670 lines
27 KiB
TypeScript
670 lines
27 KiB
TypeScript
import type { CollectedProductPayload } from '@atlas/contracts';
|
|
import {
|
|
collectOzonProduct,
|
|
isOzonProductPage,
|
|
type OzonPageJsonFetcher,
|
|
} from './ozon-parser.js';
|
|
import { collectPlatformDomAssets, normaliseImageUrl, normaliseVideoUrl } from './product-media.js';
|
|
import { formatCollectedDescription } from './description-format.js';
|
|
|
|
type SourcePlatform = '1688' | 'taobao' | 'tmall' | 'ozon';
|
|
type Json = Record<string, unknown>;
|
|
type Axis = CollectedProductPayload['variantAxes'][number];
|
|
type Variant = CollectedProductPayload['variants'][number];
|
|
|
|
export function supportedProductPlatform(
|
|
url = location.href,
|
|
): SourcePlatform | null {
|
|
if (isOzonProductPage(url)) return 'ozon';
|
|
if (/^https:\/\/detail\.1688\.com\/offer\/\d+\.html/.test(url))
|
|
return '1688';
|
|
if (/^https:\/\/item\.taobao\.com\/item\.htm/.test(url)) return 'taobao';
|
|
if (/^https:\/\/detail\.tmall\.com\/item\.htm/.test(url)) return 'tmall';
|
|
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, pageState)
|
|
: collectTaobao(root, pageUrl, platform, pageState);
|
|
}
|
|
|
|
function collect1688(root: Document, pageUrl: string, bridgedState?: unknown): CollectedProductPayload {
|
|
const state = [bridgedState, ...pageState(root)];
|
|
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 = isRecord(skuModel) ? skuModel.skuProps : findFirst(state, 'skuProps');
|
|
const axes = parse1688Axes(skuProps);
|
|
const evidence: CollectedProductPayload['evidence'] = [
|
|
{
|
|
ref: 'page:url',
|
|
kind: 'page',
|
|
path: 'location.href',
|
|
raw_value: pageUrl,
|
|
},
|
|
];
|
|
const variants: Variant[] = [];
|
|
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: isRecord(skuInfoMap) ? 'skuModel.skuInfoMap' : 'skuMapOriginal',
|
|
raw_value: `组合记录 ${skuRows.length} 条`,
|
|
});
|
|
for (const [combination, row] of skuRows) {
|
|
if (!isRecord(row)) continue;
|
|
const sku = stringAt(row, ['skuId', 'sku_id', 'skuCode'])?.trim();
|
|
const sourceSku = sku && !/^(0|null|undefined)$/.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;
|
|
variants.push({
|
|
sourceVariantKey,
|
|
...(sourceSku ? { sourceSku } : {}),
|
|
displayName:
|
|
selections
|
|
.map((item) => item.valueName)
|
|
.join(' / ') || specAttrs,
|
|
selections,
|
|
...(moneyAt(row, ['price'])
|
|
? {
|
|
sourcePrice: {
|
|
amount: moneyAt(row, ['price'])!,
|
|
currencyCode: 'CNY',
|
|
},
|
|
}
|
|
: {}),
|
|
confidence: 'verified',
|
|
evidenceRefs: ['1688:sku-map'],
|
|
});
|
|
}
|
|
}
|
|
const payload = basePayload(
|
|
'1688',
|
|
offerId(pageUrl),
|
|
root,
|
|
pageUrl,
|
|
axes,
|
|
uniqueVariants(variants),
|
|
evidence,
|
|
'1688-structured-2.1.0',
|
|
state,
|
|
);
|
|
// SKU 图片由规格值和真实可售组合关联，独立于页面当时选中的 DOM 状态。
|
|
for (const variant of payload.variants) {
|
|
for (const selection of variant.selections) {
|
|
const rawUrl = axes.find((axis) => axis.key === selection.axisKey)
|
|
?.values.find((value) => value.key === selection.valueKey)?.image_url;
|
|
const url = rawUrl && normaliseImageUrl(rawUrl, pageUrl, '1688');
|
|
if (!url || payload.assets.length >= 500) continue;
|
|
if (payload.assets.some((asset) => asset.variantSourceKey === variant.sourceVariantKey && asset.url === url)) continue;
|
|
payload.assets.push({
|
|
url,
|
|
mediaType: 'image',
|
|
role: 'source_sku',
|
|
relation: 'axis_value',
|
|
variantSourceKey: variant.sourceVariantKey,
|
|
sortOrder: payload.assets.length,
|
|
evidenceRef: '1688:sku-map',
|
|
});
|
|
}
|
|
}
|
|
return payload;
|
|
}
|
|
|
|
function collectTaobao(
|
|
root: Document,
|
|
pageUrl: string,
|
|
platform: 'taobao' | 'tmall',
|
|
bridgedState?: unknown,
|
|
): CollectedProductPayload {
|
|
const state = [bridgedState, ...pageState(root)];
|
|
const skuBase = findFirst(state, 'skuBase');
|
|
const skuCore = findFirst(state, 'skuCore');
|
|
const axes = parseTaobaoAxes(isRecord(skuBase) ? skuBase.props : undefined);
|
|
const evidence: CollectedProductPayload['evidence'] = [
|
|
{
|
|
ref: 'page:url',
|
|
kind: 'page',
|
|
path: 'location.href',
|
|
raw_value: pageUrl,
|
|
},
|
|
];
|
|
const variants: Variant[] = [];
|
|
const skus =
|
|
isRecord(skuBase) && Array.isArray(skuBase.skus) ? skuBase.skus : [];
|
|
const sku2info =
|
|
isRecord(skuCore) && isRecord(skuCore.sku2info) ? skuCore.sku2info : {};
|
|
if (skus.length)
|
|
evidence.push({
|
|
ref: 'taobao:sku-base',
|
|
kind: 'state',
|
|
path: 'skuBase.skus',
|
|
raw_value: `组合记录 ${skus.length} 条`,
|
|
});
|
|
for (const sku of skus) {
|
|
if (!isRecord(sku)) continue;
|
|
const skuId = stringAt(sku, ['skuId', 'sku_id']);
|
|
const propPath = stringAt(sku, ['propPath', 'prop_path']);
|
|
// skuId + propPath is the required combination proof. Props alone only
|
|
// become axes and cause a single default SKU on the server.
|
|
if (!skuId || !propPath) continue;
|
|
const info = isRecord(sku2info[skuId]) ? sku2info[skuId] : {};
|
|
const selections = selectionsForPropPath(propPath, axes);
|
|
variants.push({
|
|
sourceVariantKey: skuId,
|
|
sourceSku: skuId,
|
|
displayName:
|
|
selections.map((item) => item.valueName).join(' / ') ||
|
|
`SKU ${skuId}`,
|
|
selections,
|
|
...(moneyAt(info, ['price', 'promotionPrice', 'discountPrice'])
|
|
? {
|
|
sourcePrice: {
|
|
amount: moneyAt(info, [
|
|
'price',
|
|
'promotionPrice',
|
|
'discountPrice',
|
|
])!,
|
|
currencyCode: 'CNY',
|
|
},
|
|
}
|
|
: {}),
|
|
confidence: 'verified',
|
|
evidenceRefs: ['taobao:sku-base'],
|
|
});
|
|
}
|
|
return basePayload(
|
|
platform,
|
|
itemId(pageUrl),
|
|
root,
|
|
pageUrl,
|
|
axes,
|
|
uniqueVariants(variants),
|
|
evidence,
|
|
`${platform}-structured-2.0.1`,
|
|
state,
|
|
);
|
|
}
|
|
|
|
function basePayload(
|
|
platform: Exclude<SourcePlatform, 'ozon'>,
|
|
itemKey: string,
|
|
root: Document,
|
|
pageUrl: string,
|
|
axes: Axis[],
|
|
variants: Variant[],
|
|
evidence: CollectedProductPayload['evidence'],
|
|
parserVersion: string,
|
|
states: unknown[],
|
|
): CollectedProductPayload {
|
|
const canonical =
|
|
root.querySelector<HTMLLinkElement>('link[rel="canonical"]')?.href ??
|
|
pageUrl.split('#')[0];
|
|
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()
|
|
);
|
|
const assets = collectPlatformDomAssets(platform, root, canonical);
|
|
evidence.push({
|
|
ref: `${platform}:dom:image-profile`,
|
|
kind: 'dom',
|
|
path: 'platform-scoped gallery / sku / detail selectors',
|
|
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({
|
|
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 ${facts.description.length} 个字符描述` });
|
|
return {
|
|
source: {
|
|
platform,
|
|
itemKey,
|
|
canonicalUrl: canonical,
|
|
collectedAt: new Date().toISOString(),
|
|
parserVersion,
|
|
},
|
|
common: {
|
|
...(title
|
|
? {
|
|
title: {
|
|
value: title,
|
|
raw_value: title,
|
|
confidence: 'verified' as const,
|
|
},
|
|
}
|
|
: {}),
|
|
...(facts.price ? { sourcePrice: { amount: facts.price, currencyCode: 'CNY', raw_value: facts.price } } : {}),
|
|
...(facts.description ? { description: { value: facts.description, raw_value: facts.description, confidence: 'verified' as const } } : {}),
|
|
attributes: facts.attributes,
|
|
categories: facts.categories,
|
|
},
|
|
variantAxes: axes,
|
|
variants,
|
|
assets,
|
|
evidence,
|
|
};
|
|
}
|
|
|
|
/** 结构化画廊数据补充公开视频地址，与受限画廊 DOM 去重，不读取推荐或评价视频。 */
|
|
function addMarketplaceVideos(
|
|
platform: Exclude<SourcePlatform, 'ozon'>,
|
|
states: unknown[],
|
|
assets: CollectedProductPayload['assets'],
|
|
pageUrl: string,
|
|
) {
|
|
const gallery = findFirst(states, 'gallery');
|
|
const item = findFirst(states, 'item');
|
|
const galleryFields = isRecord(gallery) && isRecord(gallery.fields)
|
|
? gallery.fields
|
|
: gallery;
|
|
const raw = platform === '1688'
|
|
? [
|
|
...(isRecord(galleryFields) && Array.isArray(galleryFields.videos) ? galleryFields.videos : []),
|
|
...(isRecord(galleryFields) && galleryFields.video !== undefined ? [galleryFields.video] : []),
|
|
]
|
|
: isRecord(item) && Array.isArray(item.videos)
|
|
? item.videos
|
|
: [];
|
|
let added = 0;
|
|
const seen = new Set(assets.map((asset) => `video:${asset.url}`));
|
|
for (const entry of raw) {
|
|
if (assets.length >= 500) break;
|
|
const record = isRecord(entry) ? entry : undefined;
|
|
const 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',
|
|
...(posterUrl ? { posterUrl } : {}),
|
|
role: 'source_main',
|
|
relation: 'product_candidate',
|
|
sortOrder: assets.length,
|
|
evidenceRef: `${platform}:structured:videos`,
|
|
});
|
|
added += 1;
|
|
}
|
|
return added;
|
|
}
|
|
|
|
function directVideoUrl(value: string | undefined, pageUrl: string) {
|
|
return value ? normaliseVideoUrl(value, pageUrl) : undefined;
|
|
}
|
|
|
|
function imageUrl(value: string | undefined, pageUrl: string) {
|
|
if (!value) return undefined;
|
|
try {
|
|
const url = new URL(value, pageUrl);
|
|
return url.protocol === 'https:' ? url.href : undefined;
|
|
} catch {
|
|
return undefined;
|
|
}
|
|
}
|
|
|
|
function extractMarketplaceFacts(root: Document, states: unknown[], platform: Exclude<SourcePlatform, 'ozon'>) {
|
|
const attributes: CollectedProductPayload['common']['attributes'] = [];
|
|
const seen = new Set<string>();
|
|
for (const key of ['basicParamList', 'enhanceParamList', 'attributes', 'props']) {
|
|
const list = findFirst(states, key);
|
|
if (!Array.isArray(list)) continue;
|
|
for (const [index, entry] of list.entries()) {
|
|
if (!isRecord(entry)) continue;
|
|
const name = stringAt(entry, ['propertyName', 'name', 'key', 'title']);
|
|
const value = stringAt(entry, ['valueName', 'value', 'text']);
|
|
if (!name || !value || seen.has(`${name}\u0000${value}`)) continue;
|
|
seen.add(`${name}\u0000${value}`);
|
|
attributes.push({ key: `${platform}:${stringAt(entry, ['propertyId', 'id', 'key']) ?? index}`.slice(0, 200), name: name.slice(0, 300), value: value.slice(0, 2_000), raw_value: value.slice(0, 4_000) });
|
|
}
|
|
}
|
|
const product = jsonLdProduct(root);
|
|
const item = findFirst(states, 'item');
|
|
const temp = findFirst(states, 'tempModel');
|
|
const tradeModel = findFirst(states, 'tradeModel');
|
|
const priceVO = findFirst(states, 'priceVO');
|
|
const price = moneyAt(isRecord(tradeModel) ? tradeModel : {}, ['minPrice', 'price'])
|
|
?? moneyAt(isRecord(priceVO) ? priceVO : {}, ['priceText', 'price', 'extraPrice'])
|
|
?? decimal(product?.offers && isRecord(product.offers) ? product.offers.price : undefined);
|
|
const categoryNames = Array.from(root.querySelectorAll<HTMLAnchorElement>('[class*="breadcrumb"] a, [class*="BreadCrumb"] a, nav a[href*="category"]')).map((link) => link.textContent?.replace(/\s+/g, ' ').trim() ?? '').filter(Boolean).slice(0, 50);
|
|
return {
|
|
title: (isRecord(item) ? stringAt(item, ['title', 'name']) : undefined) ?? (isRecord(temp) ? stringAt(temp, ['offerTitle']) : undefined) ?? (product ? stringAt(product, ['name']) : undefined),
|
|
price,
|
|
description: formatCollectedDescription(
|
|
(product ? stringAt(product, ['description']) : undefined) ??
|
|
scopedDescription(root, platform) ??
|
|
'',
|
|
) || undefined,
|
|
attributes,
|
|
categories: [...new Set(categoryNames)].map((name) => ({ key: `${platform}:category:${name.toLowerCase()}`.slice(0, 300), name })),
|
|
};
|
|
}
|
|
|
|
function scopedDescription(root: Document, platform: Exclude<SourcePlatform, 'ozon'>) {
|
|
const selectors = platform === '1688'
|
|
? ['.de-description-detail', '#detailContentContainer', '.html-description']
|
|
: ['[class*="tabDetailWrap--"]', '[class*="detailInfo--"]'];
|
|
for (const selector of selectors) {
|
|
const element = root.querySelector<HTMLElement>(selector);
|
|
const value = formatCollectedDescription(
|
|
element?.innerText ?? element?.textContent ?? '',
|
|
);
|
|
if (value && value.length >= 10) return value.slice(0, 100_000);
|
|
}
|
|
return undefined;
|
|
}
|
|
|
|
function jsonLdProduct(root: Document): Json | undefined {
|
|
for (const script of Array.from(root.scripts)) {
|
|
if (script.type !== 'application/ld+json' || !script.textContent?.trim()) continue;
|
|
try { const found = findJsonLdProduct(JSON.parse(script.textContent)); if (found) return found; } catch { /* ignored */ }
|
|
}
|
|
return undefined;
|
|
}
|
|
function findJsonLdProduct(value: unknown): Json | undefined {
|
|
if (Array.isArray(value)) { for (const child of value) { const found = findJsonLdProduct(child); if (found) return found; } return undefined; }
|
|
if (!isRecord(value)) return undefined;
|
|
const type = value['@type'];
|
|
if (type === 'Product' || (Array.isArray(type) && type.includes('Product'))) return value;
|
|
return Array.isArray(value['@graph']) ? findJsonLdProduct(value['@graph']) : undefined;
|
|
}
|
|
function decimal(value: unknown) { const raw = typeof value === 'string' || typeof value === 'number' ? String(value).replace(',', '.') : ''; return /^\d+(\.\d{1,4})?$/.test(raw) ? raw : undefined; }
|
|
|
|
function pageState(root: Document) {
|
|
const all: unknown[] = [];
|
|
for (const script of Array.from(root.scripts)) {
|
|
const text = script.textContent?.trim() ?? '';
|
|
if (text.length < 2 || text.length > 5_000_000) continue;
|
|
if (script.type === 'application/ld+json') {
|
|
try {
|
|
all.push(JSON.parse(text));
|
|
} catch {
|
|
/* ignored evidence */
|
|
}
|
|
}
|
|
for (const marker of [
|
|
'gallery',
|
|
'skuModel',
|
|
'skuInfoMap',
|
|
'skuMapOriginal',
|
|
'skuBase',
|
|
'skuCore',
|
|
'skuProps',
|
|
]) {
|
|
const object = objectAfterMarker(text, marker);
|
|
if (object) all.push({ [marker]: object });
|
|
}
|
|
}
|
|
return all;
|
|
}
|
|
function objectAfterMarker(text: string, marker: string): unknown | null {
|
|
// 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;
|
|
let quote = '';
|
|
let escaped = false;
|
|
for (
|
|
let index = first;
|
|
index < text.length && index - first < 2_000_000;
|
|
index += 1
|
|
) {
|
|
const char = text[index];
|
|
if (quote) {
|
|
if (escaped) escaped = false;
|
|
else if (char === '\\') escaped = true;
|
|
else if (char === quote) quote = '';
|
|
continue;
|
|
}
|
|
if (char === '"' || char === "'") {
|
|
quote = char;
|
|
continue;
|
|
}
|
|
if (char === open) depth += 1;
|
|
if (char === close && --depth === 0) {
|
|
try {
|
|
return JSON.parse(text.slice(first, index + 1));
|
|
} catch {
|
|
return null;
|
|
}
|
|
}
|
|
}
|
|
return null;
|
|
}
|
|
function findFirst(states: unknown[], key: string): unknown {
|
|
for (const state of states) {
|
|
const found = findKey(state, key);
|
|
if (found !== undefined) return found;
|
|
}
|
|
return undefined;
|
|
}
|
|
function findKey(
|
|
value: unknown,
|
|
key: string,
|
|
seen = new Set<unknown>(),
|
|
): unknown {
|
|
if (!value || typeof value !== 'object' || seen.has(value))
|
|
return undefined;
|
|
seen.add(value);
|
|
if (isRecord(value) && key in value) return value[key];
|
|
for (const child of Array.isArray(value) ? value : Object.values(value)) {
|
|
const found = findKey(child, key, seen);
|
|
if (found !== undefined) return found;
|
|
}
|
|
return undefined;
|
|
}
|
|
function parse1688Axes(value: unknown): Axis[] {
|
|
const entries = Array.isArray(value)
|
|
? value
|
|
: isRecord(value) && Array.isArray(value.skuProps)
|
|
? value.skuProps
|
|
: [];
|
|
return entries.flatMap((entry, axisIndex) => {
|
|
if (!isRecord(entry)) return [];
|
|
const axisKey =
|
|
stringAt(entry, ['prop', 'propId', 'id']) ?? `axis:${axisIndex}`;
|
|
const name = stringAt(entry, ['name', 'propName']) ?? axisKey;
|
|
const values = (
|
|
Array.isArray(entry.value)
|
|
? entry.value
|
|
: Array.isArray(entry.values)
|
|
? entry.values
|
|
: []
|
|
).flatMap((item, valueIndex) =>
|
|
isRecord(item)
|
|
? [
|
|
{
|
|
key:
|
|
stringAt(item, ['valueId', 'vid', 'id']) ??
|
|
`${axisKey}:${valueIndex}`,
|
|
name:
|
|
stringAt(item, ['name', 'valueName']) ??
|
|
`${valueIndex + 1}`,
|
|
...(urlAt(item, [
|
|
'imageUrl',
|
|
'image',
|
|
'imageUrlLarge',
|
|
])
|
|
? {
|
|
image_url: urlAt(item, [
|
|
'imageUrl',
|
|
'image',
|
|
'imageUrlLarge',
|
|
])!,
|
|
}
|
|
: {}),
|
|
},
|
|
]
|
|
: [],
|
|
);
|
|
return values.length ? [{ key: axisKey, name, values }] : [];
|
|
});
|
|
}
|
|
function parseTaobaoAxes(value: unknown): Axis[] {
|
|
return (Array.isArray(value) ? value : []).flatMap((entry, index) => {
|
|
if (!isRecord(entry)) return [];
|
|
const key = stringAt(entry, ['pid', 'propId']) ?? `axis:${index}`;
|
|
const name = stringAt(entry, ['name', 'propName']) ?? key;
|
|
const values = (
|
|
Array.isArray(entry.values) ? entry.values : []
|
|
).flatMap((item, valueIndex) =>
|
|
isRecord(item)
|
|
? [
|
|
{
|
|
key:
|
|
stringAt(item, ['vid', 'valueId']) ??
|
|
`${key}:${valueIndex}`,
|
|
name:
|
|
stringAt(item, ['name', 'valueName']) ??
|
|
`${valueIndex + 1}`,
|
|
...(urlAt(item, ['image', 'imageUrl'])
|
|
? {
|
|
image_url: urlAt(item, [
|
|
'image',
|
|
'imageUrl',
|
|
])!,
|
|
}
|
|
: {}),
|
|
},
|
|
]
|
|
: [],
|
|
);
|
|
return values.length ? [{ key, name, values }] : [];
|
|
});
|
|
}
|
|
function selectionsForPropPath(path: string, axes: Axis[]) {
|
|
return path.split(/[;,]/).flatMap((pair) => {
|
|
const [axisKey, valueKey] = pair.split(':');
|
|
const axis = axes.find((item) => item.key === axisKey);
|
|
const value = axis?.values.find((item) => item.key === valueKey);
|
|
return axis && value
|
|
? [{ axisKey, valueKey, valueName: value.name }]
|
|
: [];
|
|
});
|
|
}
|
|
function selectionsForCombination(combination: string, axes: Axis[]) {
|
|
const ids = combination
|
|
.split(/[;,&]/)
|
|
.map((value) => value.trim())
|
|
.filter(Boolean);
|
|
return axes.flatMap((axis) => {
|
|
const value = axis.values.find((item) => ids.includes(item.key));
|
|
return value
|
|
? [
|
|
{
|
|
axisKey: axis.key,
|
|
valueKey: value.key,
|
|
valueName: value.name,
|
|
},
|
|
]
|
|
: [];
|
|
});
|
|
}
|
|
function selectionsFor1688Combination(combination: string, axes: Axis[]) {
|
|
const selected = selectionsForCombination(combination, axes);
|
|
if (selected.length === axes.length) return selected;
|
|
const parts = combination.split(/[;,&|]/).map((part) => part.trim()).filter(Boolean);
|
|
return axes.flatMap((axis) => {
|
|
const existing = selected.find((item) => item.axisKey === axis.key);
|
|
if (existing) return [existing];
|
|
const value = axis.values.find((item) =>
|
|
parts.some((part) => part === item.name || part === `${axis.name}:${item.name}`)
|
|
);
|
|
return value ? [{ axisKey: axis.key, valueKey: value.key, valueName: value.name }] : [];
|
|
});
|
|
}
|
|
function stringAt(value: Json, keys: string[]) {
|
|
for (const key of keys)
|
|
if (typeof value[key] === 'string' || typeof value[key] === 'number')
|
|
return String(value[key]);
|
|
return undefined;
|
|
}
|
|
function moneyAt(value: Json, keys: string[]) {
|
|
const raw = stringAt(value, keys)?.replace(/,/g, '.');
|
|
return raw && /^\d+(\.\d{1,4})?$/.test(raw) ? raw : undefined;
|
|
}
|
|
function urlAt(value: Json, keys: string[]) {
|
|
const raw = stringAt(value, keys);
|
|
return raw && /^https:\/\//.test(raw) ? raw : undefined;
|
|
}
|
|
function uniqueVariants(variants: Variant[]) {
|
|
const seen = new Set<string>();
|
|
return variants.filter(
|
|
(variant) =>
|
|
!seen.has(variant.sourceVariantKey) &&
|
|
!!seen.add(variant.sourceVariantKey),
|
|
);
|
|
}
|
|
function offerId(url: string) {
|
|
return (
|
|
url.match(/\/offer\/(\d+)\.html/)?.[1] ??
|
|
`1688:${new URL(url).pathname}`
|
|
);
|
|
}
|
|
function itemId(url: string) {
|
|
return new URL(url).searchParams.get('id') ?? new URL(url).pathname;
|
|
}
|
|
function isRecord(value: unknown): value is Json {
|
|
return !!value && typeof value === 'object' && !Array.isArray(value);
|
|
}
|