def delete_matching_documents(index, type, field, value, **kwargs):
# open connection if one hasn't been provided
conn = kwargs.get('conn', False)
if not conn:
conn = connect_and_create_index(index)
# a max_documents of 0 means unlimited
max_documents = kwargs.get('max_documents', 0)
# cycle through fields to find matches
documents = conn.search_raw(
indices=[index],
doc_types=[type],
query=pyes.FieldQuery(pyes.FieldParameter(field, value))
)
count = 0
if len(documents['hits']['hits']) > 0:
for hit in documents['hits']['hits']:
document_id = hit['_id']
conn.delete(index, type, document_id)
count = count + 1
if count == max_documents:
return count
return count
def connect_and_get_aip_data(uuid):
conn = connect_and_create_index('aips')
aips = conn.search(
query=pyes.FieldQuery(pyes.FieldParameter('uuid', uuid)),
fields='uuid,name,filePath,size,origin,created'
)
return aips[0]
def archival_storage_search_augment_aip_results(conn, aips):
for aip_uuid in aips:
documents = conn.search_raw(
query=pyes.FieldQuery(pyes.FieldParameter('uuid', aip_uuid.term)))
if len(documents['hits']['hits']) > 0:
aip_uuid.name = documents['hits']['hits'][0]['_source']['name']
aip_uuid.size = '{0:.2f} MB'.format(
documents['hits']['hits'][0]['_source']['size'])
aip_uuid.date = documents['hits']['hits'][0]['_source']['created']
else:
aip_uuid.name = '(data missing)'
def connect_and_delete_aip_files(uuid):
deleted = 0
conn = pyes.ES(getElasticsearchServerHostAndPort())
documents = conn.search_raw(query=pyes.FieldQuery(pyes.FieldParameter('AIPUUID', uuid)))
if len(documents['hits']['hits']) > 0:
for hit in documents['hits']['hits']:
document_id = hit['_id']
conn.delete('aips', 'aipfile', document_id)
deleted = deleted + 1
print str(deleted) + ' index documents removed.'
else:
print 'No AIP files found.'
def search_augment_aip_results(conn, aips):
for aip_uuid in aips:
documents = conn.search_raw(query=pyes.FieldQuery(
pyes.FieldParameter('uuid', aip_uuid.term)),
fields='name,size,created')
if len(documents['hits']['hits']) > 0:
aip_uuid.name = documents['hits']['hits'][0]['fields']['name']
aip_uuid.size = '{0:.2f} MB'.format(
documents['hits']['hits'][0]['fields']['size'])
aip_uuid.date = documents['hits']['hits'][0]['fields']['created']
aip_uuid.document_id_no_hyphens = documents['hits']['hits'][0][
'_id'].replace('-', '____')
else:
aip_uuid.name = '(data missing)'
def connect_and_remove_sip_transfer_files(uuid):
# get file UUIDs for each file in the SIP
sql = "SELECT fileUUID from Files WHERE sipUUID='" + MySQLdb.escape_string(uuid) + "'"
rows = databaseInterface.queryAllSQL(sql)
if len(rows) > 0:
conn = connect_and_create_index('transfers')
# cycle through file UUIDs and delete files from transfer backlog
for row in rows:
documents = conn.search_raw(query=pyes.FieldQuery(pyes.FieldParameter('fileuuid', row[0])))
if len(documents['hits']['hits']) > 0:
document_id = documents['hits']['hits'][0]['_id']
conn.delete('transfers', 'transferfile', document_id)
def connect_and_change_transfer_file_status(uuid, status):
# get file UUIDs for each file in the SIP
sql = "SELECT fileUUID from Files WHERE transferUUID='" + MySQLdb.escape_string(uuid) + "'"
rows = databaseInterface.queryAllSQL(sql)
if len(rows) > 0:
conn = connect_and_create_index('transfers')
# cycle through file UUIDs and update status
for row in rows:
documents = conn.search_raw(query=pyes.FieldQuery(pyes.FieldParameter('fileuuid', row[0])))
if len(documents['hits']['hits']) > 0:
document_id = documents['hits']['hits'][0]['_id']
conn.update({'status': status}, 'transfers', 'transferfile', document_id)
return len(rows)
def connect_and_get_aip_data(uuid):
conn = connect_and_create_index('aips')
aips = conn.search(
query=pyes.FieldQuery(pyes.FieldParameter('uuid', uuid)))
return aips[0]
