deckhearth/lib/card-import/bulk-sync.js

274 lines
8.6 KiB
JavaScript
Raw Normal View History

/**
* Bulk catalog sync using Scryfall's bulk-data endpoint.
*
* Downloads the Oracle Cards file (~168 MB) and upserts all MTG cards in one pass.
* Much faster and more complete than the per-set discovery approach — catches all
* missing sets in one shot without rate-limit concerns.
*
* Also refreshes prices, legalities, and edhrec_rank on existing cards.
*/
import { gunzipSync } from 'node:zlib';
import { sql } from '../sql.js';
import { recordPriceSnapshots } from '../pricing/price-history.js';
const BATCH_SIZE = 100;
const delay = (ms) => new Promise((resolve) => setTimeout(resolve, ms));
export function resolveBulkDownloadUri(meta) {
return meta?.download_uri ?? meta?.jsonl_download_uri ?? null;
}
async function fetchBulkDownloadUrl() {
const response = await fetch('https://api.scryfall.com/bulk-data/oracle_cards', {
headers: { 'User-Agent': 'DeckHearth/1.0', Accept: 'application/json' },
});
if (!response.ok) {
throw new Error(`Scryfall bulk-data API error: ${response.status}`);
}
return response.json();
}
/**
* Download and parse Scryfall oracle bulk (JSON array or JSONL, optionally gzip).
*/
export async function downloadBulkOracleCards(meta) {
const downloadUri = resolveBulkDownloadUri(meta);
if (!downloadUri) {
throw new Error('Scryfall bulk meta missing download_uri and jsonl_download_uri');
}
const response = await fetch(downloadUri, {
headers: { 'User-Agent': 'DeckHearth/1.0' },
});
if (!response.ok) {
throw new Error(`Bulk download failed: ${response.status}`);
}
const raw = Buffer.from(await response.arrayBuffer());
const text = downloadUri.endsWith('.gz')
? gunzipSync(raw).toString('utf8')
: raw.toString('utf8');
if (downloadUri.includes('.jsonl')) {
return text
.split('\n')
.map((line) => line.trim())
.filter(Boolean)
.map((line) => JSON.parse(line));
}
return JSON.parse(text);
}
export function mapCard(card) {
let currentPrice = null;
let marketPrice = null;
if (card.prices) {
currentPrice = card.prices.usd ? parseFloat(card.prices.usd) : null;
marketPrice = card.prices.usd_foil ? parseFloat(card.prices.usd_foil) : null;
}
return {
scryfallId: card.id,
oracleId: card.oracle_id || null,
illustrationId: card.illustration_id || null,
name: card.name,
setName: card.set_name,
setCode: card.set,
cardNumber: card.collector_number,
rarity: card.rarity,
manaCost: card.mana_cost || null,
cmc: card.cmc != null ? Math.round(card.cmc) : null,
cardType: card.type_line || null,
colors: JSON.stringify(card.colors || []),
colorIdentity: JSON.stringify(card.color_identity || []),
oracleText: card.oracle_text || null,
power: card.power || null,
toughness: card.toughness || null,
imageUrl: card.image_uris?.normal || null,
stockImageUrl: card.image_uris?.art_crop || null,
currentPrice,
marketPrice,
keywords: JSON.stringify(card.keywords || []),
legalities: JSON.stringify(card.legalities || {}),
flavorText: card.flavor_text || null,
artist: card.artist || null,
releasedAt: card.released_at || null,
layout: card.layout || null,
edhrecRank: card.edhrec_rank || null,
reserved: card.reserved || false,
reprint: card.reprint || false,
finishes: JSON.stringify(card.finishes || []),
};
}
function hasPrice(row) {
return row.currentPrice != null || row.marketPrice != null;
}
async function resolveCardIds(scryfallIds) {
if (scryfallIds.length === 0) return new Map();
const result = await sql.query(
'SELECT id, scryfall_id FROM cards WHERE scryfall_id = ANY($1)',
[scryfallIds]
);
return new Map(result.rows.map((row) => [row.scryfall_id, row.id]));
}
async function recordBatchSnapshots(cards) {
const pricedCards = cards.filter(hasPrice);
if (pricedCards.length === 0) return 0;
const idByScryfall = await resolveCardIds(pricedCards.map((row) => row.scryfallId));
const snapshots = [];
for (const row of pricedCards) {
const cardId = idByScryfall.get(row.scryfallId);
if (cardId == null) continue;
if (row.currentPrice != null) {
snapshots.push({
cardId,
source: 'scryfall',
priceType: 'market',
currency: 'USD',
amount: row.currentPrice,
});
}
if (row.marketPrice != null) {
snapshots.push({
cardId,
source: 'scryfall',
priceType: 'foil',
currency: 'USD',
amount: row.marketPrice,
});
}
}
return recordPriceSnapshots(snapshots);
}
async function upsertBatch(cards) {
if (cards.length === 0) return { upserted: 0, snapshotsWritten: 0 };
const columns = [
'scryfall_id', 'oracle_id', 'illustration_id', 'name', 'set_name', 'set_code',
'card_number', 'rarity', 'game', 'mana_cost', 'cmc', 'card_type', 'colors',
'color_identity', 'oracle_text', 'power', 'toughness', 'image_url', 'stock_image_url',
'current_price', 'market_price', 'prices_updated_at', 'keywords', 'legalities', 'flavor_text', 'artist',
'released_at', 'layout', 'edhrec_rank', 'reserved', 'reprint', 'finishes', 'verified',
];
const values = [];
const placeholders = [];
let paramIdx = 1;
for (const row of cards) {
const pricesUpdatedAt = hasPrice(row) ? new Date() : null;
const rowValues = [
row.scryfallId, row.oracleId, row.illustrationId, row.name, row.setName, row.setCode,
row.cardNumber, row.rarity, 'MTG', row.manaCost, row.cmc, row.cardType, row.colors,
row.colorIdentity, row.oracleText, row.power, row.toughness, row.imageUrl, row.stockImageUrl,
row.currentPrice, row.marketPrice, pricesUpdatedAt, row.keywords, row.legalities, row.flavorText, row.artist,
row.releasedAt, row.layout, row.edhrecRank, row.reserved, row.reprint, row.finishes, true,
];
const rowP = rowValues.map(() => `$${paramIdx++}`);
placeholders.push(`(${rowP.join(',')})`);
values.push(...rowValues);
}
const updateCols = columns
.filter((c) => c !== 'scryfall_id')
.map((c) => `${c} = EXCLUDED.${c}`)
.join(', ');
await sql.query(
`INSERT INTO cards (${columns.join(', ')})
VALUES ${placeholders.join(',\n')}
ON CONFLICT (scryfall_id) DO UPDATE SET ${updateCols}, updated_at = CURRENT_TIMESTAMP`,
values
);
const snapshotsWritten = await recordBatchSnapshots(cards);
return { upserted: cards.length, snapshotsWritten };
}
/**
* Run a full bulk MTG sync. Returns a summary compatible with the catalog sync API shape.
*/
export async function runBulkMtgSync() {
const meta = await fetchBulkDownloadUrl();
const sizeMb = ((meta.compressed_size ?? meta.size ?? 0) / 1024 / 1024).toFixed(0);
console.log(`[bulk-sync] Oracle Cards updated: ${meta.updated_at}, size: ~${sizeMb} MB`);
const allCards = await downloadBulkOracleCards(meta);
const mtgCards = allCards.filter((c) => c.lang === 'en' && !c.digital);
let upserted = 0;
let snapshotsWritten = 0;
let errors = 0;
let batch = [];
for (const card of mtgCards) {
batch.push(mapCard(card));
if (batch.length >= BATCH_SIZE) {
try {
const result = await upsertBatch(batch);
upserted += result.upserted;
snapshotsWritten += result.snapshotsWritten;
} catch (err) {
console.error('[bulk-sync] Batch error:', err.message);
errors += batch.length;
}
batch = [];
}
}
if (batch.length > 0) {
try {
const result = await upsertBatch(batch);
upserted += result.upserted;
snapshotsWritten += result.snapshotsWritten;
} catch (err) {
console.error('[bulk-sync] Final batch error:', err.message);
errors += batch.length;
}
}
// Backfill oracle_id on older per-set imported cards
const backfill = await sql`
UPDATE cards target
SET oracle_id = source.oracle_id,
color_identity = COALESCE(target.color_identity, source.color_identity),
keywords = COALESCE(target.keywords, source.keywords),
legalities = COALESCE(target.legalities, source.legalities),
layout = COALESCE(target.layout, source.layout),
edhrec_rank = COALESCE(target.edhrec_rank, source.edhrec_rank)
FROM (
SELECT DISTINCT ON (name) name, oracle_id, color_identity, keywords, legalities, layout, edhrec_rank
FROM cards WHERE oracle_id IS NOT NULL AND game = 'MTG'
) source
WHERE target.name = source.name AND target.game = 'MTG' AND target.oracle_id IS NULL
`;
const summary = {
mode: 'bulk',
totalInFile: mtgCards.length,
upserted,
snapshotsWritten,
errors,
backfilledOracleId: backfill.rowCount || 0,
updatedAt: meta.updated_at,
};
console.log('[bulk-sync]', JSON.stringify(summary));
return summary;
}