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/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;