A command line interface and Python module for accessing the CKAN Action API
Latest commit fc186b3 Sep 27, 2016 @wardi wardi Merge branch 'local-files'



A command line interface and Python module for accessing the CKAN Action API

Build Status tested under Python 2.6, 2.7, 3.2-3.5 and pypy


pip install ckanapi

ckanapi CLI

The ckanapi command line interface lets you access local and remote CKAN instances for bulk operations and simple API actions.


Simple actions with string parameters may be called directly. The response is pretty-printed to STDOUT. e.g.:

$ ckanapi action group_list -r http://demo.ckan.org

Use -r to specify the remote CKAN instance, and -a to provide an API KEY. Remote actions connect as an anonymous user by default.

Local CKAN actions may be run by specifying the config file with -c. If no remote server or config file is specified the CLI will look for a development.ini file in the current directory, much like paster commands.

Local CKAN actions are performed by the site user (default system administrator) when -u is not specified.

To perform local actions with a less privileged user use the -u option with a user name or a name that doesn't exist. This is useful if you don't want things like deleted datasets or private information to be returned.

Action Arguments

Simple action arguments may be passed in KEY=STRING form for string values or in KEY:JSON form for JSON values.

E.g. to view a dataset using a KEY=STRING parameter:

$ ckanapi action package_show id=my-dataset-name
  "name": "my-dataset-name",

E.g. to get the number of datasets for each organization using KEY:JSON parameters:

$ ckanapi action package_search 'facet.field:["organization"]' rows:0
  "facets": {
    "organization": {
      "org1": 42,
      "org2": 21,

Bulk Dumping and Loading Operations

Datasets, groups, organizations, users and related items may be dumped to JSON lines text files and created or updated from JSON lines text files.

Example: dumping datasets from CKAN into a local file with 4 processes:

$ ckanapi dump datasets --all -O datasets.jsonl.gz -z -p 4 -r http://localhost

Example: load datasets from a dataset dump file with 3 processes in parallel:

$ ckanapi load datasets -I datasets.jsonl.gz -z -p 3 -c /etc/ckan/production.ini

These bulk dumping and loading jobs can be run in parallel with multiple worker processes. The jobs in progress, the rate of job completion and any individual errors are shown on STDERR while the jobs run.

Bulk loading jobs may be resumed from the last completed record or split across multiple servers by specifying record start and max values.

Bulk Delete

Datasets, groups, organizations, users and related items may be deleted in bulk with the delete command. This command accepts ids or names on the command line or a number of different formats piped on standard input.

Delete all the datasets (JSON list of "id" or "name" values)

$ ckanapi action package_list -j | ckanapi delete datasets

Selective delete (JSON object with "results" list containing "id" values)

$ ckanapi action package_search q=ponies | ckanapi delete datasets

Processed JSON Lines (JSON objects with "id" or "name" value, one per line)

$ ckanapi dump groups --all > groups.jsonl
$ grep ponies groups.jsonl | ckanapi delete groups

Simple list of "id" or "name" values (one per line)

$ cat users_to_remove.txt
$ ckanapi delete users < users_to_remove.txt

Bulk Dataset and Resource Export - datapackage.json format

Datasets may be exported to a simplified datapackage.json format (which includes the actual resources, where available).

If the resource url is not available, the resource will be included in the datapackage.json file but the actual resource data will not be downloaded.

$ ckanapi dump datasets --all --datapackages=./output_directory/ -r http://sourceckan.example.com

Shell pipelines

Simple shell pipelines are possible with the CLI. E.g. update the title of a dataset with the help of the 'jq' command-line json tool:

$ ckanapi action package_show id=my-dataset \
  | jq '.+{"title":"New title"}' \
  | ckanapi action package_update -i

E.g. Copy all datasets from one CKAN instance to another:

$ ckanapi dump datasets --all -q -r http://sourceckan.example.com \
  | ckanapi load datasets

ckanapi Python Module

The ckanapi Python module may be used from within a CKAN extension or in a Python 2 or Python 3 application separate from CKAN.


Making a request:

from ckanapi import RemoteCKAN
ua = 'ckanapiexample/1.0 (+http://example.com/my/website)'

demo = RemoteCKAN('http://demo.ckan.org', user_agent=ua)
groups = demo.action.group_list(id='data-explorer')
print groups


[u'data-explorer', u'example-group', u'geo-examples', u'skeenawild']

The example above is using an "action shortcut". The .action object detects the method name used ("group_list" above) and converts it to a normal call_action call. This is equivalent code without using an action shortcut:

groups = demo.call_action('group_list', {'id': 'data-explorer'})

All actions in the CKAN Action API and actions added by CKAN plugins are supported by action shortcuts and call_action calls.


  • NotAuthorized - user unauthorized or accessing a deleted item
  • NotFound - name/id not found
  • ValidationError - field errors listed in .error_dict
  • SearchQueryError - error reported from SOLR index
  • SearchError
  • CKANAPIError - incorrect use of ckanapi or unable to parse response
  • ServerIncompatibleError - the remote API is not a CKAN API

When using an action shortcut or the call_action method failures are raised as exceptions just like when calling get_action from a CKAN plugin:

from ckanapi import RemoteCKAN, NotAuthorized
ua = 'ckanapiexample/1.0 (+http://example.com/my/website)'

demo = RemoteCKAN('http://demo.ckan.org', apikey='phony-key', user_agent=ua)
    pkg = demo.action.package_create(name='my-dataset', title='not going to work')
except NotAuthorized:
    print 'denied'

When it is possible to import ckan all the ckanapi exception classes are replaced with the CKAN exceptions with the same names.

File uploads

File uploads for CKAN 2.2+ are supported by passing file-like objects to action shortcut methods:

from ckanapi import RemoteCKAN
ua = 'ckanapiexample/1.0 (+http://example.com/my/website)'

mysite = RemoteCKAN('http://myckan.example.com', apikey='real-key', user_agent=ua)
    url='dummy-value',  # ignored but required by CKAN<=2.5.x
    upload=open('/path/to/file/to/upload.csv', 'rb'))

When using call_action you must pass file objects separately:

    {'package_id': 'my-dataset-with-files'},
    files={'upload': open('/path/to/file/to/upload.csv', 'rb')})

Session Control

As of ckanapi 4.0 RemoteCKAN will keep your HTTP connection open using a requests session.

For long-running scripts make sure to close your connections by using RemoteCKAN as a context manager:

from ckanapi import RemoteCKAN
ua = 'ckanapiexample/1.0 (+http://example.com/my/website)'

with RemoteCKAN('http://demo.ckan.org', user_agent=ua) as demo:
    groups = demo.action.group_list(id='data-explorer')
print groups

Or by explicitly calling RemoteCKAN.close().


A similar class is provided for accessing local CKAN instances from a plugin in the same way as remote CKAN instances. Unlike CKAN's get_action LocalCKAN prevents data from one action call leaking into the next which can cause issues that are very hard do debug.

This class defaults to using the site user with full access.

from ckanapi import LocalCKAN, ValidationError

registry = LocalCKAN()
    registry.action.package_create(name='my-dataset', title='this will work fine')
except ValidationError:
    print 'unless my-dataset already exists'

For extra caution pass a blank username to LocalCKAN and only actions allowed by anonymous users will be permitted.

from ckanapi import LocalCKAN

anon = LocalCKAN(username='')
print anon.action.status_show()


A class is provided for making action requests to a webtest.TestApp instance for use in CKAN tests:

from ckanapi import TestAppCKAN
from webtest import TestApp

test_app = TestApp(...)
demo = TestAppCKAN(test_app, apikey='my-test-key')
groups = demo.action.group_list(id='data-explorer')


🇨🇦 Government of Canada / Gouvernement du Canada

The project files are covered under Crown Copyright, Government of Canada and is distributed under the MIT license. Please see COPYING / COPYING.fr for full details.