automatically fall back to Latin-1 for old entries that aren't UTF-8

[oss/ljdump.git] / ljdump.py

diff --git a/ljdump.py b/ljdump.py
index 42f1d968df3150b1f4d3665d48aa390a4b2952a1..6a7e3a2655ff67d88814066c10332d17ef36c788 100755 (executable)
--- a/ljdump.py
+++ b/ljdump.py
@@ -4,31 +4,6 @@
 # Greg Hewgill <greg@hewgill.com> http://hewgill.com
 # Version 1.3.2
 #
-# $Id$
-#
-# This program reads the journal entries from a livejournal (or compatible)
-# blog site and archives them in a subdirectory named after the journal name.
-#
-# The configuration is read from "ljdump.config". A sample configuration is
-# provided in "ljdump.config.sample", which should be copied and then edited.
-# The configuration settings are:
-#
-#   server - The XMLRPC server URL. This should only need to be changed
-#            if you are dumping a journal that is livejournal-compatible
-#            but is not livejournal itself.
-#
-#   username - The livejournal user name. A subdirectory will be created
-#              with this same name to store the journal entries.
-#
-#   password - The account password. This password is never sent in the
-#              clear; the livejournal "challenge" password mechanism is used.
-#
-# This program may be run as often as needed to bring the backup copy up
-# to date. Both new and updated items are downloaded.
-#
-# The community http://ljdump.livejournal.com has been set up for questions
-# or comments.
-#
 # LICENSE
 #
 # This software is provided 'as-is', without any express or implied
@@ -99,7 +74,11 @@ def dumpelement(f, name, e):
         if isinstance(e[k], {}.__class__):
             dumpelement(f, k, e[k])
         else:
-            s = unicode(str(e[k]), "UTF-8")
+            try:
+                s = unicode(str(e[k]), "UTF-8")
+            except UnicodeDecodeError:
+                # fall back to Latin-1 for old entries that aren't UTF-8
+                s = unicode(str(e[k]), "cp1252")
             f.write("<%s>%s</%s>\n" % (k, saxutils.escape(s), k))
     f.write("</%s>\n" % name)
 
@@ -347,8 +326,19 @@ def ljdump(Server, Username, Password):
         print "%d errors" % errors
 
 if __name__ == "__main__":
-    config = xml.dom.minidom.parse("ljdump.config")
-    server = config.documentElement.getElementsByTagName("server")[0].childNodes[0].data
-    username = config.documentElement.getElementsByTagName("username")[0].childNodes[0].data
-    password = config.documentElement.getElementsByTagName("password")[0].childNodes[0].data
+    if os.access("ljdump.config", os.F_OK):
+        config = xml.dom.minidom.parse("ljdump.config")
+        server = config.documentElement.getElementsByTagName("server")[0].childNodes[0].data
+        username = config.documentElement.getElementsByTagName("username")[0].childNodes[0].data
+        password = config.documentElement.getElementsByTagName("password")[0].childNodes[0].data
+    else:
+        from getpass import getpass
+        print "ljdump - livejournal archiver"
+        print
+        print "Enter your Livejournal username and password."
+        print
+        server = "http://livejournal.com"
+        username = raw_input("Username: ")
+        password = getpass("Password: ")
+        print
     ljdump(server, username, password)