unrag 0.1.1 → 0.2.0

package/package.json

@@ -4,7 +4,7 @@
   "bin": {
     "unrag": "./dist/cli/index.js"
   },
-  "version": "0.1.1",
+  "version": "0.2.0",
   "private": false,
   "license": "Apache-2.0",
   "devDependencies": {

package/registry/core/context-engine.ts

@@ -1,8 +1,10 @@
+import { deleteDocuments } from "./delete";
 import { ingest } from "./ingest";
 import { retrieve } from "./retrieve";
 import { defineConfig, resolveConfig } from "./config";
 import type {
   ContextEngineConfig,
+  DeleteInput,
   IngestInput,
   IngestResult,
   ResolvedContextEngineConfig,
@@ -24,6 +26,10 @@ export class ContextEngine {
   async retrieve(input: RetrieveInput): Promise<RetrieveResult> {
     return retrieve(this.config, input);
   }
+
+  async delete(input: DeleteInput): Promise<void> {
+    return deleteDocuments(this.config, input);
+  }
 }
 
 export const createContextEngine = (config: ContextEngineConfig) =>

package/registry/core/delete.ts

@@ -0,0 +1,19 @@
+import type { DeleteInput, ResolvedContextEngineConfig } from "./types";
+
+export const deleteDocuments = async (
+  config: ResolvedContextEngineConfig,
+  input: DeleteInput
+): Promise<void> => {
+  const hasSourceId = "sourceId" in input && typeof input.sourceId === "string";
+  const hasPrefix =
+    "sourceIdPrefix" in input && typeof input.sourceIdPrefix === "string";
+
+  if (hasSourceId === hasPrefix) {
+    // Both true or both false.
+    throw new Error('Provide exactly one of "sourceId" or "sourceIdPrefix".');
+  }
+
+  await config.store.delete(input);
+};
+
+

package/registry/core/index.ts

@@ -1,4 +1,5 @@
 export { ContextEngine, createContextEngine, defineConfig } from "./context-engine";
+export { deleteDocuments } from "./delete";
 export { ingest } from "./ingest";
 export { retrieve } from "./retrieve";
 export { defaultChunker, resolveChunkingOptions } from "./chunking";

package/registry/core/types.ts

@@ -44,7 +44,29 @@ export type EmbeddingProvider = {
   embed: (input: EmbeddingInput) => Promise<number[]>;
 };
 
+export type DeleteInput =
+  | {
+      /** Delete a single logical document by exact `sourceId`. */
+      sourceId: string;
+      sourceIdPrefix?: never;
+    }
+  | {
+      /**
+       * Delete all logical documents whose `sourceId` starts with the prefix.
+       * This matches Unrag's prefix scoping behavior in retrieval.
+       */
+      sourceId?: never;
+      sourceIdPrefix: string;
+    };
+
 export type VectorStore = {
+  /**
+   * Persist (replace) a single document's chunks.
+   *
+   * The store treats `chunks[0].sourceId` as the logical identifier for the document.
+   * Calling `upsert()` multiple times with the same `sourceId` replaces the previously
+   * stored content for that document (including when the chunk count changes).
+   */
   upsert: (chunks: Chunk[]) => Promise<void>;
   query: (params: {
     embedding: number[];
@@ -53,6 +75,7 @@ export type VectorStore = {
       sourceId?: string;
     };
   }) => Promise<Array<Chunk & { score: number }>>;
+  delete: (input: DeleteInput) => Promise<void>;
 };
 
 export type IngestInput = {

package/registry/store/drizzle-postgres-pgvector/store.ts

@@ -1,6 +1,6 @@
 import { documents, chunks, embeddings } from "./schema";
 import type { Chunk, VectorStore } from "../../core/types";
-import { sql, type SQL } from "drizzle-orm";
+import { eq, like, sql, type SQL } from "drizzle-orm";
 import type { PgDatabase } from "drizzle-orm/pg-core";
 
 type DrizzleDb = PgDatabase<any, any, any>;
@@ -44,6 +44,10 @@ export const createDrizzleVectorStore = (db: DrizzleDb): VectorStore => ({
       const head = chunkItems[0]!;
       const documentRow = toDocumentRow(head);
 
+      // Replace-by-sourceId: delete any previously stored document(s) for this logical id.
+      // Cascades to chunks and embeddings.
+      await tx.delete(documents).where(eq(documents.sourceId, head.sourceId));
+
       await tx
         .insert(documents)
         .values(documentRow)
@@ -140,6 +144,17 @@ export const createDrizzleVectorStore = (db: DrizzleDb): VectorStore => ({
       score: Number(row.score),
     }));
   },
+
+  delete: async (input) => {
+    if ("sourceId" in input) {
+      await db.delete(documents).where(eq(documents.sourceId, input.sourceId));
+      return;
+    }
+
+    await db
+      .delete(documents)
+      .where(like(documents.sourceId, input.sourceIdPrefix + "%"));
+  },
 });
 
 

package/registry/store/prisma-postgres-pgvector/store.ts

@@ -21,6 +21,10 @@ export const createPrismaVectorStore = (prisma: PrismaClient): VectorStore => ({
     const documentMetadata = sanitizeMetadata(head.metadata);
 
     await prisma.$transaction(async (tx: { $executeRaw: (query: unknown) => Promise<unknown> }) => {
+      // Replace-by-sourceId: delete any previously stored document(s) for this logical id.
+      // Cascade removes chunks and embeddings.
+      await tx.$executeRaw(sql`delete from documents where source_id = ${head.sourceId}`);
+
       await tx.$executeRaw(
         sql`
           insert into documents (id, source_id, content, metadata)
@@ -128,6 +132,19 @@ export const createPrismaVectorStore = (prisma: PrismaClient): VectorStore => ({
       score: Number(row.score),
     }));
   },
+
+  delete: async (input) => {
+    if ("sourceId" in input) {
+      await prisma.$executeRaw(
+        sql`delete from documents where source_id = ${input.sourceId}`
+      );
+      return;
+    }
+
+    await prisma.$executeRaw(
+      sql`delete from documents where source_id like ${input.sourceIdPrefix + "%"}`
+    );
+  },
 });
 
 

package/registry/store/raw-sql-postgres-pgvector/store.ts

@@ -42,6 +42,12 @@ export const createRawSqlVectorStore = (pool: Pool): VectorStore => ({
       const head = chunkItems[0]!;
       const documentMetadata = sanitizeMetadata(head.metadata);
 
+      // Replace-by-sourceId: delete any previously stored document(s) for this logical id.
+      // Cascades to chunks and embeddings.
+      await client.query(`delete from documents where source_id = $1`, [
+        head.sourceId,
+      ]);
+
       await client.query(
         `
         insert into documents (id, source_id, content, metadata)
@@ -149,6 +155,21 @@ export const createRawSqlVectorStore = (pool: Pool): VectorStore => ({
       score: Number(row.score),
     }));
   },
+
+  delete: async (input) => {
+    await withTx(pool, async (client) => {
+      if ("sourceId" in input) {
+        await client.query(`delete from documents where source_id = $1`, [
+          input.sourceId,
+        ]);
+        return;
+      }
+
+      await client.query(`delete from documents where source_id like $1`, [
+        input.sourceIdPrefix + "%",
+      ]);
+    });
+  },
 });