#!/bin/bash pushd `dirname $0` > /dev/null P=`pwd -P` popd > /dev/null LANG=$1 $P/no-non-printables | sed "s|[-,\.]\{4,\}|...|g" | $P/htmlentities 2>htmlentities.$LANG.err | $P/normalize-punctuation 2>normalize-punctuation.$LANG.err | $P/tokenizer-no-escape.perl -a -b -threads 1 -l $LANG 2>tokenizer.$LANG.err | $P/lowercase.perl 2>lowercase.$LANG.err