif # with_python3

%prep
%setup -q -n %{srcdir}

%build
# copy unicode files to the right place
%{__mkdir_p} data/%{unicode_version}/auxiliary
%{__cp} %{SOURCE1001} %{SOURCE1002} %{SOURCE1003} %{SOURCE1004} \
        %{SOURCE1005} %{SOURCE1006} %{SOURCE1007} %{SOURCE1008} \
        data/%{unicode_version}/auxiliary/

# prepare unicode files
%{__mkdir_p} csv
%{__python} tools/prop2csv.py -o csv/GraphemeClusterBreak.csv data/%{unicode_version}/auxiliary/GraphemeBreakProperty.txt
%{__python} tools/test2csv.py -p GB -o csv/GraphemeClusterBreakTest.csv data/%{unicode_version}/auxiliary/GraphemeBreakTest.txt
%{__python} tools/prop2csv.py -o csv/WordBreak.csv data/%{unicode_version}/auxiliary/WordBreakProperty.txt
%{__python} tools/test2csv.py -p WB -o csv/WordBreakTest.csv data/%{unicode_version}/auxiliary/WordBreakTest.txt
%{__python} tools/prop2csv.py -o csv/SentenceBreak.csv data/%{unicode_version}/auxiliary/SentenceBreakProperty.txt
%{__python} tools/test2csv.py -p SB -o csv/SentenceBreakTest.csv data/%{unicode_version}/auxiliary/SentenceBreakTest.txt
%{__python} tools/prop2csv.py -o csv/LineBreak.csv data/%{unicode_version}/auxiliary/LineBreak.txt
%{__python} tools/test2csv.py -p LB -o csv/LineBreakTest.csv data/%{unicode_version}/auxiliary/LineBreakTest.txt

# create unicode database
sqlite3 uniseg/ucd.sqlite3 < schema.sql ||:
sqlite3 -csv uniseg/ucd.sqlite3 ".import csv/GraphemeClusterBreak.csv GraphemeClusterBreak"
sqlite3 -csv uniseg/ucd.sqlite3 ".import csv/GraphemeClusterBreakTest.csv GraphemeClusterBreakTest"
sqlite3 -csv uniseg/ucd.sqlite3 ".import csv/WordBreak.csv WordBreak"
sqlite3 -csv uniseg/ucd.sqlite3 ".import csv/WordBreakTest.csv WordBreakTest"
sqlite3 -csv uniseg/ucd.sqlite3 ".import csv/SentenceBreak.csv SentenceBreak"
sqlite3 -csv uniseg/ucd.sqlite3 ".import csv/SentenceBreakTest.csv SentenceBreakTest"
sqlite3 -csv uniseg/ucd.sqlite3 ".import csv/LineBreak.csv LineBreak"
sqlite3 -csv uniseg/ucd.sqlite3 ".import csv/LineBreakTest.csv LineBreakTest"
sqlite3 uniseg/ucd.sqlite3 "vacuum"

%install
%{__rm} -rf %{buildroot}

%{__python2} setup.py install --root %{buildroot}

%if 0%{?with_python3}
%{__python3} setup.py install --root %{buildroot}

# strip python3 from shebang line
sed -i'' -e 's,^.*#!/usr/bin/python3.*$,#!/usr/bin/python,g' %{buildroot}%{_bindir}/uniseg-dbpath
%endif # with_python3

%files
%{python2_sitelib}/*egg-info/
%{python2_sitelib}/uniseg/
%attr(0755,root,root) %{_bindir}/uniseg-dbpath

%if 0%{?with_python3}
%files -n python3-uniseg
%{python3_sitelib}/*egg-info/
%{python3_sitelib}/uniseg/
%attr(0755,root,root) %{_bindir}/uniseg-dbpath
%endif # with_python3

%changelog
* Mon Apr 27 2015 Murilo Opsfelder Araújo <muriloo@linux.vnet.ibm.com> 0.7.0-1
- Initial build