About

CKAN extension implementing ODN pipeline association to dataset functionality

Until now added features:

• Adds 'pipelines' table to DB, columns: package_id, pipeline_id, name
• Adds tab 'Pipelines' to dataset management
• Associates existing pipeline with CKAN dataset
• Associates manually created pipeline with CKAN dataset
• Associates modified copy of existing pipeline with CKAN dataset
• Shows information about last and next execution
• Links to UV functionality
• Uses ODN/UV rest API
• Adds API calls for creating / updating resources from UnifiedViews (L-Catalog DPU)
• Adds API call internal_api for proxy-ing API calls

Installation

(Optional): activate ckan virtualenv . /usr/lib/ckan/default/bin/activate

From the extension folder start the installation: python setup.py install

Add extension to ckan config: /etc/ckan/default/production.ini

ckan.plugins = odn_pipeline internal_api

• odn_pipeline - plugin for association pipeline to dataset from CKAN GUI
• internal_api - plugin to proxy API calls, needed by DPUs: L-FilesToCkan, L-RdfToCkan, L-RelationalToCkan, L-RelationalDiffToCkan

to section [app:main] add:

# pipeline
odn.uv.url = http://HOST/unifiedviews
odn.uv.api.url = http://127.0.0.1:8080/master/api/1
odn.uv.timeout = 5
# uv rest api authorization
odn.uv.api.auth.username = username
odn.uv.api.auth.password = password

# allow create pipelines from CKAN gui, default True (optional)
odn.uv.pipeline.allow.create = False

# resource update api, the URL are quoted in code
odn.storage.rdf.uri.template = http://host/sparql?query=select {?s ?p ?o} from {storage_id}

# internal_api
ckan.auth.internal_api.token = my secret token

DB init

After installing plugin and restarting apache server start db initialization:

paster --plugin=ckanext-odn-pipeline pipeline-cmd initdb --config=/etc/ckan/default/production.ini

There should be output like this:

Starting db initialization
creating pipelines table				/ Or if it was already intialized:
pipelines table created successfully	/ pipelines table already exists
End of db initialization

Uninstall

Before removing extension start:

paster --plugin=ckanext-odn-pipeline pipeline-cmd uninstall --config=/etc/ckan/default/production.ini

This will drop tables created in DB init script.

Now you can remove plugin string from: /etc/ckan/default/production.ini

Restart apache server: sudo service apache2 restart

And remove from python installed extension egg.

Resource update API for L-Catalog

This describes communication between UnifiedViews L-Catalog DPU and CKAN. This request is sent by L-Catalog to create / update resources according the pipeline id. More than one resource can be send.

POST <host>/api/3/action/resources
Accept: application/json

{
	"pipelineId": 8,
	"resources": [
		{
			"storageId": {
				"type": "RDF",
				"value": "storageID"
			},
			"resource": {
				"name": "resource name",
				"description": "description of resource",
				"last_modified" : "2014-12-05 08:38:44.019000",
			}
		}
	]
}

All keys in this request are mandatory. The only exception to this is the resource part, where only 'name' key is mandatory. The name key is used as resource identification. It is used to determine if the resource should be created or updated. The resource value corresponds to standard CKAN resource_update / resource_create api call.

storageId type: RDF / FILE

Response:

HTTP 200 OK
Content-Type: application/json

{
	"help": "...",
	"result": [
		{
			"message": null,
            "name": "resource name",
			"success": true,
			"type": "UPDATE"
		},
		{
			"message": "Error message explaining failure just for THIS resource.",
            "name": "resource name 2",
			"success": false,
			"type": "CREATE"
		}
	],
	"success": true
}

Error response:

HTTP 404 Not found
Content-Type: application/json

{
	"help": "...",
	"success": false,
	"error": {
		"message": "Not found: No dataset found with pipeline 15 assigned to it.",
		"__type": "Not Found Error"
	}
}

internal_api

Functions as proxy to other CKAN API calls. Requires following properties set up in CKAN configuration file:

• ckan.auth.internal_api.token
• odn.storage.rdf.uri.template

Request:

• POST /api/3/action/internal_api
• multipart/form-data
• parameters:
  • action - name of API call, e.g. 'resource_create'
  • pipeline_id - pipeline id used to identify the dataset the change should be applied to (optional)
  • user_id - user used for authentication (CKAN user id)
  • token - authentication token set up in CKAN conf file (ckan.auth.internal_api.token)
  • data - actual JSON data of the proxied API call (optional)
  • type - 'RDF' otherwise optional
  • value - storage id if type == 'RDF', otherwise optional

For 'package_update', 'package_show', 'resource_create' actions the pipeline_id is converted to appropriate package id. So for these actions its not necessary to add the ids to the data parameter and will be overwritten if both are given.

Added special action:

• resource_download - enabling to download file through CKAN API
  • WARNING: this change requires change in ApiController (in CKAN core) to function
  • required data parameters:
    • package_id - dataset id, that the resource belongs to
    • id - resource id
  • response if success: file, header parameter content-type indicating mimetype
  • response if failed: standard CKAN API response

Internationalization (i18n)

CKAN supports internationalization through babel (pip install babel). This tool extracts the messages from source code and html files and creates .pot file. Next using commands (step 2 or 3) it creates or updates .po files. The actual translation are in these .po files.

1. To extract new .pot file from sources

   python setup.py extract_messages
   

   This need to be done if there is no .pot file or there were some changes to messages in source code files or html files.

2. To generate .po for new localization (this example uses 'sk' localization)

   python setup.py init_catalog --locale sk
   

3. If only updating existing .po file (e.g. new messages were extracted through step 1)

   python setup.py update_catalog --locale sk
   

Licenses

GNU Affero General Public License, Version 3.0 is used for licensing of the code (see LICENSE)