[#4878] Allow custom CKAN callback URL for the DataPusher

The DataPusher pings back CKAN when performing or finishing
a job (calling the datapusher_hook action), and it does so
via an HTTP request to the host defined in ckan.site_url.

There are a number of scenarios where this does not work, eg:

* CKAN and DataPusher sitting behind a Firewall that doesn't
  allow external requests
* Standard Docker compose setup for development where the
  ckan.site_url is http://localhost:5000 or similar

This change adds a new config option that allows to define an
alternative internal URL that DataPusher can reach.

ckanext/datapusher/logic/action.py

@@ -62,7 +62,14 @@ def datapusher_submit(context, data_dict):
     datapusher_url = config.get('ckan.datapusher.url')
 
     site_url = h.url_for('/', qualified=True)
-    callback_url = h.url_for('/api/3/action/datapusher_hook', qualified=True)
+
+    callback_url_base = config.get('ckan.datapusher.callback_url_base')
+    if callback_url_base:
+        callback_url = urlparse.urljoin(
+            callback_url_base.rstrip('/'), '/api/3/action/datapusher_hook')
+    else:
+        callback_url = h.url_for(
+            '/api/3/action/datapusher_hook', qualified=True)
 
     user = p.toolkit.get_action('user_show')(context, {'id': context['user']})
 

ckanext/datapusher/tests/test.py

@@ -11,6 +11,7 @@
 import ckanext.datastore.backend.postgres as db
 import responses
 import nose
+from nose.tools import assert_equals
 import sqlalchemy.orm as orm
 from ckan.common import config
 from ckanext.datastore.tests.helpers import rebuild_all_dbs, set_url_type
@@ -232,3 +233,33 @@ def test_datapusher_hook_no_resource_id_in_metadata(self):
 
         self.app.post('/api/action/datapusher_hook', params=postparams,
                       status=409)
+
+    @responses.activate
+    @helpers.change_config(
+        'ckan.datapusher.callback_url_base', 'https://ckan.example.com')
+    @helpers.change_config(
+        'ckan.datapusher.url', 'http://datapusher.ckan.org')
+    def test_custom_callback_url_base(self):
+
+        package = model.Package.get('annakarenina')
+        resource = package.resources[0]
+
+        responses.add(
+            responses.POST,
+            'http://datapusher.ckan.org/job',
+            content_type='application/json',
+            body=json.dumps({'job_id': 'foo', 'job_key': 'barloco'})
+        )
+        responses.add_passthru(config['solr_url'])
+
+        tests.call_action_api(
+            self.app, 'datapusher_submit', apikey=self.sysadmin_user.apikey,
+            resource_id=resource.id,
+            ignore_hash=True
+        )
+
+        data = json.loads(responses.calls[-1].request.body)
+        assert_equals(
+            data['result_url'],
+            'https://ckan.example.com/api/3/action/datapusher_hook'
+        )

doc/maintaining/configuration.rst

@@ -1524,6 +1524,22 @@ running on port 8800. If you want to manually install the DataPusher, follow
 the installation `instructions <http://docs.ckan.org/projects/datapusher>`_.
 
 
+.. _ckan.datapusher.callback_url_base:
+
+ckan.datapusher.url
+^^^^^^^^^^^^^^^^^^^
+
+Example::
+
+  ckan.datapusher.callback_url_base = http://ckan:5000/
+
+Default value: Value of ``ckan.site_url``
+
+Alternative callback URL for DataPusher when performing a request to CKAN. This is
+useful on scenarios where the host where DataPusher is running can not access the
+public CKAN site URL.
+
+
 .. _ckan.datapusher.assume_task_stale_after:
 
 ckan.datapusher.assume_task_stale_after