preprocess


1
2
3
4
5

#!/bin/zsh

LANG=$1
~/scripts/htmlentities 2>htmlentities.$LANG.err | ~/scripts/normalize-punctuation 2>normalize-punctuation.$LANG.err | ~/moses/scripts/tokenizer/tokenizer.perl -a -b -threads 1 -l $LANG 2>tokenizer.$LANG.err | ~/moses/scripts/tokenizer/lowercase.perl 2>lowercase.$LANG.err