@delali/narsil 0.2.2 → 0.2.3-next.33

package/dist/index.d.ts

@@ -1,13 +1,13 @@
-import { S as StopWordOverride, C as CustomTokenizer } from './schema-CE6S_aAr.js';
-export { A as AnyDocument, B as BM25Params, E as EmbeddingFieldConfig, F as FieldType, I as IndexConfig, a as InsertOptions, P as PartitionConfig, b as SchemaDefinition, c as ScoringMode, V as VectorIndexConfig, d as VectorQuantizationMode } from './schema-CE6S_aAr.js';
+import { c as StopWordOverride, C as CustomTokenizer } from './schema-emTQfEiX.js';
+export { A as AnyDocument, B as BM25Params, E as EmbeddingFieldConfig, F as FieldType, I as IndexConfig, a as InsertOptions, P as PartitionConfig, S as SchemaDefinition, b as ScoringMode, V as VectorIndexConfig, d as VectorQuantizationMode } from './schema-emTQfEiX.js';
 import { L as LanguageModule } from './language-DB8KG1z5.js';
 export { T as TokenizerConfig } from './language-DB8KG1z5.js';
-export { A as ArrayFilter, B as BatchResult, C as ComparisonFilter, E as ErrorCode, a as ErrorCodes, F as FacetConfig, b as FacetResult, c as FieldFilter, d as FilterExpression, G as GeoFilter, e as GeoPolygonFilter, f as GeoRadiusFilter, g as GroupConfig, h as GroupReducer, i as GroupResult, H as HighlightConfig, j as HighlightMatch, k as Hit, l as HybridConfig, I as IndexInfo, m as IndexStats, M as MemoryStats, N as NarsilError, P as PartitionStatsResult, n as PreflightResult, o as PresenceFilter, p as ProcessMemoryReport, Q as QueryParams, q as QueryResult, S as ScoreComponents, r as SearchMode, s as StringFilter, t as SuggestParams, u as SuggestResult, T as TermMatchPolicy, V as VectorMaintenanceResult, v as VectorQueryConfig } from './search-DLBtiPdQ.js';
-import { N as NarsilConfig } from './config-DXypXAfN.js';
-export { A as AnalysisConfig, D as DurabilityConfig, I as IndexContext, a as InsertContext, b as NarsilPlugin, P as PartitionContext, R as RemoveContext, S as SearchContext, c as StaleAnalysis, U as UpdateContext, W as WorkerConfig, d as WorkerContext } from './config-DXypXAfN.js';
-import { N as Narsil } from './engine-CO9Wnfie.js';
-export { a as NarsilEventMap } from './engine-CO9Wnfie.js';
-export { E as EmbeddingAdapter, I as InvalidationAdapter, a as InvalidationEvent, P as PartitionStatistics, b as PersistenceAdapter } from './adapters-CVqI5NBt.js';
+export { A as ArrayFilter, B as BatchResult, C as ClientErrorCode, a as ClientErrorCodes, k as ComparisonFilter, D as DocumentProjection, E as ErrorCode, b as ErrorCodes, F as FacetConfig, l as FacetResult, m as FieldFilter, n as FilterExpression, G as GeoFilter, o as GeoPolygonFilter, p as GeoRadiusFilter, q as GroupConfig, r as GroupReducer, s as GroupResult, H as HighlightConfig, t as HighlightMatch, u as Hit, v as HybridConfig, I as IndexInfo, f as IndexStats, L as ListParams, g as ListResult, w as ListedDocument, M as MemoryStats, N as NarsilError, c as NarsilErrorCode, P as PartitionStatsResult, h as PreflightResult, x as PresenceFilter, y as ProcessMemoryReport, Q as QueryParams, e as QueryResult, z as ScoreComponents, J as SearchMode, S as ServerErrorCode, d as ServerErrorCodes, K as SortField, O as SortSpec, R as StringFilter, i as SuggestParams, j as SuggestResult, T as TermMatchPolicy, V as VectorMaintenanceResult, U as VectorQueryConfig } from './search-CQ-9bKxo.js';
+import { N as NarsilConfig } from './config-ChZ7lfqc.js';
+export { A as AnalysisConfig, D as DurabilityConfig, I as IndexContext, a as InsertContext, b as NarsilPlugin, P as PartitionContext, R as RemoveContext, S as SearchContext, c as StaleAnalysis, U as UpdateContext, W as WorkerConfig, d as WorkerContext } from './config-ChZ7lfqc.js';
+import { N as Narsil } from './engine-DScGlT2Z.js';
+export { a as NarsilEventMap } from './engine-DScGlT2Z.js';
+export { E as EmbeddingAdapter, I as InvalidationAdapter, a as InvalidationEvent, b as PartitionStatistics, P as PersistenceAdapter } from './adapters-79OIa4xm.js';
 
 /**
  * Registers a tokeniser under a name that an index config can then reference.

package/dist/index.mjs

@@ -1,8 +1,11 @@
-export { VERSION, createNarsil } from './chunk-WLR5BFL3.mjs';
-export { clearNormalizationCache, configureNormalizationCache, getLanguage, getNormalizationCacheSize, getStopWords, getTokenizer, hasStopWords, hasTokenizer, registerLanguage, registerStopWords, registerTokenizer, resetNormalizationCache } from './chunk-ASI6AOW2.mjs';
+export { VERSION, createNarsil } from './chunk-TEAJO7CU.mjs';
+import './chunk-C6VVLMRX.mjs';
+export { clearNormalizationCache, configureNormalizationCache, getLanguage, getNormalizationCacheSize, getStopWords, getTokenizer, hasStopWords, hasTokenizer, registerLanguage, registerStopWords, registerTokenizer, resetNormalizationCache } from './chunk-RO4EFWE6.mjs';
 import './chunk-TGP5M5R7.mjs';
-export { isSimdAvailable } from './chunk-S7RJJMHI.mjs';
+import './chunk-OBXS7OZM.mjs';
+export { isSimdAvailable } from './chunk-WDL632CN.mjs';
 import './chunk-KMMJMDUA.mjs';
 import './chunk-U2BKYEC7.mjs';
-export { ErrorCodes, NarsilError } from './chunk-UNJZ7ZDB.mjs';
-import './chunk-BKAI2WXI.mjs';
+import './chunk-BFIH7OV5.mjs';
+import './chunk-WTONLDU3.mjs';
+export { ClientErrorCodes, ErrorCodes, NarsilError, ServerErrorCodes } from './chunk-EMI63EXP.mjs';

package/dist/invalidation/broadcast-channel.d.ts

@@ -1,4 +1,4 @@
-import { I as InvalidationAdapter } from '../adapters-CVqI5NBt.js';
+import { I as InvalidationAdapter } from '../adapters-79OIa4xm.js';
 
 /**
  * Which channel {@link createBroadcastChannelInvalidation} publishes on.

package/dist/invalidation/filesystem.d.ts

@@ -1,4 +1,4 @@
-import { I as InvalidationAdapter } from '../adapters-CVqI5NBt.js';
+import { I as InvalidationAdapter } from '../adapters-79OIa4xm.js';
 
 /**
  * How {@link createFilesystemInvalidation} exchanges events between processes.

package/dist/invalidation/filesystem.mjs

@@ -1,3 +1,5 @@
+import { compareCodePoints } from '../chunk-BFIH7OV5.mjs';
+
 // src/invalidation/filesystem.ts
 var MARKER_MAX_AGE_MS = 6e4;
 var DEFAULT_POLL_INTERVAL = 1e3;
@@ -56,7 +58,7 @@ function createFilesystemInvalidation(config) {
       }
       throw err;
     }
-    const jsonFiles = files.filter((f) => f.endsWith(".json")).sort();
+    const jsonFiles = files.filter((f) => f.endsWith(".json")).sort(compareCodePoints);
     const now = Date.now();
     for (const file of jsonFiles) {
       const filePath = path.join(directory, file);

package/dist/invalidation/noop.d.ts

@@ -1,4 +1,4 @@
-import { I as InvalidationAdapter } from '../adapters-CVqI5NBt.js';
+import { I as InvalidationAdapter } from '../adapters-79OIa4xm.js';
 
 /**
  * Builds an invalidation adapter that discards every event, which suits a

package/dist/options-BORJg5XB.d.ts

@@ -0,0 +1,234 @@
+import { E as EmbeddingAdapter } from './adapters-79OIa4xm.js';
+import { h as TaskStore } from './tasks-CnMLpfk-.js';
+
+/**
+ * Marks that `POST /indexes/{name}/documents/_import` accepts `?async=true` and
+ * answers with a task record.
+ *
+ * @public
+ */
+declare const ASYNC_IMPORT_CAPABILITY = "documents.import.async";
+/**
+ * Marks that `POST /tasks/{id}/_cancel` asks a running task to stop.
+ *
+ * @public
+ */
+declare const TASK_CANCEL_CAPABILITY = "tasks.cancel";
+/**
+ * Marks that `GET /tasks` filters by index, type, and status, and pages through
+ * `from` and `limit`.
+ *
+ * @public
+ */
+declare const TASK_FILTER_CAPABILITY = "tasks.filter";
+/**
+ * Marks that `POST /indexes/{name}/_rebuild-analysis` reanalyses an index whose
+ * language module moved on.
+ *
+ * @public
+ */
+declare const REBUILD_ANALYSIS_CAPABILITY = "indexes.rebuildAnalysis";
+/**
+ * Every capability this server announces at `/capabilities`.
+ *
+ * A client reads this list to find out whether an optional route or mode is
+ * available before it sends a request, because a server that predates one of
+ * them answers 404 rather than explaining itself.
+ *
+ * @public
+ */
+declare const SERVER_CAPABILITIES: readonly string[];
+
+/**
+ * Context handed to {@link OnRequestHook} for every inbound request, captured
+ * synchronously before any body is read. `remoteAddress` is the peer address
+ * decoded from the socket; behind a proxy it is the proxy's address, so trust
+ * forwarded headers only when the proxy is trusted.
+ *
+ * @public
+ */
+interface RequestContext {
+    /** The request used this HTTP method, in upper case. */
+    method: string;
+    /** The request asked for this path, without the query string. */
+    path: string;
+    /** These are the request headers, keyed by lower-case name. */
+    headers: Record<string, string>;
+    /** The server decoded this peer address from the socket. */
+    remoteAddress: string;
+}
+/**
+ * Returned by an {@link OnRequestHook} to reject a request. The `status` is sent
+ * verbatim as the HTTP status; `code` and `message` populate the error envelope.
+ *
+ * @public
+ */
+interface RequestDenial {
+    /** The server answers with this status. */
+    status: number;
+    /** The error envelope carries this machine-readable code. */
+    code: string;
+    /** The error envelope carries this description. */
+    message: string;
+}
+/**
+ * Runs before every routed request, which is where authentication, rate
+ * limiting, and any other admission rule goes.
+ *
+ * @param ctx - The request's method, path, headers, and peer address.
+ * @returns A {@link RequestDenial} to reject the request, or `undefined` to
+ * let it through.
+ *
+ * @public
+ */
+type OnRequestHook = (ctx: RequestContext) => undefined | RequestDenial | Promise<undefined | RequestDenial>;
+/**
+ * Which cross-origin requests the server accepts.
+ *
+ * Passing `true` for `cors` instead of this object accepts any origin, which
+ * suits local development and little else.
+ *
+ * @public
+ */
+interface CorsOptions {
+    /** The server allows these origins, given as one origin or a list. */
+    origin?: string | string[];
+    /** The server allows these methods on a cross-origin request. */
+    methods?: string[];
+    /** A browser may send these request headers cross-origin. */
+    headers?: string[];
+}
+/**
+ * Byte and batch ceilings the server enforces before handing work to the
+ * engine. Every field has a safe default; override only to fit a deployment.
+ *
+ * @public
+ */
+interface ServerLimits {
+    /** Cap for JSON request bodies (single doc, batch, search). Excess → 413. */
+    maxBodyBytes?: number;
+    /** Cap for the NDJSON import stream and binary restore body. Excess → 413. */
+    maxImportBytes?: number;
+    /** Cap for a single NDJSON line, so one unterminated line cannot exhaust memory. Excess → 413. */
+    maxLineBytes?: number;
+    /** Documents handed to the engine per batch during NDJSON import; the loop yields between batches. */
+    importBatchSize?: number;
+    /** Maximum requests executing engine work at once; excess is shed with 429. Omit or 0 to disable. */
+    maxConcurrentRequests?: number;
+    /** Ceiling for a search's `limit`, `offset`, and `group.maxPerGroup`, so one
+     * request cannot ask for an unbounded result set. Excess → 400. Defaults to
+     * 10000, matching the cluster query result window. */
+    maxResultWindow?: number;
+    /** Ceiling for the number of document ids in one multi-get request, so one
+     * request cannot pull an unbounded number of documents. Excess → 400.
+     * Defaults to 10000. */
+    maxFetchDocuments?: number;
+    /** Ceiling for how many per-line failures one import reports and stores, so a
+     * corpus of bad records cannot produce an unbounded response or task record.
+     * Excess failures still count towards the reported total. Defaults to 100. */
+    maxImportErrors?: number;
+    /** Ceiling for how many task records one `/tasks` page returns. Excess → 400. Defaults to 1000. */
+    maxTaskPageSize?: number;
+    /** Ceiling for how many tasks this instance drives at once; excess is shed
+     * with 429. Each running task holds its own working set, and an async import
+     * holds the whole uploaded corpus until it finishes, so this is what bounds
+     * the memory a burst of task requests can claim. Defaults to 4. Set 0 to
+     * accept every task. */
+    maxConcurrentTasks?: number;
+}
+/**
+ * Everything {@link createServer} accepts.
+ *
+ * Every field is optional, and a server created without any of them binds to
+ * loopback with the default limits and no authentication.
+ *
+ * @public
+ */
+interface ServerOptions {
+    /** The server binds to this address, and to loopback by default. Any other address needs an authentication hook. */
+    host?: string;
+    /** The server binds to this port. Pass 0 to let the operating system choose one, then read {@link NarsilServer.listeningPort}. */
+    port?: number;
+    /** These settings control which cross-origin requests the server accepts, and `true` accepts any origin. */
+    cors?: boolean | CorsOptions;
+    /** Authentication or admission gate run before every routed request. */
+    onRequest?: OnRequestHook;
+    /**
+     * Build identity reported verbatim at `/version`: the package version and the
+     * git commit the server was built from, with a dirty-tree flag. Supply it from
+     * the build (a stamped env var or build arg); omit it and `/version` reports
+     * nulls. The values are descriptive only and never gate a request.
+     */
+    build?: {
+        version?: string;
+        gitSha?: string;
+        dirty?: boolean;
+    };
+    /**
+     * Named embedding adapters a JSON `createIndex` request can reference by name.
+     * Embedding adapters are functions and cannot cross JSON, so an index that
+     * needs query-time or ingest-time embedding names a server-registered adapter.
+     */
+    embeddingAdapters?: Record<string, EmbeddingAdapter>;
+    /** These ceilings apply before the server hands any work to the engine. */
+    limits?: ServerLimits;
+    /**
+     * Backing store for long-running task records. Defaults to an in-memory store
+     * that is lost on restart and not shared across instances. Supply any store
+     * that satisfies {@link TaskStore} (Redis, Upstash over HTTP, DynamoDB, a
+     * database) to survive restarts and share task status across instances.
+     */
+    taskStore?: TaskStore;
+    /**
+     * Stable identifier for this server instance, stamped on every task it owns.
+     * Supply a stable value (a pod or container name) so that after a restart the
+     * instance can mark its own previously-running tasks as failed instead of
+     * leaving them stuck. Defaults to a random id, which disables that recovery.
+     */
+    instanceId?: string;
+    /**
+     * Allows binding to a non-loopback address without an
+     * {@link ServerOptions.onRequest} auth hook. The server otherwise refuses to
+     * start in that configuration because it exposes destructive admin endpoints.
+     * Set this only when the address is on a trusted private network where access
+     * is controlled elsewhere.
+     */
+    allowInsecure?: boolean;
+}
+/**
+ * The running HTTP server, as {@link createServer} returns it.
+ *
+ * @public
+ */
+interface NarsilServer {
+    /**
+     * Binds the socket and starts answering requests.
+     *
+     * @throws An error when the address is already in use, or when a non-loopback
+     * address is configured without an authentication hook.
+     */
+    listen(): Promise<void>;
+    /** Closes the socket and ends the server. The engine keeps running. */
+    close(): Promise<void>;
+    /** The server bound to this port, which is what you read after binding to port 0. */
+    readonly listeningPort: number;
+}
+/**
+ * The body every failed response carries, which is what an HTTP client parses
+ * on a non-2xx status.
+ *
+ * @public
+ */
+interface ErrorEnvelope {
+    /** This describes the failure. */
+    error: {
+        /** This code comes from either {@link ServerErrorCodes} or {@link ErrorCodes}. */
+        code: string;
+        /** This describes the failure, for a log or a person. */
+        message: string;
+        /** These values are behind the failure, such as the field or limit involved. */
+        details?: Record<string, unknown>;
+    };
+}
+
+export { ASYNC_IMPORT_CAPABILITY as A, type CorsOptions as C, type ErrorEnvelope as E, type NarsilServer as N, type OnRequestHook as O, REBUILD_ANALYSIS_CAPABILITY as R, SERVER_CAPABILITIES as S, TASK_CANCEL_CAPABILITY as T, TASK_FILTER_CAPABILITY as a, type ServerOptions as b, type RequestContext as c, type RequestDenial as d, type ServerLimits as e };

package/dist/persistence/durability/checkpoint-worker.d.ts

@@ -1,8 +1,10 @@
-import { P as PartitionConfig, c as ScoringMode, V as VectorIndexConfig } from '../../schema-CE6S_aAr.js';
-import '../../adapters-CVqI5NBt.js';
+import { P as PartitionConfig, b as ScoringMode, V as VectorIndexConfig } from '../../schema-emTQfEiX.js';
+import '../../adapters-79OIa4xm.js';
 
 interface IndexMetadata {
     indexName: string;
+    /** The identity a cluster gave the index at creation; absent on a single engine. */
+    indexUuid?: string;
     schema: Record<string, string>;
     language: string;
     partitionCount: number;

package/dist/persistence/durability/checkpoint-worker.mjs

@@ -1,11 +1,14 @@
-import { createDurableDirectory, writeSegmentedCheckpoint } from '../../chunk-WLR5BFL3.mjs';
-import '../../chunk-ASI6AOW2.mjs';
+import { createDurableDirectory, writeSegmentedCheckpoint } from '../../chunk-TEAJO7CU.mjs';
+import '../../chunk-C6VVLMRX.mjs';
+import '../../chunk-RO4EFWE6.mjs';
 import '../../chunk-TGP5M5R7.mjs';
-import '../../chunk-S7RJJMHI.mjs';
+import '../../chunk-OBXS7OZM.mjs';
+import '../../chunk-WDL632CN.mjs';
 import '../../chunk-KMMJMDUA.mjs';
 import '../../chunk-U2BKYEC7.mjs';
-import '../../chunk-UNJZ7ZDB.mjs';
-import '../../chunk-BKAI2WXI.mjs';
+import '../../chunk-BFIH7OV5.mjs';
+import '../../chunk-WTONLDU3.mjs';
+import '../../chunk-EMI63EXP.mjs';
 
 // src/persistence/durability/rebuild.ts
 async function rebuildSnapshotFromDurable(root, metadata, targets, compactionThreshold) {