s441471-mlworkshops/calc_wer.py

import sys
import pandas as pd

correct, insertion, deletion, substitution = [] , [] , [] , []
all_wers = []


srr = 0


def wer(s,d,c,i):
    return (s + d + i )/(s + d + c)

for line in sys.stdin:
    if line.startswith('Scores'):
        line = line.split(' ') [5:]
        c = int(line[0])
        s = int(line[1])
        d = int(line[2])
        i = int(line[3])

        if d == 0 and s == 0 and i == 0:
            srr +=1

        correct.append(c)
        substitution.append(s)
        deletion.append(d)
        insertion.append(i)
        all_wers.append(wer(s,d,c,i))




all_c = sum(correct)
all_s = sum(substitution)
all_d = sum(deletion)
all_i = sum(insertion)

total_wer = wer(all_s,all_d,all_c,all_i)

print(total_wer)

data = pd.read_csv('wikiniews_results.tsv',sep = '\t',header=None)
df = pd.DataFrame(data)
df.insert(4,'',all_wers)
df.to_csv('wikiniews_results.tsv',sep = '\t')

srr = srr/len(all_wers)

with open('wer_srr.txt', 'w+') as f:
    f.write("{} {}".format(srr,total_wer))