@absolutejs/absolute 0.19.0-beta.615 → 0.19.0-beta.616

package/dist/types/ai.d.ts

@@ -11,6 +11,7 @@ export type RAGExcerptSelection = {
 export type RAGExcerptModeCounts = Record<RAGExcerptMode, number>;
 export type RAGSource = {
     chunkId: string;
+    corpusKey?: string;
     score: number;
     text: string;
     title?: string;
@@ -268,6 +269,7 @@ export type RAGAnswerWorkflowState = {
 export type RAGStreamStage = 'idle' | 'submitting' | 'retrieving' | 'retrieved' | 'streaming' | 'complete' | 'error';
 export type RAGDocumentChunk = {
     chunkId: string;
+    corpusKey?: string;
     text: string;
     title?: string;
     source?: string;
@@ -298,6 +300,7 @@ export type RAGFileExtractionInput = {
     contentType?: string;
     metadata?: Record<string, unknown>;
     chunking?: RAGChunkingOptions;
+    extractorRegistry?: RAGFileExtractorRegistryLike;
 };
 export type RAGExtractedFileDocument = RAGIngestDocument & {
     contentType?: string;
@@ -308,6 +311,27 @@ export type RAGFileExtractor = {
     supports: (input: RAGFileExtractionInput) => boolean | Promise<boolean>;
     extract: (input: RAGFileExtractionInput) => RAGExtractedFileDocument | RAGExtractedFileDocument[] | Promise<RAGExtractedFileDocument | RAGExtractedFileDocument[]>;
 };
+export type RAGFileExtractorRegistryInput = RAGFileExtractionInput & {
+    inferredContentType?: string | null;
+    inferredExtension?: string | null;
+    inferredFormat?: RAGContentFormat;
+};
+export type RAGFileExtractorRegistration = {
+    extractor: RAGFileExtractor;
+    name?: string;
+    priority?: number;
+    contentTypes?: string[];
+    extensions?: string[];
+    formats?: RAGContentFormat[];
+    names?: string[];
+    match?: (input: RAGFileExtractorRegistryInput) => boolean | Promise<boolean>;
+};
+export type RAGFileExtractorRegistry = {
+    registrations: RAGFileExtractorRegistration[];
+    includeDefaults?: boolean;
+    defaultOrder?: 'registry_first' | 'defaults_first';
+};
+export type RAGFileExtractorRegistryLike = RAGFileExtractorRegistry | RAGFileExtractorRegistration[];
 export type RAGMediaTranscriptSegment = {
     text: string;
     startMs?: number;
@@ -371,8 +395,36 @@ export type RAGChunkingOptions = {
     minChunkLength?: number;
     strategy?: RAGChunkingStrategy;
 };
+export type RAGChunkingProfileInput = {
+    document: RAGIngestDocument;
+    format: RAGContentFormat;
+    normalizedText: string;
+    metadata: Record<string, unknown>;
+    sourceNativeKind?: string;
+    defaults?: RAGChunkingOptions;
+};
+export type RAGChunkingProfile = {
+    name: string;
+    resolve: (input: RAGChunkingProfileInput) => Partial<RAGChunkingOptions> | undefined;
+};
+export type RAGChunkingProfileRegistration = {
+    name?: string;
+    documentIds?: string[];
+    formats?: RAGContentFormat[];
+    priority?: number;
+    profile: Partial<RAGChunkingOptions> | {
+        options?: Partial<RAGChunkingOptions>;
+    };
+    sourceNativeKinds?: string[];
+    sources?: string[];
+};
+export type RAGChunkingRegistry = {
+    profiles: Array<RAGChunkingProfile | RAGChunkingProfileRegistration>;
+};
+export type RAGChunkingRegistryLike = RAGChunkingRegistry | Array<RAGChunkingProfile | RAGChunkingProfileRegistration>;
 export type RAGIngestDocument = {
     text: string;
+    corpusKey?: string;
     id?: string;
     title?: string;
     source?: string;
@@ -389,14 +441,18 @@ export type RAGDocumentUrlInput = {
     metadata?: Record<string, unknown>;
     chunking?: RAGChunkingOptions;
     extractors?: RAGFileExtractor[];
+    extractorRegistry?: RAGFileExtractorRegistryLike;
 };
 export type RAGDocumentUrlIngestInput = {
     baseMetadata?: Record<string, unknown>;
     defaultChunking?: RAGChunkingOptions;
+    chunkingRegistry?: RAGChunkingRegistryLike;
     extractors?: RAGFileExtractor[];
+    extractorRegistry?: RAGFileExtractorRegistryLike;
     urls: RAGDocumentUrlInput[];
 };
 export type RAGPreparedDocument = {
+    corpusKey?: string;
     documentId: string;
     title: string;
     source: string;
@@ -409,6 +465,7 @@ export type RAGDocumentFileInput = Omit<RAGIngestDocument, 'text'> & {
     path: string;
     contentType?: string;
     extractors?: RAGFileExtractor[];
+    extractorRegistry?: RAGFileExtractorRegistryLike;
 };
 export type RAGDirectoryIngestInput = {
     directory: string;
@@ -416,7 +473,9 @@ export type RAGDirectoryIngestInput = {
     includeExtensions?: string[];
     baseMetadata?: Record<string, unknown>;
     defaultChunking?: RAGChunkingOptions;
+    chunkingRegistry?: RAGChunkingRegistryLike;
     extractors?: RAGFileExtractor[];
+    extractorRegistry?: RAGFileExtractorRegistryLike;
 };
 export type RAGQueryInput = {
     queryVector: number[];
@@ -531,6 +590,7 @@ export type RAGUpsertInput = {
 export type RAGDocumentIngestInput = {
     documents: RAGIngestDocument[];
     defaultChunking?: RAGChunkingOptions;
+    chunkingRegistry?: RAGChunkingRegistryLike;
 };
 export type RAGDocumentUploadInput = {
     name: string;
@@ -546,10 +606,13 @@ export type RAGDocumentUploadInput = {
 export type RAGDocumentUploadIngestInput = {
     baseMetadata?: Record<string, unknown>;
     defaultChunking?: RAGChunkingOptions;
+    chunkingRegistry?: RAGChunkingRegistryLike;
     extractors?: RAGFileExtractor[];
+    extractorRegistry?: RAGFileExtractorRegistryLike;
     uploads: RAGDocumentUploadInput[];
 };
 export type RAGIndexedDocument = {
+    corpusKey?: string;
     id: string;
     title: string;
     source: string;
@@ -876,21 +939,30 @@ export type RAGCorpusHealth = {
     staleDocuments: string[];
     averageChunksPerDocument: number;
     inspection?: {
+        corpusKeys: Record<string, number>;
         sourceNativeKinds: Record<string, number>;
+        extractorRegistryMatches: Record<string, number>;
+        chunkingProfiles: Record<string, number>;
         documentsWithSourceLabels: number;
         chunksWithSourceLabels: number;
         sampleDocuments: Array<{
+            corpusKey?: string;
             id: string;
             title: string;
             source: string;
             sourceNativeKind?: string;
+            extractorRegistryMatch?: string;
+            chunkingProfile?: string;
             labels?: RAGSourceLabels;
         }>;
         sampleChunks: Array<{
             chunkId: string;
+            corpusKey?: string;
             documentId?: string;
             source?: string;
             sourceNativeKind?: string;
+            extractorRegistryMatch?: string;
+            chunkingProfile?: string;
             labels?: RAGSourceLabels;
         }>;
     };
@@ -917,6 +989,22 @@ export type RAGAdminJobRecord = {
     target?: string;
     error?: string;
 };
+export type RAGJobState = {
+    adminActions: RAGAdminActionRecord[];
+    ingestJobs: RAGIngestJobRecord[];
+    adminJobs: RAGAdminJobRecord[];
+    syncJobs: RAGAdminJobRecord[];
+};
+export type RAGJobStateStore = {
+    load: () => Promise<Partial<RAGJobState> | undefined> | Partial<RAGJobState> | undefined;
+    save: (state: RAGJobState) => Promise<void> | void;
+};
+export type RAGJobHistoryRetention = {
+    maxAdminActions?: number;
+    maxAdminJobs?: number;
+    maxIngestJobs?: number;
+    maxSyncJobs?: number;
+};
 export type RAGAdminCapabilities = {
     canClearIndex: boolean;
     canCreateDocument: boolean;
@@ -931,6 +1019,46 @@ export type RAGAdminCapabilities = {
     canSyncAllSources: boolean;
     canSyncSource: boolean;
 };
+export type RAGAuthorizedAction = 'clear_index' | 'create_document' | 'delete_document' | 'ingest' | 'list_sync_sources' | 'manage_retrieval_admin' | 'manage_retrieval_baselines' | 'prune_search_traces' | 'reindex_document' | 'reindex_source' | 'reseed' | 'reset' | 'sync_all_sources' | 'sync_source';
+export type RAGAuthorizationResource = {
+    documentId?: string;
+    path?: string;
+    source?: string;
+    sourceId?: string;
+};
+export type RAGAuthorizationDecision = boolean | {
+    allowed: boolean;
+    reason?: string;
+};
+export type RAGAuthorizationContext = {
+    action: RAGAuthorizedAction;
+    request: Request;
+    resource?: RAGAuthorizationResource;
+};
+export type RAGAuthorizationProvider = (context: RAGAuthorizationContext) => Promise<RAGAuthorizationDecision> | RAGAuthorizationDecision;
+export type RAGAccessScope = {
+    allowedComparisonGroupKeys?: string[];
+    allowedCorpusKeys?: string[];
+    allowedDocumentIds?: string[];
+    allowedSourcePrefixes?: string[];
+    allowedSources?: string[];
+    allowedSyncSourceIds?: string[];
+    requiredMetadata?: Record<string, unknown>;
+};
+export type RAGAccessScopeProvider = (request: Request) => Promise<RAGAccessScope | undefined> | RAGAccessScope | undefined;
+export type RAGAccessControlContextResolver<TContext = unknown> = (request: Request) => Promise<TContext | undefined> | TContext | undefined;
+export type RAGAccessControlAuthorizeResolver<TContext = unknown> = (input: RAGAuthorizationContext & {
+    context: TContext | undefined;
+}) => Promise<RAGAuthorizationDecision> | RAGAuthorizationDecision;
+export type RAGAccessControlScopeResolver<TContext = unknown> = (input: {
+    context: TContext | undefined;
+    request: Request;
+}) => Promise<RAGAccessScope | undefined> | RAGAccessScope | undefined;
+export type CreateRAGAccessControlOptions<TContext = unknown> = {
+    resolveContext: RAGAccessControlContextResolver<TContext>;
+    authorize?: RAGAccessControlAuthorizeResolver<TContext>;
+    resolveScope?: RAGAccessControlScopeResolver<TContext>;
+};
 export type RAGSyncSourceStatus = 'idle' | 'running' | 'completed' | 'failed' | 'disabled';
 export type RAGSyncSourceRecord = {
     id: string;
@@ -1007,7 +1135,9 @@ export type RAGDirectorySyncSourceOptions = {
     description?: string;
     baseMetadata?: Record<string, unknown>;
     defaultChunking?: RAGChunkingOptions;
+    chunkingRegistry?: RAGChunkingRegistryLike;
     extractors?: RAGFileExtractor[];
+    extractorRegistry?: RAGFileExtractorRegistryLike;
     includeExtensions?: string[];
     metadata?: Record<string, unknown>;
     recursive?: boolean;
@@ -1021,7 +1151,9 @@ export type RAGUrlSyncSourceOptions = {
     description?: string;
     baseMetadata?: Record<string, unknown>;
     defaultChunking?: RAGChunkingOptions;
+    chunkingRegistry?: RAGChunkingRegistryLike;
     extractors?: RAGFileExtractor[];
+    extractorRegistry?: RAGFileExtractorRegistryLike;
     metadata?: Record<string, unknown>;
     retryAttempts?: number;
     retryDelayMs?: number;
@@ -1036,7 +1168,9 @@ export type RAGStorageSyncSourceOptions = {
     maxKeys?: number;
     baseMetadata?: Record<string, unknown>;
     defaultChunking?: RAGChunkingOptions;
+    chunkingRegistry?: RAGChunkingRegistryLike;
     extractors?: RAGFileExtractor[];
+    extractorRegistry?: RAGFileExtractorRegistryLike;
     metadata?: Record<string, unknown>;
     retryAttempts?: number;
     retryDelayMs?: number;
@@ -1086,7 +1220,9 @@ export type RAGEmailSyncSourceOptions = {
     maxResults?: number;
     baseMetadata?: Record<string, unknown>;
     defaultChunking?: RAGChunkingOptions;
+    chunkingRegistry?: RAGChunkingRegistryLike;
     extractors?: RAGFileExtractor[];
+    extractorRegistry?: RAGFileExtractorRegistryLike;
     metadata?: Record<string, unknown>;
     retryAttempts?: number;
     retryDelayMs?: number;
@@ -1306,6 +1442,7 @@ export type RAGMutationResponse = {
 export type RAGEvaluationCase = {
     id: string;
     query: string;
+    corpusKey?: string;
     topK?: number;
     model?: string;
     scoreThreshold?: number;
@@ -1569,6 +1706,7 @@ export type RAGEvaluationInput = {
 };
 export type RAGEvaluationCaseResult = {
     caseId: string;
+    corpusKey?: string;
     query: string;
     label?: string;
     status: 'pass' | 'partial' | 'fail';
@@ -1599,6 +1737,7 @@ export type RAGEvaluationSummary = {
 };
 export type RAGEvaluationResponse = {
     ok: true;
+    corpusKeys?: string[];
     cases: RAGEvaluationCaseResult[];
     summary: RAGEvaluationSummary;
     elapsedMs: number;
@@ -1684,6 +1823,7 @@ export type RAGEvaluationRunDiff = {
 };
 export type RAGEvaluationCaseTraceSnapshot = {
     caseId: string;
+    corpusKey?: string;
     label?: string;
     query: string;
     status: RAGEvaluationCaseResult['status'];
@@ -1922,6 +2062,7 @@ export type RAGRetrievalTraceTrend = {
 };
 export type RAGSearchTraceResultSnapshot = {
     chunkId: string;
+    corpusKey?: string;
     score: number;
     source?: string;
     title?: string;
@@ -2111,6 +2252,7 @@ export type RAGRetrievalComparisonSummary = {
 export type RAGRetrievalComparison = {
     suiteId: string;
     suiteLabel: string;
+    corpusKeys?: string[];
     entries: RAGRetrievalComparisonEntry[];
     summary: RAGRetrievalComparisonSummary;
     leaderboard: RAGEvaluationLeaderboardEntry[];
@@ -2139,6 +2281,7 @@ export type RAGRetrievalComparisonRun = {
     label: string;
     suiteId: string;
     suiteLabel: string;
+    corpusKeys?: string[];
     groupKey?: string;
     tags?: string[];
     startedAt: number;
@@ -3545,6 +3688,10 @@ export type RAGChatPluginConfig = AIChatPluginConfig & {
     path?: string;
     ragStore?: RAGVectorStore;
     collection?: RAGCollection;
+    jobStateStore?: RAGJobStateStore;
+    authorizeRAGAction?: RAGAuthorizationProvider;
+    resolveRAGAccessScope?: RAGAccessScopeProvider;
+    jobHistoryRetention?: RAGJobHistoryRetention;
     searchTraceStore?: RAGSearchTraceStore;
     searchTraceRetention?: RAGSearchTracePruneInput;
     searchTraceRetentionSchedule?: RAGSearchTraceRetentionSchedule;

package/dist/vue/ai/index.js

@@ -187,7 +187,7 @@ var parseAIMessage = (raw) => {
     return null;
   }
 };
-var serializeAIMessage = (msg) => JSON.stringify(msg);
+var serializeAIMessage = (message) => JSON.stringify(message);
 
 // src/ai/rag/grounding.ts
 var getContextString = (value) => typeof value === "string" && value.trim().length > 0 ? value.trim() : undefined;
@@ -1275,7 +1275,10 @@ var evaluateRAGCollectionCases = async ({
     const expectedIds = normalizeExpectedIds(mode === "chunkId" ? caseInput.expectedChunkIds ?? [] : mode === "source" ? caseInput.expectedSources ?? [] : caseInput.expectedDocumentIds ?? []);
     const topK = typeof caseInput.topK === "number" ? caseInput.topK : typeof input.topK === "number" ? input.topK : defaultTopK;
     const searchInput = {
-      filter: typeof caseInput.filter === "object" ? caseInput.filter : input.filter,
+      filter: caseInput.corpusKey ? {
+        ...(typeof caseInput.filter === "object" ? caseInput.filter : input.filter) ?? {},
+        corpusKey: caseInput.corpusKey
+      } : typeof caseInput.filter === "object" ? caseInput.filter : input.filter,
       model: caseInput.model ?? input.model,
       query,
       rerank,
@@ -1677,6 +1680,7 @@ var buildEvaluationCaseTraceSnapshot = ({
   return {
     candidateTopK: currentTrace?.candidateTopK ?? 0,
     caseId: caseResult.caseId,
+    corpusKey: caseResult.corpusKey,
     inputFilter: filter,
     finalCount: currentTrace?.resultCounts.final ?? 0,
     label: caseResult.label,
@@ -3229,11 +3233,15 @@ var persistRAGRetrievalReleaseLaneEscalationPolicyHistory = async ({
 };
 var buildRAGEvaluationResponse = (cases) => {
   const totalCases = cases.length;
+  const corpusKeys = [
+    ...new Set(cases.flatMap((entry) => entry.corpusKey ?? []))
+  ];
   const passedCases = cases.filter((entry) => entry.status === "pass").length;
   const partialCases = cases.filter((entry) => entry.status === "partial").length;
   const failedCases = cases.filter((entry) => entry.status === "fail").length;
   return {
     cases,
+    ...corpusKeys.length > 0 ? { corpusKeys } : {},
     elapsedMs: cases.reduce((sum, result) => sum + result.elapsedMs, 0),
     ok: true,
     passingRate: totalCases > 0 ? passedCases / totalCases * 100 : 0,
@@ -3459,6 +3467,9 @@ var compareRAGRetrievalStrategies = async ({
     traceSummary: entry.traceSummary
   })));
   return {
+    corpusKeys: [
+      ...new Set(entries.flatMap((entry) => entry.response.corpusKeys ?? []))
+    ],
     entries,
     leaderboard,
     summary: summarizeRAGRetrievalComparison(entries),
@@ -3491,6 +3502,7 @@ var compareRAGRetrievalTraceSummaries = (current, previous) => ({
 });
 var buildSearchTraceResultSnapshots = (results) => results.map((result) => ({
   chunkId: result.chunkId,
+  corpusKey: result.corpusKey ?? (typeof result.metadata?.corpusKey === "string" ? result.metadata.corpusKey : undefined),
   documentId: typeof result.metadata?.documentId === "string" ? result.metadata.documentId : undefined,
   score: result.score,
   source: result.source,
@@ -3690,6 +3702,7 @@ var summarizeRAGEvaluationCase = ({
   const status = expectedCount === 0 ? "partial" : matchedCount === expectedCount ? "pass" : matchedCount > 0 ? "partial" : "fail";
   return {
     caseId: caseInput.id ?? `case-${caseIndex + 1}`,
+    corpusKey: caseInput.corpusKey,
     elapsedMs,
     expectedCount,
     expectedIds,
@@ -4289,10 +4302,14 @@ var buildProvenanceLabel2 = (metadata) => {
   const transcriptSource = getContextString2(metadata.transcriptSource);
   const pdfTextMode = getContextString2(metadata.pdfTextMode);
   const ocrEngine = getContextString2(metadata.ocrEngine);
+  const extractorRegistryMatch = getContextString2(metadata.extractorRegistryMatch);
+  const chunkingProfile = getContextString2(metadata.chunkingProfile);
   const ocrConfidence = getContextNumber2(metadata.ocrRegionConfidence) ?? getContextNumber2(metadata.ocrConfidence);
   const labels = [
     pdfTextMode ? `PDF ${pdfTextMode}` : "",
     ocrEngine ? `OCR ${ocrEngine}` : "",
+    extractorRegistryMatch ? `Extractor ${extractorRegistryMatch}` : "",
+    chunkingProfile ? `Chunking ${chunkingProfile}` : "",
     typeof ocrConfidence === "number" ? `Confidence ${ocrConfidence.toFixed(2)}` : "",
     mediaKind ? `Media ${mediaKind}` : "",
     transcriptSource ? `Transcript ${transcriptSource}` : "",
@@ -5721,73 +5738,73 @@ var buildRAGAnswerGroundingHistoryPresentation = (history) => ({
 import { onUnmounted, ref, shallowRef } from "vue";
 
 // src/ai/client/actions.ts
-var serverMessageToAction = (msg) => {
-  switch (msg.type) {
+var serverMessageToAction = (message) => {
+  switch (message.type) {
     case "chunk":
       return {
-        content: msg.content,
-        conversationId: msg.conversationId,
-        messageId: msg.messageId,
+        content: message.content,
+        conversationId: message.conversationId,
+        messageId: message.messageId,
         type: "chunk"
       };
     case "thinking":
       return {
-        content: msg.content,
-        conversationId: msg.conversationId,
-        messageId: msg.messageId,
+        content: message.content,
+        conversationId: message.conversationId,
+        messageId: message.messageId,
         type: "thinking"
       };
     case "tool_status":
       return {
-        conversationId: msg.conversationId,
-        input: msg.input,
-        messageId: msg.messageId,
-        name: msg.name,
-        result: msg.result,
-        status: msg.status,
+        conversationId: message.conversationId,
+        input: message.input,
+        messageId: message.messageId,
+        name: message.name,
+        result: message.result,
+        status: message.status,
         type: "tool_status"
       };
     case "image":
       return {
-        conversationId: msg.conversationId,
-        data: msg.data,
-        format: msg.format,
-        imageId: msg.imageId,
-        isPartial: msg.isPartial,
-        messageId: msg.messageId,
-        revisedPrompt: msg.revisedPrompt,
+        conversationId: message.conversationId,
+        data: message.data,
+        format: message.format,
+        imageId: message.imageId,
+        isPartial: message.isPartial,
+        messageId: message.messageId,
+        revisedPrompt: message.revisedPrompt,
         type: "image"
       };
     case "complete":
       return {
-        conversationId: msg.conversationId,
-        durationMs: msg.durationMs,
-        messageId: msg.messageId,
-        model: msg.model,
-        sources: msg.sources,
+        conversationId: message.conversationId,
+        durationMs: message.durationMs,
+        messageId: message.messageId,
+        model: message.model,
+        sources: message.sources,
         type: "complete",
-        usage: msg.usage
+        usage: message.usage
       };
     case "rag_retrieving":
       return {
-        conversationId: msg.conversationId,
-        messageId: msg.messageId,
-        retrievalStartedAt: msg.retrievalStartedAt,
+        conversationId: message.conversationId,
+        messageId: message.messageId,
+        retrievalStartedAt: message.retrievalStartedAt,
         type: "rag_retrieving"
       };
     case "rag_retrieved":
       return {
-        conversationId: msg.conversationId,
-        messageId: msg.messageId,
-        retrievalDurationMs: msg.retrievalDurationMs,
-        retrievalStartedAt: msg.retrievalStartedAt,
-        retrievedAt: msg.retrievedAt,
-        sources: msg.sources,
-        trace: msg.trace,
+        conversationId: message.conversationId,
+        messageId: message.messageId,
+        retrievalDurationMs: message.retrievalDurationMs,
+        retrievalStartedAt: message.retrievalStartedAt,
+        retrievedAt: message.retrievedAt,
+        sources: message.sources,
+        trace: message.trace,
         type: "rag_retrieved"
       };
     case "error":
-      return { message: msg.message, type: "error" };
+      return { message: message.message, type: "error" };
     default:
       return null;
   }
@@ -6249,8 +6266,8 @@ var useAIStream = (path, conversationId) => {
     }
   };
   unsubscribeStore = store.subscribe(syncState);
-  unsubscribeConnection = connection.subscribe((msg) => {
-    const action = serverMessageToAction(msg);
+  unsubscribeConnection = connection.subscribe((message) => {
+    const action = serverMessageToAction(message);
     if (action) {
       store.dispatch(action);
     }
@@ -8223,5 +8240,5 @@ export {
   AIStreamKey
 };
 
-//# debugId=044DA3D708DF041064756E2164756E21
+//# debugId=588CD05E6FC4209364756E2164756E21
 //# sourceMappingURL=index.js.map