Update seller store experience
This commit is contained in:
R524809
@@ -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)) {
|
||||
|
||||
Reference in New Issue
Block a user