The scan terminal had grown into overlapping CSS, request-state, AI-provider, and city-card data responsibilities. This refactor separates those boundaries without changing product behavior: CSS modules are split by surface, city AI prompt/provider/fallback logic is isolated, and scan terminal request state now has reusable RemoteData adapters plus business-state tests.
Constraint: Preserve existing global scan-terminal class names and API responses during the refactor
Constraint: No new dependencies; keep this as a file-boundary cleanup
Rejected: Introduce React Query now | higher migration risk than the requested lightweight query-client path
Rejected: Rewrite AI stream behavior | progressive/fallback states are product-sensitive and were only adapter-split
Confidence: high
Scope-risk: moderate
Reversibility: clean
Directive: Keep AI stream state changes covered by business snapshots before changing fallback/cache wording
Tested: npm run test:business; npx tsc --noEmit; npm run build; python pytest -q; ruff check; py_compile targeted city AI modules
Not-tested: Live DeepSeek provider network replay and browser visual QA
Market scan state was still tracked as separate nullable payload and status strings. This moves the hook internals to RemoteData<MarketScan> so loading and error paths can preserve previous data while keeping the existing marketScan and marketStatus return values for current card consumers.
Constraint: Preserve the existing useCityMarketScan public compatibility fields.
Rejected: Update all card UI to consume marketRemote immediately | keeping the compatibility bridge avoids a broad rendering diff while the request state model lands.
Confidence: high
Scope-risk: narrow
Reversibility: clean
Tested: npm run build
Not-tested: Forced live market-scan failure with stale previous quote.
The city-card data hook still contained localStorage serialization and TTL eviction helpers alongside AI fallback and UI state transitions. Moving those helpers into scan-terminal-cache creates a reusable cache boundary for the scan terminal request layer without changing cache keys, TTLs, or payload shapes.
Constraint: Preserve existing localStorage keys and expiry behavior.
Rejected: Migrate all caches to RemoteData in this commit | cache-helper extraction is a safer intermediate step before query policy changes.
Confidence: high
Scope-risk: narrow
Reversibility: clean
Tested: npm run build
Not-tested: Browser private-mode quota edge cases beyond existing guarded behavior.
The AI city forecast hook still owned stream queueing and in-flight request dedupe after the first request-client pass. Moving that policy into scanTerminalClient keeps network concurrency, queued progress, and requestKey reuse in the request layer while leaving the hook responsible only for cached UI state and progress rendering.
Constraint: Preserve existing two-stream concurrency limit and queued user-facing progress copy.
Rejected: Move localStorage cache at the same time | cache policy should be separated from stream transport policy to keep this refactor reviewable.
Confidence: high
Scope-risk: narrow
Reversibility: clean
Tested: npm run build
Not-tested: Live multi-city SSE under production latency.
The scan terminal hooks were each carrying their own fetch, SSE parsing, error normalization, previous-data handling, and market request behavior. This adds a small scanTerminalClient plus RemoteData helpers so terminal data, city detail, market scans, and AI city streams share one request boundary without introducing React Query.
Constraint: Do not add dependencies or change backend API contracts.
Rejected: Introduce React Query immediately | too broad for this release and would force larger UI state rewrites.
Rejected: Move localStorage caches in the same pass | safer to first isolate network and stream IO before cache policy migration.
Confidence: high
Scope-risk: moderate
Reversibility: clean
Tested: npm run build
Not-tested: Live SSE cancellation against production latency.
City-card fallback reads now stop using stale METAR or official observations as strong live anchors. A stale observation no longer forces high/low revisions, and both backend and browser AI cache keys include the observation fingerprint so updated report times, receipt times, temperatures, or stale status invalidate old AI text.
Constraint: Cached city AI reads must not survive a material observation update
Rejected: Let stale METAR trigger observed-break revisions | stale reports can be older than the active temperature path
Confidence: high
Scope-risk: moderate
Tested: pytest tests/test_web_observability.py -q
Tested: npm run build
City AI can return a partially streamed JSON string when the provider truncates output. The fallback previously kept an unfinished clause such as '但TAF显示', which made the forecast explanation look broken even though earlier evidence was usable.
Constraint: Provider JSON can be truncated after useful fields have already streamed
Rejected: Drop all partial AI text | would lose valid METAR interpretation already returned before truncation
Confidence: high
Scope-risk: narrow
Tested: pytest city AI truncation regression tests
Tested: npm run build
Not-tested: Live DeepSeek provider response
2569718930@qq.com