#!/bin/bash

#zad 1
wc -l wikiniews_results.tsv > counted_lines_result.txt

#zad 3
# dzielenie linii
cut --fields=2 wikiniews_results.tsv > hypothesis.txt
cut --fields=3 wikiniews_results.tsv > reference.txt

# tworzenie plikow 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 -f 0 -r reference.trn trn -h hypothesis.trn trn -e utf-8 -i rm -o all stdout | grep Score > results

# #correct words #substitutions #deletions #insertions do obliczen WER
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

# procent niepoprawnie rozpoznanych wyrazów
awk 'BEGIN{FS=OFS="\t"}{ bad = bad + $1; all++ }END{ print bad/all }' < wer.tsv >> wer.txt

# procent całkowicie poprawnie rozpoznanych zdań
awk 'BEGIN{FS=OFS="\t"}{ if ( $1 == 0 ) good++; all++ }END{ print good/all }' < wer.tsv >> srr.txt

# zamiana treści
tail -n 50 wer.txt > wer1.txt
tail -n 50 srr.txt > srr1.txt

rm wer.txt
rm srr.txt

mv wer1.txt wer.txt
mv srr1.txt srr.txt