'''
Created on Jul 5, 2017

@author: David
'''
import codecs

index_file = "f:/parll data/data/compound-indexes.txt"
infile = "f:/parll data/data/compound_tfidf.dist"

outfile = "f:/parll data/data/compound-utile.txt"

index = 0
indexes = []

with open(index_file, "r") as f:
    for l in f:
        if not l.strip():
            continue
        indexes.append(int(l.rstrip()))
        
out = codecs.open(outfile, "w", encoding="utf-8")  
      
with codecs.open(infile, "r", encoding="utf-8") as f:
    for l in f:
        if index < indexes[0]: 
            index += 1
            continue
        print(index)
        elem = l.split("\t")[0]
        indexes.pop(0)
        out.write("{}\n".format(elem))
        
out.close()