SPB Git

spb/rareindex

Public
19.3 KB · 315 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, parseGradeFromTitle } from '@rareindex/taxonomy';4import type { Grade } from '@rareindex/shared';5import { newId, sha256, toDateOnly, type NormalizedAuctionLot, type NormalizedCatalogItem, type NormalizedListing, type NormalizedNewsItem, type NormalizedPopulationReport, type NormalizedPriceObservation, type NormalizedRecord, type NormalizedSale } from '@rareindex/shared';6import { db } from '../lib/db.ts';7import { toUsd } from '../lib/fx.ts';8import { audit } from '../lib/audit.ts';9import { enrichAttributesFromTitle, ensureVariant, resolveAsset, type Resolution } from './resolver.ts';10import { recordCertificate } from './certificates.ts';1112/** Record a slab certification sighting when a record carries one (SPEC §22); never throws into the writer. */13async function trackCert(rec: { grade: { grader: string | null; grade: string | null; qualifier: string | null; certificationNumber: string | null }; sourceId: string; connectorId: string; sourceUrl: string; rawTitle?: string | null }, 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> {14  const g = refineGrade(rec.grade, rec.rawTitle);15  if (!g.certificationNumber || !g.grader) return;16  try {17    await recordCertificate({ grader: g.grader, certNumber: g.certificationNumber, assetId, variantId, grade: g.grade, qualifier: g.qualifier, kind, targetId, sourceId: rec.sourceId, connectorId: rec.connectorId, sourceUrl: rec.sourceUrl, priceUsd, price, currency, observedAt });18  } catch {19    /* cert tracking is best-effort; the canonical write already succeeded */20  }21}2223export class FxMissingError extends Error {24  constructor(public currency: string, public date: Date) {25    super(`fx_missing:${currency}:${toDateOnly(date)}`);26  }27}2829export interface ApplyResult {30  assetId: string | null;31  variantId: string | null;32  targetId: string | null;33  method: Resolution['method'] | null;34  confidence: number | null;35  status: 'applied' | 'unmatched' | 'rejected' | 'duplicate';36  reason?: string;37  event?: 'sale_detected' | 'listing_created' | 'listing_updated' | 'entity_created';38}3940const trustCache = new Map<string, number>();41async function sourceTrust(sourceId: string): Promise<number> {42  const hit = trustCache.get(sourceId);43  if (hit !== undefined) return hit;44  const [row] = await db().select({ t: sources.trustScore }).from(sources).where(eq(sources.id, sourceId)).limit(1);45  const t = row ? Number(row.t) : 0.5;46  trustCache.set(sourceId, t);47  return t;48}4950/** Data quality 0–100 (§150): completeness × source trust × identification confidence. */51function dataQuality(fields: Record<string, unknown>, trust: number, confidence: number, extra = 0): number {52  const keys = Object.keys(fields);53  const present = keys.filter((k) => fields[k] !== null && fields[k] !== undefined && fields[k] !== '' && !(Array.isArray(fields[k]) && (fields[k] as unknown[]).length === 0)).length;54  const completeness = keys.length ? present / keys.length : 0;55  return Math.round(Math.max(0, Math.min(100, (0.4 * completeness + 0.3 * trust + 0.3 * confidence + extra) * 100)));56}5758function normCond(categorySlug: string, raw: string | null | undefined, already: string | null | undefined): string | null {59  return already ?? normalizeCondition(categorySlug, raw);60}6162async function storeImages(assetId: string, urls: string[], sourceId: string, ref: { listingId?: string; saleId?: string }): Promise<void> {63  if (!urls.length) return;64  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' }));65  await db().insert(images).values(rows).onConflictDoNothing({ target: images.url });66}6768export async function applyRecord(rec: NormalizedRecord): Promise<ApplyResult> {69  switch (rec.kind) {70    case 'catalog_item':71      return applyCatalog(rec);72    case 'sale':73      return applySale(rec);74    case 'listing':75      return applyListing(rec);76    case 'price_observation':77      return applyObservation(rec);78    case 'auction_lot':79      return applyAuctionLot(rec);80    case 'population_report':81      return applyPopulation(rec);82    case 'news_item':83      return applyNews(rec);84  }85}8687/**88 * Grade refinement (§18, §84): when a connector did not read the slab (or read the grader but not the89 * grade), parse the raw title. A grader with no readable grade stays "grader · grade unknown" — it must90 * never fall into the raw/ungraded variant, which is what compared PSA slabs against raw valuations.91 * An explicit `raw` from the connector is trusted over the title.92 */93export function refineGrade(grade: Grade, rawTitle: string | null | undefined): Grade {94  if (!rawTitle || grade.grader === 'raw' || (grade.grader && grade.grade)) return grade;95  const parsed = parseGradeFromTitle(rawTitle);96  if (!parsed.grader || parsed.grader === 'raw') return grade;97  if (grade.grader && parsed.grader !== grade.grader) return grade; // never contradict the connector's grader98  // A grade recorded without a grader is a condition-scale value (e.g. a price-guide tab), not a slab99  // grade: when the title reveals the slab, the title's grade wins.100  const g = grade.grader ? (grade.grade ?? parsed.grade) : parsed.grade;101  return { ...grade, grader: grade.grader ?? parsed.grader, grade: g, qualifier: grade.qualifier ?? parsed.qualifier };102}103104async function applyCatalog(rec: NormalizedCatalogItem): Promise<ApplyResult> {105  const r = await resolveAsset(rec.attributes, { enrich: true, imageUrls: rec.imageUrls, description: rec.description });106  if (!r) return { assetId: null, variantId: null, targetId: null, method: null, confidence: null, status: 'unmatched' };107  if (rec.imageUrls.length) await storeImages(r.assetId, rec.imageUrls.slice(0, 1), rec.sourceId, {});108  return { assetId: r.assetId, variantId: null, targetId: r.assetId, method: r.method, confidence: r.confidence, status: 'applied', event: r.created ? 'entity_created' : undefined };109}110111async function applySale(rec: NormalizedSale): Promise<ApplyResult> {112  const grade = refineGrade(rec.grade, rec.rawTitle);113  const r = await resolveAsset(enrichAttributesFromTitle(rec.attributes, rec.rawTitle), { imageUrls: rec.imageUrls });114  if (!r) return { assetId: null, variantId: null, targetId: null, method: null, confidence: null, status: 'unmatched' };115  const condition = normCond(rec.attributes.categorySlug, rec.condition.conditionRaw, rec.condition.condition);116  const variant = await ensureVariant(r.assetId, grade, { condition, completeness: rec.condition.completeness }, rec.attributes.size);117  const fx = await toUsd(rec.price, rec.currency, rec.saleDate);118  if (!fx) throw new FxMissingError(rec.currency, rec.saleDate);119  const dedupeKey = sha256(['sale', rec.sourceId, rec.externalId ?? rec.sourceUrl, toDateOnly(rec.saleDate), rec.price.toFixed(2)].join('|'));120  const trust = await sourceTrust(rec.sourceId);121  const confidence = Math.min(rec.confidence, r.confidence);122  const flags: string[] = [];123  if (rec.isBundle || rec.quantity > 1) flags.push('bundle');124  if (rec.price <= 0) flags.push('zero_price');125  if (confidence < 0.6) flags.push('low_identification_confidence');126  const status = flags.includes('bundle') || flags.includes('zero_price') ? 'excluded' : 'valid';127  const id = newId('sale');128  const inserted = await db()129    .insert(sales)130    .values({131      id,132      assetId: r.assetId,133      variantId: variant.id,134      sourceId: rec.sourceId,135      connectorId: rec.connectorId,136      sourceUrl: rec.sourceUrl,137      externalId: rec.externalId,138      saleType: rec.saleType,139      saleDate: rec.saleDate,140      price: rec.price,141      currency: rec.currency,142      priceUsd: fx.usd,143      fxRate: fx.rate,144      fxDate: fx.fxDate,145      buyerPremiumIncluded: rec.buyerPremiumIncluded,146      quantity: rec.quantity,147      isBundle: rec.isBundle,148      condition,149      grader: grade.grader,150      grade: grade.grade,151      certificationNumber: grade.certificationNumber,152      location: rec.location,153      auctionHouse: rec.auctionHouse,154      lotNumber: rec.lotNumber,155      imageUrls: rec.imageUrls,156      rawTitle: rec.rawTitle,157      confidence,158      dataQuality: dataQuality({ date: rec.saleDate, price: rec.price, currency: rec.currency, images: rec.imageUrls, grade: grade.grade, condition, externalId: rec.externalId, type: rec.saleType === 'unknown' ? null : rec.saleType }, trust, confidence),159      status,160      flags,161      dedupeKey,162    })163    .onConflictDoNothing({ target: sales.dedupeKey })164    .returning({ id: sales.id });165  if (inserted.length === 0) return { assetId: r.assetId, variantId: variant.id, targetId: null, method: r.method, confidence, status: 'duplicate' };166  if (status === 'excluded') await audit({ entityType: 'sale', entityId: id, action: 'excluded', reason: flags.join(','), details: { rawTitle: rec.rawTitle, quantity: rec.quantity } });167  if (rec.imageUrls.length) await storeImages(r.assetId, rec.imageUrls.slice(0, 3), rec.sourceId, { saleId: id });168  await trackCert(rec, 'sale', id, r.assetId, variant.id, fx.usd, rec.price, rec.currency, rec.saleDate);169  return { assetId: r.assetId, variantId: variant.id, targetId: id, method: r.method, confidence, status: 'applied', event: 'sale_detected' };170}171172async function applyListing(rec: NormalizedListing): Promise<ApplyResult> {173  const grade = refineGrade(rec.grade, rec.rawTitle);174  const r = await resolveAsset(enrichAttributesFromTitle(rec.attributes, rec.rawTitle), { imageUrls: rec.imageUrls });175  if (!r) return { assetId: null, variantId: null, targetId: null, method: null, confidence: null, status: 'unmatched' };176  const condition = normCond(rec.attributes.categorySlug, rec.condition.conditionRaw, rec.condition.condition);177  const variant = await ensureVariant(r.assetId, grade, { condition, completeness: rec.condition.completeness }, rec.attributes.size);178  let priceUsd: number | null = null;179  if (rec.price !== null && rec.currency) {180    const fx = await toUsd(rec.price, rec.currency, rec.observedAt);181    if (!fx) throw new FxMissingError(rec.currency, rec.observedAt);182    priceUsd = fx.usd;183  }184  const externalId = rec.externalId ?? sha256(rec.sourceUrl);185  const trust = await sourceTrust(rec.sourceId);186  const confidence = Math.min(rec.confidence, r.confidence);187  const now = rec.observedAt;188  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);189  const quality = dataQuality({ price: rec.price, currency: rec.currency, images: rec.imageUrls, seller: rec.seller, condition, grade: grade.grade, listedAt: rec.listedAt, description: rec.description }, trust, confidence);190  if (!existing) {191    const id = newId('listing');192    await db().insert(listings).values({193      id,194      assetId: r.assetId,195      variantId: variant.id,196      sourceId: rec.sourceId,197      connectorId: rec.connectorId,198      sourceUrl: rec.sourceUrl,199      externalId,200      listingType: rec.listingType,201      price: rec.price,202      currency: rec.currency,203      priceUsd,204      seller: rec.seller,205      sellerReputation: rec.sellerReputation,206      location: rec.location,207      shippingCost: rec.shippingCost,208      quantity: rec.quantity,209      condition,210      grader: grade.grader,211      grade: grade.grade,212      certificationNumber: grade.certificationNumber,213      imageUrls: rec.imageUrls,214      rawTitle: rec.rawTitle,215      description: rec.description,216      listedAt: rec.listedAt,217      endsAt: rec.endsAt,218      availability: rec.availability,219      bidCount: rec.bidCount,220      firstSeenAt: now,221      lastSeenAt: now,222      confidence,223      dataQuality: quality,224      flags: confidence < 0.6 ? ['low_identification_confidence'] : [],225    });226    await db().insert(listingEvents).values({ id: newId('event'), listingId: id, eventType: 'new', newPrice: rec.price, currency: rec.currency, occurredAt: now });227    if (rec.imageUrls.length) await storeImages(r.assetId, rec.imageUrls.slice(0, 3), rec.sourceId, { listingId: id });228    await trackCert(rec, 'listing', id, r.assetId, variant.id, priceUsd, rec.price, rec.currency, now);229    return { assetId: r.assetId, variantId: variant.id, targetId: id, method: r.method, confidence, status: 'applied', event: 'listing_created' };230  }231  const priceChanged = rec.price !== null && existing.price !== null && Math.abs(Number(existing.price) - rec.price) > 0.009;232  const relisted = existing.availability !== 'available' && rec.availability === 'available';233  const ended = existing.availability === 'available' && rec.availability !== 'available';234  await db()235    .update(listings)236    .set({237      price: rec.price,238      currency: rec.currency,239      priceUsd,240      availability: rec.availability,241      bidCount: rec.bidCount,242      endsAt: rec.endsAt ?? sql`${listings.endsAt}`,243      lastSeenAt: now,244      priceChangedAt: priceChanged ? now : sql`${listings.priceChangedAt}`,245      quantity: rec.quantity,246      dataQuality: quality,247      updatedAt: new Date(),248    })249    .where(eq(listings.id, existing.id));250  const ev = priceChanged ? 'price_changed' : relisted ? 'relisted' : ended ? (rec.availability === 'sold' ? 'sold' : rec.availability === 'ended' ? 'auction_ended' : 'removed') : null;251  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 });252  return { assetId: r.assetId, variantId: variant.id, targetId: existing.id, method: r.method, confidence, status: 'applied', event: ev ? 'listing_updated' : undefined };253}254255async function applyObservation(rec: NormalizedPriceObservation): Promise<ApplyResult> {256  const grade = refineGrade(rec.grade, rec.rawTitle);257  const r = await resolveAsset(rec.attributes, { imageUrls: rec.imageUrls });258  if (!r) return { assetId: null, variantId: null, targetId: null, method: null, confidence: null, status: 'unmatched' };259  const condition = normCond(rec.attributes.categorySlug, rec.condition.conditionRaw, rec.condition.condition);260  const variant = await ensureVariant(r.assetId, grade, { condition, completeness: rec.condition.completeness }, rec.attributes.size);261  const fx = await toUsd(rec.price, rec.currency, rec.observationDate);262  if (!fx) throw new FxMissingError(rec.currency, rec.observationDate);263  const dedupeKey = sha256(['obs', rec.sourceId, rec.externalId ?? rec.sourceUrl, rec.priceKind, variant.key, toDateOnly(rec.observationDate)].join('|'));264  const id = newId('valuation');265  const inserted = await db()266    .insert(priceObservations)267    .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 })268    .onConflictDoNothing({ target: priceObservations.dedupeKey })269    .returning({ id: priceObservations.id });270  if (inserted.length === 0) return { assetId: r.assetId, variantId: variant.id, targetId: null, method: r.method, confidence: r.confidence, status: 'duplicate' };271  return { assetId: r.assetId, variantId: variant.id, targetId: id, method: r.method, confidence: r.confidence, status: 'applied' };272}273274async function applyAuctionLot(rec: NormalizedAuctionLot): Promise<ApplyResult> {275  const grade = refineGrade(rec.grade, rec.rawTitle);276  const r = await resolveAsset(enrichAttributesFromTitle(rec.attributes, rec.rawTitle), { imageUrls: rec.imageUrls, allowCreate: rec.confidence >= 0.7 });277  const auctionUrl = (rec.attributes.metadata?.auctionUrl as string | undefined) ?? `${new URL(rec.sourceUrl).origin}#${rec.auctionHouse}:${rec.auctionName ?? 'auction'}`;278  const auctionId = sha256(auctionUrl).slice(0, 24);279  await db()280    .insert(auctions)281    .values({ id: `auc_${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 })282    .onConflictDoUpdate({ target: auctions.url, set: { endsAt: sql`coalesce(excluded.ends_at, ${auctions.endsAt})`, status: sql`excluded.status`, updatedAt: new Date() } });283  let variantId: string | null = null;284  if (r) variantId = (await ensureVariant(r.assetId, grade, { condition: normCond(rec.attributes.categorySlug, rec.condition.conditionRaw, rec.condition.condition), completeness: rec.condition.completeness }, rec.attributes.size)).id;285  const id = `lot_${sha256(rec.sourceUrl).slice(0, 24)}`;286  await db()287    .insert(auctionLots)288    .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: grade.grader, grade: grade.grade })289    .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() } });290  await trackCert(rec, 'auction_lot', id, r?.assetId ?? null, variantId, null, rec.currentBid, rec.currency, rec.observedAt);291  return { assetId: r?.assetId ?? null, variantId, targetId: id, method: r?.method ?? null, confidence: r?.confidence ?? null, status: r ? 'applied' : 'unmatched' };292}293294async function applyPopulation(rec: NormalizedPopulationReport): Promise<ApplyResult> {295  const r = await resolveAsset(rec.attributes, {});296  if (!r) return { assetId: null, variantId: null, targetId: null, method: null, confidence: null, status: 'unmatched' };297  const id = newId('event');298  const inserted = await db()299    .insert(populationReports)300    .values({ id, assetId: r.assetId, grader: rec.grader, sourceId: rec.sourceId, sourceUrl: rec.sourceUrl, reportDate: toDateOnly(rec.reportDate), total: rec.total, byGrade: rec.byGrade })301    .onConflictDoNothing()302    .returning({ id: populationReports.id });303  return { assetId: r.assetId, variantId: null, targetId: inserted[0]?.id ?? null, method: r.method, confidence: r.confidence, status: inserted.length ? 'applied' : 'duplicate' };304}305306async function applyNews(rec: NormalizedNewsItem): Promise<ApplyResult> {307  const id = newId('news');308  const inserted = await db()309    .insert(news)310    .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() })311    .onConflictDoNothing({ target: news.url })312    .returning({ id: news.id });313  return { assetId: null, variantId: null, targetId: inserted[0]?.id ?? null, method: null, confidence: null, status: inserted.length ? 'applied' : 'duplicate' };314}315
Files · a1e47ef1fdeca56390a87b1194cd8e95c9d7500f
Menu

Theme