@gscdump/engine-duckdb-node

Node DuckDB engine adapter — wraps the parquet storage engine as a SqlQuerySource and ships parquet/snapshot attach helpers for the browser-style attached-table path.

Install

npm install @gscdump/engine-duckdb-node @gscdump/engine

Usage

import { ROW_ANALYZERS, SQL_ANALYZERS } from '@gscdump/analysis'
import { createEngine } from '@gscdump/engine-duckdb-node'
import { createAnalyzerRegistry, runAnalyzerFromSource } from '@gscdump/engine/analyzer'

const source = createEngine({ engine, ctx })
const registry = createAnalyzerRegistry({ rows: ROW_ANALYZERS, sql: SQL_ANALYZERS })

const result = await runAnalyzerFromSource(
  source,
  { type: 'striking-distance', minImpressions: 100 },
  registry,
)

Attaching parquet for the browser-style path

import { analyzeInBrowser } from '@gscdump/analysis'
import { attachParquetIndex, attachSnapshotIndex } from '@gscdump/engine-duckdb-node'

await attachParquetIndex(conn, { files }) // per-day or per-month parquet
await attachSnapshotIndex(conn, { snapshot }) // pre-baked .duckdb snapshot

const result = await analyzeInBrowser(
  { query: (sql, params) => conn.runAndReadAll(sql, params).then(r => r.getRowObjects()) },
  { schema: 'gsc' },
  { type: 'striking-distance' },
)

Exports

• createEngine({ engine, ctx }) — builds a SqlQuerySource from a StorageEngine + TenantCtx (delegates to createEngineQuerySource from @gscdump/engine/source).
• attachParquetIndex / attachSnapshotIndex / snapshotAlias — DuckDB session attach helpers.

SQL_ANALYZERS, analyzeInBrowser, and rewriteForTableSource moved to @gscdump/analysis (they reference the analyzer instances and don't belong on the engine boundary).

Related

• @gscdump/engine — Storage engine + analyzer/source contracts.
• @gscdump/analysis — Analyzer instances + analyzeInBrowser + SQL_ANALYZERS.
• @gscdump/engine-duckdb-wasm — Browser counterpart (DuckDB-WASM).
• @gscdump/engine-sqlite — SQLite / D1 counterpart.

License

MIT