feat(ticketing): unified search (search/count/export)

[lmeilibr]

Summary

• Adds SearchApiClient and SearchService covering /api/v2/search, /api/v2/search/count and cursor-paginated /api/v2/search/export (with filter[type] and page[size]).
• Wires ticketing.search into the Ticketing facade.
• Results are returned as raw dicts because unified search is intentionally heterogeneous (tickets/users/orgs/groups).

Refs #79.

Test plan

• uv run --python 3.12 pytest tests/unit/ — 1868 passed
• 100% coverage across search_api_client / search_service
• Live integration tests on a real tenant

🤖 Generated with Claude Code

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
97.2% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[serenakeyitan]

Review

LGTM — clean, consistent addition that mirrors the existing api_client / service split.

What I checked

• SearchApiClient follows the same yield_items cursor-pagination pattern as MacroApiClient.search and the rest of the ticketing clients.
• SearchService is a thin pass-through, which is appropriate for unified search where results are intentionally heterogeneous (raw dicts).
• Facade wiring in application/services/ticketing/__init__.py matches the convention used for siblings.
• Query-string assembly via urlencode correctly handles filter[type] / page[size] bracket encoding (verified by the unit assertions).
• Unit tests cover param presence/absence, count default, export filter, page_size, and error propagation. Integration tests are thin smoke checks, which is the established style here.

Nits (non-blocking)

• Worth a one-line docstring note on search() that Zendesk's unified search caps results at ~1000 / 11 pages (already called out in pagination.py); callers iterating without bounds may otherwise be surprised.
• count() does int(data.get("count", 0)) — if the API ever returns count: null this will raise TypeError. int(data.get("count") or 0) would be marginally more defensive, but current behavior is fine for now.

No required changes. Approving.

This reply was drafted by breeze, an autonomous agent running on behalf of the account owner.