refactor(di): decompose Ctrl god object via constructor injection (#208)

[pajoma]

Summary

• Replace constructor(public ctrl: JournalController) with narrow sub-interface params in all seven action/UI modules (Inject, Parser, Writer, Reader, Dialogues, ConsoleLogger, ScanEntries)
• Ctrl becomes a two-phase object: constructor creates Configuration only; initServices(logger: ILogger) creates all services in dependency order
• 14 test files updated from ctrl.logger = logger to ctrl.initServices(logger); new J.Actions.Parser(ctrl) calls replaced by ctrl.parser

Test plan

• npm run compile — zero errors
• npm run lint — zero new warnings
• npm test — 126 tests pass (verified locally)
• grep -rn "this\.ctrl" src/actions/ src/ext/dialogues.ts src/util/logger.ts src/provider/features/scan-entries.ts — no output (verified)

Related

• Implements approved plan: docs/plans/2026-05-17-208-constructor-injection.md
• Spec: docs/specs/2026-05-17-208-constructor-injection.md
• Closes Arch: Decompose God Object Controller into Focused Dependencies #208
• Builds on Arch: Break ubiquitous circular coupling via Interface Extraction (Namespace J) #207 (merged as refactor(arch): introduce JournalController interface, break barrel cycle #213)

🤖 Generated with Claude Code

[pajoma]

Architecture Review Approval (PR #214)

I have reviewed the implementation of PR #214 and it is perfectly aligned with the approved plan.

Key Achievements:

• Dismantled God Object: The transition from ctrl: JournalController to focused constructor parameters across all seven action/UI classes is a massive win for architectural clarity.
• Improved Initialization: The two-phase initialization (initServices) correctly manages our dependency graph while preventing the circular instantiation risks we identified in the audit.
• Cleaner Tests: The migration of the test suite not only fixes the initialization pattern but also simplifies tests by utilizing the controller's internal service registry.
• SRP Integrity: Pushing ConsoleLogger construction into the Startup layer successfully keeps VS Code UI concerns out of our central controller.

This refactoring provides the definitive foundation for independent unit testing and long-term maintainability.

---

Note: Gemini was involved in creating this comment (but not without consulting me).