I don't think Beautiful soup is up to cracking oo3's xml format; it doesn't respect nesting.

This commit is contained in:
Thomas Gideon 2010-09-26 17:00:25 -04:00
parent 22971e87d3
commit 170c76ca08

View file

@ -1,2 +1,25 @@
#!/usr/bin/python #!/usr/bin/python
import gzip
from BeautifulSoup import BeautifulStoneSoup
import sys
def __main(oo_file):
g = open(oo_file)
o = open('output.xml', 'w')
try:
soup = BeautifulStoneSoup(g)
count = 0
for item in soup.outline.root:
if item.parent != soup.outline.root:
continue
count = count + 1
print count
finally:
g.close()
o.close()
if __name__ == "__main__":
__main(sys.argv[1])