def index_files(self, es_client, transfer_path, transfer_uuid):
transfer_path = elasticSearchFunctions.index_files(
es_client,
'transfers',
'transferfile',
transfer_uuid,
os.path.join(transfer_path, ''), # Expected by index_files().
status='backlog')
# archivematicaCommon
from custom_handlers import get_script_logger
import elasticSearchFunctions
from django.conf import settings as mcpclient_settings
logger = get_script_logger(
'archivematica.mcp.client.elasticSearchIndexProcessTransfer')
if __name__ == '__main__':
if not mcpclient_settings.SEARCH_ENABLED:
logger.info('Skipping indexing: indexing is currently disabled.')
sys.exit(0)
transfer_path = sys.argv[1]
transfer_uuid = sys.argv[2]
try:
status = sys.argv[3]
except IndexError:
status = ''
elasticSearchFunctions.setup_reading_from_conf(mcpclient_settings)
client = elasticSearchFunctions.get_client()
sys.exit(
elasticSearchFunctions.index_files(client,
'transfers',
'transferfile',
transfer_uuid,
transfer_path,
status=status))
def index_aip():
""" Write AIP information to ElasticSearch. """
sip_uuid = sys.argv[1] # %SIPUUID%
sip_name = sys.argv[2] # %SIPName%
sip_path = sys.argv[3] # %SIPDirectory%
sip_type = sys.argv[4] # %SIPType%
# Check if ElasticSearch is enabled
client_config_path = '/etc/archivematica/MCPClient/clientConfig.conf'
config = ConfigParser.SafeConfigParser()
config.read(client_config_path)
elastic_search_disabled = False
try:
elastic_search_disabled = config.getboolean(
'MCPClient', "disableElasticsearchIndexing")
except ConfigParser.NoOptionError:
pass
if elastic_search_disabled:
print('Skipping indexing: indexing is currently disabled in',
client_config_path)
return 0
elasticSearchFunctions.setup_reading_from_client_conf(config)
client = elasticSearchFunctions.get_client()
print('SIP UUID:', sip_uuid)
aip_info = storage_service.get_file_info(uuid=sip_uuid)
print('AIP info:', aip_info)
aip_info = aip_info[0]
mets_name = 'METS.{}.xml'.format(sip_uuid)
mets_path = os.path.join(sip_path, mets_name)
identifiers = get_identifiers(sip_path)
# If this is an AIC, find the number of AIP stored in it and index that
aips_in_aic = None
if sip_type == "AIC":
try:
uv = UnitVariable.objects.get(unittype="SIP",
unituuid=sip_uuid,
variable="AIPsinAIC")
aips_in_aic = uv.variablevalue
except UnitVariable.DoesNotExist:
pass
print('Indexing AIP info')
# Delete ES index before creating new one if reingesting
if 'REIN' in sip_type:
print('Deleting outdated entry for AIP and AIP files with UUID',
sip_uuid, 'from archival storage')
elasticSearchFunctions.delete_aip(client, sip_uuid)
elasticSearchFunctions.delete_aip_files(client, sip_uuid)
# Index AIP
elasticSearchFunctions.index_aip(client,
sip_uuid,
sip_name,
aip_info['current_full_path'],
mets_path,
size=aip_info['size'],
aips_in_aic=aips_in_aic,
identifiers=identifiers)
# Index AIP files
print('Indexing AIP files')
# Even though we treat MODS identifiers as SIP-level, we need to index them
# here because the archival storage tab actually searches on the
# aips/aipfile index.
exitCode = elasticSearchFunctions.index_files(
client,
index='aips',
type_='aipfile',
uuid=sip_uuid,
pathToArchive=sip_path,
identifiers=identifiers,
sipName=sip_name,
)
if exitCode == 1:
print('Error indexing AIP files', file=sys.stderr)
return 1
return 0
def index_aip():
""" Write AIP information to ElasticSearch. """
sip_uuid = sys.argv[1] # %SIPUUID%
sip_name = sys.argv[2] # %SIPName%
sip_path = sys.argv[3] # %SIPDirectory%
sip_type = sys.argv[4] # %SIPType%
if not mcpclient_settings.SEARCH_ENABLED:
logger.info('Skipping indexing: indexing is currently disabled.')
return 0
elasticSearchFunctions.setup_reading_from_conf(mcpclient_settings)
client = elasticSearchFunctions.get_client()
print('SIP UUID:', sip_uuid)
aip_info = storage_service.get_file_info(uuid=sip_uuid)
print('AIP info:', aip_info)
aip_info = aip_info[0]
mets_name = 'METS.{}.xml'.format(sip_uuid)
mets_path = os.path.join(sip_path, mets_name)
identifiers = get_identifiers(sip_path)
# If this is an AIC, find the number of AIP stored in it and index that
aips_in_aic = None
if sip_type == "AIC":
try:
uv = UnitVariable.objects.get(unittype="SIP",
unituuid=sip_uuid,
variable="AIPsinAIC")
aips_in_aic = uv.variablevalue
except UnitVariable.DoesNotExist:
pass
print('Indexing AIP info')
# Delete ES index before creating new one if reingesting
if 'REIN' in sip_type:
print('Deleting outdated entry for AIP and AIP files with UUID',
sip_uuid, 'from archival storage')
elasticSearchFunctions.delete_aip(client, sip_uuid)
elasticSearchFunctions.delete_aip_files(client, sip_uuid)
# Index AIP
elasticSearchFunctions.index_aip(client,
sip_uuid,
sip_name,
aip_info['current_full_path'],
mets_path,
size=aip_info['size'],
aips_in_aic=aips_in_aic,
identifiers=identifiers,
encrypted=aip_info['encrypted'])
# Index AIP files
print('Indexing AIP files')
# Even though we treat MODS identifiers as SIP-level, we need to index them
# here because the archival storage tab actually searches on the
# aips/aipfile index.
exitCode = elasticSearchFunctions.index_files(
client,
index='aips',
type_='aipfile',
uuid=sip_uuid,
pathToArchive=sip_path,
identifiers=identifiers,
sipName=sip_name,
)
if exitCode == 1:
print('Error indexing AIP files', file=sys.stderr)
return 1
return 0