@gscdump/engine 3.5.0 → 3.6.0

package/dist/entities/inspection.d.mts

@@ -7,7 +7,7 @@ import { CanonicalDifferenceKind } from "gscdump";
  *  `searchconsole_v1.Schema$UrlInspectionResult` shape but as plain JSON
  *  so storage doesn't depend on the googleapis type tree.
  */
-interface InspectionRecord {
+export interface InspectionRecord {
   url: string;
   /** ISO-8601 timestamp of when we ran the inspection. */
   inspectedAt: string;
@@ -43,7 +43,7 @@ interface InspectionRecord {
   };
 }
 /** Wire shape persisted to disk/R2. */
-interface InspectionIndex {
+export interface InspectionIndex {
   version: 1;
   /** Map of urlHash → InspectionRecord (latest only). */
   records: Record<string, InspectionRecord>;
@@ -53,7 +53,7 @@ interface InspectionIndex {
  * Keyed by UUID under the month directory — retries write a new blob,
  * never RMW an existing one. Idempotent under job retries.
  */
-interface InspectionHistoryShard {
+export interface InspectionHistoryShard {
   version: 1;
   /** Records persisted in this batch. */
   records: InspectionRecord[];
@@ -70,7 +70,7 @@ interface InspectionHistoryShard {
  * parquet sidecar exists for DuckDB JOIN seams; readers go through
  * `parquetUri`.
  */
-interface InspectionParquetRow {
+export interface InspectionParquetRow {
   [column: string]: string | number | null;
   urlHash: string;
   url: string;
@@ -103,7 +103,7 @@ interface InspectionParquetRow {
  * `firstSeenAt` preservation); every other column is taken from the
  * newest-by-`inspectedAt` event.
  */
-interface InspectionEventRow extends InspectionParquetRow {
+export interface InspectionEventRow extends InspectionParquetRow {
   /** Declared-vs-selected canonical classification, computed at inspection ingest. */
   canonicalMismatchKind: CanonicalDifferenceKind;
   crawlingUserAgent: string | null;
@@ -136,8 +136,8 @@ interface InspectionEventRow extends InspectionParquetRow {
  * batch encodes to ~10 KB so the cap is purely defensive against future
  * batch-size bumps.
  */
-declare const INSPECTION_HISTORY_MAX_BYTES: number;
-interface InspectionStore {
+export declare const INSPECTION_HISTORY_MAX_BYTES: number;
+export interface InspectionStore {
   /**
    * Append a batch of fresh inspection results as an immutable per-batch
    * shard under `history/<YYYY-MM>/<batchId>.json`. Idempotent under job
@@ -233,8 +233,7 @@ interface InspectionStore {
    */
   parquetUri: (ctx: TenantCtx) => string | undefined;
 }
-interface CreateInspectionStoreOptions {
+export interface CreateInspectionStoreOptions {
   dataSource: DataSource;
 }
-declare function createInspectionStore(opts: CreateInspectionStoreOptions): InspectionStore;
-export { CreateInspectionStoreOptions, INSPECTION_HISTORY_MAX_BYTES, InspectionEventRow, InspectionHistoryShard, InspectionIndex, InspectionParquetRow, InspectionRecord, InspectionStore, createInspectionStore };
+export declare function createInspectionStore(opts: CreateInspectionStoreOptions): InspectionStore;

package/dist/entities/query-dim.d.mts

@@ -1,7 +1,7 @@
 import { DataSource } from "../storage.mjs";
 import { queryDimMetaKey, queryDimParquetKey } from "../entity-keys.mjs";
 import { TenantCtx } from "@gscdump/contracts";
-interface QueryDimRecord {
+export interface QueryDimRecord {
   query: string;
   /** Lexical canonical, never empty: NULL/'' folds to the raw query. */
   query_canonical: string;
@@ -11,7 +11,7 @@ interface QueryDimRecord {
   intent_version: number;
 }
 /** JSON sidecar: versions + freshness, readable without decoding the parquet. */
-interface QueryDimMeta {
+export interface QueryDimMeta {
   version: 1;
   builtAt: number;
   rowCount: number;
@@ -23,7 +23,7 @@ interface QueryDimMeta {
  * passes `normalizeQuery` / `classifyIntentCode` (e.g. `encodeIntent ∘
  * classifyQueryIntent`) plus their version constants.
  */
-interface QueryDimDeps {
+export interface QueryDimDeps {
   normalizeQuery: (query: string) => string;
   normalizerVersion: number;
   /** Returns the packed intent code for a raw query. */
@@ -35,8 +35,8 @@ interface QueryDimDeps {
  * folds an empty/whitespace canonical back to the raw query so the dimension
  * is total for read-time joins.
  */
-declare function buildQueryDimRecords(queries: Iterable<string>, deps: QueryDimDeps): QueryDimRecord[];
-interface QueryDimStore {
+export declare function buildQueryDimRecords(queries: Iterable<string>, deps: QueryDimDeps): QueryDimRecord[];
+export interface QueryDimStore {
   parquetKey: (ctx: TenantCtx) => string;
   /** Write the parquet + JSON sidecar. Last-write-wins; no history. */
   write: (ctx: TenantCtx, records: readonly QueryDimRecord[], builtAt: number) => Promise<{
@@ -48,7 +48,7 @@ interface QueryDimStore {
   /** Decode the dimension rows (test/inspection; reads JOIN the parquet by key). */
   loadRecords: (ctx: TenantCtx) => Promise<QueryDimRecord[]>;
 }
-declare function createQueryDimStore({ dataSource }: {
+export declare function createQueryDimStore({ dataSource }: {
   dataSource: DataSource;
 }): QueryDimStore;
-export { QueryDimDeps, QueryDimMeta, QueryDimRecord, QueryDimStore, buildQueryDimRecords, createQueryDimStore, queryDimMetaKey, queryDimParquetKey };
+export { queryDimMetaKey, queryDimParquetKey };

package/dist/entities/sitemap-shared.d.mts

@@ -1,10 +1,10 @@
 import { DataSource } from "../storage.mjs";
 import { Row, TenantCtx } from "@gscdump/contracts";
-interface ParsedUrl {
+export interface ParsedUrl {
   loc: string;
   lastmod?: string;
 }
-interface SitemapUrlRecord {
+export interface SitemapUrlRecord {
   feedpath: string;
   feedpathHash: string;
   urlHash: string;
@@ -14,7 +14,7 @@ interface SitemapUrlRecord {
   lastSeenAt: number;
   removedAt?: number;
 }
-interface SitemapGenerationFeedManifest {
+export interface SitemapGenerationFeedManifest {
   feedpath: string;
   feedpathHash: string;
   baseKey: string;
@@ -24,7 +24,7 @@ interface SitemapGenerationFeedManifest {
   urlCount: number;
   observedAt: number;
 }
-type SitemapLegacyImportEvidence = {
+export type SitemapLegacyImportEvidence = {
   _tag: 'none';
 } | {
   _tag: 'metadata_only';
@@ -32,7 +32,7 @@ type SitemapLegacyImportEvidence = {
   recordCount: number;
   source: 'gsc_sitemaps';
 };
-interface SitemapSiteGenerationManifest {
+export interface SitemapSiteGenerationManifest {
   version: 1;
   generationId: string;
   observedAt: number;
@@ -47,13 +47,13 @@ interface SitemapSiteGenerationManifest {
   eventKeys: string[];
   feeds: Record<string, SitemapGenerationFeedManifest>;
 }
-interface SitemapStagedFeed {
+export interface SitemapStagedFeed {
   version: 1;
   generationId: string;
   observedAt: number;
   feed: SitemapGenerationFeedManifest;
 }
-type StageSitemapGenerationFeedResult = {
+export type StageSitemapGenerationFeedResult = {
   _tag: 'staged';
   generationId: string;
   feedpath: string;
@@ -72,11 +72,11 @@ type StageSitemapGenerationFeedResult = {
   _tag: 'conflict';
   reason: 'stale_generation' | 'generation_reused';
 };
-interface CompleteSitemapTraversal {
+export interface CompleteSitemapTraversal {
   _tag: 'complete';
   expectedFeedpaths: readonly string[];
 }
-type FinalizeSitemapGenerationResult = {
+export type FinalizeSitemapGenerationResult = {
   _tag: 'published';
   generationId: string;
   observedAt: number;
@@ -97,18 +97,18 @@ type FinalizeSitemapGenerationResult = {
   _tag: 'invalid_generation';
   reason: 'empty_id' | 'invalid_observed_at';
 };
-type SitemapGenerationReadResult = {
+export type SitemapGenerationReadResult = {
   _tag: 'available';
   manifest: SitemapSiteGenerationManifest;
 } | {
   _tag: 'unavailable';
   reason: 'no_generation' | 'generation_not_found';
 };
-interface IterateSitemapGenerationUrlsOptions {
+export interface IterateSitemapGenerationUrlsOptions {
   generationId?: string;
   feedpath?: string;
 }
-type IterateSitemapGenerationUrlsResult = {
+export type IterateSitemapGenerationUrlsResult = {
   _tag: 'iterator';
   generationId: string;
   observedAt: number;
@@ -124,13 +124,13 @@ type IterateSitemapGenerationUrlsResult = {
   _tag: 'unavailable';
   reason: 'no_generation' | 'generation_not_found';
 };
-interface ListSitemapGenerationUrlsOptions {
+export interface ListSitemapGenerationUrlsOptions {
   generationId?: string;
   feedpath?: string;
   cursor?: string;
   limit?: number;
 }
-type ListSitemapGenerationUrlsResult = {
+export type ListSitemapGenerationUrlsResult = {
   _tag: 'page';
   generationId: string;
   observedAt: number;
@@ -147,7 +147,7 @@ type ListSitemapGenerationUrlsResult = {
   _tag: 'unavailable';
   reason: 'no_generation' | 'generation_not_found';
 };
-type SitemapMembershipEvidence = {
+export type SitemapMembershipEvidence = {
   _tag: 'present';
   url: string;
   feedpath: string;
@@ -163,7 +163,7 @@ type SitemapMembershipEvidence = {
   url: string;
   reason: 'no_generation' | 'generation_not_found';
 };
-type QuerySitemapGenerationMembershipResult = {
+export type QuerySitemapGenerationMembershipResult = {
   _tag: 'result';
   generationId: string;
   observedAt: number;
@@ -176,7 +176,7 @@ type QuerySitemapGenerationMembershipResult = {
   _tag: 'unavailable';
   reason: 'no_generation' | 'generation_not_found';
 };
-type RecordSitemapLegacyImportResult = {
+export type RecordSitemapLegacyImportResult = {
   _tag: 'recorded';
   evidence: Extract<SitemapLegacyImportEvidence, {
     _tag: 'metadata_only';
@@ -190,12 +190,12 @@ type RecordSitemapLegacyImportResult = {
   _tag: 'invalid';
   reason: 'invalid_imported_at' | 'invalid_record_count';
 };
-interface CompleteSitemapGeneration {
+export interface CompleteSitemapGeneration {
   _tag: 'complete';
   id: string;
   observedAt: number;
 }
-interface SitemapMembershipEvent {
+export interface SitemapMembershipEvent {
   feedpath: string;
   feedpathHash: string;
   urlHash: string;
@@ -208,11 +208,11 @@ interface SitemapMembershipEvent {
   sequence: number;
   projectsState: boolean;
 }
-interface DateRange {
+export interface DateRange {
   from?: string;
   to?: string;
 }
-interface SitemapReadStore {
+export interface SitemapReadStore {
   loadEvents: (ctx: TenantCtx, dateRange?: DateRange) => AsyncIterable<SitemapMembershipEvent>;
   getSitemapGeneration: (ctx: TenantCtx, generationId?: string) => Promise<SitemapGenerationReadResult>;
   iterateSitemapGenerationUrls: (ctx: TenantCtx, opts?: IterateSitemapGenerationUrlsOptions) => Promise<IterateSitemapGenerationUrlsResult>;
@@ -222,7 +222,7 @@ interface SitemapReadStore {
     urls: readonly string[];
   }) => Promise<QuerySitemapGenerationMembershipResult>;
 }
-interface SitemapStore extends SitemapReadStore {
+export interface SitemapStore extends SitemapReadStore {
   stageSitemapGenerationFeed: (ctx: TenantCtx, generation: CompleteSitemapGeneration, feedpath: string, urls: readonly ParsedUrl[]) => Promise<StageSitemapGenerationFeedResult>;
   finalizeSitemapGeneration: (ctx: TenantCtx, generation: CompleteSitemapGeneration, traversal: CompleteSitemapTraversal) => Promise<FinalizeSitemapGenerationResult>;
   abortSitemapGeneration: (ctx: TenantCtx, generation: CompleteSitemapGeneration) => Promise<void>;
@@ -231,12 +231,11 @@ interface SitemapStore extends SitemapReadStore {
     recordCount: number;
   }) => Promise<RecordSitemapLegacyImportResult>;
 }
-interface CreateSitemapReadStoreOptions {
+export interface CreateSitemapReadStoreOptions {
   dataSource: DataSource;
 }
-type SitemapMutation = <T>(ctx: TenantCtx, effect: () => Promise<T>) => Promise<T>;
-interface CreateSitemapStoreOptions extends CreateSitemapReadStoreOptions {
+export type SitemapMutation = <T>(ctx: TenantCtx, effect: () => Promise<T>) => Promise<T>;
+export interface CreateSitemapStoreOptions extends CreateSitemapReadStoreOptions {
   withMutation: SitemapMutation;
   now?: () => number;
-}
-export { CompleteSitemapGeneration, CompleteSitemapTraversal, CreateSitemapReadStoreOptions, CreateSitemapStoreOptions, DateRange, FinalizeSitemapGenerationResult, IterateSitemapGenerationUrlsOptions, IterateSitemapGenerationUrlsResult, ListSitemapGenerationUrlsOptions, ListSitemapGenerationUrlsResult, ParsedUrl, QuerySitemapGenerationMembershipResult, RecordSitemapLegacyImportResult, SitemapGenerationFeedManifest, SitemapGenerationReadResult, SitemapLegacyImportEvidence, SitemapMembershipEvent, SitemapMembershipEvidence, SitemapMutation, SitemapReadStore, SitemapSiteGenerationManifest, SitemapStagedFeed, SitemapStore, SitemapUrlRecord, StageSitemapGenerationFeedResult };
+}

package/dist/entities/sitemap-write.d.mts

@@ -1,3 +1,2 @@
 import { CreateSitemapStoreOptions, SitemapStore } from "./sitemap-shared.mjs";
-declare function createSitemapStore(opts: CreateSitemapStoreOptions): SitemapStore;
-export { createSitemapStore };
+export declare function createSitemapStore(opts: CreateSitemapStoreOptions): SitemapStore;

package/dist/entities/sitemap.d.mts

@@ -1,3 +1,2 @@
 import { CreateSitemapReadStoreOptions, SitemapReadStore } from "./sitemap-shared.mjs";
-declare function createSitemapReadStore(opts: CreateSitemapReadStoreOptions): SitemapReadStore;
-export { createSitemapReadStore };
+export declare function createSitemapReadStore(opts: CreateSitemapReadStoreOptions): SitemapReadStore;

package/dist/entity-keys.d.mts

@@ -1,10 +1,10 @@
 import { TenantCtx } from "@gscdump/contracts";
-declare function inspectionIndexKey(ctx: TenantCtx): string;
-declare function emptyTypesKey(ctx: TenantCtx): string;
-declare function inspectionParquetKey(ctx: TenantCtx): string;
-declare function inspectionEventsPrefix(ctx: TenantCtx): string;
-declare function inspectionEventKey(ctx: TenantCtx, yearMonth: string, batchId: string): string;
-declare function inspectionBaseKey(ctx: TenantCtx): string;
+export declare function inspectionIndexKey(ctx: TenantCtx): string;
+export declare function emptyTypesKey(ctx: TenantCtx): string;
+export declare function inspectionParquetKey(ctx: TenantCtx): string;
+export declare function inspectionEventsPrefix(ctx: TenantCtx): string;
+export declare function inspectionEventKey(ctx: TenantCtx, yearMonth: string, batchId: string): string;
+export declare function inspectionBaseKey(ctx: TenantCtx): string;
 /**
  * Append-only record of URL-inspection STATE CHANGES, partitioned by month.
  *
@@ -13,42 +13,41 @@ declare function inspectionBaseKey(ctx: TenantCtx): string;
  * the moment the event file folds — this is the only durable record that a URL
  * ever left (or entered) the index.
  */
-declare function inspectionTransitionsPrefix(ctx: TenantCtx): string;
+export declare function inspectionTransitionsPrefix(ctx: TenantCtx): string;
 /** One file per calendar month (`YYYY-MM`), rewritten in place as it fills. */
-declare function inspectionTransitionsMonthKey(ctx: TenantCtx, yearMonth: string): string;
-declare function inspectionHistoryPrefix(ctx: TenantCtx, yearMonth: string): string;
-declare function inspectionHistoryShardKey(ctx: TenantCtx, yearMonth: string, batchId: string): string;
+export declare function inspectionTransitionsMonthKey(ctx: TenantCtx, yearMonth: string): string;
+export declare function inspectionHistoryPrefix(ctx: TenantCtx, yearMonth: string): string;
+export declare function inspectionHistoryShardKey(ctx: TenantCtx, yearMonth: string, batchId: string): string;
 /** Stable, URL-safe FNV-1a hash used for entity keys. */
-declare function hashUrl(url: string): string;
-declare function sitemapUrlsPrefix(ctx: TenantCtx): string;
-interface SitemapGenerationKey {
+export declare function hashUrl(url: string): string;
+export declare function sitemapUrlsPrefix(ctx: TenantCtx): string;
+export interface SitemapGenerationKey {
   id: string;
   observedAt: number;
 }
-declare function sitemapUrlsEventsPrefix(ctx: TenantCtx): string;
-declare function sitemapUrlsEventKey(ctx: TenantCtx, feedpathHash: string, generation: SitemapGenerationKey): string;
-declare function sitemapSiteManifestKey(ctx: TenantCtx): string;
-declare function sitemapSiteManifestsPrefix(ctx: TenantCtx): string;
-declare function sitemapSiteGenerationManifestKey(ctx: TenantCtx, generation: SitemapGenerationKey): string;
-declare function sitemapStagedGenerationPrefix(ctx: TenantCtx, generation: SitemapGenerationKey): string;
-declare function sitemapStagedFeedKey(ctx: TenantCtx, generation: SitemapGenerationKey, feedpathHash: string): string;
-declare function sitemapImmutableBaseKey(ctx: TenantCtx, generation: SitemapGenerationKey, feedpathHash: string): string;
-declare function sitemapLegacyImportKey(ctx: TenantCtx): string;
+export declare function sitemapUrlsEventsPrefix(ctx: TenantCtx): string;
+export declare function sitemapUrlsEventKey(ctx: TenantCtx, feedpathHash: string, generation: SitemapGenerationKey): string;
+export declare function sitemapSiteManifestKey(ctx: TenantCtx): string;
+export declare function sitemapSiteManifestsPrefix(ctx: TenantCtx): string;
+export declare function sitemapSiteGenerationManifestKey(ctx: TenantCtx, generation: SitemapGenerationKey): string;
+export declare function sitemapStagedGenerationPrefix(ctx: TenantCtx, generation: SitemapGenerationKey): string;
+export declare function sitemapStagedFeedKey(ctx: TenantCtx, generation: SitemapGenerationKey, feedpathHash: string): string;
+export declare function sitemapImmutableBaseKey(ctx: TenantCtx, generation: SitemapGenerationKey, feedpathHash: string): string;
+export declare function sitemapLegacyImportKey(ctx: TenantCtx): string;
 /** WHATWG-serialized HTTP(S) feed identity with the fragment removed. */
-type SitemapFeedIdentityResult = {
+export type SitemapFeedIdentityResult = {
   _tag: 'ok';
   url: string;
 } | {
   _tag: 'invalid';
   reason: 'credentials' | 'empty' | 'invalid_url' | 'unsupported_protocol';
 };
-declare function parseSitemapFeedIdentity(input: string): SitemapFeedIdentityResult;
+export declare function parseSitemapFeedIdentity(input: string): SitemapFeedIdentityResult;
 /** Versioned exact payload hash. Lastmod-only changes are intentionally visible. */
-declare function sitemapPayloadHashV1(urls: readonly {
+export declare function sitemapPayloadHashV1(urls: readonly {
   loc: string;
   lastmod?: string;
 }[]): string;
-declare function indexingMetadataIndexKey(ctx: TenantCtx): string;
-declare function queryDimParquetKey(ctx: TenantCtx): string;
-declare function queryDimMetaKey(ctx: TenantCtx): string;
-export { SitemapFeedIdentityResult, SitemapGenerationKey, emptyTypesKey, hashUrl, indexingMetadataIndexKey, inspectionBaseKey, inspectionEventKey, inspectionEventsPrefix, inspectionHistoryPrefix, inspectionHistoryShardKey, inspectionIndexKey, inspectionParquetKey, inspectionTransitionsMonthKey, inspectionTransitionsPrefix, parseSitemapFeedIdentity, queryDimMetaKey, queryDimParquetKey, sitemapImmutableBaseKey, sitemapLegacyImportKey, sitemapPayloadHashV1, sitemapSiteGenerationManifestKey, sitemapSiteManifestKey, sitemapSiteManifestsPrefix, sitemapStagedFeedKey, sitemapStagedGenerationPrefix, sitemapUrlsEventKey, sitemapUrlsEventsPrefix, sitemapUrlsPrefix };
+export declare function indexingMetadataIndexKey(ctx: TenantCtx): string;
+export declare function queryDimParquetKey(ctx: TenantCtx): string;
+export declare function queryDimMetaKey(ctx: TenantCtx): string;

package/dist/errors.d.mts

@@ -1,5 +1,5 @@
-type EngineErrorKind = 'analyzer-not-found' | 'analyzer-capability-missing' | 'invalid-sql-literal' | 'placeholder-arity-mismatch' | 'invalid-search-types' | 'attached-table-missing' | 'manifest-cas-exhausted' | 'invalid-snapshot-filename' | 'unsupported-snapshot-index-version' | 'invalid-schema-identifier' | 'invalid-year-month' | 'missing-attach-url' | 'manifest-cas-round-lost' | 'iceberg-table-op-failed' | 'sink-table-flush-failed' | 'rollup-build-failed' | 'lock-acquire-timeout';
-type EngineError = {
+export type EngineErrorKind = 'analyzer-not-found' | 'analyzer-capability-missing' | 'invalid-sql-literal' | 'placeholder-arity-mismatch' | 'invalid-search-types' | 'attached-table-missing' | 'manifest-cas-exhausted' | 'invalid-snapshot-filename' | 'unsupported-snapshot-index-version' | 'invalid-schema-identifier' | 'invalid-year-month' | 'missing-attach-url' | 'manifest-cas-round-lost' | 'iceberg-table-op-failed' | 'sink-table-flush-failed' | 'rollup-build-failed' | 'lock-acquire-timeout';
+export type EngineError = {
   kind: 'analyzer-not-found';
   tool: string;
   message: string;
@@ -76,7 +76,7 @@ type EngineError = {
   scope: string;
   timeoutMs: number;
 };
-declare const engineErrors: {
+export declare const engineErrors: {
   readonly analyzerNotFound: (tool: string) => EngineError;
   readonly analyzerCapabilityMissing: (tool: string, missing: readonly string[]) => EngineError;
   readonly nonFiniteNumberLiteral: (value: number) => EngineError;
@@ -102,7 +102,7 @@ declare const engineErrors: {
   readonly rollupBuildFailed: (id: string, cause: unknown) => EngineError;
   readonly lockAcquireTimeout: (scope: string, timeoutMs: number) => EngineError;
 };
-declare function isEngineError(value: unknown): value is EngineError;
+export declare function isEngineError(value: unknown): value is EngineError;
 /**
  * Re-raises an `EngineError` value as a generic `Error`, stashing the union under
  * `.engineError` for stack-walking. Used by the throwing wrappers over the
@@ -111,5 +111,4 @@ declare function isEngineError(value: unknown): value is EngineError;
  * (`AnalyzerCapabilityError`, `AttachedTableMissingError`) provide their own
  * mapper so the class identity callers match on is preserved.
  */
-declare function engineErrorToException(error: EngineError): Error;
-export { EngineError, EngineErrorKind, engineErrorToException, engineErrors, isEngineError };
+export declare function engineErrorToException(error: EngineError): Error;

package/dist/iceberg/append-sink.d.mts

@@ -1,5 +1,5 @@
 import { IcebergAppendSinkOptions, Sink } from "../sink.mjs";
-type IcebergAppendSink = Sink;
+export type IcebergAppendSink = Sink;
 /**
  * Create an `IcebergAppendSink` over the R2 Data Catalog.
  *
@@ -10,5 +10,4 @@ type IcebergAppendSink = Sink;
  * Worker callers can share the `/v1/config` context through a durable cache;
  * without it every fresh sink/isolate must probe the catalog again.
  */
-declare function createIcebergAppendSink(options: IcebergAppendSinkOptions): IcebergAppendSink;
-export { IcebergAppendSink, createIcebergAppendSink };
+export declare function createIcebergAppendSink(options: IcebergAppendSinkOptions): IcebergAppendSink;

package/dist/iceberg/catalog.d.mts

@@ -1,7 +1,7 @@
 import { EngineError } from "../errors.mjs";
 import { IcebergTableName, PartitionKeyEncoding } from "./schema.mjs";
-import { icebergCreateTable } from "../node_modules/.pnpm/icebird@0.8.27_patch_hash_651b5cc1b513184638c98e21a0852c4559044ae60f8449fbe9d81a894ea5b4d7/node_modules/icebird/types/write/write.mjs";
-import "../node_modules/.pnpm/icebird@0.8.27_patch_hash_651b5cc1b513184638c98e21a0852c4559044ae60f8449fbe9d81a894ea5b4d7/node_modules/icebird/types/index.mjs";
+import { icebergCreateTable } from "../node_modules/.pnpm/icebird@0.8.28_patch_hash_651b5cc1b513184638c98e21a0852c4559044ae60f8449fbe9d81a894ea5b4d7/node_modules/icebird/types/write/write.mjs";
+import "../node_modules/.pnpm/icebird@0.8.28_patch_hash_651b5cc1b513184638c98e21a0852c4559044ae60f8449fbe9d81a894ea5b4d7/node_modules/icebird/types/index.mjs";
 import { Result } from "gscdump/result";
 import { CommitRetryOptions, ConnectIcebergOptions, IcebergCatalogConfig, IcebergConnection, IcebergListedDataFile, QueryProfiler } from "@gscdump/lakehouse";
 type IcebirdCreateTableOptions = Parameters<typeof icebergCreateTable>[0];
@@ -13,24 +13,24 @@ type IcebirdSortOrder = NonNullable<IcebirdCreateTableOptions['sortOrder']>;
  * `ICEBERG_SCHEMAS` contract. Field ids are advisory — R2 Data Catalog
  * re-assigns them on `createTable`.
  */
-declare function icebergSchemaFor(table: IcebergTableName, encoding?: PartitionKeyEncoding): IcebirdSchema;
+export declare function icebergSchemaFor(table: IcebergTableName, encoding?: PartitionKeyEncoding): IcebirdSchema;
 /**
  * Build the icebird `PartitionSpec` for one of the 9 fact tables: the locked
  * spec `identity(site_id) + identity(search_type) + month(date)`.
  */
-declare function icebergPartitionSpecFor(table: IcebergTableName, encoding?: PartitionKeyEncoding): IcebirdPartitionSpec;
+export declare function icebergPartitionSpecFor(table: IcebergTableName, encoding?: PartitionKeyEncoding): IcebirdPartitionSpec;
 /**
  * Build the icebird `SortOrder` for a fact table from its `clusterKey`
  * (dimension-first, then `date`).
  */
-declare function icebergSortOrderFor(table: IcebergTableName, encoding?: PartitionKeyEncoding): IcebirdSortOrder;
+export declare function icebergSortOrderFor(table: IcebergTableName, encoding?: PartitionKeyEncoding): IcebirdSortOrder;
 /**
  * Outcome of a single table create/drop: the table name plus a `Result` —
  * `Ok(void)` on success, `Err(iceberg-table-op-failed)` carrying the failure
  * message when the catalog rejects the op. Per-table so a partial
  * provisioning run is fully observable.
  */
-interface IcebergTableOpResult {
+export interface IcebergTableOpResult {
   table: string;
   outcome: Result<void, EngineError>;
 }
@@ -39,7 +39,7 @@ interface IcebergTableOpResult {
  * the schema derived from {@link ICEBERG_SCHEMAS}. Per-table errors are
  * captured rather than thrown so a partial run is observable.
  */
-declare function createIcebergTables(conn: IcebergConnection, tables?: readonly IcebergTableName[], encoding?: PartitionKeyEncoding): Promise<IcebergTableOpResult[]>;
+export declare function createIcebergTables(conn: IcebergConnection, tables?: readonly IcebergTableName[], encoding?: PartitionKeyEncoding): Promise<IcebergTableOpResult[]>;
 /**
  * Drop tables from the catalog namespace, purging their data objects.
  * Defaults to every table currently in the namespace. Delegates the walk to
@@ -48,7 +48,7 @@ declare function createIcebergTables(conn: IcebergConnection, tables?: readonly
  * {@link IcebergTableOpResult} contract.
  */
 declare function dropIcebergTables$1(conn: IcebergConnection, tables?: readonly string[]): Promise<IcebergTableOpResult[]>;
-interface ListIcebergDataFilesOptions {
+export interface ListIcebergDataFilesOptions {
   table: IcebergTableName;
   /** Partition identity column. `number` for `'int'`-encoded catalogs. */
   siteId: string | number;
@@ -67,5 +67,5 @@ interface ListIcebergDataFilesOptions {
  * List the parquet data files in the current snapshot of `table`, filtered to
  * a single partition slice `(siteId, searchType, month(date) ∈ range)`.
  */
-declare function listIcebergDataFiles(conn: IcebergConnection, opts: ListIcebergDataFilesOptions): Promise<IcebergListedDataFile[]>;
-export { type CommitRetryOptions, type ConnectIcebergOptions, type IcebergCatalogConfig, IcebergTableOpResult, ListIcebergDataFilesOptions, createIcebergTables, dropIcebergTables$1 as dropIcebergTables, icebergPartitionSpecFor, icebergSchemaFor, icebergSortOrderFor, listIcebergDataFiles };
+export declare function listIcebergDataFiles(conn: IcebergConnection, opts: ListIcebergDataFilesOptions): Promise<IcebergListedDataFile[]>;
+export { type CommitRetryOptions, type ConnectIcebergOptions, type IcebergCatalogConfig, dropIcebergTables$1 as dropIcebergTables };

package/dist/iceberg/catalog.mjs

@@ -2,8 +2,8 @@ import { engineErrors } from "../errors.mjs";
 import { TABLE_METADATA } from "../drizzle-schema.mjs";
 import "../schema.mjs";
 import { DEFAULT_PARTITION_KEY_ENCODING, ICEBERG_PARTITION_SPEC, ICEBERG_TABLES, icebergSchemasFor } from "./schema.mjs";
-import { icebergCreateTable } from "../node_modules/.pnpm/icebird@0.8.27_patch_hash_651b5cc1b513184638c98e21a0852c4559044ae60f8449fbe9d81a894ea5b4d7/node_modules/icebird/src/write/write.mjs";
-import "../node_modules/.pnpm/icebird@0.8.27_patch_hash_651b5cc1b513184638c98e21a0852c4559044ae60f8449fbe9d81a894ea5b4d7/node_modules/icebird/src/index.mjs";
+import { icebergCreateTable } from "../node_modules/.pnpm/icebird@0.8.28_patch_hash_651b5cc1b513184638c98e21a0852c4559044ae60f8449fbe9d81a894ea5b4d7/node_modules/icebird/src/write/write.mjs";
+import "../node_modules/.pnpm/icebird@0.8.28_patch_hash_651b5cc1b513184638c98e21a0852c4559044ae60f8449fbe9d81a894ea5b4d7/node_modules/icebird/src/index.mjs";
 import { err, ok } from "gscdump/result";
 import { ICEBERG_TYPE_MAP, connectIcebergCatalog, dropIcebergTables, resolveIcebergDataFiles } from "@gscdump/lakehouse";
 import { icebergAppendRetrying } from "@gscdump/lakehouse/unsafe-raw";

package/dist/iceberg/overwrite-writer.d.mts

@@ -2,7 +2,7 @@ import { Row } from "../storage.mjs";
 import { IcebergS3Config } from "./schema.mjs";
 import { Sink, SinkCloseResult, SliceOverwriteWriter } from "../sink.mjs";
 /** Connection details for the Iceberg REST catalog the writer targets. */
-interface OverwriteWriterCatalogConfig {
+export interface OverwriteWriterCatalogConfig {
   /**
    * Iceberg REST catalog endpoint, e.g. `http://localhost:8181` (POC) or the
    *  R2 Data Catalog endpoint (prod).
@@ -21,7 +21,7 @@ interface OverwriteWriterCatalogConfig {
  * The JSON job wire-format shared with `scripts/iceberg-writer.py`.
  * `op` is always `overwrite` here.
  */
-interface OverwriteJob {
+export interface OverwriteJob {
   op: 'overwrite';
   catalogUri: string;
   namespace: string;
@@ -42,7 +42,7 @@ interface OverwriteJob {
  * re-backfill, without touching sibling sites. Carries no rows. `siteId` is a
  * `number` for an INT-encoded shard, a `string` for STRING encoding.
  */
-interface DeleteJob {
+export interface DeleteJob {
   op: 'delete';
   catalogUri: string;
   namespace: string;
@@ -54,9 +54,9 @@ interface DeleteJob {
   siteId: string | number;
 }
 /** Any job the shared PyIceberg backend can execute. */
-type PyIcebergJob = OverwriteJob | DeleteJob;
+export type PyIcebergJob = OverwriteJob | DeleteJob;
 /** Result wire-format from the PyIceberg backend. */
-interface OverwriteJobResult {
+export interface OverwriteJobResult {
   /** Rows written (emit/overwrite) or removed (delete). */
   rowCount?: number;
   error?: string;
@@ -68,8 +68,8 @@ interface OverwriteJobResult {
  * backend value is still assignable where the narrower `OverwriteBackend` is
  * expected (a wider-param function accepts the narrower call).
  */
-type OverwriteBackend = (job: PyIcebergJob) => Promise<OverwriteJobResult>;
-interface IcebergOverwriteWriterOptions {
+export type OverwriteBackend = (job: PyIcebergJob) => Promise<OverwriteJobResult>;
+export interface IcebergOverwriteWriterOptions {
   catalog: OverwriteWriterCatalogConfig;
   /** The transport that runs the PyIceberg job. */
   backend: OverwriteBackend;
@@ -79,7 +79,7 @@ interface IcebergOverwriteWriterOptions {
  * `SliceOverwriteWriter` contract; `overwriteSlice` builds one PyIceberg
  * overwrite job per call and runs it through the configured backend.
  */
-interface IcebergOverwriteWriter extends SliceOverwriteWriter {
+export interface IcebergOverwriteWriter extends SliceOverwriteWriter {
   /** Release any backend resources. Idempotent. */
   close: () => Promise<SinkCloseResult>;
 }
@@ -87,9 +87,9 @@ interface IcebergOverwriteWriter extends SliceOverwriteWriter {
  * Build a trailing-window Iceberg overwrite writer over the given backend.
  * Pure — all I/O is the backend's; the writer only shapes the job.
  */
-declare function createIcebergOverwriteWriter(opts: IcebergOverwriteWriterOptions): IcebergOverwriteWriter;
+export declare function createIcebergOverwriteWriter(opts: IcebergOverwriteWriterOptions): IcebergOverwriteWriter;
 /** Options for the PyIceberg subprocess backend. */
-interface SubprocessBackendOptions {
+export interface SubprocessBackendOptions {
   /** Python interpreter. Defaults to `$GSCDUMP_ICEBERG_PYTHON` then `python3`. */
   python?: string;
   /**
@@ -104,9 +104,9 @@ interface SubprocessBackendOptions {
  * tests and a Node-side scheduled job box. The script reads the job JSON on
  * stdin and writes `{rowCount}` / `{error}` on stdout.
  */
-declare function subprocessBackend(opts?: SubprocessBackendOptions): OverwriteBackend;
+export declare function subprocessBackend(opts?: SubprocessBackendOptions): OverwriteBackend;
 /** Options for the PyIceberg HTTP backend (the Cloudflare Container in prod). */
-interface HttpBackendOptions {
+export interface HttpBackendOptions {
   /**
    * URL of the PyIceberg overwrite service — a POST endpoint accepting the
    *  `OverwriteJob` JSON and returning `OverwriteJobResult`.
@@ -123,16 +123,16 @@ interface HttpBackendOptions {
  * is the prod transport: the request-path Worker holds no Python and never
  * does an Iceberg metadata commit itself.
  */
-declare function httpBackend(opts: HttpBackendOptions): OverwriteBackend;
+export declare function httpBackend(opts: HttpBackendOptions): OverwriteBackend;
 /**
  * Adapt an `IcebergOverwriteWriter` into a full `Sink` whose `emit` is also a
  * partition-overwrite — convenient for tests so a single Iceberg writer backs
  * both append and revision paths against the local stack. Prod keeps `emit`
  * on the Pipeline; only `overwriteSlice` delegates here.
  */
-declare function overwriteWriterAsSink(writer: IcebergOverwriteWriter): Sink & SliceOverwriteWriter;
+export declare function overwriteWriterAsSink(writer: IcebergOverwriteWriter): Sink & SliceOverwriteWriter;
 /** Per-table outcome of a {@link deleteSiteFromShard} run. */
-interface DeleteSiteResult {
+export interface DeleteSiteResult {
   table: string;
   /** Rows removed, when the backend reported a count. */
   rowCount?: number;
@@ -152,10 +152,9 @@ interface DeleteSiteResult {
  * failure is visible and re-runnable (delete is idempotent: re-deleting an
  * already-empty site is a no-op).
  */
-declare function deleteSiteFromShard(args: {
+export declare function deleteSiteFromShard(args: {
   catalog: OverwriteWriterCatalogConfig;
   backend: OverwriteBackend;
   siteId: string | number;
   tables: readonly string[];
-}): Promise<DeleteSiteResult[]>;
-export { DeleteJob, DeleteSiteResult, HttpBackendOptions, IcebergOverwriteWriter, IcebergOverwriteWriterOptions, OverwriteBackend, OverwriteJob, OverwriteJobResult, OverwriteWriterCatalogConfig, PyIcebergJob, SubprocessBackendOptions, createIcebergOverwriteWriter, deleteSiteFromShard, httpBackend, overwriteWriterAsSink, subprocessBackend };
+}): Promise<DeleteSiteResult[]>;