- use wvHtml instead of catdoc by default: it retains formatting

[mnogosearch-rpm] / mnogosearch.spec

%global php_apiver  %((echo 0; php -i 2>/dev/null | sed -n 's/^PHP API => //p') | tail -1)
%global php_extdir  %(php-config --extension-dir 2>/dev/null || echo "undefined")
%global php_version %(php-config --version 2>/dev/null || echo 0)

# do not add php module to provides, it is no regular library
%{?filter_setup:
%filter_provides_in %{php_extdir}/.*\.so$
%filter_setup
}

Summary:        Web indexing and search system for a small domain or intranet
Name:           mnogosearch
Version:        3.3.11
Release:        2
License:        GPLv2+
Group:          Applications/Internet
URL:            http://www.mnogosearch.org/

# md5sum 906d5e1e21ba00024574f57548d32029
Source0:        http://www.mnogosearch.org/Download/mnogosearch-%{version}.tar.gz

Source1:        mnogosearch-reindex
Source2:        mnogosearch.crond
Source3:        httpd.conf

BuildRoot:      %{_tmppath}/%{name}-%{version}-%{release}-root-%(%{__id_u} -n)

Requires:       %{name}-libs = %{version}-%{release}
Requires:       /usr/bin/wvHtml
Requires:       /usr/bin/xlhtml
Requires:       /usr/bin/ppthtml
Requires:       /usr/bin/pdftotext
Requires(pre):  shadow-utils

# user "apache" is added to mnogosearch group so we need the user to be present
Requires(post): httpd

BuildRequires:  autoconf
BuildRequires:  automake
BuildRequires:  libtool
BuildRequires:  openssl-devel
BuildRequires:  zlib-devel
BuildRequires:  expat-devel
BuildRequires:  openssl-devel
BuildRequires:  openjade
BuildRequires:  docbook-utils
BuildRequires:  mysql-devel
BuildRequires:  readline-devel
BuildRequires:  ncurses-devel
BuildRequires:  sqlite-devel
BuildRequires:  unixODBC-devel
BuildRequires:  postgresql-devel
BuildRequires:  freetds-devel
BuildRequires:  firebird-devel
BuildRequires:  catdoc
BuildRequires:  xlhtml

# build is done with unit testing by default, but it can be disabled
# by --without tests to save time
%define with_tests  %{?_without_tests:  0} %{?!_without_tests:  1}

%description
mnoGoSearch (formerly known as UdmSearch) is a full-featured Web search engine
that you can use to build search engines over HTTP, HTTPS, FTP, and NTTP
servers, local files, and database big text fields. It supports Oracle, MS SQL
Server, MySQL, PostgreSQL, InterBase/Firebird, Openlink Virtuoso, Intersystems
Cach, iODBC, EasySoft ODBC, and unixODBC database backends. It has XML, HTML,
and TEXT built-in support, and external converters support for other document
types. An automatic language/charset guesser for more 70 language/charset
combinations is included, along with basic authorization support, and you may
index password-protected intranet HTTP servers with proxy authorization
support.

mnoGoSearch is built with support for mysql (including fulltext extension),
sqlite, postgresql, firebird, unixODBC and freetds.

%package libs
Summary:        Libraries for %{name}
Group:          System Environment/Libraries

%description libs
This package contains the %{name} library files.

%package devel
Summary:        Development headers and libraries for %{name}
Group:          Development/Libraries
Requires:       %{name} = %{version}-%{release}

%description devel
This package contains the %{name} development files.

%package php
Summary:        PHP-extension for %{name}
Group:          Applications/Internet
Requires:       %{name}-libs = %{version}-%{release}
BuildRequires:  php-devel
BuildRequires:  php-cli

%if 0%{?php_zend_api}
Requires:       php(zend-abi) = %{php_zend_api}
Requires:       php(api) = %{php_core_api}
%else
Requires:       php-api = %{php_apiver}
%endif

%description php
This package contains the PHP extension for %{name}

%prep
%setup -q -n mnogosearch-%{version}

%build
autoreconf -fi

%configure \
    --sysconfdir=%{_sysconfdir}/mnogosearch  \
    --datadir=%{_datadir}/mnogosearch  \
    --localstatedir=%{_localstatedir}/lib/mnogosearch \
    --includedir=%{_includedir}/mnogosearch \
    --enable-syslog \
    --enable-syslog=LOG_LOCAL6 \
    --enable-parser \
    --enable-mp3 \
    --enable-file \
    --enable-http \
    --enable-ftp \
    --enable-news \
    --with-openssl=%{_prefix} \
    --with-zlib \
    --with-readline \
    --with-extra-charsets=all \
    --with-mysql \
    --enable-mysql-fulltext-plugin \
    --with-pgsql \
    --with-sqlite3 \
    --with-unixODBC \
    --with-freetds=%{_prefix} \
    --with-ibase \
    --with-docs \
    --enable-fhs-layout

make %{?_smp_mflags}

%install
rm -rf %{buildroot}

%{__make} install \
    DESTDIR=$RPM_BUILD_ROOT \
    doc_FILES='$(HTML_ALL)'

# no need to call the config files *-dist
(cd %{buildroot}%{_sysconfdir}/%{name}
for f in *-dist ; do
    mv -f $f `basename $f -dist`
done
)

# fix header of config file
sed -i 's|/usr/local/mnogosearch/sbin/indexer|%{_bindir}/indexer|g' \
    %{buildroot}%{_sysconfdir}/mnogosearch/indexer.conf

# change default configuration
sed -i 's|DBAddr.mysql://root@localhost/test/?dbmode=blob|DBAddr sqlite3://localhost/%{_localstatedir}/lib/mnogosearch/default.db/?dbmode=blob|g' \
    %{buildroot}%{_sysconfdir}/mnogosearch/indexer.conf \
    %{buildroot}%{_sysconfdir}/mnogosearch/search.htm \

sed -i 's|#LocalCharset UTF-8|LocalCharset UTF-8|g' \
    %{buildroot}%{_sysconfdir}/mnogosearch/indexer.conf

sed -i 's|#Server http://localhost/|Server http://localhost/|g' \
    %{buildroot}%{_sysconfdir}/mnogosearch/indexer.conf

# enable common external document helpers by default
# we require them so they will be installed

sed -i 's|#Mime application/msword      "text/html; charset=utf-8"  "wvHtml --charset=utf-8 $1 -"|Mime application/msword      "text/html; charset=utf-8"  "wvHtml --charset=utf-8 $1 -"|g' \
    %{buildroot}%{_sysconfdir}/mnogosearch/indexer.conf

sed -i 's|#Mime application/pdf          text/plain                 "pdftotext $1 -"|Mime application/pdf          text/plain                 "pdftotext $1 -"|g' \
    %{buildroot}%{_sysconfdir}/mnogosearch/indexer.conf

sed -i 's|#Mime application/vnd.ms-excel text/html                  "xlhtml $1"|Mime application/vnd.ms-excel text/html                  "xlhtml $1"|g' \
    %{buildroot}%{_sysconfdir}/mnogosearch/indexer.conf

sed -i 's|#Mime application/vnd.ms-powerpoint text/html             "ppthtml $1"|Mime application/vnd.ms-powerpoint text/html             "ppthtml $1"|g' \
    %{buildroot}%{_sysconfdir}/mnogosearch/indexer.conf

sed -i 's|Disallow \*.tex  \*.texi \*.xls  \*.doc  \*.texinfo|Disallow *.tex *.texi *.texinfo|g' \
    %{buildroot}%{_sysconfdir}/mnogosearch/indexer.conf

sed -i 's|Disallow \*.rtf  \*.pdf  \*.cdf  \*.ps|Disallow *.rtf  *.cdf  *.ps|g' \
    %{buildroot}%{_sysconfdir}/mnogosearch/indexer.conf

sed -i 's|Disallow \*.ai   \*.eps  \*.ppt  \*.hqx|Disallow *.ai   *.eps  *.hqx|g' \
    %{buildroot}%{_sysconfdir}/mnogosearch/indexer.conf

# the cgi is no regular binary, move it to libexec
install -d %{buildroot}%{_libexecdir}/mnogosearch
mv -f %{buildroot}%{_bindir}/*.cgi %{buildroot}%{_libexecdir}/mnogosearch/

# the .la's confuse libtool when building the php lib and they should not
# be packaged at all, so remove them
rm -f %{buildroot}%{_libdir}/*.la

install -m0755 %{SOURCE1} %{buildroot}%{_bindir}/

install -d %{buildroot}%{_sysconfdir}/cron.d/
install -m0644 %{SOURCE2} %{buildroot}%{_sysconfdir}/cron.d/mnogosearch

install -d %{buildroot}%{_sysconfdir}/httpd/conf.d/
install -m0644 %{SOURCE3} %{buildroot}%{_sysconfdir}/httpd/conf.d/mnogosearch.conf


# ---------------------
# make & install php stuff
# must be in install-section because it needs an already installed mnogosearch
cd php

/usr/bin/phpize
%configure --with-mnogosearch=%{buildroot}%{_prefix}

# get rid of rpath
sed -i 's|-Wl,-rpath,[^ ]*||g' Makefile

make %{?_smp_mflags} EXTRA_LDFLAGS="-L%{buildroot}%{_libdir}" EXTRA_CFLAGS="-I%{buildroot}%{_includedir}/mnogosearch"
make install \
    INSTALL_ROOT=$RPM_BUILD_ROOT
cd ..

install -d %{buildroot}%{_sysconfdir}/php.d
cat >%{buildroot}%{_sysconfdir}/php.d/mnogosearch.ini <<EOF
; Enable mnogosearch extension module
extension=mnogosearch.so
EOF

# end PHP stuff
# ---------------------

# create state-dir
install -d %{buildroot}%{_localstatedir}/lib/mnogosearch

# fix docs
rm -rf html; mkdir -p html
cp -p doc/*.html html/
rm -rf %{buildroot}/usr/doc

%check
%if %{with_tests}
# run unit tests with sqlite backend
touch ./msearch-test/sqlite.db
export UDM_TEST_DBADDR=sqlite3://`pwd`/msearch-test/sqlite.db/,sqlite3://`pwd`/msearch-test/sqlite2.db/
make check
%endif

%clean
rm -rf %{buildroot}

%pre
# create mnogosearch user and group
getent group mnogosearch >/dev/null || groupadd -r mnogosearch
getent passwd mnogosearch >/dev/null || \
    useradd -r -g mnogosearch -d %{_localstatedir}/lib/mnogosearch \
    -c "User the indexer of the mnoGoSearch search engine is executed with" \
    -s /sbin/nologin mnogosearch
exit 0

%post
# the cgi (executed with the apache user) must be able to read the database:
# the apache user must be in the mnogosearch group
usermod -a -G mnogosearch apache || :

# we brought our own apache config snippet and new group rights
service httpd condrestart > /dev/null 2>&1 || :

# create default mnogosearch database (only during install)
if [ $1 -eq 1 ] ; then
    /usr/bin/indexer --create >/dev/null
    chown mnogosearch:mnogosearch %{_localstatedir}/lib/mnogosearch/default.db
    # db must be group writeable to allow query tracking
    chmod 0660 %{_localstatedir}/lib/mnogosearch/default.db
fi

%postun
service httpd condrestart > /dev/null 2>&1 || :

%post libs -p /sbin/ldconfig

%postun libs -p /sbin/ldconfig

%files
%defattr(-,root,root,-)
%{_bindir}/indexer
%{_bindir}/mconv
%{_bindir}/mguesser
%{_bindir}/mnogosearch-reindex
%dir %{_libexecdir}/mnogosearch
%{_libexecdir}/mnogosearch/search.cgi
%config(noreplace) %verify(not size mtime md5) %{_sysconfdir}/cron.d/mnogosearch
%config(noreplace) %verify(not size mtime md5) %{_sysconfdir}/httpd/conf.d/mnogosearch.conf

%defattr (0644,root,root,0755)
%doc ChangeLog README README.XML README.CVS TODO COPYING AUTHORS INSTALL NEWS FAQ html doc/samples
%{_mandir}/man?/*
%{_datadir}/mnogosearch

# the config files of mnogosearch often contain database passwords -> should not be world readable
%defattr (0640,root,mnogosearch,0750)
%dir %{_sysconfdir}/mnogosearch

# indexer.conf is executable: has a correct shebang to run the indexer with this configuration
%attr(0755,root,mnogosearch) %config(noreplace) %verify(not size mtime md5) %{_sysconfdir}/mnogosearch/indexer.conf
%config(noreplace) %verify(not size mtime md5) %{_sysconfdir}/mnogosearch/langmap.conf
%config(noreplace) %verify(not size mtime md5) %{_sysconfdir}/mnogosearch/stopwords.conf
%config(noreplace) %verify(not size mtime md5) %{_sysconfdir}/mnogosearch/search.htm
%config(noreplace) %verify(not size mtime md5) %{_sysconfdir}/mnogosearch/node.xml

# default dir for search databases should not be world-readable:
# the search engine could have indexed secret files
%attr (0750,mnogosearch,mnogosearch) %dir %{_localstatedir}/lib/mnogosearch

%files libs
%defattr(-,root,root,-)
%{_libdir}/lib*-*.so

%files devel
%defattr(-,root,root,-)
%{_bindir}/udm-config
%{_includedir}/mnogosearch
%{_libdir}/*.a
%{_libdir}/libmnogosearch.so
%{_libdir}/libmnogocharset.so

%files php
%defattr(-,root,root,-)
%{_sysconfdir}/php.d/mnogosearch.ini
%{php_extdir}

%changelog
* Thu Jan 27 2011 Gerd v. Egidy <gerd@egidy.de> 3.3.11-2
- last minute-changes in the 3.3.11 release
- enable some external document helpers by default:
  we always have them because of requires
- require the binaries of the helpers used, not the package names
- use wvHtml instead of catdoc by default: it retains formatting
- don't disable the file types we now have helpers for

* Thu Jan 27 2011 Gerd v. Egidy <gerd@egidy.de> 3.3.11-1
- mnogosearch 3.3.11
- unit testing with two databases
- require catdoc and xlhtml for install and during build
- move make check into check-section
- fix timezone issue with test-blob unit test

* Thu Jan 27 2011 Gerd v. Egidy <gerd@egidy.de> 3.3.10-7.20110127cvs
- new cvs version with fix for directory names
- fix commandline options in mnogosearch-reindex

* Wed Jan 26 2011 Gerd v. Egidy <gerd@egidy.de> 3.3.10-6.20110126cvs
- import new upstream fixes for issues found during packaging
- group writeable default.db
- reindex option improvements by Alexander Barkov
- only create db on install
- create databases not with sqlite, but with the --create option of indexer

* Tue Jan 25 2011 Gerd v. Egidy <gerd@egidy.de> 3.3.10-5.20110124cvs
- update cvs version to get lib64 fix
- don't package *.la files as per Fedora review policy
- filter php module from provides
- create mnogosearch user and group during installation
- tighten file permissions to protect passwords and data
- create default configuration suitable for Fedora

* Fri Jan 21 2011 Gerd v. Egidy <gerd@egidy.de> 3.3.10-4.20110121cvs
- don't rename udm-config anymore
- better differentiation between release and cvs snapshots
- don't use $RPM_BUILD_DIR anymore

* Thu Jan 20 2011 Gerd v. Egidy <gerd@egidy.de> 3.3.10-3.20110120cvs
- make use of the new fhs configure option
- move language files to /usr/share
- move cgi to /usr/libexec
- clean up unneeded stuff left over from Mandriva
- link firebird database
- run supplied unit tests after build

* Mon Jan 17 2011 Gerd v. Egidy <gerd@egidy.de> 3.3.10-2.20110117cvs
- use current CVS as there are/will be some patches making packaging more easy

* Thu Jan 13 2011 Gerd v. Egidy <gerd@egidy.de> 3.3.10-1
- add building of the php module
- initial packaging for Fedora
- based on the Mandriva .spec by Oden Eriksson