ljdump.py

   1 import codecs, md5, os, pprint, sys, xml.dom.minidom, xmlrpclib
   2 from xml.sax import saxutils
   3
   4 def dochallenge(params, password):
   5     challenge = server.LJ.XMLRPC.getchallenge()
   6     params.update({
   7         'auth_method': "challenge",
   8         'auth_challenge': challenge['challenge'],
   9         'auth_response': md5.new(challenge['challenge']+md5.new(password).hexdigest()).hexdigest()
  10     })
  11     return params
  12
  13 def dumpelement(f, name, e):
  14     f.write("<%s>\n" % name)
  15     for k in e.keys():
  16         if isinstance(e[k], {}.__class__):
  17             dumpelement(f, k, e[k])
  18         else:
  19             s = unicode(str(e[k]), "UTF-8")
  20             f.write("<%s>%s</%s>\n" % (k, saxutils.escape(s), k))
  21     f.write("</%s>\n" % name)
  22
  23 def writedump(fn, event):
  24     f = codecs.open(fn, "w", "UTF-8")
  25     f.write("""<?xml version="1.0"?>\n""")
  26     dumpelement(f, "event", event)
  27     f.close()
  28
  29 config = xml.dom.minidom.parse("ljdump.config")
  30 Username = config.documentElement.getElementsByTagName("username")[0].childNodes[0].data
  31 Password = config.documentElement.getElementsByTagName("password")[0].childNodes[0].data
  32
  33 print "Fetching journal entries for: %s" % Username
  34 try:
  35     os.mkdir(Username)
  36     print "Created subdirectory: %s" % Username
  37 except:
  38     pass
  39
  40 server = xmlrpclib.ServerProxy("http://livejournal.com/interface/xmlrpc")
  41
  42 total = 0
  43 fetched = 0
  44 errors = 0
  45
  46 last = ""
  47 while True:
  48     r = server.LJ.XMLRPC.syncitems(dochallenge({
  49         'username': Username,
  50         'ver': 1,
  51         'lastsync': last,
  52     }, Password))
  53     #pprint.pprint(r)
  54     if len(r['syncitems']) == 0:
  55         break
  56     for item in r['syncitems']:
  57         if item['item'][0] == 'L':
  58             fn = "%s/%s" % (Username, item['item'])
  59             if not os.access(fn, os.F_OK):
  60                 print "Fetching journal entry %s" % item['item']
  61                 try:
  62                     e = server.LJ.XMLRPC.getevents(dochallenge({
  63                         'username': Username,
  64                         'ver': 1,
  65                         'selecttype': "one",
  66                         'itemid': item['item'][2:],
  67                     }, Password))
  68                     writedump(fn, e['events'][0])
  69                     fetched += 1
  70                 except xmlrpclib.Fault, x:
  71                     print "Error getting item: %s" % item['item']
  72                     pprint.pprint(x)
  73                     errors += 1
  74         last = item['time']
  75         total += 1
  76 print "%d total entries" % total
  77 print "%d fetched entries" % fetched
  78 if errors > 0:
  79     print "%d errors" % errors