from collections import defaultdict
import urllib.request

vasak=defaultdict(int)
parem=defaultdict(int)
ryhmasonad=set()
vasakud=defaultdict(int)
paremad=defaultdict(int)
puuduvad=defaultdict(int)
kogupuuduvad=defaultdict(int)
molemad=defaultdict(int)
ryhmitatuid=defaultdict(int)

sonukokku=0
for rida in open("Andmed/noun_hypernyms_estwn_2.3.2.txt", encoding="utf-8"):
  m=rida.strip().split("<")
  vasak[m[0]]+=1
  parem[m[1]]+=1

koos=set(vasak.keys()).intersection(set(parem.keys()))
 
m0=urllib.request.urlopen("https://minitorn.tlu.ee/~jaagup/oma/too/23/05/word2vec/vastused/jaotus1000_marksonad.txt").read().decode("utf-8").split("\n") 
for nr in range(1000):
  for sona in m0[nr*3+1].strip().split(): ryhmasonad.add(sona)
  
kokku=0
olemas=0  
sees=set()
valjas=set()
for sona in ryhmasonad:
  kokku+=1
  if sona in vasak or sona in parem: 
    olemas+=1
    sees.add(sona)
  else:
    valjas.add(sona)

print(kokku, olemas)
f2=open("puuduvad.txt", "w", encoding="utf-8")  
print(valjas, file=f2)
f2.close()