MoinMoin: attachment:gettiki_html.py of MigrationStrategy/FromTikiWiki

   1 #!/usr/local/bin/python
   2 
   3 """
   4  get wiki pages from a tikiwiki
   5  Author: Daniela Nicklas
   6 """
   7 import sys
   8 import httplib
   9 from HTMLParser import HTMLParser
  10 import string
  11 import os
  12 
  13 # config variables
  14 
  15 wikihost="hostname"
  16 wikipath="/tiki/tiki-print.php?page="
  17 targetdir="pages"
  18 pagelistfile = "pagelist"
  19 
  20 
  21 # global variables
  22 wikiroot = "" # root for all wiki pages
  23 pagelist = [] # list of all wiki pages
  24 
  25 # get pagelist
  26 pl = open(pagelistfile,'r')
  27 pagelist = []
  28 for line in pl:
  29     pagelist.append(line.strip())
  30 pl.close
  31 
  32 print "Dumping Wiki %s to directory %s"%(wikihost,targetdir)
  33 
  34 # create directory (if necessary)
  35 if not os.access(targetdir, os.F_OK):
  36     os.mkdir(targetdir)
  37 
  38 # open connection and get the titleindex
  39 httpconnection = httplib.HTTPConnection(wikihost)
  40 
  41 # fetch pages
  42 for pagename in pagelist:
  43     httpconnection.request("GET", wikipath+pagename)
  44     response = httpconnection.getresponse()
  45     if response.status != 200:
  46         print response.status, response.reason
  47     else:
  48         # get pages
  49         print pagename
  50         pagetext = response.read()
  51         f = open(targetdir+"/"+pagename+".html", "w")
  52         f.write(pagetext)
  53         f.close()
  54    
  55 # close the connection
  56 httpconnection.close()