@gscdump/engine 1.4.0 → 1.4.1

package/dist/adapters/duckdb-node.d.mts

@@ -0,0 +1,7 @@
+import { DuckDBHandle } from "../duckdb.mjs";
+interface NodeDuckDBOptions {
+  verbose?: boolean;
+}
+declare function createNodeDuckDBHandle(opts?: NodeDuckDBOptions): DuckDBHandle;
+declare function resetNodeDuckDB(): void;
+export { NodeDuckDBOptions, createNodeDuckDBHandle, resetNodeDuckDB };

package/dist/adapters/duckdb-node.mjs

@@ -0,0 +1,94 @@
+import { arrowToRows } from "../arrow-utils.mjs";
+import { createRequire } from "node:module";
+import process from "node:process";
+import { unlinkSync } from "node:fs";
+import { tmpdir } from "node:os";
+import { join } from "node:path";
+import { fileURLToPath } from "node:url";
+import { ConsoleLogger, NODE_RUNTIME, VoidLogger, createDuckDB } from "@duckdb/duckdb-wasm/dist/duckdb-node-blocking.cjs";
+const require_ = createRequire(typeof __filename !== "undefined" ? __filename : typeof import.meta !== "undefined" ? fileURLToPath(import.meta.url) : process.cwd());
+let singleton = null;
+let singletonOpts = null;
+function bundles() {
+	return {
+		mvp: {
+			mainModule: require_.resolve("@duckdb/duckdb-wasm/dist/duckdb-mvp.wasm"),
+			mainWorker: null
+		},
+		eh: {
+			mainModule: require_.resolve("@duckdb/duckdb-wasm/dist/duckdb-eh.wasm"),
+			mainWorker: null
+		}
+	};
+}
+async function initialize(opts) {
+	const logger = opts.verbose ? new ConsoleLogger() : new VoidLogger();
+	const db = await createDuckDB(bundles(), logger, NODE_RUNTIME);
+	await db.instantiate();
+	return {
+		db,
+		conn: db.connect()
+	};
+}
+function getSingleton(opts) {
+	if (!singleton) {
+		singleton = initialize(opts);
+		singletonOpts = opts;
+	}
+	return singleton;
+}
+function createNodeDuckDBHandle(opts = {}) {
+	if (singleton && opts.verbose !== void 0 && opts.verbose !== (singletonOpts?.verbose ?? false)) console.warn(`[gscdump] createNodeDuckDBHandle: ignoring verbose=${opts.verbose} — a shared DuckDB instance was already initialized with verbose=${singletonOpts?.verbose ?? false}. Call resetNodeDuckDB() before re-initializing to change it.`);
+	getSingleton(opts);
+	return {
+		async query(sql, params) {
+			const { conn } = await getSingleton(opts);
+			if (!params || params.length === 0) return arrowToRows(conn.query(sql));
+			const stmt = conn.prepare(sql);
+			try {
+				return arrowToRows(stmt.query(...params));
+			} finally {
+				stmt.close();
+			}
+		},
+		async registerFileBuffer(name, bytes) {
+			const { db } = await getSingleton(opts);
+			db.registerFileBuffer(name, bytes);
+		},
+		async copyFileToBuffer(name) {
+			const { db } = await getSingleton(opts);
+			return db.copyFileToBuffer(name);
+		},
+		async dropFiles(names) {
+			const { db } = await getSingleton(opts);
+			for (const name of names) {
+				try {
+					db.dropFile(name);
+				} catch (error) {
+					const message = error instanceof Error ? error.message : String(error);
+					if (!/not found|does not exist|unknown file/i.test(message)) throw error;
+				}
+				try {
+					unlinkSync(name);
+				} catch (error) {
+					if (error.code !== "ENOENT") throw error;
+				}
+			}
+		},
+		makeTempPath(ext) {
+			return join(tmpdir(), `gscdump-${Math.random().toString(36).slice(2, 10)}.${ext}`);
+		}
+	};
+}
+function resetNodeDuckDB() {
+	const pending = singleton;
+	singleton = null;
+	singletonOpts = null;
+	pending?.then(({ db, conn }) => {
+		conn.close();
+		db.reset();
+	}).catch((err) => {
+		console.warn("[gscdump] resetNodeDuckDB: failed to release DuckDB instance", err);
+	});
+}
+export { createNodeDuckDBHandle, resetNodeDuckDB };

package/dist/adapters/filesystem.d.mts

@@ -1,4 +1,4 @@
-import { DataSource, ManifestStore } from "../_chunks/storage.mjs";
+import { DataSource, ManifestStore } from "../storage.mjs";
 interface FilesystemDataSourceOptions {
   rootDir: string;
 }

package/dist/adapters/filesystem.mjs

@@ -1,4 +1,4 @@
-import { manifestEntryKey, matchesManifestEntryFilter, matchesSyncStateFilter, matchesWatermarkFilter, mergeSyncState, syncStateKey, watermarkKey } from "../_chunks/manifest-store-utils.mjs";
+import { manifestEntryKey, matchesManifestEntryFilter, matchesSyncStateFilter, matchesWatermarkFilter, mergeSyncState, syncStateKey, watermarkKey } from "../manifest-store-utils.mjs";
 import { dirname, join, resolve } from "node:path";
 import { Buffer } from "node:buffer";
 import { randomBytes } from "node:crypto";

package/dist/adapters/hyparquet.d.mts

@@ -1,6 +1,7 @@
-import { CodecCtx, DataSource, ParquetCodec, Row, TableName } from "../_chunks/storage.mjs";
-import { ParquetQueryFilter } from "../_chunks/libs/hyparquet.mjs";
-import { ColumnDef } from "../_chunks/schema.mjs";
+import { ParquetQueryFilter } from "../node_modules/.pnpm/hyparquet@1.26.2/node_modules/hyparquet/types/index.mjs";
+import "../node_modules/.pnpm/hyparquet@1.26.2/node_modules/hyparquet/types/node.mjs";
+import { CodecCtx, DataSource, ParquetCodec, Row, TableName } from "../storage.mjs";
+import { ColumnDef } from "../schema.mjs";
 declare function encodeRowsToParquet(table: TableName, rows: readonly Row[]): Uint8Array;
 interface EncodeFlexOptions {
   /** Columns defining the output schema + order. */

package/dist/adapters/hyparquet.mjs

@@ -1,6 +1,10 @@
-import { SCHEMAS, TABLE_METADATA } from "../_chunks/schema.mjs";
-import { parquetReadObjects } from "../_chunks/libs/hyparquet.mjs";
-import { ByteWriter, parquetWriteRows } from "../_chunks/libs/hyparquet-writer.mjs";
+import { TABLE_METADATA } from "../drizzle-schema.mjs";
+import { SCHEMAS } from "../schema.mjs";
+import { parquetReadObjects } from "../node_modules/.pnpm/hyparquet@1.26.2/node_modules/hyparquet/src/read.mjs";
+import "../node_modules/.pnpm/hyparquet@1.26.2/node_modules/hyparquet/src/node.mjs";
+import { ByteWriter } from "../node_modules/.pnpm/hyparquet-writer@0.16.1_patch_hash_7a5e451ac7634d7546ac56bcc5eadacf7715a6241640c2b5a5bc0cfc9763b8ec/node_modules/hyparquet-writer/src/bytewriter.mjs";
+import { parquetWriteRows } from "../node_modules/.pnpm/hyparquet-writer@0.16.1_patch_hash_7a5e451ac7634d7546ac56bcc5eadacf7715a6241640c2b5a5bc0cfc9763b8ec/node_modules/hyparquet-writer/src/write-rows.mjs";
+import "../node_modules/.pnpm/hyparquet-writer@0.16.1_patch_hash_7a5e451ac7634d7546ac56bcc5eadacf7715a6241640c2b5a5bc0cfc9763b8ec/node_modules/hyparquet-writer/src/node.mjs";
 const ROW_GROUP_SIZE = 25e3;
 const DEFAULT_COMPACTION_READ_CONCURRENCY = 8;
 const MAX_COMPACTION_READ_CONCURRENCY = 32;

package/dist/adapters/node-harness.d.mts

@@ -0,0 +1,38 @@
+import { DataSource, StorageEngine } from "../storage.mjs";
+import { SearchType } from "gscdump/query";
+import { Row, TableName } from "@gscdump/contracts";
+interface NodeHarnessOptions {
+  dataDir: string;
+  /** Tenant user id. Defaults to `'local'` for single-user CLI installs. */
+  userId?: string;
+  /** Name of the manifest file under `dataDir`. Defaults to `manifest.json`. */
+  manifestFilename?: string;
+}
+interface NodeHarness {
+  engine: StorageEngine;
+  /**
+   * Underlying filesystem-backed DataSource. Exposed so commands that write
+   *  derivative artifacts (rollups, exports) don't have to re-instantiate it.
+   */
+  dataSource: DataSource;
+  dataDir: string;
+  userId: string;
+  siteIdFor: (siteUrl: string) => string;
+  runRawSql: (opts: {
+    sql: string;
+    siteUrl: string;
+    table: TableName;
+    params?: unknown[];
+    /**
+     * Restrict the underlying manifest lookup to a single GSC search-type
+     * slice. Undefined keeps the legacy cross-type union.
+     */
+    searchType?: SearchType;
+  }) => Promise<{
+    rows: Row[];
+    sql: string;
+    keys: string[];
+  }>;
+}
+declare function createNodeHarness(opts: NodeHarnessOptions): NodeHarness;
+export { NodeHarness, NodeHarnessOptions, createNodeHarness };

package/dist/adapters/node-harness.mjs

@@ -0,0 +1,47 @@
+import { createDuckDBCodec, createDuckDBExecutor } from "../duckdb.mjs";
+import { createStorageEngine } from "../engine.mjs";
+import { createNodeDuckDBHandle } from "./duckdb-node.mjs";
+import { createFilesystemDataSource, createFilesystemManifestStore } from "./filesystem.mjs";
+import path from "node:path";
+import { encodeSiteId } from "gscdump/tenant";
+function createNodeHarness(opts) {
+	const dataDir = opts.dataDir;
+	const userId = opts.userId ?? "local";
+	const manifestFilename = opts.manifestFilename ?? "manifest.json";
+	const handle = createNodeDuckDBHandle();
+	const factory = { getDuckDB: async () => handle };
+	const dataSource = createFilesystemDataSource({ rootDir: dataDir });
+	const engine = createStorageEngine({
+		dataSource,
+		manifestStore: createFilesystemManifestStore({ path: path.join(dataDir, manifestFilename) }),
+		codec: createDuckDBCodec(factory),
+		executor: createDuckDBExecutor(factory)
+	});
+	async function runRawSql(runOpts) {
+		const result = await engine.runSQL({
+			ctx: {
+				userId,
+				siteId: encodeSiteId(runOpts.siteUrl)
+			},
+			table: runOpts.table,
+			fileSets: { FILES: { table: runOpts.table } },
+			sql: runOpts.sql,
+			params: runOpts.params ?? [],
+			...runOpts.searchType !== void 0 ? { searchType: runOpts.searchType } : {}
+		});
+		return {
+			rows: result.rows,
+			sql: result.sql,
+			keys: result.objectKeys
+		};
+	}
+	return {
+		engine,
+		dataSource,
+		dataDir,
+		userId,
+		siteIdFor: encodeSiteId,
+		runRawSql
+	};
+}
+export { createNodeHarness };

package/dist/adapters/node.d.mts

@@ -1,101 +1,5 @@
-import { DataSource, StorageEngine } from "../_chunks/storage.mjs";
-import { DuckDBHandle, SnapshotIndex } from "../_chunks/snapshot.mjs";
-import { SearchType } from "gscdump/query";
-import { Row, TableName } from "@gscdump/contracts";
-interface NodeDuckDBOptions {
-  verbose?: boolean;
-}
-declare function createNodeDuckDBHandle(opts?: NodeDuckDBOptions): DuckDBHandle;
-declare function resetNodeDuckDB(): void;
-interface NodeHarnessOptions {
-  dataDir: string;
-  /** Tenant user id. Defaults to `'local'` for single-user CLI installs. */
-  userId?: string;
-  /** Name of the manifest file under `dataDir`. Defaults to `manifest.json`. */
-  manifestFilename?: string;
-}
-interface NodeHarness {
-  engine: StorageEngine;
-  /**
-   * Underlying filesystem-backed DataSource. Exposed so commands that write
-   *  derivative artifacts (rollups, exports) don't have to re-instantiate it.
-   */
-  dataSource: DataSource;
-  dataDir: string;
-  userId: string;
-  siteIdFor: (siteUrl: string) => string;
-  runRawSql: (opts: {
-    sql: string;
-    siteUrl: string;
-    table: TableName;
-    params?: unknown[];
-    /**
-     * Restrict the underlying manifest lookup to a single GSC search-type
-     * slice. Undefined keeps the legacy cross-type union.
-     */
-    searchType?: SearchType;
-  }) => Promise<{
-    rows: Row[];
-    sql: string;
-    keys: string[];
-  }>;
-}
-declare function createNodeHarness(opts: NodeHarnessOptions): NodeHarness;
-/**
- * Runs arbitrary SQL and returns rows as plain objects. Caller supplies
- * this so the function works with AsyncDuckDB (browser DuckDB-WASM) or
- * @duckdb/node-api (Node) without coupling to either.
- */
-type SnapshotQueryRunner = (sql: string) => Promise<Array<Record<string, unknown>>>;
-interface AttachSnapshotOptions {
-  /** Index produced by the builder. */
-  index: SnapshotIndex;
-  /**
-   * Map from filename (`cold-YYYY-MM.duckdb`, `hot.duckdb`) to an HTTPS
-   * URL (typically a pre-signed R2 URL). Must contain an entry for every
-   * cold month in `index.cold` and — if `index.hot` — for `hot.duckdb`.
-   */
-  attachUrls: Record<string, string>;
-  /** Schema the unified views land under. Default `main`. */
-  schema?: string;
-  /**
-   * DuckDB httpfs can error with "Server sent back more data than expected"
-   * against some proxies; `force_download=true` sidesteps it. Default true.
-   */
-  forceDownload?: boolean;
-}
-interface AttachSnapshotResult {
-  schema: string;
-  /** Aliases we ATTACH'd — e.g. ['cold_2024_09', 'cold_2024_10', 'hot']. */
-  aliases: string[];
-  /** Table names with a UNION view created under `schema`. */
-  tables: string[];
-}
-/**
- * Turns a filename like `cold-2024-09.duckdb` into a valid SQL identifier
- * `cold_2024_09`. `hot.duckdb` → `hot`.
- */
-declare function snapshotAlias(fileName: string): string;
-declare function attachSnapshotIndex(runner: SnapshotQueryRunner, opts: AttachSnapshotOptions): Promise<AttachSnapshotResult>;
-interface AttachParquetIndexOptions {
-  /**
-   * Map of table name → list of Parquet URLs. The URL list may mix monthly
-   * compacted files and per-day files — DuckDB will scan all of them with
-   * `union_by_name = true`. Empty lists are skipped (no view created).
-   */
-  tables: Record<string, string[]>;
-  /** Schema the views land under. Default `main`. */
-  schema?: string;
-  /**
-   * DuckDB httpfs can error with "Server sent back more data than expected"
-   * against some proxies; `force_download=true` sidesteps it. Default true.
-   */
-  forceDownload?: boolean;
-}
-interface AttachParquetIndexResult {
-  schema: string;
-  /** Tables for which a view was created. */
-  tables: string[];
-}
-declare function attachParquetIndex(runner: SnapshotQueryRunner, opts: AttachParquetIndexOptions): Promise<AttachParquetIndexResult>;
+import { NodeDuckDBOptions, createNodeDuckDBHandle, resetNodeDuckDB } from "./duckdb-node.mjs";
+import { NodeHarness, NodeHarnessOptions, createNodeHarness } from "./node-harness.mjs";
+import { AttachSnapshotOptions, AttachSnapshotResult, SnapshotQueryRunner, attachSnapshotIndex, snapshotAlias } from "./snapshot-attach.mjs";
+import { AttachParquetIndexOptions, AttachParquetIndexResult, attachParquetIndex } from "./parquet-attach.mjs";
 export { type AttachParquetIndexOptions, type AttachParquetIndexResult, type AttachSnapshotOptions, type AttachSnapshotResult, type NodeDuckDBOptions, type NodeHarness, type NodeHarnessOptions, type SnapshotQueryRunner, attachParquetIndex, attachSnapshotIndex, createNodeDuckDBHandle, createNodeHarness, resetNodeDuckDB, snapshotAlias };

package/dist/adapters/node.mjs

@@ -1,241 +1,5 @@
-import { engineErrors } from "../errors.mjs";
-import { createDuckDBCodec, createDuckDBExecutor, createStorageEngine } from "../_chunks/engine.mjs";
-import { arrowToRows } from "../arrow-utils.mjs";
-import { createFilesystemDataSource, createFilesystemManifestStore } from "./filesystem.mjs";
-import { createRequire } from "node:module";
-import { err, ok, unwrapResult } from "gscdump/result";
-import process from "node:process";
-import { unlinkSync } from "node:fs";
-import { tmpdir } from "node:os";
-import path, { join } from "node:path";
-import { fileURLToPath } from "node:url";
-import { ConsoleLogger, NODE_RUNTIME, VoidLogger, createDuckDB } from "@duckdb/duckdb-wasm/dist/duckdb-node-blocking.cjs";
-import { encodeSiteId } from "gscdump/tenant";
-const require_ = createRequire(typeof __filename !== "undefined" ? __filename : typeof import.meta !== "undefined" ? fileURLToPath(import.meta.url) : process.cwd());
-let singleton = null;
-let singletonOpts = null;
-function bundles() {
-	return {
-		mvp: {
-			mainModule: require_.resolve("@duckdb/duckdb-wasm/dist/duckdb-mvp.wasm"),
-			mainWorker: null
-		},
-		eh: {
-			mainModule: require_.resolve("@duckdb/duckdb-wasm/dist/duckdb-eh.wasm"),
-			mainWorker: null
-		}
-	};
-}
-async function initialize(opts) {
-	const logger = opts.verbose ? new ConsoleLogger() : new VoidLogger();
-	const db = await createDuckDB(bundles(), logger, NODE_RUNTIME);
-	await db.instantiate();
-	return {
-		db,
-		conn: db.connect()
-	};
-}
-function getSingleton(opts) {
-	if (!singleton) {
-		singleton = initialize(opts);
-		singletonOpts = opts;
-	}
-	return singleton;
-}
-function createNodeDuckDBHandle(opts = {}) {
-	if (singleton && opts.verbose !== void 0 && opts.verbose !== (singletonOpts?.verbose ?? false)) console.warn(`[gscdump] createNodeDuckDBHandle: ignoring verbose=${opts.verbose} — a shared DuckDB instance was already initialized with verbose=${singletonOpts?.verbose ?? false}. Call resetNodeDuckDB() before re-initializing to change it.`);
-	getSingleton(opts);
-	return {
-		async query(sql, params) {
-			const { conn } = await getSingleton(opts);
-			if (!params || params.length === 0) return arrowToRows(conn.query(sql));
-			const stmt = conn.prepare(sql);
-			try {
-				return arrowToRows(stmt.query(...params));
-			} finally {
-				stmt.close();
-			}
-		},
-		async registerFileBuffer(name, bytes) {
-			const { db } = await getSingleton(opts);
-			db.registerFileBuffer(name, bytes);
-		},
-		async copyFileToBuffer(name) {
-			const { db } = await getSingleton(opts);
-			return db.copyFileToBuffer(name);
-		},
-		async dropFiles(names) {
-			const { db } = await getSingleton(opts);
-			for (const name of names) {
-				try {
-					db.dropFile(name);
-				} catch (error) {
-					const message = error instanceof Error ? error.message : String(error);
-					if (!/not found|does not exist|unknown file/i.test(message)) throw error;
-				}
-				try {
-					unlinkSync(name);
-				} catch (error) {
-					if (error.code !== "ENOENT") throw error;
-				}
-			}
-		},
-		makeTempPath(ext) {
-			return join(tmpdir(), `gscdump-${Math.random().toString(36).slice(2, 10)}.${ext}`);
-		}
-	};
-}
-function resetNodeDuckDB() {
-	const pending = singleton;
-	singleton = null;
-	singletonOpts = null;
-	pending?.then(({ db, conn }) => {
-		conn.close();
-		db.reset();
-	}).catch((err) => {
-		console.warn("[gscdump] resetNodeDuckDB: failed to release DuckDB instance", err);
-	});
-}
-function createNodeHarness(opts) {
-	const dataDir = opts.dataDir;
-	const userId = opts.userId ?? "local";
-	const manifestFilename = opts.manifestFilename ?? "manifest.json";
-	const handle = createNodeDuckDBHandle();
-	const factory = { getDuckDB: async () => handle };
-	const dataSource = createFilesystemDataSource({ rootDir: dataDir });
-	const engine = createStorageEngine({
-		dataSource,
-		manifestStore: createFilesystemManifestStore({ path: path.join(dataDir, manifestFilename) }),
-		codec: createDuckDBCodec(factory),
-		executor: createDuckDBExecutor(factory)
-	});
-	async function runRawSql(runOpts) {
-		const result = await engine.runSQL({
-			ctx: {
-				userId,
-				siteId: encodeSiteId(runOpts.siteUrl)
-			},
-			table: runOpts.table,
-			fileSets: { FILES: { table: runOpts.table } },
-			sql: runOpts.sql,
-			params: runOpts.params ?? [],
-			...runOpts.searchType !== void 0 ? { searchType: runOpts.searchType } : {}
-		});
-		return {
-			rows: result.rows,
-			sql: result.sql,
-			keys: result.objectKeys
-		};
-	}
-	return {
-		engine,
-		dataSource,
-		dataDir,
-		userId,
-		siteIdFor: encodeSiteId,
-		runRawSql
-	};
-}
-const IDENT_RE = /^[A-Z_][\w$]*$/i;
-async function attachParquetIndex(runner, opts) {
-	const schema = opts.schema ?? "main";
-	const forceDownload = opts.forceDownload !== false;
-	if (!IDENT_RE.test(schema)) throw new TypeError(`attachParquetIndex: invalid schema identifier ${JSON.stringify(schema)}`);
-	for (const table of Object.keys(opts.tables)) if (!IDENT_RE.test(table)) throw new TypeError(`attachParquetIndex: invalid table identifier ${JSON.stringify(table)}`);
-	await runner("LOAD httpfs").catch(() => void 0);
-	if (forceDownload) await runner("SET force_download=true");
-	await runner(`CREATE SCHEMA IF NOT EXISTS ${schema}`);
-	const created = [];
-	for (const [table, urls] of Object.entries(opts.tables)) {
-		if (urls.length === 0) continue;
-		await runner(`CREATE OR REPLACE VIEW ${schema}.${table} AS SELECT * FROM read_parquet([${urls.map((u) => `'${u.replace(/'/g, "''")}'`).join(", ")}], union_by_name = true)`);
-		created.push(table);
-	}
-	return {
-		schema,
-		tables: created
-	};
-}
-const YEAR_MONTH_RE = /^\d{4}-\d{2}$/;
-const SCHEMA_IDENT_RE = /^[A-Z_][\w$]*$/i;
-const COLD_FILENAME_RE = /^cold-(\d{4}-\d{2})\.duckdb$/;
-function snapshotAlias(fileName) {
-	if (fileName === "hot.duckdb") return "hot";
-	const m = fileName.match(COLD_FILENAME_RE);
-	if (!m?.[1]) throw new TypeError(`snapshotAlias: unrecognised filename ${JSON.stringify(fileName)}`);
-	return `cold_${m[1].replace("-", "_")}`;
-}
-const SNAPSHOT_TYPE_ERROR_KINDS = /* @__PURE__ */ new Set([
-	"invalid-snapshot-filename",
-	"unsupported-snapshot-index-version",
-	"invalid-schema-identifier",
-	"invalid-year-month"
-]);
-function snapshotAttachErrorToException(error) {
-	const exception = SNAPSHOT_TYPE_ERROR_KINDS.has(error.kind) ? new TypeError(error.message) : new Error(error.message);
-	exception.engineError = error;
-	return exception;
-}
-async function attachSnapshotIndexResult(runner, opts) {
-	const { index, attachUrls } = opts;
-	const schema = opts.schema ?? "main";
-	const forceDownload = opts.forceDownload !== false;
-	if (index?.version !== 1) return err(engineErrors.unsupportedSnapshotIndexVersion(index?.version));
-	if (!SCHEMA_IDENT_RE.test(schema)) return err(engineErrors.invalidSchemaIdentifier(schema));
-	for (const ym of index.cold) if (!YEAR_MONTH_RE.test(ym)) return err(engineErrors.invalidYearMonth(ym));
-	await runner("LOAD httpfs").catch(() => void 0);
-	if (forceDownload) await runner("SET force_download=true");
-	const plan = [];
-	for (const ym of index.cold) {
-		const fileName = `cold-${ym}.duckdb`;
-		const url = attachUrls[fileName];
-		if (!url) return err(engineErrors.missingAttachUrl(fileName));
-		plan.push({
-			fileName,
-			alias: snapshotAlias(fileName),
-			url
-		});
-	}
-	if (index.hot) {
-		const fileName = "hot.duckdb";
-		const url = attachUrls[fileName];
-		if (!url) return err(engineErrors.missingAttachUrl(fileName));
-		plan.push({
-			fileName,
-			alias: snapshotAlias(fileName),
-			url
-		});
-	}
-	const aliases = [];
-	for (const { alias, url } of plan) {
-		await runner(`ATTACH '${url.replace(/'/g, "''")}' AS ${alias} (READ_ONLY)`);
-		aliases.push(alias);
-	}
-	const aliasSet = new Set(aliases);
-	const tableRows = await runner("SELECT database_name, table_name FROM duckdb_tables()");
-	const present = /* @__PURE__ */ new Map();
-	for (const row of tableRows) {
-		const db = String(row.database_name ?? "");
-		const table = String(row.table_name ?? "");
-		if (!aliasSet.has(db) || !table) continue;
-		const list = present.get(table);
-		if (list) list.push(db);
-		else present.set(table, [db]);
-	}
-	const tables = [];
-	for (const [table, dbs] of present) {
-		if (!SCHEMA_IDENT_RE.test(table)) continue;
-		const dbsSet = new Set(dbs);
-		await runner(`CREATE OR REPLACE VIEW ${schema}.${table} AS ${aliases.filter((a) => dbsSet.has(a)).map((db) => `SELECT * FROM ${db}.${table}`).join(" UNION ALL BY NAME ")}`);
-		tables.push(table);
-	}
-	return ok({
-		schema,
-		aliases,
-		tables
-	});
-}
-async function attachSnapshotIndex(runner, opts) {
-	return unwrapResult(await attachSnapshotIndexResult(runner, opts), snapshotAttachErrorToException);
-}
+import { createNodeDuckDBHandle, resetNodeDuckDB } from "./duckdb-node.mjs";
+import { createNodeHarness } from "./node-harness.mjs";
+import { attachParquetIndex } from "./parquet-attach.mjs";
+import { attachSnapshotIndex, snapshotAlias } from "./snapshot-attach.mjs";
 export { attachParquetIndex, attachSnapshotIndex, createNodeDuckDBHandle, createNodeHarness, resetNodeDuckDB, snapshotAlias };

package/dist/adapters/parquet-attach.d.mts

@@ -0,0 +1,23 @@
+import { SnapshotQueryRunner } from "./snapshot-attach.mjs";
+interface AttachParquetIndexOptions {
+  /**
+   * Map of table name → list of Parquet URLs. The URL list may mix monthly
+   * compacted files and per-day files — DuckDB will scan all of them with
+   * `union_by_name = true`. Empty lists are skipped (no view created).
+   */
+  tables: Record<string, string[]>;
+  /** Schema the views land under. Default `main`. */
+  schema?: string;
+  /**
+   * DuckDB httpfs can error with "Server sent back more data than expected"
+   * against some proxies; `force_download=true` sidesteps it. Default true.
+   */
+  forceDownload?: boolean;
+}
+interface AttachParquetIndexResult {
+  schema: string;
+  /** Tables for which a view was created. */
+  tables: string[];
+}
+declare function attachParquetIndex(runner: SnapshotQueryRunner, opts: AttachParquetIndexOptions): Promise<AttachParquetIndexResult>;
+export { AttachParquetIndexOptions, AttachParquetIndexResult, attachParquetIndex };

package/dist/adapters/parquet-attach.mjs

@@ -0,0 +1,21 @@
+const IDENT_RE = /^[A-Z_][\w$]*$/i;
+async function attachParquetIndex(runner, opts) {
+	const schema = opts.schema ?? "main";
+	const forceDownload = opts.forceDownload !== false;
+	if (!IDENT_RE.test(schema)) throw new TypeError(`attachParquetIndex: invalid schema identifier ${JSON.stringify(schema)}`);
+	for (const table of Object.keys(opts.tables)) if (!IDENT_RE.test(table)) throw new TypeError(`attachParquetIndex: invalid table identifier ${JSON.stringify(table)}`);
+	await runner("LOAD httpfs").catch(() => void 0);
+	if (forceDownload) await runner("SET force_download=true");
+	await runner(`CREATE SCHEMA IF NOT EXISTS ${schema}`);
+	const created = [];
+	for (const [table, urls] of Object.entries(opts.tables)) {
+		if (urls.length === 0) continue;
+		await runner(`CREATE OR REPLACE VIEW ${schema}.${table} AS SELECT * FROM read_parquet([${urls.map((u) => `'${u.replace(/'/g, "''")}'`).join(", ")}], union_by_name = true)`);
+		created.push(table);
+	}
+	return {
+		schema,
+		tables: created
+	};
+}
+export { attachParquetIndex };