Transfer dataset from an old CKAN to a single organisation on datahub.io

transfer.py

#!/usr/bin/env python

# mkvirtualenv transfer
# pip install ckanapi

SOURCE = "http://br.ckan.net"
TARGET_ORGANISATION = "open-knowledge-brasil"
SOURCE_TAG = "meta.imported-from-br-ckan-net"

import ckanapi
from ckanapi.errors import NotFound, ValidationError

datahub = ckanapi.RemoteCKAN("http://datahub.io/", apikey="Ross' API KEY")
source =  ckanapi.RemoteCKAN(SOURCE)

# Get the Organisation, we only need the name
org = datahub.action.organization_show(id=TARGET_ORGANISATION)

# Get a list of all the packages on br.ckan.net
pkg_names = source.action.package_list()
print "Found {0} datasets on source".format(len(pkg_names))

for name in pkg_names:
    # Get the dataset off br.ckan.net
    dataset = source.action.package_show(id=name)
    del dataset['id']

    try:
        existing = datahub.action.package_show(id=dataset['name'])
        print "{0} already exists".format(existing['name'])
        continue
    except NotFound:
        # Great, let's carry on
        pass

    # Set the new owning organisation
    dataset['owner_org'] = org['name']

    # Some datasets have no resources, and we don't like that any more ...
    if not 'resources' in dataset:
        dataset['resources'] = []

    # Add a nice tag so we can find them all again
    dataset['tags'].append({'name': SOURCE_TAG })
    try:
        new_dataset = datahub.action.package_create(**dataset)
        print "Created {}".format(new_dataset['name'])
    except ValidationError:
        print "Failed to upload {}".format(dataset['name'])