Print worksheet from an excel workbook (.xls files).

xls2tsv.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-

'''xls2tsv - print the contents of a worksheet from an excel workbook.'''

import sys
import xlrd
import optparse

parser = optparse.OptionParser()

parser.add_option('--sheet', 
                  default='Sheet1', 
                  help='name of worksheet to be printed')
parser.add_option('--sheetnames', 
                  default=False, 
                  action='store_true', 
                  help='print all sheets in Excel file')

opts, files = parser.parse_args()

if opts.sheetnames:
    '''Print names of worksheets in file(s)'''
    if len(files) == 0:
        sys.exit('Need to specify input file(s)')
    else:
        for file in files:
            try:
                wb = xlrd.open_workbook(file)
                print "\n".join(wb.sheet_names())
            except:
                sys.exit('bailed on ' + file)
else:
    import re

    dash_pt = re.compile(u"[–—]", re.UNICODE)
    apostro_pt = re.compile(u"’", re.UNICODE)
    apostro_t_pt = re.compile(u"稚", re.UNICODE)
    ae_apostro_pt = re.compile(u"Æ", re.UNICODE)
    grave_e_pt = re.compile(u"é", re.UNICODE)
    umlaut_e_pt = re.compile(u"ë", re.UNICODE)
    cap_a_pt = re.compile(u"√¢", re.UNICODE)
    phi_e_pt = re.compile(u"Φ", re.UNICODE)
    rho_a_pt = re.compile(u"Γ", re.UNICODE)
    theta_e_pt = re.compile(u"Θ", re.UNICODE)
    unknown_pt = re.compile(u"‚ïü", re.UNICODE)

    def asciify(string):
        '''Replace any non-ascii chars with appropriate chars.'''
        try:
            return str(string)
        except UnicodeEncodeError:
            string = dash_pt.sub("–", string)
            string = apostro_pt.sub("'", string)
            string = apostro_t_pt.sub("'t", string)
            string = ae_apostro_pt.sub("'", string)
            string = grave_e_pt.sub("e", string)
            string = umlaut_e_pt.sub("e", string)
            string = cap_a_pt.sub("a", string)
            string = phi_e_pt.sub("e", string)
            string = rho_a_pt.sub("a", string)
            string = theta_e_pt.sub("e", string)
            string = unknown_pt.sub("", string)
            return string.encode('utf-8', 'replace')

    if len(files) == 0:
        sys.exit('Need to specify input file(s)')
    else:
        for file in files:
            try:
                wb = xlrd.open_workbook(file)
            except:
                sys.exit('bailed on ' + file)
            try:
                sheet = wb.sheet_by_name(opts.sheet)
            except:
                sys.exit("could not open worksheet " + opts.sheet + " in " + file)
            for row in range(sheet.nrows):
                values = [asciify(i) for i in sheet.row_values(row)]
                try:
                    print "\t".join(values)
                except:
                    sys.exit("non-ascii values in " + file)