summaryrefslogtreecommitdiff
path: root/preprocess
blob: 9273cfba8c0371e67b34c876bbf1c47d39c7486d (plain)
1
2
3
4
5
#!/bin/zsh

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