@gscdump/lakehouse 0.37.3 → 0.37.5

package/dist/_chunks/catalog.d.mts

@@ -11,6 +11,8 @@ interface CatalogCache {
    * inline so it is never cut off when the response returns.
    */
   defer?: (write: Promise<unknown>) => void;
+  /** Optional warning sink for best-effort cache driver failures. */
+  onError?: (operation: 'get' | 'set' | 'remove', key: string, error: unknown) => void;
 }
 /**
  * Read a cached value. Returns `undefined` on a miss, an expired entry, a
@@ -24,8 +26,8 @@ declare function cacheGet<T>(cache: CatalogCache, key: string, now: number): Pro
  * Returns the write promise. With a `defer` hook the write is handed to the
  * hook and a resolved promise is returned (the response is not blocked on it);
  * without one the write promise is returned for the caller to await, so a
- * fire-and-forget put is never silently dropped. Driver errors are swallowed —
- * a failed cache write must not fail the read.
+ * fire-and-forget put is never silently dropped. Driver errors are reported
+ * through `onError` (or `console.warn`) but do not fail the read.
  */
 declare function cachePut<T>(cache: CatalogCache, key: string, value: T, ttlMs: number, now: number): Promise<void>;
 /** Minimal shape of an icebird manifest-list `partitions` field-summary. */

package/dist/_chunks/catalog.mjs

@@ -11,8 +11,19 @@ function encodeJsonBigintSafe(value) {
 function coerceBigIntToNumber(value) {
 	return typeof value === "bigint" ? Number(value) : value;
 }
+function reportCatalogCacheError(cache, operation, key, error) {
+	try {
+		if (cache.onError) cache.onError(operation, key, error);
+		else console.warn(`[gscdump/lakehouse] cache ${operation} failed for ${key}`, error);
+	} catch (reportError) {
+		console.warn(`[gscdump/lakehouse] cache error reporter failed during ${operation} for ${key}`, reportError);
+	}
+}
 async function cacheGet(cache, key, now) {
-	const boxed = await cache.storage.getItem(key).catch(() => null);
+	const boxed = await cache.storage.getItem(key).catch((error) => {
+		reportCatalogCacheError(cache, "get", key, error);
+		return null;
+	});
 	if (!boxed || typeof boxed.exp !== "number" || boxed.exp <= now) return void 0;
 	return boxed.v;
 }
@@ -21,7 +32,9 @@ function cachePut(cache, key, value, ttlMs, now) {
 		v: value,
 		exp: now + ttlMs
 	};
-	const write = cache.storage.setItem(key, boxed, { ttl: Math.ceil(ttlMs / 1e3) }).catch(() => {});
+	const write = cache.storage.setItem(key, boxed, { ttl: Math.ceil(ttlMs / 1e3) }).catch((error) => {
+		reportCatalogCacheError(cache, "set", key, error);
+	});
 	if (cache.defer) {
 		cache.defer(write);
 		return Promise.resolve();
@@ -257,7 +270,10 @@ function snapshotRefKey(scope, namespace, table) {
 	return `lh-snapref\0${scope}\0${namespace}\0${table}`;
 }
 async function invalidateSnapshotRef(cache, namespace, table, cacheScope = "") {
-	await cache.storage.removeItem(snapshotRefKey(cacheScope, namespace, table)).catch(() => {});
+	const key = snapshotRefKey(cacheScope, namespace, table);
+	await cache.storage.removeItem(key).catch((error) => {
+		reportCatalogCacheError(cache, "remove", key, error);
+	});
 }
 function metadataRefKey(scope, namespace, table, snapshotId) {
 	return `lh-snapmeta\0${scope}\0${namespace}\0${table}\0${snapshotId}`;

package/dist/index.d.mts

@@ -192,6 +192,14 @@ interface RunPyIcebergWriterOptions {
   processErrorAsParseFailure?: boolean;
   rejectOnProcessError?: boolean;
 }
+/**
+ * @deprecated Import the engine-owned PyIceberg runtime instead. Retained as a
+ * compatibility shim for consumers of `@gscdump/lakehouse` 0.x.
+ */
 declare function resolvePyIcebergPython(override?: string): string;
+/**
+ * @deprecated Import the engine-owned PyIceberg runtime instead. Retained as a
+ * compatibility shim for consumers of `@gscdump/lakehouse` 0.x.
+ */
 declare function runPyIcebergWriter<T extends PyIcebergWriterResult>(options: RunPyIcebergWriterOptions): Promise<T>;
 export { type AppendCommitOptions, type AppendResult, type AppendSink, type AppendSinkCloseResult, type AppendSinkOptions, type CatalogCache, type CommitRetryOptions, type ConnectIcebergOptions, DEFAULT_PARTITION_KEY_ENCODING, type DatasetDim, type DatasetIdentity, ICEBERG_TYPE_MAP, type IcebergCatalogConfig, type IcebergColumn, type IcebergColumnType, type IcebergConnection, type IcebergDataset, type IcebergDatasetColumnDef, type IcebergDatasetDef, type IcebergDatasetLedger, type IcebergFieldSummary, type IcebergListedDataFile, type IcebergPartitionField, type IcebergPartitionSpec, type IcebergPartitionSpecField, type IcebergPartitionTransform, type IcebergPrimitiveType, type IcebergS3Config, type IcebergSchema, type IcebergSchemaField, type IcebergSortOrder, type IcebergSortOrderField, type IcebergTableOpResult, type IcebergTableSpec, type ManifestPartitionFilter, type PartitionKeyEncoding, type PartitionValueMatch, type PyIcebergWriterResult, type QueryProfiler, type ResolveDataFilesOptions, type ResolveIcebergDataFilesOptions, type RunPyIcebergWriterOptions, bigintJsonReplacer, buildManifestPartitionFilter, cacheGet, cachePut, catalogCacheScope, coerceBigIntToNumber, connectIcebergCatalog, defineIcebergDataset, deriveTableSpec, dropIcebergTables, encodeJsonBigintSafe, ensureIcebergNamespace, invalidateSnapshotRef, listIcebergTables, resolveIcebergDataFiles, resolvePyIcebergPython, runPyIcebergWriter, stringifyBigintSafe, toIcebergDayCount };

package/dist/index.mjs

@@ -1,6 +1,5 @@
 import { ICEBERG_TYPE_MAP, bigintJsonReplacer, buildManifestPartitionFilter, cacheGet, cachePut, catalogCacheScope, coerceBigIntToNumber, connectIcebergCatalog, dropIcebergTables, encodeJsonBigintSafe, ensureIcebergNamespace, icebergAppendRetrying, invalidateSnapshotRef, listIcebergTables, resolveIcebergDataFiles, stringifyBigintSafe } from "./_chunks/catalog.mjs";
 import { icebergCreateTable } from "./_chunks/libs/icebird.mjs";
-import process from "node:process";
 const INT32_MIN = -2147483648;
 const INT32_MAX = 2147483647;
 const DAY_MILLIS = 864e5;
@@ -322,11 +321,15 @@ function defineIcebergDataset(def) {
 		resolveDataFiles
 	};
 }
+function importRuntimeModule(specifier) {
+	return import(specifier);
+}
 function resolvePyIcebergPython(override) {
-	return override ?? process.env["GSCDUMP_ICEBERG_PYTHON"] ?? "python3";
+	const env = Reflect.get(globalThis, "process")?.env;
+	return override ?? env?.["GSCDUMP_ICEBERG_PYTHON"] ?? "python3";
 }
 async function runPyIcebergWriter(options) {
-	const { execFile } = await import("node:child_process");
+	const { execFile } = await importRuntimeModule("node:child_process");
 	return new Promise((resolve, reject) => {
 		execFile(options.python, [options.script], { maxBuffer: 64 * 1024 * 1024 }, (err, stdout, stderr) => {
 			let parsed;

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@gscdump/lakehouse",
   "type": "module",
-  "version": "0.37.3",
+  "version": "0.37.5",
   "description": "Dataset-agnostic Iceberg lakehouse layer + dataset registry for R2 Data Catalog producers (ADR-0021).",
   "author": {
     "name": "Harlan Wilton",