Update seller store experience

2026-09-08 16:19:38 +08:00
parent ceb644a3f9
commit 319887726e
25 changed files with 31417 additions and 127 deletions
@@ -8,6 +8,7 @@ declare module "wxt/browser" {
    | "/background.js"
    | "/content-scripts/collector.js"
    | "/content-scripts/ozon-page-json.js"
+    | "/content-scripts/source-page-state.js"
  type HtmlPublicPath = Extract<PublicPath, `${string}.html`>
  export interface WxtRuntime {
    getURL(path: PublicPath): string;
@@ -3,6 +3,7 @@ import {
    supportedProductPlatform,
 } from '../lib/source-parsers.js';
 import type { OzonPageJsonFetcher } from '../lib/ozon-parser.js';
+import type { MarketplacePageStateFetcher } from '../lib/source-parsers.js';

 export default defineContentScript({
    matches: [
@@ -72,6 +73,9 @@ export default defineContentScript({
                    supportedProductPlatform() === 'ozon'
                        ? requestOzonPageJson
                        : undefined,
+                    supportedProductPlatform() && supportedProductPlatform() !== 'ozon'
+                        ? requestMarketplacePageState
+                        : undefined,
                );
                const response = await browser.runtime.sendMessage({
                    type: 'atlas.collect',
@@ -128,6 +132,23 @@ const requestOzonPageJson: OzonPageJsonFetcher = ({ kind, sku }) =>
        );
    });

+const requestMarketplacePageState: MarketplacePageStateFetcher = (platform) =>
+    new Promise((resolve) => {
+        const requestId = crypto.randomUUID();
+        const timeout = window.setTimeout(() => {
+            window.removeEventListener('message', receive);
+            resolve({});
+        }, 3_000);
+        const receive = (event: MessageEvent<unknown>) => {
+            if (event.source !== window || !isMarketplaceStateResponse(event.data, requestId)) return;
+            window.clearTimeout(timeout);
+            window.removeEventListener('message', receive);
+            resolve(event.data.data);
+        };
+        window.addEventListener('message', receive);
+        window.postMessage({ source: 'atlas-collector', type: 'source-page-state.request', requestId, platform }, location.origin);
+    });
+
 function isBridgeResponse(
    value: unknown,
    requestId: string,
@@ -149,3 +170,10 @@ function isBridgeResponse(
        (message.ok || typeof message.error === 'string')
    );
 }
+
+function isMarketplaceStateResponse(value: unknown, requestId: string): value is { source: 'atlas-collector'; type: 'source-page-state.response'; requestId: string; data: unknown } {
+    if (!value || typeof value !== 'object' || Array.isArray(value)) return false;
+    const message = value as Record<string, unknown>;
+    return message.source === 'atlas-collector' && message.type === 'source-page-state.response'
+        && message.requestId === requestId && !!message.data && typeof message.data === 'object';
+}
@@ -0,0 +1,69 @@
+/**
+ * A deliberately narrow MAIN-world bridge for the mature 1688/Taobao/Tmall
+ * structured collectors. It never exposes cookies, headers, arbitrary window
+ * keys, or a generic fetch primitive; only public product-state projections.
+ */
+export default defineContentScript({
+    matches: [
+        '*://detail.1688.com/*',
+        '*://item.taobao.com/*',
+        '*://detail.tmall.com/*',
+    ],
+    world: 'MAIN',
+    runAt: 'document_idle',
+    main() {
+        window.addEventListener('message', (event) => {
+            if (event.source !== window || !isRequest(event.data)) return;
+            const data = event.data.platform === '1688'
+                ? project1688((window as unknown as Record<string, unknown>).context)
+                : projectTaobao(window as unknown as Record<string, unknown>);
+            window.postMessage({
+                source: 'atlas-collector',
+                type: 'source-page-state.response',
+                requestId: event.data.requestId,
+                data,
+            }, location.origin);
+        });
+    },
+});
+
+type Request = { source: 'atlas-collector'; type: 'source-page-state.request'; requestId: string; platform: '1688' | 'taobao' | 'tmall' };
+function isRequest(value: unknown): value is Request {
+    if (!value || typeof value !== 'object' || Array.isArray(value)) return false;
+    const data = value as Record<string, unknown>;
+    return data.source === 'atlas-collector' && data.type === 'source-page-state.request'
+        && typeof data.requestId === 'string' && /^[a-zA-Z0-9-]{16,120}$/.test(data.requestId)
+        && (data.platform === '1688' || data.platform === 'taobao' || data.platform === 'tmall');
+}
+
+const safeClone = (value: unknown) => { try { return JSON.parse(JSON.stringify(value)); } catch { return undefined; } };
+function find(value: unknown, key: string): unknown {
+    const pending: unknown[] = [value];
+    const seen = new Set<object>();
+    let visited = 0;
+    while (pending.length && visited < 300_000) {
+        const current = pending.shift();
+        if (!current || typeof current !== 'object' || seen.has(current)) continue;
+        seen.add(current);
+        visited += 1;
+        const record = current as Record<string, unknown>;
+        if (key in record) return record[key];
+        pending.push(...Object.values(record));
+    }
+    return undefined;
+}
+function project1688(context: unknown) {
+    const projected: Record<string, unknown> = {};
+    for (const key of ['gallery', 'tempModel', 'skuModel', 'tradeModel', 'skuMapOriginal', 'pieceWeightScaleInfo', 'productAttributes']) {
+        const value = find(context, key);
+        if (value !== undefined) projected[key] = safeClone(value);
+    }
+    return projected;
+}
+function projectTaobao(windowState: Record<string, unknown>) {
+    const ice = windowState.__ICE_APP_CONTEXT__ as Record<string, unknown> | undefined;
+    const result = ice?.loaderData && typeof ice.loaderData === 'object'
+        ? (((ice.loaderData as Record<string, unknown>).home as Record<string, unknown> | undefined)?.data as Record<string, unknown> | undefined)?.res
+        : undefined;
+    return { product: safeClone(result) };
+}
@@ -61,6 +61,7 @@ export async function collectOzonProduct(
    const localStates = readLocalWidgetStates(documentRoot);
    let attributes = extractOzonCharacteristics(localStates);
    let description = extractOzonDescription(localStates);
+    const jsonLdDescription = extractOzonJsonLdDescription(documentRoot);
    const heading = firstState(localStates, 'webProductHeading');
    const priceState = firstState(localStates, 'webPrice');
    const title =
@@ -101,6 +102,14 @@ export async function collectOzonProduct(
        'ozon:gallery:page',
        canonical,
    );
+    addGalleryVideos(
+        localStates,
+        currentSku,
+        assets,
+        assetKeys,
+        'ozon:gallery:page',
+        canonical,
+    );
    if (attributes.length) {
        evidence.push({
            ref: 'ozon:characteristics:page',
@@ -109,6 +118,14 @@ export async function collectOzonProduct(
            raw_value: `发现 ${attributes.length} 项商品特征`,
        });
    }
+    if (jsonLdDescription) {
+        evidence.push({
+            ref: 'ozon:json-ld:product.description',
+            kind: 'json_ld',
+            path: 'script[type="application/ld+json"] Product.description',
+            raw_value: `JSON-LD 商品描述 ${jsonLdDescription.value.length} 个字符`,
+        });
+    }

    const categories = collectCategories(documentRoot, canonical);
    if (categories.length) {
@@ -162,12 +179,22 @@ export async function collectOzonProduct(
                canonical,
            );
            description = description ?? extractOzonDescription(detailStates);
+            const detailAttributes = extractOzonCharacteristics(detailStates);
+            attributes = mergeOzonCharacteristics(attributes, detailAttributes);
            evidence.push({
                ref: 'ozon:detail:page',
                kind: 'state',
                path: 'webDescription|webRichContent|webFeatures image fields',
                raw_value: `发现 ${added} 张商品详情图`,
            });
+            if (detailAttributes.length) {
+                evidence.push({
+                    ref: 'ozon:characteristics:detail',
+                    kind: 'state',
+                    path: 'webShortCharacteristics|webDetailedCharacteristics|webCharacteristics.characteristics[]',
+                    raw_value: `详情段补全 ${detailAttributes.length} 项商品特征`,
+                });
+            }
        } catch (error) {
            evidence.push(
                fetchFailureEvidence('ozon:detail:page', currentSku, error),
@@ -230,6 +257,14 @@ export async function collectOzonProduct(
                    `ozon:gallery:${result.sku}`,
                    canonical,
                );
+                addGalleryVideos(
+                    result.states,
+                    result.sku,
+                    assets,
+                    assetKeys,
+                    `ozon:gallery:${result.sku}`,
+                    canonical,
+                );
                if (result.sku === currentSku) {
                    const detailedAttributes = extractOzonCharacteristics(
                        result.states,
@@ -270,6 +305,11 @@ export async function collectOzonProduct(
        });
    }

+    // Ozon's rich widget is preferred because it is the dedicated description
+    // segment. JSON-LD is the proven old-collector fallback for pages where
+    // Ozon only places the seller description in the initial document.
+    description = description ?? jsonLdDescription;
+
    addAspectCoverAssets(axes, assets, assetKeys, canonical);

    const variants = [...candidates.values()]
@@ -301,7 +341,7 @@ export async function collectOzonProduct(
            itemKey: currentSku,
            canonicalUrl: canonical,
            collectedAt: new Date().toISOString(),
-            parserVersion: 'ozon-public-2.2.0',
+            parserVersion: 'ozon-public-2.3.0',
        },
        common: {
            ...(title
@@ -452,6 +492,7 @@ function addAspectCoverAssets(
            assetKeys.add(key);
            assets.push({
                url,
+                mediaType: 'image',
                role: 'source_sku',
                relation: 'axis_value',
                sortOrder: assets.length,
@@ -491,6 +532,7 @@ function addDetailAssets(
        assetKeys.add(key);
        assets.push({
            url,
+            mediaType: 'image',
            role: 'source_detail',
            relation: 'product_candidate',
            sortOrder: added++,
@@ -621,6 +663,55 @@ export function extractOzonDescription(
    return undefined;
 }

+/** Reads only schema.org Product.description as a fallback, never arbitrary
+ * JSON-LD page copy. This mirrors the mature collector's JSON-LD path. */
+export function extractOzonJsonLdDescription(
+    root: Document,
+): OzonDescription | undefined {
+    for (const script of Array.from(root.scripts)) {
+        if (script.type !== 'application/ld+json' || !script.textContent?.trim())
+            continue;
+        try {
+            const product = findJsonLdProduct(JSON.parse(script.textContent));
+            const raw = product && typeof product.description === 'string'
+                ? product.description.trim()
+                : '';
+            if (!raw) continue;
+            return {
+                value: raw.replace(/\s+\n/g, '\n').replace(/\n{3,}/g, '\n\n').slice(0, 100_000),
+                rawValue: raw.slice(0, 120_000),
+            };
+        } catch {
+            // A malformed analytics JSON-LD block is not product evidence.
+        }
+    }
+    return undefined;
+}
+
+function findJsonLdProduct(
+    value: unknown,
+    visited = new Set<unknown>(),
+): JsonRecord | undefined {
+    if (!value || typeof value !== 'object' || visited.has(value)) return undefined;
+    visited.add(value);
+    if (Array.isArray(value)) {
+        for (const child of value) {
+            const product = findJsonLdProduct(child, visited);
+            if (product) return product;
+        }
+        return undefined;
+    }
+    const record = value as JsonRecord;
+    const type = record['@type'];
+    if (type === 'Product' || (Array.isArray(type) && type.includes('Product')))
+        return record;
+    if (Array.isArray(record['@graph'])) {
+        const product = findJsonLdProduct(record['@graph'], visited);
+        if (product) return product;
+    }
+    return undefined;
+}
+
 function findRichAnnotationJson(
    value: unknown,
    visited = new Set<unknown>(),
@@ -716,6 +807,7 @@ function addGalleryAssets(
        assetKeys.add(key);
        assets.push({
            url,
+            mediaType: 'image',
            // A gallery is product photography for this real SKU.  It is not
            // a SKU-option thumbnail merely because it is the second image.
            role: 'source_main',
@@ -727,6 +819,103 @@ function addGalleryAssets(
    }
 }

+/** Adds only Ozon's named product-gallery videos. Page-wide <video> nodes can
+ * belong to reviews or recommendations and are intentionally never considered. */
+function addGalleryVideos(
+    states: JsonRecord[],
+    sku: string,
+    assets: Asset[],
+    assetKeys: Set<string>,
+    evidenceRef: string,
+    pageUrl: string,
+) {
+    const galleries = states.filter((state) => {
+        const widget = String(state.__atlasWidgetName ?? '').replace(/^state-/, '');
+        return /^web(?:Gallery|Video)-/.test(widget);
+    });
+    for (const gallery of galleries) {
+        for (const video of extractOzonGalleryVideos(gallery, pageUrl)) {
+            if (assets.length >= 500) return;
+            const key = `video:${sku}:${video.url}`;
+            if (assetKeys.has(key)) continue;
+            assetKeys.add(key);
+            assets.push({
+                url: video.url,
+                mediaType: 'video',
+                ...(video.posterUrl ? { posterUrl: video.posterUrl } : {}),
+                role: 'source_main',
+                variantSourceKey: sku,
+                relation: 'variant_direct',
+                sortOrder: assets.length,
+                evidenceRef,
+            });
+        }
+    }
+}
+
+/** Pure helper for the strictly named Ozon gallery/video state. HLS, blobs,
+ * and non-HTTPS URLs stay out of the collection contract in this first phase. */
+export function extractOzonGalleryVideos(gallery: JsonRecord, pageUrl: string) {
+    const videos: Array<{ url: string; posterUrl?: string }> = [];
+    const seen = new Set<string>();
+    // Ozon changes whether the direct file lives in videos[], video, or a
+    // nested gallery payload. The old collector walked this *one allowlisted
+    // widget* for direct video extensions; retaining that scope is critical.
+    for (const { value, parent } of nestedVideoValues(gallery)) {
+        const url = directVideoUrl(
+            typeof value === 'string'
+                ? value
+                : stringAt(value, ['url', 'videoUrl', 'src']),
+            pageUrl,
+        );
+        if (!url) continue;
+        const posterUrl = ozonImageUrl(
+            (typeof value === 'string'
+                ? undefined
+                : stringAt(value, ['coverImage', 'coverUrl', 'poster', 'previewImage']))
+                ?? stringAt(parent, ['coverImage', 'coverUrl', 'poster', 'previewImage']),
+            pageUrl,
+        );
+        if (seen.has(url)) {
+            const existing = videos.find((video) => video.url === url);
+            if (existing && !existing.posterUrl && posterUrl)
+                existing.posterUrl = posterUrl;
+            continue;
+        }
+        seen.add(url);
+        videos.push({ url, ...(posterUrl ? { posterUrl } : {}) });
+    }
+    return videos;
+}
+
+function nestedVideoValues(root: JsonRecord) {
+    const values: Array<{ value: string | JsonRecord; parent?: JsonRecord }> = [];
+    const pending: Array<{ value: unknown; parent?: JsonRecord }> = [{ value: root }];
+    const seen = new Set<object>();
+    while (pending.length) {
+        const current = pending.pop()!;
+        if (typeof current.value === 'string') {
+            if (/\.(mp4|webm)(?:$|[?#])/i.test(current.value))
+                values.push({ value: current.value, parent: current.parent });
+            continue;
+        }
+        if (!current.value || typeof current.value !== 'object') continue;
+        if (seen.has(current.value)) continue;
+        seen.add(current.value);
+        if (Array.isArray(current.value)) {
+            for (const child of current.value)
+                pending.push({ value: child, parent: current.parent });
+            continue;
+        }
+        const record = current.value as JsonRecord;
+        if (stringAt(record, ['url', 'videoUrl', 'src']))
+            values.push({ value: record, parent: current.parent });
+        for (const child of Object.values(record))
+            pending.push({ value: child, parent: record });
+    }
+    return values;
+}
+
 /** Pure helper used by fixture tests and the structured Ozon gallery reader. */
 export function extractOzonGalleryImageUrls(
    gallery: JsonRecord,
@@ -777,7 +966,7 @@ function widgetStatesFromPageJson(value: unknown): JsonRecord[] {
    const states: JsonRecord[] = [];
    for (const [widgetName, raw] of Object.entries(value.widgetStates)) {
        if (
-            !/^web(Aspects|AspectsModal|ProductHeading|Price|Gallery|Description|RichContent|Features|ShortCharacteristics|DetailedCharacteristics|Characteristics)-/.test(
+            !/^web(Aspects|AspectsModal|ProductHeading|Price|Gallery|Video|Description|RichContent|Features|ShortCharacteristics|DetailedCharacteristics|Characteristics)-/.test(
                widgetName,
            )
        )
@@ -997,3 +1186,15 @@ function ozonImageUrl(value: string | undefined, pageUrl: string) {
    const url = normaliseImageUrl(value, pageUrl, 'ozon');
    return url?.startsWith('https://') ? url : undefined;
 }
+
+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;
+    }
+}
@@ -180,6 +180,7 @@ export function collectPlatformDomAssets(
                seen.add(key);
                assets.push({
                    url,
+                    mediaType: 'image',
                    role: group.role,
                    relation: group.relation,
                    sortOrder: assets.length,
@@ -22,22 +22,32 @@ export function supportedProductPlatform(
    return null;
 }

+/** Returns only the public, allowlisted product-state projection supplied by
+ * the platform-specific MAIN-world bridge. */
+export type MarketplacePageStateFetcher = (
+    platform: Exclude<SourcePlatform, 'ozon'>,
+) => Promise<unknown>;
+
 export async function collectCurrentProduct(
    root: Document = document,
    pageUrl = location.href,
    fetchOzonPageJson?: OzonPageJsonFetcher,
+    fetchMarketplacePageState?: MarketplacePageStateFetcher,
 ): Promise<CollectedProductPayload> {
    const platform = supportedProductPlatform(pageUrl);
    if (platform === 'ozon')
        return collectOzonProduct(root, pageUrl, fetchOzonPageJson);
    if (!platform) throw new Error('当前页面不是支持的商品详情页');
+    const pageState = fetchMarketplacePageState
+        ? await fetchMarketplacePageState(platform)
+        : undefined;
    return platform === '1688'
-        ? collect1688(root, pageUrl)
-        : collectTaobao(root, pageUrl, platform);
+        ? collect1688(root, pageUrl, pageState)
+        : collectTaobao(root, pageUrl, platform, pageState);
 }

-function collect1688(root: Document, pageUrl: string): CollectedProductPayload {
-    const state = pageState(root);
+function collect1688(root: Document, pageUrl: string, bridgedState?: unknown): CollectedProductPayload {
+    const state = [bridgedState, ...pageState(root)];
    const skuMap = findFirst(state, 'skuMapOriginal');
    const skuProps = findFirst(state, 'skuProps');
    const axes = parse1688Axes(skuProps);
@@ -50,14 +60,19 @@ function collect1688(root: Document, pageUrl: string): CollectedProductPayload {
        },
    ];
    const variants: Variant[] = [];
-    if (isRecord(skuMap)) {
+    const skuRows = Array.isArray(skuMap)
+        ? skuMap.map((row) => [stringAt(isRecord(row) ? row : {}, ['specAttrs', 'propPath', 'skuId']) ?? '', row] as const)
+        : isRecord(skuMap)
+          ? Object.entries(skuMap)
+          : [];
+    if (skuRows.length) {
        evidence.push({
            ref: '1688:sku-map',
            kind: 'state',
            path: 'skuMapOriginal',
-            raw_value: `组合记录 ${Object.keys(skuMap).length} 条`,
+            raw_value: `组合记录 ${skuRows.length} 条`,
        });
-        for (const [combination, row] of Object.entries(skuMap)) {
+        for (const [combination, row] of skuRows) {
            if (!isRecord(row)) continue;
            const sourceSku = stringAt(row, ['skuId', 'sku_id', 'skuCode']);
            // A skuMapOriginal entry itself is a source-provided sellable-combination
@@ -97,7 +112,8 @@ function collect1688(root: Document, pageUrl: string): CollectedProductPayload {
        axes,
        uniqueVariants(variants),
        evidence,
-        '1688-state-1.0.0',
+        '1688-structured-2.0.0',
+        state,
    );
 }

@@ -105,8 +121,9 @@ function collectTaobao(
    root: Document,
    pageUrl: string,
    platform: 'taobao' | 'tmall',
+    bridgedState?: unknown,
 ): CollectedProductPayload {
-    const state = pageState(root);
+    const state = [bridgedState, ...pageState(root)];
    const skuBase = findFirst(state, 'skuBase');
    const skuCore = findFirst(state, 'skuCore');
    const axes = parseTaobaoAxes(isRecord(skuBase) ? skuBase.props : undefined);
@@ -170,7 +187,8 @@ function collectTaobao(
        axes,
        uniqueVariants(variants),
        evidence,
-        `${platform}-state-1.0.0`,
+        `${platform}-structured-2.0.0`,
+        state,
    );
 }

@@ -183,15 +201,18 @@ function basePayload(
    variants: Variant[],
    evidence: CollectedProductPayload['evidence'],
    parserVersion: string,
+    states: unknown[],
 ): CollectedProductPayload {
    const canonical =
        root.querySelector<HTMLLinkElement>('link[rel="canonical"]')?.href ??
        pageUrl.split('#')[0];
-    const title =
+    const facts = extractMarketplaceFacts(root, states, platform);
+    const title = facts.title ?? (
        root.querySelector('h1')?.textContent?.replace(/\s+/g, ' ').trim() ||
        root
            .querySelector<HTMLMetaElement>('meta[property="og:title"]')
-            ?.content?.trim();
+            ?.content?.trim()
+    );
    const assets = collectPlatformDomAssets(platform, root, canonical);
    evidence.push({
        ref: `${platform}:dom:image-profile`,
@@ -199,6 +220,15 @@ function basePayload(
        path: 'platform-scoped gallery / sku / detail selectors',
        raw_value: `采集 ${assets.length} 张受限商品区域图片`,
    });
+    const videoCount = addMarketplaceVideos(platform, states, assets, canonical);
+    if (videoCount) evidence.push({
+        ref: `${platform}:structured:videos`,
+        kind: 'state',
+        path: platform === '1688' ? 'gallery.fields.video|videos' : 'item.videos',
+        raw_value: `采集 ${videoCount} 个结构化商品视频`,
+    });
+    if (facts.attributes.length) evidence.push({ ref: `${platform}:structured:attributes`, kind: 'state', path: 'allowlisted product state parameter lists', raw_value: `采集 ${facts.attributes.length} 项结构化特征` });
+    if (facts.description) evidence.push({ ref: `${platform}:json-ld:description`, kind: 'json_ld', path: 'Product.description', raw_value: `采集 ${facts.description.length} 个字符描述` });
    return {
        source: {
            platform,
@@ -217,8 +247,10 @@ function basePayload(
                      },
                  }
                : {}),
-            attributes: [],
-            categories: [],
+            ...(facts.price ? { sourcePrice: { amount: facts.price, currencyCode: 'CNY', raw_value: facts.price } } : {}),
+            ...(facts.description ? { description: { value: facts.description, raw_value: facts.description, confidence: 'verified' as const } } : {}),
+            attributes: facts.attributes,
+            categories: facts.categories,
        },
        variantAxes: axes,
        variants,
@@ -227,6 +259,137 @@ 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. */
+function addMarketplaceVideos(
+    platform: Exclude<SourcePlatform, 'ozon'>,
+    states: unknown[],
+    assets: CollectedProductPayload['assets'],
+    pageUrl: string,
+) {
+    const gallery = findFirst(states, 'gallery');
+    const item = findFirst(states, 'item');
+    const galleryFields = isRecord(gallery) && isRecord(gallery.fields)
+        ? gallery.fields
+        : gallery;
+    const raw = platform === '1688'
+        ? [
+              ...(isRecord(galleryFields) && Array.isArray(galleryFields.videos) ? galleryFields.videos : []),
+              ...(isRecord(galleryFields) && galleryFields.video !== undefined ? [galleryFields.video] : []),
+          ]
+        : isRecord(item) && Array.isArray(item.videos)
+          ? item.videos
+          : [];
+    let added = 0;
+    const seen = new Set(assets.map((asset) => `video:${asset.url}`));
+    for (const entry of raw) {
+        if (assets.length >= 500) break;
+        const record = isRecord(entry) ? entry : undefined;
+        const url = directVideoUrl(
+            typeof entry === 'string' ? entry : stringAt(record ?? {}, ['url', 'videoUrl', 'src']),
+            pageUrl,
+        );
+        if (!url || seen.has(`video:${url}`)) continue;
+        seen.add(`video:${url}`);
+        const posterUrl = imageUrl(
+            stringAt(record ?? {}, ['videoThumbnailURL', 'thumbnailUrl', 'coverUrl', 'coverImage', 'poster']),
+            pageUrl,
+        );
+        assets.push({
+            url,
+            mediaType: 'video',
+            ...(posterUrl ? { posterUrl } : {}),
+            role: 'source_main',
+            relation: 'product_candidate',
+            sortOrder: assets.length,
+            evidenceRef: `${platform}:structured:videos`,
+        });
+        added += 1;
+    }
+    return added;
+}
+
+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;
+    }
+}
+
+function imageUrl(value: string | undefined, pageUrl: string) {
+    if (!value) return undefined;
+    try {
+        const url = new URL(value, pageUrl);
+        return url.protocol === 'https:' ? url.href : undefined;
+    } catch {
+        return undefined;
+    }
+}
+
+function extractMarketplaceFacts(root: Document, states: unknown[], platform: Exclude<SourcePlatform, 'ozon'>) {
+    const attributes: CollectedProductPayload['common']['attributes'] = [];
+    const seen = new Set<string>();
+    for (const key of ['basicParamList', 'enhanceParamList', 'attributes', 'props']) {
+        const list = findFirst(states, key);
+        if (!Array.isArray(list)) continue;
+        for (const [index, entry] of list.entries()) {
+            if (!isRecord(entry)) continue;
+            const name = stringAt(entry, ['propertyName', 'name', 'key', 'title']);
+            const value = stringAt(entry, ['valueName', 'value', 'text']);
+            if (!name || !value || seen.has(`${name}\u0000${value}`)) continue;
+            seen.add(`${name}\u0000${value}`);
+            attributes.push({ key: `${platform}:${stringAt(entry, ['propertyId', 'id', 'key']) ?? index}`.slice(0, 200), name: name.slice(0, 300), value: value.slice(0, 2_000), raw_value: value.slice(0, 4_000) });
+        }
+    }
+    const product = jsonLdProduct(root);
+    const item = findFirst(states, 'item');
+    const temp = findFirst(states, 'tempModel');
+    const tradeModel = findFirst(states, 'tradeModel');
+    const priceVO = findFirst(states, 'priceVO');
+    const price = moneyAt(isRecord(tradeModel) ? tradeModel : {}, ['minPrice', 'price'])
+        ?? moneyAt(isRecord(priceVO) ? priceVO : {}, ['priceText', 'price', 'extraPrice'])
+        ?? decimal(product?.offers && isRecord(product.offers) ? product.offers.price : undefined);
+    const categoryNames = Array.from(root.querySelectorAll<HTMLAnchorElement>('[class*="breadcrumb"] a, [class*="BreadCrumb"] a, nav a[href*="category"]')).map((link) => link.textContent?.replace(/\s+/g, ' ').trim() ?? '').filter(Boolean).slice(0, 50);
+    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),
+        attributes,
+        categories: [...new Set(categoryNames)].map((name) => ({ key: `${platform}:category:${name.toLowerCase()}`.slice(0, 300), name })),
+    };
+}
+
+function scopedDescription(root: Document, platform: Exclude<SourcePlatform, 'ozon'>) {
+    const selectors = platform === '1688'
+        ? ['.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();
+        if (value && value.length >= 10) return value.slice(0, 100_000);
+    }
+    return undefined;
+}
+
+function jsonLdProduct(root: Document): Json | undefined {
+    for (const script of Array.from(root.scripts)) {
+        if (script.type !== 'application/ld+json' || !script.textContent?.trim()) continue;
+        try { const found = findJsonLdProduct(JSON.parse(script.textContent)); if (found) return found; } catch { /* ignored */ }
+    }
+    return undefined;
+}
+function findJsonLdProduct(value: unknown): Json | undefined {
+    if (Array.isArray(value)) { for (const child of value) { const found = findJsonLdProduct(child); if (found) return found; } return undefined; }
+    if (!isRecord(value)) return undefined;
+    const type = value['@type'];
+    if (type === 'Product' || (Array.isArray(type) && type.includes('Product'))) return value;
+    return Array.isArray(value['@graph']) ? findJsonLdProduct(value['@graph']) : undefined;
+}
+function decimal(value: unknown) { const raw = typeof value === 'string' || typeof value === 'number' ? String(value).replace(',', '.') : ''; return /^\d+(\.\d{1,4})?$/.test(raw) ? raw : undefined; }
+
 function pageState(root: Document) {
    const all: unknown[] = [];
    for (const script of Array.from(root.scripts)) {
@@ -5,7 +5,9 @@ import test from 'node:test';
 import {
    extractOzonCharacteristics,
    extractOzonDescription,
+    extractOzonJsonLdDescription,
    extractOzonGalleryImageUrls,
+    extractOzonGalleryVideos,
 } from '../lib/ozon-parser.ts';
 import { dedupeImageUrl, normaliseImageUrl } from '../lib/product-media.ts';

@@ -73,6 +75,41 @@ test('Ozon rich annotation converts to description text without images', () => {
    assert.equal(description?.value, '耐用的背包\n适合旅行');
 });

+test('Ozon JSON-LD Product description is the fallback when rich content is absent', () => {
+    const root = {
+        scripts: [{
+            type: 'application/ld+json',
+            textContent: JSON.stringify({
+                '@context': 'http://schema.org',
+                '@type': 'Product',
+                name: '喷雾器',
+                description: '来自 Ozon JSON-LD 的完整商品描述',
+            }),
+        }],
+    } as unknown as Document;
+    assert.equal(
+        extractOzonJsonLdDescription(root)?.value,
+        '来自 Ozon JSON-LD 的完整商品描述',
+    );
+});
+
+test('Ozon full characteristics merge with the five short header characteristics', () => {
+    const attributes = extractOzonCharacteristics([
+        {
+            __atlasWidgetName: 'webShortCharacteristics-fixture',
+            characteristics: [{ id: 'type', title: { textRs: [{ content: '类型' }] }, values: [{ text: '喷雾器' }] }],
+        },
+        {
+            __atlasWidgetName: 'webCharacteristics-fixture',
+            characteristics: [{ id: 'weight', title: { textRs: [{ content: '商品重量，克' }] }, values: [{ text: '385' }] }],
+        },
+    ]);
+    assert.deepEqual(attributes.map(({ name, value }) => ({ name, value })), [
+        { name: '类型', value: '喷雾器' },
+        { name: '商品重量，克', value: '385' },
+    ]);
+});
+
 test('Ozon thumbnail variants normalise to one original URL', () => {
    const original = 'https://ir.ozone.ru/s3/multimedia-1-9/13597160877.jpg';
    const thumbnail =
@@ -84,6 +121,29 @@ test('Ozon thumbnail variants normalise to one original URL', () => {
    );
 });

+test('Ozon gallery only accepts direct product MP4/WebM videos and retains poster', () => {
+    assert.deepEqual(
+        extractOzonGalleryVideos(
+            {
+                videos: [
+                    {
+                        url: 'https://cdn.ozon.ru/product/demo.mp4',
+                        coverImage: 'https://ir.ozone.ru/s3/multimedia-1/demo.jpg',
+                    },
+                    { videoUrl: 'https://cdn.ozon.ru/product/demo.mp4' },
+                    { url: 'https://cdn.ozon.ru/product/stream.m3u8' },
+                    { url: 'blob:https://www.ozon.ru/opaque' },
+                ],
+            },
+            'https://www.ozon.ru/product/demo-1/',
+        ),
+        [{
+            url: 'https://cdn.ozon.ru/product/demo.mp4',
+            posterUrl: 'https://ir.ozone.ru/s3/multimedia-1/demo.jpg',
+        }],
+    );
+});
+
 test('Alibaba transformed image URLs normalise before de-duplication', () => {
    const original = 'https://cbu01.alicdn.com/img/ibank/O1CN01.jpg';
    const transformed = `${original}_400x400.jpg?width=400`;