def parse_url(url, quiet=False, rid=''):
'''Parse the datasets in a DCAT format located at URL (debug)'''
if quiet:
verbose_loggers = ['rdflib', 'udata.core.dataset']
[logging.getLogger(l).setLevel(logging.ERROR) for l in verbose_loggers]
class MockSource:
url = ''
class MockJob:
items = []
class MockDatasetFactory(DatasetFactory):
'''Use DatasetFactory without .save()'''
@classmethod
def _create(cls, model_class, *args, **kwargs):
instance = model_class(*args, **kwargs)
return instance
echo(cyan('Parsing url {}'.format(url)))
source = MockSource()
source.url = url
backend = DcatBackend(source, dryrun=True)
backend.job = MockJob()
format = backend.get_format()
echo(yellow('Detected format: {}'.format(format)))
graph = backend.parse_graph(url, format)
# serialize/unserialize graph like in the job mechanism
_graph = graph.serialize(format=format, indent=None)
graph = Graph(namespace_manager=namespace_manager)
graph.parse(data=_graph, format=format)
for item in backend.job.items:
if not rid or rid in item.remote_id:
echo(magenta('Processing item {}'.format(item.remote_id)))
echo('Item kwargs: {}'.format(yellow(item.kwargs)))
node = backend.get_node_from_item(item)
dataset = MockDatasetFactory()
dataset = dataset_from_rdf(graph, dataset, node=node)
echo('')
echo(green('Dataset found!'))
echo('Title: {}'.format(yellow(dataset)))
echo('License: {}'.format(yellow(dataset.license)))
echo('Description: {}'.format(yellow(dataset.description)))
echo('Tags: {}'.format(yellow(dataset.tags)))
echo('Resources: {}'.format(yellow([(r.title, r.format, r.url) for r in dataset.resources])))
try:
dataset.validate()
except mongoengine.errors.ValidationError as e:
log.error(e, exc_info=True)
else:
echo(green('Dataset is valid ✅'))
echo('')
def migrate(record):
'''Perform database migrations'''
handler = record_migration if record else execute_migration
for plugin, package, filename in available_migrations():
migration = get_migration(plugin, filename)
if migration:
log_status(plugin, filename, cyan('Skipped'))
else:
status = purple('Recorded') if record else yellow('Apply')
log_status(plugin, filename, status)
script = resource_string(package, join('migrations', filename))
handler(plugin, filename, script)
def migrate(record, dryrun=False):
'''Perform database migrations'''
handler = record_migration if record else execute_migration
success = True
for plugin, package, filename in available_migrations():
migration = get_migration(plugin, filename)
if migration or not success:
log_status(plugin, filename, cyan('Skipped'))
else:
status = purple('Recorded') if record else yellow('Apply')
log_status(plugin, filename, status)
script = resource_string(package, join('migrations', filename))
success &= handler(plugin, filename, script, dryrun=dryrun)
def migrate(record, dry_run=False):
'''Perform database migrations'''
success = True
for migration in migrations.list_available():
if migration.record.ok or not success:
log_status(migration, cyan('Skipped'))
else:
status = magenta('Recorded') if record else yellow('Apply')
log_status(migration, status)
try:
output = migration.execute(recordonly=record, dryrun=dry_run)
except migrations.RollbackError as re:
format_output(re.migrate_exc.output, False)
log_status(migration, red('Rollback'))
format_output(re.output, not re.exc)
success = False
except migrations.MigrationError as me:
format_output(me.output, False, traceback=me.traceback)
success = False
else:
format_output(output, True)
return success
def render():
'''Force (re)rendering stored images'''
print(cyan('Rendering images'))
count = Counter()
total = Counter()
organizations = Organization.objects(logo__exists=True)
total['orgs'] = organizations.count()
print(white('Processing {0} organizations logos'.format(total['orgs'])))
for org in organizations:
count['orgs'] += render_or_skip(org, 'logo')
users = User.objects(avatar__exists=True)
total['users'] = users.count()
print(white('Processing {0} user avatars'.format(total['users'])))
for user in users:
count['users'] += render_or_skip(user, 'avatar')
posts = Post.objects(image__exists=True)
total['posts'] = posts.count()
print(white('Processing {0} post images'.format(total['posts'])))
for post in posts:
count['posts'] += render_or_skip(post, 'image')
reuses = Reuse.objects(image__exists=True)
total['reuses'] = reuses.count()
print(white('Processing {0} reuse images'.format(total['reuses'])))
for reuse in reuses:
count['reuses'] += render_or_skip(reuse, 'image')
print(white('Summary:'))
print('''
Organization logos: {count[orgs]}/{total[orgs]}
User avatars: {count[users]}/{total[users]}
Post images: {count[posts]}/{total[posts]}
Reuse images: {count[reuses]}/{total[reuses]}
'''.format(count=count, total=total))
print(green('Images rendered'))
