## With thanks to http://www.linuxjournal.com/article/9347
import zipfile
from xml.dom.minidom import parseString
if __name__ == '__main__' :
import sys
if len(sys.argv) < 2 or len(sys.argv) > 3 :
print "Usage : %s </path/to/file.ods> [</path/to/output.xml>]" %(sys.argv[0])
exit
outfile = "content.xml"
if len(sys.argv) == 3 :
outfile = sys.argv[2]
infile = sys.argv[1]
else :
infile = sys.argv[1]
output = open(outfile, "w")
file = zipfile.ZipFile(infile)
for name in file.infolist() :
if name.orig_filename == 'content.xml' :
bh = file.read(name.orig_filename)
doc = parseString(bh)
# there are rows
rows = doc.getElementsByTagName('table:table-row')
for row in rows :
# each row has a columns
output.write("<row>n")
cols = row.childNodes
for col in cols :
# each col has a cell
output.write("t<column>n")
cells = col.childNodes
for cell in cells :
# each cell has some text
texts = cell.childNodes
for text in texts :
output.write("tt<value>n")
if text.nodeType == text.TEXT_NODE :
output.write("ttt")
output.write(text.nodeValue.encode('utf-8'))
output.write("n")
output.write("tt</value>n")
output.write("t</column>n")
output.write("</row>n")
output.close()