BIDI_ML_INFO_EXTRACTION/BiddingKG/deepdive/test/gen_vec.py

#coding:utf8

import codecs
import pandas as pd

file = "C:\\Users\\User\\Desktop\\sgns.merge.word.txt"
out = "vec.tsv"

with codecs.open(file,"r",encoding="utf8") as f:
    with codecs.open(out,"w",encoding="utf8") as out:
        first = True
        while(True):
            line = f.readline().strip()
            if not line:
                break
            line_split = line.split()
            if first:
                sum = line_split[0]
                vims = line_split[1]
                first = False
                continue
            else:
                word = line_split[0]
                vec = line_split[1:]
            out.write(word+"\t")
            out.write("{"+",".join(vec)+"}")
            out.write("\n")
        out.flush()
        out.close()
        f.close()