def test_update_deletes_datasets(self, mock):
mock.action.package_list.return_value = {'success': True, 'result': [u"tst-a", u"tst-b"]}
datasets = crawler.fetch_dataset_list()
mock.action.package_list.assert_called_once_with()
mock.action.package_list.return_value = {'success': True, 'result': [u"tst-a"]}
datasets = crawler.fetch_dataset_list()
self.assertEquals(1, datasets.count())
def test_fetch_package_search_update(self, mock):
#initial call to fetch dataset list sets up 3 datasets inside the
# datastore
mock.action.package_list.return_value = [
u"tst-deleted", u"tst-not-modified", 'tst-modified'
]
crawler.fetch_dataset_list()
# the second call, we are giving a time delta, this time the registry
# has 3 modified datasets, 1 deleted, 1 modified, 1 new
mock.action.package_search.return_value = {
'count': 3,
'results': [
{'name': 'tst-deleted', 'state': 'deleted'},
{'name': 'tst-modified', 'state': 'active'},
{'name': 'tst-new', 'state': 'active'},
]
}
date = datetime.date(2000, 1, 2)
datasets = set([ds.name for ds in crawler.fetch_dataset_list(date)])
# we want to check that the result returned are only the modified/new
# datasets as these are the ones that are sent to the job queues
# normally a call without a time delta results in all the datasets
# being sent to the job queues
self.assertEquals(set(['tst-modified', 'tst-new']), datasets)
# finally we check over all the datasets to make sure nothing has
# happened to the not modified dataset, and that tst-deleted was
# actually deleted.
all_datasets = set([i.name for i in Dataset.query.all()])
self.assertEquals(
set(['tst-not-modified', 'tst-modified', 'tst-new']),
all_datasets,
)
def test_update_adds_datasets(self, mock):
mock.action.package_list.return_value = [u"tst-a"]
datasets = crawler.fetch_dataset_list()
mock.action.package_list.assert_called_once_with()
mock.action.package_list.return_value = [u"tst-a", u"tst-b"]
datasets = crawler.fetch_dataset_list()
self.assertEquals(2, datasets.count())
def test_update_deletes_datasets(self, iatikit_mock):
data_mock = iatikit_mock.return_value
data_mock.last_updated = datetime.datetime.utcnow()
data_mock.datasets = [
iatikit.Dataset("tst-a.xml"),
iatikit.Dataset("tst-b.xml")
]
datasets = crawler.fetch_dataset_list()
self.assertEquals(2, datasets.count())
data_mock.datasets = [iatikit.Dataset("tst-a.xml")]
datasets = crawler.fetch_dataset_list()
self.assertEquals(1, datasets.count())
def test_update_deletes_datasets(self, mock):
mock.action.package_list.return_value = {
'success': True,
'result': [u"tst-a", u"tst-b"]
}
datasets = crawler.fetch_dataset_list()
mock.action.package_list.assert_called_once_with()
mock.action.package_list.return_value = {
'success': True,
'result': [u"tst-a"]
}
datasets = crawler.fetch_dataset_list()
self.assertEquals(1, len(datasets))
def test_fetch_package_search_update(self, mock):
#initial call to fetch dataset list sets up 3 datasets inside the
# datastore
mock.action.package_list.return_value = {
'success': True,
'result': [u"tst-deleted", u"tst-not-modified", 'tst-modified'],
}
crawler.fetch_dataset_list()
# the second call, we are giving a time delta, this time the registry
# has 3 modified datasets, 1 deleted, 1 modified, 1 new
mock.action.package_search.return_value = {
'success': True,
'result': {
'count':
3,
'results': [
{
'name': 'tst-deleted',
'state': 'deleted'
},
{
'name': 'tst-modified',
'state': 'active'
},
{
'name': 'tst-new',
'state': 'active'
},
]
}
}
date = datetime.date(2000, 1, 2)
datasets = set([ds.name for ds in crawler.fetch_dataset_list(date)])
# we want to check that the result returned are only the modified/new
# datasets as these are the ones that are sent to the job queues
# normally a call without a time delta results in all the datasets
# being sent to the job queues
self.assertEquals(set(['tst-modified', 'tst-new']), datasets)
# finally we check over all the datasets to make sure nothing has
# happened to the not modified dataset, and that tst-deleted was
# actually deleted.
all_datasets = set([i.name for i in Dataset.query.all()])
self.assertEquals(
set(['tst-not-modified', 'tst-modified', 'tst-new']),
all_datasets,
)
def test_fetch_package_search(self, mock):
mock.action.package_search.return_value = {
'success': True,
'result': {
'count':
2,
'results': [
{
'name': 'tst-a',
'state': 'active'
},
{
'name': 'tst-b',
'state': 'active'
},
]
}
}
date = datetime.date(2000, 1, 2)
datasets = crawler.fetch_dataset_list(date)
# check the solr parameters are formed correctly
mock.action.package_search.assert_any_call(
fq='metadata_modified:[2000-01-02T00:00:00Z TO NOW]')
# check that the result of of fetch_dataset_list are only the 2 datasets
self.assertEquals(set(['tst-a', 'tst-b']),
set([ds.name for ds in datasets]))
def test_fetch_package_list(self, mock):
mock.action.package_list.return_value = {
'success': True,
'result': [u"tst-a", u"tst-b"]
}
datasets = crawler.fetch_dataset_list()
mock.action.package_list.assert_called_once_with()
self.assertIn("tst-a", [ds.name for ds in datasets])
self.assertIn("tst-b", [ds.name for ds in datasets])
def test_fetch_package_list(self, iatikit_mock):
data_mock = iatikit_mock.return_value
data_mock.last_updated = datetime.datetime.utcnow()
data_mock.datasets = [
iatikit.Dataset("tst-a.xml"),
iatikit.Dataset("tst-b.xml")
]
datasets = crawler.fetch_dataset_list()
self.assertIn("tst-a", [ds.name for ds in datasets])
self.assertIn("tst-b", [ds.name for ds in datasets])
def test_fetch_package_search(self, mock):
mock.action.package_search.return_value = {
'count': 2,
'results': [
{'name': 'tst-a', 'state': 'active'},
{'name': 'tst-b', 'state': 'active'},
]
}
date = datetime.date(2000, 1, 2)
datasets = crawler.fetch_dataset_list(date)
# check the solr parameters are formed correctly
mock.action.package_search.assert_any_call(
fq='metadata_modified:[2000-01-02T00:00:00Z TO NOW]')
# check that the result of of fetch_dataset_list are only the 2 datasets
self.assertEquals(set(['tst-a', 'tst-b']), set([ds.name for ds in datasets]))
def test_deleted_activities(self, iatikit_mock):
fac.DatasetFactory.create(
name='deleteme',
resources=[
fac.ResourceFactory.create(
url="http://yes",
activities=[
fac.ActivityFactory.build(
iati_identifier="deleted_activity",
title="orig",
)
])
])
self.assertIn("deleteme", [ds.name for ds in Dataset.query.all()])
iatikit_mock.return_value = registry
datasets = crawler.fetch_dataset_list()
self.assertNotIn("deleteme", [ds.name for ds in datasets])
self.assertIn(
"deleted_activity",
[da.iati_identifier for da in DeletedActivity.query.all()])
def test_deleted_activities(self, mock):
fac.DatasetFactory.create(
name='deleteme',
resources=[ fac.ResourceFactory.create(
url=u"http://yes",
activities=[
fac.ActivityFactory.build(
iati_identifier=u"deleted_activity",
title=u"orig",
)
]
)]
)
mock.action.package_list.return_value = [u"tst-a", u"tst-b"]
self.assertIn("deleteme", [ds.name for ds in Dataset.query.all()])
datasets = crawler.fetch_dataset_list()
self.assertNotIn("deleteme", [ds.name for ds in datasets])
self.assertIn(
"deleted_activity",
[da.iati_identifier for da in DeletedActivity.query.all()]
)
def test_deleted_activities(self, mock):
fac.DatasetFactory.create(
name='deleteme',
resources=[
fac.ResourceFactory.create(
url=u"http://yes",
activities=[
fac.ActivityFactory.build(
iati_identifier=u"deleted_activity",
title=u"orig",
)
])
])
mock.action.package_list.return_value = {
'success': True,
'result': [u"tst-a", u"tst-b"]
}
self.assertIn("deleteme", [ds.name for ds in Dataset.query.all()])
datasets = crawler.fetch_dataset_list()
self.assertNotIn("deleteme", [ds.name for ds in datasets])
self.assertIn(
"deleted_activity",
[da.iati_identifier for da in DeletedActivity.query.all()])
def test_update_adds_datasets(self, mock):
mock.return_value = [u"tst-a"]
datasets = crawler.fetch_dataset_list()
mock.return_value = [u"tst-a", u"tst-b"]
datasets = crawler.fetch_dataset_list()
self.assertEquals(2, len(datasets))
def test_fetch_package_list(self, mock):
mock.return_value = [u"tst-a", u"tst-b"]
datasets = crawler.fetch_dataset_list()
self.assertIn("tst-a", [ds.name for ds in datasets])
self.assertIn("tst-b", [ds.name for ds in datasets])
def test_fetch_package_list(self, mock):
mock.action.package_list.return_value = {'success': True, 'result': [u"tst-a", u"tst-b"]}
datasets = crawler.fetch_dataset_list()
mock.action.package_list.assert_called_once_with()
self.assertIn("tst-a", [ds.name for ds in datasets])
self.assertIn("tst-b", [ds.name for ds in datasets])