Tests for AI features — settings UI, copy generator, autofill, analyzer

[anurag629]

Testing for Phases 2-4

Comprehensive tests for all AI feature UI components and integrations.

Scope

Unit Tests

`tests/ai/prompts.test.ts`

• Title generation prompt includes category and tone
• Subtitle prompt includes the title for context
• Prompts respect character limits in instructions
• Prompt builder handles missing/optional fields gracefully

`tests/ai/generate.test.ts`

• `generateTitles()` returns array of strings
• `generateSubtitles()` returns array of strings
• JSON response parsing handles: clean JSON, markdown-wrapped JSON, plain text
• Character limit filtering removes oversized results
• Retry logic triggers on parse failure
• Empty/malformed AI responses return graceful fallback

`tests/api/ai-autofill.test.ts`

• Returns 400 for missing URL
• Returns 400 for invalid URL format
• Returns correct response shape `{ templateId, fields, colors }`
• Handles unreachable URLs gracefully

Component Tests (if React Testing Library is set up)

`tests/components/AISettingsModal.test.tsx`

• Renders all 5 providers
• Provider switch loads saved key for that provider
• Validate button triggers API call
• Shows valid/invalid status after validation
• Clear button removes key (with confirmation)
• Modal closes on ESC and backdrop click

`tests/components/AISuggestionPicker.test.tsx`

• Renders suggestions list
• Click selects a suggestion and calls onSelect
• Regenerate button triggers new generation
• Cancel closes picker
• Keyboard navigation works (arrows, Enter, Escape)

Integration Tests

`tests/ai/integration.test.ts`

• Full flow: set API key → validate → generate title → receive suggestions
• Provider switching preserves keys
• Generate with invalid key → returns auth error, doesn't crash
• All mocked — no real API calls

Mock Strategy

• Mock `fetch` globally for all AI API calls
• Mock `localStorage` for storage tests
• Provider responses should cover: success, auth error, rate limit, malformed response
• Create shared fixtures: `tests/ai/fixtures/` with sample responses per provider

Acceptance Criteria

• All test files created and passing
• Prompt builder tested with various category/context combinations
• Response parsing edge cases covered (at least 5 parse scenarios)
• UI components tested for key interactions
• Integration flow tested end-to-end (mocked)
• No real API calls in any test
• Tests run in CI without environment variables