Script to read Wordpress export file, fetch images via URL, and write locally.

extract_wordpress_images.py

import sys
import os
import urllib
from bs4 import BeautifulSoup

#TODO: pass in xml path and output folder as args
xml = open('C:\\Workspace\\blog\\geomusings_export.xml').read()
doc = BeautifulSoup(xml)

#attachments could technically be something other than images
#extract all attachment elements from Wordpress export file
for item in doc.findAll('wp:attachment_url'):
    #Contens will be URL
    lnk = item.contents[0]
    #Get just the file name
    fname = os.path.basename(lnk)
    print(os.path.basename(lnk))
    #TODO: add some logic to handle duplicate file names
    #Get file via URL and write locally
    f = open('C:\\Workspace\\blog\\images\\' + fname,'wb')
    f.write(urllib.urlopen(lnk).read())
    f.close()