@voltro/plugin-duckdb 0.85.0 → 0.87.0

package/dist/index.js

@@ -1,225 +1,17 @@
-import { Effect as e } from "effect";
-import { AnalyticsError as t } from "@voltro/runtime";
-//#region src/sink.ts
-var n = "duckdb", r = "voltro_events", i = null, a = () => i === null ? e.tryPromise({
-	try: async () => {
-		let e = await import("@duckdb/node-api");
-		return i = e, e;
-	},
-	catch: (e) => new t({
-		provider: n,
-		operation: "loadNativeModule",
-		cause: "@duckdb/node-api could not be loaded — the native binary for this platform may be missing (it is an optionalDependency). " + (e instanceof Error ? `${e.name}: ${e.message}` : String(e))
-	})
-}) : e.succeed(i), o = (e) => new i.DuckDBTimestampValue(BigInt(e.getTime()) * 1000n), s = (e) => `voltro_mirror_${e}`, c = /^[A-Za-z_][A-Za-z0-9_]*$/, l = (e, r, i) => {
-	if (!c.test(r)) throw new t({
-		provider: n,
-		operation: e,
-		cause: `invalid ${i} ${JSON.stringify(r)} — must match ${c.source} (letters, digits, underscore; no dots/quotes/spaces)`
-	});
-	return r;
-}, u = /['\\]/, d = (e) => {
-	for (let t = 0; t < e.length; t++) if (e.charCodeAt(t) < 32) return !0;
-	return !1;
-}, f = (e, r) => {
-	if (u.test(r) || d(r)) throw new t({
-		provider: n,
-		operation: e,
-		cause: `invalid property key ${JSON.stringify(r)} — must not contain quotes, backslashes, or control characters`
-	});
-	return r;
-}, p = `
-  CREATE SEQUENCE IF NOT EXISTS voltro_events_id_seq;
-  CREATE TABLE IF NOT EXISTS ${r} (
-    id          BIGINT PRIMARY KEY DEFAULT nextval('voltro_events_id_seq'),
-    name        VARCHAR NOT NULL,
-    subject_id  VARCHAR,
-    tenant_id   VARCHAR,
-    properties  JSON NOT NULL DEFAULT '{}',
-    occurred_at TIMESTAMP NOT NULL DEFAULT now()
-  );
-`.trim(), m = (r) => (i) => i.pipe(e.catchDefect((n) => n instanceof t ? e.fail(n) : e.die(n)), e.mapError((e) => e instanceof t ? e : new t({
-	provider: n,
-	operation: r,
-	cause: e instanceof Error ? `${e.name}: ${e.message}` : String(e)
-}))), h = (r) => r.conn === null ? e.fail(new t({
-	provider: n,
-	operation: "ensureConn",
-	cause: "connection not initialized — spec.initialize must run before queries"
-})) : e.succeed(r.conn), g = (e, t) => `date_trunc('${t}', ${e})`, _ = (e, t) => `json_extract_string(properties, '$.${f(e, t)}')`, v = (e, t) => `TRY_CAST(json_extract_string(properties, '$.${f(e, t)}') AS DOUBLE)`, y = (e, t) => t === "count" ? "COUNT(*)" : t === "unique" ? "COUNT(DISTINCT subject_id)" : "sum" in t ? `COALESCE(SUM(${v(e, t.sum)}), 0)` : `AVG(${v(e, t.avg)})`, b = (e, t) => t === "count" ? "COUNT(*)" : `COALESCE(SUM(${v(e, t.sum)}), 0)`, x = (e, t, n, r, a) => {
-	let { VARCHAR: s, TIMESTAMP: c } = i, l = [], u = [], d = [];
-	if (d.push("name = ?"), l.push(t), u.push(s), d.push("occurred_at >= ?"), l.push(o(n.from)), u.push(c), n.to !== void 0 && (d.push("occurred_at < ?"), l.push(o(n.to)), u.push(c)), a != null && (d.push("tenant_id = ?"), l.push(a), u.push(s)), r !== void 0) for (let [t, n] of Object.entries(r)) d.push(`${_(e, t)} = ?`), l.push(String(n)), u.push(s);
-	return {
-		sql: d.join(" AND "),
-		params: l,
-		types: u
-	};
-}, S = (e, t, n, r, a) => [
-	`INSERT INTO ${e} (id, data, version, is_deleted) VALUES (?, ?::JSON, ?::BIGINT, ?) ON CONFLICT (id) DO UPDATE SET data = EXCLUDED.data, version = EXCLUDED.version, is_deleted = EXCLUDED.is_deleted WHERE EXCLUDED.version > ${e}.version`,
-	[
-		t,
-		n,
-		String(r),
-		a
-	],
-	[
-		i.VARCHAR,
-		i.VARCHAR,
-		i.VARCHAR,
-		i.BOOLEAN
-	]
-], C = (t, n, r) => {
-	let a = /* @__PURE__ */ new Set(), o = (n, r) => e.gen(function* () {
-		if (a.has(r)) return;
-		l(n, r, "mirror table name");
-		let i = yield* h(t), o = s(r);
-		yield* e.tryPromise({
-			try: () => i.run(`CREATE TABLE IF NOT EXISTS ${o} (id VARCHAR PRIMARY KEY, data JSON NOT NULL, version BIGINT NOT NULL DEFAULT 0, is_deleted BOOLEAN NOT NULL DEFAULT false)`),
-			catch: (e) => e
-		}), a.add(r);
-	});
-	return {
-		tables: n,
-		primaryKey: r,
-		upsert: ({ table: n, row: i, version: a }) => e.gen(function* () {
-			yield* o("mirror.upsert", n);
-			let c = yield* h(t), l = s(n), u = String(i[r]);
-			yield* e.tryPromise({
-				try: () => c.run(...S(l, u, JSON.stringify(i), a, !1)),
-				catch: (e) => e
-			});
-		}).pipe(m("mirror.upsert")),
-		remove: ({ table: n, primaryKeyValue: r, version: i }) => e.gen(function* () {
-			yield* o("mirror.remove", n);
-			let a = yield* h(t), c = s(n), l = String(r);
-			yield* e.tryPromise({
-				try: () => a.run(...S(c, l, "{}", i, !0)),
-				catch: (e) => e
-			});
-		}).pipe(m("mirror.remove")),
-		maxVersion: ({ table: n, primaryKeyValue: r }) => e.gen(function* () {
-			yield* o("mirror.maxVersion", n);
-			let a = yield* h(t), c = s(n), l = (yield* e.tryPromise({
-				try: () => a.runAndReadAll(`SELECT version FROM ${c} WHERE id = ?`, [String(r)], [i.VARCHAR]),
-				catch: (e) => e
-			})).getRowObjectsJS()[0]?.version;
-			return l == null ? null : Number(l);
-		}).pipe(m("mirror.maxVersion"))
-	};
-}, w = (t, a) => {
-	let s = a.mirrorTables ?? [], c = s.length > 0 ? C(t, s, a.mirrorPrimaryKey ?? "id") : void 0;
-	return {
-		info: {
-			provider: n,
-			capabilities: {
-				track: !0,
-				aggregate: !0,
-				timeseries: !0,
-				topN: !0,
-				mirror: c !== void 0
-			},
-			detail: {
-				path: a.path ?? ":memory:",
-				...s.length > 0 ? { mirrorTables: s } : {}
-			}
-		},
-		track: (n) => e.gen(function* () {
-			let a = yield* h(t), { VARCHAR: s, TIMESTAMP: c } = i, l = n.occurredAt ?? /* @__PURE__ */ new Date(), u = JSON.stringify(n.properties), d = n.subjectId ?? null, f = n.tenantId ?? null;
-			yield* e.tryPromise({
-				try: () => a.run(`INSERT INTO ${r} (name, subject_id, tenant_id, properties, occurred_at) VALUES (?, ?, ?, ?::JSON, ?)`, [
-					n.name,
-					d,
-					f,
-					u,
-					o(l)
-				], [
-					s,
-					s,
-					s,
-					s,
-					c
-				]),
-				catch: (e) => e
-			});
-		}).pipe(m("track")),
-		aggregate: (n) => e.gen(function* () {
-			let i = yield* h(t), a = x("aggregate", n.event, n.range, n.filter, n.tenantId), o = `SELECT ${y("aggregate", n.metric)} AS value FROM ${r} WHERE ${a.sql}`, s = (yield* e.tryPromise({
-				try: () => i.runAndReadAll(o, a.params, a.types),
-				catch: (e) => e
-			})).getRowObjectsJS()[0]?.value ?? 0;
-			return typeof s == "number" ? s : Number(s ?? 0);
-		}).pipe(m("aggregate")),
-		timeseries: (n) => e.gen(function* () {
-			let i = yield* h(t), a = x("timeseries", n.event, n.range, n.filter, n.tenantId), o = g("occurred_at", n.bucket), s = `
-        SELECT ${o} AS bucket_start, ${y("timeseries", n.metric)} AS value
-        FROM ${r}
-        WHERE ${a.sql}
-        GROUP BY ${o}
-        ORDER BY ${o} ASC
-      `;
-			return (yield* e.tryPromise({
-				try: () => i.runAndReadAll(s, a.params, a.types),
-				catch: (e) => e
-			})).getRowObjectsJS().map((e) => ({
-				bucketStart: e.bucket_start instanceof Date ? e.bucket_start : new Date(e.bucket_start),
-				value: typeof e.value == "number" ? e.value : Number(e.value ?? 0)
-			}));
-		}).pipe(m("timeseries")),
-		topN: (n) => e.gen(function* () {
-			let i = yield* h(t), a = x("topN", n.event, n.range, n.filter, n.tenantId), o = Math.max(1, Math.floor(n.n)), s = _("topN", n.groupBy), c = `
-        SELECT ${s} AS key, ${b("topN", n.metric)} AS value
-        FROM ${r}
-        WHERE ${a.sql}
-        GROUP BY ${s}
-        ORDER BY value DESC
-        LIMIT ${o}
-      `;
-			return (yield* e.tryPromise({
-				try: () => i.runAndReadAll(c, a.params, a.types),
-				catch: (e) => e
-			})).getRowObjectsJS().filter((e) => e.key !== null).map((e) => ({
-				key: e.key,
-				value: typeof e.value == "number" ? e.value : Number(e.value ?? 0)
-			}));
-		}).pipe(m("topN")),
-		...c === void 0 ? {} : { mirror: c }
-	};
-}, T = (r, i) => e.gen(function* () {
-	if (r.conn !== null) return;
-	let o = yield* a(), s = yield* e.tryPromise({
-		try: async () => {
-			let e = await o.DuckDBInstance.create(i.path ?? ":memory:"), t = await e.connect();
-			return await t.run(p), {
-				instance: e,
-				connection: t
-			};
-		},
-		catch: (e) => new t({
-			provider: n,
-			operation: "initialize",
-			cause: e instanceof Error ? `${e.name}: ${e.message}` : String(e)
-		})
-	});
-	r.instance = s.instance, r.conn = s.connection;
-}), E = (t) => e.gen(function* () {
-	let { conn: r, instance: i } = t;
-	t.conn = null, t.instance = null, (r !== null || i !== null) && (yield* e.try(() => {
-		r?.closeSync(), i?.closeSync();
-	}).pipe(e.catch((t) => e.logWarning("duckdb analytics sink teardown failed", {
-		provider: n,
-		err: t
-	}))));
-}), D = (e = {}) => {
-	let t = {
+import { buildDuckdbSink as e, closeDuckdbState as t, initializeDuckdbState as n } from "./sink.js";
+//#region src/index.ts
+var r = (r = {}) => {
+	let i = {
 		conn: null,
 		instance: null
 	};
 	return {
 		kind: "duckdb",
-		build: (n) => w(t, e),
-		initialize: () => T(t, e),
-		dispose: () => E(t)
+		sql: "none",
+		build: () => e(i, r),
+		initialize: () => n(i, r),
+		dispose: () => t(i)
 	};
 };
 //#endregion
-export { D as duckdbAnalytics };
+export { r as duckdbAnalytics };

package/dist/sink.js

@@ -0,0 +1,227 @@
+import { Effect as e } from "effect";
+import { AnalyticsError as t } from "@voltro/runtime";
+//#region src/sink.ts
+var n = "duckdb", r = "voltro_events", i = null, a = () => i === null ? e.tryPromise({
+	try: async () => {
+		let e = await import("@duckdb/node-api");
+		return i = e, e;
+	},
+	catch: (e) => new t({
+		provider: n,
+		operation: "loadNativeModule",
+		cause: "@duckdb/node-api could not be loaded — the native binary for this platform may be missing (it is an optionalDependency). " + (e instanceof Error ? `${e.name}: ${e.message}` : String(e))
+	})
+}) : e.succeed(i), o = (e) => new i.DuckDBTimestampValue(BigInt(e.getTime()) * 1000n), s = (e) => `voltro_mirror_${e}`, c = /^[A-Za-z_][A-Za-z0-9_]*$/, l = (e, r, i) => {
+	if (!c.test(r)) throw new t({
+		provider: n,
+		operation: e,
+		cause: `invalid ${i} ${JSON.stringify(r)} — must match ${c.source} (letters, digits, underscore; no dots/quotes/spaces)`
+	});
+	return r;
+}, u = /['\\]/, d = (e) => {
+	for (let t = 0; t < e.length; t++) if (e.charCodeAt(t) < 32) return !0;
+	return !1;
+}, f = (e, r) => {
+	if (u.test(r) || d(r)) throw new t({
+		provider: n,
+		operation: e,
+		cause: `invalid property key ${JSON.stringify(r)} — must not contain quotes, backslashes, or control characters`
+	});
+	return r;
+}, p = `
+  CREATE SEQUENCE IF NOT EXISTS voltro_events_id_seq;
+  CREATE TABLE IF NOT EXISTS ${r} (
+    id          BIGINT PRIMARY KEY DEFAULT nextval('voltro_events_id_seq'),
+    name        VARCHAR NOT NULL,
+    subject_id  VARCHAR,
+    tenant_id   VARCHAR,
+    properties  JSON NOT NULL DEFAULT '{}',
+    occurred_at TIMESTAMP NOT NULL DEFAULT now()
+  );
+`.trim(), m = (r) => (i) => i.pipe(e.catchDefect((n) => n instanceof t ? e.fail(n) : e.die(n)), e.mapError((e) => e instanceof t ? e : new t({
+	provider: n,
+	operation: r,
+	cause: e instanceof Error ? `${e.name}: ${e.message}` : String(e)
+}))), h = (r) => r.conn === null ? e.fail(new t({
+	provider: n,
+	operation: "ensureConn",
+	cause: "connection not initialized — spec.initialize must run before queries"
+})) : e.succeed(r.conn), g = (e, t) => `date_trunc('${t}', ${e})`, _ = (e, t) => `json_extract_string(properties, '$.${f(e, t)}')`, v = (e, t) => `TRY_CAST(json_extract_string(properties, '$.${f(e, t)}') AS DOUBLE)`, y = (e, t) => t === "count" ? "COUNT(*)" : t === "unique" ? "COUNT(DISTINCT subject_id)" : "sum" in t ? `COALESCE(SUM(${v(e, t.sum)}), 0)` : `AVG(${v(e, t.avg)})`, b = (e, t) => t === "count" ? "COUNT(*)" : `COALESCE(SUM(${v(e, t.sum)}), 0)`, x = (e, t, n, r, a) => {
+	let { VARCHAR: s, TIMESTAMP: c } = i, l = [], u = [], d = [];
+	if (d.push("name = ?"), l.push(t), u.push(s), d.push("occurred_at >= ?"), l.push(o(n.from)), u.push(c), n.to !== void 0 && (d.push("occurred_at < ?"), l.push(o(n.to)), u.push(c)), a != null && (d.push("tenant_id = ?"), l.push(a), u.push(s)), r !== void 0) for (let [t, n] of Object.entries(r)) d.push(`${_(e, t)} = ?`), l.push(String(n)), u.push(s);
+	return {
+		sql: d.join(" AND "),
+		params: l,
+		types: u
+	};
+}, S = (e, t, n, r, a) => [
+	`INSERT INTO ${e} (id, data, version, is_deleted) VALUES (?, ?::JSON, ?::BIGINT, ?) ON CONFLICT (id) DO UPDATE SET data = EXCLUDED.data, version = EXCLUDED.version, is_deleted = EXCLUDED.is_deleted WHERE EXCLUDED.version > ${e}.version`,
+	[
+		t,
+		n,
+		String(r),
+		a
+	],
+	[
+		i.VARCHAR,
+		i.VARCHAR,
+		i.VARCHAR,
+		i.BOOLEAN
+	]
+], C = (t, n, r) => {
+	let a = /* @__PURE__ */ new Set(), o = (n, r) => e.gen(function* () {
+		if (a.has(r)) return;
+		l(n, r, "mirror table name");
+		let i = yield* h(t), o = s(r);
+		yield* e.tryPromise({
+			try: () => i.run(`CREATE TABLE IF NOT EXISTS ${o} (id VARCHAR PRIMARY KEY, data JSON NOT NULL, version BIGINT NOT NULL DEFAULT 0, is_deleted BOOLEAN NOT NULL DEFAULT false)`),
+			catch: (e) => e
+		}), a.add(r);
+	});
+	return {
+		tables: n,
+		primaryKey: r,
+		upsert: ({ table: n, row: i, version: a }) => e.gen(function* () {
+			yield* o("mirror.upsert", n);
+			let c = yield* h(t), l = s(n), u = String(i[r]);
+			yield* e.tryPromise({
+				try: () => c.run(...S(l, u, JSON.stringify(i), a, !1)),
+				catch: (e) => e
+			});
+		}).pipe(m("mirror.upsert")),
+		remove: ({ table: n, primaryKeyValue: r, version: i }) => e.gen(function* () {
+			yield* o("mirror.remove", n);
+			let a = yield* h(t), c = s(n), l = String(r);
+			yield* e.tryPromise({
+				try: () => a.run(...S(c, l, "{}", i, !0)),
+				catch: (e) => e
+			});
+		}).pipe(m("mirror.remove")),
+		maxVersion: ({ table: n, primaryKeyValue: r }) => e.gen(function* () {
+			yield* o("mirror.maxVersion", n);
+			let a = yield* h(t), c = s(n), l = (yield* e.tryPromise({
+				try: () => a.runAndReadAll(`SELECT version FROM ${c} WHERE id = ?`, [String(r)], [i.VARCHAR]),
+				catch: (e) => e
+			})).getRowObjectsJS()[0]?.version;
+			return l == null ? null : Number(l);
+		}).pipe(m("mirror.maxVersion"))
+	};
+}, w = (t, a) => {
+	let s = a.mirrorTables ?? [], c = s.length > 0 ? C(t, s, a.mirrorPrimaryKey ?? "id") : void 0;
+	return {
+		info: {
+			provider: n,
+			capabilities: {
+				track: !0,
+				aggregate: !0,
+				timeseries: !0,
+				topN: !0,
+				mirror: c !== void 0
+			},
+			detail: {
+				path: a.path ?? ":memory:",
+				...s.length > 0 ? { mirrorTables: s } : {}
+			}
+		},
+		track: (n) => e.gen(function* () {
+			let a = yield* h(t), { VARCHAR: s, TIMESTAMP: c } = i, l = n.occurredAt ?? /* @__PURE__ */ new Date(), u = JSON.stringify(n.properties), d = n.subjectId ?? null, f = n.tenantId ?? null;
+			yield* e.tryPromise({
+				try: () => a.run(`INSERT INTO ${r} (name, subject_id, tenant_id, properties, occurred_at) VALUES (?, ?, ?, ?::JSON, ?)`, [
+					n.name,
+					d,
+					f,
+					u,
+					o(l)
+				], [
+					s,
+					s,
+					s,
+					s,
+					c
+				]),
+				catch: (e) => e
+			});
+		}).pipe(m("track")),
+		aggregate: (n) => e.gen(function* () {
+			let i = yield* h(t), a = x("aggregate", n.event, n.range, n.filter, n.tenantId), o = `SELECT ${y("aggregate", n.metric)} AS value FROM ${r} WHERE ${a.sql}`, s = (yield* e.tryPromise({
+				try: () => i.runAndReadAll(o, a.params, a.types),
+				catch: (e) => e
+			})).getRowObjectsJS()[0]?.value ?? 0;
+			return typeof s == "number" ? s : Number(s ?? 0);
+		}).pipe(m("aggregate")),
+		timeseries: (n) => e.gen(function* () {
+			let i = yield* h(t), a = x("timeseries", n.event, n.range, n.filter, n.tenantId), o = g("occurred_at", n.bucket), s = `
+        SELECT ${o} AS bucket_start, ${y("timeseries", n.metric)} AS value
+        FROM ${r}
+        WHERE ${a.sql}
+        GROUP BY ${o}
+        ORDER BY ${o} ASC
+      `;
+			return (yield* e.tryPromise({
+				try: () => i.runAndReadAll(s, a.params, a.types),
+				catch: (e) => e
+			})).getRowObjectsJS().map((e) => ({
+				bucketStart: e.bucket_start instanceof Date ? e.bucket_start : new Date(e.bucket_start),
+				value: typeof e.value == "number" ? e.value : Number(e.value ?? 0)
+			}));
+		}).pipe(m("timeseries")),
+		topN: (n) => e.gen(function* () {
+			let i = yield* h(t), a = x("topN", n.event, n.range, n.filter, n.tenantId), o = Math.max(1, Math.floor(n.n)), s = _("topN", n.groupBy), c = `
+        SELECT ${s} AS key, ${b("topN", n.metric)} AS value
+        FROM ${r}
+        WHERE ${a.sql}
+        GROUP BY ${s}
+        ORDER BY value DESC
+        LIMIT ${o}
+      `;
+			return (yield* e.tryPromise({
+				try: () => i.runAndReadAll(c, a.params, a.types),
+				catch: (e) => e
+			})).getRowObjectsJS().filter((e) => e.key !== null).map((e) => ({
+				key: e.key,
+				value: typeof e.value == "number" ? e.value : Number(e.value ?? 0)
+			}));
+		}).pipe(m("topN")),
+		...c === void 0 ? {} : { mirror: c }
+	};
+}, T = (r, i) => e.gen(function* () {
+	if (r.conn !== null) return;
+	let o = yield* a(), s = yield* e.tryPromise({
+		try: async () => {
+			let e = await o.DuckDBInstance.create(i.path ?? ":memory:"), t;
+			try {
+				return t = await e.connect(), await t.run(p), {
+					instance: e,
+					connection: t
+				};
+			} catch (n) {
+				try {
+					t?.closeSync();
+				} finally {
+					e.closeSync();
+				}
+				throw n;
+			}
+		},
+		catch: (e) => new t({
+			provider: n,
+			operation: "initialize",
+			cause: e instanceof Error ? `${e.name}: ${e.message}` : String(e)
+		})
+	});
+	r.instance = s.instance, r.conn = s.connection;
+}).pipe(e.uninterruptible, e.onError(() => E(r))), E = (t) => e.gen(function* () {
+	let { conn: r, instance: i } = t;
+	t.conn = null, t.instance = null, (r !== null || i !== null) && (yield* e.try(() => {
+		try {
+			r?.closeSync();
+		} finally {
+			i?.closeSync();
+		}
+	}).pipe(e.catch((t) => e.logWarning("duckdb analytics sink teardown failed", {
+		provider: n,
+		err: t
+	}))));
+});
+//#endregion
+export { w as buildDuckdbSink, E as closeDuckdbState, T as initializeDuckdbState, s as mirrorTableName };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@voltro/plugin-duckdb",
-  "version": "0.85.0",
+  "version": "0.87.0",
   "description": "Embedded DuckDB AnalyticsSink. Runs DuckDB in-process for real OLAP performance without an external service. Events land in a DuckDB column-store table; aggregate / timeseries / topN queries execute vectorized.",
   "keywords": [
     "voltro",
@@ -33,7 +33,7 @@
     "node": ">=24.0.0"
   },
   "dependencies": {
-    "@voltro/runtime": "0.85.0"
+    "@voltro/runtime": "0.87.0"
   },
   "optionalDependencies": {
     "@duckdb/node-api": "1.5.5-r.5"