@gscdump/lakehouse 1.3.2 → 1.4.1

package/dist/{_chunks/libs/unstorage.d.mts → node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/dist/shared/unstorage.DqlWKU2I.d.mts}

@@ -1,7 +1,3 @@
-import "./db0.mjs";
-import "./chokidar.mjs";
-import "./lru-cache.mjs";
-import "./ioredis.mjs";
 type StorageValue = null | string | number | boolean | object;
 type WatchEvent = "update" | "remove";
 type WatchCallback = (event: WatchEvent, key: string) => any;
@@ -61,7 +57,7 @@ type StorageDefinition = {
 };
 type StorageItemMap<T> = T extends StorageDefinition ? T["items"] : T;
 type StorageItemType<T, K> = K extends keyof StorageItemMap<T> ? StorageItemMap<T>[K] : T extends StorageDefinition ? StorageValue : T;
-interface Storage$1<T extends StorageValue = StorageValue> {
+interface Storage<T extends StorageValue = StorageValue> {
   hasItem<U extends Extract<T, StorageDefinition>, K extends keyof StorageItemMap<U>>(key: K, opts?: TransactionOptions): Promise<boolean>;
   hasItem(key: string, opts?: TransactionOptions): Promise<boolean>;
   getItem<U extends Extract<T, StorageDefinition>, K extends string & keyof StorageItemMap<U>>(key: K, ops?: TransactionOptions): Promise<StorageItemType<T, K> | null>;
@@ -102,7 +98,7 @@ interface Storage$1<T extends StorageValue = StorageValue> {
   dispose: () => Promise<void>;
   watch: (callback: WatchCallback) => Promise<Unwatch>;
   unwatch: () => Promise<void>;
-  mount: (base: string, driver: Driver) => Storage$1;
+  mount: (base: string, driver: Driver) => Storage;
   unmount: (base: string, dispose?: boolean) => Promise<void>;
   getMount: (key?: string) => {
     base: string;
@@ -114,11 +110,11 @@ interface Storage$1<T extends StorageValue = StorageValue> {
     base: string;
     driver: Driver;
   }[];
-  keys: Storage$1["getKeys"];
-  get: Storage$1<T>["getItem"];
-  set: Storage$1<T>["setItem"];
-  has: Storage$1<T>["hasItem"];
-  del: Storage$1<T>["removeItem"];
-  remove: Storage$1<T>["removeItem"];
+  keys: Storage["getKeys"];
+  get: Storage<T>["getItem"];
+  set: Storage<T>["setItem"];
+  has: Storage<T>["hasItem"];
+  del: Storage<T>["removeItem"];
+  remove: Storage<T>["removeItem"];
 }
-export { Storage$1 as Storage };
+export type { Driver as D, GetKeysOptions as G, StorageValue as S, TransactionOptions as T, Unwatch as U, WatchCallback as W, Storage as a, DriverFlags as b, StorageMeta as c, WatchEvent as d };

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/dist/shared/unstorage2.DqlWKU2I.d.mts

@@ -0,0 +1 @@
+export { };

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/azure-app-configuration.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/azure-cosmos.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/azure-key-vault.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/azure-storage-blob.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/azure-storage-table.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/capacitor-preferences.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/cloudflare-kv-binding.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/cloudflare-kv-http.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/cloudflare-r2-binding.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/db0.d.mts

@@ -0,0 +1,2 @@
+import "../../../../db0@0.3.4_drizzle-orm@1.0.0-rc.3_@cloudflare_workers-types@5.20260722.1_valibot@1.4.2_typescript@7.0.2__zod@4.4.3_/node_modules/db0/dist/index.mjs";
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/deno-kv-node.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/deno-kv.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/fs-lite.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/fs.d.mts

@@ -0,0 +1,2 @@
+import "../../../../chokidar@5.0.0/node_modules/chokidar/index.mjs";
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/github.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/http.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/indexedb.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/localstorage.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/localstorage2.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/lru-cache.d.mts

@@ -0,0 +1,2 @@
+import "../../../../lru-cache@11.5.0/node_modules/lru-cache/dist/esm/index.mjs";
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/mongodb.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/netlify-blobs.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/overlay.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/planetscale.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/redis.d.mts

@@ -0,0 +1,2 @@
+import "../../../../ioredis@5.10.1_supports-color@10.2.2/node_modules/ioredis/built/index.mjs";
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/s3.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/session-storage.d.mts

@@ -0,0 +1,2 @@
+import "./localstorage2.mjs";
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/uploadthing.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/upstash.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/vercel-blob.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/vercel-kv.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/node_modules/.pnpm/unstorage@1.17.5_aws4fetch@1.0.20_db0@0.3.4_ioredis@5.10.1_supports-color@10.2.2_/node_modules/unstorage/drivers/vercel-runtime-cache.d.mts

@@ -0,0 +1 @@
+import "../dist/index.mjs";

package/dist/orphan-sweep.d.mts

@@ -0,0 +1,94 @@
+import { IcebergConnection } from "./catalog.mjs";
+/** One listed object under a data prefix. */
+interface SweepListedObject {
+  key: string;
+  uploaded: Date;
+}
+/** One page of a prefix listing. */
+interface SweepListPage {
+  objects: readonly SweepListedObject[];
+  truncated: boolean;
+  cursor?: string;
+}
+/**
+ * Minimal storage client `sweepUncommittedOrphans` needs: list-by-prefix +
+ * batch delete against the warehouse bucket. Deliberately NOT an S3 SDK
+ * client shape — this interface is structurally satisfied by Cloudflare's
+ * native `R2Bucket` Workers binding (`env.R2_ANALYTICS.list({ prefix })` /
+ * `env.R2_ANALYTICS.delete(keys)`), so a caller running inside a Worker can
+ * pass the binding straight through with no adapter. Any other
+ * S3-compatible client can be wrapped to this shape.
+ */
+interface SweepStorageClient {
+  list: (options: {
+    prefix: string;
+    cursor?: string;
+  }) => Promise<SweepListPage>;
+  delete: (keys: string[]) => Promise<void>;
+}
+interface SweepUncommittedOrphansOptions {
+  conn: IcebergConnection;
+  s3: SweepStorageClient;
+  /**
+   * Bare R2 bucket name backing `conn`'s warehouse. Every scanned table's
+   * `location` must resolve into this exact bucket; a table whose location
+   * points elsewhere is SKIPPED — never guessed at, never touched.
+   */
+  bucket: string;
+  /**
+   * Restrict the sweep to these tables; defaults to every table currently
+   * in `conn.namespace` (`conn` already carries the namespace scope, so
+   * there's no separate `namespace` field here).
+   */
+  tables?: readonly string[];
+  /** Only delete objects older than this many hours. Default 48. */
+  graceHours?: number;
+  /** Hard cap on deletions in one run. Default 500. */
+  maxDeletes?: number;
+  /** Compute the delete set but never call `s3.delete`. Default false. */
+  dryRun?: boolean;
+  /** Injectable clock (ms epoch). Defaults to `Date.now`. */
+  now?: () => number;
+  /** Maximum catalog/object-store reads held in flight. Default 4, max 32. */
+  ioConcurrency?: number;
+}
+interface SweepUncommittedOrphansResult {
+  /** Tables whose retained snapshots were successfully walked. */
+  scannedTables: string[];
+  /** Tables skipped because their `location` didn't resolve into `bucket` — never scanned, never touched. */
+  skippedTables: string[];
+  /** Count of live (referenced-by-some-retained-snapshot) files found across `scannedTables`. */
+  liveFileCount: number;
+  /** Objects under a scanned table's data prefix with no snapshot reference anywhere, past `graceHours`. */
+  candidateCount: number;
+  /** Keys deleted (or, under `dryRun`, that WOULD be deleted). */
+  deleted: string[];
+  dryRun: boolean;
+  /** True when `candidateCount` exceeded `maxDeletes` — remaining orphans are left for the next run. */
+  cappedAtLimit: boolean;
+  /** Set when the sweep did nothing because the table scope listed zero tables (fail-closed — see class docs). */
+  reason?: 'zero-tables';
+}
+/**
+ * Sweep data files written under a table's `data/` prefix that were NEVER
+ * referenced by any snapshot the catalog still retains — a crashed/partial
+ * writer's leftovers. This is the one orphan class Cloudflare's own R2 Data
+ * Catalog snapshot-expiration GC does not cover.
+ *
+ * HARD SAFETY:
+ * - A listing failure (table load / manifest walk throwing) PROPAGATES.
+ *   It is never caught and treated as "no live files" — that would make
+ *   every real data file look orphaned. Callers must let this reject and
+ *   must not proceed to delete on catch.
+ * - Zero tables in scope is treated as "couldn't determine what's live",
+ *   not "nothing to protect" — the sweep no-ops with `reason: 'zero-tables'`
+ *   rather than listing/deleting anything.
+ * - Deletions are capped at `maxDeletes` per run (default 500).
+ * - `dryRun` computes the exact candidate set without ever calling `s3.delete`.
+ * - Only ever lists/deletes keys under a scanned table's own
+ *   `<table.location>/data/` prefix, and every candidate is re-checked to
+ *   actually start with that prefix before being considered for deletion —
+ *   never a bare bucket-root listing, never a key outside a data prefix.
+ */
+declare function sweepUncommittedOrphans(opts: SweepUncommittedOrphansOptions): Promise<SweepUncommittedOrphansResult>;
+export { SweepListPage, SweepListedObject, SweepStorageClient, SweepUncommittedOrphansOptions, SweepUncommittedOrphansResult, sweepUncommittedOrphans };

package/dist/orphan-sweep.mjs

@@ -0,0 +1,141 @@
+import { restCatalogListTables, restCatalogLoadTable } from "./node_modules/.pnpm/icebird@0.8.15_patch_hash_d1902372376af18885e7c7dce807f1ecab86c2f86a9d1ff65d51e3b63b64b057/node_modules/icebird/src/catalog/rest.mjs";
+import { icebergManifests } from "./node_modules/.pnpm/icebird@0.8.15_patch_hash_d1902372376af18885e7c7dce807f1ecab86c2f86a9d1ff65d51e3b63b64b057/node_modules/icebird/src/manifest.mjs";
+import "./node_modules/.pnpm/icebird@0.8.15_patch_hash_d1902372376af18885e7c7dce807f1ecab86c2f86a9d1ff65d51e3b63b64b057/node_modules/icebird/src/index.mjs";
+const DEFAULT_GRACE_HOURS = 48;
+const DEFAULT_MAX_DELETES = 500;
+const DEFAULT_IO_CONCURRENCY = 4;
+const MAX_IO_CONCURRENCY = 32;
+const HOUR_MS = 3600 * 1e3;
+function createIoLimiter(requested) {
+	const concurrency = typeof requested === "number" && Number.isFinite(requested) ? Math.max(1, Math.min(MAX_IO_CONCURRENCY, Math.floor(requested))) : DEFAULT_IO_CONCURRENCY;
+	let active = 0;
+	const waiters = [];
+	return async (operation) => {
+		if (active >= concurrency) await new Promise((resolve) => waiters.push(resolve));
+		active++;
+		try {
+			return await operation();
+		} finally {
+			active--;
+			waiters.shift()?.();
+		}
+	};
+}
+async function settleOrThrow(promises) {
+	const settled = await Promise.allSettled(promises);
+	const failure = settled.find((result) => result.status === "rejected");
+	if (failure) throw failure.reason;
+	return settled.map((result) => result.value);
+}
+function splitBucketKey(filePath) {
+	if (!filePath.startsWith("s3://")) return null;
+	const rest = filePath.slice(5);
+	const slash = rest.indexOf("/");
+	if (slash < 0) return null;
+	return {
+		bucket: rest.slice(0, slash),
+		key: rest.slice(slash + 1)
+	};
+}
+async function scanTable(conn, table, bucket, io) {
+	const { metadata } = await io(() => restCatalogLoadTable(conn.catalog, {
+		namespace: conn.namespace,
+		table
+	}));
+	const location = splitBucketKey(String(metadata.location ?? ""));
+	if (!location || location.bucket !== bucket) return {
+		dataPrefix: null,
+		liveKeys: /* @__PURE__ */ new Set()
+	};
+	const dataPrefix = `${location.key.replace(/\/+$/, "")}/data/`;
+	const liveKeys = /* @__PURE__ */ new Set();
+	const manifestsBySnapshot = await settleOrThrow((metadata.snapshots ?? []).map((snapshot) => io(() => icebergManifests({
+		metadata,
+		resolver: conn.resolver,
+		snapshotId: snapshot["snapshot-id"]
+	}))));
+	for (const manifests of manifestsBySnapshot) for (const manifest of manifests) for (const entry of manifest.entries) {
+		if (entry.status === 2) continue;
+		const filePath = entry.data_file?.file_path;
+		if (!filePath) continue;
+		const parsed = splitBucketKey(filePath);
+		if (parsed && parsed.bucket === bucket) liveKeys.add(parsed.key);
+	}
+	return {
+		dataPrefix,
+		liveKeys
+	};
+}
+async function listAllUnderPrefix(s3, prefix, io) {
+	const out = [];
+	let cursor;
+	do {
+		const page = await io(() => s3.list({
+			prefix,
+			cursor
+		}));
+		out.push(...page.objects);
+		cursor = page.truncated ? page.cursor : void 0;
+	} while (cursor);
+	return out;
+}
+async function sweepUncommittedOrphans(opts) {
+	const { conn, s3, bucket } = opts;
+	const graceHours = opts.graceHours ?? DEFAULT_GRACE_HOURS;
+	const maxDeletes = opts.maxDeletes ?? DEFAULT_MAX_DELETES;
+	const dryRun = opts.dryRun ?? false;
+	const graceCutoff = (opts.now ?? Date.now)() - graceHours * HOUR_MS;
+	const io = createIoLimiter(opts.ioConcurrency);
+	const tables = opts.tables ?? (await io(() => restCatalogListTables(conn.catalog, { namespace: conn.namespace }))).map((t) => t.name);
+	if (tables.length === 0) return {
+		scannedTables: [],
+		skippedTables: [],
+		liveFileCount: 0,
+		candidateCount: 0,
+		deleted: [],
+		dryRun,
+		cappedAtLimit: false,
+		reason: "zero-tables"
+	};
+	const tableResults = await settleOrThrow(tables.map(async (table) => {
+		const { dataPrefix, liveKeys } = await scanTable(conn, table, bucket, io);
+		if (!dataPrefix) return {
+			table,
+			skipped: true,
+			liveFileCount: 0,
+			candidates: []
+		};
+		const listed = await listAllUnderPrefix(s3, dataPrefix, io);
+		const tableCandidates = [];
+		for (const obj of listed) {
+			if (!obj.key.startsWith(dataPrefix)) continue;
+			if (liveKeys.has(obj.key)) continue;
+			if (obj.uploaded.getTime() > graceCutoff) continue;
+			tableCandidates.push(obj);
+		}
+		return {
+			table,
+			skipped: false,
+			liveFileCount: liveKeys.size,
+			candidates: tableCandidates
+		};
+	}));
+	const scannedTables = tableResults.filter((result) => !result.skipped).map((result) => result.table);
+	const skippedTables = tableResults.filter((result) => result.skipped).map((result) => result.table);
+	const liveFileCount = tableResults.reduce((sum, result) => sum + result.liveFileCount, 0);
+	const candidates = tableResults.flatMap((result) => result.candidates);
+	const candidateCount = candidates.length;
+	const toDelete = candidates.slice(0, maxDeletes);
+	const cappedAtLimit = candidateCount > maxDeletes;
+	if (!dryRun && toDelete.length > 0) await s3.delete(toDelete.map((o) => o.key));
+	return {
+		scannedTables,
+		skippedTables,
+		liveFileCount,
+		candidateCount,
+		deleted: toDelete.map((o) => o.key),
+		dryRun,
+		cappedAtLimit
+	};
+}
+export { sweepUncommittedOrphans };

package/dist/partition-prune.d.mts

@@ -0,0 +1,35 @@
+import { IcebergPartitionField } from "./schema.mjs";
+/** Minimal shape of an icebird manifest-list `partitions` field-summary. */
+interface IcebergFieldSummary {
+  contains_null: boolean;
+  contains_nan?: boolean | null;
+  lower_bound?: Uint8Array | null;
+  upper_bound?: Uint8Array | null;
+}
+/** Predicate handed to icebird's patched `icebergManifests({ partitionFilter })`. */
+type ManifestPartitionFilter = (partitions: IcebergFieldSummary[] | undefined) => boolean;
+/**
+ * One identity/dims value to prune an `identity`-transform partition field by.
+ *
+ * `'int32'`-encoded fields are deliberately NOT pruned here (mirrors the
+ * original engine behavior): an INT column's bound bytes are a 4-byte int, but
+ * per-team catalogs are single-tenant, so identity pruning on them saves ~nothing
+ * — the per-file partition check in the resolver remains the authoritative
+ * correctness filter. Only `'string'`-encoded identity fields are pruned by
+ * lexicographic UTF-8 bound comparison (the truncated-string-stats case R2 SQL
+ * needs the workaround for).
+ */
+interface PartitionValueMatch {
+  /** Partition field name as declared in the partition spec (e.g. `'site_id'`). */
+  field: string;
+  value: string | number;
+  encoding: 'string' | 'int32';
+}
+/**
+ * Build the `partitionFilter` predicate for a slice of `matches` (identity/dims
+ * value equality) plus an optional `wantedMonths` set (for a `month`-transform
+ * field in `partitionSpec`). Returns `false` to skip a manifest, `true` to keep
+ * it. Keep-all when a manifest carries no `partitions` summaries.
+ */
+declare function buildManifestPartitionFilter(partitionSpec: readonly IcebergPartitionField[], matches: readonly PartitionValueMatch[], wantedMonths?: ReadonlySet<number>): ManifestPartitionFilter;
+export { IcebergFieldSummary, ManifestPartitionFilter, PartitionValueMatch, buildManifestPartitionFilter };

package/dist/partition-prune.mjs

@@ -0,0 +1,57 @@
+function toUint8(bytes) {
+	if (bytes == null) return null;
+	return bytes instanceof Uint8Array ? bytes : new Uint8Array(bytes);
+}
+const UTF8_DECODER = new TextDecoder();
+function decodeString(bytes) {
+	const u = toUint8(bytes);
+	return u == null ? null : UTF8_DECODER.decode(u);
+}
+function decodeMonthInt(bytes) {
+	const u = toUint8(bytes);
+	if (u == null) return null;
+	return new DataView(u.buffer, u.byteOffset, u.byteLength).getInt32(0, true);
+}
+function buildManifestPartitionFilter(partitionSpec, matches, wantedMonths) {
+	const fieldIndex = (name) => partitionSpec.findIndex((f) => f.name === name || f.sourceColumn === name);
+	const monthFieldIndex = partitionSpec.findIndex((f) => f.transform === "month");
+	const stringMatches = matches.flatMap((match) => {
+		if (match.encoding !== "string") return [];
+		const index = fieldIndex(match.field);
+		return index < 0 ? [] : [{
+			index,
+			value: String(match.value)
+		}];
+	});
+	const wantedMonthValues = wantedMonths ? [...wantedMonths].filter(Number.isFinite).sort((a, b) => a - b) : [];
+	const hasWantedMonthInRange = (lo, hi) => {
+		let left = 0;
+		let right = wantedMonthValues.length;
+		while (left < right) {
+			const middle = left + right >>> 1;
+			if (wantedMonthValues[middle] < lo) left = middle + 1;
+			else right = middle;
+		}
+		return left < wantedMonthValues.length && wantedMonthValues[left] <= hi;
+	};
+	return (partitions) => {
+		if (!partitions || partitions.length === 0) return true;
+		for (const match of stringMatches) {
+			const summary = partitions[match.index];
+			if (!summary || summary.lower_bound == null && summary.upper_bound == null) continue;
+			const lo = decodeString(summary.lower_bound);
+			const hi = decodeString(summary.upper_bound);
+			if (lo != null && hi != null && (match.value < lo || match.value > hi)) return false;
+		}
+		if (wantedMonthValues.length > 0 && monthFieldIndex >= 0) {
+			const monthSummary = partitions[monthFieldIndex];
+			if (monthSummary && (monthSummary.lower_bound != null || monthSummary.upper_bound != null)) {
+				const lo = decodeMonthInt(monthSummary.lower_bound);
+				const hi = decodeMonthInt(monthSummary.upper_bound);
+				if (lo != null && hi != null && !hasWantedMonthInRange(lo, hi)) return false;
+			}
+		}
+		return true;
+	};
+}
+export { buildManifestPartitionFilter };

package/dist/provisioning/allocator.d.mts

@@ -0,0 +1,44 @@
+/**
+ * Team-scoped Catalog Site Id allocator (ADR-0021 amendment 11).
+ *
+ * A "Catalog Site Id" is the small INT32 join key every dataset's `'site-int'`
+ * identity partitions on (nuxtseo ADR-0091's `gscdumpSiteIntId` / gscdump's
+ * `user_sites.int_id`). One allocator, owned here, so no consumer hand-rolls
+ * its own next-int sequence (the exact copy-paste class ADR-0021 exists to
+ * close). Pure — the caller persists the result; this module has no DB.
+ *
+ * Two entry points into a team's id space:
+ *  - ADOPT a `supplied` id (e.g. gscdump's existing `int_id` for a site that
+ *    predates the allocator) — validated for TEAM-SCOPE uniqueness, never
+ *    silently reassigned.
+ *  - ALLOCATE fresh — next-int strictly above every existing id in the team
+ *    (adopted ids included), so a later allocation can never collide with an
+ *    earlier adoption.
+ */
+type AllocateCatalogSiteIdResult = {
+  _tag: 'allocated';
+  id: number;
+} | {
+  _tag: 'adopted';
+  id: number;
+} | {
+  _tag: 'collision';
+  id: number;
+};
+interface AllocateCatalogSiteIdOptions {
+  /** Every Catalog Site Id already in use for this team (adopted + previously allocated). */
+  existingIds: Iterable<number>;
+  /**
+   * Adopt this id instead of allocating a fresh one (e.g. gscdump's existing
+   * `int_id`). Validated against `existingIds` for team-scope uniqueness —
+   * a collision is returned as a value, never silently resolved.
+   */
+  supplied?: number;
+}
+/**
+ * Adopt `supplied`, or allocate the next unused int strictly above every id
+ * already in `existingIds`. Never mutates the caller's set; never reuses an
+ * id already present.
+ */
+declare function allocateCatalogSiteId(opts: AllocateCatalogSiteIdOptions): AllocateCatalogSiteIdResult;
+export { AllocateCatalogSiteIdOptions, AllocateCatalogSiteIdResult, allocateCatalogSiteId };

package/dist/provisioning/allocator.mjs

@@ -0,0 +1,24 @@
+const INT32_MAX = 2147483647;
+function allocateCatalogSiteId(opts) {
+	const existing = new Set(opts.existingIds);
+	if (opts.supplied != null) {
+		if (existing.has(opts.supplied)) return {
+			_tag: "collision",
+			id: opts.supplied
+		};
+		return {
+			_tag: "adopted",
+			id: opts.supplied
+		};
+	}
+	let max = 0;
+	for (const id of existing) if (id > max) max = id;
+	let next = max + 1;
+	while (existing.has(next)) next++;
+	if (next > INT32_MAX) throw new RangeError(`allocateCatalogSiteId: exhausted the INT32 id space (next=${next})`);
+	return {
+		_tag: "allocated",
+		id: next
+	};
+}
+export { allocateCatalogSiteId };

package/dist/provisioning/index.d.mts

@@ -1,47 +1,5 @@
-import { PartitionKeyEncoding } from "../_chunks/schema.mjs";
-/**
- * Team-scoped Catalog Site Id allocator (ADR-0021 amendment 11).
- *
- * A "Catalog Site Id" is the small INT32 join key every dataset's `'site-int'`
- * identity partitions on (nuxtseo ADR-0091's `gscdumpSiteIntId` / gscdump's
- * `user_sites.int_id`). One allocator, owned here, so no consumer hand-rolls
- * its own next-int sequence (the exact copy-paste class ADR-0021 exists to
- * close). Pure — the caller persists the result; this module has no DB.
- *
- * Two entry points into a team's id space:
- *  - ADOPT a `supplied` id (e.g. gscdump's existing `int_id` for a site that
- *    predates the allocator) — validated for TEAM-SCOPE uniqueness, never
- *    silently reassigned.
- *  - ALLOCATE fresh — next-int strictly above every existing id in the team
- *    (adopted ids included), so a later allocation can never collide with an
- *    earlier adoption.
- */
-type AllocateCatalogSiteIdResult = {
-  _tag: 'allocated';
-  id: number;
-} | {
-  _tag: 'adopted';
-  id: number;
-} | {
-  _tag: 'collision';
-  id: number;
-};
-interface AllocateCatalogSiteIdOptions {
-  /** Every Catalog Site Id already in use for this team (adopted + previously allocated). */
-  existingIds: Iterable<number>;
-  /**
-   * Adopt this id instead of allocating a fresh one (e.g. gscdump's existing
-   * `int_id`). Validated against `existingIds` for team-scope uniqueness —
-   * a collision is returned as a value, never silently resolved.
-   */
-  supplied?: number;
-}
-/**
- * Adopt `supplied`, or allocate the next unused int strictly above every id
- * already in `existingIds`. Never mutates the caller's set; never reuses an
- * id already present.
- */
-declare function allocateCatalogSiteId(opts: AllocateCatalogSiteIdOptions): AllocateCatalogSiteIdResult;
+import { PartitionKeyEncoding } from "../schema.mjs";
+import { AllocateCatalogSiteIdOptions, AllocateCatalogSiteIdResult, allocateCatalogSiteId } from "./allocator.mjs";
 interface CloudflareProvisionCreds {
   accountId: string;
   /** Account-scoped management API token (bucket CRUD). */

package/dist/provisioning/index.mjs

@@ -1,26 +1,4 @@
-const INT32_MAX = 2147483647;
-function allocateCatalogSiteId(opts) {
-	const existing = new Set(opts.existingIds);
-	if (opts.supplied != null) {
-		if (existing.has(opts.supplied)) return {
-			_tag: "collision",
-			id: opts.supplied
-		};
-		return {
-			_tag: "adopted",
-			id: opts.supplied
-		};
-	}
-	let max = 0;
-	for (const id of existing) if (id > max) max = id;
-	let next = max + 1;
-	while (existing.has(next)) next++;
-	if (next > INT32_MAX) throw new RangeError(`allocateCatalogSiteId: exhausted the INT32 id space (next=${next})`);
-	return {
-		_tag: "allocated",
-		id: next
-	};
-}
+import { allocateCatalogSiteId } from "./allocator.mjs";
 const CF_API_BASE = "https://api.cloudflare.com/client/v4";
 const COMPACTION_TARGET_SIZE_MB = "128";
 const SNAPSHOT_MIN_SNAPSHOTS_TO_KEEP = 10;