Wordpress post comparison

wp_compare.py

# Compares 2 WP post tables to find published posts with the same name but different content
# Used after merging an old WP install into an existing one
import csv, difflib, hashlib, sys
file = open(1, encoding='utf-8')
main = [row for row in csv.reader(file)]
file = open(2, encoding='utf-8')
xen = [row for row in csv.reader(file)]

# There is no doubt a better way to do this using difflib, but I did this before finding out about it
main_posts = {}
for row in main:
  if row[17] == '0':
    hash = hashlib.sha1(row[4].encode("utf-8")).hexdigest()
    main_posts.update({row[5]:row[4]})

xen_posts = {}
for row in xen:
  if row[17] == '0':
    hash = hashlib.sha1(row[4].encode("utf-8")).hexdigest()
    xen_posts.update({row[5]:row[4]})

mismatch = []
match = []
for key in xen_posts:
  main_hash = main_posts.get(key)
  xen_hash = xen_posts.get(key)
  if main_hash != xen_hash:
    mismatch.append(key)
  else:
    match.append(key)

for post in mismatch:
  os.system("cls")
  print("%s: " % post)
  main_post = ["%s\n" % i for i in main_posts.get(post).splitlines()]
  xen_post = ["%s\n" % i for i in xen_posts.get(post).splitlines()]
  sys.stdout.writelines([i for i in difflib.ndiff(main_post, xen_post) if not i.startswith("  ")])
  input()

# This is identical to using difflib.ndiff
#sys.stdout.writelines(difflib.Differ().compare(main_post, xen_post))