SPB Git

spb/rareindex

Public
17.9 KB · 292 lines typescript
1import { and, eq, sql } from 'drizzle-orm';2import { auctionLots, auctions, images, listingEvents, listings, news, populationReports, priceObservations, sales, sources } from '@rareindex/database';3import { normalizeCondition } from '@rareindex/taxonomy';4import { newId, sha256, toDateOnly, type NormalizedAuctionLot, type NormalizedCatalogItem, type NormalizedListing, type NormalizedNewsItem, type NormalizedPopulationReport, type NormalizedPriceObservation, type NormalizedRecord, type NormalizedSale } from '@rareindex/shared';5import { db } from '../lib/db.ts';6import { toUsd } from '../lib/fx.ts';7import { audit } from '../lib/audit.ts';8import { enrichAttributesFromTitle, ensureVariant, resolveAsset, type Resolution } from './resolver.ts';9import { recordCertificate } from './certificates.ts';1011/** Record a slab certification sighting when a record carries one (SPEC §22); never throws into the writer. */12async function trackCert(rec: { grade: { grader: string | null; grade: string | null; qualifier: string | null; certificationNumber: string | null }; sourceId: string; connectorId: string; sourceUrl: string }, kind: 'sale' | 'listing' | 'auction_lot', targetId: string | null, assetId: string | null, variantId: string | null, priceUsd: number | null, price: number | null, currency: string | null, observedAt: Date): Promise<void> {13  if (!rec.grade.certificationNumber || !rec.grade.grader) return;14  try {15    await recordCertificate({ grader: rec.grade.grader, certNumber: rec.grade.certificationNumber, assetId, variantId, grade: rec.grade.grade, qualifier: rec.grade.qualifier, kind, targetId, sourceId: rec.sourceId, connectorId: rec.connectorId, sourceUrl: rec.sourceUrl, priceUsd, price, currency, observedAt });16  } catch {17    /* cert tracking is best-effort; the canonical write already succeeded */18  }19}2021export class FxMissingError extends Error {22  constructor(public currency: string, public date: Date) {23    super(`fx_missing:${currency}:${toDateOnly(date)}`);24  }25}2627export interface ApplyResult {28  assetId: string | null;29  variantId: string | null;30  targetId: string | null;31  method: Resolution['method'] | null;32  confidence: number | null;33  status: 'applied' | 'unmatched' | 'rejected' | 'duplicate';34  reason?: string;35  event?: 'sale_detected' | 'listing_created' | 'listing_updated' | 'entity_created';36}3738const trustCache = new Map<string, number>();39async function sourceTrust(sourceId: string): Promise<number> {40  const hit = trustCache.get(sourceId);41  if (hit !== undefined) return hit;42  const [row] = await db().select({ t: sources.trustScore }).from(sources).where(eq(sources.id, sourceId)).limit(1);43  const t = row ? Number(row.t) : 0.5;44  trustCache.set(sourceId, t);45  return t;46}4748/** Data quality 0–100 (§150): completeness × source trust × identification confidence. */49function dataQuality(fields: Record<string, unknown>, trust: number, confidence: number, extra = 0): number {50  const keys = Object.keys(fields);51  const present = keys.filter((k) => fields[k] !== null && fields[k] !== undefined && fields[k] !== '' && !(Array.isArray(fields[k]) && (fields[k] as unknown[]).length === 0)).length;52  const completeness = keys.length ? present / keys.length : 0;53  return Math.round(Math.max(0, Math.min(100, (0.4 * completeness + 0.3 * trust + 0.3 * confidence + extra) * 100)));54}5556function normCond(categorySlug: string, raw: string | null | undefined, already: string | null | undefined): string | null {57  return already ?? normalizeCondition(categorySlug, raw);58}5960async function storeImages(assetId: string, urls: string[], sourceId: string, ref: { listingId?: string; saleId?: string }): Promise<void> {61  if (!urls.length) return;62  const rows = urls.slice(0, 12).map((url, i) => ({ id: newId('image'), assetId, listingId: ref.listingId ?? null, saleId: ref.saleId ?? null, sourceId, url, role: i === 0 && !ref.listingId && !ref.saleId ? 'hero' : 'source' }));63  await db().insert(images).values(rows).onConflictDoNothing({ target: images.url });64}6566export async function applyRecord(rec: NormalizedRecord): Promise<ApplyResult> {67  switch (rec.kind) {68    case 'catalog_item':69      return applyCatalog(rec);70    case 'sale':71      return applySale(rec);72    case 'listing':73      return applyListing(rec);74    case 'price_observation':75      return applyObservation(rec);76    case 'auction_lot':77      return applyAuctionLot(rec);78    case 'population_report':79      return applyPopulation(rec);80    case 'news_item':81      return applyNews(rec);82  }83}8485async function applyCatalog(rec: NormalizedCatalogItem): Promise<ApplyResult> {86  const r = await resolveAsset(rec.attributes, { enrich: true, imageUrls: rec.imageUrls, description: rec.description });87  if (!r) return { assetId: null, variantId: null, targetId: null, method: null, confidence: null, status: 'unmatched' };88  if (rec.imageUrls.length) await storeImages(r.assetId, rec.imageUrls.slice(0, 1), rec.sourceId, {});89  return { assetId: r.assetId, variantId: null, targetId: r.assetId, method: r.method, confidence: r.confidence, status: 'applied', event: r.created ? 'entity_created' : undefined };90}9192async function applySale(rec: NormalizedSale): Promise<ApplyResult> {93  const r = await resolveAsset(enrichAttributesFromTitle(rec.attributes, rec.rawTitle), { imageUrls: rec.imageUrls });94  if (!r) return { assetId: null, variantId: null, targetId: null, method: null, confidence: null, status: 'unmatched' };95  const condition = normCond(rec.attributes.categorySlug, rec.condition.conditionRaw, rec.condition.condition);96  const variant = await ensureVariant(r.assetId, rec.grade, { condition, completeness: rec.condition.completeness }, rec.attributes.size);97  const fx = await toUsd(rec.price, rec.currency, rec.saleDate);98  if (!fx) throw new FxMissingError(rec.currency, rec.saleDate);99  const dedupeKey = sha256(['sale', rec.sourceId, rec.externalId ?? rec.sourceUrl, toDateOnly(rec.saleDate), rec.price.toFixed(2)].join('|'));100  const trust = await sourceTrust(rec.sourceId);101  const confidence = Math.min(rec.confidence, r.confidence);102  const flags: string[] = [];103  if (rec.isBundle || rec.quantity > 1) flags.push('bundle');104  if (rec.price <= 0) flags.push('zero_price');105  if (confidence < 0.6) flags.push('low_identification_confidence');106  const status = flags.includes('bundle') || flags.includes('zero_price') ? 'excluded' : 'valid';107  const id = newId('sale');108  const inserted = await db()109    .insert(sales)110    .values({111      id,112      assetId: r.assetId,113      variantId: variant.id,114      sourceId: rec.sourceId,115      connectorId: rec.connectorId,116      sourceUrl: rec.sourceUrl,117      externalId: rec.externalId,118      saleType: rec.saleType,119      saleDate: rec.saleDate,120      price: rec.price,121      currency: rec.currency,122      priceUsd: fx.usd,123      fxRate: fx.rate,124      fxDate: fx.fxDate,125      buyerPremiumIncluded: rec.buyerPremiumIncluded,126      quantity: rec.quantity,127      isBundle: rec.isBundle,128      condition,129      grader: rec.grade.grader,130      grade: rec.grade.grade,131      certificationNumber: rec.grade.certificationNumber,132      location: rec.location,133      auctionHouse: rec.auctionHouse,134      lotNumber: rec.lotNumber,135      imageUrls: rec.imageUrls,136      rawTitle: rec.rawTitle,137      confidence,138      dataQuality: dataQuality({ date: rec.saleDate, price: rec.price, currency: rec.currency, images: rec.imageUrls, grade: rec.grade.grade, condition, externalId: rec.externalId, type: rec.saleType === 'unknown' ? null : rec.saleType }, trust, confidence),139      status,140      flags,141      dedupeKey,142    })143    .onConflictDoNothing({ target: sales.dedupeKey })144    .returning({ id: sales.id });145  if (inserted.length === 0) return { assetId: r.assetId, variantId: variant.id, targetId: null, method: r.method, confidence, status: 'duplicate' };146  if (status === 'excluded') await audit({ entityType: 'sale', entityId: id, action: 'excluded', reason: flags.join(','), details: { rawTitle: rec.rawTitle, quantity: rec.quantity } });147  if (rec.imageUrls.length) await storeImages(r.assetId, rec.imageUrls.slice(0, 3), rec.sourceId, { saleId: id });148  await trackCert(rec, 'sale', id, r.assetId, variant.id, fx.usd, rec.price, rec.currency, rec.saleDate);149  return { assetId: r.assetId, variantId: variant.id, targetId: id, method: r.method, confidence, status: 'applied', event: 'sale_detected' };150}151152async function applyListing(rec: NormalizedListing): Promise<ApplyResult> {153  const r = await resolveAsset(enrichAttributesFromTitle(rec.attributes, rec.rawTitle), { imageUrls: rec.imageUrls });154  if (!r) return { assetId: null, variantId: null, targetId: null, method: null, confidence: null, status: 'unmatched' };155  const condition = normCond(rec.attributes.categorySlug, rec.condition.conditionRaw, rec.condition.condition);156  const variant = await ensureVariant(r.assetId, rec.grade, { condition, completeness: rec.condition.completeness }, rec.attributes.size);157  let priceUsd: number | null = null;158  if (rec.price !== null && rec.currency) {159    const fx = await toUsd(rec.price, rec.currency, rec.observedAt);160    if (!fx) throw new FxMissingError(rec.currency, rec.observedAt);161    priceUsd = fx.usd;162  }163  const externalId = rec.externalId ?? sha256(rec.sourceUrl);164  const trust = await sourceTrust(rec.sourceId);165  const confidence = Math.min(rec.confidence, r.confidence);166  const now = rec.observedAt;167  const [existing] = await db().select({ id: listings.id, price: listings.price, availability: listings.availability }).from(listings).where(and(eq(listings.sourceId, rec.sourceId), eq(listings.externalId, externalId))).limit(1);168  const quality = dataQuality({ price: rec.price, currency: rec.currency, images: rec.imageUrls, seller: rec.seller, condition, grade: rec.grade.grade, listedAt: rec.listedAt, description: rec.description }, trust, confidence);169  if (!existing) {170    const id = newId('listing');171    await db().insert(listings).values({172      id,173      assetId: r.assetId,174      variantId: variant.id,175      sourceId: rec.sourceId,176      connectorId: rec.connectorId,177      sourceUrl: rec.sourceUrl,178      externalId,179      listingType: rec.listingType,180      price: rec.price,181      currency: rec.currency,182      priceUsd,183      seller: rec.seller,184      sellerReputation: rec.sellerReputation,185      location: rec.location,186      shippingCost: rec.shippingCost,187      quantity: rec.quantity,188      condition,189      grader: rec.grade.grader,190      grade: rec.grade.grade,191      certificationNumber: rec.grade.certificationNumber,192      imageUrls: rec.imageUrls,193      rawTitle: rec.rawTitle,194      description: rec.description,195      listedAt: rec.listedAt,196      endsAt: rec.endsAt,197      availability: rec.availability,198      bidCount: rec.bidCount,199      firstSeenAt: now,200      lastSeenAt: now,201      confidence,202      dataQuality: quality,203      flags: confidence < 0.6 ? ['low_identification_confidence'] : [],204    });205    await db().insert(listingEvents).values({ id: newId('event'), listingId: id, eventType: 'new', newPrice: rec.price, currency: rec.currency, occurredAt: now });206    if (rec.imageUrls.length) await storeImages(r.assetId, rec.imageUrls.slice(0, 3), rec.sourceId, { listingId: id });207    await trackCert(rec, 'listing', id, r.assetId, variant.id, priceUsd, rec.price, rec.currency, now);208    return { assetId: r.assetId, variantId: variant.id, targetId: id, method: r.method, confidence, status: 'applied', event: 'listing_created' };209  }210  const priceChanged = rec.price !== null && existing.price !== null && Math.abs(Number(existing.price) - rec.price) > 0.009;211  const relisted = existing.availability !== 'available' && rec.availability === 'available';212  const ended = existing.availability === 'available' && rec.availability !== 'available';213  await db()214    .update(listings)215    .set({216      price: rec.price,217      currency: rec.currency,218      priceUsd,219      availability: rec.availability,220      bidCount: rec.bidCount,221      endsAt: rec.endsAt ?? sql`${listings.endsAt}`,222      lastSeenAt: now,223      priceChangedAt: priceChanged ? now : sql`${listings.priceChangedAt}`,224      quantity: rec.quantity,225      dataQuality: quality,226      updatedAt: new Date(),227    })228    .where(eq(listings.id, existing.id));229  const ev = priceChanged ? 'price_changed' : relisted ? 'relisted' : ended ? (rec.availability === 'sold' ? 'sold' : rec.availability === 'ended' ? 'auction_ended' : 'removed') : null;230  if (ev) await db().insert(listingEvents).values({ id: newId('event'), listingId: existing.id, eventType: ev, oldPrice: existing.price, newPrice: rec.price, currency: rec.currency, occurredAt: now });231  return { assetId: r.assetId, variantId: variant.id, targetId: existing.id, method: r.method, confidence, status: 'applied', event: ev ? 'listing_updated' : undefined };232}233234async function applyObservation(rec: NormalizedPriceObservation): Promise<ApplyResult> {235  const r = await resolveAsset(rec.attributes, { imageUrls: rec.imageUrls });236  if (!r) return { assetId: null, variantId: null, targetId: null, method: null, confidence: null, status: 'unmatched' };237  const condition = normCond(rec.attributes.categorySlug, rec.condition.conditionRaw, rec.condition.condition);238  const variant = await ensureVariant(r.assetId, rec.grade, { condition, completeness: rec.condition.completeness }, rec.attributes.size);239  const fx = await toUsd(rec.price, rec.currency, rec.observationDate);240  if (!fx) throw new FxMissingError(rec.currency, rec.observationDate);241  const dedupeKey = sha256(['obs', rec.sourceId, rec.externalId ?? rec.sourceUrl, rec.priceKind, variant.key, toDateOnly(rec.observationDate)].join('|'));242  const id = newId('valuation');243  const inserted = await db()244    .insert(priceObservations)245    .values({ id, assetId: r.assetId, variantId: variant.id, sourceId: rec.sourceId, connectorId: rec.connectorId, sourceUrl: rec.sourceUrl, priceKind: rec.priceKind, price: rec.price, currency: rec.currency, priceUsd: fx.usd, observationDate: toDateOnly(rec.observationDate), sampleSize: rec.sampleSize, dedupeKey })246    .onConflictDoNothing({ target: priceObservations.dedupeKey })247    .returning({ id: priceObservations.id });248  if (inserted.length === 0) return { assetId: r.assetId, variantId: variant.id, targetId: null, method: r.method, confidence: r.confidence, status: 'duplicate' };249  return { assetId: r.assetId, variantId: variant.id, targetId: id, method: r.method, confidence: r.confidence, status: 'applied' };250}251252async function applyAuctionLot(rec: NormalizedAuctionLot): Promise<ApplyResult> {253  const r = await resolveAsset(enrichAttributesFromTitle(rec.attributes, rec.rawTitle), { imageUrls: rec.imageUrls, allowCreate: rec.confidence >= 0.7 });254  const auctionUrl = (rec.attributes.metadata?.auctionUrl as string | undefined) ?? `${new URL(rec.sourceUrl).origin}#${rec.auctionHouse}:${rec.auctionName ?? 'auction'}`;255  const auctionId = sha256(auctionUrl).slice(0, 24);256  await db()257    .insert(auctions)258    .values({ id: `auc_${auctionId}`, sourceId: rec.sourceId, auctionHouse: rec.auctionHouse, name: rec.auctionName ?? rec.auctionHouse, url: auctionUrl, startsAt: rec.startsAt, endsAt: rec.endsAt, location: rec.location, categorySlugs: [rec.attributes.categorySlug], status: rec.status === 'unknown' ? 'upcoming' : rec.status, currency: rec.currency })259    .onConflictDoUpdate({ target: auctions.url, set: { endsAt: sql`coalesce(excluded.ends_at, ${auctions.endsAt})`, status: sql`excluded.status`, updatedAt: new Date() } });260  let variantId: string | null = null;261  if (r) variantId = (await ensureVariant(r.assetId, rec.grade, { condition: normCond(rec.attributes.categorySlug, rec.condition.conditionRaw, rec.condition.condition), completeness: rec.condition.completeness }, rec.attributes.size)).id;262  const id = `lot_${sha256(rec.sourceUrl).slice(0, 24)}`;263  await db()264    .insert(auctionLots)265    .values({ id, auctionId: `auc_${auctionId}`, assetId: r?.assetId ?? null, variantId, sourceId: rec.sourceId, lotNumber: rec.lotNumber, title: rec.rawTitle, url: rec.sourceUrl, estimateLow: rec.estimateLow, estimateHigh: rec.estimateHigh, currentBid: rec.currentBid, currency: rec.currency, startsAt: rec.startsAt, endsAt: rec.endsAt, status: rec.status === 'unknown' ? 'upcoming' : rec.status, imageUrls: rec.imageUrls, grader: rec.grade.grader, grade: rec.grade.grade })266    .onConflictDoUpdate({ target: auctionLots.url, set: { currentBid: sql`excluded.current_bid`, status: sql`excluded.status`, endsAt: sql`coalesce(excluded.ends_at, ${auctionLots.endsAt})`, assetId: sql`coalesce(${auctionLots.assetId}, excluded.asset_id)`, updatedAt: new Date() } });267  await trackCert(rec, 'auction_lot', id, r?.assetId ?? null, variantId, null, rec.currentBid, rec.currency, rec.observedAt);268  return { assetId: r?.assetId ?? null, variantId, targetId: id, method: r?.method ?? null, confidence: r?.confidence ?? null, status: r ? 'applied' : 'unmatched' };269}270271async function applyPopulation(rec: NormalizedPopulationReport): Promise<ApplyResult> {272  const r = await resolveAsset(rec.attributes, {});273  if (!r) return { assetId: null, variantId: null, targetId: null, method: null, confidence: null, status: 'unmatched' };274  const id = newId('event');275  const inserted = await db()276    .insert(populationReports)277    .values({ id, assetId: r.assetId, grader: rec.grader, sourceId: rec.sourceId, sourceUrl: rec.sourceUrl, reportDate: toDateOnly(rec.reportDate), total: rec.total, byGrade: rec.byGrade })278    .onConflictDoNothing()279    .returning({ id: populationReports.id });280  return { assetId: r.assetId, variantId: null, targetId: inserted[0]?.id ?? null, method: r.method, confidence: r.confidence, status: inserted.length ? 'applied' : 'duplicate' };281}282283async function applyNews(rec: NormalizedNewsItem): Promise<ApplyResult> {284  const id = newId('news');285  const inserted = await db()286    .insert(news)287    .values({ id, sourceId: rec.sourceId, url: rec.sourceUrl, title: rec.title, summary: rec.summary, publishedAt: rec.publishedAt, categorySlugs: rec.categorySlugs, imageUrl: rec.imageUrl, fetchedAt: new Date() })288    .onConflictDoNothing({ target: news.url })289    .returning({ id: news.id });290  return { assetId: null, variantId: null, targetId: inserted[0]?.id ?? null, method: null, confidence: null, status: inserted.length ? 'applied' : 'duplicate' };291}292
Files · d732f70ce5489245a263ca6bcf09fbebc0d2be77
Menu

Theme