def test_timeseries_resource_url_non_empty(self):
"""Asserts that timeseries resources must have url defined"""
data = {'timeseries_resources__0__description': 'description with no url',
'timeseries_resources__0__date': 'date with no url',
'timeseries_resources__0__format': 'format with no url'}
response = self._form_client.post_form(data)
assert_in('URL:</b> Missing value', response.body, response.body)
def test_cannot_create_a_non_ogl_dataset_without_specifying_license_name(self):
"""
Asserts that the user specify a license name if the license is non-OGL
"""
data = {'license_id': ""}
response = self._form_client.post_form(data)
assert_in('Licence:</b> Please enter the access constraints.', response.body)
def test_cannot_create_a_non_ogl_dataset_without_specifying_license_name(self):
"""
Asserts that the user specify a license name if the license is non-OGL
"""
data = {"license_id": ""}
response = self._form_client.post_form(data)
assert_in("Licence:</b> Please enter the access constraints.", response.body)
def test_groups_allowed_to_be_its_parent(self):
groups = model.Group.by_name(u'national-health-service').\
groups_allowed_to_be_its_parent(type=group_type)
names = names_from_groups(groups)
assert_in('department-of-health', names)
assert_in('cabinet-office', names)
assert_not_in('natonal-health-service', names)
assert_not_in('nhs-wirral-ccg', names)
def test_get_children_group_hierarchy__from_top_2(self):
groups = model.Group.by_name(u'department-of-health').\
get_children_group_hierarchy(type=group_type)
# the first group must be NHS or Food Standards Agency - i.e. on the
# first level down
nhs = groups[0]
assert_in(nhs[1], ('national-health-service', 'food-standards-agency'))
assert_equal(model.Group.get(nhs[3]).name, 'department-of-health')
def test_groups_allowed_to_be_its_parent(self):
groups = model.Group.by_name(u'national-health-service').\
groups_allowed_to_be_its_parent(type=group_type)
names = names_from_groups(groups)
assert_in('department-of-health', names)
assert_in('cabinet-office', names)
assert_not_in('natonal-health-service', names)
assert_not_in('nhs-wirral-ccg', names)
def test_get_children_group_hierarchy__from_top_2(self):
groups = model.Group.by_name(u'department-of-health').\
get_children_group_hierarchy(type=group_type)
# the first group must be NHS or Food Standards Agency - i.e. on the
# first level down
nhs = groups[0]
assert_in(nhs[1], ('national-health-service', 'food-standards-agency'))
assert_equal(model.Group.get(nhs[3]).name, 'department-of-health')
def test_publisher_cannot_edit_a_UKLP_dataset(self):
"""Test that a publisher cannot access the edit page of a UKLP dataset"""
offset = url_for(controller="package", action="edit", id=self.uklp_dataset["id"])
response = None
try:
response = self.app.get(offset, extra_environ={"REMOTE_USER": self.tester})
except paste.fixture.AppError, e:
assert_in("401 Unauthorized", str(e))
def test_publisher_cannot_edit_a_UKLP_dataset(self):
"""Test that a publisher cannot access the edit page of a UKLP dataset"""
offset = url_for(controller='package', action='edit', id=self.uklp_dataset['id'])
response = None
try:
response = self.app.get(offset, extra_environ={'REMOTE_USER': self.tester})
except paste.fixture.AppError, e:
assert_in('401 Unauthorized', str(e))
def test_timeseries_resource_url_non_empty(self):
"""Asserts that timeseries resources must have url defined"""
data = {
"timeseries_resources__0__description": "description with no url",
"timeseries_resources__0__date": "date with no url",
"timeseries_resources__0__format": "format with no url",
}
response = self._form_client.post_form(data)
assert_in("URL:</b> Missing value", response.body, response.body)
def test_get_children_groups(self):
res = model.Group.by_name(u'department-of-health').\
get_children_groups(type=group_type)
# check groups
assert_equal(name_set_from_groups(res),
set(('national-health-service',
'food-standards-agency')))
# check each group is a Group
assert isinstance(res[0], model.Group)
assert_in(res[0].name, ('national-health-service', 'food-standards-agency'))
assert_in(res[0].title, ('National Health Service', 'Food Standards Agency'))
def test_additional_resources_disallows_non_documentation_resource_types(self):
"""
Asserts that non-documentation resource types are not allowed for additional resources.
"""
for resource_type in ["bad", "api", "file"]:
data = {
"additional_resources__0__description": "Additional resource 1",
"additional_resources__0__url": "http://example.com/doc1",
"additional_resources__0__resource_type": resource_type,
}
response = self._form_client.post_form(data)
assert_in("Invalid resource type: %s" % resource_type, response.body)
def test_get_children_groups(self):
res = model.Group.by_name(u'department-of-health').\
get_children_groups(type=group_type)
# check groups
assert_equal(name_set_from_groups(res),
set(('national-health-service', 'food-standards-agency')))
# check each group is a Group
assert isinstance(res[0], model.Group)
assert_in(res[0].name,
('national-health-service', 'food-standards-agency'))
assert_in(res[0].title,
('National Health Service', 'Food Standards Agency'))
def test_additional_resources_disallows_non_documentation_resource_types(self):
"""
Asserts that non-documentation resource types are not allowed for additional resources.
"""
for resource_type in ["bad", "api", "file"]:
data = {
'additional_resources__0__description': 'Additional resource 1',
'additional_resources__0__url': 'http://example.com/doc1',
'additional_resources__0__resource_type': resource_type,
}
response = self._form_client.post_form(data)
assert_in("Invalid resource type: %s" % resource_type,
response.body)
def test_21_package_dictization_with_deleted_group(self):
"""
Ensure that the dictization does not return groups that the dataset has
been removed from.
"""
# Create a new dataset and 2 new groups
model.repo.new_revision()
pkg = model.Package(name='testing-deleted-groups')
group_1 = model.Group(name='test-group-1')
group_2 = model.Group(name='test-group-2')
model.Session.add(pkg)
model.Session.add(group_1)
model.Session.add(group_2)
model.Session.flush()
# Add the dataset to group_1, and signal that the dataset used
# to be a member of group_2 by setting its membership state to 'deleted'
membership_1 = model.Member(table_id = pkg.id,
table_name = 'package',
group = group_1,
group_id = group_1.id,
state = 'active')
membership_2 = model.Member(table_id = pkg.id,
table_name = 'package',
group = group_2,
group_id = group_2.id,
state = 'deleted')
model.Session.add(membership_1)
model.Session.add(membership_2)
model.repo.commit()
# Dictize the dataset
context = {"model": model,
"session": model.Session}
result = package_dictize(pkg, context)
self.remove_changable_columns(result)
assert_not_in('test-group-2', [ g['name'] for g in result['groups'] ])
assert_in('test-group-1', [ g['name'] for g in result['groups'] ])
def test_21_package_dictization_with_deleted_group(self):
"""
Ensure that the dictization does not return groups that the dataset has
been removed from.
"""
# Create a new dataset and 2 new groups
model.repo.new_revision()
pkg = model.Package(name='testing-deleted-groups')
group_1 = model.Group(name='test-group-1')
group_2 = model.Group(name='test-group-2')
model.Session.add(pkg)
model.Session.add(group_1)
model.Session.add(group_2)
model.Session.flush()
# Add the dataset to group_1, and signal that the dataset used
# to be a member of group_2 by setting its membership state to 'deleted'
membership_1 = model.Member(table_id = pkg.id,
table_name = 'package',
group = group_1,
group_id = group_1.id,
state = 'active')
membership_2 = model.Member(table_id = pkg.id,
table_name = 'package',
group = group_2,
group_id = group_2.id,
state = 'deleted')
model.Session.add(membership_1)
model.Session.add(membership_2)
model.repo.commit()
# Dictize the dataset
context = {"model": model,
"session": model.Session}
result = package_dictize(pkg, context)
self.remove_changable_columns(result)
assert_not_in('test-group-2', [ g['name'] for g in result['groups'] ])
assert_in('test-group-1', [ g['name'] for g in result['groups'] ])
def test_data_resources_disallows_non_file_or_api_resource_types(self):
"""
Asserts that non-{file,api} resource types are not allowed for data resources.
"""
for resource_type in ["bad", "documentation"]:
data = {
"timeseries_resources__0__description": "Timeseries resource 1",
"timeseries_resources__0__url": "http://example.com/data1",
"timeseries_resources__0__date": "Summer",
"timeseries_resources__0__resource_type": resource_type,
}
response = self._form_client.post_form(data)
assert_in("Invalid resource type: %s" % resource_type, response.body)
data = {
"individual_resources__0__description": "Individual resource 1",
"individual_resources__0__url": "http://example.com/data1",
"individual_resources__0__resource_type": resource_type,
}
response = self._form_client.post_form(data)
assert_in("Invalid resource type: %s" % resource_type, response.body)
def test_data_resources_disallows_non_file_or_api_resource_types(self):
"""
Asserts that non-{file,api} resource types are not allowed for data resources.
"""
for resource_type in ["bad", "documentation"]:
data = {
'timeseries_resources__0__description': 'Timeseries resource 1',
'timeseries_resources__0__url': 'http://example.com/data1',
'timeseries_resources__0__date': 'Summer',
'timeseries_resources__0__resource_type': resource_type,
}
response = self._form_client.post_form(data)
assert_in("Invalid resource type: %s" % resource_type,
response.body)
data = {
'individual_resources__0__description': 'Individual resource 1',
'individual_resources__0__url': 'http://example.com/data1',
'individual_resources__0__resource_type': resource_type,
}
response = self._form_client.post_form(data)
assert_in("Invalid resource type: %s" % resource_type,
response.body)
def test_url_with_404(self, url):
result = self.check_link(url)
assert_in('Server returned HTTP error status: 404 Not Found', result['url_errors'])
def test_file_url(self):
url = u'file:///home/root/test.txt'
result = self.check_link(url)
assert_in(u'Invalid url scheme. Please use one of: ftp http https',
result['url_errors'])
def test_url_with_404(self, url):
result = self.check_link(url)
assert_in('Server returned HTTP error status: 404 Not Found', result['url_errors'])
def test_url_with_503(self, url):
result = self.check_link(url)
assert_in('Server returned HTTP error status: 503 Service Unavailable', result['url_errors'])
def test_empty_url(self):
url = u''
result = self.check_link(url)
assert_in("URL parsing failure - did not find a host name", result['url_errors'])
def test_a_full_timeseries_dataset(self):
"""
Tests the submission of a fully-completed timeseries dataset.
"""
package_data = _EXAMPLE_TIMESERIES_DATA
package_name = package_data['name']
CreateTestData.flag_for_deletion(package_name)
assert not self.get_package_by_name(package_name),\
'Package "%s" already exists' % package_name
# create package via form
response = self._form_client.post_form(package_data)
# ensure it's correct
pkg = self.get_package_by_name(package_name)
assert pkg, response.body
assert_equal(package_data['title'], pkg.title)
assert_equal(package_data['notes'], pkg.notes)
publisher = pkg.as_dict()['groups'][0]
assert_equal(package_data['groups__0__name'], publisher)
# Extra data
# Timeseries data
expected_timeseries_keys = filter(lambda k: k.startswith('timeseries_resources'),
package_data.keys())
timeseries_resources = ckanext.dgu.lib.helpers.timeseries_resources(pkg.as_dict())
assert_equal(len(timeseries_resources), 4)
for key in expected_timeseries_keys:
index, field = key.split('__')[1:]
index = int(index)
assert_equal(package_data[key],
timeseries_resources[index][field])
# Publisher / contact details
# The contact-email should not be an extra-field on the dataset as it's the
# same as the publisher group's contact-email. ie - it hasn't been overridden.
# The resof the information should be in the extras fields
assert_not_in('contact-email', pkg.extras)
assert_equal(package_data['contact-name'], pkg.extras['contact-name'])
assert_equal(package_data['contact-phone'], pkg.extras['contact-phone'])
assert_equal(package_data['foi-name'], pkg.extras['foi-name'])
assert_equal(package_data['foi-email'], pkg.extras['foi-email'])
assert_equal(package_data['foi-phone'], pkg.extras['foi-phone'])
# Themes and tags
assert_equal(package_data['theme-primary'], pkg.extras['theme-primary'])
assert_equal(set(package_data['theme-secondary']),
set(pkg.extras['theme-secondary']))
# Health and Education are from the primary and secondary themes, which
# end up in the tags
assert_equal(set(['tag1', 'tag2', 'a multi word tag', 'Health', 'Education']),
set(tag.name for tag in pkg.get_tags()))
# Additional resources
expected_additional_keys = filter(lambda k: k.startswith('additional_resources'),
package_data.keys())
additional_resources = ckanext.dgu.lib.helpers.additional_resources(pkg.as_dict())
assert_equal(len(additional_resources), 2)
for key in expected_additional_keys:
index, field = key.split('__')[1:]
index = int(index)
assert_equal(package_data[key],
additional_resources[index][field])
assert_equal(package_data['mandate'], pkg.extras['mandate'])
assert_equal(package_data['access_constraints'], pkg.license_id)
assert_equal(package_data['temporal_coverage-from'], DateType.db_to_form(pkg.extras['temporal_coverage-from']))
assert_equal(package_data['temporal_coverage-to'], DateType.db_to_form(pkg.extras['temporal_coverage-to']))
assert_in('England', pkg.extras['geographic_coverage'])
def test_a_full_timeseries_dataset(self):
"""
Tests the submission of a fully-completed timeseries dataset.
"""
package_data = _EXAMPLE_TIMESERIES_DATA
package_name = package_data["name"]
CreateTestData.flag_for_deletion(package_name)
assert not self.get_package_by_name(package_name), 'Package "%s" already exists' % package_name
# create package via form
response = self._form_client.post_form(package_data)
# ensure it's correct
pkg = self.get_package_by_name(package_name)
assert pkg, response.body
assert_equal(package_data["title"], pkg.title)
assert_equal(package_data["notes"], pkg.notes)
publisher = pkg.as_dict()["groups"][0]
assert_equal(package_data["groups__0__name"], publisher)
# Extra data
# Timeseries data
expected_timeseries_keys = filter(lambda k: k.startswith("timeseries_resources"), package_data.keys())
timeseries_resources = ckanext.dgu.lib.helpers.timeseries_resources(pkg.as_dict())
assert_equal(len(timeseries_resources), 4)
for key in expected_timeseries_keys:
index, field = key.split("__")[1:]
index = int(index)
assert_equal(package_data[key], timeseries_resources[index][field])
# Publisher / contact details
# The contact-email should not be an extra-field on the dataset as it's the
# same as the publisher group's contact-email. ie - it hasn't been overridden.
# The resof the information should be in the extras fields
assert_not_in("contact-email", pkg.extras)
assert_equal(package_data["contact-name"], pkg.extras["contact-name"])
assert_equal(package_data["contact-phone"], pkg.extras["contact-phone"])
assert_equal(package_data["foi-name"], pkg.extras["foi-name"])
assert_equal(package_data["foi-email"], pkg.extras["foi-email"])
assert_equal(package_data["foi-phone"], pkg.extras["foi-phone"])
# Themes and tags
assert_equal(package_data["theme-primary"], pkg.extras["theme-primary"])
assert_equal(set(package_data["theme-secondary"]), set(pkg.extras["theme-secondary"]))
# Health and Education are from the primary and secondary themes, which
# end up in the tags
assert_equal(
set(["tag1", "tag2", "a multi word tag", "Health", "Education"]), set(tag.name for tag in pkg.get_tags())
)
# Additional resources
expected_additional_keys = filter(lambda k: k.startswith("additional_resources"), package_data.keys())
additional_resources = ckanext.dgu.lib.helpers.additional_resources(pkg.as_dict())
assert_equal(len(additional_resources), 2)
for key in expected_additional_keys:
index, field = key.split("__")[1:]
index = int(index)
assert_equal(package_data[key], additional_resources[index][field])
assert_equal(package_data["mandate"], pkg.extras["mandate"])
assert_equal(package_data["access_constraints"], pkg.license_id)
assert_equal(package_data["temporal_coverage-from"], DateType.db_to_form(pkg.extras["temporal_coverage-from"]))
assert_equal(package_data["temporal_coverage-to"], DateType.db_to_form(pkg.extras["temporal_coverage-to"]))
assert_in("England", pkg.extras["geographic_coverage"])
def test_empty_url(self):
url = u''
result = self.check_link(url)
assert_in("URL parsing failure - did not find a host name", result['url_errors'])
def test_url_with_503(self, url):
result = self.check_link(url)
assert_in('Server returned HTTP error status: 503 Service Unavailable', result['url_errors'])
def test_file_url(self):
url = u'file:///home/root/test.txt'
result = self.check_link(url)
assert_in(u'Invalid url scheme. Please use one of: ftp http https',
result['url_errors'])
