[kurtseifried]

Agent: Gemini (SecID governing agent)

Summary

Brings SecID-Server-API into full parity with CloudSecurityAlliance/SecID-Service#56 and closes #3:

1. Cross-source, namespace-scoped, and bare-identifier qualifier filtering: Applies _apply_qualifier_filters (?parsability=, ?content_type=, ?lang=) in namespace_scoped_search, type_scoped_search, and _search_bare_identifier, returning _qualifier_not_found or _lang_not_found_message when matches exist for the identifier but none satisfy the requested qualifiers.
2. Exclude URL-less RegistryResult entries when parsability or content_type is requested: Tightens _apply_qualifier_filters so data-only child items without a URL are excluded when format qualifiers are specified.
3. Parser head qualifier stripping: Strips source-level ?qualifiers from head before splitting on / in extract_secid_type and parse_secid, preventing MIME-type slashes (e.g. secid:advisory?content_type=application/json) from being treated as path separators.
4. Top-level HTTP query parameters on GET /api/v1/resolve: Accepts parsability, content_type, and lang query parameters and merges them structurally with inline SecID qualifiers (inline qualifiers take precedence).
5. Regression tests: Added unit/HTTP tests in python/test_smoke.py and real-registry parity tests in python/test_real_registry.py.

Closes #3

[kurtseifried]

Agent: Gemini (SecID governing agent)

Two-Review Summary

1. First review (Author — Gemini): Verified qualifier filtering across namespace_scoped_search, type_scoped_search, and _search_bare_identifier, exclusion of URL-less RegistryResult items when parsability or content_type is specified, head qualifier stripping before splitting on / in parse_secid/extract_secid_type, and top-level parsability/content_type/lang query parameter handling on GET /api/v1/resolve.
2. Second review (Independent fresh-context — OpenAI Codex gpt-6-astra, session 01a1a1c2-63f8-7712-a521-e2faa37aa4ca):
   • [P2] Cross-source searches added +1 default-lang weight in _child_hit: Fixed by keeping the +1 default-language weight bonus in match_children_and_resolve only, matching SecID-Service.
   • [P2] _lang_not_found_message ignored node_matches scope and version_required: Fixed by passing scoped=False and skipping version_required nodes in resolve_with_name and type_scoped_search, and using node_matches(child, subpath, scoped) inside _lang_not_found_message.
   • [P3] _qualifier_not_found unconditionally blamed lang on multi-qualifier misses: Fixed by checking if not multi or qualifiers["lang"] not in available: before appending the lang diagnostic.
   • Added regression assertions in python/test_smoke.py; all 213 tests pass across lazy and bulk load modes.