#!/usr/bin/env python

import xlrd
import MeCab
import sys

args = sys.argv

#Open excel file
book = xlrd.open_workbook(args[1])
sh = book.sheet_by_index(0)

# header
print("\t".join(('text','price')))

#Perspective with word-separation option
t = MeCab.Tagger ("-Owakati")

#About each line
for rx in range(1, sh.nrows):

    #Pick up the columns you need
    text = sh.cell_value(rowx=rx, colx=1)
    price = sh.cell_value(rowx=rx, colx=2)

    #Delete line breaks
    text = text.replace('\n','').replace('\r','')

    try:
        #Perspective and line break removal
        m = t.parse(text).replace('\n','')

        #output
        print( "\t".join((m, price)) )

    except RuntimeError as e:
        print("RuntimeError:" + e)

Recommended Posts