feat: enhance FilteredFileAdapter to handle flexible filtering for policies and roles

[HashCookie]

Fix: #361

• The old logic's handling of the g rules is too simplistic. If filter[1] is a null value (indicating there are no g rules to filter), the old logic will still continue to call filter_words for filtering, which may lead to unnecessary misfiltering.
• When filtering g rules, the filter may be a null value. The old logic did not explicitly handle this situation, which could lead to all g rules being skipped (misfiltered).

Changes

1. Improved Filtering Logic:

• Refactored filter_line and filter_words to distinguish between p and g rules.
• Added explicit handling for empty filters, ensuring that g rules can still be loaded when adapter.filter.G is empty.

2. Optimized Role Rule Handling:

• Ensured that g rules are only filtered when adapter.filter.G contains non-empty criteria, preventing unintentional exclusion of role inheritance relationships.

Test

import casbin

if __name__ == "__main__":
     adapter = casbin.persist.adapters.filtered_file_adapter.FilteredFileAdapter(
         "policy.csv"
     )
     adapter.filter.P = ["", "domain1", "", ""]
     adapter.filter.G = ["", "", "domain1"]

     model = casbin.Enforcer.new_model("model.conf")
     filtered_enforcer = casbin.Enforcer(model, adapter)

     filtered_enforcer.load_filtered_policy(filtered_enforcer.adapter.filter)
     print(f"filtered_enforcer.get_policy(): {filtered_enforcer.get_policy()}")

     unfiltered_enforcer = casbin.Enforcer("model.conf", "policy.csv")
     print(f"unfiltered_enforcer.get_policy(): {unfiltered_enforcer.get_policy()}")

     request_vals = ["alice", "domain1", "data1", "read"]

     print(f"request to filtered_enforcer: {filtered_enforcer.enforce(*request_vals)} (expected True)") 
     print(f"request to unfiltered_enforcer: {unfiltered_enforcer.enforce(*request_vals)} (expected True)") 

     print("Filtered Policy Rules:", filtered_enforcer.get_policy())
     print("Filtered Role Rules:", filtered_enforcer.get_grouping_policy())

     print("Before filtering:", unfiltered_enforcer.get_policy())
     print("After filtering:", filtered_enforcer.get_policy())
     print("Role assignments:", filtered_enforcer.get_grouping_policy())

Return

filtered_enforcer.get_policy(): [['admin', 'domain1', 'data1', 'read'], ['admin', 'domain1', 'data1', 'write']]
unfiltered_enforcer.get_policy(): [['admin', 'domain1', 'data1', 'read'], ['admin', 'domain1', 'data1', 'write'], ['admin', 'domain2', 'data2', 'read'], ['admin', 'domain2', 'data2', 'write']]
request to filtered_enforcer: True (expected True)
request to unfiltered_enforcer: True (expected True)
Filtered Policy Rules: [['admin', 'domain1', 'data1', 'read'], ['admin', 'domain1', 'data1', 'write']]
Filtered Role Rules: [['alice', 'admin', 'domain1']]
Before filtering: [['admin', 'domain1', 'data1', 'read'], ['admin', 'domain1', 'data1', 'write'], ['admin', 'domain2', 'data2', 'read'], ['admin', 'domain2', 'data2', 'write']]
After filtering: [['admin', 'domain1', 'data1', 'read'], ['admin', 'domain1', 'data1', 'write']]
Role assignments: [['alice', 'admin', 'domain1']]

[hsluoyz]

@HashCookie add a test case for it

[github-actions]

🎉 This PR is included in version 1.37.0 🎉

The release is available on:

• v1.37.0
• GitHub release

Your semantic-release bot 📦🚀