task-116(任务与重复分析中心): 实现重复检查分布视图

新增 duplicate-distribution.ts：按国家/站点聚合撞款分布与撞款 ASIN top 店铺。 TDD: task-116.test.ts 8 用例先 RED 后 GREEN。
2026-09-05 17:24:34 +08:00
parent 61b2d7152d
commit b5a1a51403
2 changed files with 115 additions and 0 deletions
@@ -0,0 +1,46 @@
+/** 重复检查分布视图（任务 116）：按国家/站点聚合撞款与 top 店铺分布；纯逻辑。 */
+import type { DuplicateShop } from './duplicate-model.ts'
+
+export interface CountryDistributionRow {
+  country: string
+  asinCount: number
+  shopCount: number
+}
+
+/** 按国家聚合分布(每店铺 country_codes 全量计入)。 */
+export function countryDistribution(shops: readonly DuplicateShop[]): CountryDistributionRow[] {
+  const byCountry = new Map<string, { asinCount: number; shopCount: number }>()
+  for (const item of shops) {
+    for (const code of item.countryCodes) {
+      const country = code.trim().toUpperCase()
+      if (!country) continue
+      const entry = byCountry.get(country) ?? { asinCount: 0, shopCount: 0 }
+      entry.asinCount += item.asinCount
+      entry.shopCount += 1
+      byCountry.set(country, entry)
+    }
+  }
+  return [...byCountry.entries()]
+    .map(([country, entry]) => ({ country, asinCount: entry.asinCount, shopCount: entry.shopCount }))
+    .sort((a, b) => b.asinCount - a.asinCount || a.country.localeCompare(b.country))
+}
+
+export interface SiteDistributionRow {
+  site: string
+  asinCount: number
+  shopCount: number
+}
+
+/** 按站点(site)聚合：以单国家维度汇总撞款，站点取国家码。 */
+export function duplicateSiteDistribution(shops: readonly DuplicateShop[]): SiteDistributionRow[] {
+  const rows = countryDistribution(shops)
+  return rows.map((row) => ({ site: row.country, asinCount: row.asinCount, shopCount: row.shopCount }))
+}
+
+/** 按撞款 ASIN 数取 top 店铺(降序)，limit<=0 返回空。 */
+export function topShopsByAsin(shops: readonly DuplicateShop[], limit: number): DuplicateShop[] {
+  if (!(limit > 0)) return []
+  return [...shops]
+    .sort((a, b) => b.asinCount - a.asinCount || a.shopName.localeCompare(b.shopName))
+    .slice(0, Math.floor(limit))
+}