from wordcloud import WordCloud
Z=[]
sonad=[]
sonakogused={}
for rida in open("sonaseosed.txt", encoding="utf-8"):
  m=rida.strip().split(",")
  Z.append([int(m[1]), int(m[2]), float(m[3]), int(m[4])])
for rida in open("sonaindeksid.txt", encoding="utf-8"):
  sonad.append(rida.strip().split(",")[1])
for rida in open("sonakogused.txt", encoding="utf-8"):
  m=rida.strip().split(" ")
  try:
    sonakogused[m[0]]=int(m[1])
  except:
    print(m)

def uuriPuu(indeks, tase=0):
  if indeks<len(sonad):
    vastus=sonad[indeks]
  else:
    koht=indeks-len(sonad)
    rida=Z[koht]
#    print(tase*" "+str(indeks)+" "+str(rida[2]))
    v1=uuriPuu(rida[0], tase+1)
    v2=uuriPuu(rida[1], tase+1)
    vastus=(v1+" "+v2).strip()
#    if len(vastus.split())>50:
    if 15 < len(vastus.split())<=50 :
      sagedused={sona:sonakogused[sona] if sona in sonakogused else 1
                for sona in vastus.split()}
      WordCloud().generate_from_frequencies(sagedused).to_file("pilved5/"+str(indeks)+".png")
#  print(tase*" "+vastus)
  return vastus
    
uuriPuu(len(sonad)+len(Z)-1)