Merge branch 'harvest-object-create'

ckanext/harvest/logic/action/create.py

@@ -1,18 +1,27 @@
 import logging
 
+import ckan
 from ckan import logic
 
 from ckan.logic import NotFound, check_access
 from ckanext.harvest.logic import HarvestJobExists
 
 from ckanext.harvest.plugin import DATASET_TYPE_NAME
-from ckanext.harvest.model import (HarvestSource, HarvestJob)
-from ckanext.harvest.logic.dictization import harvest_job_dictize
-from ckanext.harvest.logic.schema import harvest_source_show_package_schema
+from ckanext.harvest.model import (HarvestSource, HarvestJob, HarvestObject,
+    HarvestObjectExtra)
+from ckanext.harvest.logic.dictization import (harvest_job_dictize,
+    harvest_object_dictize)
+from ckanext.harvest.logic.schema import (harvest_source_show_package_schema,
+    harvest_object_create_schema)
 from ckanext.harvest.logic.action.get import harvest_source_list,harvest_job_list
 
 log = logging.getLogger(__name__)
 
+_validate = ckan.lib.navl.dictization_functions.validate
+
+class InactiveSource(Exception):
+    pass
+
 def harvest_source_create(context,data_dict):
     '''
     Creates a new harvest source
@@ -136,3 +145,32 @@ def _check_for_existing_jobs(context, source_id):
     exist = len(exist_new + exist_running) > 0
 
     return exist
+
+def harvest_object_create(context, data_dict):
+    """ Create a new harvest object
+
+    :type guid: string (optional)
+    :type content: string (optional)
+    :type job_id: string 
+    :type source_id: string (optional)
+    :type package_id: string (optional)
+    :type extras: dict (optional)
+    """
+    check_access('harvest_object_create', context, data_dict)
+    data, errors = _validate(data_dict, harvest_object_create_schema(), context)
+
+    if errors:
+        raise logic.ValidationError(errors)
+
+    obj = HarvestObject(
+        guid=data.get('guid'),
+        content=data.get('content'),
+        job=data['job_id'],
+        harvest_source_id=data.get('source_id'),
+        package_id=data.get('package_id'),
+        extras=[ HarvestObjectExtra(key=k, value=v) 
+            for k, v in data.get('extras', {}).items() ]
+    )
+
+    obj.save()
+    return harvest_object_dictize(obj, context)

ckanext/harvest/logic/auth/create.py

@@ -52,3 +52,13 @@ def harvest_job_create_all(context, data_dict):
         return {'success': False, 'msg': pt._('Only sysadmins can create harvest jobs for all sources')}
     else:
         return {'success': True}
+
+def harvest_object_create(context, data_dict):
+    """
+        Auth check for creating a harvest object
+
+        only the sysadmins can create harvest objects
+    """
+    # sysadmins can run all actions if we've got to this point we're not a sysadmin
+    return {'success': False, 'msg': pt._('Only the sysadmins can create harvest objects')}
+

ckanext/harvest/logic/schema.py

@@ -20,6 +20,9 @@
                                               harvest_source_frequency_exists,
                                               dataset_type_exists,
                                               harvest_source_convert_from_config,
+                                              harvest_source_id_exists,
+                                              harvest_job_exists,
+                                              harvest_object_extras_validator,
                                               )
 
 def harvest_source_schema():
@@ -74,3 +77,16 @@ def harvest_source_show_package_schema():
     })
 
     return schema
+
+def harvest_object_create_schema():
+    schema = {
+        'guid': [ignore_missing, unicode],
+        'content': [ignore_missing, unicode],
+        'state': [ignore_missing, unicode],
+        'job_id': [harvest_job_exists],
+        'source_id': [ignore_missing, harvest_source_id_exists],
+        'package_id': [ignore_missing, package_id_exists],
+        'extras': [ignore_missing, harvest_object_extras_validator],
+    }
+    return schema
+

ckanext/harvest/logic/validators.py

@@ -7,7 +7,7 @@
 from ckan.plugins import PluginImplementations
 
 from ckanext.harvest.plugin import DATASET_TYPE_NAME
-from ckanext.harvest.model import HarvestSource, UPDATE_FREQUENCIES
+from ckanext.harvest.model import HarvestSource, UPDATE_FREQUENCIES, HarvestJob
 from ckanext.harvest.interfaces import IHarvester
 
 from ckan.lib.navl.validators import keep_extras
@@ -22,6 +22,14 @@ def harvest_source_id_exists(value, context):
         raise Invalid('Harvest Source with id %r does not exist.' % str(value))
     return value
 
+def harvest_job_exists(value, context):
+    """Check if a harvest job exists and returns the model if it does"""
+    result = HarvestJob.get(value, None)
+
+    if not result:
+        raise Invalid('Harvest Job with id %r does not exist.' % str(value))
+    return result
+
 def _normalize_url(url):
     o = urlparse.urlparse(url)
 
@@ -196,3 +204,11 @@ def dataset_type_exists(value):
     if value != DATASET_TYPE_NAME:
         value = DATASET_TYPE_NAME
     return value
+
+def harvest_object_extras_validator(value, context):
+    if not isinstance(value, dict):
+        raise Invalid('extras must be a dict')
+    for v in value.values():
+        if not isinstance(v, basestring):
+            raise Invalid('extras must be a dict of strings')
+    return value

ckanext/harvest/tests/factories.py

@@ -0,0 +1,14 @@
+import factory
+from ckanext.harvest.model import HarvestSource, HarvestJob
+
+class HarvestSourceFactory(factory.Factory):
+    FACTORY_FOR = HarvestSource
+
+    url = "http://harvest.test.com"
+    type = "test-harvest-source"
+
+class HarvestJobFactory(factory.Factory):
+    FACTORY_FOR = HarvestJob
+
+    status = "New"
+    source = factory.SubFactory(HarvestSourceFactory)

ckanext/harvest/tests/test_action.py

@@ -3,12 +3,16 @@
 import ckan
 import paste
 import pylons.test
+import factories
+import unittest
 
 from ckan import tests
 from ckan import plugins as p
+from ckan.plugins import toolkit
 from ckanext.harvest.interfaces import IHarvester
 import ckanext.harvest.model as harvest_model
 
+
 class MockHarvesterForActionTests(p.SingletonPlugin):
     p.implements(IHarvester)
     def info(self):
@@ -203,3 +207,58 @@ def test_update(self):
         assert source.url == source_dict['url']
         assert source.type == source_dict['source_type']
 
+class TestHarvestObject(unittest.TestCase):
+    @classmethod
+    def setup_class(cls):
+        harvest_model.setup()
+
+    @classmethod
+    def teardown_class(cls):
+        ckan.model.repo.rebuild_db()
+
+    def test_create(self):
+        job = factories.HarvestJobFactory()
+        job.save()
+
+        context = {
+            'model' : ckan.model,
+            'session': ckan.model.Session,
+            'ignore_auth': True,
+        }
+        data_dict = {
+            'guid' : 'guid',
+            'content' : 'content',
+            'job_id' : job.id,
+            'extras' : { 'a key' : 'a value' },
+        }
+        harvest_object = toolkit.get_action('harvest_object_create')(
+            context, data_dict)
+
+        # fetch the object from database to check it was created
+        created_object = harvest_model.HarvestObject.get(harvest_object['id'])
+        assert created_object.guid == harvest_object['guid'] == data_dict['guid']
+
+    def test_create_bad_parameters(self):
+        source_a = factories.HarvestSourceFactory()
+        source_a.save()
+        job = factories.HarvestJobFactory()
+        job.save()
+
+        context = {
+            'model' : ckan.model,
+            'session': ckan.model.Session,
+            'ignore_auth': True,
+        }
+        data_dict = {
+            'job_id' : job.id,
+            'source_id' : source_a.id,
+            'extras' : 1
+        }
+        harvest_object_create = toolkit.get_action('harvest_object_create')
+        self.assertRaises(ckan.logic.ValidationError, harvest_object_create,
+            context, data_dict)
+
+        data_dict['extras'] = {'test': 1 }
+
+        self.assertRaises(ckan.logic.ValidationError, harvest_object_create,
+            context, data_dict)

pip-requirements.txt

@@ -1 +1,2 @@
 pika==0.9.8
+factory-boy>=2