def command(self):
self._load_config()
from ckan.lib.search import rebuild, check, show, clear
if not self.args:
# default to printing help
print self.usage
return
cmd = self.args[0]
if cmd == 'rebuild':
if len(self.args) > 1:
rebuild(self.args[1])
else:
rebuild()
elif cmd == 'check':
check()
elif cmd == 'show':
if not len(self.args) == 2:
import pdb; pdb.set_trace()
self.args
show(self.args[1])
elif cmd == 'clear':
clear()
else:
print 'Command %s not recognized' % cmd
def command(self):
self._load_config()
from ckan.lib.search import rebuild, check, show, clear
if not self.args:
# default to printing help
print self.usage
return
cmd = self.args[0]
if cmd == 'rebuild':
if len(self.args) > 1:
rebuild(self.args[1])
else:
rebuild()
elif cmd == 'check':
check()
elif cmd == 'show':
if not len(self.args) == 2:
import pdb
pdb.set_trace()
self.args
show(self.args[1])
elif cmd == 'clear':
clear()
else:
print 'Command %s not recognized' % cmd
def test_package_update_race_condition(self, lc_mock, dae_mock):
"""
Handling of package updates during extraction.
"""
res_dict = factories.Resource(**RES_DICT)
sysadmin = factories.Sysadmin()
def download_and_extract(*args, **kwargs):
# Simulate a change to the package by another party during
# the download and extraction process.
toolkit.get_action('package_patch')({
'user': sysadmin['name']
}, {
'id': res_dict['package_id'],
'title': 'A changed title'
})
return {'fulltext': 'foobar'}
dae_mock.side_effect = download_and_extract
extract(config['__file__'], res_dict)
# Make sure that the changed package metadata is kept and indexed
pkg_dict = toolkit.get_action('package_show')(
{}, {
'id': res_dict['package_id']
})
assert_equal(pkg_dict['title'], 'A changed title')
indexed_pkg_dict = search.show(res_dict['package_id'])
assert_equal(indexed_pkg_dict['title'], 'A changed title')
def show(self):
from ckan.lib.search import show
if not len(self.args) == 2:
print 'Missing parameter: dataset-name'
return
index = show(self.args[1])
pprint(index)
def test_indexed_package_stores_resource_type(self):
index = search.index.PackageSearchIndex()
pkg_dict = self._get_pkg_dict_with_resources()
index.index_package(pkg_dict)
indexed_pkg = search.show(pkg_dict['name'])
# Resource types are indexed
assert_equal(indexed_pkg['res_type'], ['doc', 'file'])
def test_index_package_stores_unvalidated_data_dict_without_validated_data_dict(self):
# This is a regression test for #2208
index = search.index.PackageSearchIndex()
pkg_dict = self._get_pkg_dict()
index.index_package(pkg_dict)
data_dict = json.loads(search.show(pkg_dict['name'])['data_dict'])
assert_not_in('validated_data_dict', data_dict)
def test_indexed_package_stores_resource_type(self):
index = search.index.PackageSearchIndex()
pkg_dict = self._get_pkg_dict_with_resources()
index.index_package(pkg_dict)
indexed_pkg = search.show(pkg_dict['name'])
# Resource types are indexed
assert_equal(indexed_pkg['res_type'], ['doc', 'file'])
def test_index_package_stores_unvalidated_data_dict_without_validated_data_dict(self):
# This is a regression test for #2208
index = search.index.PackageSearchIndex()
pkg_dict = self._get_pkg_dict()
index.index_package(pkg_dict)
data_dict = json.loads(search.show(pkg_dict['name'])['data_dict'])
assert_not_in('validated_data_dict', data_dict)
def test_indexed_package_stores_resource_type(self):
index = search.index.PackageSearchIndex()
pkg_dict = self._get_pkg_dict_with_resources()
index.index_package(pkg_dict)
indexed_pkg = search.show(pkg_dict["name"])
# Resource types are indexed
assert indexed_pkg["res_type"] == ["doc", "file"]
def test_index_package_stores_unvalidated_data_dict_without_validated_data_dict(
self, ):
# This is a regression test for #2208
index = search.index.PackageSearchIndex()
pkg_dict = self._get_pkg_dict()
index.index_package(pkg_dict)
data_dict = json.loads(search.show(pkg_dict["name"])["data_dict"])
assert "validated_data_dict" not in data_dict
def command(self):
self._load_config()
from ckan.lib.search import rebuild, check, show
if not self.args:
# default to run
cmd = 'rebuild'
else:
cmd = self.args[0]
if cmd == 'rebuild':
rebuild()
elif cmd == 'check':
check()
elif cmd == 'show':
if not len(self.args) == 2:
import pdb; pdb.set_trace()
self.args
show(self.args[1])
else:
print 'Command %s not recognized' % cmd
def test_index_package_stores_basic_solr_fields(self):
index = search.index.PackageSearchIndex()
pkg_dict = self._get_pkg_dict()
index.index_package(pkg_dict)
indexed_pkg = search.show(pkg_dict["name"])
# At root level are the fields that SOLR uses
assert indexed_pkg["name"] == "river-quality"
assert indexed_pkg["metadata_modified"] == "2014-06-10T08:24:12.782Z"
assert indexed_pkg["entity_type"] == "package"
assert indexed_pkg["dataset_type"] == "dataset"
def test_index_package_stores_validated_data_dict(self):
index = search.index.PackageSearchIndex()
pkg_dict = self._get_pkg_dict()
index.index_package(pkg_dict)
indexed_pkg = search.show(pkg_dict["name"])
# validated_data_dict is the result of package_show, validated
validated_data_dict = json.loads(indexed_pkg["validated_data_dict"])
assert validated_data_dict["name"] == "river-quality"
# title is inserted (copied from the name) during validation
# so its presence shows it is validated
assert "title" in validated_data_dict
def test_index_package_stores_validated_data_dict(self):
index = search.index.PackageSearchIndex()
pkg_dict = self._get_pkg_dict()
index.index_package(pkg_dict)
indexed_pkg = search.show(pkg_dict['name'])
# validated_data_dict is the result of package_show, validated
validated_data_dict = json.loads(indexed_pkg['validated_data_dict'])
assert_equal(validated_data_dict['name'], 'river-quality')
# title is inserted (copied from the name) during validation
# so its presence shows it is validated
assert_in('title', validated_data_dict)
def test_index_package_stores_basic_solr_fields(self):
index = search.index.PackageSearchIndex()
pkg_dict = self._get_pkg_dict()
index.index_package(pkg_dict)
indexed_pkg = search.show(pkg_dict['name'])
# At root level are the fields that SOLR uses
assert_equal(indexed_pkg['name'], 'river-quality')
assert_equal(indexed_pkg['metadata_modified'],
'2014-06-10T08:24:12.782Z')
assert_equal(indexed_pkg['entity_type'], 'package')
assert_equal(indexed_pkg['dataset_type'], 'dataset')
def test_index_package_stores_basic_solr_fields(self):
index = search.index.PackageSearchIndex()
pkg_dict = self._get_pkg_dict()
index.index_package(pkg_dict)
indexed_pkg = search.show(pkg_dict['name'])
# At root level are the fields that SOLR uses
assert_equal(indexed_pkg['name'], 'river-quality')
assert_equal(indexed_pkg['metadata_modified'],
'2014-06-10T08:24:12.782Z')
assert_equal(indexed_pkg['entity_type'], 'package')
assert_equal(indexed_pkg['dataset_type'], 'dataset')
def test_index_package_stores_validated_data_dict(self):
index = search.index.PackageSearchIndex()
pkg_dict = self._get_pkg_dict()
index.index_package(pkg_dict)
indexed_pkg = search.show(pkg_dict['name'])
# validated_data_dict is the result of package_show, validated
validated_data_dict = json.loads(indexed_pkg['validated_data_dict'])
assert_equal(validated_data_dict['name'], 'river-quality')
# title is inserted (copied from the name) during validation
# so its presence shows it is validated
assert_in('title', validated_data_dict)
def test_index_package_stores_resource_extras_in_config_file(self):
index = search.index.PackageSearchIndex()
pkg_dict = self._get_pkg_dict_with_resources()
index.index_package(pkg_dict)
indexed_pkg = search.show(pkg_dict['name'])
# Resource fields given by ckan.extra_resource_fields are indexed
assert_equal(indexed_pkg['res_extras_alt_url'],
['http://www.bar.com/riverquality.pdf',
'http://www.bar.com/riverquality.csv'])
# Other resource fields are ignored
assert_equal(indexed_pkg.get('res_extras_institution', None), None)
assert_equal(indexed_pkg.get('res_extras_city', None), None)
def test_index_package_stores_resource_extras_in_config_file(self):
index = search.index.PackageSearchIndex()
pkg_dict = self._get_pkg_dict_with_resources()
index.index_package(pkg_dict)
indexed_pkg = search.show(pkg_dict['name'])
# Resource fields given by ckan.extra_resource_fields are indexed
assert_equal(indexed_pkg['res_extras_alt_url'],
['http://www.bar.com/riverquality.pdf',
'http://www.bar.com/riverquality.csv'])
# Other resource fields are ignored
assert_equal(indexed_pkg.get('res_extras_institution', None), None)
assert_equal(indexed_pkg.get('res_extras_city', None), None)
def test_index_package_stores_resource_extras_in_config_file(self):
index = search.index.PackageSearchIndex()
pkg_dict = self._get_pkg_dict_with_resources()
index.index_package(pkg_dict)
indexed_pkg = search.show(pkg_dict["name"])
# Resource fields given by ckan.extra_resource_fields are indexed
assert indexed_pkg["res_extras_alt_url"] == [
"http://www.bar.com/riverquality.pdf",
"http://www.bar.com/riverquality.csv",
]
# Other resource fields are ignored
assert indexed_pkg.get("res_extras_institution", None) is None
assert indexed_pkg.get("res_extras_city", None) is None
def test_package_update_race_condition(self, lc_mock, dae_mock):
"""
Handling of package updates during extraction.
"""
res_dict = factories.Resource(**RES_DICT)
sysadmin = factories.Sysadmin()
def download_and_extract(*args, **kwargs):
# Simulate a change to the package by another party during
# the download and extraction process.
toolkit.get_action('package_patch')({'user': sysadmin['name']},
{'id': res_dict['package_id'],
'title': 'A changed title'})
return {'fulltext': 'foobar'}
dae_mock.side_effect = download_and_extract
extract(config['__file__'], res_dict)
# Make sure that the changed package metadata is kept and indexed
pkg_dict = toolkit.get_action('package_show')(
{}, {'id': res_dict['package_id']})
assert_equal(pkg_dict['title'], 'A changed title')
indexed_pkg_dict = search.show(res_dict['package_id'])
assert_equal(indexed_pkg_dict['title'], 'A changed title')
def _modification_time_in_the_search_index(pkg):
pkg_dict = show(pkg.name)
mod_time_str = pkg_dict['last_major_modification']
return dateutil.parser.parse(mod_time_str).replace(tzinfo=None)
def test_dataset_properties_in_solr(self):
try:
search_result = search.show('test_private_dataset_1')
except Exception, e:
assert False, 'search query needs to succeed'
def show(dataset_name: str):
from ckan.lib.search import show
index = show(dataset_name)
click.echo(index)
def _modification_time_in_the_search_index(pkg):
pkg_dict = show(pkg.name)
mod_time_str = pkg_dict['last_major_modification']
return dateutil.parser.parse(mod_time_str).replace(tzinfo=None)
def show(dataset_name):
from ckan.lib.search import show
index = show(dataset_name)
click.echo(index)
def package_show(context, data_dict):
'''Return the metadata of a dataset (package) and its resources.
:param id: the id or name of the dataset
:type id: string
:param use_default_schema: use default package schema instead of
a custom schema defined with an IDatasetForm plugin (default: False)
:type use_default_schema: bool
:param include_tracking: add tracking information to dataset and
resources (default: False)
:type include_tracking: bool
:rtype: dictionary
'''
model = context['model']
context['session'] = model.Session
name_or_id = data_dict.get("id") or _get_or_bust(data_dict, 'name_or_id')
pkg = model.Package.get(name_or_id)
if pkg is None:
raise NotFound
context['package'] = pkg
_check_access('package_show', context, data_dict)
if data_dict.get('use_default_schema', False):
context['schema'] = ckan.logic.schema.default_show_package_schema()
include_tracking = asbool(data_dict.get('include_tracking', False))
package_dict = None
use_cache = (context.get('use_cache', True)
and not 'revision_id' in context
and not 'revision_date' in context)
if use_cache:
try:
search_result = search.show(name_or_id)
except (search.SearchError, socket.error):
pass
else:
use_validated_cache = 'schema' not in context
if use_validated_cache and 'validated_data_dict' in search_result:
package_json = search_result['validated_data_dict']
package_dict = json.loads(package_json)
package_dict_validated = True
else:
package_dict = json.loads(search_result['data_dict'])
package_dict_validated = False
metadata_modified = pkg.metadata_modified.isoformat()
search_metadata_modified = search_result['metadata_modified']
# solr stores less precice datetime,
# truncate to 22 charactors to get good enough match
if metadata_modified[:22] != search_metadata_modified[:22]:
package_dict = None
if not package_dict:
package_dict = model_dictize.package_dictize(pkg, context)
package_dict_validated = False
if include_tracking:
# page-view tracking summary data
package_dict['tracking_summary'] = (
model.TrackingSummary.get_for_package(package_dict['id']))
for resource_dict in package_dict['resources']:
_add_tracking_summary_to_resource_dict(resource_dict, model)
if context.get('for_view'):
for item in plugins.PluginImplementations(plugins.IPackageController):
package_dict = item.before_view(package_dict)
for item in plugins.PluginImplementations(plugins.IPackageController):
item.read(pkg)
for item in plugins.PluginImplementations(plugins.IResourceController):
for resource_dict in package_dict['resources']:
item.before_show(resource_dict)
if not package_dict_validated:
package_plugin = lib_plugins.lookup_package_plugin(
package_dict['type'])
if 'schema' in context:
schema = context['schema']
else:
schema = package_plugin.show_package_schema()
if schema and context.get('validate', True):
package_dict, errors = lib_plugins.plugin_validate(
package_plugin, context, package_dict, schema, 'package_show')
for item in plugins.PluginImplementations(plugins.IPackageController):
item.after_show(context, package_dict)
return package_dict
