def delete(self, userId, bucket, archiveid):
if not self.archive_clients:
raise Exception("archive not initialized")
try:
url = None
with session_scope() as dbsession:
meta = db_archivemetadata.get(userId,
bucket,
archiveid,
session=dbsession)
if meta:
url = meta.get("content_url")
db_archivemetadata.delete(userId,
bucket,
archiveid,
session=dbsession)
# Remove the data itself. Can result in orphaned data if system fails here but better than deleting the content but not the meta, leaving a confused state.
if url:
scheme = urllib.parse.urlparse(url).scheme
if scheme in self.archive_clients:
return self.archive_clients[scheme].delete(
userId, bucket, archiveid)
else:
logger.warn(
"Deleting archive document {}/{}/{}, but found no content url for backend delete so skipping backend operation"
.format(userId, bucket, archiveid))
return url is not None
except Exception as err:
raise err
def get(self, userId, bucket, archiveid):
if not self.archive_clients:
raise Exception("archive not initialized")
try:
with session_scope() as dbsession:
result = db_archivemetadata.get(
userId, bucket, archiveid, session=dbsession
)
if not result:
return None
url = result.get("content_url")
if url is None:
raise Exception(
"Null reference url for valid metadata record for {}/{}/{}".format(
userId, bucket, archiveid
)
)
else:
is_compressed = result.get("is_compressed", False)
expected = result.get("digest")
# get the raw data from driver, note content is bytes (not str)
content = self._client_for(url).get_by_uri(url)
found_size = len(content)
if expected:
found = hashlib.md5(content).hexdigest()
else:
found = None
if expected and found != expected:
logger.error(
"Digest mismatch:\nDB Record: {}\nContent digest: {}\n, Content size: {}".format(
result, found, found_size, content
)
)
raise Exception(
"Detected digest mismatch on content fetch from backend. Expected: {}, Got: {}".format(
expected, found
)
)
content = self._do_decompress(is_compressed, content)
return content
except Exception as err:
import traceback
traceback.print_exc()
logger.debug("cannot get data: exception - " + str(err))
raise err
def get_document_meta(self, userId, bucket, archiveId):
if not self.archive_clients:
raise Exception("archive not initialized")
with session_scope() as dbsession:
ret = db_archivemetadata.get(userId,
bucket,
archiveId,
session=dbsession)
return (ret)
def get_document_meta(self, userId, bucket, archiveId):
"""
Return the metadata for the doc, or None if not found
:param userId:
:param bucket:
:param archiveId:
:return:
"""
if not self.archive_clients:
raise Exception("archive not initialized")
with session_scope() as dbsession:
ret = db_archivemetadata.get(userId, bucket, archiveId, session=dbsession)
if ret is None:
return None
else:
return ret
