s426086-mlworkshops/countlines-metrics.sh

#!/bin/sh
#zad 1 - liczenie lini
wc -l wikiniews_results.tsv > countlines_result.txt

#zad 3 
#dane do txt
cut --fields=2 wikiniews_results.tsv > hypothesis.txt
cut --fields=3 wikiniews_results.tsv > reference.txt
#do trn
awk 'BEGIN{FS=OFS="\t"}{print $0,"(sp1_"NR")"}' < hypothesis.txt > hypothesis.trn
awk 'BEGIN{FS=OFS="\t"}{print $0,"(sp1_"NR")"}' < reference.txt > reference.trn

#Sclite sluzy do obliczenia WER
sclite -f 0 -r reference.trn trn -h hypothesis.trn trn -e utf-8 -i rm -o all stdout | grep Score > results

#pobranie #C #S #D #I do obliczenia WER i oblicznie
awk 'BEGIN{OFS="\t"}{print $6, $7, $8, $9}' < results | awk '{ print ($2 + $3 + $4)/($2 + $3 + $1) }' > wer.tsv

paste wikiniews_results.tsv wer.tsv > wikiniews_results.tsv
# tail -n 49 wer.txt >> wer_tmp.txt
# rm wer.txt
# mv wer_tmp.txt wer.txt
# tail -n 49 srr.txt >> srr_tmp.txt
# rm srr.txt
# mv srr_tmp.txt srr.txt
awk 'BEGIN{FS=OFS="\t"}{ bad = bad + $1; all++ }END{ print bad/all }' < wer.tsv >> wer.txt 
awk 'BEGIN{FS=OFS="\t"}{ if ( $1 == 0 ) good++; all++ }END{ print good/all }' < wer.tsv >> srr.txt