Apertium on openSUSE

Introduction

This note explains how to install the latest svn version of Apertium on a stock openSUSE 12.3 Milestone 1. Apertium is installed here into a directory in the user's homedir - in my case it was /home/kevin/downloads/linguistics, but it can be anything you choose. In this note, this choice will be referred to as [homedir].

(For general installation help, see Installation.)

Install required packages

If you choose the development pattern at or after your openSUSE install (using YaST, Smart, zypper or whatever), you will get most of the packages required for development, and may not have to add all of those listed below. Ensure that the following packages are present:

• subversion
• libxml2 and libxml2-devel
• libxslt1 and libxslt-devel
• libtool
• libcpcre1 and pcre-devel
• flex
• make
• automake
• autoconf
• gcc-c++

Download Apertium

You can download the full package or only the bare essentials. I`m recommending you the second option and later downloading language pairs that you are interested in.

Open a terminal, go to your [homedir] and checkout the latest version:

cd [homedir]
svn co https://apertium.svn.sourceforge.net/svnroot/apertium

This will take some time to fully download, and at the end you should have a directory [homedir]/apertium/ with some directiories in it, some containing tools and some containing language-pair data.

If you want to download only the bare essentials to build the (eg) Welsh translator, the commands would be:

cd [homedir]
svn co https://apertium.svn.sourceforge.net/svnroot/apertium/trunk/apertium
svn co https://apertium.svn.sourceforge.net/svnroot/apertium/trunk/lttoolbox
svn co https://apertium.svn.sourceforge.net/svnroot/apertium/trunk/apertium-cy-en

You can replace cy-en by different language pair. For the list of language pairs go here.

Set up your install dirs

From [homedir]:

mkdir local

Install lttoolbox

From [homedir]:

cd lttoolbox/
./autogen.sh

Obviously, replace [homedir] here and in subsequent instances by your real path.

Set the local dir as part of the configure process:

./configure --prefix=[homedir]/local
make
make install

Your [homedir]/local dir should now contain four folders: bin, include, lib, and share.

Install apertium

We will set PKG_CONFIG_PATH as part of the configure process. We will also set up a library path to ensure Apertium can find the lttoolbox libraries we just built.

From [homedir]:

cd apertium
export LD_LIBRARY_PATH=[homedir]/local/lib
export PKG_CONFIG_PATH=[homedir]/local/lib/pkgconfig
export PATH=[homedir]/local/bin:$PATH
./autogen.sh 
./configure --prefix=[homedir]/local
make
make install

In case your architecture is x86_64, then you should check if there is directory lib or lib64, and if it is lib64, you have to do the follow instructions:

From [homedir]:

cd apertium/trunk/apertium
export LD_LIBRARY_PATH=[homedir]/local/lib64
export PKG_CONFIG_PATH=[homedir]/local/lib64/pkgconfig
export PATH=[homedir]/local/bin:$PATH
./autogen.sh 
./configure --prefix=[homedir]/local
make
make install

Your [homedir]/local/bin dir should now contain a number of apertium binaries.

Install VISL CG3

Apertium for Welsh now requires this constraint grammar package to help with disambiguation. For install instructions go here.

Tips: <prefix> - it must be [homedir]/local/

tmalloc is required, so you have to install also the last package.

The packages can have different names than in that instructions, for example: boost-devel instead of libboost-dev gperftools-devel instead of libgoogle-perftools-dev

Install language-pair data

In this instance we will install the cy-en (Welsh-English) pair. First, we need to ensure that the system can find the new Apertium binaries:

export PATH=[homedir]/local/bin:$PATH

(This can be made permanent later, by adding the path to ~/.bash_profile, or whatever is your preferred method for doing this.)

From [homedir] you go to directory with choosen language-pair. The path will be different for full package and different for minimal installation:

Full package:

cd [homedir]/trunk/apertium-cy-en

Minimal installation:

cd apertium-cy-en

In this directory:

export LD_LIBRARY_PATH=[homedir]/local/lib
export PKG_CONFIG_PATH=[homedir]/local/lib/pkgconfig 
export PATH=[homedir]/local/bin:$PATH
./autogen.sh 
./configure --prefix=[homedir]/local
make
make install

In case your architecture is x86_64, then you should check if there is directory lib or lib64, and if it is lib64, you have to do the follow instructions in directory with choosen language-pair:

From [homedir]:

export LD_LIBRARY_PATH=[homedir]/local/lib64
export PKG_CONFIG_PATH=[homedir]/local/lib64/pkgconfig 
export PATH=[homedir]/local/bin:$PATH
./autogen.sh 
./configure --prefix=[homedir]/local
make
make install

Your [homedir]/local/share/apertium should now contain a new folder apertium-cy-en, and the modes folder should contain a file for the Welsh-English language direction (cy-en.mode).

Note that if you are using an AMD64 processor, you must use version 3.0.4 or above of Apertium. Versions below this have a bug which will mean the make fails.

Testing the new install

From [homedir]:

echo "prawf" | apertium cy-en

You should see test returned. Currently there is no English-Welsh direction.