Bug #92
closedneed access control tracking for metadata documents
0%
Description
Need to create a mechanism for tracking access control information for metadata
documents. This might be doen by creating a new set of attributes and tables to
store access control lists (acl), or we might try to store this information in
the SDSC MCAT along with the existing ACL info for datasets. The latter would
likely be more complicated, so initial attempts will probably create local
storage for the ACLs.
Also, need an easy-to-use API for the MetaCatServlet through which clients (like
dmanclient) can query and change the access control lists (either the owner of
document is only person who can change acls, or there needs to be a specific
permission granting the ability to change acls).
Related issues
Updated by Jivka Bojilova over 23 years ago
DONE
1. ACL info for resources (data and metadata docs) is applied through access
files only.
2. Access files are submitted to Metacat in the way like other metadata
docs with the difference that in background records are created in
xml_access and xml_relation tables for use.
3. Access files could be submitted to Metacat only after submission of the
resources specified within them by <resourceIdentifier> tags. Resources are
specified with their whole metacat URLs. docids are extracted from the URL by
parsing the URL query string using MetaCatUtil.parseQuery(URL murl.getQuery())
routine.
4. Access file can specify acl info for resources only owned by the user
submitting the access file or having "all" permissions on all off them. In other
case submission of the access file is rejected.
5. Access files itself get by default public read access as all metadata docs
which is convenient for now during development by probably should be made
optional in order to be specified by the client. (?)
6. It is possible same permission for a user on given resource to be specified
from different access files. In this case the permission that is the most (by
time duration, by perm_order: "allowFirst" or "denyFirst") is used (simple
algorithm is implemented).
7. "accesfileid" attr in xml_access table stores docid of the access
file. Also "docid" attr in xml_relation table is again the docid of the access
file that brings the relationships (<accessfile, 'isaclfor', resource>).
These attrs are convenient when access file is updated or deleted to delete the
related records from xml_access and xml_relation tables first (these 2 tables
keep only the current/last version of the access file)
8. "read" action checks the user for having "read" permission on the found docs
(if not publicly readable or not his/her own). Only docs on which user have
"read" permission are extracted (along with his/her own and publicly readable).
9. "update" and "delete" actions check for "write" permission on the manipulated
doc (owned docs are permitted again).