README

Google makes it pretty easy to migrate from one Google Account to
another with Google Reader, letting you export a OPML file of your
subscriptions from one account and to import it into the next - it
even gets folders right!

Alas, other state is not transferred, so if you used the 'starred'
items as a bookmarking feature over the past few years and have
built up a hundreds of starred items, moving accounts might sound
like an unhappy prospect.

Fortunately, Google Reader has a well-documented API, and changing
the 'starred' status of an item is supported through the API.

To use this script, you need to get a copy of your starred items
as an atom document.  Log into Google Reader with the account you're 
moving away from, and then, in another tab, open this:
http://www.google.com/reader/atom/user/-/state/com.google/starred?n=500

It should give you back an XML document with your first 500 starred
items. Save that file. If you've got more than 500 starred items,
increase the number at the end.  There is a limit on the number of
items you can request in one shot, and while it may be a large
number (10,000) but it's still a limit.  If you hit it, you may
need to look into the continuation token.  It's not clear to me if
the API may return less than the items you request and give you a
continutation token even if you're not at the official limit - I
only had 465 starred items and so it was never an issue.

(If you want to know how many items you've got starred, visit this URL:
http://www.google.com/reader/api/0/stream/items/count?s=user/-/state/com.google/starred

Then, assuming you saved the XML document as 'starred.xml', run
python star-greader-posts.py starred.xml your-new-google-reader-account your-password

If you use two-step authentication, you'll need to get an application
specific password. Go to https://www.google.com/settings/, then "click
Authorizing Applications & Sites", then "Generate new application-specific
password".

It will spin over each entry in the atom document and star the
appropriate entry.

Useful references:

We use the ClientLogin authentication method, and not OAuth - this
looks like it'd be a great reference if we really wanted to use
OAuth
http://asktherelic.com/2010/04/26/accessing-google-reader-via-oauth-and-python/

This script is heavily based on these two posts:
http://stackoverflow.com/questions/52880/google-reader-api-unread-count
and
http://blog.yjl.im/2010/08/using-python-to-get-google-reader.html

In particular, if you run up against some sort of limit in the
number of items you can return in one shot, you'll need to modify
the code to use the Continuation Token to download your starred.xml
file in chunks.

Note that these two scripts only get an authenticated session that
will let you read data, and not write it - you need to get a token
(not a continuation token - there are multiple things called
'token's) in order to star an item.

The API Documentation is very, very helpful
http://blog.martindoms.com/2009/08/15/using-the-google-reader-api-part-1/
http://code.google.com/p/pyrfeed/wiki/GoogleReaderAPI
http://code.google.com/p/google-reader-api/
This page, in particular, finally made it clear where the token
I needed was going to come from.
http://code.google.com/p/google-reader-api/wiki/Authentication

Thanks to Reece Hart for the improvements on order-handling and
a pointer to the app-specific passwords.