feat(edit): 采集插件、编辑页SKU 属性动态展示、颜色展示开发

apps/extension/lib/description-format.ts

@@ -0,0 +1,22 @@
+/**
+ * 将采集到的描述规范为可阅读的纯文本。保留原有段落和列表边界；只有来源
+ * 已把结构压成一行时，才依据常见商品段落标题补出最小必要的换行。
+ */
+export function formatCollectedDescription(value: string): string {
+    const withoutMarkup = value
+        .replace(/\r\n?/g, '\n')
+        .replace(/<\s*br\s*\/?\s*>/gi, '\n')
+        .replace(/<\s*\/\s*(?:p|div|section|article|li|h[1-6])\s*>/gi, '\n')
+        .replace(/<\s*li\b[^>]*>/gi, '\n- ')
+        .replace(/<[^>]*>/g, '');
+    const formatted = withoutMarkup
+        .replace(
+            /\s*(Описание(?:\s+товара)?|Характеристики|Преимущества|Особенности|Комплектация|Технические\s+характеристики|Specifications?|Features?|Package\s+contents?|商品描述|产品描述|商品信息|规格参数|产品参数|产品特点|包装清单)\s*([:：])\s*/gi,
+            '\n\n$1$2\n',
+        )
+        .replace(/([^\n])\s*[-•·]\s+(?=[\p{L}\p{N}])/gu, '$1\n- ')
+        .replace(/[\t\f\v ]*\n[\t\f\v ]*/g, '\n')
+        .replace(/\n{3,}/g, '\n\n')
+        .trim();
+    return formatted;
+}

apps/extension/lib/ozon-parser.ts

@@ -1,5 +1,6 @@
 import type { CollectedProductPayload } from '@atlas/contracts';
 import { dedupeImageUrl, normaliseImageUrl } from './product-media.js';
+import { formatCollectedDescription } from './description-format.js';
 
 export type OzonPageJsonRequest = {
     kind: 'modal-aspects' | 'product' | 'product-detail';
@@ -341,7 +342,7 @@ export async function collectOzonProduct(
             itemKey: currentSku,
             canonicalUrl: canonical,
             collectedAt: new Date().toISOString(),
-            parserVersion: 'ozon-public-2.3.0',
+            parserVersion: 'ozon-public-2.3.1',
         },
         common: {
             ...(title
@@ -650,10 +651,7 @@ export function extractOzonDescription(
         } catch {
             // Some older page responses already contain plain text here.
         }
-        value = value
-            .replace(/\s+\n/g, '\n')
-            .replace(/\n{3,}/g, '\n\n')
-            .trim();
+        value = formatCollectedDescription(value);
         if (!value) continue;
         return {
             value: value.slice(0, 100_000),
@@ -678,7 +676,7 @@ export function extractOzonJsonLdDescription(
                 : '';
             if (!raw) continue;
             return {
-                value: raw.replace(/\s+\n/g, '\n').replace(/\n{3,}/g, '\n\n').slice(0, 100_000),
+                value: formatCollectedDescription(raw).slice(0, 100_000),
                 rawValue: raw.slice(0, 120_000),
             };
         } catch {

apps/extension/lib/source-parsers.ts

@@ -5,6 +5,7 @@ import {
     type OzonPageJsonFetcher,
 } from './ozon-parser.js';
 import { collectPlatformDomAssets } from './product-media.js';
+import { formatCollectedDescription } from './description-format.js';
 
 type SourcePlatform = '1688' | 'taobao' | 'tmall' | 'ozon';
 type Json = Record<string, unknown>;
@@ -112,7 +113,7 @@ function collect1688(root: Document, pageUrl: string, bridgedState?: unknown): C
         axes,
         uniqueVariants(variants),
         evidence,
-        '1688-structured-2.0.0',
+        '1688-structured-2.0.1',
         state,
     );
 }
@@ -187,7 +188,7 @@ function collectTaobao(
         axes,
         uniqueVariants(variants),
         evidence,
-        `${platform}-structured-2.0.0`,
+        `${platform}-structured-2.0.1`,
         state,
     );
 }
@@ -357,7 +358,11 @@ function extractMarketplaceFacts(root: Document, states: unknown[], platform: Ex
     return {
         title: (isRecord(item) ? stringAt(item, ['title', 'name']) : undefined) ?? (isRecord(temp) ? stringAt(temp, ['offerTitle']) : undefined) ?? (product ? stringAt(product, ['name']) : undefined),
         price,
-        description: (product ? stringAt(product, ['description']) : undefined) ?? scopedDescription(root, platform),
+        description: formatCollectedDescription(
+            (product ? stringAt(product, ['description']) : undefined) ??
+                scopedDescription(root, platform) ??
+                '',
+        ) || undefined,
         attributes,
         categories: [...new Set(categoryNames)].map((name) => ({ key: `${platform}:category:${name.toLowerCase()}`.slice(0, 300), name })),
     };
@@ -368,7 +373,10 @@ function scopedDescription(root: Document, platform: Exclude<SourcePlatform, 'oz
         ? ['.de-description-detail', '#detailContentContainer', '.html-description']
         : ['[class*="tabDetailWrap--"]', '[class*="detailInfo--"]'];
     for (const selector of selectors) {
-        const value = root.querySelector(selector)?.textContent?.replace(/\s+/g, ' ').trim();
+        const element = root.querySelector<HTMLElement>(selector);
+        const value = formatCollectedDescription(
+            element?.innerText ?? element?.textContent ?? '',
+        );
         if (value && value.length >= 10) return value.slice(0, 100_000);
     }
     return undefined;

apps/extension/test/product-media.test.mts

@@ -9,6 +9,7 @@ import {
     extractOzonGalleryImageUrls,
     extractOzonGalleryVideos,
 } from '../lib/ozon-parser.ts';
+import { formatCollectedDescription } from '../lib/description-format.ts';
 import { dedupeImageUrl, normaliseImageUrl } from '../lib/product-media.ts';
 
 test('Ozon gallery fixture only returns product-gallery images', async () => {
@@ -93,6 +94,15 @@ test('Ozon JSON-LD Product description is the fallback when rich content is abse
     );
 });
 
+test('采集描述保留段落和列表，并为压平的 Ozon 结构标题补出换行', () => {
+    assert.equal(
+        formatCollectedDescription(
+            'Описание товара:适合户外使用。Характеристики:- 容量：25 L- 材质：尼龙Преимущества:- 防水\n\nКомплектация:<br>背包',
+        ),
+        'Описание товара:\n适合户外使用。\n\nХарактеристики:\n- 容量：25 L\n- 材质：尼龙\n\nПреимущества:\n- 防水\n\nКомплектация:\n背包',
+    );
+});
+
 test('Ozon full characteristics merge with the five short header characteristics', () => {
     const attributes = extractOzonCharacteristics([
         {