Dump the workflow states

Thankfully, I have lots of backups. Even better, I have a QA server, which has a pretty recent snapshot from Production.  If I can see what states the items had before I messed them up, then I could reset the ones that are wrong.  (If it's too many to do manually, I can always hack a python script, right?)

I tried several different approaches.  This one actually worked:

def dump_catalog_by_state():
    #Import stuff and set up environment.
    from Products.CMFCore.utils import getToolByName
    from AccessControl.SecurityManagement import newSecurityManager
    from Products.CMFCore.tests.base.security import OmnipotentUser
    admin = OmnipotentUser()
    newSecurityManager(None, admin)
    OUTPUT_PATH = '/var/tmp/dump/'
    #Now do it.
    print "Getting list of plone sites...",
    #getPloneSites is our own custom Script python.
    #You could give it a list of site objects, I suppose.
    sites = app.getPloneSites() 
    print "Found %s site(s)." % len(sites)
    for site in sites:
        catalog = getToolByName(site, 'portal_catalog')
        states = ['open', 'pending', 'private', 'restricted', ]
        for state in states:
            print "Querying '%s' catalog for items in the '%s' state" % (site.id, state),
            items = catalog({'review_state': state})
            print "Found %s item(s)." % len(items)
            urls = [item.getPath() + '\n' for item in items]
            #Dump to file.
            filename = "%s%s-%s" % (OUTPUT_PATH, site.id, state)
            print "Dumping to %s" % filename,
            dump = file(filename, "w")
            dump.writelines(urls)
            dump.close()
            print "Dumped."

I copied-and-pasted this function into the debug shell (bin/instance debug) and ran it.  Out came a lot of text files!  This is exactly the information I needed.