def irods_transform_meta_data(ibcontext, **kwargs):
"""
Transform meta data
"""
cfg = ibcontext['irods'].get_config(kwargs)
key = {
'irods_zone': get_irods_zone(cfg),
'irods_collection': cfg['irods_collection']
}
data = ibcontext['cache'].read(key)
if data is None:
raise ValueError(
'collection not found in cache: {0} (zone: {1})'.format(
key['irods_collection'], key['irods_zone']))
data = extract_irods_collecion_data(ibcontext['cache'].read(key),
cfg['irods_collection'])
md = {
cfg['irods_collection']:
transform_meta_data(data['meta_data'], COLLECTION_META_DATA_MAPPING)
}
for obj in data['objects']:
obj['meta_data']['name'] = os.path.basename(obj['path'])
md[obj['path']] = transform_meta_data(obj['meta_data'],
OBJECT_META_DATA_MAPPING)
ibcontext['cache'].update(key, {'target_meta_data': md})
def irods_validate_meta_data(ibcontext, **kwargs):
"""
Validate meta data
"""
cfg = ibcontext['irods'].get_config(kwargs)
key = {
'irods_zone': get_irods_zone(cfg),
'irods_collection': cfg['irods_collection']
}
data = ibcontext['cache'].read(key)
if data is None:
raise ValueError(
'collection not found in cache: {0} (zone: {1})'.format(
key['irods_collection'], key['irods_zone']))
validate_collection_meta_data(data, cfg['irods_collection'],
COLLECTION_META_DATA_VALIDATION)
validate_objects_meta_data(data, cfg['irods_collection'],
OBJECT_META_DATA_VALIDATION)
def create_package(ibcontext, **kwargs):
logger = logging.getLogger('ipublish')
irods_cfg = ibcontext['irods'].get_config(kwargs)
ckan_cfg = ibcontext['ckan'].get_config(kwargs)
key = {
'irods_zone': get_irods_zone(irods_cfg),
'irods_collection': irods_cfg['irods_collection']
}
data = ibcontext['cache'].read(key)
if data is None:
raise KeyError('No such key {0}'.format(key))
collection = [
x for x in data.get('irods_data')
if x['path'] == irods_cfg['irods_collection']
][0]
target_meta_data = data['target_meta_data']
md = target_meta_data[irods_cfg['irods_collection']].copy()
md['owner_org'] = ckan_cfg['ckan_org']
result = ibcontext['ckan'].action('package_create', data=md, method='post')
id = result['result']['id']
logger.info('created package with id {0}'.format(id))
logger.debug(result['result'])
result['result']['resources'] = []
try:
for obj in collection['objects']:
obj_md = target_meta_data[obj['path']]
obj_md['package_id'] = id
obj_res = ibcontext['ckan'].action('resource_create',
data=obj_md,
method='post')
result['result']['resources'].append(obj_res)
except Exception:
ibcontext['ckan'].action('package_delete',
data={'id': id},
method='post')
raise
ibcontext['cache'].update(key, {'ckan_package': result['result']})
repo_url = ckan_cfg['ckan_dataset_url'] + '/' + md['name']
repository_info = {'PUBLIC_URL': repo_url, 'PUBLIC_REPO_TYPE': 'CKAN'}
ibcontext['cache'].update(key, {'repository_info': repository_info})
def create_draft(ibcontext, **kwargs):
logger = logging.getLogger('ipublish')
irods_cfg = ibcontext['irods'].get_config(kwargs)
config = ibcontext['b2share'].get_config(kwargs)
key = {'irods_zone': get_irods_zone(irods_cfg),
'irods_collection': irods_cfg['irods_collection']}
logger.debug(config)
meta_data = ibcontext['cache'].read(key)\
.get('target_meta_data')\
.get(irods_cfg['irods_collection'])
logger.debug(meta_data)
api_url = config.get('b2share_api_url')
access_token = config.get('b2share_access_token')
community = config.get('b2share_community')
url = urljoin(api_url, 'records') + '/'
data = {'titles': [{'title': meta_data['title']}],
'community': community,
'open_access': True,
'community_specific': {}}
logger.debug('url: %s', url)
logger.debug(data)
req = requests.post(url,
data=json.dumps(data),
params={"access_token": access_token},
headers={"Content-Type": "application/json"})
req.raise_for_status()
if req.status_code != 201:
raise ValueError('unexpected status code %d', req.status_code)
b2share_id = req.json()['id']
logger.debug('created draft %s', b2share_id)
ibcontext['cache'].update(key,
{'b2share_id': b2share_id})
repo_url = urljoin(api_url,
'records',
b2share_id,
'draft')
repository_info = {'PUBLIC_URL': repo_url,
'PUBLIC_REPO_TYPE': 'B2SHARE'}
ibcontext['cache'].update(key,
{'repository_info': repository_info})
def patch_draft(ibcontext, **kwargs):
logger = logging.getLogger('ipublish')
irods_cfg = ibcontext['irods'].get_config(kwargs)
config = ibcontext['b2share'].get_config(kwargs)
api_url = config.get('b2share_api_url')
access_token = config.get('b2share_access_token')
key = {'irods_zone': get_irods_zone(irods_cfg),
'irods_collection': irods_cfg['irods_collection']}
logger.debug(config)
meta_data = ibcontext['cache'].read(key)\
.get('target_meta_data')\
.get(irods_cfg['irods_collection'])
b2share_id = ibcontext['cache'].read(key)\
.get('b2share_id')
logger.debug('b2share id: %s', b2share_id)
pids = []
for item in ibcontext['cache'].read(key)\
.get('target_meta_data').values():
pids.append({'alternate_identifier': item['pid'],
'alternate_identifier_type': 'http://hdl.handle.net/'})
patch = [{'op': 'add',
'path': '/creators',
'value': [{'creator_name': meta_data['author']}]},
{'op': 'add',
'path': '/alternate_identifiers',
'value': pids}]
logger.debug(patch)
url = urljoin(api_url,
'records',
b2share_id,
'draft')
req = requests.patch(url=url,
params={'access_token':
access_token},
headers={"Content-Type":
"application/json-patch+json"},
data=json.dumps(patch))
req.raise_for_status()
def wait_for_confirmation(ibcontext, **kwargs):
logger = logging.getLogger('ipublish')
b2share_cfg = ibcontext['b2share'].get_config(kwargs)
irods_cfg = ibcontext['irods'].get_config(kwargs)
api_url = b2share_cfg.get('b2share_api_url')
access_token = b2share_cfg.get('b2share_access_token')
key = {'irods_zone': get_irods_zone(irods_cfg),
'irods_collection': irods_cfg['irods_collection']}
b2share_id = ibcontext['cache'].read(key)\
.get('b2share_id')
logger.debug('b2share id: %s', b2share_id)
url = urljoin(api_url,
'records',
b2share_id,
'draft')
state = 'draft'
while state == 'draft':
req = requests.get(url=url,
params={'access_token': access_token})
req.raise_for_status()
state = req.json().get('metadata', {}).get('publication_state', None)
logger.debug('url: %s:%s', url, state)
if state == 'draft':
time.sleep(5)