feat(edit): 编辑上报细节优化

apps/extension/lib/source-parsers.ts

@@ -4,7 +4,7 @@ import {
     isOzonProductPage,
     type OzonPageJsonFetcher,
 } from './ozon-parser.js';
-import { collectPlatformDomAssets, normaliseImageUrl } from './product-media.js';
+import { collectPlatformDomAssets, normaliseImageUrl, normaliseVideoUrl } from './product-media.js';
 import { formatCollectedDescription } from './description-format.js';
 
 type SourcePlatform = '1688' | 'taobao' | 'tmall' | 'ozon';
@@ -49,11 +49,12 @@ export async function collectCurrentProduct(
 
 function collect1688(root: Document, pageUrl: string, bridgedState?: unknown): CollectedProductPayload {
     const state = [bridgedState, ...pageState(root)];
-    const skuInfoMap = findFirst(state, 'skuInfoMap');
+    const skuModel = findFirst(state, 'skuModel');
+    const skuInfoMap = isRecord(skuModel) ? skuModel.skuInfoMap : findFirst(state, 'skuInfoMap');
     const skuMap = isRecord(skuInfoMap) && Object.keys(skuInfoMap).length
         ? skuInfoMap
         : findFirst(state, 'skuMapOriginal');
-    const skuProps = findFirst(state, 'skuProps');
+    const skuProps = isRecord(skuModel) ? skuModel.skuProps : findFirst(state, 'skuProps');
     const axes = parse1688Axes(skuProps);
     const evidence: CollectedProductPayload['evidence'] = [
         {
@@ -78,9 +79,14 @@ function collect1688(root: Document, pageUrl: string, bridgedState?: unknown): C
         });
         for (const [combination, row] of skuRows) {
             if (!isRecord(row)) continue;
-            const sourceSku = stringAt(row, ['skuId', 'sku_id', 'skuCode']);
+            const sku = stringAt(row, ['skuId', 'sku_id', 'skuCode'])?.trim();
+            const sourceSku = sku && !/^(0|null|undefined)$/.test(sku) ? sku : undefined;
             const specAttrs = stringAt(row, ['specAttrs', 'propPath']) ?? combination;
             const selections = selectionsFor1688Combination(specAttrs, axes);
+            // 元数据对象不是 SKU；无来源 ID 时须同时具备完整规格匹配和可售记录证据。
+            const hasSaleEvidence = moneyAt(row, ['price']) !== undefined ||
+                (typeof row.canBookCount === 'number' && Number.isFinite(row.canBookCount) && row.canBookCount >= 0);
+            if (!sourceSku && !(axes.length > 0 && selections.length === axes.length && hasSaleEvidence)) continue;
             // 来源记录本身证明可售组合，不能用规格轴的笛卡尔积扩充 SKU。
             const sourceVariantKey = sourceSku ?? `1688:${combination}`;
             if (!combination || combination === 'undefined') continue;
@@ -239,7 +245,7 @@ function basePayload(
         ref: `${platform}:dom:image-profile`,
         kind: 'dom',
         path: 'platform-scoped gallery / sku / detail selectors',
-        raw_value: `采集 ${assets.length} 张受限商品区域图片`,
+        raw_value: `采集 ${assets.filter((asset) => asset.mediaType === 'image').length} 张受限商品区域图片、${assets.filter((asset) => asset.mediaType === 'video').length} 个画廊视频`,
     });
     const videoCount = addMarketplaceVideos(platform, states, assets, canonical);
     if (videoCount) evidence.push({
@@ -280,8 +286,7 @@ function basePayload(
     };
 }
 
-/** The old collector's proven video sources are structured product state. Do
- * not broaden this to page video elements: they also contain feeds and reviews. */
+/** 结构化画廊数据补充公开视频地址，与受限画廊 DOM 去重，不读取推荐或评价视频。 */
 function addMarketplaceVideos(
     platform: Exclude<SourcePlatform, 'ozon'>,
     states: unknown[],
@@ -306,16 +311,21 @@ function addMarketplaceVideos(
     for (const entry of raw) {
         if (assets.length >= 500) break;
         const record = isRecord(entry) ? entry : undefined;
-        const url = directVideoUrl(
-            typeof entry === 'string' ? entry : stringAt(record ?? {}, ['url', 'videoUrl', 'src']),
-            pageUrl,
-        );
-        if (!url || seen.has(`video:${url}`)) continue;
-        seen.add(`video:${url}`);
+        const candidates = typeof entry === 'string' ? [entry]
+            : ['url', 'videoUrl', 'videoURL', 'videoAddress', 'address', 'src', 'playUrl']
+                .map((key) => stringAt(record ?? {}, [key]));
+        const url = candidates.map((candidate) => directVideoUrl(candidate, pageUrl)).find(Boolean);
+        if (!url) continue;
         const posterUrl = imageUrl(
             stringAt(record ?? {}, ['videoThumbnailURL', 'thumbnailUrl', 'coverUrl', 'coverImage', 'poster']),
             pageUrl,
         );
+        if (seen.has(`video:${url}`)) {
+            const existing = assets.find((asset) => asset.mediaType === 'video' && asset.url === url);
+            if (existing && !existing.posterUrl && posterUrl) existing.posterUrl = posterUrl;
+            continue;
+        }
+        seen.add(`video:${url}`);
         assets.push({
             url,
             mediaType: 'video',
@@ -331,14 +341,7 @@ function addMarketplaceVideos(
 }
 
 function directVideoUrl(value: string | undefined, pageUrl: string) {
-    if (!value) return undefined;
-    try {
-        const url = new URL(value, pageUrl);
-        if (url.protocol !== 'https:' || /\.m3u8(?:$|[?#])/i.test(url.pathname)) return undefined;
-        return /\.(mp4|webm)(?:$|[?#])/i.test(url.pathname) ? url.href : undefined;
-    } catch {
-        return undefined;
-    }
+    return value ? normaliseVideoUrl(value, pageUrl) : undefined;
 }
 
 function imageUrl(value: string | undefined, pageUrl: string) {
@@ -431,6 +434,8 @@ function pageState(root: Document) {
             }
         }
         for (const marker of [
+            'gallery',
+            'skuModel',
             'skuInfoMap',
             'skuMapOriginal',
             'skuBase',
@@ -444,11 +449,16 @@ function pageState(root: Document) {
     return all;
 }
 function objectAfterMarker(text: string, marker: string): unknown | null {
-    const position = text.indexOf(marker);
-    if (position < 0) return null;
-    const start = text.slice(position).search(/[\[{]/);
-    if (start < 0) return null;
-    const first = position + start;
+    // 只读取对象属性的直接字面值，避免从字段名字符串或 null 跳到后续无关对象。
+    const property = new RegExp(`(?:^|[,{])\\s*(?:"${marker}"|'${marker}'|${marker})\\s*:\\s*(?=[\\[{])`, 'g');
+    for (const match of text.matchAll(property)) {
+        const first = match.index! + match[0].length;
+        const parsed = objectLiteralAt(text, first);
+        if (parsed !== null) return parsed;
+    }
+    return null;
+}
+function objectLiteralAt(text: string, first: number): unknown | null {
     const open = text[first];
     const close = open === '{' ? '}' : ']';
     let depth = 0;