@absolutejs/rag 0.0.27 → 0.0.28

package/dist/src/index.d.ts

@@ -18,7 +18,7 @@ export { buildRAGEvaluationLeaderboard, buildRAGEvaluationResponse, buildRAGEval
 export { buildRAGAdminActionPresentation, buildRAGAdminActionPresentations, buildRAGAdminJobPresentation, buildRAGAdminJobPresentations, buildRAGAnswerGroundingCaseSnapshotPresentations, buildRAGAnswerGroundingEntityQualityPresentation, buildRAGAnswerGroundingHistoryPresentation, buildRAGAnswerGroundingHistoryRows, buildRAGCitations, buildRAGCitationReferenceMap, buildRAGChunkExcerpts, buildRAGChunkGraphNavigation, buildRAGChunkPreviewGraph, buildRAGChunkPreviewNavigation, buildRAGChunkGraph, buildRAGComparisonTraceDiffRows, buildRAGComparisonTraceSummaryRows, buildRAGCorpusHealthPresentation, buildRAGEvaluationCaseTracePresentations, buildRAGEvaluationEntityQualityPresentation, buildRAGEvaluationHistoryPresentation, buildRAGEvaluationHistoryRows, buildRAGEvaluationSuiteSnapshotHistoryPresentation, buildRAGEvaluationSuiteSnapshotPresentations, buildRAGEvaluationSuiteSnapshotRows, buildRAGGroundedAnswer, buildRAGGroundedAnswerSectionSummaries, buildRAGGroundingOverviewPresentation, buildRAGGroundingProviderCaseComparisonPresentations, buildRAGGroundingProviderOverviewPresentation, buildRAGGroundingProviderPresentations, buildRAGGroundingReferences, buildRAGQualityOverviewPresentation, buildRAGReadinessPresentation, buildRAGSourceLabels, buildRAGSectionRetrievalDiagnostics, buildRAGRerankerComparisonOverviewPresentation, buildRAGRerankerComparisonPresentations, buildRAGRerankerOverviewPresentation, buildRAGRetrievalComparisonOverviewPresentation, buildRAGRetrievalComparisonPresentations, buildRAGRetrievalOverviewPresentation, buildRAGRetrievalTracePresentation, buildRAGSourceGroups, buildRAGSourceSummaries, buildRAGSyncOverviewPresentation, buildRAGSyncSourcePresentation, buildRAGSyncSourcePresentations, } from "./presentation/ui";
 export { createRAGBunS3SyncClient, createRAGDirectorySyncSource, createRAGEmailSyncSource, createRAGFeedSyncSource, createRAGFileSyncStateStore, createRAGGitHubSyncSource, createRAGLinkedConnectorSyncSource, createRAGLinkedGmailEmailSyncSource, createRAGSiteDiscoverySyncSource, createRAGSitemapSyncSource, createRAGStaticEmailSyncClient, previewRAGSyncConflictResolutions, previewRAGSyncExtractionRecovery, resolveRAGSyncExtractionRecovery, resolveRAGSyncConflictResolutions, createRAGStorageSyncSource, createRAGSyncManager, createRAGSyncScheduler, createRAGUrlSyncSource, } from "./sync/sync";
 export { createRAGFileJobStateStore } from "./internal/jobState";
-export { createRAGCollection, ingestDocuments, ingestRAGDocuments, searchDocuments, } from "./retrieval/collection";
+export { createRAGCollection, ingestDocuments, ingestRAGDocuments, ingestRAGSource, removeRAGSource, searchDocuments, } from "./retrieval/collection";
 export { createInMemoryRAGStore } from "./adapters/inMemory";
 export { createSyncRAGStore } from "./adapters/sync";
 export type { SyncRAGStore } from "./adapters/sync";

package/dist/src/retrieval/collection.d.ts

@@ -1,6 +1,8 @@
-import type { RAGCollection, RAGCollectionSearchParams, RAGDocumentIngestInput, RAGQueryResult, RAGUpsertInput } from "../../types/engine";
+import type { RAGCollection, RAGCollectionSearchParams, RAGDocumentIngestInput, RAGIngestSourceInput, RAGIngestSourceResult, RAGQueryResult, RAGRemoveSourceInput, RAGRemoveSourceResult, RAGUpsertInput } from "../../types/engine";
 import type { CreateRAGCollectionOptions } from "../../types/retrieval";
 export declare const createRAGCollection: (options: CreateRAGCollectionOptions) => RAGCollection;
 export declare const ingestDocuments: (collection: RAGCollection, input: RAGUpsertInput) => Promise<void>;
 export declare const ingestRAGDocuments: (collection: RAGCollection, input: RAGDocumentIngestInput) => Promise<void>;
 export declare const searchDocuments: (collection: RAGCollection, input: RAGCollectionSearchParams) => Promise<RAGQueryResult[]>;
+export declare const ingestRAGSource: (collection: RAGCollection, input: RAGIngestSourceInput) => Promise<RAGIngestSourceResult>;
+export declare const removeRAGSource: (collection: RAGCollection, input: RAGRemoveSourceInput) => Promise<RAGRemoveSourceResult>;

package/dist/types/engine.d.ts

@@ -277,10 +277,19 @@ export type RAGDocumentChunkEmbeddingVariant = {
     metadata?: Record<string, unknown>;
     embedding?: number[];
 };
+export type RAGEmbeddingKind = "query" | "passage";
 export type RAGEmbeddingInput = {
     text: string;
     model?: string;
     signal?: AbortSignal;
+    /**
+     * Asymmetric-embedding hint for providers that distinguish how the input is
+     * used (e.g. Pinecone Inference `input_type`, Cohere `input_type`, voyage
+     * `input_type`). The collection passes `"query"` when embedding a search
+     * query and `"passage"` when embedding a chunk/document for ingestion.
+     * Providers that do not support asymmetric embedding ignore this field.
+     */
+    kind?: RAGEmbeddingKind;
 };
 export type RAGEmbeddingFunction = (input: RAGEmbeddingInput) => Promise<number[]>;
 export type RAGEmbeddingProvider = {
@@ -4195,11 +4204,84 @@ export type RAGRetrievalComparisonRuntime = {
     relatedSearchTraces?: RAGSearchTraceStats;
     relatedPruneRun?: RAGSearchTracePruneRun;
 };
+export type RAGIngestSourceInput = {
+    /**
+     * Stable identifier the consumer owns for this document/source. All chunk ids
+     * are derived deterministically from it (`${sourceId}#${index}`) and it is
+     * stamped into every chunk's metadata as `sourceId`, so the whole set can be
+     * deleted or replaced later without the consumer tracking individual ids.
+     */
+    sourceId: string;
+    /** Ingest from in-memory file bytes (any extractor-supported type). */
+    upload?: RAGDocumentUploadInput;
+    /** Ingest from a URL (fetched + extracted by content type). */
+    url?: RAGDocumentUrlInput;
+    /** Ingest from an already-extracted text document. */
+    document?: RAGIngestDocument;
+    /** Extra metadata merged into every chunk (alongside `sourceId`). */
+    metadata?: Record<string, unknown>;
+    /** Chunking options applied to this source. */
+    chunking?: RAGChunkingOptions;
+    chunkingRegistry?: RAGChunkingRegistryLike;
+    /** Custom file extractors (for upload/url ingest). */
+    extractors?: RAGFileExtractor[];
+    extractorRegistry?: RAGFileExtractorRegistryLike;
+    /**
+     * Override the embedding `kind` used for the chunks. Defaults to `"passage"`
+     * so asymmetric embedders index document text with the document/passage
+     * embedding while search continues to use the query embedding.
+     */
+    embedKind?: RAGEmbeddingKind;
+    /**
+     * Whether to remove any existing chunks for this sourceId before ingesting
+     * (idempotent replace). Defaults to `true`.
+     */
+    replace?: boolean;
+    /**
+     * Chunk count from a prior ingest of this sourceId. Used when replacing on
+     * stores that cannot metadata-filter-delete, so the previous deterministic id
+     * set is removed even if the new ingest produces fewer chunks.
+     */
+    previousChunkCount?: number;
+};
+export type RAGIngestSourceResult = {
+    sourceId: string;
+    chunkCount: number;
+    chunkIds: string[];
+};
+export type RAGRemoveSourceInput = {
+    sourceId: string;
+    /**
+     * Number of chunks the source produced (from `RAGIngestSourceResult`).
+     * Required to delete by deterministic id on stores that cannot
+     * metadata-filter-delete (e.g. Pinecone serverless).
+     */
+    chunkCount?: number;
+    /** Explicit chunk ids to delete (takes precedence over `chunkCount`). */
+    chunkIds?: string[];
+    /**
+     * Also issue a metadata-filter delete on `{ sourceId }`. Defaults to `true`;
+     * harmless (and cheap) on filter-capable stores, ignored when unsupported.
+     */
+    filterDelete?: boolean;
+};
+export type RAGRemoveSourceResult = {
+    sourceId: string;
+    deleted: number;
+};
 export type RAGCollection = {
     store: RAGVectorStore;
     search: (input: RAGCollectionSearchParams) => Promise<RAGQueryResult[]>;
     searchWithTrace: (input: RAGCollectionSearchParams) => Promise<RAGCollectionSearchResult>;
     ingest: (input: RAGUpsertInput) => Promise<void>;
+    /**
+     * Ingest a single document as a tracked source: extract → chunk → embed with
+     * the passage/document kind → upsert with deterministic, source-derived chunk
+     * ids. Returns the ids/count so the set is deletable via `removeSource`.
+     */
+    ingestSource: (input: RAGIngestSourceInput) => Promise<RAGIngestSourceResult>;
+    /** Delete every chunk previously ingested under `sourceId`. */
+    removeSource: (input: RAGRemoveSourceInput) => Promise<RAGRemoveSourceResult>;
     clear?: () => Promise<void> | void;
     getStatus?: () => RAGVectorStoreStatus;
     getCapabilities?: () => RAGBackendCapabilities;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@absolutejs/rag",
-  "version": "0.0.27",
+  "version": "0.0.28",
   "homepage": "https://github.com/absolutejs/rag",
   "bugs": {
     "url": "https://github.com/absolutejs/rag/issues"