Release 0.4.12 (#1013)

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>

.github/workflows/preview.yml

@@ -0,0 +1,28 @@
+name: Publish Preview
+on: [pull_request]
+
+jobs:
+  pre_release:
+    name: Pre Release
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Checkout Repo
+        uses: actions/checkout@v4
+
+      - uses: pnpm/action-setup@v3
+
+      - 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
+
+      - name: Pre Release
+        run: pnpx pkg-pr-new publish ./packages/*

.github/workflows/publish.yml

@@ -1,36 +0,0 @@
-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@v3
-      - 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/llamaindex
-        run: npx jsr publish --allow-slow-types
-        working-directory: packages/llamaindex
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

.github/workflows/release.yml

@@ -58,9 +58,11 @@ jobs:
 
       # Refs: https://github.com/changesets/changesets/issues/421
       - name: Update lock file
+        continue-on-error: true
         run: pnpm install --lockfile-only
 
       - name: Commit lock file
+        continue-on-error: true
         uses: stefanzweifel/git-auto-commit-action@v5
         with:
           commit_message: "chore: update lock file"

.github/workflows/test.yml

@@ -128,6 +128,12 @@ jobs:
         run: pnpm run build
       - name: Copy examples
         run: rsync -rv --exclude=node_modules ./examples ${{ runner.temp }}
+      - name: Pack @llamaindex/cloud
+        run: pnpm pack --pack-destination ${{ runner.temp }}
+        working-directory: packages/cloud
+      - name: Pack @llamaindex/core
+        run: pnpm pack --pack-destination ${{ runner.temp }}
+        working-directory: packages/core
       - name: Pack @llamaindex/env
         run: pnpm pack --pack-destination ${{ runner.temp }}
         working-directory: packages/env

.gitignore

@@ -48,3 +48,6 @@ playwright/.cache/
 
 # intellij
 **/.idea
+
+# generated API
+packages/cloud/src/client

README.md

@@ -76,7 +76,7 @@ main();
 node --import tsx ./main.ts
 ```
 
-### Next.js
+### React Server Component (Next.js, Waku, Redwood.JS...)
 
 First, you will need to add a llamaindex plugin to your Next.js project.
 
@@ -154,40 +154,6 @@ export async function chatWithAgent(
 }
 ```
 
-### Cloudflare Workers
-
-```ts
-// src/index.ts
-export default {
-  async fetch(
-    request: Request,
-    env: Env,
-    ctx: ExecutionContext,
-  ): Promise<Response> {
-    const { setEnvs } = await import("@llamaindex/env");
-    // set environment variables so that the OpenAIAgent can use them
-    setEnvs(env);
-    const { OpenAIAgent } = await import("llamaindex");
-    const agent = new OpenAIAgent({
-      tools: [],
-    });
-    const responseStream = await agent.chat({
-      stream: true,
-      message: "Hello? What is the weather today?",
-    });
-    const textEncoder = new TextEncoder();
-    const response = responseStream.pipeThrough(
-      new TransformStream({
-        transform: (chunk, controller) => {
-          controller.enqueue(textEncoder.encode(chunk.response.delta));
-        },
-      }),
-    );
-    return new Response(response);
-  },
-};
-```
-
 ## Playground
 
 Check out our NextJS playground at https://llama-playground.vercel.app/. The source is available at https://github.com/run-llama/ts-playground

apps/docs/CHANGELOG.md

@@ -1,5 +1,71 @@
 # docs
 
+## 0.0.38
+
+### Patch Changes
+
+- Updated dependencies [f326ab8]
+  - llamaindex@0.4.12
+
+## 0.0.37
+
+### Patch Changes
+
+- Updated dependencies [8bf5b4a]
+  - llamaindex@0.4.11
+
+## 0.0.36
+
+### Patch Changes
+
+- Updated dependencies [7dce3d2]
+  - llamaindex@0.4.10
+
+## 0.0.35
+
+### Patch Changes
+
+- Updated dependencies [3a96a48]
+  - llamaindex@0.4.9
+
+## 0.0.34
+
+### Patch Changes
+
+- Updated dependencies [83ebdfb]
+  - llamaindex@0.4.8
+
+## 0.0.33
+
+### Patch Changes
+
+- Updated dependencies [41fe871]
+- Updated dependencies [321c39d]
+- Updated dependencies [f7f1af0]
+  - llamaindex@0.4.7
+
+## 0.0.32
+
+### Patch Changes
+
+- Updated dependencies [1feb23b]
+- Updated dependencies [08c55ec]
+  - llamaindex@0.4.6
+
+## 0.0.31
+
+### Patch Changes
+
+- Updated dependencies [6c3e5d0]
+  - llamaindex@0.4.5
+
+## 0.0.30
+
+### Patch Changes
+
+- Updated dependencies [42eb73a]
+  - llamaindex@0.4.4
+
 ## 0.0.29
 
 ### Patch Changes

apps/docs/docs/examples/agent_gemini.mdx

@@ -0,0 +1,6 @@
+# Gemini Agent
+
+import CodeBlock from "@theme/CodeBlock";
+import CodeSourceGemini from "!raw-loader!../../../../examples/gemini/agent.ts";
+
+<CodeBlock language="ts">{CodeSourceGemini}</CodeBlock>

apps/docs/docs/modules/agent/index.md

@@ -12,12 +12,14 @@ An “agent” is an automated reasoning and decision engine. It takes in a user
 LlamaIndex.TS comes with a few built-in agents, but you can also create your own. The built-in agents include:
 
 - OpenAI Agent
-- Anthropic Agent
+- Anthropic Agent both via Anthropic and Bedrock (in `@llamaIndex/community`)
+- Gemini Agent
 - ReACT Agent
 
 ## Examples
 
 - [OpenAI Agent](../../examples/agent.mdx)
+- [Gemini Agent](../../examples/agent_gemini.mdx)
 
 ## Api References
 

apps/docs/docs/modules/llms/available_llms/bedrock.md

@@ -15,7 +15,7 @@ Settings.llm = new Bedrock({
 });
 ```
 
-Currently only supports Anthropic models:
+Currently only supports Anthropic and Meta models:
 
 ```ts
 ANTHROPIC_CLAUDE_INSTANT_1 = "anthropic.claude-instant-v1";
@@ -25,6 +25,10 @@ ANTHROPIC_CLAUDE_3_SONNET = "anthropic.claude-3-sonnet-20240229-v1:0";
 ANTHROPIC_CLAUDE_3_HAIKU = "anthropic.claude-3-haiku-20240307-v1:0";
 ANTHROPIC_CLAUDE_3_OPUS = "anthropic.claude-3-opus-20240229-v1:0"; // available on us-west-2
 ANTHROPIC_CLAUDE_3_5_SONNET = "anthropic.claude-3-5-sonnet-20240620-v1:0";
+META_LLAMA2_13B_CHAT = "meta.llama2-13b-chat-v1";
+META_LLAMA2_70B_CHAT = "meta.llama2-70b-chat-v1";
+META_LLAMA3_8B_INSTRUCT = "meta.llama3-8b-instruct-v1:0";
+META_LLAMA3_70B_INSTRUCT = "meta.llama3-70b-instruct-v1:0";
 ```
 
 Sonnet, Haiku and Opus are multimodal, image_url only supports base64 data url format, e.g. `data:image/jpeg;base64,SGVsbG8sIFdvcmxkIQ==`

apps/docs/package.json

@@ -1,6 +1,6 @@
 {
   "name": "docs",
-  "version": "0.0.29",
+  "version": "0.0.38",
   "private": true,
   "scripts": {
     "docusaurus": "docusaurus",

examples/README.md

@@ -9,7 +9,7 @@ make sure you have basic knowledge of the [LlamaIndexTS](https://ts.llamaindex.a
 # export your API key
 export OPENAI_API_KEY="sk-..."
 
-npx ts-node ./chatEngine.ts
+npx tsx ./chatEngine.ts
 ```
 
 ## Build your own RAG app

examples/agent/azure_dynamic_session.ts

@@ -0,0 +1,54 @@
+import "dotenv/config";
+
+import {
+  DefaultAzureCredential,
+  getBearerTokenProvider,
+} from "@azure/identity";
+import { AzureDynamicSessionTool, OpenAI, ReActAgent } from "llamaindex";
+
+async function main() {
+  const credential = new DefaultAzureCredential();
+  const azureADTokenProvider = getBearerTokenProvider(
+    credential,
+    "https://cognitiveservices.azure.com/.default",
+  );
+
+  const azure = {
+    azureADTokenProvider,
+    deployment: process.env.AZURE_OPENAI_DEPLOYMENT ?? "gpt-35-turbo",
+  };
+
+  // configure LLM model
+  const llm = new OpenAI({
+    azure,
+  });
+
+  const azureDynamicSession = new AzureDynamicSessionTool();
+
+  // Create an ReActAgent with the azure dynamic session tool
+  const agent = new ReActAgent({
+    llm,
+    tools: [azureDynamicSession],
+    // verbose: true,
+    systemPrompt: `You are a Python interpreter.
+      - You are given tasks to complete and you run python code to solve them.
+      - The python code runs by the python runtime. Every time you call $(interpreter) tool, the python code is executed in a separate cell. It's okay to make multiple calls to $(interpreter).
+      - You can run any python code you want in a secure environment.
+      - For images, return the full URL, not the base64 data.
+      - Return any image content as an HTML tag with the src attribute set to the URL of the image.`,
+  });
+
+  // Chat with the agent
+  const response = await agent.chat({
+    message:
+      "plot a chart of 5 random numbers and save it to /mnt/data/chart.png",
+    stream: false,
+  });
+
+  // Print the response
+  console.log({ response });
+}
+
+void main().then(() => {
+  console.log("Done");
+});

examples/astradb/README.md

@@ -24,7 +24,7 @@ Here are two sample scripts which work well with the sample data in the Astra Po
 
 Loads and queries a simple vectorstore with some documents about Astra DB
 
-run `ts-node astradb/example`
+run `tsx astradb/example`
 
 ## Movie Reviews Example
 
@@ -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 `npx ts-node astradb/load`
+run `npx tsx 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 `npx ts-node astradb/query`
+run `npx tsx astradb/query`

examples/chromadb/README.md

@@ -6,7 +6,7 @@ Export your OpenAI API Key using `export OPEN_API_KEY=insert your api key here`
 
 If you haven't installed chromadb, run `pip install chromadb`. Start the server using `chroma run`.
 
-Now, open a new terminal window and inside `examples`, run `pnpm dlx ts-node chromadb/test.ts`.
+Now, open a new terminal window and inside `examples`, run `pnpm dlx tsx chromadb/test.ts`.
 
 Here's the output for the input query `Tell me about Godfrey Cheshire's rating of La Sapienza.`:
 

examples/cloud/README.md

@@ -21,7 +21,7 @@ export LLAMA_CLOUD_BASE_URL="https://api.staging.llamaindex.ai"
 This example is using the managed index named `test` from the project `default` to create a chat engine.
 
 ```shell
-pnpx ts-node cloud/chat.ts
+pnpx tsx cloud/chat.ts
 ```
 
 ## Query Engine
@@ -29,7 +29,7 @@ pnpx ts-node cloud/chat.ts
 This example shows how to use the managed index with a query engine.
 
 ```shell
-pnpx ts-node cloud/query.ts
+pnpx tsx cloud/query.ts
 ```
 
 ## Pipeline
@@ -37,5 +37,5 @@ pnpx ts-node cloud/query.ts
 This example shows how to create a managed index with a pipeline.
 
 ```shell
-pnpx ts-node cloud/pipeline.ts
+pnpx tsx cloud/pipeline.ts
 ```

examples/cloud/chat.ts

@@ -6,7 +6,7 @@ import { ContextChatEngine, LlamaCloudIndex } from "llamaindex";
 async function main() {
   const index = new LlamaCloudIndex({
     name: "test",
-    projectName: "default",
+    projectName: "Default",
     baseUrl: process.env.LLAMA_CLOUD_BASE_URL,
     apiKey: process.env.LLAMA_CLOUD_API_KEY,
   });
@@ -19,10 +19,10 @@ async function main() {
   while (true) {
     const query = await rl.question("User: ");
     const stream = await chatEngine.chat({ message: query, stream: true });
-    console.log();
     for await (const chunk of stream) {
       process.stdout.write(chunk.response);
     }
+    process.stdout.write("\n");
   }
 }
 

examples/gemini/agent.ts

@@ -0,0 +1,65 @@
+import { FunctionTool, Gemini, GEMINI_MODEL, LLMAgent } from "llamaindex";
+
+const sumNumbers = FunctionTool.from(
+  ({ a, b }: { a: number; b: number }) => `${a + b}`,
+  {
+    name: "sumNumbers",
+    description: "Use this function to sum two numbers",
+    parameters: {
+      type: "object",
+      properties: {
+        a: {
+          type: "number",
+          description: "The first number",
+        },
+        b: {
+          type: "number",
+          description: "The second number",
+        },
+      },
+      required: ["a", "b"],
+    },
+  },
+);
+
+const divideNumbers = FunctionTool.from(
+  ({ a, b }: { a: number; b: number }) => `${a / b}`,
+  {
+    name: "divideNumbers",
+    description: "Use this function to divide two numbers",
+    parameters: {
+      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() {
+  const gemini = new Gemini({
+    model: GEMINI_MODEL.GEMINI_PRO,
+  });
+  const agent = new LLMAgent({
+    llm: gemini,
+    tools: [sumNumbers, divideNumbers],
+  });
+
+  const response = await agent.chat({
+    message: "How much is 5 + 5? then divide by 2",
+  });
+
+  console.log(response.message);
+}
+
+void main().then(() => {
+  console.log("Done");
+});

examples/milvus/README.md

@@ -25,10 +25,10 @@ Here are two sample scripts which work with loading and querying data from a Mil
 
 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`
+run `npx tsx 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`
+run `npx tsx milvus/query`

examples/mongodb/README.md

@@ -34,7 +34,7 @@ MONGODB_COLLECTION=tiny_tweets_collection
 You are now ready to import our ready-made data set into Mongo. This is the file `tinytweets.json`, a selection of approximately 1000 tweets from @seldo on Twitter in mid-2019. With your environment set up you can do this by running
 
 ```
-npx ts-node mongodb/1_import.ts
+npx tsx mongodb/1_import.ts
 ```
 
 If you don't want to use tweets, you can replace `json_file` with any other array of JSON objects, but you will need to modify some code later to make sure the correct field gets indexed. There is no LlamaIndex-specific code here; you can load your data into Mongo any way you want to.
@@ -59,7 +59,7 @@ MONGODB_VECTOR_INDEX=tiny_tweets_vector_index
 If the data you're indexing is the tweets we gave you, you're ready to go:
 
 ```bash
-npx ts-node mongodb/2_load_and_index.ts
+npx tsx mongodb/2_load_and_index.ts
 ```
 
 > Note: this script is running a couple of minutes and currently doesn't show any progress.
@@ -112,7 +112,7 @@ Now you're ready to query your data!
 You can do this by running
 
 ```bash
-npx ts-node mongodb/3_query.ts
+npx tsx mongodb/3_query.ts
 ```
 
 This sets up a connection to Atlas just like `2_load_and_index.ts` did, then it creates a [query engine](https://docs.llamaindex.ai/en/stable/understanding/querying/querying.html#getting-started) and runs a query against it.

examples/package.json

@@ -4,6 +4,7 @@
   "version": "0.0.6",
   "dependencies": {
     "@aws-crypto/sha256-js": "^5.2.0",
+    "@azure/identity": "^4.2.1",
     "@datastax/astra-db-ts": "^1.2.1",
     "@notionhq/client": "^2.2.15",
     "@pinecone-database/pinecone": "^2.2.2",
@@ -18,7 +19,6 @@
   },
   "devDependencies": {
     "@types/node": "^20.14.1",
-    "ts-node": "^10.9.2",
     "tsx": "^4.15.6",
     "typescript": "^5.5.2"
   },

examples/pg-vector-store/README.md

@@ -37,7 +37,7 @@ Read and follow the instructions in the README.md file located one directory up
 
 To import documents and save the embedding vectors to your database:
 
-> `npx ts-node pg-vector-store/load-docs.ts data`
+> `npx tsx pg-vector-store/load-docs.ts data`
 
 where data is the directory containing your input files. Using the `data` directory in the example above will read all of the files in that directory using the LlamaIndexTS default readers for each file type.
 
@@ -45,6 +45,6 @@ where data is the directory containing your input files. Using the `data` direct
 
 To query using the resulting vector store:
 
-> `npx ts-node pg-vector-store/query.ts`
+> `npx tsx pg-vector-store/query.ts`
 
 The script will prompt for a question, then process and present the answer using the PGVectorStore data and your OpenAI API key. It will continue to prompt until you enter `q`, `quit` or `exit` as the next query.

examples/pinecone-vector-store/README.md

@@ -19,7 +19,7 @@ Read and follow the instructions in the README.md file located one directory up
 
 To import documents and save the embedding vectors to your database:
 
-> `npx ts-node pinecone-vector-store/load-docs.ts data`
+> `npx tsx pinecone-vector-store/load-docs.ts data`
 
 where data is the directory containing your input files. Using the _data_ directory in the example above will read all of the files in that directory using the llamaindexTS default readers for each file type.
 
@@ -29,6 +29,6 @@ where data is the directory containing your input files. Using the _data_ direct
 
 To query using the resulting vector store:
 
-> `npx ts-node pinecone-vector-store/query.ts`
+> `npx tsx pinecone-vector-store/query.ts`
 
 The script will prompt for a question, then process and present the answer using the PineconeVectorStore data and your OpenAI API key. It will continue to prompt until you enter `q`, `quit` or `exit` as the next query.

examples/qdrantdb/README.md

@@ -8,4 +8,4 @@ Add your OpenAI API Key into a file called `.env` in the parent folder of this d
 OPEN_API_KEY=sk-you-key
 ```
 
-Now, open a new terminal window and inside `examples`, run `npx ts-node qdrantdb/preFilters.ts`.
+Now, open a new terminal window and inside `examples`, run `npx tsx qdrantdb/preFilters.ts`.

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

@@ -7,7 +7,7 @@ import {
 import { TextFileReader } from "llamaindex/readers/TextFileReader";
 
 class ZipReader extends FileReader {
-  loadDataAsContent(fileContent: Buffer): Promise<Document<Metadata>[]> {
+  loadDataAsContent(fileContent: Uint8Array): Promise<Document<Metadata>[]> {
     throw new Error("Implement me");
   }
 }

package.json

@@ -12,10 +12,9 @@
     "e2e": "turbo run e2e",
     "test": "turbo run test",
     "type-check": "tsc -b --diagnostics",
-    "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.mjs",
-    "new-version": "changeset version && pnpm run check-minor-version && pnpm format:write && pnpm run build:release",
+    "release": "pnpm run build:release && changeset publish",
+    "release-snapshot": "pnpm run build:release && changeset publish --tag snapshot",
+    "new-version": "changeset version && pnpm format:write && pnpm run build:release",
     "new-snapshot": "pnpm run build:release && changeset version --snapshot"
   },
   "devDependencies": {
@@ -24,14 +23,14 @@
     "eslint": "^8.57.0",
     "eslint-config-next": "^14.2.4",
     "eslint-config-prettier": "^9.1.0",
-    "eslint-config-turbo": "^1.13.4",
+    "eslint-config-turbo": "^2.0.5",
     "eslint-plugin-react": "7.34.1",
     "husky": "^9.0.11",
     "lint-staged": "^15.2.7",
     "madge": "^7.0.0",
     "prettier": "^3.3.2",
     "prettier-plugin-organize-imports": "^3.2.4",
-    "turbo": "^1.13.4",
+    "turbo": "^2.0.5",
     "typescript": "^5.5.2"
   },
   "packageManager": "pnpm@9.4.0",
@@ -40,6 +39,9 @@
       "trim": "1.0.1",
       "@babel/traverse": "7.23.2",
       "protobufjs": "7.2.6"
+    },
+    "patchedDependencies": {
+      "bunchee@5.2.1": "patches/bunchee@5.2.1.patch"
     }
   },
   "lint-staged": {

packages/autotool/examples/02_nextjs/CHANGELOG.md

@@ -1,5 +1,80 @@
 # @llamaindex/autotool-02-next-example
 
+## 0.1.22
+
+### Patch Changes
+
+- Updated dependencies [f326ab8]
+  - llamaindex@0.4.12
+  - @llamaindex/autotool@1.0.0
+
+## 0.1.21
+
+### Patch Changes
+
+- Updated dependencies [8bf5b4a]
+  - llamaindex@0.4.11
+  - @llamaindex/autotool@1.0.0
+
+## 0.1.20
+
+### Patch Changes
+
+- Updated dependencies [7dce3d2]
+  - llamaindex@0.4.10
+  - @llamaindex/autotool@1.0.0
+
+## 0.1.19
+
+### Patch Changes
+
+- Updated dependencies [3a96a48]
+  - llamaindex@0.4.9
+  - @llamaindex/autotool@1.0.0
+
+## 0.1.18
+
+### Patch Changes
+
+- Updated dependencies [83ebdfb]
+  - llamaindex@0.4.8
+  - @llamaindex/autotool@1.0.0
+
+## 0.1.17
+
+### Patch Changes
+
+- Updated dependencies [41fe871]
+- Updated dependencies [321c39d]
+- Updated dependencies [f7f1af0]
+  - llamaindex@0.4.7
+  - @llamaindex/autotool@1.0.0
+
+## 0.1.16
+
+### Patch Changes
+
+- Updated dependencies [1feb23b]
+- Updated dependencies [08c55ec]
+  - llamaindex@0.4.6
+  - @llamaindex/autotool@1.0.0
+
+## 0.1.15
+
+### Patch Changes
+
+- Updated dependencies [6c3e5d0]
+  - llamaindex@0.4.5
+  - @llamaindex/autotool@1.0.0
+
+## 0.1.14
+
+### Patch Changes
+
+- Updated dependencies [42eb73a]
+  - llamaindex@0.4.4
+  - @llamaindex/autotool@1.0.0
+
 ## 0.1.13
 
 ### Patch Changes

packages/autotool/examples/02_nextjs/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@llamaindex/autotool-02-next-example",
   "private": true,
-  "version": "0.1.13",
+  "version": "0.1.22",
   "scripts": {
     "dev": "next dev",
     "build": "next build",

packages/autotool/package.json

@@ -51,7 +51,7 @@
     "unplugin": "^1.10.1"
   },
   "peerDependencies": {
-    "llamaindex": "^0.4.3",
+    "llamaindex": "^0.4.12",
     "openai": "^4",
     "typescript": "^4"
   },

packages/cloud/CHANGELOG.md

@@ -0,0 +1,13 @@
+# @llamaindex/cloud
+
+## 0.1.2
+
+### Patch Changes
+
+- f326ab8: chore: bump version
+
+## 0.1.1
+
+### Patch Changes
+
+- 321c39d: fix: generate api as class

packages/cloud/README.md

@@ -0,0 +1,18 @@
+# @llamaindex/cloud
+
+> LlamaCloud is a new generation of managed parsing, ingestion, and retrieval services, designed to bring production-grade context-augmentation to your LLM and RAG applications.
+
+## Usage
+
+```ts
+import { OpenAPI } from "@llamaindex/cloud/api";
+OpenAPI.TOKEN = "YOUR_API_KEY";
+OpenAPI.BASE = "https://api.cloud.llamaindex.ai/";
+// ...
+```
+
+For more information, see the [API documentation](https://docs.cloud.llamaindex.ai/).
+
+## License
+
+MIT

packages/cloud/openapi-ts.config.ts

@@ -0,0 +1,18 @@
+import { defineConfig } from "@hey-api/openapi-ts";
+
+export default defineConfig({
+  // you can download this file to get the latest version of the OpenAPI document
+  // @link https://api.cloud.llamaindex.ai/api/openapi.json
+  input: "./openapi.json",
+  output: {
+    path: "./src/client",
+    format: "prettier",
+    lint: "eslint",
+  },
+  services: {
+    asClass: true,
+  },
+  types: {
+    enums: "javascript",
+  },
+});

packages/cloud/package.json

@@ -0,0 +1,40 @@
+{
+  "name": "@llamaindex/cloud",
+  "version": "0.1.2",
+  "type": "module",
+  "license": "MIT",
+  "scripts": {
+    "generate": "pnpm dlx @hey-api/openapi-ts",
+    "build": "pnpm run generate && bunchee"
+  },
+  "files": [
+    "openapi.json",
+    "dist"
+  ],
+  "exports": {
+    "./openapi.json": "./openapi.json",
+    "./api": {
+      "require": {
+        "types": "./dist/api.d.cts",
+        "default": "./dist/api.cjs"
+      },
+      "import": {
+        "types": "./dist/api.d.ts",
+        "default": "./dist/api.js"
+      },
+      "default": {
+        "types": "./dist/api.d.ts",
+        "default": "./dist/api.js"
+      }
+    }
+  },
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/himself65/LlamaIndexTS.git",
+    "directory": "packages/cloud"
+  },
+  "devDependencies": {
+    "@hey-api/openapi-ts": "^0.48.0",
+    "bunchee": "^5.2.1"
+  }
+}

packages/cloud/src/api.ts

@@ -0,0 +1 @@
+export * from "./client";

packages/cloud/tsconfig.json

@@ -0,0 +1,15 @@
+{
+  "extends": "../../tsconfig.json",
+  "compilerOptions": {
+    "rootDir": "./src",
+    "outDir": "./dist/type",
+    "tsBuildInfoFile": "./dist/.tsbuildinfo",
+    "emitDeclarationOnly": true,
+    "moduleResolution": "Bundler",
+    "skipLibCheck": true,
+    "strict": true,
+    "lib": ["DOM", "ESNext"]
+  },
+  "include": ["./src"],
+  "exclude": ["node_modules"]
+}

packages/community/CHANGELOG.md

@@ -1,5 +1,72 @@
 # @llamaindex/community
 
+## 0.0.16
+
+### Patch Changes
+
+- f326ab8: chore: bump version
+- Updated dependencies [f326ab8]
+  - llamaindex@0.4.12
+
+## 0.0.15
+
+### Patch Changes
+
+- Updated dependencies [8bf5b4a]
+  - llamaindex@0.4.11
+
+## 0.0.14
+
+### Patch Changes
+
+- Updated dependencies [7dce3d2]
+  - llamaindex@0.4.10
+
+## 0.0.13
+
+### Patch Changes
+
+- Updated dependencies [3a96a48]
+  - llamaindex@0.4.9
+
+## 0.0.12
+
+### Patch Changes
+
+- Updated dependencies [83ebdfb]
+  - llamaindex@0.4.8
+
+## 0.0.11
+
+### Patch Changes
+
+- Updated dependencies [41fe871]
+- Updated dependencies [321c39d]
+- Updated dependencies [f7f1af0]
+  - llamaindex@0.4.7
+
+## 0.0.10
+
+### Patch Changes
+
+- Updated dependencies [1feb23b]
+- Updated dependencies [08c55ec]
+  - llamaindex@0.4.6
+
+## 0.0.9
+
+### Patch Changes
+
+- Updated dependencies [6c3e5d0]
+  - llamaindex@0.4.5
+
+## 0.0.8
+
+### Patch Changes
+
+- Updated dependencies [42eb73a]
+  - llamaindex@0.4.4
+
 ## 0.0.7
 
 ### Patch Changes

packages/community/README.md

@@ -5,6 +5,7 @@
 ## Current Features:
 
 - Bedrock support for the Anthropic Claude Models [usage](https://ts.llamaindex.ai/modules/llms/available_llms/bedrock)
+- Bedrock support for the Meta LLama 2 and 3 Models [usage](https://ts.llamaindex.ai/modules/llms/available_llms/bedrock)
 
 ## LICENSE
 

packages/community/jsr.json

@@ -1,8 +0,0 @@
-{
-  "name": "@llamaindex/community",
-  "version": "0.0.5",
-  "exports": {
-    ".": "./src/index.ts",
-    "./type": "./src/type.ts"
-  }
-}

packages/community/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@llamaindex/community",
   "description": "Community package for LlamaIndexTS",
-  "version": "0.0.7",
+  "version": "0.0.16",
   "type": "module",
   "types": "dist/type/index.d.ts",
   "main": "dist/cjs/index.js",

packages/community/src/llm/bedrock/base.ts

@@ -1,53 +1,34 @@
 import {
   BedrockRuntimeClient,
+  type BedrockRuntimeClientConfig,
   InvokeModelCommand,
   InvokeModelWithResponseStreamCommand,
-  ResponseStream,
-  type BedrockRuntimeClientConfig,
-  type InvokeModelCommandInput,
-  type InvokeModelWithResponseStreamCommandInput,
 } from "@aws-sdk/client-bedrock-runtime";
 import type {
-  BaseTool,
   ChatMessage,
   ChatResponse,
-  ChatResponseChunk,
   CompletionResponse,
   LLMChatParamsNonStreaming,
   LLMChatParamsStreaming,
   LLMCompletionParamsNonStreaming,
   LLMCompletionParamsStreaming,
   LLMMetadata,
-  PartialToolCall,
-  ToolCall,
   ToolCallLLMMessageOptions,
 } from "llamaindex";
-import { ToolCallLLM, streamConverter, wrapLLMEvent } from "llamaindex";
-import type {
-  AnthropicNoneStreamingResponse,
-  AnthropicTextContent,
-  StreamEvent,
-  ToolBlock,
-  ToolChoice,
-} from "./types.js";
+import { streamConverter, ToolCallLLM, wrapLLMEvent } from "llamaindex";
 import {
-  mapBaseToolsToAnthropicTools,
-  mapChatMessagesToAnthropicMessages,
-  mapMessageContentToMessageContentDetails,
-  toUtf8,
-} from "./utils.js";
-
-export type BedrockAdditionalChatOptions = { toolChoice: ToolChoice };
+  type BedrockAdditionalChatOptions,
+  type BedrockChatStreamResponse,
+  Provider,
+} from "./provider";
+import { PROVIDERS } from "./providers";
+import { mapMessageContentToMessageContentDetails } from "./utils.js";
 
 export type BedrockChatParamsStreaming = LLMChatParamsStreaming<
   BedrockAdditionalChatOptions,
   ToolCallLLMMessageOptions
 >;
 
-export type BedrockChatStreamResponse = AsyncIterable<
-  ChatResponseChunk<ToolCallLLMMessageOptions>
->;
-
 export type BedrockChatParamsNonStreaming = LLMChatParamsNonStreaming<
   BedrockAdditionalChatOptions,
   ToolCallLLMMessageOptions
@@ -151,174 +132,6 @@ export const TOOL_CALL_MODELS = [
   BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_5_SONNET,
 ];
 
-abstract class Provider<ProviderStreamEvent extends {} = {}> {
-  abstract getTextFromResponse(response: Record<string, any>): string;
-
-  abstract getToolsFromResponse<T extends {} = {}>(
-    response: Record<string, any>,
-  ): T[];
-
-  getStreamingEventResponse(
-    response: Record<string, any>,
-  ): ProviderStreamEvent | undefined {
-    return response.chunk?.bytes
-      ? (JSON.parse(toUtf8(response.chunk?.bytes)) as ProviderStreamEvent)
-      : undefined;
-  }
-
-  async *reduceStream(
-    stream: AsyncIterable<ResponseStream>,
-  ): BedrockChatStreamResponse {
-    yield* streamConverter(stream, (response) => {
-      return {
-        delta: this.getTextFromStreamResponse(response),
-        raw: response,
-      };
-    });
-  }
-
-  getTextFromStreamResponse(response: Record<string, any>): string {
-    return this.getTextFromResponse(response);
-  }
-
-  abstract getRequestBody<T extends ChatMessage>(
-    metadata: LLMMetadata,
-    messages: T[],
-    tools?: BaseTool[],
-    options?: BedrockAdditionalChatOptions,
-  ): InvokeModelCommandInput | InvokeModelWithResponseStreamCommandInput;
-}
-
-class AnthropicProvider extends Provider<StreamEvent> {
-  getResultFromResponse(
-    response: Record<string, any>,
-  ): AnthropicNoneStreamingResponse {
-    return JSON.parse(toUtf8(response.body));
-  }
-
-  getToolsFromResponse<AnthropicToolContent>(
-    response: Record<string, any>,
-  ): AnthropicToolContent[] {
-    const result = this.getResultFromResponse(response);
-    return result.content
-      .filter((item) => item.type === "tool_use")
-      .map((item) => item as AnthropicToolContent);
-  }
-
-  getTextFromResponse(response: Record<string, any>): string {
-    const result = this.getResultFromResponse(response);
-    return result.content
-      .filter((item) => item.type === "text")
-      .map((item) => (item as AnthropicTextContent).text)
-      .join(" ");
-  }
-
-  getTextFromStreamResponse(response: Record<string, any>): string {
-    const event = this.getStreamingEventResponse(response);
-    if (event?.type === "content_block_delta") {
-      if (event.delta.type === "text_delta") return event.delta.text;
-      if (event.delta.type === "input_json_delta")
-        return event.delta.partial_json;
-    }
-    return "";
-  }
-
-  async *reduceStream(
-    stream: AsyncIterable<ResponseStream>,
-  ): BedrockChatStreamResponse {
-    let collecting = [];
-    let tool: ToolBlock | undefined = undefined;
-    // #TODO this should be broken down into a separate consumer
-    for await (const response of stream) {
-      const event = this.getStreamingEventResponse(response);
-      if (
-        event?.type === "content_block_start" &&
-        event.content_block.type === "tool_use"
-      ) {
-        tool = event.content_block;
-        continue;
-      }
-
-      if (
-        event?.type === "content_block_delta" &&
-        event.delta.type === "input_json_delta"
-      ) {
-        collecting.push(event.delta.partial_json);
-      }
-
-      let options: undefined | ToolCallLLMMessageOptions = undefined;
-      if (tool && collecting.length) {
-        const input = collecting.filter((item) => item).join("");
-        // We have all we need to parse the tool_use json
-        if (event?.type === "content_block_stop") {
-          options = {
-            toolCall: [
-              {
-                id: tool.id,
-                name: tool.name,
-                input: JSON.parse(input),
-              } as ToolCall,
-            ],
-          };
-          // reset the collection/tool
-          collecting = [];
-          tool = undefined;
-        } else {
-          options = {
-            toolCall: [
-              {
-                id: tool.id,
-                name: tool.name,
-                input,
-              } as PartialToolCall,
-            ],
-          };
-        }
-      }
-      const delta = this.getTextFromStreamResponse(response);
-      if (!delta && !options) continue;
-
-      yield {
-        delta,
-        options,
-        raw: response,
-      };
-    }
-  }
-
-  getRequestBody<T extends ChatMessage<ToolCallLLMMessageOptions>>(
-    metadata: LLMMetadata,
-    messages: T[],
-    tools?: BaseTool[],
-    options?: BedrockAdditionalChatOptions,
-  ): InvokeModelCommandInput | InvokeModelWithResponseStreamCommandInput {
-    const extra: Record<string, unknown> = {};
-    if (options?.toolChoice) {
-      extra["tool_choice"] = options?.toolChoice;
-    }
-    const mapped = mapChatMessagesToAnthropicMessages(messages);
-    return {
-      modelId: metadata.model,
-      contentType: "application/json",
-      accept: "application/json",
-      body: JSON.stringify({
-        anthropic_version: "bedrock-2023-05-31",
-        messages: mapped,
-        tools: mapBaseToolsToAnthropicTools(tools),
-        max_tokens: metadata.maxTokens,
-        temperature: metadata.temperature,
-        top_p: metadata.topP,
-        ...extra,
-      }),
-    };
-  }
-}
-
-// Other providers could go here
-const PROVIDERS: { [key: string]: Provider } = {
-  anthropic: new AnthropicProvider(),
-};
-
 const getProvider = (model: string): Provider => {
   const providerName = model.split(".")[0];
   if (!(providerName in PROVIDERS)) {
@@ -373,6 +186,10 @@ export class Bedrock extends ToolCallLLM<BedrockAdditionalChatOptions> {
     this.temperature = temperature ?? DEFAULT_BEDROCK_PARAMS.temperature;
     this.topP = topP ?? DEFAULT_BEDROCK_PARAMS.topP;
     this.client = new BedrockRuntimeClient(params);
+
+    if (!this.supportToolCall) {
+      console.warn(`The model "${this.model}" doesn't support ToolCall`);
+    }
   }
 
   get supportToolCall(): boolean {
@@ -402,10 +219,13 @@ export class Bedrock extends ToolCallLLM<BedrockAdditionalChatOptions> {
     );
     const command = new InvokeModelCommand(input);
     const response = await this.client.send(command);
-    const tools = this.provider.getToolsFromResponse(response);
-    const options: ToolCallLLMMessageOptions = tools.length
-      ? { toolCall: tools }
-      : {};
+    let options: ToolCallLLMMessageOptions = {};
+    if (this.supportToolCall) {
+      const tools = this.provider.getToolsFromResponse(response);
+      if (tools.length) {
+        options = { toolCall: tools };
+      }
+    }
     return {
       raw: response,
       message: {

packages/community/src/llm/bedrock/provider.ts

@@ -0,0 +1,59 @@
+import {
+  type InvokeModelCommandInput,
+  type InvokeModelWithResponseStreamCommandInput,
+  ResponseStream,
+} from "@aws-sdk/client-bedrock-runtime";
+import {
+  type BaseTool,
+  type ChatMessage,
+  type ChatResponseChunk,
+  type LLMMetadata,
+  streamConverter,
+  type ToolCallLLMMessageOptions,
+} from "llamaindex";
+import type { ToolChoice } from "./types";
+import { toUtf8 } from "./utils";
+
+export type BedrockAdditionalChatOptions = { toolChoice: ToolChoice };
+
+export type BedrockChatStreamResponse = AsyncIterable<
+  ChatResponseChunk<ToolCallLLMMessageOptions>
+>;
+
+export abstract class Provider<ProviderStreamEvent extends {} = {}> {
+  abstract getTextFromResponse(response: Record<string, any>): string;
+
+  abstract getToolsFromResponse<T extends {} = {}>(
+    response: Record<string, any>,
+  ): T[];
+
+  getStreamingEventResponse(
+    response: Record<string, any>,
+  ): ProviderStreamEvent | undefined {
+    return response.chunk?.bytes
+      ? (JSON.parse(toUtf8(response.chunk?.bytes)) as ProviderStreamEvent)
+      : undefined;
+  }
+
+  async *reduceStream(
+    stream: AsyncIterable<ResponseStream>,
+  ): BedrockChatStreamResponse {
+    yield* streamConverter(stream, (response) => {
+      return {
+        delta: this.getTextFromStreamResponse(response),
+        raw: response,
+      };
+    });
+  }
+
+  getTextFromStreamResponse(response: Record<string, any>): string {
+    return this.getTextFromResponse(response);
+  }
+
+  abstract getRequestBody<T extends ChatMessage>(
+    metadata: LLMMetadata,
+    messages: T[],
+    tools?: BaseTool[],
+    options?: BedrockAdditionalChatOptions,
+  ): InvokeModelCommandInput | InvokeModelWithResponseStreamCommandInput;
+}

packages/community/src/llm/bedrock/providers/anthropic.ts

@@ -0,0 +1,154 @@
+import {
+  type InvokeModelCommandInput,
+  type InvokeModelWithResponseStreamCommandInput,
+  ResponseStream,
+} from "@aws-sdk/client-bedrock-runtime";
+import type {
+  BaseTool,
+  ChatMessage,
+  LLMMetadata,
+  PartialToolCall,
+  ToolCall,
+  ToolCallLLMMessageOptions,
+} from "llamaindex";
+import {
+  type BedrockAdditionalChatOptions,
+  type BedrockChatStreamResponse,
+  Provider,
+} from "../provider";
+import type {
+  AnthropicNoneStreamingResponse,
+  AnthropicStreamEvent,
+  AnthropicTextContent,
+  ToolBlock,
+} from "../types";
+import {
+  mapBaseToolsToAnthropicTools,
+  mapChatMessagesToAnthropicMessages,
+  toUtf8,
+} from "../utils";
+
+export class AnthropicProvider extends Provider<AnthropicStreamEvent> {
+  getResultFromResponse(
+    response: Record<string, any>,
+  ): AnthropicNoneStreamingResponse {
+    return JSON.parse(toUtf8(response.body));
+  }
+
+  getToolsFromResponse<AnthropicToolContent>(
+    response: Record<string, any>,
+  ): AnthropicToolContent[] {
+    const result = this.getResultFromResponse(response);
+    return result.content
+      .filter((item) => item.type === "tool_use")
+      .map((item) => item as AnthropicToolContent);
+  }
+
+  getTextFromResponse(response: Record<string, any>): string {
+    const result = this.getResultFromResponse(response);
+    return result.content
+      .filter((item) => item.type === "text")
+      .map((item) => (item as AnthropicTextContent).text)
+      .join(" ");
+  }
+
+  getTextFromStreamResponse(response: Record<string, any>): string {
+    const event = this.getStreamingEventResponse(response);
+    if (event?.type === "content_block_delta") {
+      if (event.delta.type === "text_delta") return event.delta.text;
+      if (event.delta.type === "input_json_delta")
+        return event.delta.partial_json;
+    }
+    return "";
+  }
+
+  async *reduceStream(
+    stream: AsyncIterable<ResponseStream>,
+  ): BedrockChatStreamResponse {
+    let collecting = [];
+    let tool: ToolBlock | undefined = undefined;
+    // #TODO this should be broken down into a separate consumer
+    for await (const response of stream) {
+      const event = this.getStreamingEventResponse(response);
+      if (
+        event?.type === "content_block_start" &&
+        event.content_block.type === "tool_use"
+      ) {
+        tool = event.content_block;
+        continue;
+      }
+
+      if (
+        event?.type === "content_block_delta" &&
+        event.delta.type === "input_json_delta"
+      ) {
+        collecting.push(event.delta.partial_json);
+      }
+
+      let options: undefined | ToolCallLLMMessageOptions = undefined;
+      if (tool && collecting.length) {
+        const input = collecting.filter((item) => item).join("");
+        // We have all we need to parse the tool_use json
+        if (event?.type === "content_block_stop") {
+          options = {
+            toolCall: [
+              {
+                id: tool.id,
+                name: tool.name,
+                input: JSON.parse(input),
+              } as ToolCall,
+            ],
+          };
+          // reset the collection/tool
+          collecting = [];
+          tool = undefined;
+        } else {
+          options = {
+            toolCall: [
+              {
+                id: tool.id,
+                name: tool.name,
+                input,
+              } as PartialToolCall,
+            ],
+          };
+        }
+      }
+      const delta = this.getTextFromStreamResponse(response);
+      if (!delta && !options) continue;
+
+      yield {
+        delta,
+        options,
+        raw: response,
+      };
+    }
+  }
+
+  getRequestBody<T extends ChatMessage<ToolCallLLMMessageOptions>>(
+    metadata: LLMMetadata,
+    messages: T[],
+    tools?: BaseTool[],
+    options?: BedrockAdditionalChatOptions,
+  ): InvokeModelCommandInput | InvokeModelWithResponseStreamCommandInput {
+    const extra: Record<string, unknown> = {};
+    if (options?.toolChoice) {
+      extra["tool_choice"] = options?.toolChoice;
+    }
+    const mapped = mapChatMessagesToAnthropicMessages(messages);
+    return {
+      modelId: metadata.model,
+      contentType: "application/json",
+      accept: "application/json",
+      body: JSON.stringify({
+        anthropic_version: "bedrock-2023-05-31",
+        messages: mapped,
+        tools: mapBaseToolsToAnthropicTools(tools),
+        max_tokens: metadata.maxTokens,
+        temperature: metadata.temperature,
+        top_p: metadata.topP,
+        ...extra,
+      }),
+    };
+  }
+}

packages/community/src/llm/bedrock/providers/index.ts

@@ -0,0 +1,9 @@
+import { Provider } from "../provider";
+import { AnthropicProvider } from "./anthropic";
+import { MetaProvider } from "./meta";
+
+// Other providers should go here
+export const PROVIDERS: { [key: string]: Provider } = {
+  anthropic: new AnthropicProvider(),
+  meta: new MetaProvider(),
+};

packages/community/src/llm/bedrock/providers/meta.ts

@@ -0,0 +1,69 @@
+import type {
+  InvokeModelCommandInput,
+  InvokeModelWithResponseStreamCommandInput,
+} from "@aws-sdk/client-bedrock-runtime";
+import type { ChatMessage, LLMMetadata } from "llamaindex";
+import type { MetaNoneStreamingResponse, MetaStreamEvent } from "../types";
+import {
+  mapChatMessagesToMetaLlama2Messages,
+  mapChatMessagesToMetaLlama3Messages,
+  toUtf8,
+} from "../utils";
+
+import { Provider } from "../provider";
+
+export class MetaProvider extends Provider<MetaStreamEvent> {
+  constructor() {
+    super();
+  }
+
+  getResultFromResponse(
+    response: Record<string, any>,
+  ): MetaNoneStreamingResponse {
+    return JSON.parse(toUtf8(response.body));
+  }
+
+  getToolsFromResponse(_response: Record<string, any>): never {
+    throw new Error("Not supported by this provider.");
+  }
+
+  getTextFromResponse(response: Record<string, any>): string {
+    const result = this.getResultFromResponse(response);
+    return result.generation;
+  }
+
+  getTextFromStreamResponse(response: Record<string, any>): string {
+    const event = this.getStreamingEventResponse(response);
+    if (event?.generation) {
+      return event.generation;
+    }
+    return "";
+  }
+
+  getRequestBody<T extends ChatMessage>(
+    metadata: LLMMetadata,
+    messages: T[],
+  ): InvokeModelCommandInput | InvokeModelWithResponseStreamCommandInput {
+    let promptFunction: (messages: ChatMessage[]) => string;
+
+    if (metadata.model.startsWith("meta.llama3")) {
+      promptFunction = mapChatMessagesToMetaLlama3Messages;
+    } else if (metadata.model.startsWith("meta.llama2")) {
+      promptFunction = mapChatMessagesToMetaLlama2Messages;
+    } else {
+      throw new Error(`Meta model ${metadata.model} is not supported`);
+    }
+
+    return {
+      modelId: metadata.model,
+      contentType: "application/json",
+      accept: "application/json",
+      body: JSON.stringify({
+        prompt: promptFunction(messages),
+        max_gen_len: metadata.maxTokens,
+        temperature: metadata.temperature,
+        top_p: metadata.topP,
+      }),
+    };
+  }
+}

packages/community/src/llm/bedrock/types.ts

@@ -79,7 +79,7 @@ export type ToolChoice =
   | { type: "auto" }
   | { type: "tool"; name: string };
 
-export type StreamEvent =
+export type AnthropicStreamEvent =
   | { type: "message_start"; message: Message }
   | ContentBlockStart
   | ContentBlockDelta
@@ -93,6 +93,8 @@ export type AnthropicContent =
   | AnthropicToolContent
   | AnthropicToolResultContent;
 
+export type MetaTextContent = string;
+
 export type AnthropicTextContent = {
   type: "text";
   text: string;
@@ -133,6 +135,11 @@ export type AnthropicMessage = {
   content: AnthropicContent[];
 };
 
+export type MetaMessage = {
+  role: "user" | "assistant" | "system";
+  content: MetaTextContent;
+};
+
 export type AnthropicNoneStreamingResponse = {
   id: string;
   type: "message";
@@ -143,3 +150,16 @@ export type AnthropicNoneStreamingResponse = {
   stop_sequence?: string;
   usage: { input_tokens: number; output_tokens: number };
 };
+
+type MetaResponse = {
+  generation: string;
+  prompt_token_count: number;
+  generation_token_count: number;
+  stop_reason: "stop" | "length";
+};
+
+export type MetaStreamEvent = MetaResponse & {
+  "amazon-bedrock-invocationMetrics": InvocationMetrics;
+};
+
+export type MetaNoneStreamingResponse = MetaResponse;

packages/community/src/llm/bedrock/utils.ts

@@ -4,6 +4,7 @@ import type {
   JSONObject,
   MessageContent,
   MessageContentDetail,
+  MessageContentTextDetail,
   ToolCallLLMMessageOptions,
   ToolMetadata,
 } from "llamaindex";
@@ -13,6 +14,7 @@ import type {
   AnthropicMediaTypes,
   AnthropicMessage,
   AnthropicTextContent,
+  MetaMessage,
 } from "./types.js";
 
 const ACCEPTED_IMAGE_MIME_TYPES = [
@@ -148,6 +150,85 @@ export const mapChatMessagesToAnthropicMessages = <
   return mergeNeighboringSameRoleMessages(mapped);
 };
 
+export const mapChatMessagesToMetaMessages = <T extends ChatMessage>(
+  messages: T[],
+): MetaMessage[] => {
+  return messages.map((msg) => {
+    let content: string;
+    if (typeof msg.content === "string") {
+      content = msg.content;
+    } else {
+      content = (msg.content[0] as MessageContentTextDetail).text;
+    }
+    return {
+      role:
+        msg.role === "assistant"
+          ? "assistant"
+          : msg.role === "user"
+            ? "user"
+            : "system",
+      content,
+    };
+  });
+};
+
+/**
+ * Documentation at https://llama.meta.com/docs/model-cards-and-prompt-formats/meta-llama-3
+ */
+export const mapChatMessagesToMetaLlama3Messages = <T extends ChatMessage>(
+  messages: T[],
+): string => {
+  const mapped = mapChatMessagesToMetaMessages(messages).map((message) => {
+    const text = message.content;
+    return `<|start_header_id|>${message.role}<|end_header_id|>\n${text}\n<|eot_id|>\n`;
+  });
+  return (
+    "<|begin_of_text|>" +
+    mapped.join("\n") +
+    "\n<|start_header_id|>assistant<|end_header_id|>\n"
+  );
+};
+
+/**
+ * Documentation at https://llama.meta.com/docs/model-cards-and-prompt-formats/meta-llama-2
+ */
+export const mapChatMessagesToMetaLlama2Messages = <T extends ChatMessage>(
+  messages: T[],
+): string => {
+  const mapped = mapChatMessagesToMetaMessages(messages);
+  let output = "<s>";
+  let insideInst = false;
+  let needsStartAgain = false;
+  for (const message of mapped) {
+    if (needsStartAgain) {
+      output += "<s>";
+      needsStartAgain = false;
+    }
+    const text = message.content;
+    if (message.role === "system") {
+      if (!insideInst) {
+        output += "[INST] ";
+        insideInst = true;
+      }
+      output += `<<SYS>>\n${text}\n<</SYS>>\n`;
+    } else if (message.role === "user") {
+      output += text;
+      if (insideInst) {
+        output += " [/INST]";
+        insideInst = false;
+      }
+    } else if (message.role === "assistant") {
+      if (insideInst) {
+        output += " [/INST]";
+        insideInst = false;
+      }
+      output += ` ${text} </s>\n`;
+      needsStartAgain = true;
+    }
+  }
+  return output;
+};
+
 export const mapTextContent = (text: string): AnthropicTextContent => {
   return { type: "text", text };
 };

packages/core/CHANGELOG.md

@@ -0,0 +1,17 @@
+# @llamaindex/core
+
+## 0.0.3
+
+### Patch Changes
+
+- f326ab8: chore: bump version
+- Updated dependencies [f326ab8]
+  - @llamaindex/env@0.1.8
+
+## 0.0.2
+
+### Patch Changes
+
+- f10b41d: fix: release files
+- Updated dependencies [41fe871]
+  - @llamaindex/env@0.1.7

packages/core/package.json

@@ -0,0 +1,85 @@
+{
+  "name": "@llamaindex/core",
+  "type": "module",
+  "version": "0.0.3",
+  "description": "LlamaIndex Core Module",
+  "exports": {
+    "./llms": {
+      "require": {
+        "types": "./dist/llms/index.d.cts",
+        "default": "./dist/llms/index.cjs"
+      },
+      "import": {
+        "types": "./dist/llms/index.d.ts",
+        "default": "./dist/llms/index.js"
+      },
+      "default": {
+        "types": "./dist/llms/index.d.ts",
+        "default": "./dist/llms/index.js"
+      }
+    },
+    "./decorator": {
+      "require": {
+        "types": "./dist/decorator/index.d.cts",
+        "default": "./dist/decorator/index.cjs"
+      },
+      "import": {
+        "types": "./dist/decorator/index.d.ts",
+        "default": "./dist/decorator/index.js"
+      },
+      "default": {
+        "types": "./dist/decorator/index.d.ts",
+        "default": "./dist/decorator/index.js"
+      }
+    },
+    "./global": {
+      "require": {
+        "types": "./dist/global/index.d.cts",
+        "default": "./dist/global/index.cjs"
+      },
+      "import": {
+        "types": "./dist/global/index.d.ts",
+        "default": "./dist/global/index.js"
+      },
+      "default": {
+        "types": "./dist/global/index.d.ts",
+        "default": "./dist/global/index.js"
+      }
+    },
+    "./schema": {
+      "require": {
+        "types": "./dist/schema/index.d.cts",
+        "default": "./dist/schema/index.cjs"
+      },
+      "import": {
+        "types": "./dist/schema/index.d.ts",
+        "default": "./dist/schema/index.js"
+      },
+      "default": {
+        "types": "./dist/schema/index.d.ts",
+        "default": "./dist/schema/index.js"
+      }
+    }
+  },
+  "files": [
+    "dist"
+  ],
+  "scripts": {
+    "dev": "bunchee --watch",
+    "build": "bunchee"
+  },
+  "repository": {
+    "type": "git",
+    "directory": "packages/core",
+    "url": "https://github.com/himself65/LlamaIndexTS.git"
+  },
+  "devDependencies": {
+    "ajv": "^8.16.0",
+    "bunchee": "^5.2.1"
+  },
+  "dependencies": {
+    "@llamaindex/env": "workspace:*",
+    "@types/node": "^20.14.9",
+    "zod": "^3.23.8"
+  }
+}

packages/core/src/decorator/index.ts

@@ -1,11 +1,11 @@
 import { getEnv } from "@llamaindex/env";
-import type { BaseNode } from "../../Node.js";
-import { getChunkSize } from "../settings/chunk-size.js";
+import { Settings } from "../global";
+import type { BaseNode } from "../schema/node";
 
 const emitOnce = false;
 
 export function chunkSizeCheck<
-  This extends BaseNode,
+  This extends { id_: string },
   Args extends any[],
   Return,
 >(
@@ -17,7 +17,7 @@ export function chunkSizeCheck<
 ) {
   return function (this: This, ...args: Args) {
     const content = contentGetter.call(this, ...args);
-    const chunkSize = getChunkSize();
+    const chunkSize = Settings.chunkSize;
     const enableChunkSizeCheck = getEnv("ENABLE_CHUNK_SIZE_CHECK") === "true";
     if (
       enableChunkSizeCheck &&
@@ -25,7 +25,7 @@ export function chunkSizeCheck<
       content.length > chunkSize
     ) {
       console.warn(
-        `Node (${this.id_}) is larger than chunk size: ${content.length}`,
+        `Node (${this.id_}) is larger than chunk size: ${content.length} > ${chunkSize}`,
       );
       if (!emitOnce) {
         console.warn(

packages/core/src/global/index.ts

@@ -0,0 +1 @@
+export { Settings } from "./settings";

packages/core/src/global/settings.ts

@@ -0,0 +1,17 @@
+import {
+  getChunkSize,
+  setChunkSize,
+  withChunkSize,
+} from "./settings/chunk-size";
+
+export const Settings = {
+  get chunkSize(): number | undefined {
+    return getChunkSize();
+  },
+  set chunkSize(chunkSize: number | undefined) {
+    setChunkSize(chunkSize);
+  },
+  withChunkSize<Result>(chunkSize: number, fn: () => Result): Result {
+    return withChunkSize(chunkSize, fn);
+  },
+};

packages/core/src/global/settings/chunk-size.ts

@@ -1,14 +1,16 @@
 import { AsyncLocalStorage } from "@llamaindex/env";
 
 const chunkSizeAsyncLocalStorage = new AsyncLocalStorage<number | undefined>();
-const globalChunkSize: number | null = null;
+let globalChunkSize: number | null = null;
 
 export function getChunkSize(): number | undefined {
   return globalChunkSize ?? chunkSizeAsyncLocalStorage.getStore();
 }
 
 export function setChunkSize(chunkSize: number | undefined) {
-  chunkSizeAsyncLocalStorage.enterWith(chunkSize);
+  if (chunkSize !== undefined) {
+    globalChunkSize = chunkSize;
+  }
 }
 
 export function withChunkSize<Result>(

packages/core/src/global/type.ts

@@ -0,0 +1,9 @@
+export type UUID = `${string}-${string}-${string}-${string}-${string}`;
+
+export type JSONValue = string | number | boolean | JSONObject | JSONArray;
+
+export type JSONObject = {
+  [key: string]: JSONValue;
+};
+
+export type JSONArray = Array<JSONValue>;

packages/core/src/llms/index.ts

@@ -0,0 +1,31 @@
+export type {
+  BaseTool,
+  BaseToolWithCall,
+  ChatMessage,
+  ChatResponse,
+  ChatResponseChunk,
+  CompletionResponse,
+  LLM,
+  LLMChat,
+  LLMChatParamsBase,
+  LLMChatParamsNonStreaming,
+  LLMChatParamsStreaming,
+  LLMCompletionParamsBase,
+  LLMCompletionParamsNonStreaming,
+  LLMCompletionParamsStreaming,
+  LLMMetadata,
+  MessageContent,
+  MessageContentDetail,
+  MessageContentImageDetail,
+  MessageContentTextDetail,
+  MessageType,
+  PartialToolCall,
+  TextChatMessage,
+  ToolCall,
+  ToolCallLLMMessageOptions,
+  ToolCallOptions,
+  ToolMetadata,
+  ToolOutput,
+  ToolResult,
+  ToolResultOptions,
+} from "./type";

packages/core/src/llms/type.ts

@@ -0,0 +1,245 @@
+import type { Tokenizers } from "@llamaindex/env";
+import type { JSONSchemaType } from "ajv";
+import type { JSONObject, JSONValue } from "../global/type";
+
+/**
+ * @internal
+ */
+export interface LLMChat<
+  AdditionalChatOptions extends object = object,
+  AdditionalMessageOptions extends object = object,
+> {
+  chat(
+    params:
+      | LLMChatParamsStreaming<AdditionalChatOptions>
+      | LLMChatParamsNonStreaming<AdditionalChatOptions>,
+  ): Promise<
+    | ChatResponse<AdditionalMessageOptions>
+    | AsyncIterable<ChatResponseChunk<AdditionalMessageOptions>>
+  >;
+}
+
+/**
+ * Unified language model interface
+ */
+export interface LLM<
+  AdditionalChatOptions extends object = object,
+  AdditionalMessageOptions extends object = object,
+> extends LLMChat<AdditionalChatOptions> {
+  metadata: LLMMetadata;
+  /**
+   * Get a chat response from the LLM
+   */
+  chat(
+    params: LLMChatParamsStreaming<
+      AdditionalChatOptions,
+      AdditionalMessageOptions
+    >,
+  ): Promise<AsyncIterable<ChatResponseChunk>>;
+  chat(
+    params: LLMChatParamsNonStreaming<
+      AdditionalChatOptions,
+      AdditionalMessageOptions
+    >,
+  ): Promise<ChatResponse<AdditionalMessageOptions>>;
+
+  /**
+   * Get a prompt completion from the LLM
+   */
+  complete(
+    params: LLMCompletionParamsStreaming,
+  ): Promise<AsyncIterable<CompletionResponse>>;
+  complete(
+    params: LLMCompletionParamsNonStreaming,
+  ): Promise<CompletionResponse>;
+}
+
+export type MessageType = "user" | "assistant" | "system" | "memory";
+
+export type TextChatMessage<AdditionalMessageOptions extends object = object> =
+  {
+    content: string;
+    role: MessageType;
+    options?: undefined | AdditionalMessageOptions;
+  };
+
+export type ChatMessage<AdditionalMessageOptions extends object = object> = {
+  content: MessageContent;
+  role: MessageType;
+  options?: undefined | AdditionalMessageOptions;
+};
+
+export interface ChatResponse<
+  AdditionalMessageOptions extends object = object,
+> {
+  message: ChatMessage<AdditionalMessageOptions>;
+  /**
+   * Raw response from the LLM
+   *
+   * If LLM response an iterable of chunks, this will be an array of those chunks
+   */
+  raw: object | null;
+}
+
+export type ChatResponseChunk<
+  AdditionalMessageOptions extends object = object,
+> = {
+  raw: object | null;
+  delta: string;
+  options?: undefined | AdditionalMessageOptions;
+};
+
+export interface CompletionResponse {
+  text: string;
+  /**
+   * Raw response from the LLM
+   *
+   * It's possible that this is `null` if the LLM response an iterable of chunks
+   */
+  raw: object | null;
+}
+
+export type LLMMetadata = {
+  model: string;
+  temperature: number;
+  topP: number;
+  maxTokens?: number;
+  contextWindow: number;
+  tokenizer: Tokenizers | undefined;
+};
+
+export interface LLMChatParamsBase<
+  AdditionalChatOptions extends object = object,
+  AdditionalMessageOptions extends object = object,
+> {
+  messages: ChatMessage<AdditionalMessageOptions>[];
+  additionalChatOptions?: AdditionalChatOptions;
+  tools?: BaseTool[];
+}
+
+export interface LLMChatParamsStreaming<
+  AdditionalChatOptions extends object = object,
+  AdditionalMessageOptions extends object = object,
+> extends LLMChatParamsBase<AdditionalChatOptions, AdditionalMessageOptions> {
+  stream: true;
+}
+
+export interface LLMChatParamsNonStreaming<
+  AdditionalChatOptions extends object = object,
+  AdditionalMessageOptions extends object = object,
+> extends LLMChatParamsBase<AdditionalChatOptions, AdditionalMessageOptions> {
+  stream?: false;
+}
+
+export interface LLMCompletionParamsBase {
+  prompt: MessageContent;
+}
+
+export interface LLMCompletionParamsStreaming extends LLMCompletionParamsBase {
+  stream: true;
+}
+
+export interface LLMCompletionParamsNonStreaming
+  extends LLMCompletionParamsBase {
+  stream?: false | null;
+}
+
+export type MessageContentTextDetail = {
+  type: "text";
+  text: string;
+};
+
+export type MessageContentImageDetail = {
+  type: "image_url";
+  image_url: { url: string };
+};
+
+export type MessageContentDetail =
+  | MessageContentTextDetail
+  | MessageContentImageDetail;
+
+/**
+ * Extended type for the content of a message that allows for multi-modal messages.
+ */
+export type MessageContent = string | MessageContentDetail[];
+
+export type ToolCall = {
+  name: string;
+  input: JSONObject;
+  id: string;
+};
+
+// happened in streaming response, the tool call is not ready yet
+export type PartialToolCall = {
+  name: string;
+  id: string;
+  // input is not ready yet, JSON.parse(input) will throw an error
+  input: string;
+};
+
+export type ToolResult = {
+  id: string;
+  result: string;
+  isError: boolean;
+};
+
+export type ToolCallOptions = {
+  toolCall: (ToolCall | PartialToolCall)[];
+};
+
+export type ToolResultOptions = {
+  toolResult: ToolResult;
+};
+
+export type ToolCallLLMMessageOptions =
+  | ToolResultOptions
+  | ToolCallOptions
+  | {};
+
+type Known =
+  | { [key: string]: Known }
+  | [Known, ...Known[]]
+  | Known[]
+  | number
+  | string
+  | boolean
+  | null;
+
+export type ToolMetadata<
+  Parameters extends Record<string, unknown> = Record<string, unknown>,
+> = {
+  description: string;
+  name: string;
+  /**
+   * OpenAI uses JSON Schema to describe the parameters that a tool can take.
+   * @link https://json-schema.org/understanding-json-schema
+   */
+  parameters?: Parameters;
+};
+
+/**
+ * Simple Tool interface. Likely to change.
+ */
+export interface BaseTool<Input = any> {
+  /**
+   * This could be undefined if the implementation is not provided,
+   *  which might be the case when communicating with a llm.
+   *
+   * @return {JSONValue | Promise<JSONValue>} The output of the tool.
+   */
+  call?: (input: Input) => JSONValue | Promise<JSONValue>;
+  metadata: // if user input any, we cannot check the schema
+  Input extends Known ? ToolMetadata<JSONSchemaType<Input>> : ToolMetadata;
+}
+
+export type BaseToolWithCall<Input = any> = Omit<BaseTool<Input>, "call"> & {
+  call: NonNullable<Pick<BaseTool<Input>, "call">["call"]>;
+};
+
+export type ToolOutput = {
+  tool: BaseTool | undefined;
+  // all of existing function calling LLMs only support object input
+  input: JSONObject;
+  output: JSONValue;
+  isError: boolean;
+};

packages/core/src/schema/index.ts

@@ -0,0 +1,2 @@
+export * from "./node";
+export * from "./zod";

packages/core/src/schema/node.ts

@@ -0,0 +1,452 @@
+import { createSHA256, path, randomUUID } from "@llamaindex/env";
+import { chunkSizeCheck, lazyInitHash } from "../decorator";
+
+export enum NodeRelationship {
+  SOURCE = "SOURCE",
+  PREVIOUS = "PREVIOUS",
+  NEXT = "NEXT",
+  PARENT = "PARENT",
+  CHILD = "CHILD",
+}
+
+export enum ObjectType {
+  TEXT = "TEXT",
+  IMAGE = "IMAGE",
+  INDEX = "INDEX",
+  DOCUMENT = "DOCUMENT",
+  IMAGE_DOCUMENT = "IMAGE_DOCUMENT",
+}
+
+export enum MetadataMode {
+  ALL = "ALL",
+  EMBED = "EMBED",
+  LLM = "LLM",
+  NONE = "NONE",
+}
+
+export type Metadata = Record<string, any>;
+
+export interface RelatedNodeInfo<T extends Metadata = Metadata> {
+  nodeId: string;
+  nodeType?: ObjectType;
+  metadata: T;
+  hash?: string;
+}
+
+export type RelatedNodeType<T extends Metadata = Metadata> =
+  | RelatedNodeInfo<T>
+  | RelatedNodeInfo<T>[];
+
+export type BaseNodeParams<T extends Metadata = Metadata> = {
+  id_?: string;
+  metadata?: T;
+  excludedEmbedMetadataKeys?: string[];
+  excludedLlmMetadataKeys?: string[];
+  relationships?: Partial<Record<NodeRelationship, RelatedNodeType<T>>>;
+  hash?: string;
+  embedding?: number[];
+};
+
+/**
+ * Generic abstract class for retrievable nodes
+ */
+export abstract class BaseNode<T extends Metadata = Metadata> {
+  /**
+   * The unique ID of the Node/Document. The trailing underscore is here
+   * to avoid collisions with the id keyword in Python.
+   *
+   * Set to a UUID by default.
+   */
+  id_: string;
+  embedding?: number[];
+
+  // Metadata fields
+  metadata: T;
+  excludedEmbedMetadataKeys: string[];
+  excludedLlmMetadataKeys: string[];
+  relationships: Partial<Record<NodeRelationship, RelatedNodeType<T>>>;
+
+  @lazyInitHash
+  accessor hash: string = "";
+
+  protected constructor(init?: BaseNodeParams<T>) {
+    const {
+      id_,
+      metadata,
+      excludedEmbedMetadataKeys,
+      excludedLlmMetadataKeys,
+      relationships,
+      hash,
+      embedding,
+    } = init || {};
+    this.id_ = id_ ?? randomUUID();
+    this.metadata = metadata ?? ({} as T);
+    this.excludedEmbedMetadataKeys = excludedEmbedMetadataKeys ?? [];
+    this.excludedLlmMetadataKeys = excludedLlmMetadataKeys ?? [];
+    this.relationships = relationships ?? {};
+    this.embedding = embedding;
+  }
+
+  abstract get type(): ObjectType;
+
+  abstract getContent(metadataMode: MetadataMode): string;
+
+  abstract getMetadataStr(metadataMode: MetadataMode): string;
+
+  // todo: set value as a generic type
+  abstract setContent(value: unknown): void;
+
+  get sourceNode(): RelatedNodeInfo<T> | undefined {
+    const relationship = this.relationships[NodeRelationship.SOURCE];
+
+    if (Array.isArray(relationship)) {
+      throw new Error("Source object must be a single RelatedNodeInfo object");
+    }
+
+    return relationship;
+  }
+
+  get prevNode(): RelatedNodeInfo<T> | undefined {
+    const relationship = this.relationships[NodeRelationship.PREVIOUS];
+
+    if (Array.isArray(relationship)) {
+      throw new Error(
+        "Previous object must be a single RelatedNodeInfo object",
+      );
+    }
+
+    return relationship;
+  }
+
+  get nextNode(): RelatedNodeInfo<T> | undefined {
+    const relationship = this.relationships[NodeRelationship.NEXT];
+
+    if (Array.isArray(relationship)) {
+      throw new Error("Next object must be a single RelatedNodeInfo object");
+    }
+
+    return relationship;
+  }
+
+  get parentNode(): RelatedNodeInfo<T> | undefined {
+    const relationship = this.relationships[NodeRelationship.PARENT];
+
+    if (Array.isArray(relationship)) {
+      throw new Error("Parent object must be a single RelatedNodeInfo object");
+    }
+
+    return relationship;
+  }
+
+  get childNodes(): RelatedNodeInfo<T>[] | undefined {
+    const relationship = this.relationships[NodeRelationship.CHILD];
+
+    if (!Array.isArray(relationship)) {
+      throw new Error(
+        "Child object must be a an array of RelatedNodeInfo objects",
+      );
+    }
+
+    return relationship;
+  }
+
+  abstract generateHash(): string;
+
+  getEmbedding(): number[] {
+    if (this.embedding === undefined) {
+      throw new Error("Embedding not set");
+    }
+
+    return this.embedding;
+  }
+
+  asRelatedNodeInfo(): RelatedNodeInfo<T> {
+    return {
+      nodeId: this.id_,
+      metadata: this.metadata,
+      hash: this.hash,
+    };
+  }
+
+  /**
+   * Called by built in JSON.stringify (see https://javascript.info/json)
+   * Properties are read-only as they are not deep-cloned (not necessary for stringification).
+   * @see toMutableJSON - use to return a mutable JSON instead
+   */
+  toJSON(): Record<string, any> {
+    return {
+      ...this,
+      type: this.type,
+      // hash is an accessor property, so it's not included in the rest operator
+      hash: this.hash,
+    };
+  }
+
+  clone(): BaseNode {
+    return jsonToNode(this.toMutableJSON()) as BaseNode;
+  }
+
+  /**
+   * Converts the object to a JSON representation.
+   * Properties can be safely modified as a deep clone of the properties are created.
+   * @return {Record<string, any>} - The JSON representation of the object.
+   */
+  toMutableJSON(): Record<string, any> {
+    return structuredClone(this.toJSON());
+  }
+}
+
+export type TextNodeParams<T extends Metadata = Metadata> =
+  BaseNodeParams<T> & {
+    text?: string;
+    textTemplate?: string;
+    startCharIdx?: number;
+    endCharIdx?: number;
+    metadataSeparator?: string;
+  };
+
+/**
+ * TextNode is the default node type for text. Most common node type in LlamaIndex.TS
+ */
+export class TextNode<T extends Metadata = Metadata> extends BaseNode<T> {
+  text: string;
+  textTemplate: string;
+
+  startCharIdx?: number;
+  endCharIdx?: number;
+  // textTemplate: NOTE write your own formatter if needed
+  // metadataTemplate: NOTE write your own formatter if needed
+  metadataSeparator: string;
+
+  constructor(init: TextNodeParams<T> = {}) {
+    super(init);
+    const { text, textTemplate, startCharIdx, endCharIdx, metadataSeparator } =
+      init;
+    this.text = text ?? "";
+    this.textTemplate = textTemplate ?? "";
+    if (startCharIdx) {
+      this.startCharIdx = startCharIdx;
+    }
+    if (endCharIdx) {
+      this.endCharIdx = endCharIdx;
+    }
+    this.metadataSeparator = metadataSeparator ?? "\n";
+  }
+
+  /**
+   * Generate a hash of the text node.
+   * The ID is not part of the hash as it can change independent of content.
+   * @returns
+   */
+  generateHash() {
+    const hashFunction = createSHA256();
+    hashFunction.update(`type=${this.type}`);
+    hashFunction.update(
+      `startCharIdx=${this.startCharIdx} endCharIdx=${this.endCharIdx}`,
+    );
+    hashFunction.update(this.getContent(MetadataMode.ALL));
+    return hashFunction.digest();
+  }
+
+  get type() {
+    return ObjectType.TEXT;
+  }
+
+  @chunkSizeCheck
+  getContent(metadataMode: MetadataMode = MetadataMode.NONE): string {
+    const metadataStr = this.getMetadataStr(metadataMode).trim();
+    return `${metadataStr}\n\n${this.text}`.trim();
+  }
+
+  getMetadataStr(metadataMode: MetadataMode): string {
+    if (metadataMode === MetadataMode.NONE) {
+      return "";
+    }
+
+    const usableMetadataKeys = new Set(Object.keys(this.metadata).sort());
+    if (metadataMode === MetadataMode.LLM) {
+      for (const key of this.excludedLlmMetadataKeys) {
+        usableMetadataKeys.delete(key);
+      }
+    } else if (metadataMode === MetadataMode.EMBED) {
+      for (const key of this.excludedEmbedMetadataKeys) {
+        usableMetadataKeys.delete(key);
+      }
+    }
+
+    return [...usableMetadataKeys]
+      .map((key) => `${key}: ${this.metadata[key]}`)
+      .join(this.metadataSeparator);
+  }
+
+  setContent(value: string) {
+    this.text = value;
+    this.hash = this.generateHash();
+  }
+
+  getNodeInfo() {
+    return { start: this.startCharIdx, end: this.endCharIdx };
+  }
+
+  getText() {
+    return this.getContent(MetadataMode.NONE);
+  }
+}
+
+export type IndexNodeParams<T extends Metadata = Metadata> =
+  TextNodeParams<T> & {
+    indexId: string;
+  };
+
+export class IndexNode<T extends Metadata = Metadata> extends TextNode<T> {
+  indexId: string;
+
+  constructor(init?: IndexNodeParams<T>) {
+    super(init);
+    const { indexId } = init || {};
+    this.indexId = indexId ?? "";
+  }
+
+  get type() {
+    return ObjectType.INDEX;
+  }
+}
+
+/**
+ * A document is just a special text node with a docId.
+ */
+export class Document<T extends Metadata = Metadata> extends TextNode<T> {
+  constructor(init?: TextNodeParams<T>) {
+    super(init);
+  }
+
+  get type() {
+    return ObjectType.DOCUMENT;
+  }
+}
+
+export function jsonToNode(json: any, type?: ObjectType) {
+  if (!json.type && !type) {
+    throw new Error("Node type not found");
+  }
+  const nodeType = type || json.type;
+
+  switch (nodeType) {
+    case ObjectType.TEXT:
+      return new TextNode(json);
+    case ObjectType.INDEX:
+      return new IndexNode(json);
+    case ObjectType.DOCUMENT:
+      return new Document(json);
+    case ObjectType.IMAGE_DOCUMENT:
+      return new ImageDocument(json);
+    default:
+      throw new Error(`Invalid node type: ${nodeType}`);
+  }
+}
+
+export type ImageType = string | Blob | URL;
+
+export type ImageNodeParams<T extends Metadata = Metadata> =
+  TextNodeParams<T> & {
+    image: ImageType;
+  };
+
+export class ImageNode<T extends Metadata = Metadata> extends TextNode<T> {
+  image: ImageType; // image as blob
+
+  constructor(init: ImageNodeParams<T>) {
+    super(init);
+    const { image } = init;
+    this.image = image;
+  }
+
+  get type() {
+    return ObjectType.IMAGE;
+  }
+
+  getUrl(): URL {
+    // id_ stores the relative path, convert it to the URL of the file
+    const absPath = path.resolve(this.id_);
+    return new URL(`file://${absPath}`);
+  }
+
+  // Calculates the image part of the hash
+  private generateImageHash() {
+    const hashFunction = createSHA256();
+
+    if (this.image instanceof Blob) {
+      // TODO: ideally we should use the blob's content to calculate the hash:
+      // hashFunction.update(new Uint8Array(await this.image.arrayBuffer()));
+      // as this is async, we're using the node's ID for the time being
+      hashFunction.update(this.id_);
+    } else if (this.image instanceof URL) {
+      hashFunction.update(this.image.toString());
+    } else if (typeof this.image === "string") {
+      hashFunction.update(this.image);
+    } else {
+      throw new Error(
+        `Unknown image type: ${typeof this.image}. Can't calculate hash`,
+      );
+    }
+
+    return hashFunction.digest();
+  }
+
+  generateHash() {
+    const hashFunction = createSHA256();
+    // calculates hash based on hash of both components (image and text)
+    hashFunction.update(super.generateHash());
+    hashFunction.update(this.generateImageHash());
+
+    return hashFunction.digest();
+  }
+}
+
+export class ImageDocument<T extends Metadata = Metadata> extends ImageNode<T> {
+  constructor(init: ImageNodeParams<T>) {
+    super(init);
+  }
+
+  get type() {
+    return ObjectType.IMAGE_DOCUMENT;
+  }
+}
+
+/**
+ * A node with a similarity score
+ */
+export interface NodeWithScore<T extends Metadata = Metadata> {
+  node: BaseNode<T>;
+  score?: number;
+}
+
+export enum ModalityType {
+  TEXT = "TEXT",
+  IMAGE = "IMAGE",
+}
+
+type NodesByType = {
+  [P in ModalityType]?: BaseNode[];
+};
+
+export function splitNodesByType(nodes: BaseNode[]): NodesByType {
+  const result: NodesByType = {};
+
+  for (const node of nodes) {
+    let type: ModalityType;
+    if (node instanceof ImageNode) {
+      type = ModalityType.IMAGE;
+    } else if (node instanceof TextNode) {
+      type = ModalityType.TEXT;
+    } else {
+      throw new Error(`Unknown node type: ${node.type}`);
+    }
+    if (type in result) {
+      result[type]?.push(node);
+    } else {
+      result[type] = [node];
+    }
+  }
+  return result;
+}

packages/core/src/schema/zod.ts

@@ -0,0 +1,18 @@
+import { z } from "zod";
+
+export const anyFunctionSchema = z.function(z.tuple([]).rest(z.any()), z.any());
+
+export const toolMetadataSchema = z.object({
+  description: z.string(),
+  name: z.string(),
+  parameters: z.record(z.any()),
+});
+
+export const baseToolSchema = z.object({
+  call: anyFunctionSchema.optional(),
+  metadata: toolMetadataSchema,
+});
+
+export const baseToolWithCallSchema = baseToolSchema.extend({
+  call: z.function(),
+});

packages/core/tests/decorator.test.ts

@@ -0,0 +1,25 @@
+import { Settings } from "@llamaindex/core/global";
+import { TextNode } from "@llamaindex/core/schema";
+import { env } from "process";
+import { afterEach, describe, expect, test, vi } from "vitest";
+describe("chunkSizeCheck", () => {
+  afterEach(() => {
+    Settings.chunkSize = undefined;
+    env.ENABLE_CHUNK_SIZE_CHECK = undefined;
+  });
+
+  test("should warn when content is larger than chunk size", () => {
+    env.ENABLE_CHUNK_SIZE_CHECK = "true";
+
+    let message = "";
+    const consoleMock = vi
+      .spyOn(console, "warn")
+      .mockImplementation((msg) => (message += msg + "\n"));
+
+    Settings.chunkSize = 0;
+    const node = new TextNode();
+    expect(message).toEqual("");
+    node.setContent("a".repeat(1024));
+    expect(message).toContain("is larger than chunk size");
+  });
+});

packages/core/tests/package.json

@@ -0,0 +1,12 @@
+{
+  "name": "@llamaindex/core-tests",
+  "private": true,
+  "type": "module",
+  "scripts": {
+    "test": "vitest"
+  },
+  "devDependencies": {
+    "@llamaindex/core": "workspace:*",
+    "vitest": "^1.6.0"
+  }
+}

packages/core/tests/schema-node.test.ts

@@ -1,4 +1,4 @@
-import { Document, TextNode } from "llamaindex/Node";
+import { Document, TextNode } from "@llamaindex/core/schema";
 import { beforeEach, describe, expect, test } from "vitest";
 
 describe("Document", () => {
@@ -50,7 +50,6 @@ describe("TextNode", () => {
       `
       {
         "embedding": undefined,
-        "endCharIdx": undefined,
         "excludedEmbedMetadataKeys": [],
         "excludedLlmMetadataKeys": [],
         "hash": "Z6SWgFPlalaeblMGQGw0KS3qKgmZdEWXKfzEp/K+QN0=",
@@ -62,7 +61,6 @@ describe("TextNode", () => {
         "metadataSeparator": "
       ",
         "relationships": {},
-        "startCharIdx": undefined,
         "text": "Hello World",
         "textTemplate": "",
         "type": "TEXT",

packages/core/tests/tsconfig.json

@@ -0,0 +1,18 @@
+{
+  "extends": "../../../tsconfig.json",
+  "compilerOptions": {
+    "outDir": "./lib",
+    "module": "node16",
+    "moduleResolution": "node16",
+    "target": "ESNext"
+  },
+  "include": ["./**/*.ts"],
+  "references": [
+    {
+      "path": "../tsconfig.json"
+    },
+    {
+      "path": "../../env/tsconfig.json"
+    }
+  ]
+}

packages/core/tsconfig.json

@@ -0,0 +1,15 @@
+{
+  "extends": "../../tsconfig.json",
+  "compilerOptions": {
+    "rootDir": "./src",
+    "outDir": "./dist/type",
+    "tsBuildInfoFile": "./dist/.tsbuildinfo",
+    "emitDeclarationOnly": true,
+    "moduleResolution": "Bundler",
+    "skipLibCheck": true,
+    "strict": true,
+    "types": ["node"]
+  },
+  "include": ["./src"],
+  "exclude": ["node_modules"]
+}

packages/env/CHANGELOG.md

@@ -1,5 +1,17 @@
 # @llamaindex/env
 
+## 0.1.8
+
+### Patch Changes
+
+- f326ab8: chore: bump version
+
+## 0.1.7
+
+### Patch Changes
+
+- 41fe871: Add support for azure dynamic session tool
+
 ## 0.1.6
 
 ### Patch Changes

packages/env/jsr.json

@@ -1,10 +0,0 @@
-{
-  "name": "@llamaindex/env",
-  "version": "0.1.6",
-  "exports": {
-    ".": "./src/index.edge-light.ts"
-  },
-  "publish": {
-    "include": ["LICENSE", "README.md", "src/**/*", "jsr.json"]
-  }
-}

packages/env/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@llamaindex/env",
   "description": "environment wrapper, supports all JS environment including node, deno, bun, edge runtime, and cloudflare worker",
-  "version": "0.1.6",
+  "version": "0.1.8",
   "type": "module",
   "types": "dist/type/index.d.ts",
   "main": "dist/cjs/index.js",

packages/env/src/fs/node.ts

@@ -5,6 +5,7 @@
  *
  * @module
  */
+import { createWriteStream } from "node:fs";
 import fs from "node:fs/promises";
 
-export { fs };
+export { createWriteStream, fs };

packages/env/src/index.ts

@@ -15,12 +15,14 @@ import { ok } from "node:assert";
 import { createHash, randomUUID } from "node:crypto";
 import { EOL } from "node:os";
 import path from "node:path";
+import { Readable } from "node:stream";
 import {
   ReadableStream,
   TransformStream,
   WritableStream,
 } from "node:stream/web";
-import { fs } from "./fs/node.js";
+import { fileURLToPath } from "node:url";
+import { createWriteStream, fs } from "./fs/node.js";
 import type { SHA256 } from "./polyfill.js";
 
 export function createSHA256(): SHA256 {
@@ -38,11 +40,14 @@ export function createSHA256(): SHA256 {
 export { Tokenizers, tokenizers, type Tokenizer } from "./tokenizers/node.js";
 export { AsyncLocalStorage, CustomEvent, getEnv, setEnvs } from "./utils.js";
 export {
+  createWriteStream,
   EOL,
+  fileURLToPath,
   fs,
   ok,
   path,
   randomUUID,
+  Readable,
   ReadableStream,
   TransformStream,
   WritableStream,

packages/experimental/CHANGELOG.md

@@ -1,5 +1,71 @@
 # @llamaindex/experimental
 
+## 0.0.47
+
+### Patch Changes
+
+- Updated dependencies [f326ab8]
+  - llamaindex@0.4.12
+
+## 0.0.46
+
+### Patch Changes
+
+- Updated dependencies [8bf5b4a]
+  - llamaindex@0.4.11
+
+## 0.0.45
+
+### Patch Changes
+
+- Updated dependencies [7dce3d2]
+  - llamaindex@0.4.10
+
+## 0.0.44
+
+### Patch Changes
+
+- Updated dependencies [3a96a48]
+  - llamaindex@0.4.9
+
+## 0.0.43
+
+### Patch Changes
+
+- Updated dependencies [83ebdfb]
+  - llamaindex@0.4.8
+
+## 0.0.42
+
+### Patch Changes
+
+- Updated dependencies [41fe871]
+- Updated dependencies [321c39d]
+- Updated dependencies [f7f1af0]
+  - llamaindex@0.4.7
+
+## 0.0.41
+
+### Patch Changes
+
+- Updated dependencies [1feb23b]
+- Updated dependencies [08c55ec]
+  - llamaindex@0.4.6
+
+## 0.0.40
+
+### Patch Changes
+
+- Updated dependencies [6c3e5d0]
+  - llamaindex@0.4.5
+
+## 0.0.39
+
+### Patch Changes
+
+- Updated dependencies [42eb73a]
+  - llamaindex@0.4.4
+
 ## 0.0.38
 
 ### Patch Changes

packages/experimental/jsr.json

@@ -1,8 +0,0 @@
-{
-  "name": "@llamaindex/experimental",
-  "version": "0.0.5",
-  "exports": {
-    ".": "./src/index.ts",
-    "./type": "./src/type.ts"
-  }
-}

packages/experimental/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@llamaindex/experimental",
   "description": "Experimental package for LlamaIndexTS",
-  "version": "0.0.38",
+  "version": "0.0.47",
   "type": "module",
   "types": "dist/type/index.d.ts",
   "main": "dist/cjs/index.js",

packages/llamaindex/CHANGELOG.md

@@ -1,5 +1,72 @@
 # llamaindex
 
+## 0.4.12
+
+### Patch Changes
+
+- f326ab8: chore: bump version
+- Updated dependencies [f326ab8]
+  - @llamaindex/cloud@0.1.2
+  - @llamaindex/core@0.0.3
+  - @llamaindex/env@0.1.8
+
+## 0.4.11
+
+### Patch Changes
+
+- 8bf5b4a: fix: llama parse input spreadsheet
+
+## 0.4.10
+
+### Patch Changes
+
+- 7dce3d2: fix: disable External Filters for Gemini
+
+## 0.4.9
+
+### Patch Changes
+
+- 3a96a48: fix: anthroipic image input
+
+## 0.4.8
+
+### Patch Changes
+
+- 83ebdfb: fix: next.js build error
+
+## 0.4.7
+
+### Patch Changes
+
+- 41fe871: Add support for azure dynamic session tool
+- 321c39d: fix: generate api as class
+- f7f1af0: fix: throw error when no pipeline found
+- Updated dependencies [41fe871]
+- Updated dependencies [f10b41d]
+- Updated dependencies [321c39d]
+  - @llamaindex/env@0.1.7
+  - @llamaindex/core@0.0.2
+  - @llamaindex/cloud@0.1.1
+
+## 0.4.6
+
+### Patch Changes
+
+- 1feb23b: feat: Gemini tool calling for agent support
+- 08c55ec: Add metadata to PDFs and use Uint8Array for readers content
+
+## 0.4.5
+
+### Patch Changes
+
+- 6c3e5d0: fix: switch to correct reference for a static function
+
+## 0.4.4
+
+### Patch Changes
+
+- 42eb73a: Fix IngestionPipeline not working without vectorStores
+
 ## 0.4.3
 
 ### Patch Changes

packages/llamaindex/e2e/examples/cloudflare-worker-agent/CHANGELOG.md

@@ -1,5 +1,71 @@
 # @llamaindex/cloudflare-worker-agent-test
 
+## 0.0.31
+
+### Patch Changes
+
+- Updated dependencies [f326ab8]
+  - llamaindex@0.4.12
+
+## 0.0.30
+
+### Patch Changes
+
+- Updated dependencies [8bf5b4a]
+  - llamaindex@0.4.11
+
+## 0.0.29
+
+### Patch Changes
+
+- Updated dependencies [7dce3d2]
+  - llamaindex@0.4.10
+
+## 0.0.28
+
+### Patch Changes
+
+- Updated dependencies [3a96a48]
+  - llamaindex@0.4.9
+
+## 0.0.27
+
+### Patch Changes
+
+- Updated dependencies [83ebdfb]
+  - llamaindex@0.4.8
+
+## 0.0.26
+
+### Patch Changes
+
+- Updated dependencies [41fe871]
+- Updated dependencies [321c39d]
+- Updated dependencies [f7f1af0]
+  - llamaindex@0.4.7
+
+## 0.0.25
+
+### Patch Changes
+
+- Updated dependencies [1feb23b]
+- Updated dependencies [08c55ec]
+  - llamaindex@0.4.6
+
+## 0.0.24
+
+### Patch Changes
+
+- Updated dependencies [6c3e5d0]
+  - llamaindex@0.4.5
+
+## 0.0.23
+
+### Patch Changes
+
+- Updated dependencies [42eb73a]
+  - llamaindex@0.4.4
+
 ## 0.0.22
 
 ### Patch Changes

packages/llamaindex/e2e/examples/cloudflare-worker-agent/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@llamaindex/cloudflare-worker-agent-test",
-  "version": "0.0.22",
+  "version": "0.0.31",
   "type": "module",
   "private": true,
   "scripts": {

packages/llamaindex/e2e/examples/nextjs-agent/CHANGELOG.md

@@ -1,5 +1,71 @@
 # @llamaindex/next-agent-test
 
+## 0.1.31
+
+### Patch Changes
+
+- Updated dependencies [f326ab8]
+  - llamaindex@0.4.12
+
+## 0.1.30
+
+### Patch Changes
+
+- Updated dependencies [8bf5b4a]
+  - llamaindex@0.4.11
+
+## 0.1.29
+
+### Patch Changes
+
+- Updated dependencies [7dce3d2]
+  - llamaindex@0.4.10
+
+## 0.1.28
+
+### Patch Changes
+
+- Updated dependencies [3a96a48]
+  - llamaindex@0.4.9
+
+## 0.1.27
+
+### Patch Changes
+
+- Updated dependencies [83ebdfb]
+  - llamaindex@0.4.8
+
+## 0.1.26
+
+### Patch Changes
+
+- Updated dependencies [41fe871]
+- Updated dependencies [321c39d]
+- Updated dependencies [f7f1af0]
+  - llamaindex@0.4.7
+
+## 0.1.25
+
+### Patch Changes
+
+- Updated dependencies [1feb23b]
+- Updated dependencies [08c55ec]
+  - llamaindex@0.4.6
+
+## 0.1.24
+
+### Patch Changes
+
+- Updated dependencies [6c3e5d0]
+  - llamaindex@0.4.5
+
+## 0.1.23
+
+### Patch Changes
+
+- Updated dependencies [42eb73a]
+  - llamaindex@0.4.4
+
 ## 0.1.22
 
 ### Patch Changes

packages/llamaindex/e2e/examples/nextjs-agent/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@llamaindex/next-agent-test",
-  "version": "0.1.22",
+  "version": "0.1.31",
   "private": true,
   "scripts": {
     "dev": "next dev",

packages/llamaindex/e2e/examples/nextjs-agent/src/actions/index.tsx

@@ -1,7 +1,7 @@
 "use server";
 import { createStreamableUI } from "ai/rsc";
+import type { ChatMessage } from "llamaindex";
 import { OpenAIAgent } from "llamaindex";
-import type { ChatMessage } from "llamaindex/llm/types";
 
 export async function chatWithAgent(
   question: string,

packages/llamaindex/e2e/examples/nextjs-edge-runtime/CHANGELOG.md

@@ -1,5 +1,71 @@
 # test-edge-runtime
 
+## 0.1.30
+
+### Patch Changes
+
+- Updated dependencies [f326ab8]
+  - llamaindex@0.4.12
+
+## 0.1.29
+
+### Patch Changes
+
+- Updated dependencies [8bf5b4a]
+  - llamaindex@0.4.11
+
+## 0.1.28
+
+### Patch Changes
+
+- Updated dependencies [7dce3d2]
+  - llamaindex@0.4.10
+
+## 0.1.27
+
+### Patch Changes
+
+- Updated dependencies [3a96a48]
+  - llamaindex@0.4.9
+
+## 0.1.26
+
+### Patch Changes
+
+- Updated dependencies [83ebdfb]
+  - llamaindex@0.4.8
+
+## 0.1.25
+
+### Patch Changes
+
+- Updated dependencies [41fe871]
+- Updated dependencies [321c39d]
+- Updated dependencies [f7f1af0]
+  - llamaindex@0.4.7
+
+## 0.1.24
+
+### Patch Changes
+
+- Updated dependencies [1feb23b]
+- Updated dependencies [08c55ec]
+  - llamaindex@0.4.6
+
+## 0.1.23
+
+### Patch Changes
+
+- Updated dependencies [6c3e5d0]
+  - llamaindex@0.4.5
+
+## 0.1.22
+
+### Patch Changes
+
+- Updated dependencies [42eb73a]
+  - llamaindex@0.4.4
+
 ## 0.1.21
 
 ### Patch Changes

packages/llamaindex/e2e/examples/nextjs-edge-runtime/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@llamaindex/nextjs-edge-runtime-test",
-  "version": "0.1.21",
+  "version": "0.1.30",
   "private": true,
   "scripts": {
     "dev": "next dev",

packages/llamaindex/e2e/examples/nextjs-node-runtime/CHANGELOG.md

@@ -1,5 +1,72 @@
 # @llamaindex/next-node-runtime
 
+## 0.0.12
+
+### Patch Changes
+
+- Updated dependencies [f326ab8]
+  - llamaindex@0.4.12
+
+## 0.0.11
+
+### Patch Changes
+
+- Updated dependencies [8bf5b4a]
+  - llamaindex@0.4.11
+
+## 0.0.10
+
+### Patch Changes
+
+- Updated dependencies [7dce3d2]
+  - llamaindex@0.4.10
+
+## 0.0.9
+
+### Patch Changes
+
+- Updated dependencies [3a96a48]
+  - llamaindex@0.4.9
+
+## 0.0.8
+
+### Patch Changes
+
+- 83ebdfb: fix: next.js build error
+- Updated dependencies [83ebdfb]
+  - llamaindex@0.4.8
+
+## 0.0.7
+
+### Patch Changes
+
+- Updated dependencies [41fe871]
+- Updated dependencies [321c39d]
+- Updated dependencies [f7f1af0]
+  - llamaindex@0.4.7
+
+## 0.0.6
+
+### Patch Changes
+
+- Updated dependencies [1feb23b]
+- Updated dependencies [08c55ec]
+  - llamaindex@0.4.6
+
+## 0.0.5
+
+### Patch Changes
+
+- Updated dependencies [6c3e5d0]
+  - llamaindex@0.4.5
+
+## 0.0.4
+
+### Patch Changes
+
+- Updated dependencies [42eb73a]
+  - llamaindex@0.4.4
+
 ## 0.0.3
 
 ### Patch Changes

packages/llamaindex/e2e/examples/nextjs-node-runtime/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@llamaindex/next-node-runtime-test",
-  "version": "0.0.3",
+  "version": "0.0.12",
   "private": true,
   "scripts": {
     "dev": "next dev",

packages/llamaindex/e2e/examples/nextjs-node-runtime/src/actions/openai.ts

@@ -0,0 +1,68 @@
+"use server";
+import {
+  OpenAI,
+  OpenAIAgent,
+  QueryEngineTool,
+  Settings,
+  VectorStoreIndex,
+} from "llamaindex";
+import { HuggingFaceEmbedding } from "llamaindex/embeddings/HuggingFaceEmbedding";
+import { SimpleDirectoryReader } from "llamaindex/readers/SimpleDirectoryReader";
+
+Settings.llm = new OpenAI({
+  // eslint-disable-next-line turbo/no-undeclared-env-vars
+  apiKey: process.env.NEXT_PUBLIC_OPENAI_KEY ?? "FAKE_KEY_TO_PASS_TESTS",
+  model: "gpt-4o",
+});
+Settings.embedModel = new HuggingFaceEmbedding({
+  modelType: "BAAI/bge-small-en-v1.5",
+  quantized: false,
+});
+Settings.callbackManager.on("llm-tool-call", (event) => {
+  console.log(event.detail.payload);
+});
+Settings.callbackManager.on("llm-tool-result", (event) => {
+  console.log(event.detail.payload);
+});
+
+export async function getOpenAIModelRequest(query: string) {
+  try {
+    const currentDir = __dirname;
+
+    // load our data and create a query engine
+    const reader = new SimpleDirectoryReader();
+    const documents = await reader.loadData(currentDir);
+    const index = await VectorStoreIndex.fromDocuments(documents);
+    const retriever = index.asRetriever({
+      similarityTopK: 10,
+    });
+    const queryEngine = index.asQueryEngine({
+      retriever,
+    });
+
+    // define the query engine as a tool
+    const tools = [
+      new QueryEngineTool({
+        queryEngine: queryEngine,
+        metadata: {
+          name: "deployment_details_per_env",
+          description: `This tool can answer detailed questions about deployments happened in various environments.`,
+        },
+      }),
+    ];
+    // create the agent
+    const agent = new OpenAIAgent({ tools });
+
+    const { response } = await agent.chat({
+      message: query,
+    });
+    return {
+      message: response,
+    };
+  } catch (err) {
+    console.error(err);
+    return {
+      errors: "Error Calling OpenAI Model",
+    };
+  }
+}

packages/llamaindex/e2e/examples/nextjs-node-runtime/src/app/api/openai/route.ts

@@ -0,0 +1,10 @@
+import { getOpenAIModelRequest } from "@/actions/openai";
+import { NextRequest, NextResponse } from "next/server";
+
+// POST /api/openai
+export async function POST(request: NextRequest) {
+  const body = await request.json();
+  const content = await getOpenAIModelRequest(body.query);
+
+  return NextResponse.json(content, { status: 200 });
+}

packages/llamaindex/e2e/examples/waku-query-engine/CHANGELOG.md

@@ -1,5 +1,71 @@
 # @llamaindex/waku-query-engine-test
 
+## 0.0.31
+
+### Patch Changes
+
+- Updated dependencies [f326ab8]
+  - llamaindex@0.4.12
+
+## 0.0.30
+
+### Patch Changes
+
+- Updated dependencies [8bf5b4a]
+  - llamaindex@0.4.11
+
+## 0.0.29
+
+### Patch Changes
+
+- Updated dependencies [7dce3d2]
+  - llamaindex@0.4.10
+
+## 0.0.28
+
+### Patch Changes
+
+- Updated dependencies [3a96a48]
+  - llamaindex@0.4.9
+
+## 0.0.27
+
+### Patch Changes
+
+- Updated dependencies [83ebdfb]
+  - llamaindex@0.4.8
+
+## 0.0.26
+
+### Patch Changes
+
+- Updated dependencies [41fe871]
+- Updated dependencies [321c39d]
+- Updated dependencies [f7f1af0]
+  - llamaindex@0.4.7
+
+## 0.0.25
+
+### Patch Changes
+
+- Updated dependencies [1feb23b]
+- Updated dependencies [08c55ec]
+  - llamaindex@0.4.6
+
+## 0.0.24
+
+### Patch Changes
+
+- Updated dependencies [6c3e5d0]
+  - llamaindex@0.4.5
+
+## 0.0.23
+
+### Patch Changes
+
+- Updated dependencies [42eb73a]
+  - llamaindex@0.4.4
+
 ## 0.0.22
 
 ### Patch Changes

packages/llamaindex/e2e/examples/waku-query-engine/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@llamaindex/waku-query-engine-test",
-  "version": "0.0.22",
+  "version": "0.0.31",
   "type": "module",
   "private": true,
   "scripts": {

packages/llamaindex/e2e/fixtures/llm/openai.ts

@@ -7,7 +7,7 @@ import type {
   LLMChatParamsStreaming,
   LLMCompletionParamsNonStreaming,
   LLMCompletionParamsStreaming,
-} from "llamaindex/llm/types";
+} from "llamaindex";
 import { extractText } from "llamaindex/llm/utils";
 import { deepStrictEqual, strictEqual } from "node:assert";
 import { llmCompleteMockStorage } from "../../node/utils.js";

packages/llamaindex/jsr.json

@@ -1,11 +0,0 @@
-{
-  "name": "@llamaindex/llamaindex",
-  "version": "0.4.3",
-  "exports": "./src/index.ts",
-  "imports": {
-    "@llamaindex/env": "jsr:@llamaindex/env@0.1.6"
-  },
-  "publish": {
-    "include": ["LICENSE", "README.md", "src/**/*", "jsr.json"]
-  }
-}

packages/llamaindex/package.json

@@ -1,6 +1,6 @@
 {
   "name": "llamaindex",
-  "version": "0.4.3",
+  "version": "0.4.12",
   "license": "MIT",
   "type": "module",
   "keywords": [
@@ -22,17 +22,20 @@
   "dependencies": {
     "@anthropic-ai/sdk": "^0.21.1",
     "@aws-crypto/sha256-js": "^5.2.0",
+    "@azure/identity": "^4.2.1",
     "@datastax/astra-db-ts": "^1.2.1",
     "@google-cloud/vertexai": "^1.2.0",
     "@google/generative-ai": "^0.12.0",
     "@grpc/grpc-js": "^1.10.8",
     "@huggingface/inference": "^2.7.0",
-    "@llamaindex/cloud": "0.0.7",
+    "@llamaindex/cloud": "workspace:*",
+    "@llamaindex/core": "workspace:*",
     "@llamaindex/env": "workspace:*",
     "@mistralai/mistralai": "^0.4.0",
     "@pinecone-database/pinecone": "^2.2.2",
     "@qdrant/js-client-rest": "^1.9.0",
-    "@types/lodash": "^4.17.5",
+    "@types/lodash": "^4.17.4",
+    "@types/node": "^20.14.5",
     "@types/papaparse": "^5.3.14",
     "@types/pg": "^8.11.6",
     "@xenova/transformers": "^2.17.2",
@@ -60,7 +63,8 @@
     "tiktoken": "^1.0.15",
     "unpdf": "^0.10.1",
     "wikipedia": "^2.1.2",
-    "wink-nlp": "^2.3.0"
+    "wink-nlp": "^2.3.0",
+    "zod": "^3.23.8"
   },
   "peerDependencies": {
     "@notionhq/client": "^2.2.15"
@@ -148,7 +152,7 @@
   "repository": {
     "type": "git",
     "url": "https://github.com/run-llama/LlamaIndexTS.git",
-    "directory": "packages/core"
+    "directory": "packages/llamaindex"
   },
   "scripts": {
     "lint": "eslint .",

packages/llamaindex/src/ChatHistory.ts

@@ -1,8 +1,8 @@
+import type { ChatMessage, LLM, MessageType } from "@llamaindex/core/llms";
 import { tokenizers, type Tokenizer } from "@llamaindex/env";
 import type { SummaryPrompt } from "./Prompt.js";
 import { defaultSummaryPrompt, messagesToHistoryStr } from "./Prompt.js";
 import { OpenAI } from "./llm/openai.js";
-import type { ChatMessage, LLM, MessageType } from "./llm/types.js";
 import { extractText } from "./llm/utils.js";
 
 /**

packages/llamaindex/src/EngineResponse.ts

@@ -1,9 +1,9 @@
-import type { NodeWithScore } from "./Node.js";
 import type {
   ChatMessage,
   ChatResponse,
   ChatResponseChunk,
-} from "./llm/types.js";
+} from "@llamaindex/core/llms";
+import type { NodeWithScore } from "@llamaindex/core/schema";
 import { extractText } from "./llm/utils.js";
 
 export class EngineResponse implements ChatResponse, ChatResponseChunk {
@@ -64,7 +64,7 @@ export class EngineResponse implements ChatResponse, ChatResponseChunk {
     sourceNodes?: NodeWithScore[],
   ): EngineResponse {
     return new EngineResponse(
-      this.toChatResponse(chunk.delta, chunk.raw),
+      EngineResponse.toChatResponse(chunk.delta, chunk.raw),
       true,
       sourceNodes,
     );

packages/llamaindex/src/Node.ts

@@ -1,448 +1,6 @@
-import { createSHA256, path, randomUUID } from "@llamaindex/env";
-import { chunkSizeCheck, lazyInitHash } from "./internal/decorator/node.js";
+console.warn(
+  'Package "llamaindex/Node" is deprecated, and will be removed in the next major release.',
+);
+console.warn("Please import from the @llamaindex/core/schema package instead.");
 
-export enum NodeRelationship {
-  SOURCE = "SOURCE",
-  PREVIOUS = "PREVIOUS",
-  NEXT = "NEXT",
-  PARENT = "PARENT",
-  CHILD = "CHILD",
-}
-
-export enum ObjectType {
-  TEXT = "TEXT",
-  IMAGE = "IMAGE",
-  INDEX = "INDEX",
-  DOCUMENT = "DOCUMENT",
-  IMAGE_DOCUMENT = "IMAGE_DOCUMENT",
-}
-
-export enum MetadataMode {
-  ALL = "ALL",
-  EMBED = "EMBED",
-  LLM = "LLM",
-  NONE = "NONE",
-}
-
-export type Metadata = Record<string, any>;
-
-export interface RelatedNodeInfo<T extends Metadata = Metadata> {
-  nodeId: string;
-  nodeType?: ObjectType;
-  metadata: T;
-  hash?: string;
-}
-
-export type RelatedNodeType<T extends Metadata = Metadata> =
-  | RelatedNodeInfo<T>
-  | RelatedNodeInfo<T>[];
-
-export type BaseNodeParams<T extends Metadata = Metadata> = {
-  id_?: string;
-  metadata?: T;
-  excludedEmbedMetadataKeys?: string[];
-  excludedLlmMetadataKeys?: string[];
-  relationships?: Partial<Record<NodeRelationship, RelatedNodeType<T>>>;
-  hash?: string;
-  embedding?: number[];
-};
-
-/**
- * Generic abstract class for retrievable nodes
- */
-export abstract class BaseNode<T extends Metadata = Metadata> {
-  /**
-   * The unique ID of the Node/Document. The trailing underscore is here
-   * to avoid collisions with the id keyword in Python.
-   *
-   * Set to a UUID by default.
-   */
-  id_: string;
-  embedding?: number[];
-
-  // Metadata fields
-  metadata: T;
-  excludedEmbedMetadataKeys: string[];
-  excludedLlmMetadataKeys: string[];
-  relationships: Partial<Record<NodeRelationship, RelatedNodeType<T>>>;
-
-  @lazyInitHash
-  accessor hash: string = "";
-
-  protected constructor(init?: BaseNodeParams<T>) {
-    const {
-      id_,
-      metadata,
-      excludedEmbedMetadataKeys,
-      excludedLlmMetadataKeys,
-      relationships,
-      hash,
-      embedding,
-    } = init || {};
-    this.id_ = id_ ?? randomUUID();
-    this.metadata = metadata ?? ({} as T);
-    this.excludedEmbedMetadataKeys = excludedEmbedMetadataKeys ?? [];
-    this.excludedLlmMetadataKeys = excludedLlmMetadataKeys ?? [];
-    this.relationships = relationships ?? {};
-    this.embedding = embedding;
-  }
-
-  abstract get type(): ObjectType;
-
-  abstract getContent(metadataMode: MetadataMode): string;
-  abstract getMetadataStr(metadataMode: MetadataMode): string;
-  // todo: set value as a generic type
-  abstract setContent(value: unknown): void;
-
-  get sourceNode(): RelatedNodeInfo<T> | undefined {
-    const relationship = this.relationships[NodeRelationship.SOURCE];
-
-    if (Array.isArray(relationship)) {
-      throw new Error("Source object must be a single RelatedNodeInfo object");
-    }
-
-    return relationship;
-  }
-
-  get prevNode(): RelatedNodeInfo<T> | undefined {
-    const relationship = this.relationships[NodeRelationship.PREVIOUS];
-
-    if (Array.isArray(relationship)) {
-      throw new Error(
-        "Previous object must be a single RelatedNodeInfo object",
-      );
-    }
-
-    return relationship;
-  }
-
-  get nextNode(): RelatedNodeInfo<T> | undefined {
-    const relationship = this.relationships[NodeRelationship.NEXT];
-
-    if (Array.isArray(relationship)) {
-      throw new Error("Next object must be a single RelatedNodeInfo object");
-    }
-
-    return relationship;
-  }
-
-  get parentNode(): RelatedNodeInfo<T> | undefined {
-    const relationship = this.relationships[NodeRelationship.PARENT];
-
-    if (Array.isArray(relationship)) {
-      throw new Error("Parent object must be a single RelatedNodeInfo object");
-    }
-
-    return relationship;
-  }
-
-  get childNodes(): RelatedNodeInfo<T>[] | undefined {
-    const relationship = this.relationships[NodeRelationship.CHILD];
-
-    if (!Array.isArray(relationship)) {
-      throw new Error(
-        "Child object must be a an array of RelatedNodeInfo objects",
-      );
-    }
-
-    return relationship;
-  }
-
-  abstract generateHash(): string;
-
-  getEmbedding(): number[] {
-    if (this.embedding === undefined) {
-      throw new Error("Embedding not set");
-    }
-
-    return this.embedding;
-  }
-
-  asRelatedNodeInfo(): RelatedNodeInfo<T> {
-    return {
-      nodeId: this.id_,
-      metadata: this.metadata,
-      hash: this.hash,
-    };
-  }
-
-  /**
-   * Called by built in JSON.stringify (see https://javascript.info/json)
-   * Properties are read-only as they are not deep-cloned (not necessary for stringification).
-   * @see toMutableJSON - use to return a mutable JSON instead
-   */
-  toJSON(): Record<string, any> {
-    return {
-      ...this,
-      type: this.type,
-      // hash is an accessor property, so it's not included in the rest operator
-      hash: this.hash,
-    };
-  }
-
-  clone(): BaseNode {
-    return jsonToNode(this.toMutableJSON()) as BaseNode;
-  }
-
-  /**
-   * Converts the object to a JSON representation.
-   * Properties can be safely modified as a deep clone of the properties are created.
-   * @return {Record<string, any>} - The JSON representation of the object.
-   */
-  toMutableJSON(): Record<string, any> {
-    return structuredClone(this.toJSON());
-  }
-}
-
-export type TextNodeParams<T extends Metadata = Metadata> =
-  BaseNodeParams<T> & {
-    text?: string;
-    textTemplate?: string;
-    startCharIdx?: number;
-    endCharIdx?: number;
-    metadataSeparator?: string;
-  };
-
-/**
- * TextNode is the default node type for text. Most common node type in LlamaIndex.TS
- */
-export class TextNode<T extends Metadata = Metadata> extends BaseNode<T> {
-  text: string;
-  textTemplate: string;
-
-  startCharIdx?: number;
-  endCharIdx?: number;
-  // textTemplate: NOTE write your own formatter if needed
-  // metadataTemplate: NOTE write your own formatter if needed
-  metadataSeparator: string;
-
-  constructor(init: TextNodeParams<T> = {}) {
-    super(init);
-    const { text, textTemplate, startCharIdx, endCharIdx, metadataSeparator } =
-      init;
-    this.text = text ?? "";
-    this.textTemplate = textTemplate ?? "";
-    if (startCharIdx) {
-      this.startCharIdx = startCharIdx;
-    }
-    this.endCharIdx = endCharIdx;
-    this.metadataSeparator = metadataSeparator ?? "\n";
-  }
-
-  /**
-   * Generate a hash of the text node.
-   * The ID is not part of the hash as it can change independent of content.
-   * @returns
-   */
-  generateHash() {
-    const hashFunction = createSHA256();
-    hashFunction.update(`type=${this.type}`);
-    hashFunction.update(
-      `startCharIdx=${this.startCharIdx} endCharIdx=${this.endCharIdx}`,
-    );
-    hashFunction.update(this.getContent(MetadataMode.ALL));
-    return hashFunction.digest();
-  }
-
-  get type() {
-    return ObjectType.TEXT;
-  }
-
-  @chunkSizeCheck
-  getContent(metadataMode: MetadataMode = MetadataMode.NONE): string {
-    const metadataStr = this.getMetadataStr(metadataMode).trim();
-    return `${metadataStr}\n\n${this.text}`.trim();
-  }
-
-  getMetadataStr(metadataMode: MetadataMode): string {
-    if (metadataMode === MetadataMode.NONE) {
-      return "";
-    }
-
-    const usableMetadataKeys = new Set(Object.keys(this.metadata).sort());
-    if (metadataMode === MetadataMode.LLM) {
-      for (const key of this.excludedLlmMetadataKeys) {
-        usableMetadataKeys.delete(key);
-      }
-    } else if (metadataMode === MetadataMode.EMBED) {
-      for (const key of this.excludedEmbedMetadataKeys) {
-        usableMetadataKeys.delete(key);
-      }
-    }
-
-    return [...usableMetadataKeys]
-      .map((key) => `${key}: ${this.metadata[key]}`)
-      .join(this.metadataSeparator);
-  }
-
-  setContent(value: string) {
-    this.text = value;
-    this.hash = this.generateHash();
-  }
-
-  getNodeInfo() {
-    return { start: this.startCharIdx, end: this.endCharIdx };
-  }
-
-  getText() {
-    return this.getContent(MetadataMode.NONE);
-  }
-}
-
-export type IndexNodeParams<T extends Metadata = Metadata> =
-  TextNodeParams<T> & {
-    indexId: string;
-  };
-
-export class IndexNode<T extends Metadata = Metadata> extends TextNode<T> {
-  indexId: string;
-
-  constructor(init?: IndexNodeParams<T>) {
-    super(init);
-    const { indexId } = init || {};
-    this.indexId = indexId ?? "";
-  }
-
-  get type() {
-    return ObjectType.INDEX;
-  }
-}
-
-/**
- * A document is just a special text node with a docId.
- */
-export class Document<T extends Metadata = Metadata> extends TextNode<T> {
-  constructor(init?: TextNodeParams<T>) {
-    super(init);
-  }
-
-  get type() {
-    return ObjectType.DOCUMENT;
-  }
-}
-
-export function jsonToNode(json: any, type?: ObjectType) {
-  if (!json.type && !type) {
-    throw new Error("Node type not found");
-  }
-  const nodeType = type || json.type;
-
-  switch (nodeType) {
-    case ObjectType.TEXT:
-      return new TextNode(json);
-    case ObjectType.INDEX:
-      return new IndexNode(json);
-    case ObjectType.DOCUMENT:
-      return new Document(json);
-    case ObjectType.IMAGE_DOCUMENT:
-      return new ImageDocument(json);
-    default:
-      throw new Error(`Invalid node type: ${nodeType}`);
-  }
-}
-
-export type ImageType = string | Blob | URL;
-
-export type ImageNodeParams<T extends Metadata = Metadata> =
-  TextNodeParams<T> & {
-    image: ImageType;
-  };
-
-export class ImageNode<T extends Metadata = Metadata> extends TextNode<T> {
-  image: ImageType; // image as blob
-
-  constructor(init: ImageNodeParams<T>) {
-    super(init);
-    const { image } = init;
-    this.image = image;
-  }
-
-  get type() {
-    return ObjectType.IMAGE;
-  }
-
-  getUrl(): URL {
-    // id_ stores the relative path, convert it to the URL of the file
-    const absPath = path.resolve(this.id_);
-    return new URL(`file://${absPath}`);
-  }
-
-  // Calculates the image part of the hash
-  private generateImageHash() {
-    const hashFunction = createSHA256();
-
-    if (this.image instanceof Blob) {
-      // TODO: ideally we should use the blob's content to calculate the hash:
-      // hashFunction.update(new Uint8Array(await this.image.arrayBuffer()));
-      // as this is async, we're using the node's ID for the time being
-      hashFunction.update(this.id_);
-    } else if (this.image instanceof URL) {
-      hashFunction.update(this.image.toString());
-    } else if (typeof this.image === "string") {
-      hashFunction.update(this.image);
-    } else {
-      throw new Error(
-        `Unknown image type: ${typeof this.image}. Can't calculate hash`,
-      );
-    }
-
-    return hashFunction.digest();
-  }
-
-  generateHash() {
-    const hashFunction = createSHA256();
-    // calculates hash based on hash of both components (image and text)
-    hashFunction.update(super.generateHash());
-    hashFunction.update(this.generateImageHash());
-
-    return hashFunction.digest();
-  }
-}
-
-export class ImageDocument<T extends Metadata = Metadata> extends ImageNode<T> {
-  constructor(init: ImageNodeParams<T>) {
-    super(init);
-  }
-
-  get type() {
-    return ObjectType.IMAGE_DOCUMENT;
-  }
-}
-
-/**
- * A node with a similarity score
- */
-export interface NodeWithScore<T extends Metadata = Metadata> {
-  node: BaseNode<T>;
-  score?: number;
-}
-
-export enum ModalityType {
-  TEXT = "TEXT",
-  IMAGE = "IMAGE",
-}
-
-type NodesByType = {
-  [P in ModalityType]?: BaseNode[];
-};
-
-export function splitNodesByType(nodes: BaseNode[]): NodesByType {
-  const result: NodesByType = {};
-
-  for (const node of nodes) {
-    let type: ModalityType;
-    if (node instanceof ImageNode) {
-      type = ModalityType.IMAGE;
-    } else if (node instanceof TextNode) {
-      type = ModalityType.TEXT;
-    } else {
-      throw new Error(`Unknown node type: ${node.type}`);
-    }
-    if (type in result) {
-      result[type]?.push(node);
-    } else {
-      result[type] = [node];
-    }
-  }
-  return result;
-}
+export * from "@llamaindex/core/schema";

packages/llamaindex/src/Prompt.ts

@@ -1,6 +1,5 @@
+import type { ChatMessage, ToolMetadata } from "@llamaindex/core/llms";
 import type { SubQuestion } from "./engines/query/types.js";
-import type { ChatMessage } from "./llm/types.js";
-import type { ToolMetadata } from "./types.js";
 
 /**
  * A SimplePrompt is a function that takes a dictionary of inputs and returns a string.

packages/llamaindex/src/QuestionGenerator.ts

@@ -1,3 +1,4 @@
+import type { LLM, ToolMetadata } from "@llamaindex/core/llms";
 import { SubQuestionOutputParser } from "./OutputParser.js";
 import type { SubQuestionPrompt } from "./Prompt.js";
 import { buildToolsText, defaultSubQuestionPrompt } from "./Prompt.js";
@@ -6,13 +7,8 @@ import type {
   SubQuestion,
 } from "./engines/query/types.js";
 import { OpenAI } from "./llm/openai.js";
-import type { LLM } from "./llm/types.js";
 import { PromptMixin } from "./prompts/index.js";
-import type {
-  BaseOutputParser,
-  StructuredOutput,
-  ToolMetadata,
-} from "./types.js";
+import type { BaseOutputParser, StructuredOutput } from "./types.js";
 
 /**
  * LLMQuestionGenerator uses the LLM to generate new questions for the LLM using tools and a user query.

packages/llamaindex/src/Retriever.ts

@@ -1,4 +1,4 @@
-import type { NodeWithScore } from "./Node.js";
+import type { NodeWithScore } from "@llamaindex/core/schema";
 import type { ServiceContext } from "./ServiceContext.js";
 import type { MessageContent } from "./index.edge.js";
 

packages/llamaindex/src/ServiceContext.ts

@@ -1,8 +1,8 @@
+import type { LLM } from "@llamaindex/core/llms";
 import { PromptHelper } from "./PromptHelper.js";
 import { OpenAIEmbedding } from "./embeddings/OpenAIEmbedding.js";
 import type { BaseEmbedding } from "./embeddings/types.js";
 import { OpenAI } from "./llm/openai.js";
-import type { LLM } from "./llm/types.js";
 import { SimpleNodeParser } from "./nodeParsers/SimpleNodeParser.js";
 import type { NodeParser } from "./nodeParsers/types.js";
 

packages/llamaindex/src/Settings.ts

@@ -1,9 +1,11 @@
+import { Settings as CoreSettings } from "@llamaindex/core/global";
 import { CallbackManager } from "./callbacks/CallbackManager.js";
 import { OpenAI } from "./llm/openai.js";
 
 import { PromptHelper } from "./PromptHelper.js";
 import { SimpleNodeParser } from "./nodeParsers/SimpleNodeParser.js";
 
+import type { LLM } from "@llamaindex/core/llms";
 import { AsyncLocalStorage, getEnv } from "@llamaindex/env";
 import type { ServiceContext } from "./ServiceContext.js";
 import type { BaseEmbedding } from "./embeddings/types.js";
@@ -17,12 +19,6 @@ import {
   setEmbeddedModel,
   withEmbeddedModel,
 } from "./internal/settings/EmbedModel.js";
-import {
-  getChunkSize,
-  setChunkSize,
-  withChunkSize,
-} from "./internal/settings/chunk-size.js";
-import type { LLM } from "./llm/types.js";
 import type { NodeParser } from "./nodeParsers/types.js";
 
 export type PromptConfig = {
@@ -148,15 +144,15 @@ class GlobalSettings implements Config {
   }
 
   set chunkSize(chunkSize: number | undefined) {
-    setChunkSize(chunkSize);
+    CoreSettings.chunkSize = chunkSize;
   }
 
   get chunkSize(): number | undefined {
-    return getChunkSize();
+    return CoreSettings.chunkSize;
   }
 
   withChunkSize<Result>(chunkSize: number, fn: () => Result): Result {
-    return withChunkSize(chunkSize, fn);
+    return CoreSettings.withChunkSize(chunkSize, fn);
   }
 
   get chunkOverlap(): number | undefined {

packages/llamaindex/src/agent/base.ts

@@ -1,3 +1,10 @@
+import type {
+  BaseToolWithCall,
+  ChatMessage,
+  LLM,
+  MessageContent,
+  ToolOutput,
+} from "@llamaindex/core/llms";
 import { ReadableStream, TransformStream, randomUUID } from "@llamaindex/env";
 import { ChatHistory } from "../ChatHistory.js";
 import { EngineResponse } from "../EngineResponse.js";
@@ -11,8 +18,7 @@ import { wrapEventCaller } from "../internal/context/EventCaller.js";
 import { consoleLogger, emptyLogger } from "../internal/logger.js";
 import { getCallbackManager } from "../internal/settings/CallbackManager.js";
 import { isAsyncIterable } from "../internal/utils.js";
-import type { ChatMessage, LLM, MessageContent } from "../llm/index.js";
-import type { BaseToolWithCall, ToolOutput } from "../types.js";
+import { ObjectRetriever } from "../objects/index.js";
 import type {
   AgentTaskContext,
   TaskHandler,
@@ -126,12 +132,21 @@ export type AgentParamsBase<
   >
     ? AdditionalMessageOptions
     : never,
-> = {
-  llm?: AI;
-  chatHistory?: ChatMessage<AdditionalMessageOptions>[];
-  systemPrompt?: MessageContent;
-  verbose?: boolean;
-};
+> =
+  | {
+      llm?: AI;
+      chatHistory?: ChatMessage<AdditionalMessageOptions>[];
+      systemPrompt?: MessageContent;
+      verbose?: boolean;
+      tools: BaseToolWithCall[];
+    }
+  | {
+      llm?: AI;
+      chatHistory?: ChatMessage<AdditionalMessageOptions>[];
+      systemPrompt?: MessageContent;
+      verbose?: boolean;
+      toolRetriever: ObjectRetriever<BaseToolWithCall>;
+    };
 
 /**
  * Worker will schedule tasks and handle the task execution
@@ -219,13 +234,12 @@ export abstract class AgentRunner<
     const { llm, getTools, stream } = step.context;
     const lastMessage = step.context.store.messages.at(-1)!.content;
     const tools = await getTools(lastMessage);
-    const response = await llm.chat({
-      // @ts-expect-error
-      stream,
-      tools,
-      messages: [...step.context.store.messages],
-    });
     if (!stream) {
+      const response = await llm.chat({
+        stream,
+        tools,
+        messages: [...step.context.store.messages],
+      });
       await stepTools<LLM>({
         response,
         tools,
@@ -233,6 +247,11 @@ export abstract class AgentRunner<
         enqueueOutput,
       });
     } else {
+      const response = await llm.chat({
+        stream,
+        tools,
+        messages: [...step.context.store.messages],
+      });
       await stepToolsStreaming<LLM>({
         response,
         tools,

packages/llamaindex/src/agent/llm.ts

@@ -1,8 +1,8 @@
-import type { LLM } from "../llm/index.js";
+import type { BaseToolWithCall, LLM } from "@llamaindex/core/llms";
 import { ObjectRetriever } from "../objects/index.js";
 import { Settings } from "../Settings.js";
-import type { BaseToolWithCall } from "../types.js";
 import { AgentRunner, AgentWorker, type AgentParamsBase } from "./base.js";
+import { validateAgentParams } from "./utils.js";
 
 type LLMParamsBase = AgentParamsBase<LLM>;
 
@@ -22,6 +22,7 @@ export class LLMAgentWorker extends AgentWorker<LLM> {
 
 export class LLMAgent extends AgentRunner<LLM> {
   constructor(params: LLMAgentParams) {
+    validateAgentParams(params);
     const llm = params.llm ?? (Settings.llm ? (Settings.llm as LLM) : null);
     if (!llm)
       throw new Error(