Release 0.7.1 (#1342)

Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
Co-authored-by: himself65 <himself65@users.noreply.github.com>

.github/workflows/preview.yml

@@ -25,4 +25,4 @@ jobs:
         run: pnpm run build
 
       - name: Pre Release
-        run: pnpx pkg-pr-new publish ./packages/*
+        run: pnpx pkg-pr-new publish ./packages/* ./packages/providers/*

.github/workflows/test.yml

@@ -136,27 +136,26 @@ 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/openai
-        run: pnpm pack --pack-destination ${{ runner.temp }}
-        working-directory: packages/llm/openai
-      - name: Pack @llamaindex/groq
-        run: pnpm pack --pack-destination ${{ runner.temp }}
-        working-directory: packages/llm/groq
-      - name: Pack @llamaindex/ollama
-        run: pnpm pack --pack-destination ${{ runner.temp }}
-        working-directory: packages/llm/ollama
-      - 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
-      - name: Pack llamaindex
-        run: pnpm pack --pack-destination ${{ runner.temp }}
-        working-directory: packages/llamaindex
+      - name: Pack packages
+        run: |
+          for dir in packages/*; do
+            if [ -d "$dir" ] && [ -f "$dir/package.json" ] && [[ ! "$dir" =~ autotool ]]; then
+              echo "Packing $dir"
+              pnpm pack --pack-destination ${{ runner.temp }} -C $dir
+            else
+              echo "Skipping $dir, no package.json found"
+            fi
+          done
+      - name: Pack provider packages
+        run: |
+          for dir in packages/providers/*; do
+            if [ -d "$dir" ] && [ -f "$dir/package.json" ]; then
+              echo "Packing $dir"
+              pnpm pack --pack-destination ${{ runner.temp }} -C $dir
+            else
+              echo "Skipping $dir, no package.json found"
+            fi
+          done
       - name: Install
         run: npm add ${{ runner.temp }}/*.tgz
         working-directory: ${{ runner.temp }}/examples

apps/docs/CHANGELOG.md

@@ -1,5 +1,108 @@
 # docs
 
+## 0.0.93
+
+### Patch Changes
+
+- Updated dependencies [ae49ff4]
+- Updated dependencies [4c38c1b]
+- Updated dependencies [a75af83]
+- Updated dependencies [a75af83]
+  - llamaindex@0.7.1
+
+## 0.0.92
+
+### Patch Changes
+
+- Updated dependencies [1364e8e]
+- Updated dependencies [3b7736f]
+- Updated dependencies [96fc69c]
+  - llamaindex@0.7.0
+  - @llamaindex/examples@0.0.9
+
+## 0.0.91
+
+### Patch Changes
+
+- Updated dependencies [5729bd9]
+  - llamaindex@0.6.22
+
+## 0.0.90
+
+### Patch Changes
+
+- Updated dependencies [6f75306]
+- Updated dependencies [94cb4ad]
+  - llamaindex@0.6.21
+
+## 0.0.89
+
+### Patch Changes
+
+- Updated dependencies [6a9a7b1]
+  - llamaindex@0.6.20
+
+## 0.0.88
+
+### Patch Changes
+
+- Updated dependencies [62cba52]
+- Updated dependencies [d265e96]
+- Updated dependencies [d30bbf7]
+- Updated dependencies [53fd00a]
+  - llamaindex@0.6.19
+
+## 0.0.87
+
+### Patch Changes
+
+- Updated dependencies [5f67820]
+- Updated dependencies [fe08d04]
+  - llamaindex@0.6.18
+
+## 0.0.86
+
+### Patch Changes
+
+- Updated dependencies [ee697fb]
+  - llamaindex@0.6.17
+
+## 0.0.85
+
+### Patch Changes
+
+- Updated dependencies [63e9846]
+- Updated dependencies [6f3a31c]
+  - llamaindex@0.6.16
+
+## 0.0.84
+
+### Patch Changes
+
+- Updated dependencies [2a82413]
+  - llamaindex@0.6.15
+
+## 0.0.83
+
+### Patch Changes
+
+- llamaindex@0.6.14
+
+## 0.0.82
+
+### Patch Changes
+
+- llamaindex@0.6.13
+
+## 0.0.81
+
+### Patch Changes
+
+- Updated dependencies [f7b4e94]
+- Updated dependencies [78037a6]
+- Updated dependencies [1d9e3b1]
+  - llamaindex@0.6.12
+
 ## 0.0.80
 
 ### Patch Changes

apps/docs/docs/modules/data_loaders/llama_parse/index.mdx

@@ -13,7 +13,7 @@ Official documentation for LlamaParse can be found [here](https://docs.cloud.lla
 ## Usage
 
 You can then use the `LlamaParseReader` class to load local files and convert them into a parsed document that can be used by LlamaIndex.
-See [LlamaParseReader.ts](https://github.com/run-llama/LlamaIndexTS/blob/main/packages/llamaindex/src/readers/LlamaParseReader.ts) for a list of supported file types:
+See [reader.ts](https://github.com/run-llama/LlamaIndexTS/blob/main/packages/cloud/src/reader.ts) for a list of supported file types:
 
 <CodeBlock language="ts">{CodeSource}</CodeBlock>
 

apps/docs/package.json

@@ -1,6 +1,6 @@
 {
   "name": "docs",
-  "version": "0.0.80",
+  "version": "0.0.93",
   "private": true,
   "scripts": {
     "docusaurus": "docusaurus",

examples/CHANGELOG.md

@@ -1,5 +1,16 @@
 # examples
 
+## 0.0.9
+
+### Patch Changes
+
+- Updated dependencies [1364e8e]
+- Updated dependencies [96fc69c]
+- Updated dependencies [3b7736f]
+- Updated dependencies [96fc69c]
+  - llamaindex@0.7.0
+  - @llamaindex/core@0.3.0
+
 ## 0.0.8
 
 ### Patch Changes

examples/agent/large_toolcall_with_gpt4o.ts

@@ -13,7 +13,7 @@ import { FunctionTool, OpenAI, ToolCallOptions } from "llamaindex";
   }
 })();
 
-async function callLLM(init: Partial<OpenAI>) {
+async function callLLM(init: { model: string }) {
   const csvData =
     "Country,Average Height (cm)\nNetherlands,156\nDenmark,158\nNorway,160";
 

examples/astradb/example.ts

@@ -1,6 +1,7 @@
 import {
   AstraDBVectorStore,
   Document,
+  MetadataFilters,
   storageContextFromDefaults,
   VectorStoreIndex,
 } from "llamaindex";
@@ -42,8 +43,10 @@ async function main() {
     const index = await VectorStoreIndex.fromDocuments(docs, {
       storageContext: ctx,
     });
-
-    const queryEngine = index.asQueryEngine();
+    const preFilters: MetadataFilters = {
+      filters: [{ key: "id", operator: "in", value: [123, 789] }],
+    }; // try changing the filters to see the different results
+    const queryEngine = index.asQueryEngine({ preFilters });
     const response = await queryEngine.query({
       query: "Describe AstraDB.",
     });

examples/chromadb/preFilters.ts

@@ -1,57 +1,83 @@
 import {
   ChromaVectorStore,
   Document,
+  MetadataFilters,
   VectorStoreIndex,
   storageContextFromDefaults,
 } from "llamaindex";
 
-const collectionName = "dog_colors";
+const collectionName = "dogs_with_color";
 
 async function main() {
   try {
-    const docs = [
-      new Document({
-        text: "The dog is brown",
-        metadata: {
-          dogId: "1",
-        },
-      }),
-      new Document({
-        text: "The dog is red",
-        metadata: {
-          dogId: "2",
-        },
-      }),
-    ];
-
-    console.log("Creating ChromaDB vector store");
     const chromaVS = new ChromaVectorStore({ collectionName });
-    const ctx = await storageContextFromDefaults({ vectorStore: chromaVS });
+    const index = await VectorStoreIndex.fromVectorStore(chromaVS);
 
-    console.log("Embedding documents and adding to index");
-    const index = await VectorStoreIndex.fromDocuments(docs, {
-      storageContext: ctx,
-    });
+    const queryFn = async (filters?: MetadataFilters) => {
+      console.log("\nQuerying dogs by filters: ", JSON.stringify(filters));
+      const query = "List all colors of dogs";
+      const queryEngine = index.asQueryEngine({
+        preFilters: filters,
+        similarityTopK: 3,
+      });
+      const response = await queryEngine.query({ query });
+      console.log(response.toString());
+    };
 
-    console.log("Querying index");
-    const queryEngine = index.asQueryEngine({
-      preFilters: {
-        filters: [
-          {
-            key: "dogId",
-            value: "2",
-            operator: "==",
-          },
-        ],
-      },
-    });
-    const response = await queryEngine.query({
-      query: "What is the color of the dog?",
-    });
-    console.log(response.toString());
+    await queryFn(); // red, brown, yellow
+    await queryFn({ filters: [{ key: "dogId", value: "1", operator: "==" }] }); // brown
+    await queryFn({ filters: [{ key: "dogId", value: "1", operator: "!=" }] }); // red, yellow
+    await queryFn({
+      filters: [
+        { key: "dogId", value: "1", operator: "==" },
+        { key: "dogId", value: "3", operator: "==" },
+      ],
+      condition: "or",
+    }); // brown, yellow
+    await queryFn({
+      filters: [{ key: "dogId", value: ["1", "2"], operator: "in" }],
+    }); // red, brown
   } catch (e) {
     console.error(e);
   }
 }
 
-void main();
+async function generate() {
+  const docs = [
+    new Document({
+      id_: "doc1",
+      text: "The dog is brown",
+      metadata: {
+        dogId: "1",
+      },
+    }),
+    new Document({
+      id_: "doc2",
+      text: "The dog is red",
+      metadata: {
+        dogId: "2",
+      },
+    }),
+    new Document({
+      id_: "doc3",
+      text: "The dog is yellow",
+      metadata: {
+        dogId: "3",
+      },
+    }),
+  ];
+
+  console.log("Creating ChromaDB vector store");
+  const chromaVS = new ChromaVectorStore({ collectionName });
+  const ctx = await storageContextFromDefaults({ vectorStore: chromaVS });
+
+  console.log("Embedding documents and adding to index");
+  await VectorStoreIndex.fromDocuments(docs, {
+    storageContext: ctx,
+  });
+}
+
+(async () => {
+  await generate();
+  await main();
+})();

examples/metadata-filter/preFilters.ts

@@ -0,0 +1,51 @@
+import {
+  Document,
+  MetadataFilters,
+  Settings,
+  SimpleDocumentStore,
+  VectorStoreIndex,
+  storageContextFromDefaults,
+} from "llamaindex";
+
+async function getDataSource() {
+  const docs = [
+    new Document({ text: "The dog is brown", metadata: { dogId: "1" } }),
+    new Document({ text: "The dog is yellow", metadata: { dogId: "2" } }),
+  ];
+  const storageContext = await storageContextFromDefaults({
+    persistDir: "./cache",
+  });
+  const numberOfDocs = Object.keys(
+    (storageContext.docStore as SimpleDocumentStore).toDict(),
+  ).length;
+  if (numberOfDocs === 0) {
+    return await VectorStoreIndex.fromDocuments(docs, { storageContext });
+  }
+  return await VectorStoreIndex.init({
+    storageContext,
+  });
+}
+
+Settings.callbackManager.on("retrieve-end", (event) => {
+  const { nodes, query } = event.detail;
+  console.log(`${query.query} - Number of retrieved nodes:`, nodes.length);
+});
+
+async function main() {
+  const index = await getDataSource();
+  const filters: MetadataFilters = {
+    filters: [{ key: "dogId", value: "2", operator: "==" }],
+  };
+
+  const retriever = index.asRetriever({ similarityTopK: 3, filters });
+  const queryEngine = index.asQueryEngine({
+    similarityTopK: 3,
+    preFilters: filters,
+  });
+
+  console.log("Retriever and query engine should only retrieve 1 node:");
+  await retriever.retrieve({ query: "Retriever: get dog" });
+  await queryEngine.query({ query: "QueryEngine: get dog" });
+}
+
+void main();

examples/package.json

@@ -1,12 +1,12 @@
 {
   "name": "@llamaindex/examples",
   "private": true,
-  "version": "0.0.8",
+  "version": "0.0.9",
   "dependencies": {
     "@aws-crypto/sha256-js": "^5.2.0",
     "@azure/identity": "^4.4.1",
     "@datastax/astra-db-ts": "^1.4.1",
-    "@llamaindex/core": "^0.2.0",
+    "@llamaindex/core": "^0.3.0",
     "@notionhq/client": "^2.2.15",
     "@pinecone-database/pinecone": "^3.0.2",
     "@vercel/postgres": "^0.10.0",
@@ -15,7 +15,7 @@
     "commander": "^12.1.0",
     "dotenv": "^16.4.5",
     "js-tiktoken": "^1.0.14",
-    "llamaindex": "^0.6.0",
+    "llamaindex": "^0.7.0",
     "mongodb": "^6.7.0",
     "pathe": "^1.1.2",
     "postgres": "^3.4.4"

examples/qdrantdb/preFilters.ts

@@ -39,6 +39,12 @@ async function main() {
           dogId: "2",
         },
       }),
+      new Document({
+        text: "The dog is black",
+        metadata: {
+          dogId: "3",
+        },
+      }),
     ];
     console.log("Creating QdrantDB vector store");
     const qdrantVs = new QdrantVectorStore({ url: qdrantUrl, collectionName });
@@ -73,6 +79,42 @@ async function main() {
       query: "What is the color of the dog?",
     });
     console.log("Filter with dogId 2 response:", response.toString());
+
+    console.log("Querying index with dogId !=2: Expected output: Not red");
+    const queryEngineNotDogId2 = index.asQueryEngine({
+      preFilters: {
+        filters: [
+          {
+            key: "dogId",
+            value: "2",
+            operator: "!=",
+          },
+        ],
+      },
+    });
+    const responseNotDogId2 = await queryEngineNotDogId2.query({
+      query: "What is the color of the dog?",
+    });
+    console.log(responseNotDogId2.toString());
+
+    console.log(
+      "Querying index with dogId 2 or 3: Expected output: Red, Black",
+    );
+    const queryEngineIn = index.asQueryEngine({
+      preFilters: {
+        filters: [
+          {
+            key: "dogId",
+            value: ["2", "3"],
+            operator: "in",
+          },
+        ],
+      },
+    });
+    const responseIn = await queryEngineIn.query({
+      query: "List all dogs",
+    });
+    console.log(responseIn.toString());
   } catch (e) {
     console.error(e);
   }

examples/vectorIndexCustomize.ts

@@ -25,12 +25,9 @@ async function main() {
     similarityCutoff: 0.7,
   });
   // TODO: cannot pass responseSynthesizer into retriever query engine
-  const queryEngine = new RetrieverQueryEngine(
-    retriever,
-    undefined,
-    undefined,
-    [nodePostprocessor],
-  );
+  const queryEngine = new RetrieverQueryEngine(retriever, undefined, [
+    nodePostprocessor,
+  ]);
 
   const response = await queryEngine.query({
     query: "What did the author do growing up?",

examples/vectorIndexFromVectorStore.ts

@@ -1,4 +1,5 @@
 import {
+  BaseVectorStore,
   getResponseSynthesizer,
   OpenAI,
   OpenAIEmbedding,
@@ -6,7 +7,6 @@ import {
   Settings,
   TextNode,
   VectorIndexRetriever,
-  VectorStore,
   VectorStoreIndex,
   VectorStoreQuery,
   VectorStoreQueryResult,
@@ -24,7 +24,7 @@ Settings.llm = new OpenAI({
  * Please do not use this class in production; it's only for demonstration purposes.
  */
 class PineconeVectorStore<T extends RecordMetadata = RecordMetadata>
-  implements VectorStore
+  implements BaseVectorStore
 {
   storesText = true;
   isEmbeddingQuery = false;
@@ -165,9 +165,7 @@ async function main() {
     });
 
     const responseSynthesizer = getResponseSynthesizer("tree_summarize");
-    return new RetrieverQueryEngine(retriever, responseSynthesizer, {
-      filter,
-    });
+    return new RetrieverQueryEngine(retriever, responseSynthesizer);
   };
 
   // whatever is a key from your metadata

packages/autotool/CHANGELOG.md

@@ -1,5 +1,108 @@
 # @llamaindex/autotool
 
+## 4.0.1
+
+### Patch Changes
+
+- a75af83: refactor: move some llm and embedding to single package
+- Updated dependencies [ae49ff4]
+- Updated dependencies [4c38c1b]
+- Updated dependencies [a75af83]
+- Updated dependencies [a75af83]
+  - llamaindex@0.7.1
+
+## 4.0.0
+
+### Patch Changes
+
+- Updated dependencies [1364e8e]
+- Updated dependencies [3b7736f]
+- Updated dependencies [96fc69c]
+  - llamaindex@0.7.0
+
+## 3.0.22
+
+### Patch Changes
+
+- Updated dependencies [5729bd9]
+  - llamaindex@0.6.22
+
+## 3.0.21
+
+### Patch Changes
+
+- Updated dependencies [6f75306]
+- Updated dependencies [94cb4ad]
+  - llamaindex@0.6.21
+
+## 3.0.20
+
+### Patch Changes
+
+- Updated dependencies [6a9a7b1]
+  - llamaindex@0.6.20
+
+## 3.0.19
+
+### Patch Changes
+
+- Updated dependencies [62cba52]
+- Updated dependencies [d265e96]
+- Updated dependencies [d30bbf7]
+- Updated dependencies [53fd00a]
+  - llamaindex@0.6.19
+
+## 3.0.18
+
+### Patch Changes
+
+- Updated dependencies [5f67820]
+- Updated dependencies [fe08d04]
+  - llamaindex@0.6.18
+
+## 3.0.17
+
+### Patch Changes
+
+- Updated dependencies [ee697fb]
+  - llamaindex@0.6.17
+
+## 3.0.16
+
+### Patch Changes
+
+- Updated dependencies [63e9846]
+- Updated dependencies [6f3a31c]
+  - llamaindex@0.6.16
+
+## 3.0.15
+
+### Patch Changes
+
+- Updated dependencies [2a82413]
+  - llamaindex@0.6.15
+
+## 3.0.14
+
+### Patch Changes
+
+- llamaindex@0.6.14
+
+## 3.0.13
+
+### Patch Changes
+
+- llamaindex@0.6.13
+
+## 3.0.12
+
+### Patch Changes
+
+- Updated dependencies [f7b4e94]
+- Updated dependencies [78037a6]
+- Updated dependencies [1d9e3b1]
+  - llamaindex@0.6.12
+
 ## 3.0.11
 
 ### Patch Changes

packages/autotool/examples/01_node/CHANGELOG.md

@@ -1,5 +1,120 @@
 # @llamaindex/autotool-01-node-example
 
+## 0.0.33
+
+### Patch Changes
+
+- Updated dependencies [ae49ff4]
+- Updated dependencies [4c38c1b]
+- Updated dependencies [a75af83]
+- Updated dependencies [a75af83]
+  - llamaindex@0.7.1
+  - @llamaindex/autotool@4.0.1
+
+## 0.0.32
+
+### Patch Changes
+
+- Updated dependencies [1364e8e]
+- Updated dependencies [3b7736f]
+- Updated dependencies [96fc69c]
+  - llamaindex@0.7.0
+  - @llamaindex/autotool@4.0.0
+
+## 0.0.31
+
+### Patch Changes
+
+- Updated dependencies [5729bd9]
+  - llamaindex@0.6.22
+  - @llamaindex/autotool@3.0.22
+
+## 0.0.30
+
+### Patch Changes
+
+- Updated dependencies [6f75306]
+- Updated dependencies [94cb4ad]
+  - llamaindex@0.6.21
+  - @llamaindex/autotool@3.0.21
+
+## 0.0.29
+
+### Patch Changes
+
+- Updated dependencies [6a9a7b1]
+  - llamaindex@0.6.20
+  - @llamaindex/autotool@3.0.20
+
+## 0.0.28
+
+### Patch Changes
+
+- Updated dependencies [62cba52]
+- Updated dependencies [d265e96]
+- Updated dependencies [d30bbf7]
+- Updated dependencies [53fd00a]
+  - llamaindex@0.6.19
+  - @llamaindex/autotool@3.0.19
+
+## 0.0.27
+
+### Patch Changes
+
+- Updated dependencies [5f67820]
+- Updated dependencies [fe08d04]
+  - llamaindex@0.6.18
+  - @llamaindex/autotool@3.0.18
+
+## 0.0.26
+
+### Patch Changes
+
+- Updated dependencies [ee697fb]
+  - llamaindex@0.6.17
+  - @llamaindex/autotool@3.0.17
+
+## 0.0.25
+
+### Patch Changes
+
+- Updated dependencies [63e9846]
+- Updated dependencies [6f3a31c]
+  - llamaindex@0.6.16
+  - @llamaindex/autotool@3.0.16
+
+## 0.0.24
+
+### Patch Changes
+
+- Updated dependencies [2a82413]
+  - llamaindex@0.6.15
+  - @llamaindex/autotool@3.0.15
+
+## 0.0.23
+
+### Patch Changes
+
+- llamaindex@0.6.14
+- @llamaindex/autotool@3.0.14
+
+## 0.0.22
+
+### Patch Changes
+
+- llamaindex@0.6.13
+- @llamaindex/autotool@3.0.13
+
+## 0.0.21
+
+### Patch Changes
+
+- Updated dependencies [f7b4e94]
+- Updated dependencies [78037a6]
+- Updated dependencies [1d9e3b1]
+  - llamaindex@0.6.12
+  - @llamaindex/autotool@3.0.12
+
 ## 0.0.20
 
 ### Patch Changes

packages/autotool/examples/01_node/package.json

@@ -13,5 +13,5 @@
   "scripts": {
     "start": "node --import tsx --import @llamaindex/autotool/node ./src/index.ts"
   },
-  "version": "0.0.20"
+  "version": "0.0.33"
 }

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

@@ -1,5 +1,120 @@
 # @llamaindex/autotool-02-next-example
 
+## 0.1.77
+
+### Patch Changes
+
+- Updated dependencies [ae49ff4]
+- Updated dependencies [4c38c1b]
+- Updated dependencies [a75af83]
+- Updated dependencies [a75af83]
+  - llamaindex@0.7.1
+  - @llamaindex/autotool@4.0.1
+
+## 0.1.76
+
+### Patch Changes
+
+- Updated dependencies [1364e8e]
+- Updated dependencies [3b7736f]
+- Updated dependencies [96fc69c]
+  - llamaindex@0.7.0
+  - @llamaindex/autotool@4.0.0
+
+## 0.1.75
+
+### Patch Changes
+
+- Updated dependencies [5729bd9]
+  - llamaindex@0.6.22
+  - @llamaindex/autotool@3.0.22
+
+## 0.1.74
+
+### Patch Changes
+
+- Updated dependencies [6f75306]
+- Updated dependencies [94cb4ad]
+  - llamaindex@0.6.21
+  - @llamaindex/autotool@3.0.21
+
+## 0.1.73
+
+### Patch Changes
+
+- Updated dependencies [6a9a7b1]
+  - llamaindex@0.6.20
+  - @llamaindex/autotool@3.0.20
+
+## 0.1.72
+
+### Patch Changes
+
+- Updated dependencies [62cba52]
+- Updated dependencies [d265e96]
+- Updated dependencies [d30bbf7]
+- Updated dependencies [53fd00a]
+  - llamaindex@0.6.19
+  - @llamaindex/autotool@3.0.19
+
+## 0.1.71
+
+### Patch Changes
+
+- Updated dependencies [5f67820]
+- Updated dependencies [fe08d04]
+  - llamaindex@0.6.18
+  - @llamaindex/autotool@3.0.18
+
+## 0.1.70
+
+### Patch Changes
+
+- Updated dependencies [ee697fb]
+  - llamaindex@0.6.17
+  - @llamaindex/autotool@3.0.17
+
+## 0.1.69
+
+### Patch Changes
+
+- Updated dependencies [63e9846]
+- Updated dependencies [6f3a31c]
+  - llamaindex@0.6.16
+  - @llamaindex/autotool@3.0.16
+
+## 0.1.68
+
+### Patch Changes
+
+- Updated dependencies [2a82413]
+  - llamaindex@0.6.15
+  - @llamaindex/autotool@3.0.15
+
+## 0.1.67
+
+### Patch Changes
+
+- llamaindex@0.6.14
+- @llamaindex/autotool@3.0.14
+
+## 0.1.66
+
+### Patch Changes
+
+- llamaindex@0.6.13
+- @llamaindex/autotool@3.0.13
+
+## 0.1.65
+
+### Patch Changes
+
+- Updated dependencies [f7b4e94]
+- Updated dependencies [78037a6]
+- Updated dependencies [1d9e3b1]
+  - llamaindex@0.6.12
+  - @llamaindex/autotool@3.0.12
+
 ## 0.1.64
 
 ### Patch Changes

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

@@ -1,7 +1,7 @@
 {
   "name": "@llamaindex/autotool-02-next-example",
   "private": true,
-  "version": "0.1.64",
+  "version": "0.1.77",
   "scripts": {
     "dev": "next dev",
     "build": "next build",

packages/autotool/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@llamaindex/autotool",
   "type": "module",
-  "version": "3.0.11",
+  "version": "4.0.1",
   "description": "auto transpile your JS function to LLM Agent compatible",
   "files": [
     "dist",
@@ -70,7 +70,7 @@
     "@swc/types": "^0.1.12",
     "@types/json-schema": "^7.0.15",
     "@types/node": "^22.5.1",
-    "bunchee": "5.3.2",
+    "bunchee": "5.5.1",
     "llamaindex": "workspace:*",
     "next": "14.2.11",
     "rollup": "^4.21.2",

packages/cloud/CHANGELOG.md

@@ -1,5 +1,59 @@
 # @llamaindex/cloud
 
+## 1.0.1
+
+### Patch Changes
+
+- 4c38c1b: fix(cloud): do not detect file type in llama parse
+- 24d065f: Log Parse Job Errors when verbose is enabled
+- a75af83: refactor: move some llm and embedding to single package
+- Updated dependencies [ae49ff4]
+- Updated dependencies [a75af83]
+  - @llamaindex/env@0.1.14
+  - @llamaindex/core@0.3.1
+
+## 1.0.0
+
+### Patch Changes
+
+- Updated dependencies [1364e8e]
+- Updated dependencies [96fc69c]
+  - @llamaindex/core@0.3.0
+
+## 0.2.14
+
+### Patch Changes
+
+- Updated dependencies [5f67820]
+  - @llamaindex/core@0.2.12
+
+## 0.2.13
+
+### Patch Changes
+
+- Updated dependencies [ee697fb]
+  - @llamaindex/core@0.2.11
+
+## 0.2.12
+
+### Patch Changes
+
+- Updated dependencies [3489e7d]
+- Updated dependencies [468bda5]
+  - @llamaindex/core@0.2.10
+
+## 0.2.11
+
+### Patch Changes
+
+- 0b20ff9: fix: package.json format
+
+## 0.2.10
+
+### Patch Changes
+
+- 981811e: fix(cloud): llama parse reader save image incorrectly
+
 ## 0.2.9
 
 ### Patch Changes

packages/cloud/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@llamaindex/cloud",
-  "version": "0.2.9",
+  "version": "1.0.1",
   "type": "module",
   "license": "MIT",
   "scripts": {
@@ -51,15 +51,12 @@
   "devDependencies": {
     "@hey-api/client-fetch": "^0.2.4",
     "@hey-api/openapi-ts": "^0.53.0",
-    "@llamaindex/core": "workspace:^0.2.8",
-    "@llamaindex/env": "workspace:^0.1.13",
-    "bunchee": "5.3.2"
+    "@llamaindex/core": "workspace:*",
+    "@llamaindex/env": "workspace:*",
+    "bunchee": "5.5.1"
   },
   "peerDependencies": {
-    "@llamaindex/core": "workspace:^0.2.8",
-    "@llamaindex/env": "workspace:^0.1.13"
-  },
-  "dependencies": {
-    "magic-bytes.js": "^1.10.0"
+    "@llamaindex/core": "workspace:*",
+    "@llamaindex/env": "workspace:*"
   }
 }

packages/cloud/src/reader.ts

@@ -1,11 +1,10 @@
-import { createClient, createConfig, type Client } from "@hey-api/client-fetch";
+import { type Client, createClient, createConfig } from "@hey-api/client-fetch";
 import { Document, FileReader } from "@llamaindex/core/schema";
-import { fs, getEnv } from "@llamaindex/env";
-import { filetypeinfo } from "magic-bytes.js";
+import { fs, getEnv, path } from "@llamaindex/env";
 import {
-  ParsingService,
   type Body_upload_file_api_v1_parsing_upload_post,
   type ParserLanguages,
+  ParsingService,
 } from "./api";
 import { sleep } from "./utils";
 
@@ -13,99 +12,6 @@ export type Language = ParserLanguages;
 
 export type ResultType = "text" | "markdown" | "json";
 
-const SUPPORT_FILE_EXT: string[] = [
-  ".pdf",
-  // document and presentations
-  ".602",
-  ".abw",
-  ".cgm",
-  ".cwk",
-  ".doc",
-  ".docx",
-  ".docm",
-  ".dot",
-  ".dotm",
-  ".hwp",
-  ".key",
-  ".lwp",
-  ".mw",
-  ".mcw",
-  ".pages",
-  ".pbd",
-  ".ppt",
-  ".pptm",
-  ".pptx",
-  ".pot",
-  ".potm",
-  ".potx",
-  ".rtf",
-  ".sda",
-  ".sdd",
-  ".sdp",
-  ".sdw",
-  ".sgl",
-  ".sti",
-  ".sxi",
-  ".sxw",
-  ".stw",
-  ".sxg",
-  ".txt",
-  ".uof",
-  ".uop",
-  ".uot",
-  ".vor",
-  ".wpd",
-  ".wps",
-  ".xml",
-  ".zabw",
-  ".epub",
-  // images
-  ".jpg",
-  ".jpeg",
-  ".png",
-  ".gif",
-  ".bmp",
-  ".svg",
-  ".tiff",
-  ".webp",
-  // web
-  ".htm",
-  ".html",
-  // spreadsheets
-  ".xlsx",
-  ".xls",
-  ".xlsm",
-  ".xlsb",
-  ".xlw",
-  ".csv",
-  ".dif",
-  ".sylk",
-  ".slk",
-  ".prn",
-  ".numbers",
-  ".et",
-  ".ods",
-  ".fods",
-  ".uos1",
-  ".uos2",
-  ".dbf",
-  ".wk1",
-  ".wk2",
-  ".wk3",
-  ".wk4",
-  ".wks",
-  ".123",
-  ".wq1",
-  ".wq2",
-  ".wb1",
-  ".wb2",
-  ".wb3",
-  ".qpw",
-  ".xlr",
-  ".eth",
-  ".tsv",
-];
-
 //todo: should move into @llamaindex/env
 type WriteStream = {
   write: (text: string) => void;
@@ -239,17 +145,12 @@ export class LlamaParseReader extends FileReader {
 
   // Create a job for the LlamaParse API
   private async createJob(data: Uint8Array): Promise<string> {
-    // Load data, set the mime type
-    const { mime } = await LlamaParseReader.getMimeType(data);
-
     if (this.verbose) {
       console.log("Started uploading the file");
     }
 
     const body = {
-      file: new Blob([data], {
-        type: mime,
-      }),
+      file: new Blob([data]),
       language: this.language,
       parsing_instruction: this.parsingInstruction,
       skip_diagonal_text: this.skipDiagonalText,
@@ -368,6 +269,11 @@ export class LlamaParseReader extends FileReader {
         }
         tries++;
       } else {
+        if (this.verbose) {
+          console.error(
+            `Recieved Error response ${status} for job ${jobId}.  Got Error Code: ${data.error_code} and Error Message: ${data.error_message}`,
+          );
+        }
         throw new Error(
           `Failed to parse the file: ${jobId}, status: ${status}`,
         );
@@ -510,14 +416,7 @@ export class LlamaParseReader extends FileReader {
     jobId: string,
     imageName: string,
   ): Promise<string> {
-    // Get the full path
-    let imagePath = `${downloadPath}/${jobId}-${imageName}`;
-    // Get a valid image path
-    if (!imagePath.endsWith(".png") && !imagePath.endsWith(".jpg")) {
-      imagePath += ".png";
-    }
-
-    return imagePath;
+    return path.join(downloadPath, `${jobId}-${imageName}`);
   }
 
   private async fetchAndSaveImage(
@@ -538,10 +437,9 @@ export class LlamaParseReader extends FileReader {
     if (response.error) {
       throw new Error(`Failed to download image: ${response.error.detail}`);
     }
-    const arrayBuffer = (await response.data) as ArrayBuffer;
-    const buffer = new Uint8Array(arrayBuffer);
+    const blob = (await response.data) as Blob;
     // Write the image buffer to the specified imagePath
-    await fs.writeFile(imagePath, buffer);
+    await fs.writeFile(imagePath, new Uint8Array(await blob.arrayBuffer()));
   }
 
   // Filters out invalid values (null, undefined, empty string) of specific params.
@@ -572,24 +470,4 @@ export class LlamaParseReader extends FileReader {
         }),
     );
   }
-
-  static async getMimeType(
-    data: Uint8Array,
-  ): Promise<{ mime: string; extension: string }> {
-    const typeinfos = filetypeinfo(data);
-    // find the first type info that matches the supported MIME types
-    // It could be happened that docx file is recognized as zip file, so we need to check the mime type
-    const info = typeinfos.find((info) => {
-      if (info.extension && SUPPORT_FILE_EXT.includes(`.${info.extension}`)) {
-        return info;
-      }
-    });
-    if (!info || !info.mime || !info.extension) {
-      const ext = SUPPORT_FILE_EXT.join(", ");
-      throw new Error(
-        `File has type which does not match supported MIME Types. Supported formats include: ${ext}`,
-      );
-    }
-    return { mime: info.mime, extension: info.extension };
-  }
 }

packages/community/CHANGELOG.md

@@ -1,5 +1,58 @@
 # @llamaindex/community
 
+## 0.0.49
+
+### Patch Changes
+
+- a75af83: refactor: move some llm and embedding to single package
+- Updated dependencies [ae49ff4]
+- Updated dependencies [a75af83]
+  - @llamaindex/env@0.1.14
+  - @llamaindex/core@0.3.1
+
+## 0.0.48
+
+### Patch Changes
+
+- Updated dependencies [1364e8e]
+- Updated dependencies [96fc69c]
+  - @llamaindex/core@0.3.0
+
+## 0.0.47
+
+### Patch Changes
+
+- Updated dependencies [5f67820]
+  - @llamaindex/core@0.2.12
+
+## 0.0.46
+
+### Patch Changes
+
+- Updated dependencies [ee697fb]
+  - @llamaindex/core@0.2.11
+
+## 0.0.45
+
+### Patch Changes
+
+- Updated dependencies [3489e7d]
+- Updated dependencies [468bda5]
+  - @llamaindex/core@0.2.10
+
+## 0.0.44
+
+### Patch Changes
+
+- Updated dependencies [b17d439]
+  - @llamaindex/core@0.2.9
+
+## 0.0.43
+
+### Patch Changes
+
+- 2774e80: feat: added meta3.2 support via Bedrock including vision, tool call and inference region support
+
 ## 0.0.42
 
 ### Patch Changes

packages/community/README.md

@@ -5,9 +5,11 @@
 ## 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, 3 and 3.1 Models [usage](https://ts.llamaindex.ai/modules/llms/available_llms/bedrock)
-- Meta LLama3.1 405b tool call support
+- Bedrock support for the Meta LLama 2, 3, 3.1 and 3.2 Models [usage](https://ts.llamaindex.ai/modules/llms/available_llms/bedrock)
+- Meta LLama3.1 405b and Llama3.2 tool call support
+- Meta 3.2 11B and 90B vision support
 - Bedrock support for querying Knowledge Base
+- Bedrock: [Supported Regions and models for cross-region inference](https://docs.aws.amazon.com/bedrock/latest/userguide/cross-region-inference-support.html)
 
 ## LICENSE
 

packages/community/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@llamaindex/community",
   "description": "Community package for LlamaIndexTS",
-  "version": "0.0.42",
+  "version": "0.0.49",
   "type": "module",
   "types": "dist/type/index.d.ts",
   "main": "dist/cjs/index.js",
@@ -43,11 +43,11 @@
   },
   "devDependencies": {
     "@types/node": "^22.5.1",
-    "bunchee": "5.3.2"
+    "bunchee": "5.5.1"
   },
   "dependencies": {
-    "@aws-sdk/client-bedrock-runtime": "^3.642.0",
     "@aws-sdk/client-bedrock-agent-runtime": "^3.642.0",
+    "@aws-sdk/client-bedrock-runtime": "^3.642.0",
     "@llamaindex/core": "workspace:*",
     "@llamaindex/env": "workspace:*"
   }

packages/community/src/index.ts

@@ -2,5 +2,7 @@ export {
   BEDROCK_MODELS,
   BEDROCK_MODEL_MAX_TOKENS,
   Bedrock,
+  INFERENCE_BEDROCK_MODELS,
+  INFERENCE_TO_BEDROCK_MAP,
 } from "./llm/bedrock/index.js";
 export { AmazonKnowledgeBaseRetriever } from "./retrievers/bedrock.js";

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

@@ -6,7 +6,10 @@ import type {
   MessageContentDetail,
   ToolCallLLMMessageOptions,
 } from "@llamaindex/core/llms";
-import { mapMessageContentToMessageContentDetails } from "../utils";
+import {
+  extractDataUrlComponents,
+  mapMessageContentToMessageContentDetails,
+} from "../utils";
 import type {
   AnthropicContent,
   AnthropicImageContent,
@@ -143,27 +146,6 @@ export const mapTextContent = (text: string): AnthropicTextContent => {
   return { type: "text", text };
 };
 
-export const extractDataUrlComponents = (
-  dataUrl: string,
-): {
-  mimeType: string;
-  base64: string;
-} => {
-  const parts = dataUrl.split(";base64,");
-
-  if (parts.length !== 2 || !parts[0]!.startsWith("data:")) {
-    throw new Error("Invalid data URL");
-  }
-
-  const mimeType = parts[0]!.slice(5);
-  const base64 = parts[1]!;
-
-  return {
-    mimeType,
-    base64,
-  };
-};
-
 export const mapImageContent = (imageUrl: string): AnthropicImageContent => {
   if (!imageUrl.startsWith("data:"))
     throw new Error(

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

@@ -47,35 +47,96 @@ export type BedrockChatParamsNonStreaming = LLMChatParamsNonStreaming<
 export type BedrockChatNonStreamResponse =
   ChatResponse<ToolCallLLMMessageOptions>;
 
-export enum BEDROCK_MODELS {
-  AMAZON_TITAN_TG1_LARGE = "amazon.titan-tg1-large",
-  AMAZON_TITAN_TEXT_EXPRESS_V1 = "amazon.titan-text-express-v1",
-  AI21_J2_GRANDE_INSTRUCT = "ai21.j2-grande-instruct",
-  AI21_J2_JUMBO_INSTRUCT = "ai21.j2-jumbo-instruct",
-  AI21_J2_MID = "ai21.j2-mid",
-  AI21_J2_MID_V1 = "ai21.j2-mid-v1",
-  AI21_J2_ULTRA = "ai21.j2-ultra",
-  AI21_J2_ULTRA_V1 = "ai21.j2-ultra-v1",
-  COHERE_COMMAND_TEXT_V14 = "cohere.command-text-v14",
-  ANTHROPIC_CLAUDE_INSTANT_1 = "anthropic.claude-instant-v1",
-  ANTHROPIC_CLAUDE_1 = "anthropic.claude-v1", // EOF: No longer supported
-  ANTHROPIC_CLAUDE_2 = "anthropic.claude-v2",
-  ANTHROPIC_CLAUDE_2_1 = "anthropic.claude-v2:1",
-  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",
-  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",
-  META_LLAMA3_1_8B_INSTRUCT = "meta.llama3-1-8b-instruct-v1:0",
-  META_LLAMA3_1_70B_INSTRUCT = "meta.llama3-1-70b-instruct-v1:0",
-  META_LLAMA3_1_405B_INSTRUCT = "meta.llama3-1-405b-instruct-v1:0",
-  MISTRAL_7B_INSTRUCT = "mistral.mistral-7b-instruct-v0:2",
-  MISTRAL_MIXTRAL_7B_INSTRUCT = "mistral.mixtral-8x7b-instruct-v0:1",
-  MISTRAL_MIXTRAL_LARGE_2402 = "mistral.mistral-large-2402-v1:0",
-}
+export const BEDROCK_MODELS = {
+  AMAZON_TITAN_TG1_LARGE: "amazon.titan-tg1-large",
+  AMAZON_TITAN_TEXT_EXPRESS_V1: "amazon.titan-text-express-v1",
+  AI21_J2_GRANDE_INSTRUCT: "ai21.j2-grande-instruct",
+  AI21_J2_JUMBO_INSTRUCT: "ai21.j2-jumbo-instruct",
+  AI21_J2_MID: "ai21.j2-mid",
+  AI21_J2_MID_V1: "ai21.j2-mid-v1",
+  AI21_J2_ULTRA: "ai21.j2-ultra",
+  AI21_J2_ULTRA_V1: "ai21.j2-ultra-v1",
+  COHERE_COMMAND_TEXT_V14: "cohere.command-text-v14",
+  ANTHROPIC_CLAUDE_INSTANT_1: "anthropic.claude-instant-v1",
+  ANTHROPIC_CLAUDE_1: "anthropic.claude-v1", // EOF: No longer supported
+  ANTHROPIC_CLAUDE_2: "anthropic.claude-v2",
+  ANTHROPIC_CLAUDE_2_1: "anthropic.claude-v2:1",
+  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",
+  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",
+  META_LLAMA3_1_8B_INSTRUCT: "meta.llama3-1-8b-instruct-v1:0",
+  META_LLAMA3_1_70B_INSTRUCT: "meta.llama3-1-70b-instruct-v1:0",
+  META_LLAMA3_1_405B_INSTRUCT: "meta.llama3-1-405b-instruct-v1:0",
+  META_LLAMA3_2_1B_INSTRUCT: "meta.llama3-2-1b-instruct-v1:0",
+  META_LLAMA3_2_3B_INSTRUCT: "meta.llama3-2-3b-instruct-v1:0",
+  META_LLAMA3_2_11B_INSTRUCT: "meta.llama3-2-11b-instruct-v1:0",
+  META_LLAMA3_2_90B_INSTRUCT: "meta.llama3-2-90b-instruct-v1:0",
+  MISTRAL_7B_INSTRUCT: "mistral.mistral-7b-instruct-v0:2",
+  MISTRAL_MIXTRAL_7B_INSTRUCT: "mistral.mixtral-8x7b-instruct-v0:1",
+  MISTRAL_MIXTRAL_LARGE_2402: "mistral.mistral-large-2402-v1:0",
+};
+export type BEDROCK_MODELS =
+  (typeof BEDROCK_MODELS)[keyof typeof BEDROCK_MODELS];
+
+export const INFERENCE_BEDROCK_MODELS = {
+  US_ANTHROPIC_CLAUDE_3_HAIKU: "us.anthropic.claude-3-haiku-20240307-v1:0",
+  US_ANTHROPIC_CLAUDE_3_OPUS: "us.anthropic.claude-3-opus-20240229-v1:0",
+  US_ANTHROPIC_CLAUDE_3_SONNET: "us.anthropic.claude-3-sonnet-20240229-v1:0",
+  US_ANTHROPIC_CLAUDE_3_5_SONNET:
+    "us.anthropic.claude-3-5-sonnet-20240620-v1:0",
+  US_META_LLAMA_3_2_1B_INSTRUCT: "us.meta.llama3-2-1b-instruct-v1:0",
+  US_META_LLAMA_3_2_3B_INSTRUCT: "us.meta.llama3-2-3b-instruct-v1:0",
+  US_META_LLAMA_3_2_11B_INSTRUCT: "us.meta.llama3-2-11b-instruct-v1:0",
+  US_META_LLAMA_3_2_90B_INSTRUCT: "us.meta.llama3-2-90b-instruct-v1:0",
+
+  EU_ANTHROPIC_CLAUDE_3_HAIKU: "eu.anthropic.claude-3-haiku-20240307-v1:0",
+  EU_ANTHROPIC_CLAUDE_3_SONNET: "eu.anthropic.claude-3-sonnet-20240229-v1:0",
+  EU_ANTHROPIC_CLAUDE_3_5_SONNET:
+    "eu.anthropic.claude-3-5-sonnet-20240620-v1:0",
+  EU_META_LLAMA_3_2_1B_INSTRUCT: "eu.meta.llama3-2-1b-instruct-v1:0",
+  EU_META_LLAMA_3_2_3B_INSTRUCT: "eu.meta.llama3-2-3b-instruct-v1:0",
+};
+
+export type INFERENCE_BEDROCK_MODELS =
+  (typeof INFERENCE_BEDROCK_MODELS)[keyof typeof INFERENCE_BEDROCK_MODELS];
+
+export const INFERENCE_TO_BEDROCK_MAP: Record<
+  INFERENCE_BEDROCK_MODELS,
+  BEDROCK_MODELS
+> = {
+  [INFERENCE_BEDROCK_MODELS.US_ANTHROPIC_CLAUDE_3_HAIKU]:
+    BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_HAIKU,
+  [INFERENCE_BEDROCK_MODELS.US_ANTHROPIC_CLAUDE_3_OPUS]:
+    BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_OPUS,
+  [INFERENCE_BEDROCK_MODELS.US_ANTHROPIC_CLAUDE_3_SONNET]:
+    BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_SONNET,
+  [INFERENCE_BEDROCK_MODELS.US_ANTHROPIC_CLAUDE_3_5_SONNET]:
+    BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_5_SONNET,
+  [INFERENCE_BEDROCK_MODELS.US_META_LLAMA_3_2_1B_INSTRUCT]:
+    BEDROCK_MODELS.META_LLAMA3_2_1B_INSTRUCT,
+  [INFERENCE_BEDROCK_MODELS.US_META_LLAMA_3_2_3B_INSTRUCT]:
+    BEDROCK_MODELS.META_LLAMA3_2_3B_INSTRUCT,
+  [INFERENCE_BEDROCK_MODELS.US_META_LLAMA_3_2_11B_INSTRUCT]:
+    BEDROCK_MODELS.META_LLAMA3_2_11B_INSTRUCT,
+  [INFERENCE_BEDROCK_MODELS.US_META_LLAMA_3_2_90B_INSTRUCT]:
+    BEDROCK_MODELS.META_LLAMA3_2_90B_INSTRUCT,
+
+  [INFERENCE_BEDROCK_MODELS.EU_ANTHROPIC_CLAUDE_3_HAIKU]:
+    BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_HAIKU,
+  [INFERENCE_BEDROCK_MODELS.EU_ANTHROPIC_CLAUDE_3_SONNET]:
+    BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_SONNET,
+  [INFERENCE_BEDROCK_MODELS.EU_ANTHROPIC_CLAUDE_3_5_SONNET]:
+    BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_5_SONNET,
+  [INFERENCE_BEDROCK_MODELS.EU_META_LLAMA_3_2_1B_INSTRUCT]:
+    BEDROCK_MODELS.META_LLAMA3_2_1B_INSTRUCT,
+  [INFERENCE_BEDROCK_MODELS.EU_META_LLAMA_3_2_3B_INSTRUCT]:
+    BEDROCK_MODELS.META_LLAMA3_2_3B_INSTRUCT,
+};
 
 /*
  * Values taken from https://docs.aws.amazon.com/bedrock/latest/userguide/model-parameters.html#model-parameters-claude
@@ -109,6 +170,10 @@ const CHAT_ONLY_MODELS = {
   [BEDROCK_MODELS.META_LLAMA3_1_8B_INSTRUCT]: 128000,
   [BEDROCK_MODELS.META_LLAMA3_1_70B_INSTRUCT]: 128000,
   [BEDROCK_MODELS.META_LLAMA3_1_405B_INSTRUCT]: 128000,
+  [BEDROCK_MODELS.META_LLAMA3_2_1B_INSTRUCT]: 131000,
+  [BEDROCK_MODELS.META_LLAMA3_2_3B_INSTRUCT]: 131000,
+  [BEDROCK_MODELS.META_LLAMA3_2_11B_INSTRUCT]: 128000,
+  [BEDROCK_MODELS.META_LLAMA3_2_90B_INSTRUCT]: 128000,
   [BEDROCK_MODELS.MISTRAL_7B_INSTRUCT]: 32000,
   [BEDROCK_MODELS.MISTRAL_MIXTRAL_7B_INSTRUCT]: 32000,
   [BEDROCK_MODELS.MISTRAL_MIXTRAL_LARGE_2402]: 32000,
@@ -139,17 +204,25 @@ export const STREAMING_MODELS = new Set([
   BEDROCK_MODELS.META_LLAMA3_1_8B_INSTRUCT,
   BEDROCK_MODELS.META_LLAMA3_1_70B_INSTRUCT,
   BEDROCK_MODELS.META_LLAMA3_1_405B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_2_1B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_2_3B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_2_11B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_2_90B_INSTRUCT,
   BEDROCK_MODELS.MISTRAL_7B_INSTRUCT,
   BEDROCK_MODELS.MISTRAL_MIXTRAL_7B_INSTRUCT,
   BEDROCK_MODELS.MISTRAL_MIXTRAL_LARGE_2402,
 ]);
 
-export const TOOL_CALL_MODELS = [
+export const TOOL_CALL_MODELS: BEDROCK_MODELS[] = [
   BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_SONNET,
   BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_HAIKU,
   BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_OPUS,
   BEDROCK_MODELS.ANTHROPIC_CLAUDE_3_5_SONNET,
   BEDROCK_MODELS.META_LLAMA3_1_405B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_2_1B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_2_3B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_2_11B_INSTRUCT,
+  BEDROCK_MODELS.META_LLAMA3_2_90B_INSTRUCT,
 ];
 
 const getProvider = (model: string): Provider => {
@@ -166,7 +239,7 @@ const getProvider = (model: string): Provider => {
 };
 
 export type BedrockModelParams = {
-  model: keyof typeof BEDROCK_FOUNDATION_LLMS;
+  model: BEDROCK_MODELS | INFERENCE_BEDROCK_MODELS;
   temperature?: number;
   topP?: number;
   maxTokens?: number;
@@ -185,6 +258,10 @@ export const BEDROCK_MODEL_MAX_TOKENS: Partial<Record<BEDROCK_MODELS, number>> =
     [BEDROCK_MODELS.META_LLAMA3_1_8B_INSTRUCT]: 2048,
     [BEDROCK_MODELS.META_LLAMA3_1_70B_INSTRUCT]: 2048,
     [BEDROCK_MODELS.META_LLAMA3_1_405B_INSTRUCT]: 2048,
+    [BEDROCK_MODELS.META_LLAMA3_2_1B_INSTRUCT]: 2048,
+    [BEDROCK_MODELS.META_LLAMA3_2_3B_INSTRUCT]: 2048,
+    [BEDROCK_MODELS.META_LLAMA3_2_11B_INSTRUCT]: 2048,
+    [BEDROCK_MODELS.META_LLAMA3_2_90B_INSTRUCT]: 2048,
   };
 
 const DEFAULT_BEDROCK_PARAMS = {
@@ -193,14 +270,15 @@ const DEFAULT_BEDROCK_PARAMS = {
   maxTokens: 1024, // required by anthropic
 };
 
-export type BedrockParams = BedrockModelParams & BedrockRuntimeClientConfig;
+export type BedrockParams = BedrockRuntimeClientConfig & BedrockModelParams;
 
 /**
  * ToolCallLLM for Bedrock
  */
 export class Bedrock extends ToolCallLLM<BedrockAdditionalChatOptions> {
   private client: BedrockRuntimeClient;
-  model: keyof typeof BEDROCK_FOUNDATION_LLMS;
+  protected actualModel: BEDROCK_MODELS | INFERENCE_BEDROCK_MODELS;
+  model: BEDROCK_MODELS;
   temperature: number;
   topP: number;
   maxTokens?: number;
@@ -217,8 +295,8 @@ export class Bedrock extends ToolCallLLM<BedrockAdditionalChatOptions> {
     ...params
   }: BedrockParams) {
     super();
-
-    this.model = model;
+    this.actualModel = model;
+    this.model = INFERENCE_TO_BEDROCK_MAP[model] ?? model;
     this.provider = getProvider(this.model);
     this.maxTokens = maxTokens ?? DEFAULT_BEDROCK_PARAMS.maxTokens;
     this.temperature = temperature ?? DEFAULT_BEDROCK_PARAMS.temperature;
@@ -241,7 +319,7 @@ export class Bedrock extends ToolCallLLM<BedrockAdditionalChatOptions> {
       temperature: this.temperature,
       topP: this.topP,
       maxTokens: this.maxTokens,
-      contextWindow: BEDROCK_FOUNDATION_LLMS[this.model],
+      contextWindow: BEDROCK_FOUNDATION_LLMS[this.model] ?? 128000,
       tokenizer: undefined,
     };
   }
@@ -256,6 +334,8 @@ export class Bedrock extends ToolCallLLM<BedrockAdditionalChatOptions> {
       params.additionalChatOptions,
     );
     const command = new InvokeModelCommand(input);
+    command.input.modelId = this.actualModel;
+
     const response = await this.client.send(command);
     let options: ToolCallLLMMessageOptions = {};
     if (this.supportToolCall) {
@@ -287,6 +367,8 @@ export class Bedrock extends ToolCallLLM<BedrockAdditionalChatOptions> {
       params.additionalChatOptions,
     );
     const command = new InvokeModelWithResponseStreamCommand(input);
+    command.input.modelId = this.actualModel;
+
     const response = await this.client.send(command);
 
     if (response.body) yield* this.provider.reduceStream(response.body);

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

@@ -67,21 +67,26 @@ export class MetaProvider extends Provider<MetaStreamEvent> {
     for await (const response of stream) {
       const event = this.getStreamingEventResponse(response);
       const delta = this.getTextFromStreamResponse(response);
+
       // odd quirk of llama3.1, start token is \n\n
       if (
+        !toolId &&
         !event?.generation.trim() &&
         event?.generation_token_count === 1 &&
-        event.prompt_token_count !== null
+        event?.prompt_token_count !== null
       )
         continue;
 
-      if (delta === TOKENS.TOOL_CALL) {
+      if (delta.startsWith(TOKENS.TOOL_CALL)) {
         toolId = randomUUID();
+        const parts = delta.split(TOKENS.TOOL_CALL).filter((part) => part);
+        collecting.push(...parts);
         continue;
       }
 
       let options: undefined | ToolCallLLMMessageOptions = undefined;
       if (toolId && event?.stop_reason === "stop") {
+        if (delta) collecting.push(delta);
         const tool = JSON.parse(collecting.join(""));
         options = {
           toolCall: [
@@ -110,11 +115,18 @@ export class MetaProvider extends Provider<MetaStreamEvent> {
   getRequestBody<T extends ChatMessage>(
     metadata: LLMMetadata,
     messages: T[],
-    tools?: BaseTool[],
+    tools: BaseTool[] = [],
   ): InvokeModelCommandInput | InvokeModelWithResponseStreamCommandInput {
     let prompt: string = "";
+    let images: string[] = [];
     if (metadata.model.startsWith("meta.llama3")) {
-      prompt = mapChatMessagesToMetaLlama3Messages(messages, tools);
+      const mapped = mapChatMessagesToMetaLlama3Messages({
+        messages,
+        tools,
+        model: metadata.model,
+      });
+      prompt = mapped.prompt;
+      images = mapped.images;
     } else if (metadata.model.startsWith("meta.llama2")) {
       prompt = mapChatMessagesToMetaLlama2Messages(messages);
     } else {
@@ -127,6 +139,7 @@ export class MetaProvider extends Provider<MetaStreamEvent> {
       accept: "application/json",
       body: JSON.stringify({
         prompt,
+        images: images.length ? images : undefined,
         max_gen_len: metadata.maxTokens,
         temperature: metadata.temperature,
         top_p: metadata.topP,

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

@@ -1,9 +1,12 @@
 import type {
   BaseTool,
   ChatMessage,
+  LLMMetadata,
   MessageContentTextDetail,
   ToolCallLLMMessageOptions,
 } from "@llamaindex/core/llms";
+import { extractDataUrlComponents } from "../utils";
+import { TOKENS } from "./constants";
 import type { MetaMessage } from "./types";
 
 const getToolCallInstructionString = (tool: BaseTool): string => {
@@ -24,7 +27,7 @@ const getToolCallParametersString = (tool: BaseTool): string => {
 
 // ported from https://github.com/meta-llama/llama-agentic-system/blob/main/llama_agentic_system/system_prompt.py
 // NOTE: using json instead of the above xml style tool calling works more reliability
-export const getToolsPrompt = (tools?: BaseTool[]) => {
+export const getToolsPrompt_3_1 = (tools?: BaseTool[]) => {
   if (!tools?.length) return "";
 
   const customToolParams = tools.map((tool) => {
@@ -77,6 +80,46 @@ Reminder:
   `;
 };
 
+export const getToolsPrompt_3_2 = (tools?: BaseTool[]) => {
+  if (!tools?.length) return "";
+  return `
+You are an expert in composing functions. You are given a question and a set of possible functions.
+Based on the question, you will need to make one or more function/tool calls to achieve the purpose.
+If none of the function can be used, point it out. If the given question lacks the parameters required by the function,
+also point it out. You should only return the function call in tools call sections.
+
+If you decide to invoke any of the function(s), you MUST put it in the format of and start with the token: ${TOKENS.TOOL_CALL}:
+{
+  "name": function_name,
+  "parameters": parameters,
+}
+where
+
+{
+  "name": function_name,
+  "parameters": parameters, => a JSON dict with the function argument name as key and function argument value as value.
+}
+
+Here is an example,
+
+{
+  "name": "example_function_name",
+  "parameters": {"example_name": "example_value"}
+}
+
+Reminder:
+- Function calls MUST follow the specified format
+- Required parameters MUST be specified
+- Only call one function at a time
+- You SHOULD NOT include any other text in the response
+- Put the entire function call reply on one line
+
+Here is a list of functions in JSON format that you can invoke.
+
+${JSON.stringify(tools)}
+`;
+};
+
 export const mapChatRoleToMetaRole = (
   role: ChatMessage["role"],
 ): MetaMessage["role"] => {
@@ -125,16 +168,46 @@ export const mapChatMessagesToMetaMessages = <
 /**
  * Documentation at https://llama.meta.com/docs/model-cards-and-prompt-formats/meta-llama-3
  */
-export const mapChatMessagesToMetaLlama3Messages = <T extends ChatMessage>(
-  messages: T[],
-  tools?: BaseTool[],
-): string => {
+export const mapChatMessagesToMetaLlama3Messages = <T extends ChatMessage>({
+  messages,
+  model,
+  tools,
+}: {
+  messages: T[];
+  model: LLMMetadata["model"];
+  tools?: BaseTool[];
+}): { prompt: string; images: string[] } => {
+  const images: string[] = [];
+  const textMessages: T[] = [];
+
+  messages.forEach((message) => {
+    if (Array.isArray(message.content)) {
+      message.content.forEach((content) => {
+        if (content.type === "image_url") {
+          const { base64 } = extractDataUrlComponents(content.image_url.url);
+          images.push(base64);
+        } else {
+          textMessages.push(message);
+        }
+      });
+    } else {
+      textMessages.push(message);
+    }
+  });
+
   const parts: string[] = [];
-  if (tools?.length) {
+
+  let toolsPrompt = "";
+  if (model.startsWith("meta.llama3-2")) {
+    toolsPrompt = getToolsPrompt_3_2(tools);
+  } else if (model.startsWith("meta.llama3-1")) {
+    toolsPrompt = getToolsPrompt_3_1(tools);
+  }
+  if (toolsPrompt) {
     parts.push(
       "<|begin_of_text|>",
       "<|start_header_id|>system<|end_header_id|>",
-      getToolsPrompt(tools),
+      toolsPrompt,
       "<|eot_id|>",
     );
   }
@@ -154,7 +227,9 @@ export const mapChatMessagesToMetaLlama3Messages = <T extends ChatMessage>(
     ...mapped,
     "<|start_header_id|>assistant<|end_header_id|>",
   );
-  return parts.join("\n");
+
+  const prompt = parts.join("\n");
+  return { prompt, images };
 };
 
 /**

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

@@ -11,3 +11,24 @@ export const mapMessageContentToMessageContentDetails = (
 
 export const toUtf8 = (input: Uint8Array): string =>
   new TextDecoder("utf-8").decode(input);
+
+export const extractDataUrlComponents = (
+  dataUrl: string,
+): {
+  mimeType: string;
+  base64: string;
+} => {
+  const parts = dataUrl.split(";base64,");
+
+  if (parts.length !== 2 || !parts[0]!.startsWith("data:")) {
+    throw new Error("Invalid data URL");
+  }
+
+  const mimeType = parts[0]!.slice(5);
+  const base64 = parts[1]!;
+
+  return {
+    mimeType,
+    base64,
+  };
+};

packages/core/CHANGELOG.md

@@ -1,5 +1,46 @@
 # @llamaindex/core
 
+## 0.3.1
+
+### Patch Changes
+
+- a75af83: refactor: move some llm and embedding to single package
+- Updated dependencies [ae49ff4]
+- Updated dependencies [a75af83]
+  - @llamaindex/env@0.1.14
+
+## 0.3.0
+
+### Minor Changes
+
+- 1364e8e: update metadata extractors to use PromptTemplate
+- 96fc69c: add defaultQuestionExtractPrompt
+
+## 0.2.12
+
+### Patch Changes
+
+- 5f67820: Fix that node parsers generate nodes with UUIDs
+
+## 0.2.11
+
+### Patch Changes
+
+- ee697fb: fix: generate uuid when inserting to Qdrant
+
+## 0.2.10
+
+### Patch Changes
+
+- 3489e7d: fix: num output incorrect in prompt helper
+- 468bda5: fix: correct warning when chunk size smaller than 0
+
+## 0.2.9
+
+### Patch Changes
+
+- b17d439: Fix #1278: resolved issue where the id\_ was not correctly passed as the id when creating a TextNode. As a result, the upsert operation to the vector database was using a generated ID instead of the provided document ID, if available.
+
 ## 0.2.8
 
 ### Patch Changes

packages/core/data-structs/package.json

@@ -0,0 +1,8 @@
+{
+  "type": "module",
+  "main": "./dist/index.cjs",
+  "module": "./dist/index.js",
+  "types": "./dist/index.d.ts",
+  "exports": "./dist/index.js",
+  "private": true
+}

packages/core/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@llamaindex/core",
   "type": "module",
-  "version": "0.2.8",
+  "version": "0.3.1",
   "description": "LlamaIndex Core Module",
   "exports": {
     "./agent": {
@@ -258,16 +258,44 @@
     },
     "./vector-store": {
       "require": {
-        "types": "./dist/vector-store/index.d.cts",
-        "default": "./dist/vector-store/index.cjs"
+        "types": "./vector-store/dist/index.d.cts",
+        "default": "./vector-store/dist/index.cjs"
       },
       "import": {
-        "types": "./dist/vector-store/index.d.ts",
-        "default": "./dist/vector-store/index.js"
+        "types": "./vector-store/dist/index.d.ts",
+        "default": "./vector-store/dist/index.js"
       },
       "default": {
-        "types": "./dist/vector-store/index.d.ts",
-        "default": "./dist/vector-store/index.js"
+        "types": "./vector-store/dist/index.d.ts",
+        "default": "./vector-store/dist/index.js"
+      }
+    },
+    "./tools": {
+      "require": {
+        "types": "./tools/dist/index.d.cts",
+        "default": "./tools/dist/index.cjs"
+      },
+      "import": {
+        "types": "./tools/dist/index.d.ts",
+        "default": "./tools/dist/index.js"
+      },
+      "default": {
+        "types": "./tools/dist/index.d.ts",
+        "default": "./tools/dist/index.js"
+      }
+    },
+    "./data-structs": {
+      "require": {
+        "types": "./data-structs/dist/index.d.cts",
+        "default": "./data-structs/dist/index.cjs"
+      },
+      "import": {
+        "types": "./data-structs/dist/index.d.ts",
+        "default": "./data-structs/dist/index.js"
+      },
+      "default": {
+        "types": "./data-structs/dist/index.d.ts",
+        "default": "./data-structs/dist/index.js"
       }
     }
   },
@@ -289,7 +317,10 @@
     "./storage",
     "./response-synthesizers",
     "./chat-engine",
-    "./retriever"
+    "./retriever",
+    "./vector-store",
+    "./tools",
+    "./data-structs"
   ],
   "scripts": {
     "dev": "bunchee --watch",
@@ -303,7 +334,7 @@
   "devDependencies": {
     "@edge-runtime/vm": "^4.0.3",
     "ajv": "^8.17.1",
-    "bunchee": "5.3.2",
+    "bunchee": "5.5.1",
     "happy-dom": "^15.7.4",
     "natural": "^8.0.1",
     "python-format-js": "^1.4.3"
@@ -312,6 +343,7 @@
     "@llamaindex/env": "workspace:*",
     "@types/node": "^22.5.1",
     "magic-bytes.js": "^1.10.0",
-    "zod": "^3.23.8"
+    "zod": "^3.23.8",
+    "zod-to-json-schema": "^3.23.3"
   }
 }

packages/core/src/data-structs/data-structs.ts

@@ -0,0 +1,67 @@
+import { randomUUID } from "@llamaindex/env";
+import type { UUID } from "../global";
+import { IndexStructType } from "./struct-type";
+
+export abstract class IndexStruct {
+  indexId: string;
+  summary: string | undefined;
+
+  constructor(
+    indexId: UUID = randomUUID(),
+    summary: string | undefined = undefined,
+  ) {
+    this.indexId = indexId;
+    this.summary = summary;
+  }
+
+  toJson(): Record<string, unknown> {
+    return {
+      indexId: this.indexId,
+      summary: this.summary,
+    };
+  }
+
+  getSummary(): string {
+    if (this.summary === undefined) {
+      throw new Error("summary field of the index struct is not set");
+    }
+    return this.summary;
+  }
+}
+
+// A table of keywords mapping keywords to text chunks.
+export class KeywordTable extends IndexStruct {
+  table: Map<string, Set<string>> = new Map();
+  type: IndexStructType = IndexStructType.KEYWORD_TABLE;
+
+  addNode(keywords: string[], nodeId: string): void {
+    keywords.forEach((keyword) => {
+      if (!this.table.has(keyword)) {
+        this.table.set(keyword, new Set());
+      }
+      this.table.get(keyword)!.add(nodeId);
+    });
+  }
+
+  deleteNode(keywords: string[], nodeId: string) {
+    keywords.forEach((keyword) => {
+      if (this.table.has(keyword)) {
+        this.table.get(keyword)!.delete(nodeId);
+      }
+    });
+  }
+
+  toJson(): Record<string, unknown> {
+    return {
+      ...super.toJson(),
+      table: Array.from(this.table.entries()).reduce(
+        (acc, [keyword, nodeIds]) => {
+          acc[keyword] = Array.from(nodeIds);
+          return acc;
+        },
+        {} as Record<string, string[]>,
+      ),
+      type: this.type,
+    };
+  }
+}

packages/core/src/data-structs/index.ts

@@ -0,0 +1,2 @@
+export { IndexStruct, KeywordTable } from "./data-structs";
+export { IndexStructType } from "./struct-type";

packages/core/src/data-structs/struct-type.ts

@@ -0,0 +1,39 @@
+export const IndexStructType = {
+  NODE: "node",
+  TREE: "tree",
+  LIST: "list",
+  KEYWORD_TABLE: "keyword_table",
+  DICT: "dict",
+  SIMPLE_DICT: "simple_dict",
+  WEAVIATE: "weaviate",
+  PINECONE: "pinecone",
+  QDRANT: "qdrant",
+  LANCEDB: "lancedb",
+  MILVUS: "milvus",
+  CHROMA: "chroma",
+  MYSCALE: "myscale",
+  CLICKHOUSE: "clickhouse",
+  VECTOR_STORE: "vector_store",
+  OPENSEARCH: "opensearch",
+  DASHVECTOR: "dashvector",
+  CHATGPT_RETRIEVAL_PLUGIN: "chatgpt_retrieval_plugin",
+  DEEPLAKE: "deeplake",
+  EPSILLA: "epsilla",
+  MULTIMODAL_VECTOR_STORE: "multimodal",
+  SQL: "sql",
+  KG: "kg",
+  SIMPLE_KG: "simple_kg",
+  SIMPLE_LPG: "simple_lpg",
+  NEBULAGRAPH: "nebulagraph",
+  FALKORDB: "falkordb",
+  EMPTY: "empty",
+  COMPOSITE: "composite",
+  PANDAS: "pandas",
+  DOCUMENT_SUMMARY: "document_summary",
+  VECTARA: "vectara",
+  ZILLIZ_CLOUD_PIPELINE: "zilliz_cloud_pipeline",
+  POSTGRESML: "postgresml",
+} as const;
+
+export type IndexStructType =
+  (typeof IndexStructType)[keyof typeof IndexStructType];

packages/core/src/indices/prompt-helper.ts

@@ -8,18 +8,16 @@ import {
   Settings,
 } from "../global";
 import type { LLMMetadata } from "../llms";
-import { SentenceSplitter } from "../node-parser";
-import type { PromptTemplate } from "../prompts";
+import { TextSplitter, TokenTextSplitter, truncateText } from "../node-parser";
+import { BasePromptTemplate, PromptTemplate } from "../prompts";
 
 /**
  * Get the empty prompt text given a prompt.
  */
-function getEmptyPromptTxt(prompt: PromptTemplate) {
-  return prompt.format({
-    ...Object.fromEntries(
-      [...prompt.templateVars.keys()].map((key) => [key, ""]),
-    ),
-  });
+function getEmptyPromptTxt(prompt: PromptTemplate): string {
+  return prompt.format(
+    Object.fromEntries([...prompt.templateVars.keys()].map((key) => [key, ""])),
+  );
 }
 
 /**
@@ -35,24 +33,24 @@ export function getBiggestPrompt(prompts: PromptTemplate[]): PromptTemplate {
 }
 
 export type PromptHelperOptions = {
-  contextWindow?: number;
-  numOutput?: number;
-  chunkOverlapRatio?: number;
-  chunkSizeLimit?: number;
-  tokenizer?: Tokenizer;
-  separator?: string;
+  contextWindow?: number | undefined;
+  numOutput?: number | undefined;
+  chunkOverlapRatio?: number | undefined;
+  chunkSizeLimit?: number | undefined;
+  tokenizer?: Tokenizer | undefined;
+  separator?: string | undefined;
 };
 
 /**
  * A collection of helper functions for working with prompts.
  */
 export class PromptHelper {
-  contextWindow = DEFAULT_CONTEXT_WINDOW;
-  numOutput = DEFAULT_NUM_OUTPUTS;
-  chunkOverlapRatio = DEFAULT_CHUNK_OVERLAP_RATIO;
+  contextWindow: number;
+  numOutput: number;
+  chunkOverlapRatio: number;
   chunkSizeLimit: number | undefined;
   tokenizer: Tokenizer;
-  separator = " ";
+  separator: string;
 
   constructor(options: PromptHelperOptions = {}) {
     const {
@@ -72,68 +70,93 @@ export class PromptHelper {
   }
 
   /**
-   * Given a prompt, return the maximum size of the inputs to the prompt.
-   * @param prompt
-   * @returns
+   * Calculate the available context size based on the number of prompt tokens.
    */
-  private getAvailableContextSize(prompt: PromptTemplate) {
-    const emptyPromptText = getEmptyPromptTxt(prompt);
-    const promptTokens = this.tokenizer.encode(emptyPromptText);
-    const numPromptTokens = promptTokens.length;
-
-    return this.contextWindow - numPromptTokens - this.numOutput;
-  }
-
-  /**
-   * Find the maximum size of each chunk given a prompt.
-   */
-  private getAvailableChunkSize(
-    prompt: PromptTemplate,
-    numChunks = 1,
-    padding = 5,
-  ): number {
-    const availableContextSize = this.getAvailableContextSize(prompt);
-
-    const result = Math.floor(availableContextSize / numChunks) - padding;
-
-    if (this.chunkSizeLimit) {
-      return Math.min(this.chunkSizeLimit, result);
-    } else {
-      return result;
+  #getAvailableContextSize(numPromptTokens: number): number {
+    const contextSizeTokens =
+      this.contextWindow - numPromptTokens - this.numOutput;
+    if (contextSizeTokens < 0) {
+      throw new Error(
+        `Calculated available context size ${contextSizeTokens} is not non-negative.`,
+      );
     }
+    return contextSizeTokens;
   }
 
   /**
-   * Creates a text splitter with the correct chunk sizes and overlaps given a prompt.
+   * Calculate the available chunk size based on the prompt and other parameters.
+   */
+  #getAvailableChunkSize<Template extends BasePromptTemplate>(
+    prompt: Template,
+    numChunks: number = 1,
+    padding: number = 5,
+  ): number {
+    let numPromptTokens = 0;
+
+    if (prompt instanceof PromptTemplate) {
+      numPromptTokens = this.tokenizer.encode(getEmptyPromptTxt(prompt)).length;
+    }
+
+    const availableContextSize = this.#getAvailableContextSize(numPromptTokens);
+    let result = Math.floor(availableContextSize / numChunks) - padding;
+
+    if (this.chunkSizeLimit !== undefined) {
+      result = Math.min(this.chunkSizeLimit, result);
+    }
+
+    return result;
+  }
+
+  /**
+   * Creates a text splitter configured to maximally pack the available context window.
    */
   getTextSplitterGivenPrompt(
-    prompt: PromptTemplate,
-    numChunks = 1,
-    padding = DEFAULT_PADDING,
-  ) {
-    const chunkSize = this.getAvailableChunkSize(prompt, numChunks, padding);
-    if (chunkSize === 0) {
-      throw new Error("Got 0 as available chunk size");
+    prompt: BasePromptTemplate,
+    numChunks: number = 1,
+    padding: number = DEFAULT_PADDING,
+  ): TextSplitter {
+    const chunkSize = this.#getAvailableChunkSize(prompt, numChunks, padding);
+    if (chunkSize <= 0) {
+      throw new TypeError(`Chunk size ${chunkSize} is not positive.`);
     }
-    const chunkOverlap = this.chunkOverlapRatio * chunkSize;
-    return new SentenceSplitter({
+    const chunkOverlap = Math.floor(this.chunkOverlapRatio * chunkSize);
+    return new TokenTextSplitter({
+      separator: this.separator,
       chunkSize,
       chunkOverlap,
-      separator: this.separator,
       tokenizer: this.tokenizer,
     });
   }
 
   /**
-   * Repack resplits the strings based on the optimal text splitter.
+   * Truncate text chunks to fit within the available context window.
+   */
+  truncate(
+    prompt: BasePromptTemplate,
+    textChunks: string[],
+    padding: number = DEFAULT_PADDING,
+  ): string[] {
+    const textSplitter = this.getTextSplitterGivenPrompt(
+      prompt,
+      textChunks.length,
+      padding,
+    );
+    return textChunks.map((chunk) => truncateText(chunk, textSplitter));
+  }
+
+  /**
+   * Repack text chunks to better utilize the available context window.
    */
   repack(
-    prompt: PromptTemplate,
+    prompt: BasePromptTemplate,
     textChunks: string[],
-    padding = DEFAULT_PADDING,
-  ) {
+    padding: number = DEFAULT_PADDING,
+  ): string[] {
     const textSplitter = this.getTextSplitterGivenPrompt(prompt, 1, padding);
-    const combinedStr = textChunks.join("\n\n");
+    const combinedStr = textChunks
+      .map((c) => c.trim())
+      .filter((c) => c.length > 0)
+      .join("\n\n");
     return textSplitter.splitText(combinedStr);
   }
 
@@ -154,7 +177,8 @@ export class PromptHelper {
     } = options ?? {};
     return new PromptHelper({
       contextWindow: metadata.contextWindow,
-      numOutput: metadata.maxTokens ?? DEFAULT_NUM_OUTPUTS,
+      // fixme: numOutput is not in LLMMetadata
+      numOutput: DEFAULT_NUM_OUTPUTS,
       chunkOverlapRatio,
       chunkSizeLimit,
       tokenizer,

packages/core/src/node-parser/index.ts

@@ -13,6 +13,7 @@ export { MetadataAwareTextSplitter, NodeParser, TextSplitter } from "./base";
 export { MarkdownNodeParser } from "./markdown";
 export { SentenceSplitter } from "./sentence-splitter";
 export { SentenceWindowNodeParser } from "./sentence-window";
+export { TokenTextSplitter } from "./token-text-splitter";
 export type { SplitterParams } from "./type";
 export {
   splitByChar,
@@ -20,5 +21,6 @@ export {
   splitByRegex,
   splitBySentenceTokenizer,
   splitBySep,
+  truncateText,
 } from "./utils";
 export type { TextSplitterFn } from "./utils";

packages/core/src/node-parser/token-text-splitter.ts

@@ -0,0 +1,206 @@
+import type { Tokenizer } from "@llamaindex/env";
+import { z } from "zod";
+import { DEFAULT_CHUNK_OVERLAP, DEFAULT_CHUNK_SIZE, Settings } from "../global";
+import { MetadataAwareTextSplitter } from "./base";
+import type { SplitterParams } from "./type";
+import { splitByChar, splitBySep } from "./utils";
+
+const DEFAULT_METADATA_FORMAT_LEN = 2;
+
+const tokenTextSplitterSchema = z.object({
+  chunkSize: z.number().positive().default(DEFAULT_CHUNK_SIZE),
+  chunkOverlap: z.number().nonnegative().default(DEFAULT_CHUNK_OVERLAP),
+  separator: z.string().default(" "),
+  backupSeparators: z.array(z.string()).default(["\n"]),
+});
+
+export class TokenTextSplitter extends MetadataAwareTextSplitter {
+  chunkSize: number = DEFAULT_CHUNK_SIZE;
+  chunkOverlap: number = DEFAULT_CHUNK_OVERLAP;
+  separator: string = " ";
+  backupSeparators: string[] = ["\n"];
+  #tokenizer: Tokenizer;
+  #splitFns: Array<(text: string) => string[]> = [];
+
+  constructor(
+    params?: SplitterParams & Partial<z.infer<typeof tokenTextSplitterSchema>>,
+  ) {
+    super();
+
+    if (params) {
+      const parsedParams = tokenTextSplitterSchema.parse(params);
+      this.chunkSize = parsedParams.chunkSize;
+      this.chunkOverlap = parsedParams.chunkOverlap;
+      this.separator = parsedParams.separator;
+      this.backupSeparators = parsedParams.backupSeparators;
+    }
+
+    if (this.chunkOverlap > this.chunkSize) {
+      throw new Error(
+        `Got a larger chunk overlap (${this.chunkOverlap}) than chunk size (${this.chunkSize}), should be smaller.`,
+      );
+    }
+
+    this.#tokenizer = params?.tokenizer ?? Settings.tokenizer;
+
+    const allSeparators = [this.separator, ...this.backupSeparators];
+    this.#splitFns = allSeparators.map((sep) => splitBySep(sep));
+    this.#splitFns.push(splitByChar());
+  }
+
+  /**
+   * Split text into chunks, reserving space required for metadata string.
+   * @param text The text to split.
+   * @param metadata The metadata string.
+   * @returns An array of text chunks.
+   */
+  splitTextMetadataAware(text: string, metadata: string): string[] {
+    const metadataLength =
+      this.tokenSize(metadata) + DEFAULT_METADATA_FORMAT_LEN;
+    const effectiveChunkSize = this.chunkSize - metadataLength;
+
+    if (effectiveChunkSize <= 0) {
+      throw new Error(
+        `Metadata length (${metadataLength}) is longer than chunk size (${this.chunkSize}). ` +
+          `Consider increasing the chunk size or decreasing the size of your metadata to avoid this.`,
+      );
+    } else if (effectiveChunkSize < 50) {
+      console.warn(
+        `Metadata length (${metadataLength}) is close to chunk size (${this.chunkSize}). ` +
+          `Resulting chunks are less than 50 tokens. Consider increasing the chunk size or decreasing the size of your metadata to avoid this.`,
+      );
+    }
+
+    return this._splitText(text, effectiveChunkSize);
+  }
+
+  /**
+   * Split text into chunks.
+   * @param text The text to split.
+   * @returns An array of text chunks.
+   */
+  splitText(text: string): string[] {
+    return this._splitText(text, this.chunkSize);
+  }
+
+  /**
+   * Internal method to split text into chunks up to a specified size.
+   * @param text The text to split.
+   * @param chunkSize The maximum size of each chunk.
+   * @returns An array of text chunks.
+   */
+  private _splitText(text: string, chunkSize: number): string[] {
+    if (text === "") return [text];
+
+    // Dispatch chunking start event
+    Settings.callbackManager.dispatchEvent("chunking-start", { text: [text] });
+
+    const splits = this._split(text, chunkSize);
+    const chunks = this._merge(splits, chunkSize);
+
+    Settings.callbackManager.dispatchEvent("chunking-end", { chunks });
+
+    return chunks;
+  }
+
+  /**
+   * Break text into splits that are smaller than the chunk size.
+   * @param text The text to split.
+   * @param chunkSize The maximum size of each split.
+   * @returns An array of text splits.
+   */
+  private _split(text: string, chunkSize: number): string[] {
+    if (this.tokenSize(text) <= chunkSize) {
+      return [text];
+    }
+
+    for (const splitFn of this.#splitFns) {
+      const splits = splitFn(text);
+      if (splits.length > 1) {
+        const newSplits: string[] = [];
+        for (const split of splits) {
+          const splitLen = this.tokenSize(split);
+          if (splitLen <= chunkSize) {
+            newSplits.push(split);
+          } else {
+            newSplits.push(...this._split(split, chunkSize));
+          }
+        }
+        return newSplits;
+      }
+    }
+
+    return [text];
+  }
+
+  /**
+   * Merge splits into chunks with overlap.
+   * @param splits The array of text splits.
+   * @param chunkSize The maximum size of each chunk.
+   * @returns An array of merged text chunks.
+   */
+  private _merge(splits: string[], chunkSize: number): string[] {
+    const chunks: string[] = [];
+    let currentChunk: string[] = [];
+    let currentLength = 0;
+
+    for (const split of splits) {
+      const splitLength = this.tokenSize(split);
+
+      if (splitLength > chunkSize) {
+        console.warn(
+          `Got a split of size ${splitLength}, larger than chunk size ${chunkSize}.`,
+        );
+      }
+
+      if (currentLength + splitLength > chunkSize) {
+        const chunk = currentChunk.join("").trim();
+        if (chunk) {
+          chunks.push(chunk);
+        }
+
+        currentChunk = [];
+        currentLength = 0;
+
+        const overlapTokens = this.chunkOverlap;
+        const overlapSplits: string[] = [];
+
+        let overlapLength = 0;
+        while (
+          overlapSplits.length < splits.length &&
+          overlapLength < overlapTokens
+        ) {
+          const overlapSplit = currentChunk.shift();
+          if (!overlapSplit) break;
+          overlapSplits.push(overlapSplit);
+          overlapLength += this.tokenSize(overlapSplit);
+        }
+
+        for (const overlapSplit of overlapSplits.reverse()) {
+          currentChunk.push(overlapSplit);
+          currentLength += this.tokenSize(overlapSplit);
+          if (currentLength >= overlapTokens) break;
+        }
+      }
+
+      currentChunk.push(split);
+      currentLength += splitLength;
+    }
+
+    const finalChunk = currentChunk.join("").trim();
+    if (finalChunk) {
+      chunks.push(finalChunk);
+    }
+
+    return chunks;
+  }
+
+  /**
+   * Calculate the number of tokens in the text using the tokenizer.
+   * @param text The text to tokenize.
+   * @returns The number of tokens.
+   */
+  private tokenSize(text: string): number {
+    return this.#tokenizer.encode(text).length;
+  }
+}

packages/core/src/node-parser/utils.ts

@@ -3,7 +3,10 @@ import SentenceTokenizer from "./sentence_tokenizer";
 
 export type TextSplitterFn = (text: string) => string[];
 
-const truncateText = (text: string, textSplitter: TextSplitter): string => {
+export const truncateText = (
+  text: string,
+  textSplitter: TextSplitter,
+): string => {
   const chunks = textSplitter.splitText(text);
   return chunks[0] ?? text;
 };

packages/core/src/prompts/index.ts

@@ -12,11 +12,15 @@ export {
   defaultCondenseQuestionPrompt,
   defaultContextSystemPrompt,
   defaultKeywordExtractPrompt,
+  defaultNodeTextTemplate,
   defaultQueryKeywordExtractPrompt,
+  defaultQuestionExtractPrompt,
   defaultRefinePrompt,
   defaultSubQuestionPrompt,
   defaultSummaryPrompt,
   defaultTextQAPrompt,
+  defaultTitleCombinePromptTemplate,
+  defaultTitleExtractorPromptTemplate,
   defaultTreeSummarizePrompt,
 } from "./prompt";
 export type {
@@ -25,9 +29,12 @@ export type {
   ContextSystemPrompt,
   KeywordExtractPrompt,
   QueryKeywordExtractPrompt,
+  QuestionExtractPrompt,
   RefinePrompt,
   SubQuestionPrompt,
   SummaryPrompt,
   TextQAPrompt,
+  TitleCombinePrompt,
+  TitleExtractorPrompt,
   TreeSummarizePrompt,
 } from "./prompt";

packages/core/src/prompts/prompt.ts

@@ -13,8 +13,12 @@ export type CondenseQuestionPrompt = PromptTemplate<
   ["chatHistory", "question"]
 >;
 export type ContextSystemPrompt = PromptTemplate<["context"]>;
-export type KeywordExtractPrompt = PromptTemplate<["context"]>;
+export type KeywordExtractPrompt = PromptTemplate<["context", "maxKeywords"]>;
 export type QueryKeywordExtractPrompt = PromptTemplate<["question"]>;
+export type QuestionExtractPrompt = PromptTemplate<["context", "numQuestions"]>;
+export type TitleExtractorPrompt = PromptTemplate<["context"]>;
+export type TitleCombinePrompt = PromptTemplate<["context"]>;
+export type KeywordExtractorPrompt = PromptTemplate<["context", "numKeywords"]>;
 
 export const defaultTextQAPrompt: TextQAPrompt = new PromptTemplate({
   templateVars: ["context", "query"],
@@ -64,11 +68,13 @@ export const defaultRefinePrompt: RefinePrompt = new PromptTemplate({
   templateVars: ["query", "existingAnswer", "context"],
   template: `The original query is as follows: {query}
 We have provided an existing answer: {existingAnswer}
-We have the opportunity to refine the existing answer (only if needed) with some more context below.
+We have the opportunity to refine the existing answer
+(only if needed) with some more context below.
 ------------
 {context}
 ------------
-Given the new context, refine the original answer to better answer the query. If the context isn't useful, return the original answer.
+Given the new context, refine the original answer to better answer the query.
+If the context isn't useful, return the original answer.
 Refined Answer:`,
 });
 
@@ -251,3 +257,55 @@ export const defaultQueryKeywordExtractPrompt = new PromptTemplate({
 }).partialFormat({
   maxKeywords: "10",
 });
+
+export const defaultQuestionExtractPrompt = new PromptTemplate({
+  templateVars: ["numQuestions", "context"],
+  template: `(
+  "Given the contextual informations below, generate {numQuestions} questions this context can provides specific answers to which are unlikely to be found else where. Higher-level summaries of surrounding context may be provided as well. "
+  "Try using these summaries to generate better questions that this context can answer."
+  "---------------------"
+  "{context}"
+  "---------------------"
+  "Provide questions in the following format: 'QUESTIONS: <questions>'"
+)`,
+}).partialFormat({
+  numQuestions: "5",
+});
+
+export const defaultTitleExtractorPromptTemplate = new PromptTemplate({
+  templateVars: ["context"],
+  template: `{context}
+Give a title that summarizes all of the unique entities, titles or themes found in the context. 
+Title: `,
+});
+
+export const defaultTitleCombinePromptTemplate = new PromptTemplate({
+  templateVars: ["context"],
+  template: `{context} 
+Based on the above candidate titles and contents, what is the comprehensive title for this document? 
+Title: `,
+});
+
+export const defaultKeywordExtractorPromptTemplate = new PromptTemplate({
+  templateVars: ["context", "numKeywords"],
+  template: `{context}
+Give {numKeywords} unique keywords for this document. 
+Format as comma separated. 
+Keywords: `,
+}).partialFormat({
+  keywordCount: "5",
+});
+
+export const defaultNodeTextTemplate = new PromptTemplate({
+  templateVars: ["metadataStr", "content"],
+  template: `[Excerpt from document]
+{metadataStr}
+Excerpt:
+-----
+{content}
+-----
+`,
+}).partialFormat({
+  metadataStr: "",
+  content: "",
+});

packages/core/src/response-synthesizers/factory.ts

@@ -403,27 +403,27 @@ class MultiModal extends BaseSynthesizer {
   }
 }
 
-export function getResponseSynthesizer(
-  mode: ResponseMode,
+const modeToSynthesizer = {
+  compact: CompactAndRefine,
+  refine: Refine,
+  tree_summarize: TreeSummarize,
+  multi_modal: MultiModal,
+} as const;
+
+export function getResponseSynthesizer<Mode extends ResponseMode>(
+  mode: Mode,
   options: BaseSynthesizerOptions & {
     textQATemplate?: TextQAPrompt;
     refineTemplate?: RefinePrompt;
     summaryTemplate?: TreeSummarizePrompt;
     metadataMode?: MetadataMode;
   } = {},
-) {
-  switch (mode) {
-    case "compact": {
-      return new CompactAndRefine(options);
-    }
-    case "refine": {
-      return new Refine(options);
-    }
-    case "tree_summarize": {
-      return new TreeSummarize(options);
-    }
-    case "multi_modal": {
-      return new MultiModal(options);
-    }
+): InstanceType<(typeof modeToSynthesizer)[Mode]> {
+  const Synthesizer: (typeof modeToSynthesizer)[Mode] = modeToSynthesizer[mode];
+  if (!Synthesizer) {
+    throw new Error(`Invalid response mode: ${mode}`);
   }
+  return new Synthesizer(options) as InstanceType<
+    (typeof modeToSynthesizer)[Mode]
+  >;
 }

packages/core/src/tools/function-tool.ts

@@ -0,0 +1,62 @@
+import type { JSONSchemaType } from "ajv";
+import { z } from "zod";
+import { zodToJsonSchema } from "zod-to-json-schema";
+import type { JSONValue } from "../global";
+import type { BaseTool, ToolMetadata } from "../llms";
+
+const kOriginalFn = Symbol("originalFn");
+
+export class FunctionTool<T, R extends JSONValue | Promise<JSONValue>>
+  implements BaseTool<T>
+{
+  [kOriginalFn]?: (input: T) => R;
+
+  #fn: (input: T) => R;
+  #metadata: ToolMetadata<JSONSchemaType<T>>;
+  // todo: for the future, we can use zod to validate the input parameters
+  #zodType: z.ZodType<T> | null = null;
+  constructor(
+    fn: (input: T) => R,
+    metadata: ToolMetadata<JSONSchemaType<T>>,
+    zodType?: z.ZodType<T>,
+  ) {
+    this.#fn = fn;
+    this.#metadata = metadata;
+    if (zodType) {
+      this.#zodType = zodType;
+    }
+  }
+
+  static from<T>(
+    fn: (input: T) => JSONValue | Promise<JSONValue>,
+    schema: ToolMetadata<JSONSchemaType<T>>,
+  ): FunctionTool<T, JSONValue | Promise<JSONValue>>;
+  static from<T, R extends z.ZodType<T>>(
+    fn: (input: T) => JSONValue | Promise<JSONValue>,
+    schema: Omit<ToolMetadata, "parameters"> & {
+      parameters: R;
+    },
+  ): FunctionTool<T, JSONValue>;
+  static from(fn: any, schema: any): any {
+    if (schema.parameter instanceof z.ZodSchema) {
+      const jsonSchema = zodToJsonSchema(schema.parameter);
+      return new FunctionTool(
+        fn,
+        {
+          ...schema,
+          parameters: jsonSchema,
+        },
+        schema.parameter,
+      );
+    }
+    return new FunctionTool(fn, schema);
+  }
+
+  get metadata(): BaseTool<T>["metadata"] {
+    return this.#metadata as BaseTool<T>["metadata"];
+  }
+
+  call(input: T) {
+    return this.#fn.call(null, input);
+  }
+}

packages/core/src/tools/index.ts

@@ -0,0 +1 @@
+export { FunctionTool } from "./function-tool";

packages/core/tests/node-parser/text-splitter.test.ts

@@ -2,6 +2,7 @@ import {
   SentenceSplitter,
   splitBySentenceTokenizer,
 } from "@llamaindex/core/node-parser";
+import { Document } from "@llamaindex/core/schema";
 import { describe, expect, test } from "vitest";
 
 describe("sentence splitter", () => {
@@ -115,4 +116,26 @@ describe("sentence splitter", () => {
     const split = splitBySentenceTokenizer();
     expect(split(text)).toEqual([text]);
   });
+
+  test("split nodes with UUID IDs and correct relationships", () => {
+    const UUID_REGEX =
+      /^[0-9a-f]{8}-[0-9a-f]{4}-4[0-9a-f]{3}-[89ab][0-9a-f]{3}-[0-9a-f]{12}$/i;
+    const sentenceSplitter = new SentenceSplitter();
+    const docId = "test-doc-id";
+    const doc = new Document({
+      id_: docId,
+      text: "This is a test sentence. This is another test sentence.",
+    });
+    const nodes = sentenceSplitter.getNodesFromDocuments([doc]);
+    nodes.forEach((node) => {
+      // test node id should match uuid regex
+      expect(node.id_).toMatch(UUID_REGEX);
+
+      // test source reference to the doc ID
+      const source = node.relationships?.SOURCE;
+      expect(source).toBeDefined();
+      expect(source).toHaveProperty("nodeId");
+      expect((source as { nodeId: string }).nodeId).toEqual(docId);
+    });
+  });
 });

packages/core/tests/tools.test.ts

@@ -0,0 +1,35 @@
+import { FunctionTool } from "@llamaindex/core/tools";
+import { describe, test } from "vitest";
+import { z } from "zod";
+
+describe("FunctionTool", () => {
+  test("type system", () => {
+    FunctionTool.from((input: string) => input, {
+      name: "test",
+      description: "test",
+    });
+    FunctionTool.from(({ input }: { input: string }) => input, {
+      name: "test",
+      description: "test",
+      parameters: {
+        type: "object",
+        properties: {
+          input: {
+            type: "string",
+          },
+        },
+        required: ["input"],
+      },
+    });
+    const inputSchema = z
+      .object({
+        input: z.string(),
+      })
+      .required();
+    FunctionTool.from(({ input }: { input: string }) => input, {
+      name: "test",
+      description: "test",
+      parameters: inputSchema,
+    });
+  });
+});

packages/core/tests/workflow.test.ts

@@ -166,4 +166,68 @@ describe("Workflow", () => {
       greeting: "Hello Alice, you are 30 years old!",
     });
   });
+
+  test("workflow with two concurrent steps", async () => {
+    const concurrentFlow = new Workflow({ verbose: true });
+
+    const step1 = vi.fn(async (_context, _ev: StartEvent) => {
+      await new Promise((resolve) => setTimeout(resolve, 200));
+      return new StopEvent({ result: "Step 1 completed" });
+    });
+
+    const step2 = vi.fn(async (_context, _ev: StartEvent) => {
+      await new Promise((resolve) => setTimeout(resolve, 100));
+      return new StopEvent({ result: "Step 2 completed" });
+    });
+
+    concurrentFlow.addStep(StartEvent, step1);
+    concurrentFlow.addStep(StartEvent, step2);
+
+    const startTime = new Date();
+    const result = await concurrentFlow.run("start");
+    const endTime = new Date();
+    const duration = endTime.getTime() - startTime.getTime();
+
+    expect(step1).toHaveBeenCalledTimes(1);
+    expect(step2).toHaveBeenCalledTimes(1);
+    expect(duration).toBeLessThan(200);
+    expect(result.data.result).toBe("Step 2 completed");
+  });
+
+  test("workflow with two concurrent cyclic steps", async () => {
+    const concurrentCyclicFlow = new Workflow({ verbose: true });
+
+    class Step1Event extends WorkflowEvent {}
+    class Step2Event extends WorkflowEvent {}
+
+    let step2Count = 0;
+
+    const step1 = vi.fn(async (_context, ev: StartEvent | Step1Event) => {
+      await new Promise((resolve) => setTimeout(resolve, 1000));
+      return new Step1Event({ result: "Step 1 completed" });
+    });
+
+    const step2 = vi.fn(async (_context, ev: StartEvent | Step2Event) => {
+      await new Promise((resolve) => setTimeout(resolve, 100));
+      step2Count++;
+      if (step2Count >= 5) {
+        return new StopEvent({ result: "Step 2 completed 5 times" });
+      }
+      return new Step2Event({ result: "Step 2 completed" });
+    });
+
+    concurrentCyclicFlow.addStep([StartEvent, Step1Event], step1);
+    concurrentCyclicFlow.addStep([StartEvent, Step2Event], step2);
+
+    const startTime = new Date();
+    const result = await concurrentCyclicFlow.run("start");
+    const endTime = new Date();
+    const duration = endTime.getTime() - startTime.getTime();
+
+    expect(step1).toHaveBeenCalledTimes(1);
+    expect(step2).toHaveBeenCalledTimes(5);
+    expect(duration).toBeGreaterThan(500); // At least 5 * 100ms for step2
+    expect(duration).toBeLessThan(1000); // Less than 1 second
+    expect(result.data.result).toBe("Step 2 completed 5 times");
+  });
 });

packages/core/tools/package.json

@@ -0,0 +1,8 @@
+{
+  "type": "module",
+  "main": "./dist/index.cjs",
+  "module": "./dist/index.js",
+  "types": "./dist/index.d.ts",
+  "exports": "./dist/index.js",
+  "private": true
+}

packages/core/vector-store/package.json

@@ -0,0 +1,8 @@
+{
+  "type": "module",
+  "main": "./dist/index.cjs",
+  "module": "./dist/index.js",
+  "types": "./dist/index.d.ts",
+  "exports": "./dist/index.js",
+  "private": true
+}

packages/env/CHANGELOG.md

@@ -1,5 +1,12 @@
 # @llamaindex/env
 
+## 0.1.14
+
+### Patch Changes
+
+- ae49ff4: feat: use `gpt-tokenizer`
+- a75af83: refactor: move some llm and embedding to single package
+
 ## 0.1.13
 
 ### Patch Changes

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": "0.1.13",
+  "version": "0.1.14",
   "type": "module",
   "types": "dist/type/index.d.ts",
   "main": "dist/cjs/index.js",
@@ -75,8 +75,8 @@
     "@swc/core": "^1.7.22",
     "@xenova/transformers": "^2.17.2",
     "concurrently": "^8.2.2",
+    "gpt-tokenizer": "^2.5.0",
     "pathe": "^1.1.2",
-    "tiktoken": "^1.0.16",
     "vitest": "^2.0.5"
   },
   "dependencies": {
@@ -85,9 +85,9 @@
   "peerDependencies": {
     "@aws-crypto/sha256-js": "^5.2.0",
     "@xenova/transformers": "^2.17.2",
+    "gpt-tokenizer": "^2.5.0",
     "js-tiktoken": "^1.0.12",
-    "pathe": "^1.1.2",
-    "tiktoken": "^1.0.15"
+    "pathe": "^1.1.2"
   },
   "peerDependenciesMeta": {
     "@aws-crypto/sha256-js": {

packages/env/src/index.browser.ts

@@ -14,7 +14,7 @@ export {
   type OnLoad,
 } from "./multi-model/index.browser.js";
 export { Tokenizers, tokenizers, type Tokenizer } from "./tokenizers/js.js";
-
+export { NotSupportCurrentRuntimeClass } from "./utils/shared.js";
 // @ts-expect-error
 if (typeof window === "undefined") {
   console.warn(

packages/env/src/index.edge-light.ts

@@ -14,3 +14,4 @@ export {
   type OnLoad,
 } from "./multi-model/index.non-nodejs.js";
 export { Tokenizers, tokenizers, type Tokenizer } from "./tokenizers/js.js";
+export { NotSupportCurrentRuntimeClass } from "./utils/shared.js";

packages/env/src/index.ts

@@ -47,6 +47,7 @@ export {
   getEnv,
   setEnvs,
 } from "./utils/index.js";
+export { NotSupportCurrentRuntimeClass } from "./utils/shared.js";
 export {
   createWriteStream,
   EOL,

packages/env/src/index.workerd.ts

@@ -7,6 +7,8 @@
  */
 import { INTERNAL_ENV } from "./utils/index.js";
 
+export { NotSupportCurrentRuntimeClass } from "./utils/shared.js";
+
 export * from "./node-polyfill.js";
 
 export function getEnv(name: string): string | undefined {

packages/env/src/tokenizers/node.ts

@@ -2,21 +2,18 @@
 import type { Tokenizer } from "./types.js";
 import { Tokenizers } from "./types.js";
 
-import { get_encoding } from "tiktoken";
+import cl100kBase from "gpt-tokenizer";
 
 class TokenizerSingleton {
-  private defaultTokenizer: Tokenizer;
+  #defaultTokenizer: Tokenizer;
 
   constructor() {
-    const encoding = get_encoding("cl100k_base");
-
-    this.defaultTokenizer = {
-      encode: (text: string) => {
-        return encoding.encode(text);
+    this.#defaultTokenizer = {
+      encode: (text: string): Uint32Array => {
+        return new Uint32Array(cl100kBase.encode(text));
       },
       decode: (tokens: Uint32Array) => {
-        const text = encoding.decode(tokens);
-        return new TextDecoder().decode(text);
+        return cl100kBase.decode(tokens);
       },
     };
   }
@@ -26,7 +23,7 @@ class TokenizerSingleton {
       throw new Error(`Tokenizer encoding ${encoding} not yet supported`);
     }
 
-    return this.defaultTokenizer;
+    return this.#defaultTokenizer;
   }
 }
 

packages/env/src/utils/shared.ts

@@ -0,0 +1,13 @@
+export class NotSupportCurrentRuntimeClass {
+  constructor(runtime: string) {
+    throw new Error(`Current environment ${runtime} is not supported`);
+  }
+
+  static bind(runtime: string) {
+    return class extends NotSupportCurrentRuntimeClass {
+      constructor(...args: any[]) {
+        super(runtime);
+      }
+    } as any;
+  }
+}

packages/env/tests/tokenizer.test.ts

@@ -0,0 +1,11 @@
+import { describe, expect, it } from "vitest";
+import { tokenizers } from "../src/tokenizers/node.js";
+
+describe("tokenizer", () => {
+  it("should tokenize text", () => {
+    const tokenizer = tokenizers.tokenizer();
+    expect(tokenizer.decode(tokenizer.encode("hello world"))).toBe(
+      "hello world",
+    );
+  });
+});

packages/experimental/CHANGELOG.md

@@ -1,5 +1,108 @@
 # @llamaindex/experimental
 
+## 0.0.102
+
+### Patch Changes
+
+- a75af83: refactor: move some llm and embedding to single package
+- Updated dependencies [ae49ff4]
+- Updated dependencies [4c38c1b]
+- Updated dependencies [a75af83]
+- Updated dependencies [a75af83]
+  - llamaindex@0.7.1
+
+## 0.0.101
+
+### Patch Changes
+
+- Updated dependencies [1364e8e]
+- Updated dependencies [3b7736f]
+- Updated dependencies [96fc69c]
+  - llamaindex@0.7.0
+
+## 0.0.100
+
+### Patch Changes
+
+- Updated dependencies [5729bd9]
+  - llamaindex@0.6.22
+
+## 0.0.99
+
+### Patch Changes
+
+- Updated dependencies [6f75306]
+- Updated dependencies [94cb4ad]
+  - llamaindex@0.6.21
+
+## 0.0.98
+
+### Patch Changes
+
+- Updated dependencies [6a9a7b1]
+  - llamaindex@0.6.20
+
+## 0.0.97
+
+### Patch Changes
+
+- Updated dependencies [62cba52]
+- Updated dependencies [d265e96]
+- Updated dependencies [d30bbf7]
+- Updated dependencies [53fd00a]
+  - llamaindex@0.6.19
+
+## 0.0.96
+
+### Patch Changes
+
+- Updated dependencies [5f67820]
+- Updated dependencies [fe08d04]
+  - llamaindex@0.6.18
+
+## 0.0.95
+
+### Patch Changes
+
+- Updated dependencies [ee697fb]
+  - llamaindex@0.6.17
+
+## 0.0.94
+
+### Patch Changes
+
+- Updated dependencies [63e9846]
+- Updated dependencies [6f3a31c]
+  - llamaindex@0.6.16
+
+## 0.0.93
+
+### Patch Changes
+
+- Updated dependencies [2a82413]
+  - llamaindex@0.6.15
+
+## 0.0.92
+
+### Patch Changes
+
+- llamaindex@0.6.14
+
+## 0.0.91
+
+### Patch Changes
+
+- llamaindex@0.6.13
+
+## 0.0.90
+
+### Patch Changes
+
+- Updated dependencies [f7b4e94]
+- Updated dependencies [78037a6]
+- Updated dependencies [1d9e3b1]
+  - llamaindex@0.6.12
+
 ## 0.0.89
 
 ### Patch Changes

packages/experimental/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@llamaindex/experimental",
   "description": "Experimental package for LlamaIndexTS",
-  "version": "0.0.89",
+  "version": "0.0.102",
   "type": "module",
   "types": "dist/type/index.d.ts",
   "main": "dist/cjs/index.js",

packages/llamaindex/CHANGELOG.md

@@ -1,5 +1,154 @@
 # llamaindex
 
+## 0.7.1
+
+### Patch Changes
+
+- ae49ff4: feat: use `gpt-tokenizer`
+- 4c38c1b: fix(cloud): do not detect file type in llama parse
+- a75af83: feat: allow passing perform setup in pg vector store
+- a75af83: refactor: move some llm and embedding to single package
+- Updated dependencies [ae49ff4]
+- Updated dependencies [4c38c1b]
+- Updated dependencies [24d065f]
+- Updated dependencies [a75af83]
+  - @llamaindex/env@0.1.14
+  - @llamaindex/cloud@1.0.1
+  - @llamaindex/huggingface@0.0.2
+  - @llamaindex/portkey-ai@0.0.2
+  - @llamaindex/anthropic@0.0.2
+  - @llamaindex/deepinfra@0.0.2
+  - @llamaindex/replicate@0.0.2
+  - @llamaindex/ollama@0.0.9
+  - @llamaindex/openai@0.1.18
+  - @llamaindex/clip@0.0.2
+  - @llamaindex/groq@0.0.17
+  - @llamaindex/core@0.3.1
+
+## 0.7.0
+
+### Minor Changes
+
+- 1364e8e: update metadata extractors to use PromptTemplate
+- 96fc69c: Correct initialization of QuestionsAnsweredExtractor so that it uses the promptTemplate arg when passed in
+
+### Patch Changes
+
+- 3b7736f: feat: added gemini 002 support
+- Updated dependencies [1364e8e]
+- Updated dependencies [96fc69c]
+  - @llamaindex/core@0.3.0
+  - @llamaindex/cloud@1.0.0
+  - @llamaindex/ollama@0.0.8
+  - @llamaindex/openai@0.1.17
+  - @llamaindex/groq@0.0.16
+
+## 0.6.22
+
+### Patch Changes
+
+- 5729bd9: Fix LlamaCloud API calls for ensuring an index and for file uploads
+
+## 0.6.21
+
+### Patch Changes
+
+- 6f75306: feat: support metadata filters for AstraDB
+- 94cb4ad: feat: Add metadata filters to ChromaDb and update to 1.9.2
+
+## 0.6.20
+
+### Patch Changes
+
+- 6a9a7b1: fix: take init api key into account
+- Updated dependencies [6a9a7b1]
+  - @llamaindex/openai@0.1.16
+  - @llamaindex/groq@0.0.15
+
+## 0.6.19
+
+### Patch Changes
+
+- 62cba52: Add ensureIndex function to LlamaCloudIndex
+- d265e96: fix: ignore resolving unpdf for nextjs
+- d30bbf7: Convert undefined values to null in LlamaCloud filters
+- 53fd00a: Fix getPipelineId in LlamaCloudIndex
+
+## 0.6.18
+
+### Patch Changes
+
+- 5f67820: Fix that node parsers generate nodes with UUIDs
+- fe08d04: Fix LlamaCloud retrieval with multiple pipelines
+- Updated dependencies [5f67820]
+  - @llamaindex/core@0.2.12
+  - @llamaindex/cloud@0.2.14
+  - @llamaindex/ollama@0.0.7
+  - @llamaindex/openai@0.1.15
+  - @llamaindex/groq@0.0.14
+
+## 0.6.17
+
+### Patch Changes
+
+- ee697fb: fix: generate uuid when inserting to Qdrant
+- Updated dependencies [ee697fb]
+  - @llamaindex/core@0.2.11
+  - @llamaindex/cloud@0.2.13
+  - @llamaindex/ollama@0.0.6
+  - @llamaindex/openai@0.1.14
+  - @llamaindex/groq@0.0.13
+
+## 0.6.16
+
+### Patch Changes
+
+- 63e9846: fix: preFilters does not work with asQueryEngine
+- 6f3a31c: feat: add metadata filters for Qdrant vector store
+- Updated dependencies [3489e7d]
+- Updated dependencies [468bda5]
+  - @llamaindex/core@0.2.10
+  - @llamaindex/cloud@0.2.12
+  - @llamaindex/ollama@0.0.5
+  - @llamaindex/openai@0.1.13
+  - @llamaindex/groq@0.0.12
+
+## 0.6.15
+
+### Patch Changes
+
+- 2a82413: fix(core): set `Settings.llm` to OpenAI by default and support lazy load openai
+- Updated dependencies [2a82413]
+- Updated dependencies [0b20ff9]
+  - @llamaindex/groq@0.0.11
+  - @llamaindex/openai@0.1.12
+  - @llamaindex/cloud@0.2.11
+
+## 0.6.14
+
+### Patch Changes
+
+- Updated dependencies [b17d439]
+  - @llamaindex/core@0.2.9
+  - @llamaindex/ollama@0.0.4
+  - @llamaindex/openai@0.1.11
+  - @llamaindex/groq@0.0.10
+
+## 0.6.13
+
+### Patch Changes
+
+- Updated dependencies [981811e]
+  - @llamaindex/cloud@0.2.10
+
+## 0.6.12
+
+### Patch Changes
+
+- f7b4e94: feat: add filters for pinecone
+- 78037a6: fix: bypass service context embed model
+- 1d9e3b1: fix: export llama reader in non-nodejs runtime
+
 ## 0.6.11
 
 ### Patch Changes

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

@@ -1,5 +1,107 @@
 # @llamaindex/cloudflare-worker-agent-test
 
+## 0.0.86
+
+### Patch Changes
+
+- Updated dependencies [ae49ff4]
+- Updated dependencies [4c38c1b]
+- Updated dependencies [a75af83]
+- Updated dependencies [a75af83]
+  - llamaindex@0.7.1
+
+## 0.0.85
+
+### Patch Changes
+
+- Updated dependencies [1364e8e]
+- Updated dependencies [3b7736f]
+- Updated dependencies [96fc69c]
+  - llamaindex@0.7.0
+
+## 0.0.84
+
+### Patch Changes
+
+- Updated dependencies [5729bd9]
+  - llamaindex@0.6.22
+
+## 0.0.83
+
+### Patch Changes
+
+- Updated dependencies [6f75306]
+- Updated dependencies [94cb4ad]
+  - llamaindex@0.6.21
+
+## 0.0.82
+
+### Patch Changes
+
+- Updated dependencies [6a9a7b1]
+  - llamaindex@0.6.20
+
+## 0.0.81
+
+### Patch Changes
+
+- Updated dependencies [62cba52]
+- Updated dependencies [d265e96]
+- Updated dependencies [d30bbf7]
+- Updated dependencies [53fd00a]
+  - llamaindex@0.6.19
+
+## 0.0.80
+
+### Patch Changes
+
+- Updated dependencies [5f67820]
+- Updated dependencies [fe08d04]
+  - llamaindex@0.6.18
+
+## 0.0.79
+
+### Patch Changes
+
+- Updated dependencies [ee697fb]
+  - llamaindex@0.6.17
+
+## 0.0.78
+
+### Patch Changes
+
+- Updated dependencies [63e9846]
+- Updated dependencies [6f3a31c]
+  - llamaindex@0.6.16
+
+## 0.0.77
+
+### Patch Changes
+
+- Updated dependencies [2a82413]
+  - llamaindex@0.6.15
+
+## 0.0.76
+
+### Patch Changes
+
+- llamaindex@0.6.14
+
+## 0.0.75
+
+### Patch Changes
+
+- llamaindex@0.6.13
+
+## 0.0.74
+
+### Patch Changes
+
+- Updated dependencies [f7b4e94]
+- Updated dependencies [78037a6]
+- Updated dependencies [1d9e3b1]
+  - llamaindex@0.6.12
+
 ## 0.0.73
 
 ### Patch Changes

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

@@ -1,6 +1,6 @@
 {
   "name": "@llamaindex/cloudflare-worker-agent-test",
-  "version": "0.0.73",
+  "version": "0.0.86",
   "type": "module",
   "private": true,
   "scripts": {

packages/llamaindex/e2e/examples/llama-parse-browser/CHANGELOG.md

@@ -1,5 +1,52 @@
 # @llamaindex/llama-parse-browser-test
 
+## 0.0.12
+
+### Patch Changes
+
+- Updated dependencies [4c38c1b]
+- Updated dependencies [24d065f]
+- Updated dependencies [a75af83]
+  - @llamaindex/cloud@1.0.1
+
+## 0.0.11
+
+### Patch Changes
+
+- @llamaindex/cloud@1.0.0
+
+## 0.0.10
+
+### Patch Changes
+
+- @llamaindex/cloud@0.2.14
+
+## 0.0.9
+
+### Patch Changes
+
+- @llamaindex/cloud@0.2.13
+
+## 0.0.8
+
+### Patch Changes
+
+- @llamaindex/cloud@0.2.12
+
+## 0.0.7
+
+### Patch Changes
+
+- Updated dependencies [0b20ff9]
+  - @llamaindex/cloud@0.2.11
+
+## 0.0.6
+
+### Patch Changes
+
+- Updated dependencies [981811e]
+  - @llamaindex/cloud@0.2.10
+
 ## 0.0.5
 
 ### Patch Changes

packages/llamaindex/e2e/examples/llama-parse-browser/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@llamaindex/llama-parse-browser-test",
   "private": true,
-  "version": "0.0.5",
+  "version": "0.0.12",
   "type": "module",
   "scripts": {
     "dev": "vite",

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

@@ -1,5 +1,107 @@
 # @llamaindex/next-agent-test
 
+## 0.1.86
+
+### Patch Changes
+
+- Updated dependencies [ae49ff4]
+- Updated dependencies [4c38c1b]
+- Updated dependencies [a75af83]
+- Updated dependencies [a75af83]
+  - llamaindex@0.7.1
+
+## 0.1.85
+
+### Patch Changes
+
+- Updated dependencies [1364e8e]
+- Updated dependencies [3b7736f]
+- Updated dependencies [96fc69c]
+  - llamaindex@0.7.0
+
+## 0.1.84
+
+### Patch Changes
+
+- Updated dependencies [5729bd9]
+  - llamaindex@0.6.22
+
+## 0.1.83
+
+### Patch Changes
+
+- Updated dependencies [6f75306]
+- Updated dependencies [94cb4ad]
+  - llamaindex@0.6.21
+
+## 0.1.82
+
+### Patch Changes
+
+- Updated dependencies [6a9a7b1]
+  - llamaindex@0.6.20
+
+## 0.1.81
+
+### Patch Changes
+
+- Updated dependencies [62cba52]
+- Updated dependencies [d265e96]
+- Updated dependencies [d30bbf7]
+- Updated dependencies [53fd00a]
+  - llamaindex@0.6.19
+
+## 0.1.80
+
+### Patch Changes
+
+- Updated dependencies [5f67820]
+- Updated dependencies [fe08d04]
+  - llamaindex@0.6.18
+
+## 0.1.79
+
+### Patch Changes
+
+- Updated dependencies [ee697fb]
+  - llamaindex@0.6.17
+
+## 0.1.78
+
+### Patch Changes
+
+- Updated dependencies [63e9846]
+- Updated dependencies [6f3a31c]
+  - llamaindex@0.6.16
+
+## 0.1.77
+
+### Patch Changes
+
+- Updated dependencies [2a82413]
+  - llamaindex@0.6.15
+
+## 0.1.76
+
+### Patch Changes
+
+- llamaindex@0.6.14
+
+## 0.1.75
+
+### Patch Changes
+
+- llamaindex@0.6.13
+
+## 0.1.74
+
+### Patch Changes
+
+- Updated dependencies [f7b4e94]
+- Updated dependencies [78037a6]
+- Updated dependencies [1d9e3b1]
+  - llamaindex@0.6.12
+
 ## 0.1.73
 
 ### Patch Changes

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

@@ -1,6 +1,6 @@
 {
   "name": "@llamaindex/next-agent-test",
-  "version": "0.1.73",
+  "version": "0.1.86",
   "private": true,
   "scripts": {
     "dev": "next dev",

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

@@ -1,5 +1,107 @@
 # test-edge-runtime
 
+## 0.1.85
+
+### Patch Changes
+
+- Updated dependencies [ae49ff4]
+- Updated dependencies [4c38c1b]
+- Updated dependencies [a75af83]
+- Updated dependencies [a75af83]
+  - llamaindex@0.7.1
+
+## 0.1.84
+
+### Patch Changes
+
+- Updated dependencies [1364e8e]
+- Updated dependencies [3b7736f]
+- Updated dependencies [96fc69c]
+  - llamaindex@0.7.0
+
+## 0.1.83
+
+### Patch Changes
+
+- Updated dependencies [5729bd9]
+  - llamaindex@0.6.22
+
+## 0.1.82
+
+### Patch Changes
+
+- Updated dependencies [6f75306]
+- Updated dependencies [94cb4ad]
+  - llamaindex@0.6.21
+
+## 0.1.81
+
+### Patch Changes
+
+- Updated dependencies [6a9a7b1]
+  - llamaindex@0.6.20
+
+## 0.1.80
+
+### Patch Changes
+
+- Updated dependencies [62cba52]
+- Updated dependencies [d265e96]
+- Updated dependencies [d30bbf7]
+- Updated dependencies [53fd00a]
+  - llamaindex@0.6.19
+
+## 0.1.79
+
+### Patch Changes
+
+- Updated dependencies [5f67820]
+- Updated dependencies [fe08d04]
+  - llamaindex@0.6.18
+
+## 0.1.78
+
+### Patch Changes
+
+- Updated dependencies [ee697fb]
+  - llamaindex@0.6.17
+
+## 0.1.77
+
+### Patch Changes
+
+- Updated dependencies [63e9846]
+- Updated dependencies [6f3a31c]
+  - llamaindex@0.6.16
+
+## 0.1.76
+
+### Patch Changes
+
+- Updated dependencies [2a82413]
+  - llamaindex@0.6.15
+
+## 0.1.75
+
+### Patch Changes
+
+- llamaindex@0.6.14
+
+## 0.1.74
+
+### Patch Changes
+
+- llamaindex@0.6.13
+
+## 0.1.73
+
+### Patch Changes
+
+- Updated dependencies [f7b4e94]
+- Updated dependencies [78037a6]
+- Updated dependencies [1d9e3b1]
+  - llamaindex@0.6.12
+
 ## 0.1.72
 
 ### Patch Changes

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

@@ -1,6 +1,6 @@
 {
   "name": "@llamaindex/nextjs-edge-runtime-test",
-  "version": "0.1.72",
+  "version": "0.1.85",
   "private": true,
   "scripts": {
     "dev": "next dev",

packages/llamaindex/e2e/examples/nextjs-edge-runtime/src/app/page.tsx

@@ -1,19 +1,12 @@
-import { tokenizerResultPromise } from "@/utils/llm";
-import { use } from "react";
+import "@/utils/llm";
 
 export const runtime = "edge";
 
 export default function Home() {
-  const result = use(tokenizerResultPromise);
   return (
     <main>
       <div>
         <h1>Next.js Edge Runtime</h1>
-        <div>
-          {result.map((value, index) => (
-            <span key={index}>{value}</span>
-          ))}
-        </div>
       </div>
     </main>
   );

packages/llamaindex/e2e/examples/nextjs-edge-runtime/src/utils/llm.ts

@@ -1,23 +1,8 @@
 // test runtime
 import "llamaindex";
-import { ClipEmbedding } from "llamaindex";
 import "llamaindex/readers/SimpleDirectoryReader";
 
 // @ts-expect-error
 if (typeof EdgeRuntime !== "string") {
   throw new Error("Expected run in EdgeRuntime");
 }
-
-export const tokenizerResultPromise = new Promise<number[]>(
-  (resolve, reject) => {
-    const embedding = new ClipEmbedding();
-    //#region make sure @xenova/transformers is working in edge runtime
-    embedding
-      .getTokenizer()
-      .then((tokenizer) => {
-        resolve(tokenizer.encode("hello world"));
-      })
-      .catch(reject);
-    //#endregion
-  },
-);

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

@@ -1,5 +1,107 @@
 # @llamaindex/next-node-runtime
 
+## 0.0.67
+
+### Patch Changes
+
+- Updated dependencies [ae49ff4]
+- Updated dependencies [4c38c1b]
+- Updated dependencies [a75af83]
+- Updated dependencies [a75af83]
+  - llamaindex@0.7.1
+
+## 0.0.66
+
+### Patch Changes
+
+- Updated dependencies [1364e8e]
+- Updated dependencies [3b7736f]
+- Updated dependencies [96fc69c]
+  - llamaindex@0.7.0
+
+## 0.0.65
+
+### Patch Changes
+
+- Updated dependencies [5729bd9]
+  - llamaindex@0.6.22
+
+## 0.0.64
+
+### Patch Changes
+
+- Updated dependencies [6f75306]
+- Updated dependencies [94cb4ad]
+  - llamaindex@0.6.21
+
+## 0.0.63
+
+### Patch Changes
+
+- Updated dependencies [6a9a7b1]
+  - llamaindex@0.6.20
+
+## 0.0.62
+
+### Patch Changes
+
+- Updated dependencies [62cba52]
+- Updated dependencies [d265e96]
+- Updated dependencies [d30bbf7]
+- Updated dependencies [53fd00a]
+  - llamaindex@0.6.19
+
+## 0.0.61
+
+### Patch Changes
+
+- Updated dependencies [5f67820]
+- Updated dependencies [fe08d04]
+  - llamaindex@0.6.18
+
+## 0.0.60
+
+### Patch Changes
+
+- Updated dependencies [ee697fb]
+  - llamaindex@0.6.17
+
+## 0.0.59
+
+### Patch Changes
+
+- Updated dependencies [63e9846]
+- Updated dependencies [6f3a31c]
+  - llamaindex@0.6.16
+
+## 0.0.58
+
+### Patch Changes
+
+- Updated dependencies [2a82413]
+  - llamaindex@0.6.15
+
+## 0.0.57
+
+### Patch Changes
+
+- llamaindex@0.6.14
+
+## 0.0.56
+
+### Patch Changes
+
+- llamaindex@0.6.13
+
+## 0.0.55
+
+### Patch Changes
+
+- Updated dependencies [f7b4e94]
+- Updated dependencies [78037a6]
+- Updated dependencies [1d9e3b1]
+  - llamaindex@0.6.12
+
 ## 0.0.54
 
 ### Patch Changes

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

@@ -1,6 +1,6 @@
 {
   "name": "@llamaindex/next-node-runtime-test",
-  "version": "0.0.54",
+  "version": "0.0.67",
   "private": true,
   "scripts": {
     "dev": "next dev",

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

@@ -1,5 +1,107 @@
 # @llamaindex/waku-query-engine-test
 
+## 0.0.86
+
+### Patch Changes
+
+- Updated dependencies [ae49ff4]
+- Updated dependencies [4c38c1b]
+- Updated dependencies [a75af83]
+- Updated dependencies [a75af83]
+  - llamaindex@0.7.1
+
+## 0.0.85
+
+### Patch Changes
+
+- Updated dependencies [1364e8e]
+- Updated dependencies [3b7736f]
+- Updated dependencies [96fc69c]
+  - llamaindex@0.7.0
+
+## 0.0.84
+
+### Patch Changes
+
+- Updated dependencies [5729bd9]
+  - llamaindex@0.6.22
+
+## 0.0.83
+
+### Patch Changes
+
+- Updated dependencies [6f75306]
+- Updated dependencies [94cb4ad]
+  - llamaindex@0.6.21
+
+## 0.0.82
+
+### Patch Changes
+
+- Updated dependencies [6a9a7b1]
+  - llamaindex@0.6.20
+
+## 0.0.81
+
+### Patch Changes
+
+- Updated dependencies [62cba52]
+- Updated dependencies [d265e96]
+- Updated dependencies [d30bbf7]
+- Updated dependencies [53fd00a]
+  - llamaindex@0.6.19
+
+## 0.0.80
+
+### Patch Changes
+
+- Updated dependencies [5f67820]
+- Updated dependencies [fe08d04]
+  - llamaindex@0.6.18
+
+## 0.0.79
+
+### Patch Changes
+
+- Updated dependencies [ee697fb]
+  - llamaindex@0.6.17
+
+## 0.0.78
+
+### Patch Changes
+
+- Updated dependencies [63e9846]
+- Updated dependencies [6f3a31c]
+  - llamaindex@0.6.16
+
+## 0.0.77
+
+### Patch Changes
+
+- Updated dependencies [2a82413]
+  - llamaindex@0.6.15
+
+## 0.0.76
+
+### Patch Changes
+
+- llamaindex@0.6.14
+
+## 0.0.75
+
+### Patch Changes
+
+- llamaindex@0.6.13
+
+## 0.0.74
+
+### Patch Changes
+
+- Updated dependencies [f7b4e94]
+- Updated dependencies [78037a6]
+- Updated dependencies [1d9e3b1]
+  - llamaindex@0.6.12
+
 ## 0.0.73
 
 ### Patch Changes

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

@@ -1,6 +1,6 @@
 {
   "name": "@llamaindex/waku-query-engine-test",
-  "version": "0.0.73",
+  "version": "0.0.86",
   "type": "module",
   "private": true,
   "scripts": {
@@ -10,17 +10,16 @@
   },
   "dependencies": {
     "llamaindex": "workspace:*",
-    "react": "19.0.0-rc-7771d3a7-20240827",
-    "react-dom": "19.0.0-rc-7771d3a7-20240827",
-    "react-server-dom-webpack": "19.0.0-rc-7771d3a7-20240827",
-    "waku": "0.21.1"
+    "react": "19.0.0-rc-bf7e210c-20241017",
+    "react-dom": "19.0.0-rc-bf7e210c-20241017",
+    "react-server-dom-webpack": "19.0.0-rc-bf7e210c-20241017",
+    "waku": "0.21.4"
   },
   "devDependencies": {
-    "@types/react": "18.3.5",
-    "@types/react-dom": "18.3.0",
-    "autoprefixer": "10.4.20",
-    "tailwindcss": "3.4.10",
-    "typescript": "5.6.2",
-    "vite-plugin-wasm": "^3.3.0"
+    "@types/react": "18.3.11",
+    "@types/react-dom": "18.3.1",
+    "autoprefixer": "^10.4.20",
+    "tailwindcss": "^3.4.14",
+    "typescript": "5.6.2"
   }
 }

packages/llamaindex/e2e/examples/waku-query-engine/vite.config.ts

@@ -1,8 +0,0 @@
-import wasm from "vite-plugin-wasm";
-
-export default {
-  plugins: [wasm()],
-  ssr: {
-    external: ["tiktoken"],
-  },
-};

packages/llamaindex/e2e/node/vector-store/pg-vector-store.e2e.ts

@@ -105,3 +105,22 @@ await test("simple node", async (t) => {
     assert.deepStrictEqual(result.nodes, []);
   }
 });
+
+await test("no setup", async (t) => {
+  // @ts-expect-error private method
+  assert.ok(PGVectorStore.prototype.checkSchema);
+  // @ts-expect-error private method
+  const Mock = class extends PGVectorStore {
+    private override async checkSchema(): Promise<any> {
+      throw new Error("should not be called");
+    }
+  };
+  const vectorStore = new Mock({
+    clientConfig: pgConfig,
+    performSetup: false,
+  });
+  const db = await vectorStore.client();
+  t.after(async () => {
+    await db.close();
+  });
+});

packages/llamaindex/package.json

@@ -1,6 +1,6 @@
 {
   "name": "llamaindex",
-  "version": "0.6.11",
+  "version": "0.7.1",
   "license": "MIT",
   "type": "module",
   "keywords": [
@@ -29,13 +29,18 @@
     "@google-cloud/vertexai": "1.2.0",
     "@google/generative-ai": "0.12.0",
     "@grpc/grpc-js": "^1.11.1",
-    "@huggingface/inference": "^2.8.0",
+    "@llamaindex/anthropic": "workspace:*",
+    "@llamaindex/clip": "workspace:*",
     "@llamaindex/cloud": "workspace:*",
     "@llamaindex/core": "workspace:*",
+    "@llamaindex/deepinfra": "workspace:*",
     "@llamaindex/env": "workspace:*",
     "@llamaindex/groq": "workspace:*",
+    "@llamaindex/huggingface": "workspace:*",
     "@llamaindex/ollama": "workspace:*",
     "@llamaindex/openai": "workspace:*",
+    "@llamaindex/portkey-ai": "workspace:*",
+    "@llamaindex/replicate": "workspace:^0.0.2",
     "@mistralai/mistralai": "^1.0.4",
     "@mixedbread-ai/sdk": "^2.2.11",
     "@pinecone-database/pinecone": "^3.0.2",
@@ -48,9 +53,10 @@
     "@zilliz/milvus2-sdk-node": "^2.4.6",
     "ajv": "^8.17.1",
     "assemblyai": "^4.7.0",
-    "chromadb": "1.8.1",
+    "chromadb": "1.9.2",
     "cohere-ai": "7.13.0",
     "discord-api-types": "^0.37.98",
+    "gpt-tokenizer": "^2.5.0",
     "groq-sdk": "^0.6.1",
     "js-tiktoken": "^1.0.14",
     "lodash": "^4.17.21",
@@ -62,10 +68,8 @@
     "openai": "^4.60.0",
     "papaparse": "^5.4.1",
     "pathe": "^1.1.2",
-    "portkey-ai": "0.1.16",
     "rake-modified": "^1.0.8",
     "string-strip-html": "^13.4.8",
-    "tiktoken": "^1.0.15",
     "unpdf": "^0.11.0",
     "weaviate-client": "^3.1.4",
     "wikipedia": "^2.1.2",

packages/llamaindex/src/Settings.ts

@@ -2,7 +2,6 @@ import {
   type CallbackManager,
   Settings as CoreSettings,
 } from "@llamaindex/core/global";
-import { OpenAI } from "@llamaindex/openai";
 
 import { PromptHelper } from "@llamaindex/core/indices";
 
@@ -12,7 +11,6 @@ import {
   type NodeParser,
   SentenceSplitter,
 } from "@llamaindex/core/node-parser";
-import type { LoadTransformerEvent } from "@llamaindex/env";
 import { AsyncLocalStorage } from "@llamaindex/env";
 import type { ServiceContext } from "./ServiceContext.js";
 import {
@@ -21,12 +19,6 @@ import {
   withEmbeddedModel,
 } from "./internal/settings/EmbedModel.js";
 
-declare module "@llamaindex/core/global" {
-  interface LlamaIndexEventMaps {
-    "load-transformers": LoadTransformerEvent;
-  }
-}
-
 export type PromptConfig = {
   llm?: string;
   lang?: string;
@@ -61,12 +53,6 @@ class GlobalSettings implements Config {
   }
 
   get llm(): LLM {
-    // fixme: we might need check internal error instead of try-catch here
-    try {
-      CoreSettings.llm;
-    } catch (error) {
-      CoreSettings.llm = new OpenAI();
-    }
     return CoreSettings.llm;
   }
 

packages/llamaindex/src/agent/anthropic.ts

@@ -1,43 +1 @@
-import {
-  LLMAgent,
-  LLMAgentWorker,
-  type LLMAgentParams,
-} from "@llamaindex/core/agent";
-import type {
-  NonStreamingChatEngineParams,
-  StreamingChatEngineParams,
-} from "@llamaindex/core/chat-engine";
-import type { EngineResponse } from "@llamaindex/core/schema";
-import { Settings } from "../Settings.js";
-import { Anthropic } from "../llm/anthropic.js";
-
-export type AnthropicAgentParams = LLMAgentParams;
-
-export class AnthropicAgentWorker extends LLMAgentWorker {}
-
-export class AnthropicAgent extends LLMAgent {
-  constructor(params: AnthropicAgentParams) {
-    const llm =
-      params.llm ??
-      (Settings.llm instanceof Anthropic
-        ? (Settings.llm as Anthropic)
-        : new Anthropic());
-    super({
-      ...params,
-      llm,
-    });
-  }
-
-  async chat(params: NonStreamingChatEngineParams): Promise<EngineResponse>;
-  async chat(params: StreamingChatEngineParams): Promise<never>;
-  override async chat(
-    params: NonStreamingChatEngineParams | StreamingChatEngineParams,
-  ) {
-    const { stream } = params;
-    if (stream) {
-      // Anthropic does support this, but looks like it's not supported in the LITS LLM
-      throw new Error("Anthropic does not support streaming");
-    }
-    return super.chat(params);
-  }
-}
+export * from "@llamaindex/anthropic";

packages/llamaindex/src/cloud/LLamaCloudFileService.ts

@@ -41,7 +41,7 @@ export class LLamaCloudFileService {
   ) {
     initService();
     const { data: file } = await FilesService.uploadFileApiV1FilesPost({
-      path: { project_id: projectId },
+      query: { project_id: projectId },
       body: {
         upload_file: uploadFile,
       },
@@ -85,7 +85,7 @@ export class LLamaCloudFileService {
       await new Promise((resolve) => setTimeout(resolve, 100)); // Sleep for 100ms
     }
     throw new Error(
-      `File processing did not complete after ${maxAttempts} attempts.`,
+      `File processing did not complete after ${maxAttempts} attempts. Check your LlamaCloud index at https://cloud.llamaindex.ai/project/${projectId}/deploy/${pipelineId} for more details.`,
     );
   }
 

packages/llamaindex/src/cloud/LlamaCloudIndex.ts

@@ -1,19 +1,21 @@
 import type { BaseQueryEngine } from "@llamaindex/core/query-engine";
 import type { BaseSynthesizer } from "@llamaindex/core/response-synthesizers";
-import type { Document, TransformComponent } from "@llamaindex/core/schema";
+import type { Document } from "@llamaindex/core/schema";
 import { RetrieverQueryEngine } from "../engines/query/RetrieverQueryEngine.js";
 import type { BaseNodePostprocessor } from "../postprocessors/types.js";
 import type { CloudRetrieveParams } from "./LlamaCloudRetriever.js";
 import { LlamaCloudRetriever } from "./LlamaCloudRetriever.js";
-import { getPipelineCreate } from "./config.js";
 import type { CloudConstructorParams } from "./type.js";
-import { getAppBaseUrl, getProjectId, initService } from "./utils.js";
+import {
+  getAppBaseUrl,
+  getPipelineId,
+  getProjectId,
+  initService,
+} from "./utils.js";
 
-import { PipelinesService, ProjectsService } from "@llamaindex/cloud/api";
-import { SentenceSplitter } from "@llamaindex/core/node-parser";
+import { PipelinesService, type PipelineCreate } from "@llamaindex/cloud/api";
 import type { BaseRetriever } from "@llamaindex/core/retriever";
 import { getEnv } from "@llamaindex/env";
-import { OpenAIEmbedding } from "@llamaindex/openai";
 import { Settings } from "../Settings.js";
 
 export class LlamaCloudIndex {
@@ -28,10 +30,7 @@ export class LlamaCloudIndex {
     verbose = Settings.debug,
     raiseOnError = false,
   ): Promise<void> {
-    const pipelineId = await this.getPipelineId(
-      this.params.name,
-      this.params.projectName,
-    );
+    const pipelineId = await this.getPipelineId();
 
     if (verbose) {
       console.log("Waiting for pipeline ingestion: ");
@@ -78,10 +77,7 @@ export class LlamaCloudIndex {
     verbose = Settings.debug,
     raiseOnError = false,
   ): Promise<void> {
-    const pipelineId = await this.getPipelineId(
-      this.params.name,
-      this.params.projectName,
-    );
+    const pipelineId = await this.getPipelineId();
 
     if (verbose) {
       console.log("Loading data: ");
@@ -143,17 +139,13 @@ export class LlamaCloudIndex {
   public async getPipelineId(
     name?: string,
     projectName?: string,
+    organizationId?: string,
   ): Promise<string> {
-    const { data: pipelines } =
-      await PipelinesService.searchPipelinesApiV1PipelinesGet({
-        path: {
-          project_id: await this.getProjectId(projectName),
-          project_name: name ?? this.params.name,
-        },
-        throwOnError: true,
-      });
-
-    return pipelines[0]!.id;
+    return await getPipelineId(
+      name ?? this.params.name,
+      projectName ?? this.params.projectName,
+      organizationId ?? this.params.organizationId,
+    );
   }
 
   public async getProjectId(
@@ -166,75 +158,42 @@ export class LlamaCloudIndex {
     );
   }
 
+  /**
+   * Adds documents to the given index parameters. If the index does not exist, it will be created.
+   *
+   * @param params - An object containing the following properties:
+   *   - documents: An array of Document objects to be added to the index.
+   *   - verbose: Optional boolean to enable verbose logging.
+   *   - Additional properties from CloudConstructorParams.
+   * @returns A Promise that resolves to a new LlamaCloudIndex instance.
+   */
   static async fromDocuments(
     params: {
       documents: Document[];
-      transformations?: TransformComponent[];
       verbose?: boolean;
     } & CloudConstructorParams,
+    config?: {
+      embedding: PipelineCreate["embedding_config"];
+      transform: PipelineCreate["transform_config"];
+    },
   ): Promise<LlamaCloudIndex> {
-    initService(params);
-    const defaultTransformations: TransformComponent[] = [
-      new SentenceSplitter(),
-      new OpenAIEmbedding({
-        apiKey: getEnv("OPENAI_API_KEY"),
-      }),
-    ];
+    const index = new LlamaCloudIndex({ ...params });
+    await index.ensureIndex({ ...config, verbose: params.verbose ?? false });
+    await index.addDocuments(params.documents, params.verbose);
+    return index;
+  }
+
+  async addDocuments(documents: Document[], verbose?: boolean): Promise<void> {
     const apiUrl = getAppBaseUrl();
-
-    const pipelineCreateParams = await getPipelineCreate({
-      pipelineName: params.name,
-      pipelineType: "MANAGED",
-      inputNodes: params.documents,
-      transformations: params.transformations ?? defaultTransformations,
-    });
-
-    const { data: project } =
-      await ProjectsService.upsertProjectApiV1ProjectsPut({
-        path: {
-          organization_id: params.organizationId,
-        },
-        body: {
-          name: params.projectName ?? "default",
-        },
-        throwOnError: true,
-      });
-
-    if (!project.id) {
-      throw new Error("Project ID should be defined");
-    }
-
-    const { data: pipeline } =
-      await PipelinesService.upsertPipelineApiV1PipelinesPut({
-        path: {
-          project_id: project.id,
-        },
-        body: pipelineCreateParams.configured_transformations
-          ? {
-              name: params.name,
-              configured_transformations:
-                pipelineCreateParams.configured_transformations,
-            }
-          : {
-              name: params.name,
-            },
-        throwOnError: true,
-      });
-
-    if (!pipeline.id) {
-      throw new Error("Pipeline ID must be defined");
-    }
-
-    if (params.verbose) {
-      console.log(`Created pipeline ${pipeline.id} with name ${params.name}`);
-    }
+    const projectId = await this.getProjectId();
+    const pipelineId = await this.getPipelineId();
 
     await PipelinesService.upsertBatchPipelineDocumentsApiV1PipelinesPipelineIdDocumentsPut(
       {
         path: {
-          pipeline_id: pipeline.id,
+          pipeline_id: pipelineId,
         },
-        body: params.documents.map((doc) => ({
+        body: documents.map((doc) => ({
           metadata: doc.metadata,
           text: doc.text,
           excluded_embed_metadata_keys: doc.excludedEmbedMetadataKeys,
@@ -248,7 +207,7 @@ export class LlamaCloudIndex {
       const { data: pipelineStatus } =
         await PipelinesService.getPipelineStatusApiV1PipelinesPipelineIdStatusGet(
           {
-            path: { pipeline_id: pipeline.id },
+            path: { pipeline_id: pipelineId },
             throwOnError: true,
           },
         );
@@ -262,32 +221,30 @@ export class LlamaCloudIndex {
 
       if (pipelineStatus.status === "ERROR") {
         console.error(
-          `Some documents failed to ingest, check your pipeline logs at ${apiUrl}/project/${project.id}/deploy/${pipeline.id}`,
+          `Some documents failed to ingest, check your pipeline logs at ${apiUrl}/project/${projectId}/deploy/${pipelineId}`,
         );
         throw new Error("Some documents failed to ingest");
       }
 
       if (pipelineStatus.status === "PARTIAL_SUCCESS") {
         console.info(
-          `Documents ingestion partially succeeded, to check a more complete status check your pipeline at ${apiUrl}/project/${project.id}/deploy/${pipeline.id}`,
+          `Documents ingestion partially succeeded, to check a more complete status check your pipeline at ${apiUrl}/project/${projectId}/deploy/${pipelineId}`,
         );
         break;
       }
 
-      if (params.verbose) {
+      if (verbose) {
         process.stdout.write(".");
       }
 
       await new Promise((resolve) => setTimeout(resolve, 1000));
     }
 
-    if (params.verbose) {
+    if (verbose) {
       console.info(
-        `Ingestion completed, find your index at ${apiUrl}/project/${project.id}/deploy/${pipeline.id}`,
+        `Ingestion completed, find your index at ${apiUrl}/project/${projectId}/deploy/${pipelineId}`,
       );
     }
-
-    return new LlamaCloudIndex({ ...params });
   }
 
   asRetriever(params: CloudRetrieveParams = {}): BaseRetriever {
@@ -308,20 +265,12 @@ export class LlamaCloudIndex {
     return new RetrieverQueryEngine(
       retriever,
       params?.responseSynthesizer,
-      params?.preFilters,
       params?.nodePostprocessors,
     );
   }
 
   async insert(document: Document) {
-    const pipelineId = await this.getPipelineId(
-      this.params.name,
-      this.params.projectName,
-    );
-
-    if (!pipelineId) {
-      throw new Error("We couldn't find the pipeline ID for the given name");
-    }
+    const pipelineId = await this.getPipelineId();
 
     await PipelinesService.createBatchPipelineDocumentsApiV1PipelinesPipelineIdDocumentsPost(
       {
@@ -344,14 +293,7 @@ export class LlamaCloudIndex {
   }
 
   async delete(document: Document) {
-    const pipelineId = await this.getPipelineId(
-      this.params.name,
-      this.params.projectName,
-    );
-
-    if (!pipelineId) {
-      throw new Error("We couldn't find the pipeline ID for the given name");
-    }
+    const pipelineId = await this.getPipelineId();
 
     await PipelinesService.deletePipelineDocumentApiV1PipelinesPipelineIdDocumentsDocumentIdDelete(
       {
@@ -366,14 +308,7 @@ export class LlamaCloudIndex {
   }
 
   async refreshDoc(document: Document) {
-    const pipelineId = await this.getPipelineId(
-      this.params.name,
-      this.params.projectName,
-    );
-
-    if (!pipelineId) {
-      throw new Error("We couldn't find the pipeline ID for the given name");
-    }
+    const pipelineId = await this.getPipelineId();
 
     await PipelinesService.upsertBatchPipelineDocumentsApiV1PipelinesPipelineIdDocumentsPut(
       {
@@ -394,4 +329,71 @@ export class LlamaCloudIndex {
 
     await this.waitForDocumentIngestion([document.id_]);
   }
+
+  public async ensureIndex(config?: {
+    embedding?: PipelineCreate["embedding_config"];
+    transform?: PipelineCreate["transform_config"];
+    verbose?: boolean;
+  }): Promise<void> {
+    const projectId = await this.getProjectId();
+
+    const { data: pipelines } =
+      await PipelinesService.searchPipelinesApiV1PipelinesGet({
+        query: {
+          project_id: projectId,
+          pipeline_name: this.params.name,
+        },
+        throwOnError: true,
+      });
+
+    if (pipelines.length === 0) {
+      // no pipeline found, create a new one
+      let embeddingConfig = config?.embedding;
+      if (!embeddingConfig) {
+        // no embedding config provided, use OpenAI as default
+        const openAIApiKey = getEnv("OPENAI_API_KEY");
+        const embeddingModel = getEnv("EMBEDDING_MODEL");
+        if (!openAIApiKey || !embeddingModel) {
+          throw new Error(
+            "No embedding configuration provided. Fallback to OpenAI embedding model. OPENAI_API_KEY and EMBEDDING_MODEL environment variables must be set.",
+          );
+        }
+        embeddingConfig = {
+          type: "OPENAI_EMBEDDING",
+          component: {
+            api_key: openAIApiKey,
+            model_name: embeddingModel,
+          },
+        };
+      }
+
+      let transformConfig = config?.transform;
+      if (!transformConfig) {
+        transformConfig = {
+          mode: "auto",
+          chunk_size: 1024,
+          chunk_overlap: 200,
+        };
+      }
+
+      const { data: pipeline } =
+        await PipelinesService.upsertPipelineApiV1PipelinesPut({
+          query: {
+            project_id: projectId,
+          },
+          body: {
+            name: this.params.name,
+            embedding_config: embeddingConfig,
+            transform_config: transformConfig,
+          },
+          throwOnError: true,
+        });
+
+      if (config?.verbose) {
+        console.log(
+          `Created pipeline ${pipeline.id} with name ${pipeline.name}`,
+        );
+      }
+    }
+  }
 }

packages/llamaindex/src/cloud/LlamaCloudRetriever.ts

@@ -1,4 +1,5 @@
 import {
+  type MetadataFilter,
   type MetadataFilters,
   PipelinesService,
   type RetrievalParams,
@@ -11,7 +12,7 @@ import type { NodeWithScore } from "@llamaindex/core/schema";
 import { jsonToNode, ObjectType } from "@llamaindex/core/schema";
 import { extractText } from "@llamaindex/core/utils";
 import type { ClientParams, CloudConstructorParams } from "./type.js";
-import { getProjectId, initService } from "./utils.js";
+import { getPipelineId, initService } from "./utils.js";
 
 export type CloudRetrieveParams = Omit<
   RetrievalParams,
@@ -42,6 +43,24 @@ export class LlamaCloudRetriever extends BaseRetriever {
     });
   }
 
+  // LlamaCloud expects null values for filters, but LlamaIndexTS uses undefined for empty values
+  // This function converts the undefined values to null
+  private convertFilter(filters?: MetadataFilters): MetadataFilters | null {
+    if (!filters) return null;
+
+    const processFilter = (
+      filter: MetadataFilter | MetadataFilters,
+    ): MetadataFilter | MetadataFilters => {
+      if ("filters" in filter) {
+        // type MetadataFilters
+        return { ...filter, filters: filter.filters.map(processFilter) };
+      }
+      return { ...filter, value: filter.value ?? null };
+    };
+
+    return { ...filters, filters: filters.filters.map(processFilter) };
+  }
+
   constructor(params: CloudConstructorParams & CloudRetrieveParams) {
     super();
     this.clientParams = { apiKey: params.apiKey, baseUrl: params.baseUrl };
@@ -57,45 +76,24 @@ export class LlamaCloudRetriever extends BaseRetriever {
   }
 
   async _retrieve(query: QueryBundle): Promise<NodeWithScore[]> {
-    const { data: pipelines } =
-      await PipelinesService.searchPipelinesApiV1PipelinesGet({
-        query: {
-          project_id: await getProjectId(this.projectName, this.organizationId),
-          project_name: this.pipelineName,
-        },
-        throwOnError: true,
-      });
+    const pipelineId = await getPipelineId(
+      this.pipelineName,
+      this.projectName,
+      this.organizationId,
+    );
 
-    if (pipelines.length === 0 || !pipelines[0]!.id) {
-      throw new Error(
-        `No pipeline found with name ${this.pipelineName} in project ${this.projectName}`,
-      );
-    }
-
-    const { data: pipeline } =
-      await PipelinesService.getPipelineApiV1PipelinesPipelineIdGet({
-        path: {
-          pipeline_id: pipelines[0]!.id,
-        },
-        throwOnError: true,
-      });
-
-    if (!pipeline) {
-      throw new Error(
-        `No pipeline found with name ${this.pipelineName} in project ${this.projectName}`,
-      );
-    }
+    const filters = this.convertFilter(this.retrieveParams.filters);
 
     const { data: results } =
       await PipelinesService.runSearchApiV1PipelinesPipelineIdRetrievePost({
         throwOnError: true,
         path: {
-          pipeline_id: pipeline.id,
+          pipeline_id: pipelineId,
         },
         body: {
           ...this.retrieveParams,
           query: extractText(query),
-          search_filters: this.retrieveParams.filters as MetadataFilters,
+          search_filters: filters,
           dense_similarity_top_k: this.retrieveParams.similarityTopK!,
         },
       });

packages/llamaindex/src/cloud/config.ts

@@ -1,55 +0,0 @@
-import type {
-  ConfiguredTransformationItem,
-  PipelineCreate,
-  PipelineType,
-} from "@llamaindex/cloud/api";
-import { SentenceSplitter } from "@llamaindex/core/node-parser";
-import { BaseNode, type TransformComponent } from "@llamaindex/core/schema";
-import { OpenAIEmbedding } from "@llamaindex/openai";
-
-export type GetPipelineCreateParams = {
-  pipelineName: string;
-  pipelineType: PipelineType;
-  transformations?: TransformComponent[];
-  inputNodes?: BaseNode[];
-};
-
-function getTransformationConfig(
-  transformation: TransformComponent,
-): ConfiguredTransformationItem {
-  if (transformation instanceof SentenceSplitter) {
-    return {
-      configurable_transformation_type: "SENTENCE_AWARE_NODE_PARSER",
-      component: {
-        chunk_size: transformation.chunkSize, // TODO: set to public in SentenceSplitter
-        chunk_overlap: transformation.chunkOverlap, // TODO: set to public in SentenceSplitter
-        include_metadata: transformation.includeMetadata,
-        include_prev_next_rel: transformation.includePrevNextRel,
-      },
-    };
-  }
-  if (transformation instanceof OpenAIEmbedding) {
-    return {
-      configurable_transformation_type: "OPENAI_EMBEDDING",
-      component: {
-        model: transformation.model,
-        api_key: transformation.apiKey,
-        embed_batch_size: transformation.embedBatchSize,
-        dimensions: transformation.dimensions,
-      },
-    };
-  }
-  throw new Error(`Unsupported transformation: ${typeof transformation}`);
-}
-
-export async function getPipelineCreate(
-  params: GetPipelineCreateParams,
-): Promise<PipelineCreate> {
-  const { pipelineName, pipelineType, transformations = [] } = params;
-
-  return {
-    name: pipelineName,
-    configured_transformations: transformations.map(getTransformationConfig),
-    pipeline_type: pipelineType,
-  };
-}

packages/llamaindex/src/cloud/utils.ts

@@ -1,4 +1,8 @@
-import { client, ProjectsService } from "@llamaindex/cloud/api";
+import {
+  client,
+  PipelinesService,
+  ProjectsService,
+} from "@llamaindex/cloud/api";
 import { DEFAULT_BASE_URL } from "@llamaindex/core/global";
 import { getEnv } from "@llamaindex/env";
 import type { ClientParams } from "./type.js";
@@ -40,9 +44,9 @@ export async function getProjectId(
 ): Promise<string> {
   const { data: projects } = await ProjectsService.listProjectsApiV1ProjectsGet(
     {
-      path: {
+      query: {
         project_name: projectName,
-        organization_id: organizationId,
+        organization_id: organizationId ?? null,
       },
       throwOnError: true,
     },
@@ -66,3 +70,26 @@ export async function getProjectId(
 
   return project.id;
 }
+
+export async function getPipelineId(
+  name: string,
+  projectName: string,
+  organizationId?: string,
+): Promise<string> {
+  const { data: pipelines } =
+    await PipelinesService.searchPipelinesApiV1PipelinesGet({
+      query: {
+        project_id: await getProjectId(projectName, organizationId),
+        pipeline_name: name,
+      },
+      throwOnError: true,
+    });
+
+  if (pipelines.length === 0 || !pipelines[0]!.id) {
+    throw new Error(
+      `No pipeline found with name ${name} in project ${projectName}`,
+    );
+  }
+
+  return pipelines[0]!.id;
+}

packages/llamaindex/src/embeddings/ClipEmbedding.ts

@@ -1,139 +1 @@
-import { MultiModalEmbedding } from "@llamaindex/core/embeddings";
-import type { ImageType } from "@llamaindex/core/schema";
-import _ from "lodash";
-// only import type, to avoid bundling error
-import { loadTransformers } from "@llamaindex/env";
-import type {
-  CLIPTextModelWithProjection,
-  CLIPVisionModelWithProjection,
-  PreTrainedTokenizer,
-  Processor,
-} from "@xenova/transformers";
-import { Settings } from "../Settings.js";
-
-async function readImage(input: ImageType) {
-  const { RawImage } = await loadTransformers((transformer) => {
-    Settings.callbackManager.dispatchEvent(
-      "load-transformers",
-      {
-        transformer,
-      },
-      true,
-    );
-  });
-  if (input instanceof Blob) {
-    return await RawImage.fromBlob(input);
-  } else if (_.isString(input) || input instanceof URL) {
-    return await RawImage.fromURL(input);
-  } else {
-    throw new Error(`Unsupported input type: ${typeof input}`);
-  }
-}
-
-export enum ClipEmbeddingModelType {
-  XENOVA_CLIP_VIT_BASE_PATCH32 = "Xenova/clip-vit-base-patch32",
-  XENOVA_CLIP_VIT_BASE_PATCH16 = "Xenova/clip-vit-base-patch16",
-}
-
-export class ClipEmbedding extends MultiModalEmbedding {
-  modelType: ClipEmbeddingModelType =
-    ClipEmbeddingModelType.XENOVA_CLIP_VIT_BASE_PATCH16;
-
-  private tokenizer: PreTrainedTokenizer | null = null;
-  private processor: Processor | null = null;
-  private visionModel: CLIPVisionModelWithProjection | null = null;
-  private textModel: CLIPTextModelWithProjection | null = null;
-
-  constructor() {
-    super();
-  }
-
-  async getTokenizer() {
-    const { AutoTokenizer } = await loadTransformers((transformer) => {
-      Settings.callbackManager.dispatchEvent(
-        "load-transformers",
-        {
-          transformer,
-        },
-        true,
-      );
-    });
-    if (!this.tokenizer) {
-      this.tokenizer = await AutoTokenizer.from_pretrained(this.modelType);
-    }
-    return this.tokenizer;
-  }
-
-  async getProcessor() {
-    const { AutoProcessor } = await loadTransformers((transformer) => {
-      Settings.callbackManager.dispatchEvent(
-        "load-transformers",
-        {
-          transformer,
-        },
-        true,
-      );
-    });
-    if (!this.processor) {
-      this.processor = await AutoProcessor.from_pretrained(this.modelType);
-    }
-    return this.processor;
-  }
-
-  async getVisionModel() {
-    const { CLIPVisionModelWithProjection } = await loadTransformers(
-      (transformer) => {
-        Settings.callbackManager.dispatchEvent(
-          "load-transformers",
-          {
-            transformer,
-          },
-          true,
-        );
-      },
-    );
-    if (!this.visionModel) {
-      this.visionModel = await CLIPVisionModelWithProjection.from_pretrained(
-        this.modelType,
-      );
-    }
-
-    return this.visionModel;
-  }
-
-  async getTextModel() {
-    const { CLIPTextModelWithProjection } = await loadTransformers(
-      (transformer) => {
-        Settings.callbackManager.dispatchEvent(
-          "load-transformers",
-          {
-            transformer,
-          },
-          true,
-        );
-      },
-    );
-    if (!this.textModel) {
-      this.textModel = await CLIPTextModelWithProjection.from_pretrained(
-        this.modelType,
-      );
-    }
-
-    return this.textModel;
-  }
-
-  async getImageEmbedding(image: ImageType): Promise<number[]> {
-    const loadedImage = await readImage(image);
-    const imageInputs = await (await this.getProcessor())(loadedImage);
-    const { image_embeds } = await (await this.getVisionModel())(imageInputs);
-    return Array.from(image_embeds.data);
-  }
-
-  async getTextEmbedding(text: string): Promise<number[]> {
-    const textInputs = await (
-      await this.getTokenizer()
-    )([text], { padding: true, truncation: true });
-    const { text_embeds } = await (await this.getTextModel())(textInputs);
-    return text_embeds.data;
-  }
-}
+export * from "@llamaindex/clip";

packages/llamaindex/src/embeddings/CloudflareWorkerEmbedding.ts

@@ -1,17 +0,0 @@
-import { MultiModalEmbedding } from "@llamaindex/core/embeddings";
-import type { ImageType } from "@llamaindex/core/schema";
-
-/**
- * Cloudflare worker doesn't support image embeddings for now
- */
-export class CloudflareWorkerMultiModalEmbedding extends MultiModalEmbedding {
-  constructor() {
-    super();
-  }
-  getImageEmbedding(images: ImageType): Promise<number[]> {
-    throw new Error("Method not implemented.");
-  }
-  getTextEmbedding(text: string): Promise<number[]> {
-    throw new Error("Method not implemented.");
-  }
-}

packages/llamaindex/src/embeddings/DeepInfraEmbedding.ts

@@ -1,152 +1 @@
-import { BaseEmbedding } from "@llamaindex/core/embeddings";
-import type { MessageContentDetail } from "@llamaindex/core/llms";
-import { extractSingleText } from "@llamaindex/core/utils";
-import { getEnv } from "@llamaindex/env";
-
-const DEFAULT_MODEL = "sentence-transformers/clip-ViT-B-32";
-
-const API_TOKEN_ENV_VARIABLE_NAME = "DEEPINFRA_API_TOKEN";
-
-const API_ROOT = "https://api.deepinfra.com/v1/inference";
-
-const DEFAULT_TIMEOUT = 60 * 1000;
-
-const DEFAULT_MAX_RETRIES = 5;
-
-export interface DeepInfraEmbeddingResponse {
-  embeddings: number[][];
-  request_id: string;
-  inference_status: InferenceStatus;
-}
-
-export interface InferenceStatus {
-  status: string;
-  runtime_ms: number;
-  cost: number;
-  tokens_input: number;
-}
-
-const mapPrefixWithInputs = (prefix: string, inputs: string[]): string[] => {
-  return inputs.map((input) => (prefix ? `${prefix} ${input}` : input));
-};
-
-/**
- * DeepInfraEmbedding is an alias for DeepInfra that implements the BaseEmbedding interface.
- */
-export class DeepInfraEmbedding extends BaseEmbedding {
-  /**
-   * DeepInfra model to use
-   * @default "sentence-transformers/clip-ViT-B-32"
-   * @see https://deepinfra.com/models/embeddings
-   */
-  model: string;
-
-  /**
-   * DeepInfra API token
-   * @see https://deepinfra.com/dash/api_keys
-   * If not provided, it will try to get the token from the environment variable `DEEPINFRA_API_TOKEN`
-   *
-   */
-  apiToken: string;
-
-  /**
-   * Prefix to add to the query
-   * @default ""
-   */
-  queryPrefix: string;
-
-  /**
-   * Prefix to add to the text
-   * @default ""
-   */
-  textPrefix: string;
-
-  /**
-   *
-   * @default 5
-   */
-  maxRetries: number;
-
-  /**
-   *
-   * @default 60 * 1000
-   */
-  timeout: number;
-
-  constructor(init?: Partial<DeepInfraEmbedding>) {
-    super();
-
-    this.model = init?.model ?? DEFAULT_MODEL;
-    this.apiToken = init?.apiToken ?? getEnv(API_TOKEN_ENV_VARIABLE_NAME) ?? "";
-    this.queryPrefix = init?.queryPrefix ?? "";
-    this.textPrefix = init?.textPrefix ?? "";
-    this.maxRetries = init?.maxRetries ?? DEFAULT_MAX_RETRIES;
-    this.timeout = init?.timeout ?? DEFAULT_TIMEOUT;
-  }
-
-  async getTextEmbedding(text: string): Promise<number[]> {
-    const texts = mapPrefixWithInputs(this.textPrefix, [text]);
-    const embeddings = await this.getDeepInfraEmbedding(texts);
-    return embeddings[0]!;
-  }
-
-  async getQueryEmbedding(
-    query: MessageContentDetail,
-  ): Promise<number[] | null> {
-    const text = extractSingleText(query);
-    if (text) {
-      const queries = mapPrefixWithInputs(this.queryPrefix, [text]);
-      const embeddings = await this.getDeepInfraEmbedding(queries);
-      return embeddings[0]!;
-    } else {
-      return null;
-    }
-  }
-
-  getTextEmbeddings = async (texts: string[]): Promise<number[][]> => {
-    const textsWithPrefix = mapPrefixWithInputs(this.textPrefix, texts);
-    return this.getDeepInfraEmbedding(textsWithPrefix);
-  };
-
-  async getQueryEmbeddings(queries: string[]): Promise<number[][]> {
-    const queriesWithPrefix = mapPrefixWithInputs(this.queryPrefix, queries);
-    return await this.getDeepInfraEmbedding(queriesWithPrefix);
-  }
-
-  private async getDeepInfraEmbedding(inputs: string[]): Promise<number[][]> {
-    const url = this.getUrl(this.model);
-
-    for (let attempt = 0; attempt < this.maxRetries; attempt++) {
-      const controller = new AbortController();
-      const id = setTimeout(() => controller.abort(), this.timeout);
-
-      try {
-        const response = await fetch(url, {
-          method: "POST",
-          headers: {
-            "Content-Type": "application/json",
-            Authorization: `Bearer ${this.apiToken}`,
-          },
-          body: JSON.stringify({ inputs }),
-          signal: controller.signal,
-        });
-        if (!response.ok) {
-          throw new Error(`Request failed with status ${response.status}`);
-        }
-
-        const responseJson: DeepInfraEmbeddingResponse = await response.json();
-        return responseJson.embeddings;
-      } catch (error) {
-        console.error(`Attempt ${attempt + 1} failed: ${error}`);
-      } finally {
-        clearTimeout(id);
-      }
-    }
-
-    throw new Error("Exceeded maximum retries");
-  }
-
-  private getUrl(model: string): string {
-    return `${API_ROOT}/${model}`;
-  }
-}
+export * from "@llamaindex/deepinfra";

packages/llamaindex/src/embeddings/HuggingFaceEmbedding.ts

@@ -1,110 +1 @@
-import { HfInference } from "@huggingface/inference";
-import { BaseEmbedding } from "@llamaindex/core/embeddings";
-import { loadTransformers } from "@llamaindex/env";
-import { Settings } from "../Settings.js";
-
-export enum HuggingFaceEmbeddingModelType {
-  XENOVA_ALL_MINILM_L6_V2 = "Xenova/all-MiniLM-L6-v2",
-  XENOVA_ALL_MPNET_BASE_V2 = "Xenova/all-mpnet-base-v2",
-}
-
-/**
- * Uses feature extraction from '@xenova/transformers' to generate embeddings.
- * Per default the model [XENOVA_ALL_MINILM_L6_V2](https://huggingface.co/Xenova/all-MiniLM-L6-v2) is used.
- *
- * Can be changed by setting the `modelType` parameter in the constructor, e.g.:
- * ```
- * new HuggingFaceEmbedding({
- *     modelType: HuggingFaceEmbeddingModelType.XENOVA_ALL_MPNET_BASE_V2,
- * });
- * ```
- *
- * @extends BaseEmbedding
- */
-export class HuggingFaceEmbedding extends BaseEmbedding {
-  modelType: string = HuggingFaceEmbeddingModelType.XENOVA_ALL_MINILM_L6_V2;
-  quantized: boolean = true;
-
-  private extractor: any;
-
-  constructor(init?: Partial<HuggingFaceEmbedding>) {
-    super();
-    Object.assign(this, init);
-  }
-
-  async getExtractor() {
-    if (!this.extractor) {
-      const { pipeline } = await loadTransformers((transformer) => {
-        Settings.callbackManager.dispatchEvent(
-          "load-transformers",
-          {
-            transformer,
-          },
-          true,
-        );
-      });
-      this.extractor = await pipeline("feature-extraction", this.modelType, {
-        quantized: this.quantized,
-      });
-    }
-    return this.extractor;
-  }
-
-  override async getTextEmbedding(text: string): Promise<number[]> {
-    const extractor = await this.getExtractor();
-    const output = await extractor(text, { pooling: "mean", normalize: true });
-    return Array.from(output.data);
-  }
-}
-
-// Workaround to get the Options type from @huggingface/inference@2.7.0
-type HfInferenceOptions = ConstructorParameters<typeof HfInference>[1];
-
-export type HFConfig = HfInferenceOptions & {
-  model: string;
-  accessToken: string;
-  endpoint?: string;
-};
-
-/**
- * Uses feature extraction from Hugging Face's Inference API to generate embeddings.
- *
- * Set the `model` and `accessToken` parameter in the constructor, e.g.:
- * ```
- * new HuggingFaceInferenceAPIEmbedding({
- *     model: HuggingFaceEmbeddingModelType.XENOVA_ALL_MPNET_BASE_V2,
- *     accessToken: "<your-access-token>"
- * });
- * ```
- *
- * @extends BaseEmbedding
- */
-export class HuggingFaceInferenceAPIEmbedding extends BaseEmbedding {
-  model: string;
-  hf: HfInference;
-
-  constructor(init: HFConfig) {
-    super();
-    const { model, accessToken, endpoint, ...hfInferenceOpts } = init;
-
-    this.hf = new HfInference(accessToken, hfInferenceOpts);
-    this.model = model;
-    if (endpoint) this.hf.endpoint(endpoint);
-  }
-
-  async getTextEmbedding(text: string): Promise<number[]> {
-    const res = await this.hf.featureExtraction({
-      model: this.model,
-      inputs: text,
-    });
-    return res as number[];
-  }
-
-  getTextEmbeddings = async (texts: string[]): Promise<Array<number[]>> => {
-    const res = await this.hf.featureExtraction({
-      model: this.model,
-      inputs: texts,
-    });
-    return res as number[][];
-  };
-}
+export * from "@llamaindex/huggingface";

packages/llamaindex/src/embeddings/fireworks.ts

@@ -2,7 +2,7 @@ import { getEnv } from "@llamaindex/env";
 import { OpenAIEmbedding } from "@llamaindex/openai";
 
 export class FireworksEmbedding extends OpenAIEmbedding {
-  constructor(init?: Partial<OpenAIEmbedding>) {
+  constructor(init?: Omit<Partial<OpenAIEmbedding>, "session">) {
     const {
       apiKey = getEnv("FIREWORKS_API_KEY"),
       additionalSessionOptions = {},

packages/llamaindex/src/embeddings/index.ts

@@ -1,4 +1,5 @@
 export * from "@llamaindex/core/embeddings";
+export { ClipEmbedding, ClipEmbeddingModelType } from "./ClipEmbedding.js";
 export { DeepInfraEmbedding } from "./DeepInfraEmbedding.js";
 export { FireworksEmbedding } from "./fireworks.js";
 export * from "./GeminiEmbedding.js";
@@ -9,5 +10,3 @@ export * from "./MixedbreadAIEmbeddings.js";
 export { OllamaEmbedding } from "./OllamaEmbedding.js";
 export * from "./OpenAIEmbedding.js";
 export { TogetherEmbedding } from "./together.js";
-// ClipEmbedding might not work in non-node.js runtime, but it doesn't have side effects
-export { ClipEmbedding, ClipEmbeddingModelType } from "./ClipEmbedding.js";

packages/llamaindex/src/embeddings/together.ts

@@ -2,7 +2,7 @@ import { getEnv } from "@llamaindex/env";
 import { OpenAIEmbedding } from "@llamaindex/openai";
 
 export class TogetherEmbedding extends OpenAIEmbedding {
-  constructor(init?: Partial<OpenAIEmbedding>) {
+  constructor(init?: Omit<Partial<OpenAIEmbedding>, "session">) {
     const {
       apiKey = getEnv("TOGETHER_API_KEY"),
       additionalSessionOptions = {},

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

@@ -14,12 +14,10 @@ export class RetrieverQueryEngine extends BaseQueryEngine {
   retriever: BaseRetriever;
   responseSynthesizer: BaseSynthesizer;
   nodePostprocessors: BaseNodePostprocessor[];
-  preFilters?: unknown;
 
   constructor(
     retriever: BaseRetriever,
     responseSynthesizer?: BaseSynthesizer,
-    preFilters?: unknown,
     nodePostprocessors?: BaseNodePostprocessor[],
   ) {
     super(async (strOrQueryBundle, stream) => {
@@ -52,7 +50,6 @@ export class RetrieverQueryEngine extends BaseQueryEngine {
     this.retriever = retriever;
     this.responseSynthesizer =
       responseSynthesizer || getResponseSynthesizer("compact");
-    this.preFilters = preFilters;
     this.nodePostprocessors = nodePostprocessors || [];
   }
 

packages/llamaindex/src/extractors/MetadataExtractors.ts

@@ -1,14 +1,20 @@
 import type { LLM } from "@llamaindex/core/llms";
+import {
+  PromptTemplate,
+  defaultKeywordExtractPrompt,
+  defaultQuestionExtractPrompt,
+  defaultSummaryPrompt,
+  defaultTitleCombinePromptTemplate,
+  defaultTitleExtractorPromptTemplate,
+  type KeywordExtractPrompt,
+  type QuestionExtractPrompt,
+  type SummaryPrompt,
+  type TitleCombinePrompt,
+  type TitleExtractorPrompt,
+} from "@llamaindex/core/prompts";
 import type { BaseNode } from "@llamaindex/core/schema";
 import { MetadataMode, TextNode } from "@llamaindex/core/schema";
 import { OpenAI } from "@llamaindex/openai";
-import {
-  defaultKeywordExtractorPromptTemplate,
-  defaultQuestionAnswerPromptTemplate,
-  defaultSummaryExtractorPromptTemplate,
-  defaultTitleCombinePromptTemplate,
-  defaultTitleExtractorPromptTemplate,
-} from "./prompts.js";
 import { BaseExtractor } from "./types.js";
 
 const STRIP_REGEX = /(\r\n|\n|\r)/gm;
@@ -16,6 +22,7 @@ const STRIP_REGEX = /(\r\n|\n|\r)/gm;
 type KeywordExtractArgs = {
   llm?: LLM;
   keywords?: number;
+  promptTemplate?: KeywordExtractPrompt["template"];
 };
 
 type ExtractKeyword = {
@@ -39,6 +46,12 @@ export class KeywordExtractor extends BaseExtractor {
    */
   keywords: number = 5;
 
+  /**
+   * The prompt template to use for the question extractor.
+   * @type {string}
+   */
+  promptTemplate: KeywordExtractPrompt;
+
   /**
    * Constructor for the KeywordExtractor class.
    * @param {LLM} llm LLM instance.
@@ -53,6 +66,12 @@ export class KeywordExtractor extends BaseExtractor {
 
     this.llm = options?.llm ?? new OpenAI();
     this.keywords = options?.keywords ?? 5;
+    this.promptTemplate = options?.promptTemplate
+      ? new PromptTemplate({
+          templateVars: ["context", "maxKeywords"],
+          template: options.promptTemplate,
+        })
+      : defaultKeywordExtractPrompt;
   }
 
   /**
@@ -66,9 +85,9 @@ export class KeywordExtractor extends BaseExtractor {
     }
 
     const completion = await this.llm.complete({
-      prompt: defaultKeywordExtractorPromptTemplate({
-        contextStr: node.getContent(MetadataMode.ALL),
-        keywords: this.keywords,
+      prompt: this.promptTemplate.format({
+        context: node.getContent(MetadataMode.ALL),
+        maxKeywords: this.keywords.toString(),
       }),
     });
 
@@ -93,8 +112,8 @@ export class KeywordExtractor extends BaseExtractor {
 type TitleExtractorsArgs = {
   llm?: LLM;
   nodes?: number;
-  nodeTemplate?: string;
-  combineTemplate?: string;
+  nodeTemplate?: TitleExtractorPrompt["template"];
+  combineTemplate?: TitleCombinePrompt["template"];
 };
 
 type ExtractTitle = {
@@ -129,19 +148,19 @@ export class TitleExtractor extends BaseExtractor {
    * The prompt template to use for the title extractor.
    * @type {string}
    */
-  nodeTemplate: string;
+  nodeTemplate: TitleExtractorPrompt;
 
   /**
    * The prompt template to merge title with..
    * @type {string}
    */
-  combineTemplate: string;
+  combineTemplate: TitleCombinePrompt;
 
   /**
    * Constructor for the TitleExtractor class.
    * @param {LLM} llm LLM instance.
    * @param {number} nodes Number of nodes to extract titles from.
-   * @param {string} nodeTemplate The prompt template to use for the title extractor.
+   * @param {TitleExtractorPrompt} nodeTemplate The prompt template to use for the title extractor.
    * @param {string} combineTemplate The prompt template to merge title with..
    */
   constructor(options?: TitleExtractorsArgs) {
@@ -150,10 +169,19 @@ export class TitleExtractor extends BaseExtractor {
     this.llm = options?.llm ?? new OpenAI();
     this.nodes = options?.nodes ?? 5;
 
-    this.nodeTemplate =
-      options?.nodeTemplate ?? defaultTitleExtractorPromptTemplate();
-    this.combineTemplate =
-      options?.combineTemplate ?? defaultTitleCombinePromptTemplate();
+    this.nodeTemplate = options?.nodeTemplate
+      ? new PromptTemplate({
+          templateVars: ["context"],
+          template: options.nodeTemplate,
+        })
+      : defaultTitleExtractorPromptTemplate;
+
+    this.combineTemplate = options?.combineTemplate
+      ? new PromptTemplate({
+          templateVars: ["context"],
+          template: options.combineTemplate,
+        })
+      : defaultTitleCombinePromptTemplate;
   }
 
   /**
@@ -218,8 +246,8 @@ export class TitleExtractor extends BaseExtractor {
       const titleCandidates = await this.getTitlesCandidates(nodes);
       const combinedTitles = titleCandidates.join(", ");
       const completion = await this.llm.complete({
-        prompt: defaultTitleCombinePromptTemplate({
-          contextStr: combinedTitles,
+        prompt: this.combineTemplate.format({
+          context: combinedTitles,
         }),
       });
 
@@ -232,8 +260,8 @@ export class TitleExtractor extends BaseExtractor {
   private async getTitlesCandidates(nodes: BaseNode[]): Promise<string[]> {
     const titleJobs = nodes.map(async (node) => {
       const completion = await this.llm.complete({
-        prompt: defaultTitleExtractorPromptTemplate({
-          contextStr: node.getContent(MetadataMode.ALL),
+        prompt: this.nodeTemplate.format({
+          context: node.getContent(MetadataMode.ALL),
         }),
       });
 
@@ -247,7 +275,7 @@ export class TitleExtractor extends BaseExtractor {
 type QuestionAnswerExtractArgs = {
   llm?: LLM;
   questions?: number;
-  promptTemplate?: string;
+  promptTemplate?: QuestionExtractPrompt["template"];
   embeddingOnly?: boolean;
 };
 
@@ -276,7 +304,7 @@ export class QuestionsAnsweredExtractor extends BaseExtractor {
    * The prompt template to use for the question extractor.
    * @type {string}
    */
-  promptTemplate: string;
+  promptTemplate: QuestionExtractPrompt;
 
   /**
    * Wheter to use metadata for embeddings only
@@ -289,7 +317,7 @@ export class QuestionsAnsweredExtractor extends BaseExtractor {
    * Constructor for the QuestionsAnsweredExtractor class.
    * @param {LLM} llm LLM instance.
    * @param {number} questions Number of questions to generate.
-   * @param {string} promptTemplate The prompt template to use for the question extractor.
+   * @param {TextQAPrompt} promptTemplate The prompt template to use for the question extractor.
    * @param {boolean} embeddingOnly Wheter to use metadata for embeddings only.
    */
   constructor(options?: QuestionAnswerExtractArgs) {
@@ -300,12 +328,14 @@ export class QuestionsAnsweredExtractor extends BaseExtractor {
 
     this.llm = options?.llm ?? new OpenAI();
     this.questions = options?.questions ?? 5;
-    this.promptTemplate =
-      options?.promptTemplate ??
-      defaultQuestionAnswerPromptTemplate({
-        numQuestions: this.questions,
-        contextStr: "",
-      });
+    this.promptTemplate = options?.promptTemplate
+      ? new PromptTemplate({
+          templateVars: ["numQuestions", "context"],
+          template: options.promptTemplate,
+        }).partialFormat({
+          numQuestions: "5",
+        })
+      : defaultQuestionExtractPrompt;
     this.embeddingOnly = options?.embeddingOnly ?? false;
   }
 
@@ -323,9 +353,9 @@ export class QuestionsAnsweredExtractor extends BaseExtractor {
 
     const contextStr = node.getContent(this.metadataMode);
 
-    const prompt = defaultQuestionAnswerPromptTemplate({
-      contextStr,
-      numQuestions: this.questions,
+    const prompt = this.promptTemplate.format({
+      context: contextStr,
+      numQuestions: this.questions.toString(),
     });
 
     const questions = await this.llm.complete({
@@ -356,7 +386,7 @@ export class QuestionsAnsweredExtractor extends BaseExtractor {
 type SummaryExtractArgs = {
   llm?: LLM;
   summaries?: string[];
-  promptTemplate?: string;
+  promptTemplate?: SummaryPrompt["template"];
 };
 
 type ExtractSummary = {
@@ -385,7 +415,7 @@ export class SummaryExtractor extends BaseExtractor {
    * The prompt template to use for the summary extractor.
    * @type {string}
    */
-  promptTemplate: string;
+  promptTemplate: SummaryPrompt;
 
   private selfSummary: boolean;
   private prevSummary: boolean;
@@ -404,8 +434,12 @@ export class SummaryExtractor extends BaseExtractor {
 
     this.llm = options?.llm ?? new OpenAI();
     this.summaries = summaries;
-    this.promptTemplate =
-      options?.promptTemplate ?? defaultSummaryExtractorPromptTemplate();
+    this.promptTemplate = options?.promptTemplate
+      ? new PromptTemplate({
+          templateVars: ["context"],
+          template: options.promptTemplate,
+        })
+      : defaultSummaryPrompt;
 
     this.selfSummary = summaries?.includes("self") ?? false;
     this.prevSummary = summaries?.includes("prev") ?? false;
@@ -422,10 +456,10 @@ export class SummaryExtractor extends BaseExtractor {
       return "";
     }
 
-    const contextStr = node.getContent(this.metadataMode);
+    const context = node.getContent(this.metadataMode);
 
-    const prompt = defaultSummaryExtractorPromptTemplate({
-      contextStr,
+    const prompt = this.promptTemplate.format({
+      context,
     });
 
     const summary = await this.llm.complete({

packages/llamaindex/src/extractors/prompts.ts

@@ -1,74 +0,0 @@
-export interface DefaultPromptTemplate {
-  contextStr: string;
-}
-
-export interface DefaultKeywordExtractorPromptTemplate
-  extends DefaultPromptTemplate {
-  keywords: number;
-}
-
-export interface DefaultQuestionAnswerPromptTemplate
-  extends DefaultPromptTemplate {
-  numQuestions: number;
-}
-
-export interface DefaultNodeTextTemplate {
-  metadataStr: string;
-  content: string;
-}
-
-export const defaultKeywordExtractorPromptTemplate = ({
-  contextStr = "",
-  keywords = 5,
-}: DefaultKeywordExtractorPromptTemplate) => `${contextStr} 
-Give ${keywords} unique keywords for this document. 
-Format as comma separated. 
-Keywords: `;
-
-export const defaultTitleExtractorPromptTemplate = (
-  { contextStr = "" }: DefaultPromptTemplate = {
-    contextStr: "",
-  },
-) => `${contextStr}
-Give a title that summarizes all of the unique entities, titles or themes found in the context. 
-Title: `;
-
-export const defaultTitleCombinePromptTemplate = (
-  { contextStr = "" }: DefaultPromptTemplate = {
-    contextStr: "",
-  },
-) => `${contextStr} 
-Based on the above candidate titles and contents, what is the comprehensive title for this document? 
-Title: `;
-
-export const defaultQuestionAnswerPromptTemplate = (
-  { contextStr = "", numQuestions = 5 }: DefaultQuestionAnswerPromptTemplate = {
-    contextStr: "",
-    numQuestions: 5,
-  },
-) => `${contextStr}
-Given the contextual informations, generate ${numQuestions} questions this context can provides specific answers to which are unlikely to be found else where. Higher-level summaries of surrounding context may be provideds as well. 
-Try using these summaries to generate better questions that this context can answer.
-`;
-
-export const defaultSummaryExtractorPromptTemplate = (
-  { contextStr = "" }: DefaultPromptTemplate = {
-    contextStr: "",
-  },
-) => `${contextStr}
-Summarize the key topics and entities of the sections.
-Summary: `;
-
-export const defaultNodeTextTemplate = ({
-  metadataStr = "",
-  content = "",
-}: {
-  metadataStr?: string;
-  content?: string;
-} = {}) => `[Excerpt from document]
-${metadataStr}
-Excerpt:
------
-${content}
------
-`;

packages/llamaindex/src/extractors/types.ts

@@ -1,10 +1,10 @@
+import { defaultNodeTextTemplate } from "@llamaindex/core/prompts";
 import {
   BaseNode,
   MetadataMode,
   TextNode,
   TransformComponent,
 } from "@llamaindex/core/schema";
-import { defaultNodeTextTemplate } from "./prompts.js";
 
 /*
  * Abstract class for all extractors.
@@ -71,7 +71,7 @@ export abstract class BaseExtractor extends TransformComponent {
         if (newNodes[idx] instanceof TextNode) {
           newNodes[idx] = new TextNode({
             ...newNodes[idx],
-            textTemplate: defaultNodeTextTemplate(),
+            textTemplate: defaultNodeTextTemplate.format(),
           });
         }
       }

packages/llamaindex/src/index.edge.ts

@@ -1,3 +1,20 @@
+//#region initial setup for OpenAI
+import { OpenAI } from "@llamaindex/openai";
+import { Settings } from "./Settings.js";
+
+try {
+  Settings.llm;
+} catch {
+  Settings.llm = new OpenAI();
+}
+
+//#endregion
+
+export {
+  LlamaParseReader,
+  type Language,
+  type ResultType,
+} from "@llamaindex/cloud/reader";
 export * from "@llamaindex/core/agent";
 export * from "@llamaindex/core/chat-engine";
 export {
@@ -23,12 +40,12 @@ export type {
   JSONArray,
   JSONObject,
   JSONValue,
+  LlamaIndexEventMaps,
   LLMEndEvent,
   LLMStartEvent,
   LLMStreamEvent,
   LLMToolCallEvent,
   LLMToolResultEvent,
-  LlamaIndexEventMaps,
 } from "@llamaindex/core/global";
 export * from "@llamaindex/core/indices";
 export * from "@llamaindex/core/llms";
@@ -56,7 +73,7 @@ export * from "./postprocessors/index.js";
 export * from "./QuestionGenerator.js";
 export * from "./selectors/index.js";
 export * from "./ServiceContext.js";
-export { Settings } from "./Settings.js";
 export * from "./storage/StorageContext.js";
 export * from "./tools/index.js";
 export * from "./types.js";
+export { Settings };