import json v={} for tyyp in ["gpe", "loc", "org", "per", "prd"]: for rida in open(tyyp+"_kustutada_ja_kat_asendused.csv", encoding="utf-8").readlines()[1:]: m=rida.strip().split(",") if len(m)!=4: raise Exception("pikkus "+len(m)) voti=tyyp+"_"+m[0].replace(" ", "_") if m[2]=="" and m[3]=="": v[voti]="" if m[2]=="" and m[3]!="": v[voti]=(m[3]+"_"+m[0]).lower().replace(" ", "_") if m[2]!="" and m[3]!="": v[voti]=(m[3]+"_"+m[2]).lower().replace(" ", "_") if m[2]!="" and m[3]=="": v[voti]=(tyyp+"_"+m[2]).lower().replace(" ", "_") #v[m[0]] for rida in open(tyyp+"_lemma_asendused.csv", encoding="utf-8").readlines()[1:]: m=rida.strip().split(",") if len(m)!=4: raise Exception("pikkus "+len(m)) voti=tyyp+"_"+m[0].replace(" ", "_") vastus="" if m[2]=="" and m[3]=="": vastus="" if m[2]=="" and m[3]!="": vastus=(m[3]+"_"+m[0]).lower().replace(" ", "_") if m[2]!="" and m[3]!="": vastus=(m[3]+"_"+m[2]).lower().replace(" ", "_") if m[2]!="" and m[3]=="": vastus=(tyyp+"_"+m[2]).lower().replace(" ", "_") if voti in v: if vastus!=v[voti]: raise Exception("erinevus "+voti) else: v[voti]=vastus with open("asendused1.json", "w", encoding="utf-8") as f2: json.dump(v, f2, ensure_ascii=False, indent=2)