fix: fix syncing edge with core version

Co-authored-by: Michael Schramm <michael@tucan.ai>

.changeset/eighty-waves-march.md

@@ -0,0 +1,5 @@
+---
+"llamaindex": patch
+---
+
+Add auto create milvus collection and add milvus node metadata

.changeset/fluffy-squids-play.md

@@ -0,0 +1,5 @@
+---
+"llamaindex": patch
+---
+
+Use Pinecone namespaces for all operations

.changeset/yellow-chicken-study.md

@@ -0,0 +1,5 @@
+---
+"llamaindex": patch
+---
+
+Add support for edge runtime by using @llamaindex/edge

.eslintrc.js

@@ -11,5 +11,13 @@ module.exports = {
     "max-params": ["error", 4],
     "prefer-const": "error",
   },
+  overrides: [
+    {
+      files: ["examples/**/*.ts"],
+      rules: {
+        "turbo/no-undeclared-env-vars": "off",
+      },
+    },
+  ],
   ignorePatterns: ["dist/", "lib/"],
 };

.github/workflows/e2e.yml

@@ -1,68 +0,0 @@
-name: E2E Tests
-on:
-  push:
-    branches: [main]
-  pull_request:
-    paths:
-      - "packages/create-llama/**"
-      - ".github/workflows/e2e.yml"
-    branches: [main]
-
-env:
-  POETRY_VERSION: "1.6.1"
-
-jobs:
-  e2e:
-    name: create-llama
-    timeout-minutes: 60
-    strategy:
-      fail-fast: true
-      matrix:
-        node-version: [18, 20]
-        python-version: ["3.11"]
-        os: [macos-latest, windows-latest]
-    defaults:
-      run:
-        shell: bash
-    runs-on: ${{ matrix.os }}
-    steps:
-      - uses: actions/checkout@v4
-      - name: Set up python ${{ matrix.python-version }}
-        uses: actions/setup-python@v4
-        with:
-          python-version: ${{ matrix.python-version }}
-      - name: Install Poetry
-        uses: snok/install-poetry@v1
-        with:
-          version: ${{ env.POETRY_VERSION }}
-      - uses: pnpm/action-setup@v2
-      - name: Setup Node.js ${{ matrix.node-version }}
-        uses: actions/setup-node@v4
-        with:
-          node-version: ${{ matrix.node-version }}
-          cache: "pnpm"
-      - name: Install dependencies
-        run: pnpm install
-      - name: Install Playwright Browsers
-        run: pnpm exec playwright install --with-deps
-        working-directory: ./packages/create-llama
-      - name: Build create-llama
-        run: pnpm run build
-        working-directory: ./packages/create-llama
-      - name: Pack
-        run: pnpm pack --pack-destination ./output
-        working-directory: ./packages/create-llama
-      - name: Extract Pack
-        run: tar -xvzf ./output/*.tgz -C ./output
-        working-directory: ./packages/create-llama
-      - name: Run Playwright tests
-        run: pnpm exec playwright test
-        env:
-          OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
-        working-directory: ./packages/create-llama
-      - uses: actions/upload-artifact@v3
-        if: always()
-        with:
-          name: playwright-report
-          path: ./packages/create-llama/playwright-report/
-          retention-days: 30

.github/workflows/publish.yml

@@ -0,0 +1,36 @@
+name: Publish
+on:
+  push:
+    branches:
+      - main
+
+jobs:
+  publish:
+    runs-on: ubuntu-latest
+
+    permissions:
+      contents: read
+      id-token: write
+
+    steps:
+      - uses: actions/checkout@v4
+      - uses: pnpm/action-setup@v2
+      - name: Setup Node.js
+        uses: actions/setup-node@v4
+        with:
+          node-version-file: ".nvmrc"
+          cache: "pnpm"
+      - name: Install dependencies
+        run: pnpm install
+
+      - name: Publish @llamaindex/env
+        run: npx jsr publish
+        working-directory: packages/env
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Publish @llamaindex/core
+        run: npx jsr publish --allow-slow-types
+        working-directory: packages/core
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

.github/workflows/test.yml

@@ -44,6 +44,24 @@ jobs:
           name: typecheck-build-dist
           path: ./packages/core/dist
           if-no-files-found: error
+  core-edge-runtime:
+    runs-on: ubuntu-latest
+
+    steps:
+      - uses: actions/checkout@v4
+      - uses: pnpm/action-setup@v2
+      - name: Setup Node.js
+        uses: actions/setup-node@v4
+        with:
+          node-version-file: ".nvmrc"
+          cache: "pnpm"
+      - name: Install dependencies
+        run: pnpm install
+      - name: Build
+        run: pnpm run build --filter @llamaindex/edge
+      - name: Build Edge Runtime
+        run: pnpm run build
+        working-directory: ./packages/edge/e2e/test-edge-runtime
   typecheck-examples:
     runs-on: ubuntu-latest
 

.gitignore

@@ -44,6 +44,7 @@ test-results/
 playwright-report/
 blob-report/
 playwright/.cache/
+.tsbuildinfo
 
 # intellij
 **/.idea

CONTRIBUTING.md

@@ -84,8 +84,7 @@ Any changes you make should be reflected in the browser. If you need to regenera
 To publish a new version of the library, run
 
 ```shell
-pnpm new-llamaindex
-pnpm new-create-llama
+pnpm new-version
 pnpm release
 git push # push to the main branch
 git push --tags

README.md

@@ -121,6 +121,42 @@ const nextConfig = {
 module.exports = nextConfig;
 ```
 
+### NextJS with Milvus:
+
+As proto files are not loaded per default in NextJS, you'll need to add the following to your next.config.js to have it load the proto files.
+
+```js
+const path = require("path");
+const CopyWebpackPlugin = require("copy-webpack-plugin");
+
+// next.config.js
+/** @type {import('next').NextConfig} */
+const nextConfig = {
+  webpack: (config, { isServer }) => {
+    if (isServer) {
+      // Copy the proto files to the server build directory
+      config.plugins.push(
+        new CopyWebpackPlugin({
+          patterns: [
+            {
+              from: path.join(
+                __dirname,
+                "node_modules/@zilliz/milvus2-sdk-node/dist",
+              ),
+              to: path.join(__dirname, ".next"),
+            },
+          ],
+        }),
+      );
+    }
+    // Important: return the modified config
+    return config;
+  },
+};
+
+module.exports = nextConfig;
+```
+
 ## Supported LLMs:
 
 - OpenAI GPT-3.5-turbo and GPT-4

apps/docs/CHANGELOG.md

@@ -1,5 +1,12 @@
 # docs
 
+## 0.0.4
+
+### Patch Changes
+
+- Updated dependencies [5116ad8]
+  - @llamaindex/env@0.0.5
+
 ## 0.0.3
 
 ### Patch Changes

apps/docs/docs/modules/embeddings/available_embeddings/huggingface.md

@@ -23,3 +23,15 @@ const results = await queryEngine.query({
   query,
 });
 ```
+
+Per default, `HuggingFaceEmbedding` is using the `Xenova/all-MiniLM-L6-v2` model. You can change the model by passing the `modelType` parameter to the constructor.
+If you're not using a quantized model, set the `quantized` parameter to `false`.
+
+For example, to use the not quantized `BAAI/bge-small-en-v1.5` model, you can use the following code:
+
+```
+const embedModel = new HuggingFaceEmbedding({
+  modelType: "BAAI/bge-small-en-v1.5",
+  quantized: false,
+});
+```

apps/docs/docs/modules/evaluation/modules/correctness.md

@@ -53,10 +53,6 @@ const evaluator = new CorrectnessEvaluator({
   serviceContext: ctx,
 });
 
-const response = await queryEngine.query({
-  query,
-});
-
 const result = await evaluator.evaluateResponse({
   query,
   response,

apps/docs/docs/modules/ingestion_pipeline/transformations.md

@@ -1,6 +1,6 @@
 # Transformations
 
-A transformation is something that takes a list of nodes as an input, and returns a list of nodes. Each component that implements the Transformatio class has both a `transform` definition responsible for transforming the nodes
+A transformation is something that takes a list of nodes as an input, and returns a list of nodes. Each component that implements the Transformation class has both a `transform` definition responsible for transforming the nodes.
 
 Currently, the following components are Transformation objects:
 

apps/docs/docs/modules/node_postprocessors/index.md

@@ -36,7 +36,7 @@ const processor = new SimilarityPostprocessor({
   similarityCutoff: 0.7,
 });
 
-const filteredNodes = processor.postprocessNodes(nodes);
+const filteredNodes = await processor.postprocessNodes(nodes);
 
 // cohere rerank: rerank nodes given query using trained model
 const reranker = new CohereRerank({
@@ -100,7 +100,7 @@ const response = await queryEngine.query("<user_query>");
 ```ts
 import { SimilarityPostprocessor } from "llamaindex";
 
-nodes = await index.asRetriever().retrieve("test query str");
+nodes = await index.asRetriever().retrieve({ query: "test query str" });
 
 const processor = new SimilarityPostprocessor({
   similarityCutoff: 0.7,

apps/docs/docs/modules/retriever.md

@@ -11,7 +11,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Fetch nodes!
-const nodesWithScore = await retriever.retrieve("query string");
+const nodesWithScore = await retriever.retrieve({ query: "query string" });
 ```
 
 ## API Reference

apps/docs/i18n/ar/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // جلب العقد!
-const nodesWithScore = await retriever.retrieve("سلسلة الاستعلام");
+const nodesWithScore = await retriever.retrieve({ query: "سلسلة الاستعلام" });
 ```
 
 ## مرجع الواجهة البرمجية (API Reference)

apps/docs/i18n/bg/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Извличане на върхове!
-const nodesWithScore = await retriever.retrieve("query string");
+const nodesWithScore = await retriever.retrieve({ query: "query string" });
 ```
 
 ## API Reference (API справка)

apps/docs/i18n/cat/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const recuperador = vector_index.asRetriever();
 recuperador.similarityTopK = 3;
 
 // Obteniu els nodes!
-const nodesAmbPuntuació = await recuperador.retrieve("cadena de consulta");
+const nodesAmbPuntuació = await recuperador.retrieve({ query: "cadena de consulta" });
 ```
 
 ## Referència de l'API

apps/docs/i18n/cs/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Získání uzlů!
-const nodesWithScore = await retriever.retrieve("dotazovací řetězec");
+const nodesWithScore = await retriever.retrieve({ query: "dotazovací řetězec" });
 ```
 
 ## API Reference (Odkazy na rozhraní)

apps/docs/i18n/da/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Hent noder!
-const nodesWithScore = await retriever.retrieve("forespørgselsstreng");
+const nodesWithScore = await retriever.retrieve({ query: "forespørgselsstreng" });
 ```
 
 ## API Reference

apps/docs/i18n/de/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Knoten abrufen!
-const nodesWithScore = await retriever.retrieve("Abfragezeichenfolge");
+const nodesWithScore = await retriever.retrieve({ query: "Abfragezeichenfolge" });
 ```
 
 ## API-Referenz

apps/docs/i18n/el/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Ανάκτηση κόμβων!
-const nodesWithScore = await retriever.retrieve("συμβολοσειρά ερωτήματος");
+const nodesWithScore = await retriever.retrieve({ query: "συμβολοσειρά ερωτήματος" });
 ```
 
 ## Αναφορά API

apps/docs/i18n/es/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const recuperador = vector_index.asRetriever();
 recuperador.similarityTopK = 3;
 
 // ¡Obtener nodos!
-const nodosConPuntuación = await recuperador.retrieve("cadena de consulta");
+const nodosConPuntuación = await recuperador.retrieve({ query: "cadena de consulta" });
 ```
 
 ## Referencia de la API

apps/docs/i18n/et/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Too sõlmed!
-const nodesWithScore = await retriever.retrieve("päringu string");
+const nodesWithScore = await retriever.retrieve({ query: "päringu string" });
 ```
 
 ## API viide

apps/docs/i18n/fa/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // بازیابی گره ها!
-const nodesWithScore = await retriever.retrieve("رشته پرس و جو");
+const nodesWithScore = await retriever.retrieve({ query: "رشته پرس و جو" });
 ```
 
 ## مرجع API

apps/docs/i18n/fi/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Hae solmut!
-const nodesWithScore = await retriever.retrieve("kyselymerkkijono");
+const nodesWithScore = await retriever.retrieve({ query: "kyselymerkkijono" });
 ```
 
 ## API-viite

apps/docs/i18n/fr/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -11,7 +11,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Récupérer les nœuds !
-const nodesWithScore = await retriever.retrieve("chaîne de requête");
+const nodesWithScore = await retriever.retrieve({ query: "chaîne de requête" });
 ```
 
 ## Référence de l'API

apps/docs/i18n/he/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // אחזור צמתים!
-const nodesWithScore = await retriever.retrieve("מחרוזת שאילתה");
+const nodesWithScore = await retriever.retrieve({ query: "מחרוזת שאילתה" });
 ```
 
 ## מדריך לממשק API

apps/docs/i18n/hi/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // नोड्स प्राप्त करें!
-const nodesWithScore = await retriever.retrieve("क्वेरी स्ट्रिंग");
+const nodesWithScore = await retriever.retrieve({ query: "क्वेरी स्ट्रिंग" });
 ```
 
 ## एपीआई संदर्भ (API Reference)

apps/docs/i18n/hr/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const dohvatnik = vector_index.asRetriever();
 dohvatnik.similarityTopK = 3;
 
 // Dohvati čvorove!
-const čvoroviSaRezultatom = await dohvatnik.retrieve("upitni niz");
+const čvoroviSaRezultatom = await dohvatnik.retrieve({ query: "upitni niz" });
 ```
 
 ## API Referenca

apps/docs/i18n/hu/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Node-ok lekérése!
-const nodesWithScore = await retriever.retrieve("lekérdezési karakterlánc");
+const nodesWithScore = await retriever.retrieve({ query: "lekérdezési karakterlánc" });
 ```
 
 ## API Referencia

apps/docs/i18n/in/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Mengambil node!
-const nodesWithScore = await retriever.retrieve("string query");
+const nodesWithScore = await retriever.retrieve({ query: "string query" });
 ```
 
 ## Referensi API

apps/docs/i18n/it/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Recupera i nodi!
-const nodesWithScore = await retriever.retrieve("stringa di query");
+const nodesWithScore = await retriever.retrieve({ query: "stringa di query" });
 ```
 
 ## Riferimento API

apps/docs/i18n/ja/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // ノードを取得します！
-const nodesWithScore = await retriever.retrieve("クエリ文字列");
+const nodesWithScore = await retriever.retrieve({ query: "クエリ文字列" });
 ```
 
 ## API リファレンス

apps/docs/i18n/ko/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // 노드를 가져옵니다!
-const nodesWithScore = await retriever.retrieve("쿼리 문자열");
+const nodesWithScore = await retriever.retrieve({ query: "쿼리 문자열" });
 ```
 
 ## API 참조

apps/docs/i18n/lt/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const gavėjas = vector_index.asRetriever();
 gavėjas.similarityTopK = 3;
 
 // Išgaunami mazgai!
-const mazgaiSuRezultatu = await gavėjas.retrieve("užklausos eilutė");
+const mazgaiSuRezultatu = await gavėjas.retrieve({ query: "užklausos eilutė" });
 ```
 
 ## API nuorodos (API Reference)

apps/docs/i18n/nl/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Haal knooppunten op!
-const nodesWithScore = await retriever.retrieve("zoekopdracht");
+const nodesWithScore = await retriever.retrieve({ query: "zoekopdracht" });
 ```
 
 ## API Referentie

apps/docs/i18n/no/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Hent noder!
-const nodesWithScore = await retriever.retrieve("spørringsstreng");
+const nodesWithScore = await retriever.retrieve({ query: "spørringsstreng" });
 ```
 
 ## API-referanse

apps/docs/i18n/pl/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Pobierz węzły!
-const nodesWithScore = await retriever.retrieve("ciąg zapytania");
+const nodesWithScore = await retriever.retrieve({ query: "ciąg zapytania" });
 ```
 
 ## Dokumentacja interfejsu API

apps/docs/i18n/pt/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const recuperador = vector_index.asRetriever();
 recuperador.similarityTopK = 3;
 
 // Buscar nós!
-const nósComPontuação = await recuperador.retrieve("string de consulta");
+const nósComPontuação = await recuperador.retrieve({ query: "string de consulta" });
 ```
 
 ## Referência da API

apps/docs/i18n/ro/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const recuperator = vector_index.asRetriever();
 recuperator.similarityTopK = 3;
 
 // Preia nodurile!
-const noduriCuScor = await recuperator.retrieve("șir de interogare");
+const noduriCuScor = await recuperator.retrieve({ query: "șir de interogare" });
 ```
 
 ## Referință API

apps/docs/i18n/ru/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Получение узлов!
-const nodesWithScore = await retriever.retrieve("строка запроса");
+const nodesWithScore = await retriever.retrieve({ query: "строка запроса" });
 ```
 
 ## Справочник по API

apps/docs/i18n/se/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Dohvati čvorove!
-const nodesWithScore = await retriever.retrieve("upitni niz");
+const nodesWithScore = await retriever.retrieve({ query: "upitni niz" });
 ```
 
 ## API Referenca

apps/docs/i18n/sk/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const pridobitelj = vector_index.asRetriever();
 pridobitelj.similarityTopK = 3;
 
 // Pridobivanje vozlišč!
-const vozliščaZRezultatom = await pridobitelj.retrieve("poizvedbeni niz");
+const vozliščaZRezultatom = await pridobitelj.retrieve({ query: "poizvedbeni niz" });
 ```
 
 ## API Sklic

apps/docs/i18n/sl/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Získajte uzly!
-const nodesWithScore = await retriever.retrieve("reťazec dotazu");
+const nodesWithScore = await retriever.retrieve({ query: "reťazec dotazu" });
 ```
 
 ## API Referencia

apps/docs/i18n/sv/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Hämta noder!
-const nodesWithScore = await retriever.retrieve("frågesträng");
+const nodesWithScore = await retriever.retrieve({ query: "frågesträng" });
 ```
 
 ## API-referens

apps/docs/i18n/th/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // เรียกคืนโหนด!
-const nodesWithScore = await retriever.retrieve("query string");
+const nodesWithScore = await retriever.retrieve({ query: "query string" });
 ```
 
 ## API Reference (การอ้างอิง API)

apps/docs/i18n/tr/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Düğümleri getir!
-const nodesWithScore = await retriever.retrieve("sorgu dizesi");
+const nodesWithScore = await retriever.retrieve({ query: "sorgu dizesi" });
 ```
 
 ## API Referansı

apps/docs/i18n/uk/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Отримати вузли!
-const nodesWithScore = await retriever.retrieve("рядок запиту");
+const nodesWithScore = await retriever.retrieve({ query: "рядок запиту" });
 ```
 
 ## Довідник API

apps/docs/i18n/vi/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // Lấy các node!
-const nodesWithScore = await retriever.retrieve("chuỗi truy vấn");
+const nodesWithScore = await retriever.retrieve({ query: "chuỗi truy vấn" });
 ```
 
 ## Tài liệu tham khảo API

apps/docs/i18n/zh-Hans/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -11,7 +11,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // 获取节点！
-const nodesWithScore = await retriever.retrieve("查询字符串");
+const nodesWithScore = await retriever.retrieve({ query: "查询字符串" });
 ```
 
 ## API 参考

apps/docs/i18n/zh_tw/docusaurus-plugin-content-docs/current/modules/low_level/retriever.md

@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;
 
 // 提取節點！
-const nodesWithScore = await retriever.retrieve("查詢字串");
+const nodesWithScore = await retriever.retrieve({ query: "查詢字串" });
 ```
 
 ## API 參考

apps/docs/package.json

@@ -1,6 +1,6 @@
 {
   "name": "docs",
-  "version": "0.0.3",
+  "version": "0.0.4",
   "private": true,
   "scripts": {
     "docusaurus": "docusaurus",

examples/CHANGELOG.md

@@ -0,0 +1,14 @@
+# examples
+
+## 0.0.4
+
+### Patch Changes
+
+- d2e8d0c: add support for Milvus vector store
+- Updated dependencies [d2e8d0c]
+- Updated dependencies [aefc326]
+- Updated dependencies [484a710]
+- Updated dependencies [d766bd0]
+- Updated dependencies [dd95927]
+- Updated dependencies [bf583a7]
+  - llamaindex@0.2.0

examples/agent/query_openai_agent.ts

@@ -8,7 +8,7 @@ import {
 async function main() {
   // Load the documents
   const documents = await new SimpleDirectoryReader().loadData({
-    directoryPath: "node_modules/llamaindex/examples/",
+    directoryPath: "node_modules/llamaindex/examples",
   });
 
   // Create a vector index from the documents

examples/agent/step_wise_openai.ts

@@ -0,0 +1,95 @@
+import { FunctionTool, OpenAIAgent } from "llamaindex";
+
+// Define a function to sum two numbers
+function sumNumbers({ a, b }: { a: number; b: number }): number {
+  return a + b;
+}
+
+// Define a function to divide two numbers
+function divideNumbers({ a, b }: { a: number; b: number }): number {
+  return a / b;
+}
+
+// Define the parameters of the sum function as a JSON schema
+const sumJSON = {
+  type: "object",
+  properties: {
+    a: {
+      type: "number",
+      description: "The first number",
+    },
+    b: {
+      type: "number",
+      description: "The second number",
+    },
+  },
+  required: ["a", "b"],
+};
+
+const divideJSON = {
+  type: "object",
+  properties: {
+    a: {
+      type: "number",
+      description: "The dividend a to divide",
+    },
+    b: {
+      type: "number",
+      description: "The divisor b to divide by",
+    },
+  },
+  required: ["a", "b"],
+};
+
+async function main() {
+  // Create a function tool from the sum function
+  const functionTool = new FunctionTool(sumNumbers, {
+    name: "sumNumbers",
+    description: "Use this function to sum two numbers",
+    parameters: sumJSON,
+  });
+
+  // Create a function tool from the divide function
+  const functionTool2 = new FunctionTool(divideNumbers, {
+    name: "divideNumbers",
+    description: "Use this function to divide two numbers",
+    parameters: divideJSON,
+  });
+
+  // Create an OpenAIAgent with the function tools
+  const agent = new OpenAIAgent({
+    tools: [functionTool, functionTool2],
+    verbose: true,
+  });
+
+  // Create a task to sum and divide numbers
+  const task = agent.createTask("How much is 5 + 5? then divide by 2");
+
+  let count = 0;
+
+  while (true) {
+    const stepOutput = await agent.runStep(task.taskId);
+
+    console.log(`Runnning step ${count++}`);
+    console.log(`======== OUTPUT ==========`);
+    if (stepOutput.output.response) {
+      console.log(stepOutput.output.response);
+    } else {
+      console.log(stepOutput.output.sources);
+    }
+    console.log(`==========================`);
+
+    if (stepOutput.isLast) {
+      const finalResponse = await agent.finalizeResponse(
+        task.taskId,
+        stepOutput,
+      );
+      console.log({ finalResponse });
+      break;
+    }
+  }
+}
+
+main().then(() => {
+  console.log("Done");
+});

examples/agent/step_wise_query_tool.ts

@@ -8,7 +8,7 @@ import {
 async function main() {
   // Load the documents
   const documents = await new SimpleDirectoryReader().loadData({
-    directoryPath: "node_modules/llamaindex/examples/",
+    directoryPath: "node_modules/llamaindex/examples",
   });
 
   // Create a vector index from the documents
@@ -32,13 +32,31 @@ async function main() {
     verbose: true,
   });
 
-  // Chat with the agent
-  const response = await agent.chat({
-    message: "What was his salary?",
-  });
+  const task = agent.createTask("What was his salary?");
 
-  // Print the response
-  console.log(String(response));
+  let count = 0;
+
+  while (true) {
+    const stepOutput = await agent.runStep(task.taskId);
+
+    console.log(`Runnning step ${count++}`);
+    console.log(`======== OUTPUT ==========`);
+    if (stepOutput.output.response) {
+      console.log(stepOutput.output.response);
+    } else {
+      console.log(stepOutput.output.sources);
+    }
+    console.log(`==========================`);
+
+    if (stepOutput.isLast) {
+      const finalResponse = await agent.finalizeResponse(
+        task.taskId,
+        stepOutput,
+      );
+      console.log({ finalResponse });
+      break;
+    }
+  }
 }
 
 main().then(() => {

examples/agent/step_wise_react.ts

@@ -0,0 +1,90 @@
+import { FunctionTool, ReActAgent } from "llamaindex";
+
+// Define a function to sum two numbers
+function sumNumbers({ a, b }: { a: number; b: number }): number {
+  return a + b;
+}
+
+// Define a function to divide two numbers
+function divideNumbers({ a, b }: { a: number; b: number }): number {
+  return a / b;
+}
+
+// Define the parameters of the sum function as a JSON schema
+const sumJSON = {
+  type: "object",
+  properties: {
+    a: {
+      type: "number",
+      description: "The first number",
+    },
+    b: {
+      type: "number",
+      description: "The second number",
+    },
+  },
+  required: ["a", "b"],
+};
+
+const divideJSON = {
+  type: "object",
+  properties: {
+    a: {
+      type: "number",
+      description: "The dividend",
+    },
+    b: {
+      type: "number",
+      description: "The divisor",
+    },
+  },
+  required: ["a", "b"],
+};
+
+async function main() {
+  // Create a function tool from the sum function
+  const functionTool = new FunctionTool(sumNumbers, {
+    name: "sumNumbers",
+    description: "Use this function to sum two numbers",
+    parameters: sumJSON,
+  });
+
+  // Create a function tool from the divide function
+  const functionTool2 = new FunctionTool(divideNumbers, {
+    name: "divideNumbers",
+    description: "Use this function to divide two numbers",
+    parameters: divideJSON,
+  });
+
+  // Create an OpenAIAgent with the function tools
+  const agent = new ReActAgent({
+    tools: [functionTool, functionTool2],
+    verbose: true,
+  });
+
+  const task = agent.createTask("Divide 16 by 2 then add 20");
+
+  let count = 0;
+
+  while (true) {
+    const stepOutput = await agent.runStep(task.taskId);
+
+    console.log(`Runnning step ${count++}`);
+    console.log(`======== OUTPUT ==========`);
+    console.log(stepOutput.output);
+    console.log(`==========================`);
+
+    if (stepOutput.isLast) {
+      const finalResponse = await agent.finalizeResponse(
+        task.taskId,
+        stepOutput,
+      );
+      console.log({ finalResponse });
+      break;
+    }
+  }
+}
+
+main().then(() => {
+  console.log("Done");
+});

examples/agent/stream_openai_agent.ts

@@ -0,0 +1,77 @@
+import { FunctionTool, OpenAIAgent } from "llamaindex";
+
+// Define a function to sum two numbers
+function sumNumbers({ a, b }: { a: number; b: number }): number {
+  return a + b;
+}
+
+// Define a function to divide two numbers
+function divideNumbers({ a, b }: { a: number; b: number }): number {
+  return a / b;
+}
+
+// Define the parameters of the sum function as a JSON schema
+const sumJSON = {
+  type: "object",
+  properties: {
+    a: {
+      type: "number",
+      description: "The first number",
+    },
+    b: {
+      type: "number",
+      description: "The second number",
+    },
+  },
+  required: ["a", "b"],
+};
+
+const divideJSON = {
+  type: "object",
+  properties: {
+    a: {
+      type: "number",
+      description: "The dividend",
+    },
+    b: {
+      type: "number",
+      description: "The divisor",
+    },
+  },
+  required: ["a", "b"],
+};
+
+async function main() {
+  // Create a function tool from the sum function
+  const functionTool = new FunctionTool(sumNumbers, {
+    name: "sumNumbers",
+    description: "Use this function to sum two numbers",
+    parameters: sumJSON,
+  });
+
+  // Create a function tool from the divide function
+  const functionTool2 = new FunctionTool(divideNumbers, {
+    name: "divideNumbers",
+    description: "Use this function to divide two numbers",
+    parameters: divideJSON,
+  });
+
+  // Create an OpenAIAgent with the function tools
+  const agent = new OpenAIAgent({
+    tools: [functionTool, functionTool2],
+    verbose: false,
+  });
+
+  const stream = await agent.chat({
+    message: "Divide 16 by 2 then add 20",
+    stream: true,
+  });
+
+  for await (const chunk of stream.response) {
+    process.stdout.write(chunk.response);
+  }
+}
+
+main().then(() => {
+  console.log("\nDone");
+});

examples/anthropic/chat.ts

@@ -3,6 +3,7 @@ import { Anthropic } from "llamaindex";
 (async () => {
   const anthropic = new Anthropic({
     apiKey: process.env.ANTHROPIC_API_KEY,
+    model: "claude-3-opus",
   });
   const result = await anthropic.chat({
     messages: [

examples/anthropic/chat_interactive.ts

@@ -0,0 +1,34 @@
+import { Anthropic, SimpleChatEngine, SimpleChatHistory } from "llamaindex";
+import { stdin as input, stdout as output } from "node:process";
+import readline from "node:readline/promises";
+
+(async () => {
+  const llm = new Anthropic({
+    apiKey: process.env.ANTHROPIC_API_KEY,
+    model: "claude-3-opus",
+  });
+  // chatHistory will store all the messages in the conversation
+  const chatHistory = new SimpleChatHistory({
+    messages: [
+      {
+        content: "You want to talk in rhymes.",
+        role: "system",
+      },
+    ],
+  });
+  const chatEngine = new SimpleChatEngine({
+    llm,
+    chatHistory,
+  });
+  const rl = readline.createInterface({ input, output });
+
+  while (true) {
+    const query = await rl.question("User: ");
+    process.stdout.write("Assistant: ");
+    const stream = await chatEngine.chat({ message: query, stream: true });
+    for await (const chunk of stream) {
+      process.stdout.write(chunk.response);
+    }
+    process.stdout.write("\n");
+  }
+})();

examples/anthropic/haiku.ts

@@ -0,0 +1,19 @@
+import { Anthropic } from "llamaindex";
+
+(async () => {
+  const anthropic = new Anthropic({
+    apiKey: process.env.ANTHROPIC_API_KEY,
+    model: "claude-3-haiku",
+  });
+  const result = await anthropic.chat({
+    messages: [
+      { content: "You want to talk in rhymes.", role: "system" },
+      {
+        content:
+          "How much wood would a woodchuck chuck if a woodchuck could chuck wood?",
+        role: "user",
+      },
+    ],
+  });
+  console.log(result);
+})();

examples/anthropic/stream.ts

@@ -0,0 +1,23 @@
+import { Anthropic } from "llamaindex";
+
+(async () => {
+  const anthropic = new Anthropic({
+    apiKey: process.env.ANTHROPIC_API_KEY,
+    model: "claude-instant-1.2",
+  });
+  const stream = await anthropic.chat({
+    messages: [
+      { content: "You want to talk in rhymes.", role: "system" },
+      {
+        content:
+          "How much wood would a woodchuck chuck if a woodchuck could chuck wood?",
+        role: "user",
+      },
+    ],
+    stream: true,
+  });
+
+  for await (const chunk of stream) {
+    process.stdout.write(chunk.delta);
+  }
+})();

examples/astradb/README.md

@@ -32,10 +32,10 @@ run `ts-node astradb/example`
 
 This sample loads the same dataset of movie reviews as the Astra Portal sample dataset. (Feel free to load the data in your the Astra Data Explorer to compare)
 
-run `ts-node astradb/load`
+run `npx ts-node astradb/load`
 
 ### Use RAG to Query the data
 
 Check out your data in the Astra Data Explorer and change the sample query as you see fit.
 
-run `ts-node astradb/query`
+run `npx ts-node astradb/query`

examples/milvus/README.md

@@ -0,0 +1,34 @@
+# Milvus Vector Store
+
+Here are two sample scripts which work with loading and querying data from a Milvus Vector Store.
+
+## Prerequisites
+
+- An Milvus Vector Database
+  - Hosted https://milvus.io/
+  - Self Hosted https://milvus.io/docs/install_standalone-docker.md
+- An OpenAI API Key
+
+## Setup
+
+1. Set your env variables:
+
+- `MILVUS_ADDRESS`: Address of your Milvus Vector Store (like localhost:19530)
+- `MILVUS_USERNAME`: empty or username for your Milvus Vector Store
+- `MILVUS_PASSWORD`: empty or password for your Milvus Vector Store
+- `OPENAI_API_KEY`: Your OpenAI key
+
+2. `cd` Into the `examples` directory
+3. run `npm i`
+
+## Load the data
+
+This sample loads the same dataset of movie reviews as sample dataset. You can install https://github.com/zilliztech/attu to inspect the loaded data.
+
+run `npx ts-node milvus/load`
+
+## Use RAG to Query the data
+
+Check out your data in Attu and change the sample query as you see fit.
+
+run `npx ts-node milvus/query`

examples/milvus/load.ts

@@ -0,0 +1,26 @@
+import {
+  MilvusVectorStore,
+  PapaCSVReader,
+  storageContextFromDefaults,
+  VectorStoreIndex,
+} from "llamaindex";
+
+const collectionName = "movie_reviews";
+
+async function main() {
+  try {
+    const reader = new PapaCSVReader(false);
+    const docs = await reader.loadData("./data/movie_reviews.csv");
+
+    const vectorStore = new MilvusVectorStore({ collection: collectionName });
+
+    const ctx = await storageContextFromDefaults({ vectorStore });
+    const index = await VectorStoreIndex.fromDocuments(docs, {
+      storageContext: ctx,
+    });
+  } catch (e) {
+    console.error(e);
+  }
+}
+
+main();

examples/milvus/query.ts

@@ -0,0 +1,30 @@
+import {
+  MilvusVectorStore,
+  serviceContextFromDefaults,
+  VectorStoreIndex,
+} from "llamaindex";
+
+const collectionName = "movie_reviews";
+
+async function main() {
+  try {
+    const milvus = new MilvusVectorStore({ collection: collectionName });
+
+    const ctx = serviceContextFromDefaults();
+    const index = await VectorStoreIndex.fromVectorStore(milvus, ctx);
+
+    const retriever = await index.asRetriever({ similarityTopK: 20 });
+
+    const queryEngine = await index.asQueryEngine({ retriever });
+
+    const results = await queryEngine.query({
+      query: "What is the best reviewed movie?",
+    });
+
+    console.log(results.response);
+  } catch (e) {
+    console.error(e);
+  }
+}
+
+main();

examples/multimodal/retrieve.ts

@@ -27,9 +27,9 @@ async function main() {
   // retrieve documents using the index
   const index = await createIndex();
   const retriever = index.asRetriever({ similarityTopK: 3 });
-  const results = await retriever.retrieve(
-    "what are Vincent van Gogh's famous paintings",
-  );
+  const results = await retriever.retrieve({
+    query: "what are Vincent van Gogh's famous paintings",
+  });
   for (const result of results) {
     const node = result.node;
     if (!node) {

examples/openai.ts

@@ -1,4 +1,4 @@
-import { OpenAI } from "llamaindex";
+import { OpenAI, OpenAIEmbedding } from "llamaindex";
 
 (async () => {
   const llm = new OpenAI({ model: "gpt-4-1106-preview", temperature: 0.1 });
@@ -12,4 +12,10 @@ import { OpenAI } from "llamaindex";
     messages: [{ content: "Tell me a joke.", role: "user" }],
   });
   console.log(response2.message.content);
+
+  // embeddings
+  const embedModel = new OpenAIEmbedding();
+  const texts = ["hello", "world"];
+  const embeddings = await embedModel.getTextEmbeddingsBatch(texts);
+  console.log(`\nWe have ${embeddings.length} embeddings`);
 })();

examples/package.json

@@ -1,16 +1,19 @@
 {
   "name": "examples",
   "private": true,
-  "version": "0.0.3",
+  "version": "0.0.4",
   "dependencies": {
+    "@aws-crypto/sha256-js": "^5.2.0",
     "@datastax/astra-db-ts": "^0.1.4",
     "@notionhq/client": "^2.2.14",
     "@pinecone-database/pinecone": "^1.1.3",
+    "@zilliz/milvus2-sdk-node": "^2.3.5",
     "chromadb": "^1.8.1",
     "commander": "^11.1.0",
     "dotenv": "^16.4.1",
     "llamaindex": "latest",
-    "mongodb": "^6.2.0"
+    "mongodb": "^6.2.0",
+    "pathe": "^1.1.2"
   },
   "devDependencies": {
     "@types/node": "^18.19.10",

examples/pinecone-vector-store/README.md

@@ -7,8 +7,9 @@ There are two scripts available here: load-docs.ts and query.ts
 You'll need a Pinecone account, project, and index. Pinecone does not allow automatic creation of indexes on the free plan,
 so this vector store does not check and create the index (unlike, e.g., the PGVectorStore)
 
-Set the **PINECONE_API_KEY** and **PINECONE_ENVIRONMENT** environment variables to match your specific values. You will likely also need to set **PINECONE_INDEX_NAME**, unless your
-index is the default value "llama".
+Set the **PINECONE_API_KEY** and **PINECONE_ENVIRONMENT** environment variables to match your specific values.
+You will likely also need to set **PINECONE_INDEX_NAME**, unless your index is the default value "llama".
+By default, all operations take place inside the default namespace '', but you can set **PINECONE_NAMESPACE** to a different value if you need to.
 
 You'll also need a value for OPENAI_API_KEY in your environment.
 

examples/qdrantdb/README.md

@@ -0,0 +1,11 @@
+# Qdrant Vector Store Example
+
+How to run `examples/qdrantdb/preFilters.ts`:
+
+Add your OpenAI API Key into a file called `.env` in the parent folder of this directory. It should look like this:
+
+```
+OPEN_API_KEY=sk-you-key
+```
+
+Now, open a new terminal window and inside `examples`, run `npx ts-node qdrantdb/preFilters.ts`.

examples/qdrantdb/preFilters.ts

@@ -0,0 +1,82 @@
+import * as dotenv from "dotenv";
+import {
+  CallbackManager,
+  Document,
+  MetadataMode,
+  QdrantVectorStore,
+  VectorStoreIndex,
+  serviceContextFromDefaults,
+  storageContextFromDefaults,
+} from "llamaindex";
+
+// Load environment variables from local .env file
+dotenv.config();
+
+const collectionName = "dog_colors";
+const qdrantUrl = "http://127.0.0.1:6333";
+
+async function main() {
+  try {
+    const docs = [
+      new Document({
+        text: "The dog is brown",
+        metadata: {
+          dogId: "1",
+        },
+      }),
+      new Document({
+        text: "The dog is red",
+        metadata: {
+          dogId: "2",
+        },
+      }),
+    ];
+    console.log("Creating QdrantDB vector store");
+    const qdrantVs = new QdrantVectorStore({ url: qdrantUrl, collectionName });
+    const ctx = await storageContextFromDefaults({ vectorStore: qdrantVs });
+
+    console.log("Embedding documents and adding to index");
+    const index = await VectorStoreIndex.fromDocuments(docs, {
+      storageContext: ctx,
+      serviceContext: serviceContextFromDefaults({
+        callbackManager: new CallbackManager({
+          onRetrieve: (data) => {
+            console.log(
+              "The retrieved nodes are:",
+              data.nodes.map((node) => node.node.getContent(MetadataMode.NONE)),
+            );
+          },
+        }),
+      }),
+    });
+
+    console.log(
+      "Querying index with no filters: Expected output: Brown probably",
+    );
+    const queryEngineNoFilters = index.asQueryEngine();
+    const noFilterResponse = await queryEngineNoFilters.query({
+      query: "What is the color of the dog?",
+    });
+    console.log("No filter response:", noFilterResponse.toString());
+    console.log("Querying index with dogId 2: Expected output: Red");
+    const queryEngineDogId2 = index.asQueryEngine({
+      preFilters: {
+        filters: [
+          {
+            key: "dogId",
+            value: "2",
+            filterType: "ExactMatch",
+          },
+        ],
+      },
+    });
+    const response = await queryEngineDogId2.query({
+      query: "What is the color of the dog?",
+    });
+    console.log("Filter with dogId 2 response:", response.toString());
+  } catch (e) {
+    console.error(e);
+  }
+}
+
+main();

examples/readers/CHANGELOG.md

@@ -0,0 +1,11 @@
+# llamaindex-loader-example
+
+### Patch Changes
+
+- Updated dependencies [d2e8d0c]
+- Updated dependencies [aefc326]
+- Updated dependencies [484a710]
+- Updated dependencies [d766bd0]
+- Updated dependencies [dd95927]
+- Updated dependencies [bf583a7]
+  - llamaindex@0.2.0

examples/readers/src/custom-simple-directory-reader.ts

@@ -2,8 +2,8 @@ import type { BaseReader, Document, Metadata } from "llamaindex";
 import {
   FILE_EXT_TO_READER,
   SimpleDirectoryReader,
-  TextFileReader,
 } from "llamaindex/readers/SimpleDirectoryReader";
+import { TextFileReader } from "llamaindex/readers/TextFileReader";
 
 class ZipReader implements BaseReader {
   loadData(...args: any[]): Promise<Document<Metadata>[]> {

package.json

@@ -11,10 +11,12 @@
     "prepare": "husky",
     "test": "turbo run test",
     "type-check": "tsc -b --diagnostics",
-    "release": "pnpm run build:release && changeset publish",
-    "new-llamaindex": "pnpm run build:release && changeset version --ignore create-llama",
-    "new-create-llama": "pnpm run build:release && changeset version --ignore llamaindex",
-    "new-snapshots": "pnpm run build:release && changeset version --snapshot"
+    "release": "pnpm run check-minor-version && pnpm run build:release && changeset publish",
+    "release-snapshot": "pnpm run check-minor-version && pnpm run build:release && changeset publish --tag snapshot",
+    "check-minor-version": "node ./scripts/check-minor-version",
+    "update-version": "node ./scripts/update-version",
+    "new-version": "pnpm run build:release && changeset version && pnpm run check-minor-version && pnpm run update-version",
+    "new-snapshot": "pnpm run build:release && changeset version --snapshot && pnpm run update-version"
   },
   "devDependencies": {
     "@changesets/cli": "^2.27.1",

packages/core/.gitignore

@@ -1 +1,3 @@
 .turbo
+README.md
+LICENSE

packages/core/CHANGELOG.md

@@ -1,5 +1,49 @@
 # llamaindex
 
+## 0.2.0
+
+### Minor Changes
+
+- bf583a7: Use parameter object for retrieve function of Retriever (to align usage with query function of QueryEngine)
+
+### Patch Changes
+
+- d2e8d0c: add support for Milvus vector store
+- aefc326: feat: experimental package + json query engine
+- 484a710: - Add missing exports:
+  - `IndexStructType`,
+  - `IndexDict`,
+  - `jsonToIndexStruct`,
+  - `IndexList`,
+  - `IndexStruct`
+  - Fix `IndexDict.toJson()` method
+- d766bd0: Add streaming to agents
+- dd95927: add Claude Haiku support and update anthropic SDK
+
+## 0.1.21
+
+### Patch Changes
+
+- 552a61a: Add quantized parameter to HuggingFaceEmbedding
+- d824876: Add support for Claude 3
+
+## 0.1.20
+
+### Patch Changes
+
+- 64683a5: fix: prefix messages always true
+- 698cd9c: fix: step wise agent + examples
+- 7257751: fixed removeRefDocNode and persist store on delete
+- 5116ad8: fix: compatibility issue with Deno
+- Updated dependencies [5116ad8]
+  - @llamaindex/env@0.0.5
+
+## 0.1.19
+
+### Patch Changes
+
+- 026d068: feat: enhance pinecone usage
+
 ## 0.1.18
 
 ### Patch Changes

packages/core/jsr.json

@@ -0,0 +1,8 @@
+{
+  "name": "@llamaindex/core",
+  "version": "0.1.21",
+  "exports": "./src/index.ts",
+  "imports": {
+    "@llamaindex/env": "jsr:@llamaindex/env@0.0.5"
+  }
+}

packages/core/package.json

@@ -1,29 +1,32 @@
 {
   "name": "llamaindex",
-  "version": "0.1.18",
+  "version": "0.2.0",
+  "expectedMinorVersion": "2",
   "license": "MIT",
   "type": "module",
   "dependencies": {
-    "@anthropic-ai/sdk": "^0.13.0",
+    "@anthropic-ai/sdk": "^0.18.0",
     "@aws-crypto/sha256-js": "^5.2.0",
     "@datastax/astra-db-ts": "^0.1.4",
-    "@types/lodash": "^4.14.202",
-    "@types/node": "^18.19.14",
-    "@types/papaparse": "^5.3.14",
-    "@types/pg": "^8.11.0",
-    "@llamaindex/cloud": "^0.0.1",
+    "@grpc/grpc-js": "^1.10.2",
+    "@llamaindex/cloud": "0.0.4",
     "@llamaindex/env": "workspace:*",
     "@mistralai/mistralai": "^0.0.10",
     "@notionhq/client": "^2.2.14",
     "@pinecone-database/pinecone": "^2.0.1",
     "@qdrant/js-client-rest": "^1.7.0",
+    "@types/lodash": "^4.14.202",
+    "@types/node": "^18.19.14",
+    "@types/papaparse": "^5.3.14",
+    "@types/pg": "^8.11.0",
     "@xenova/transformers": "^2.15.0",
+    "@zilliz/milvus2-sdk-node": "^2.3.5",
     "assemblyai": "^4.2.2",
     "chromadb": "~1.7.3",
     "cohere-ai": "^7.7.5",
-    "file-type": "^18.7.0",
     "js-tiktoken": "^1.0.10",
     "lodash": "^4.17.21",
+    "magic-bytes.js": "^1.10.0",
     "mammoth": "^1.6.0",
     "md-utils-ts": "^2.0.0",
     "mongodb": "^6.3.0",
@@ -59,10 +62,6 @@
         "types": "./dist/type/index.d.ts",
         "default": "./dist/index.js"
       },
-      "edge-light": {
-        "types": "./dist/type/index.d.ts",
-        "default": "./dist/index.edge-light.js"
-      },
       "require": {
         "types": "./dist/type/index.d.ts",
         "default": "./dist/cjs/index.js"
@@ -92,10 +91,11 @@
   "scripts": {
     "lint": "eslint .",
     "build": "rm -rf ./dist && pnpm run build:esm && pnpm run build:cjs && pnpm run build:type",
-    "build:esm": "swc src -d dist --strip-leading-paths --config-file .swcrc",
-    "build:cjs": "swc src -d dist/cjs --strip-leading-paths --config-file .cjs.swcrc",
-    "build:type": "pnpm run -w type-check",
-    "postbuild": "node -e \"require('fs').writeFileSync('./dist/cjs/package.json', JSON.stringify({ type: 'commonjs' }))\"",
+    "build:esm": "swc src -d dist --strip-leading-paths --config-file ../../.swcrc",
+    "build:cjs": "swc src -d dist/cjs --strip-leading-paths --config-file ../../.cjs.swcrc",
+    "build:type": "tsc -p tsconfig.json",
+    "copy": "cp -r ../../README.md ../../LICENSE .",
+    "postbuild": "pnpm run copy && node -e \"require('fs').writeFileSync('./dist/cjs/package.json', JSON.stringify({ type: 'commonjs' }))\"",
     "circular-check": "madge -c ./src/index.ts",
     "dev": "concurrently \"pnpm run build:esm --watch\" \"pnpm run build:cjs --watch\" \"pnpm run build:type --watch\""
   }

packages/core/src/Retriever.ts

@@ -2,14 +2,16 @@ import type { Event } from "./callbacks/CallbackManager.js";
 import type { NodeWithScore } from "./Node.js";
 import type { ServiceContext } from "./ServiceContext.js";
 
+export type RetrieveParams = {
+  query: string;
+  parentEvent?: Event;
+  preFilters?: unknown;
+};
+
 /**
  * Retrievers retrieve the nodes that most closely match our query in similarity.
  */
 export interface BaseRetriever {
-  retrieve(
-    query: string,
-    parentEvent?: Event,
-    preFilters?: unknown,
-  ): Promise<NodeWithScore[]>;
+  retrieve(params: RetrieveParams): Promise<NodeWithScore[]>;
   getServiceContext(): ServiceContext;
 }

packages/core/src/agent/openai/base.ts

@@ -37,8 +37,6 @@ export class OpenAIAgent extends AgentRunner {
     toolRetriever,
     systemPrompt,
   }: OpenAIAgentParams) {
-    prefixMessages = prefixMessages || [];
-
     llm = llm ?? new OpenAI({ model: "gpt-3.5-turbo-0613" });
 
     if (systemPrompt) {

packages/core/src/agent/openai/worker.ts

@@ -1,10 +1,10 @@
-// Assuming that the necessary interfaces and classes (like BaseTool, OpenAI, ChatMessage, CallbackManager, etc.) are defined elsewhere
-
 import { randomUUID } from "@llamaindex/env";
+import { Response } from "../../Response.js";
 import type { CallbackManager } from "../../callbacks/CallbackManager.js";
 import {
   AgentChatResponse,
   ChatResponseMode,
+  StreamingAgentChatResponse,
 } from "../../engines/chat/types.js";
 import type {
   ChatMessage,
@@ -12,6 +12,7 @@ import type {
   ChatResponseChunk,
 } from "../../llm/index.js";
 import { OpenAI } from "../../llm/index.js";
+import { streamConverter, streamReducer } from "../../llm/utils.js";
 import { ChatMemoryBuffer } from "../../memory/ChatMemoryBuffer.js";
 import type { ObjectRetriever } from "../../objects/base.js";
 import type { ToolOutput } from "../../tools/types.js";
@@ -192,13 +193,40 @@ export class OpenAIAgentWorker implements AgentWorker {
   private _processMessage(
     task: Task,
     chatResponse: ChatResponse,
-  ): AgentChatResponse | AsyncIterable<ChatResponseChunk> {
+  ): AgentChatResponse {
     const aiMessage = chatResponse.message;
     task.extraState.newMemory.put(aiMessage);
 
     return new AgentChatResponse(aiMessage.content, task.extraState.sources);
   }
 
+  private async _getStreamAiResponse(
+    task: Task,
+    llmChatKwargs: any,
+  ): Promise<StreamingAgentChatResponse> {
+    const stream = await this.llm.chat({
+      stream: true,
+      ...llmChatKwargs,
+    });
+
+    const iterator = streamConverter(
+      streamReducer({
+        stream,
+        initialValue: "",
+        reducer: (accumulator, part) => (accumulator += part.delta),
+        finished: (accumulator) => {
+          task.extraState.newMemory.put({
+            content: accumulator,
+            role: "assistant",
+          });
+        },
+      }),
+      (r: ChatResponseChunk) => new Response(r.delta),
+    );
+
+    return new StreamingAgentChatResponse(iterator, task.extraState.sources);
+  }
+
   /**
    * Get agent response.
    * @param task: task
@@ -210,7 +238,7 @@ export class OpenAIAgentWorker implements AgentWorker {
     task: Task,
     mode: ChatResponseMode,
     llmChatKwargs: any,
-  ): Promise<AgentChatResponse> {
+  ): Promise<AgentChatResponse | StreamingAgentChatResponse> {
     if (mode === ChatResponseMode.WAIT) {
       const chatResponse = (await this.llm.chat({
         stream: false,
@@ -218,9 +246,11 @@ export class OpenAIAgentWorker implements AgentWorker {
       })) as unknown as ChatResponse;
 
       return this._processMessage(task, chatResponse) as AgentChatResponse;
-    } else {
-      throw new Error("Not implemented");
+    } else if (mode === ChatResponseMode.STREAM) {
+      return this._getStreamAiResponse(task, llmChatKwargs);
     }
+
+    throw new Error("Invalid mode");
   }
 
   /**

packages/core/src/agent/react/worker.ts

@@ -1,4 +1,4 @@
-import { randomUUID } from "crypto";
+import { randomUUID } from "@llamaindex/env";
 import { CallbackManager } from "../../callbacks/CallbackManager.js";
 import { AgentChatResponse } from "../../engines/chat/index.js";
 import type { ChatResponse, LLM } from "../../llm/index.js";
@@ -17,7 +17,6 @@ import {
   ObservationReasoningStep,
   ResponseReasoningStep,
 } from "./types.js";
-
 type ReActAgentWorkerParams = {
   tools: BaseTool[];
   llm?: LLM;

packages/core/src/agent/runner/base.ts

@@ -1,9 +1,10 @@
-import { randomUUID } from "crypto";
+import { randomUUID } from "@llamaindex/env";
 import { CallbackManager } from "../../callbacks/CallbackManager.js";
 import type { ChatEngineAgentParams } from "../../engines/chat/index.js";
 import {
   AgentChatResponse,
   ChatResponseMode,
+  StreamingAgentChatResponse,
 } from "../../engines/chat/index.js";
 import type { ChatMessage, LLM } from "../../llm/index.js";
 import { ChatMemoryBuffer } from "../../memory/ChatMemoryBuffer.js";
@@ -11,10 +12,9 @@ import type { BaseMemory } from "../../memory/types.js";
 import type { AgentWorker, TaskStepOutput } from "../types.js";
 import { Task, TaskStep } from "../types.js";
 import { AgentState, BaseAgentRunner, TaskState } from "./types.js";
-
 const validateStepFromArgs = (
   taskId: string,
-  input: string,
+  input?: string | null,
   step?: any,
   kwargs?: any,
 ): TaskStep | undefined => {
@@ -24,6 +24,7 @@ const validateStepFromArgs = (
     }
     return step;
   } else {
+    if (!input) return;
     return new TaskStep(taskId, step, input, kwargs);
   }
 };
@@ -194,7 +195,7 @@ export class AgentRunner extends BaseAgentRunner {
    */
   async runStep(
     taskId: string,
-    input: string,
+    input?: string | null,
     step?: TaskStep,
     kwargs: any = {},
   ): Promise<TaskStepOutput> {
@@ -230,23 +231,26 @@ export class AgentRunner extends BaseAgentRunner {
     taskId: string,
     stepOutput: TaskStepOutput,
     kwargs?: any,
-  ): Promise<AgentChatResponse> {
+  ): Promise<AgentChatResponse | StreamingAgentChatResponse> {
     if (!stepOutput) {
       stepOutput =
         this.getCompletedSteps(taskId)[
           this.getCompletedSteps(taskId).length - 1
         ];
     }
+
     if (!stepOutput.isLast) {
       throw new Error(
         "finalizeResponse can only be called on the last step output",
       );
     }
 
-    if (!(stepOutput.output instanceof AgentChatResponse)) {
-      throw new Error(
-        `When \`isLast\` is True, cur_step_output.output must be AGENT_CHAT_RESPONSE_TYPE: ${stepOutput.output}`,
-      );
+    if (!(stepOutput.output instanceof StreamingAgentChatResponse)) {
+      if (!(stepOutput.output instanceof AgentChatResponse)) {
+        throw new Error(
+          `When \`isLast\` is True, cur_step_output.output must be AGENT_CHAT_RESPONSE_TYPE: ${stepOutput.output}`,
+        );
+      }
     }
 
     this.agentWorker.finalizeTask(this.getTask(taskId), kwargs);
@@ -261,20 +265,32 @@ export class AgentRunner extends BaseAgentRunner {
   protected async _chat({
     message,
     toolChoice,
-  }: ChatEngineAgentParams & { mode: ChatResponseMode }) {
+    stream,
+  }: ChatEngineAgentParams): Promise<AgentChatResponse>;
+  protected async _chat({
+    message,
+    toolChoice,
+    stream,
+  }: ChatEngineAgentParams & {
+    stream: true;
+  }): Promise<StreamingAgentChatResponse>;
+  protected async _chat({
+    message,
+    toolChoice,
+    stream,
+  }: ChatEngineAgentParams): Promise<
+    AgentChatResponse | StreamingAgentChatResponse
+  > {
     const task = this.createTask(message as string);
 
     let resultOutput;
 
+    const mode = stream ? ChatResponseMode.STREAM : ChatResponseMode.WAIT;
+
     while (true) {
-      const curStepOutput = await this._runStep(
-        task.taskId,
-        undefined,
-        ChatResponseMode.WAIT,
-        {
-          toolChoice,
-        },
-      );
+      const curStepOutput = await this._runStep(task.taskId, undefined, mode, {
+        toolChoice,
+      });
 
       if (curStepOutput.isLast) {
         resultOutput = curStepOutput;
@@ -298,7 +314,26 @@ export class AgentRunner extends BaseAgentRunner {
     message,
     chatHistory,
     toolChoice,
-  }: ChatEngineAgentParams): Promise<AgentChatResponse> {
+    stream,
+  }: ChatEngineAgentParams & {
+    stream?: false;
+  }): Promise<AgentChatResponse>;
+  public async chat({
+    message,
+    chatHistory,
+    toolChoice,
+    stream,
+  }: ChatEngineAgentParams & {
+    stream: true;
+  }): Promise<StreamingAgentChatResponse>;
+  public async chat({
+    message,
+    chatHistory,
+    toolChoice,
+    stream,
+  }: ChatEngineAgentParams): Promise<
+    AgentChatResponse | StreamingAgentChatResponse
+  > {
     if (!toolChoice) {
       toolChoice = this.defaultToolChoice;
     }
@@ -307,7 +342,7 @@ export class AgentRunner extends BaseAgentRunner {
       message,
       chatHistory,
       toolChoice,
-      mode: ChatResponseMode.WAIT,
+      stream,
     });
 
     return chatResponse;

packages/core/src/agent/runner/types.ts

@@ -1,4 +1,7 @@
-import type { AgentChatResponse } from "../../engines/chat/index.js";
+import type {
+  AgentChatResponse,
+  StreamingAgentChatResponse,
+} from "../../engines/chat/index.js";
 import type { Task, TaskStep, TaskStepOutput } from "../types.js";
 import { BaseAgent } from "../types.js";
 
@@ -57,7 +60,7 @@ export abstract class BaseAgentRunner extends BaseAgent {
     taskId: string,
     stepOutput: TaskStepOutput,
     kwargs?: any,
-  ): Promise<AgentChatResponse>;
+  ): Promise<AgentChatResponse | StreamingAgentChatResponse>;
 
   abstract undoStep(taskId: string): void;
 }

packages/core/src/agent/types.ts

@@ -1,7 +1,9 @@
 import type {
   AgentChatResponse,
   ChatEngineAgentParams,
+  StreamingAgentChatResponse,
 } from "../engines/chat/index.js";
+
 import type { QueryEngineParamsNonStreaming } from "../types.js";
 
 export interface AgentWorker {
@@ -12,11 +14,15 @@ export interface AgentWorker {
 }
 
 interface BaseChatEngine {
-  chat(params: ChatEngineAgentParams): Promise<AgentChatResponse>;
+  chat(
+    params: ChatEngineAgentParams,
+  ): Promise<AgentChatResponse | StreamingAgentChatResponse>;
 }
 
 interface BaseQueryEngine {
-  query(params: QueryEngineParamsNonStreaming): Promise<AgentChatResponse>;
+  query(
+    params: QueryEngineParamsNonStreaming,
+  ): Promise<AgentChatResponse | StreamingAgentChatResponse>;
 }
 
 /**
@@ -31,7 +37,10 @@ export abstract class BaseAgent implements BaseChatEngine, BaseQueryEngine {
     return [];
   }
 
-  abstract chat(params: ChatEngineAgentParams): Promise<AgentChatResponse>;
+  abstract chat(
+    params: ChatEngineAgentParams,
+  ): Promise<AgentChatResponse | StreamingAgentChatResponse>;
+
   abstract reset(): void;
 
   /**
@@ -41,7 +50,7 @@ export abstract class BaseAgent implements BaseChatEngine, BaseQueryEngine {
    */
   async query(
     params: QueryEngineParamsNonStreaming,
-  ): Promise<AgentChatResponse> {
+  ): Promise<AgentChatResponse | StreamingAgentChatResponse> {
     // Handle non-streaming query
     const agentResponse = await this.chat({
       message: params.query,
@@ -161,13 +170,13 @@ export class TaskStep implements ITaskStep {
  * @param isLast: isLast
  */
 export class TaskStepOutput {
-  output: unknown;
+  output: any;
   taskStep: TaskStep;
   nextSteps: TaskStep[];
   isLast: boolean;
 
   constructor(
-    output: unknown,
+    output: any,
     taskStep: TaskStep,
     nextSteps: TaskStep[],
     isLast: boolean = false,

packages/core/src/cloud/LlamaCloudIndex.ts

@@ -3,7 +3,7 @@ import { RetrieverQueryEngine } from "../engines/query/RetrieverQueryEngine.js";
 import type { BaseNodePostprocessor } from "../postprocessors/types.js";
 import type { BaseSynthesizer } from "../synthesizers/types.js";
 import type { BaseQueryEngine } from "../types.js";
-import type { RetrieveParams } from "./LlamaCloudRetriever.js";
+import type { CloudRetrieveParams } from "./LlamaCloudRetriever.js";
 import { LlamaCloudRetriever } from "./LlamaCloudRetriever.js";
 import type { CloudConstructorParams } from "./types.js";
 
@@ -14,7 +14,7 @@ export class LlamaCloudIndex {
     this.params = params;
   }
 
-  asRetriever(params: RetrieveParams = {}): BaseRetriever {
+  asRetriever(params: CloudRetrieveParams = {}): BaseRetriever {
     return new LlamaCloudRetriever({ ...this.params, ...params });
   }
 
@@ -23,7 +23,7 @@ export class LlamaCloudIndex {
       responseSynthesizer?: BaseSynthesizer;
       preFilters?: unknown;
       nodePostprocessors?: BaseNodePostprocessor[];
-    } & RetrieveParams,
+    } & CloudRetrieveParams,
   ): BaseQueryEngine {
     const retriever = new LlamaCloudRetriever({
       ...this.params,

packages/core/src/cloud/LlamaCloudRetriever.ts

@@ -2,15 +2,14 @@ import type { PlatformApi, PlatformApiClient } from "@llamaindex/cloud";
 import { globalsHelper } from "../GlobalsHelper.js";
 import type { NodeWithScore } from "../Node.js";
 import { ObjectType, jsonToNode } from "../Node.js";
-import type { BaseRetriever } from "../Retriever.js";
+import type { BaseRetriever, RetrieveParams } from "../Retriever.js";
 import type { ServiceContext } from "../ServiceContext.js";
 import { serviceContextFromDefaults } from "../ServiceContext.js";
-import type { Event } from "../callbacks/CallbackManager.js";
 import type { ClientParams, CloudConstructorParams } from "./types.js";
 import { DEFAULT_PROJECT_NAME } from "./types.js";
 import { getClient } from "./utils.js";
 
-export type RetrieveParams = Omit<
+export type CloudRetrieveParams = Omit<
   PlatformApi.RetrievalParams,
   "query" | "searchFilters" | "pipelineId" | "className"
 > & { similarityTopK?: number };
@@ -18,7 +17,7 @@ export type RetrieveParams = Omit<
 export class LlamaCloudRetriever implements BaseRetriever {
   client?: PlatformApiClient;
   clientParams: ClientParams;
-  retrieveParams: RetrieveParams;
+  retrieveParams: CloudRetrieveParams;
   projectName: string = DEFAULT_PROJECT_NAME;
   pipelineName: string;
   serviceContext: ServiceContext;
@@ -35,7 +34,7 @@ export class LlamaCloudRetriever implements BaseRetriever {
     });
   }
 
-  constructor(params: CloudConstructorParams & RetrieveParams) {
+  constructor(params: CloudConstructorParams & CloudRetrieveParams) {
     this.clientParams = { apiKey: params.apiKey, baseUrl: params.baseUrl };
     if (params.similarityTopK) {
       params.denseSimilarityTopK = params.similarityTopK;
@@ -55,11 +54,11 @@ export class LlamaCloudRetriever implements BaseRetriever {
     return this.client;
   }
 
-  async retrieve(
-    query: string,
-    parentEvent?: Event | undefined,
-    preFilters?: unknown,
-  ): Promise<NodeWithScore[]> {
+  async retrieve({
+    query,
+    parentEvent,
+    preFilters,
+  }: RetrieveParams): Promise<NodeWithScore[]> {
     const pipelines = await (
       await this.getClient()
     ).pipeline.searchPipelines({

packages/core/src/cloud/utils.ts

@@ -1,4 +1,5 @@
 import type { PlatformApiClient } from "@llamaindex/cloud";
+import { getEnv } from "@llamaindex/env";
 import type { ClientParams } from "./types.js";
 import { DEFAULT_BASE_URL } from "./types.js";
 
@@ -7,8 +8,8 @@ export async function getClient({
   baseUrl,
 }: ClientParams = {}): Promise<PlatformApiClient> {
   // Get the environment variables or use defaults
-  baseUrl = baseUrl ?? process.env.LLAMA_CLOUD_BASE_URL ?? DEFAULT_BASE_URL;
-  apiKey = apiKey ?? process.env.LLAMA_CLOUD_API_KEY;
+  baseUrl = baseUrl ?? getEnv("LLAMA_CLOUD_BASE_URL") ?? DEFAULT_BASE_URL;
+  apiKey = apiKey ?? getEnv("LLAMA_CLOUD_API_KEY");
 
   const { PlatformApiClient } = await import("@llamaindex/cloud");
 

packages/core/src/embeddings/HuggingFaceEmbedding.ts

@@ -20,6 +20,7 @@ export enum HuggingFaceEmbeddingModelType {
  */
 export class HuggingFaceEmbedding extends BaseEmbedding {
   modelType: string = HuggingFaceEmbeddingModelType.XENOVA_ALL_MINILM_L6_V2;
+  quantized: boolean = true;
 
   private extractor: any;
 
@@ -31,7 +32,9 @@ export class HuggingFaceEmbedding extends BaseEmbedding {
   async getExtractor() {
     if (!this.extractor) {
       const { pipeline } = await import("@xenova/transformers");
-      this.extractor = await pipeline("feature-extraction", this.modelType);
+      this.extractor = await pipeline("feature-extraction", this.modelType, {
+        quantized: this.quantized,
+      });
     }
     return this.extractor;
   }

packages/core/src/embeddings/fireworks.ts

@@ -1,9 +1,10 @@
+import { getEnv } from "@llamaindex/env";
 import { OpenAIEmbedding } from "./OpenAIEmbedding.js";
 
 export class FireworksEmbedding extends OpenAIEmbedding {
   constructor(init?: Partial<OpenAIEmbedding>) {
     const {
-      apiKey = process.env.FIREWORKS_API_KEY,
+      apiKey = getEnv("FIREWORKS_API_KEY"),
       additionalSessionOptions = {},
       model = "nomic-ai/nomic-embed-text-v1.5",
       ...rest

packages/core/src/embeddings/together.ts

@@ -1,9 +1,10 @@
+import { getEnv } from "@llamaindex/env";
 import { OpenAIEmbedding } from "./OpenAIEmbedding.js";
 
 export class TogetherEmbedding extends OpenAIEmbedding {
   constructor(init?: Partial<OpenAIEmbedding>) {
     const {
-      apiKey = process.env.TOGETHER_API_KEY,
+      apiKey = getEnv("TOGETHER_API_KEY"),
       additionalSessionOptions = {},
       model = "togethercomputer/m2-bert-80M-32k-retrieval",
       ...rest

packages/core/src/embeddings/utils.ts

@@ -1,5 +1,6 @@
 import { defaultFS } from "@llamaindex/env";
 import _ from "lodash";
+import { filetypemime } from "magic-bytes.js";
 import type { ImageType } from "../Node.js";
 import { DEFAULT_SIMILARITY_TOP_K } from "../constants.js";
 import { VectorStoreQueryMode } from "../storage/vectorStore/types.js";
@@ -199,13 +200,12 @@ export function getTopKMMREmbeddings(
 }
 
 async function blobToDataUrl(input: Blob) {
-  const { fileTypeFromBuffer } = await import("file-type");
   const buffer = Buffer.from(await input.arrayBuffer());
-  const type = await fileTypeFromBuffer(buffer);
-  if (!type) {
+  const mimes = filetypemime(buffer);
+  if (mimes.length < 1) {
     throw new Error("Unsupported image type");
   }
-  return "data:" + type.mime + ";base64," + buffer.toString("base64");
+  return "data:" + mimes[0] + ";base64," + buffer.toString("base64");
 }
 
 export async function readImage(input: ImageType) {

packages/core/src/engines/chat/DefaultContextGenerator.ts

@@ -64,10 +64,10 @@ export class DefaultContextGenerator
         tags: ["final"],
       };
     }
-    const sourceNodesWithScore = await this.retriever.retrieve(
-      message,
+    const sourceNodesWithScore = await this.retriever.retrieve({
+      query: message,
       parentEvent,
-    );
+    });
 
     const nodes = await this.applyNodePostprocessors(
       sourceNodesWithScore,

packages/core/src/engines/chat/types.ts

@@ -27,6 +27,7 @@ export interface ChatEngineParamsNonStreaming extends ChatEngineParamsBase {
 
 export interface ChatEngineAgentParams extends ChatEngineParamsBase {
   toolChoice?: string | Record<string, any>;
+  stream?: boolean;
 }
 
 /**
@@ -86,3 +87,20 @@ export class AgentChatResponse {
     return this.response ?? "";
   }
 }
+
+export class StreamingAgentChatResponse {
+  response: AsyncIterable<Response>;
+
+  sources: ToolOutput[];
+  sourceNodes?: BaseNode[];
+
+  constructor(
+    response: AsyncIterable<Response>,
+    sources?: ToolOutput[],
+    sourceNodes?: BaseNode[],
+  ) {
+    this.response = response;
+    this.sources = sources ?? [];
+    this.sourceNodes = sourceNodes ?? [];
+  }
+}

packages/core/src/engines/query/RetrieverQueryEngine.ts

@@ -63,11 +63,11 @@ export class RetrieverQueryEngine
   }
 
   private async retrieve(query: string, parentEvent: Event) {
-    const nodes = await this.retriever.retrieve(
+    const nodes = await this.retriever.retrieve({
       query,
       parentEvent,
-      this.preFilters,
-    );
+      preFilters: this.preFilters,
+    });
 
     return await this.applyNodePostprocessors(nodes, query);
   }