@gscdump/cloudflare 2.1.0 → 2.2.0

package/dist/arrow.mjs

@@ -43,7 +43,8 @@ function rowsToArrowIPC(rows, schemaColumns) {
 		}
 		if (hasString || !hasValue) types[name] = utf8();
 	}
-	const ipc = tableToIPC(tableFromArrays(data, { types }), { format: "stream" });
+	const table = tableFromArrays(data, { types });
+	const ipc = tableToIPC(table, { format: "stream" });
 	if (!ipc) throw new Error("rowsToArrowIPC: tableToIPC returned null");
 	return ipc;
 }

package/dist/engine.mjs

@@ -10,20 +10,24 @@ function createAnalyticsEngineRuntime() {
 			bucket: env.R2_DATA,
 			bucketName: env.R2_BUCKET_NAME
 		});
+		const dataSource = hooks.onR2Write ? {
+			...baseDataSource,
+			async write(key, bytes) {
+				hooks.onR2Write(bytes.byteLength);
+				return baseDataSource.write(key, bytes);
+			}
+		} : baseDataSource;
+		const manifestStore = createD1ManifestStore(db);
+		const codec = createDucklingsCodec(env);
+		const executor = createDucklingsExecutor(env, {
+			rowCache,
+			...hooks.duckdbRpcTimeoutMs !== void 0 ? { rpcTimeoutMs: hooks.duckdbRpcTimeoutMs } : {}
+		});
 		return createStorageEngine({
-			dataSource: hooks.onR2Write ? {
-				...baseDataSource,
-				async write(key, bytes) {
-					hooks.onR2Write(bytes.byteLength);
-					return baseDataSource.write(key, bytes);
-				}
-			} : baseDataSource,
-			manifestStore: createD1ManifestStore(db),
-			codec: createDucklingsCodec(env),
-			executor: createDucklingsExecutor(env, {
-				rowCache,
-				...hooks.duckdbRpcTimeoutMs !== void 0 ? { rpcTimeoutMs: hooks.duckdbRpcTimeoutMs } : {}
-			})
+			dataSource,
+			manifestStore,
+			codec,
+			executor
 		});
 	} };
 }

package/dist/server-tail/duckdb-iceberg-executor.mjs

@@ -48,16 +48,20 @@ function createDuckDbIcebergExecutor(config) {
 		return unwrapResult(await sendResult(sql), duckDbIcebergErrorToException);
 	}
 	function runSqlResult(sql, params = []) {
-		return sendResult(bindLiterals(resolveTablePlaceholders(sql, config), params));
+		const resolved = resolveTablePlaceholders(sql, config);
+		return sendResult(bindLiterals(resolved, params));
 	}
 	function runSql(sql, params = []) {
-		return send(bindLiterals(resolveTablePlaceholders(sql, config), params));
+		const resolved = resolveTablePlaceholders(sql, config);
+		return send(bindLiterals(resolved, params));
 	}
 	function runPlan(plan) {
-		return send(bindLiterals(plan.sql.split(TABLE_PLACEHOLDER).join(icebergTableRef(config, plan.table)), plan.params));
+		const resolved = plan.sql.split(TABLE_PLACEHOLDER).join(icebergTableRef(config, plan.table));
+		return send(bindLiterals(resolved, plan.params));
 	}
 	function runPlanResult(plan) {
-		return sendResult(bindLiterals(plan.sql.split(TABLE_PLACEHOLDER).join(icebergTableRef(config, plan.table)), plan.params));
+		const resolved = plan.sql.split(TABLE_PLACEHOLDER).join(icebergTableRef(config, plan.table));
+		return sendResult(bindLiterals(resolved, plan.params));
 	}
 	function runArchetypeResult(query) {
 		if (query.archetype === "arbitrary-sql") return runSqlResult(query.sql, query.params ?? []);

package/dist/workers-duckdb.mjs

@@ -10,12 +10,12 @@ function resolveSvc(env) {
 }
 const DUCKDB_RPC_TIMEOUT_MS = 22e3;
 const WORKER_R2_MAX_FILES = 96;
-const WORKER_R2_MAX_BYTES = 64 * 1024 * 1024;
+const WORKER_R2_MAX_BYTES = 67108864;
 const WORKER_R2_DECODE_CONCURRENCY = 2;
 const WORKER_R2_HEAD_CONCURRENCY = 4;
-const IPC_CHUNK_BUDGET = 8 * 1024 * 1024;
-const IPC_DIRECT_CALL_BUDGET = 30 * 1024 * 1024;
-const IPC_STAGED_TOTAL_BUDGET = 64 * 1024 * 1024;
+const IPC_CHUNK_BUDGET = 8388608;
+const IPC_DIRECT_CALL_BUDGET = 31457280;
+const IPC_STAGED_TOTAL_BUDGET = 67108864;
 var DuckDBServiceTimeoutError = class extends Error {
 	name = "DuckDBServiceTimeoutError";
 	constructor(timeoutMs) {
@@ -94,7 +94,7 @@ function assertWorkerReadBudget(opts) {
 	}
 	if (totalBytes > maxBytes) throw new Error(`createDucklingsExecutor: planned read spans ${totalBytes} bytes, exceeding the ${maxBytes} byte Worker budget. Narrow the date range or route through a background/windowed query.`);
 }
-const ROW_CACHE_MAX_BYTES = 16 * 1024 * 1024;
+const ROW_CACHE_MAX_BYTES = 16777216;
 function estimateRowsBytes(rows) {
 	if (rows.length === 0) return 0;
 	const cols = Object.keys(rows[0]).length;
@@ -297,7 +297,8 @@ function createDucklingsExecutor(env, opts = {}) {
 			});
 			const merged = [];
 			for (const rows of perFile) merged.push(...rows);
-			const chunks = rowsToArrowIPCChunks(merged, SCHEMAS[placeholderTables?.[placeholder] ?? table]?.columns, {
+			const placeholderTable = placeholderTables?.[placeholder] ?? table;
+			const chunks = rowsToArrowIPCChunks(merged, SCHEMAS[placeholderTable]?.columns, {
 				maxChunkBytes,
 				placeholder
 			});
@@ -309,11 +310,12 @@ function createDucklingsExecutor(env, opts = {}) {
 			tableChunks[tmp] = chunks;
 		}
 		signal?.throwIfAborted();
-		const finalSql = bindLiterals(sql.replace(READ_PARQUET_PLACEHOLDER, (_, placeholder) => {
+		const rewritten = sql.replace(READ_PARQUET_PLACEHOLDER, (_, placeholder) => {
 			const tmp = tempNames[placeholder];
 			if (!tmp) throw new Error(`createDucklingsExecutor: SQL references {{${placeholder}}} but no fileKeys entry provided`);
 			return tmp;
-		}), params);
+		});
+		const finalSql = bindLiterals(rewritten, params);
 		const canInlineTables = totalIpcBytes <= maxDirectCallBytes && Object.values(tableChunks).every((chunks) => chunks.length === 1);
 		let result;
 		if (canInlineTables) {

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@gscdump/cloudflare",
   "type": "module",
-  "version": "2.1.0",
+  "version": "2.2.0",
   "description": "Cloudflare-Workers-flavored helpers for the gscdump analytics stack: AnalyticsEnv binding contract, R2 SigV4 presigner, DuckDB Workers shims, engine factory.",
   "author": {
     "name": "Harlan Wilton",
@@ -47,10 +47,10 @@
   },
   "dependencies": {
     "@uwdata/flechette": "^2.5.0",
-    "@gscdump/contracts": "^2.1.0",
-    "@gscdump/engine": "^2.1.0",
-    "@gscdump/engine-sqlite": "^2.1.0",
-    "gscdump": "^2.1.0"
+    "@gscdump/contracts": "^2.2.0",
+    "@gscdump/engine": "^2.2.0",
+    "@gscdump/engine-sqlite": "^2.2.0",
+    "gscdump": "^2.2.0"
   },
   "devDependencies": {
     "@cloudflare/vitest-pool-workers": "^0.18.8",