@100xprompt/chitta 0.1.0

package/src/service.ts

@@ -0,0 +1,40 @@
+// Wires config → fetch-based adapters → the native ACL/retrieval moat. This is
+// the only place that knows about concrete backends; everything above depends on
+// the interfaces in provider.ts. No Python, no SDKs - direct HTTP to Arango/Qdrant
+// and the embedding server.
+
+import { ArangoHttpClient, type ArangoConfig } from "./arango-client"
+import { QdrantVectorService, type QdrantConfig } from "./qdrant-vector"
+import { HttpEmbeddingProvider, type EmbeddingConfig } from "./embeddings"
+import { ArangoGraphProvider } from "./arango-graph-provider"
+import { RetrievalService } from "./retrieval"
+
+export interface ContextConfig {
+  arango: ArangoConfig
+  qdrant: QdrantConfig
+  embeddings: EmbeddingConfig
+  collectionName: string
+}
+
+export interface ContextLog {
+  info: (m: string) => void
+  debug: (m: string) => void
+  error: (m: string, ...a: unknown[]) => void
+}
+
+/** Build a fully-wired retrieval service from config. */
+export function buildContextService(cfg: ContextConfig, log?: ContextLog) {
+  const graph = new ArangoGraphProvider(new ArangoHttpClient(cfg.arango), log)
+  const vector = new QdrantVectorService(cfg.qdrant)
+  const embeddings = new HttpEmbeddingProvider(cfg.embeddings)
+  const retrieval = new RetrievalService({
+    graph,
+    vector,
+    embeddings,
+    collectionName: cfg.collectionName,
+    log,
+  })
+  return { graph, vector, embeddings, retrieval }
+}
+
+export type ContextService = ReturnType<typeof buildContextService>

package/src/types.ts

@@ -0,0 +1,91 @@
+// Shared types for the context-retrieval layer. Ported from the data shapes used
+// across PipesHub's Query service (`retrieval_service.py`) and Arango provider.
+
+/** virtualRecordId -> recordId. Each virtual id maps to the ONE record the user
+ *  is actually permitted to see (the cross-connector leak guard lives here). */
+export type AccessibleMap = Record<string, string>
+
+/** Optional retrieval filters. `kb`/`apps` scope the corpus; the rest are
+ *  metadata facets resolved against graph edges. */
+export interface RetrievalFilters {
+  kb?: string[]
+  apps?: string[]
+  departments?: string[]
+  categories?: string[]
+  subcategories1?: string[]
+  subcategories2?: string[]
+  subcategories3?: string[]
+  languages?: string[]
+  topics?: string[]
+}
+
+/** Metadata filters = everything except kb/apps. */
+export type MetadataFilters = Omit<RetrievalFilters, "kb" | "apps">
+
+export interface RecordDoc {
+  _key: string
+  virtualRecordId?: string
+  origin?: string
+  connectorName?: string
+  connectorId?: string
+  kbId?: string
+  webUrl?: string
+  recordName?: string
+  recordType?: string
+  mimeType?: string
+  previewRenderable?: boolean
+  hideWeburl?: boolean
+  [k: string]: unknown
+}
+
+export interface UserDoc {
+  _key?: string
+  id?: string
+  userId?: string
+  email?: string
+  [k: string]: unknown
+}
+
+export interface SearchResultMeta {
+  virtualRecordId?: string
+  recordId?: string
+  orgId?: string
+  origin?: string
+  connector?: string | null
+  connectorId?: string | null
+  kbId?: string | null
+  webUrl?: string
+  recordName?: string
+  mimeType?: string
+  extension?: string
+  point_id?: string | number
+  [k: string]: unknown
+}
+
+export interface SearchResult {
+  score: number
+  citationType: string
+  metadata: SearchResultMeta
+  content: string
+}
+
+export enum RetrievalStatus {
+  SUCCESS = "SUCCESS",
+  ERROR = "ERROR",
+  EMPTY_RESPONSE = "EMPTY_RESPONSE",
+  ACCESSIBLE_RECORDS_NOT_FOUND = "ACCESSIBLE_RECORDS_NOT_FOUND",
+  VECTOR_DB_EMPTY = "VECTOR_DB_EMPTY",
+}
+
+export interface RetrievalResponse {
+  searchResults: SearchResult[]
+  records: RecordDoc[]
+  status: RetrievalStatus
+  statusCode: number
+  message: string
+  appliedFilters?: { kb: string[]; kb_count: number }
+}
+
+export const ACCESSIBLE_RECORDS_NOT_FOUND_MESSAGE =
+  "No documents are available for you to search yet. Upload files in Collections " +
+  "and/or connect a data source under Connectors so content can be indexed."