Compiling Unitex Corpus Processor

27 Oct 2012

Unitex Corpus Processor is a very useful software component for some NLP tasks.

If you have ever needed to compile it as a dynamic library in Unix, with the following command:

make 64BITS=yes LIBRARY=yes

maybe you have found an error similar to this when compiling Unitex 2.1:

g++ *.o -Wall -W -O3 -fPIC -D_NOT_UNDER_WINDOWS -DUNITEX_LIBRARY
-I"../include_tre" -Ilibtre/include -I".."  -Llibtre/lib -ltre -lpthread
-shared -o ../bin/libunitex.so/usr/bin/ld: libtre/lib/libtre.a(regcomp.o):
relocation R_X86_64_32 against `a local symbol' can not be used when making
a shared object; recompile with -fPIC
libtre/lib/libtre.a: could not read symbols: Bad value
collect2: ld returned 1 exit status
make: *** [.libunitex] Error 1

Or maybe your error looks like this (Unitex 3.0):

g++ *.o -Wall -O3 -fPIC -D_NOT_UNDER_WINDOWS -DUNITEX_LIBRARY
-I"../include_tre" -Ilibtre/include -I".."  -Llibtre/lib -ltre   -lpthread
-shared -o ../bin/libunitex.so
/usr/bin/ld: libtre/lib/libtre.a(regerror.o): relocation R_X86_64_32S
against `.rodata' can not be used when making a shared object;
recompile with -fPIC
libtre/lib/libtre.a: could not read symbols: Bad value
collect2: ld returned 1 exit status
make: *** [.libunitex] Error 1

I could solve this issue unzipping and installing the "tre" library included in Unitex distribution (<UNITEX_DIR>/Src/C++/tre-0.8.0.tar.gz):

cd ..
tar xvfz tre-0.8.0.tar.gz
cd tre-0.8.0

./configure
make

And, as "root":

make install

Then, return to <UNITEX_DIR>/Src/C++/build and execute the following command (all in one line):

g++ *.o -Wall -W -O3 -fPIC -D_NOT_UNDER_WINDOWS -DUNITEX_LIBRARY -I"../include_tre"
-Ilibtre/include -I".."  -L/usr/local/lib -ltre -lpthread -shared
-o ../bin/libunitex.so

If all goes well, you will have your Unitex library in <UNITEX_DIR>/Src/C++/bin/libunitex.so

Hope it helps!

This entry was tagged as nlp unitex issues unix