Release 0.6.19 (#1318)

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

apps/docs/CHANGELOG.md

@@ -1,5 +1,38 @@
 # docs
 
+## 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

apps/docs/package.json

@@ -1,6 +1,6 @@
 {
   "name": "docs",
-  "version": "0.0.84",
+  "version": "0.0.88",
   "private": true,
   "scripts": {
     "docusaurus": "docusaurus",

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/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

@@ -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 +1,38 @@
 # @llamaindex/autotool
 
+## 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

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

@@ -1,5 +1,42 @@
 # @llamaindex/autotool-01-node-example
 
+## 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

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

@@ -13,5 +13,5 @@
   "scripts": {
     "start": "node --import tsx --import @llamaindex/autotool/node ./src/index.ts"
   },
-  "version": "0.0.24"
+  "version": "0.0.28"
 }

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

@@ -1,5 +1,42 @@
 # @llamaindex/autotool-02-next-example
 
+## 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

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

@@ -1,7 +1,7 @@
 {
   "name": "@llamaindex/autotool-02-next-example",
   "private": true,
-  "version": "0.1.68",
+  "version": "0.1.72",
   "scripts": {
     "dev": "next dev",
     "build": "next build",

packages/autotool/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@llamaindex/autotool",
   "type": "module",
-  "version": "3.0.15",
+  "version": "3.0.19",
   "description": "auto transpile your JS function to LLM Agent compatible",
   "files": [
     "dist",

packages/cloud/CHANGELOG.md

@@ -1,5 +1,27 @@
 # @llamaindex/cloud
 
+## 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

packages/cloud/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@llamaindex/cloud",
-  "version": "0.2.11",
+  "version": "0.2.14",
   "type": "module",
   "license": "MIT",
   "scripts": {

packages/community/CHANGELOG.md

@@ -1,5 +1,27 @@
 # @llamaindex/community
 
+## 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

packages/community/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@llamaindex/community",
   "description": "Community package for LlamaIndexTS",
-  "version": "0.0.44",
+  "version": "0.0.47",
   "type": "module",
   "types": "dist/type/index.d.ts",
   "main": "dist/cjs/index.js",

packages/core/CHANGELOG.md

@@ -1,5 +1,24 @@
 # @llamaindex/core
 
+## 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

packages/core/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@llamaindex/core",
   "type": "module",
-  "version": "0.2.9",
+  "version": "0.2.12",
   "description": "LlamaIndex Core Module",
   "exports": {
     "./agent": {

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/prompt.ts

@@ -64,11 +64,13 @@ 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:`,
 });
 

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/schema/node.ts

@@ -479,7 +479,7 @@ export function buildNodeFromSplits(
     ) {
       const imageDoc = doc as ImageNode;
       const imageNode = new ImageNode({
-        id_: imageDoc.id_ ?? idGenerator(i, imageDoc),
+        id_: idGenerator(i, imageDoc),
         text: textChunk,
         image: imageDoc.image,
         embedding: imageDoc.embedding,
@@ -496,7 +496,7 @@ export function buildNodeFromSplits(
     ) {
       const textDoc = doc as TextNode;
       const node = new TextNode({
-        id_: textDoc.id_ ?? idGenerator(i, textDoc),
+        id_: idGenerator(i, textDoc),
         text: textChunk,
         embedding: textDoc.embedding,
         excludedEmbedMetadataKeys: [...textDoc.excludedEmbedMetadataKeys],

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/experimental/CHANGELOG.md

@@ -1,5 +1,38 @@
 # @llamaindex/experimental
 
+## 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

packages/experimental/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@llamaindex/experimental",
   "description": "Experimental package for LlamaIndexTS",
-  "version": "0.0.93",
+  "version": "0.0.97",
   "type": "module",
   "types": "dist/type/index.d.ts",
   "main": "dist/cjs/index.js",

packages/llamaindex/CHANGELOG.md

@@ -1,5 +1,53 @@
 # llamaindex
 
+## 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

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

@@ -1,5 +1,38 @@
 # @llamaindex/cloudflare-worker-agent-test
 
+## 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

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

@@ -1,6 +1,6 @@
 {
   "name": "@llamaindex/cloudflare-worker-agent-test",
-  "version": "0.0.77",
+  "version": "0.0.81",
   "type": "module",
   "private": true,
   "scripts": {

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

@@ -1,5 +1,23 @@
 # @llamaindex/llama-parse-browser-test
 
+## 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

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

@@ -1,7 +1,7 @@
 {
   "name": "@llamaindex/llama-parse-browser-test",
   "private": true,
-  "version": "0.0.7",
+  "version": "0.0.10",
   "type": "module",
   "scripts": {
     "dev": "vite",

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

@@ -1,5 +1,38 @@
 # @llamaindex/next-agent-test
 
+## 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

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

@@ -1,6 +1,6 @@
 {
   "name": "@llamaindex/next-agent-test",
-  "version": "0.1.77",
+  "version": "0.1.81",
   "private": true,
   "scripts": {
     "dev": "next dev",

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

@@ -1,5 +1,38 @@
 # test-edge-runtime
 
+## 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

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

@@ -1,6 +1,6 @@
 {
   "name": "@llamaindex/nextjs-edge-runtime-test",
-  "version": "0.1.76",
+  "version": "0.1.80",
   "private": true,
   "scripts": {
     "dev": "next dev",

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

@@ -1,5 +1,38 @@
 # @llamaindex/next-node-runtime
 
+## 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

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

@@ -1,6 +1,6 @@
 {
   "name": "@llamaindex/next-node-runtime-test",
-  "version": "0.0.58",
+  "version": "0.0.62",
   "private": true,
   "scripts": {
     "dev": "next dev",

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

@@ -1,5 +1,38 @@
 # @llamaindex/waku-query-engine-test
 
+## 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

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

@@ -1,6 +1,6 @@
 {
   "name": "@llamaindex/waku-query-engine-test",
-  "version": "0.0.77",
+  "version": "0.0.81",
   "type": "module",
   "private": true,
   "scripts": {

packages/llamaindex/package.json

@@ -1,6 +1,6 @@
 {
   "name": "llamaindex",
-  "version": "0.6.15",
+  "version": "0.6.19",
   "license": "MIT",
   "type": "module",
   "keywords": [

packages/llamaindex/src/cloud/LLamaCloudFileService.ts

@@ -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({
+          path: {
+            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/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/indices/keyword/index.ts

@@ -246,7 +246,6 @@ export class KeywordTableIndex extends BaseIndex<KeywordTable> {
     return new RetrieverQueryEngine(
       retriever ?? this.asRetriever(),
       responseSynthesizer,
-      options?.preFilters,
       options?.nodePostprocessors,
     );
   }

packages/llamaindex/src/indices/summary/index.ts

@@ -189,7 +189,6 @@ export class SummaryIndex extends BaseIndex<IndexList> {
     return new RetrieverQueryEngine(
       retriever,
       responseSynthesizer,
-      options?.preFilters,
       options?.nodePostprocessors,
     );
   }

packages/llamaindex/src/indices/vectorStore/index.ts

@@ -298,9 +298,8 @@ export class VectorStoreIndex extends BaseIndex<IndexDict> {
       similarityTopK,
     } = options ?? {};
     return new RetrieverQueryEngine(
-      retriever ?? this.asRetriever({ similarityTopK }),
+      retriever ?? this.asRetriever({ similarityTopK, filters: preFilters }),
       responseSynthesizer,
-      preFilters,
       nodePostprocessors,
     );
   }
@@ -387,7 +386,7 @@ export type VectorIndexRetrieverOptions = {
   index: VectorStoreIndex;
   similarityTopK?: number | undefined;
   topK?: TopKMap | undefined;
-  filters?: MetadataFilters;
+  filters?: MetadataFilters | undefined;
 };
 
 export class VectorIndexRetriever extends BaseRetriever {

packages/llamaindex/src/next.ts

@@ -37,6 +37,7 @@ export default function withLlamaIndex(config: any) {
     webpackConfig.resolve.alias = {
       ...webpackConfig.resolve.alias,
       "@google-cloud/vertexai": false,
+      unpdf: false,
     };
     // Following lines will fix issues with onnxruntime-node when using pnpm
     // See: https://github.com/vercel/next.js/issues/43433

packages/llamaindex/src/vector-store/QdrantVectorStore.ts

@@ -1,16 +1,22 @@
 import type { BaseNode } from "@llamaindex/core/schema";
 import {
+  FilterCondition,
+  FilterOperator,
   VectorStoreBase,
   type IEmbedModel,
+  type MetadataFilters,
   type VectorStoreNoEmbedModel,
   type VectorStoreQuery,
   type VectorStoreQueryResult,
 } from "./types.js";
 
-import type { QdrantClientParams } from "@qdrant/js-client-rest";
+import type { QdrantClientParams, Schemas } from "@qdrant/js-client-rest";
 import { QdrantClient } from "@qdrant/js-client-rest";
 import { metadataDictToNode, nodeToMetadata } from "./utils.js";
 
+type QdrantFilter = Schemas["Filter"];
+type QdrantMustConditions = QdrantFilter["must"];
+
 type PointStruct = {
   id: string;
   payload: Record<string, string>;
@@ -272,7 +278,7 @@ export class QdrantVectorStore
   ): Promise<VectorStoreQueryResult> {
     const qdrantFilters = options?.qdrant_filters;
 
-    let queryFilters;
+    let queryFilters: QdrantFilter | undefined;
 
     if (!query.queryEmbedding) {
       throw new Error("No query embedding provided");
@@ -281,7 +287,7 @@ export class QdrantVectorStore
     if (qdrantFilters) {
       queryFilters = qdrantFilters;
     } else {
-      queryFilters = await this.buildQueryFilter(query);
+      queryFilters = buildQueryFilter(query);
     }
 
     const result = (await this.db.search(this.collectionName, {
@@ -292,58 +298,118 @@ export class QdrantVectorStore
 
     return this.parseToQueryResult(result);
   }
+}
 
-  /**
-   * Qdrant filter builder
-   * @param query The VectorStoreQuery to be used
-   */
-  private async buildQueryFilter(query: VectorStoreQuery) {
-    if (!query.docIds && !query.queryStr && !query.filters) {
-      return null;
-    }
+/**
+ * Qdrant filter builder
+ * @param query The VectorStoreQuery to be used
+ */
+function buildQueryFilter(query: VectorStoreQuery): QdrantFilter | undefined {
+  if (!query.docIds && !query.queryStr && !query.filters) return undefined;
 
-    const mustConditions = [];
+  const mustConditions: QdrantMustConditions = [];
+  if (query.docIds) {
+    mustConditions.push({
+      key: "doc_id",
+      match: { any: query.docIds },
+    });
+  }
 
-    if (query.docIds) {
-      mustConditions.push({
-        key: "doc_id",
-        match: {
-          any: query.docIds,
-        },
-      });
-    }
+  const metadataFilters = toQdrantMetadataFilters(query.filters);
+  if (metadataFilters) {
+    mustConditions.push(metadataFilters);
+  }
 
-    if (!query.filters) {
-      return {
-        must: mustConditions,
-      };
-    }
+  return { must: mustConditions };
+}
 
-    const metadataFilters = query.filters.filters;
+/**
+ * Converts metadata filters to Qdrant-compatible filters
+ * @param subFilters The metadata filters to be converted
+ * @returns A QdrantFilter object or undefined if no valid filters are provided
+ */
+function toQdrantMetadataFilters(
+  subFilters?: MetadataFilters,
+): QdrantFilter | undefined {
+  if (!subFilters?.filters.length) return undefined;
 
-    for (let i = 0; i < metadataFilters.length; i++) {
-      const filter = metadataFilters[i]!;
+  const conditions: QdrantMustConditions = [];
 
-      if (typeof filter.key === "number") {
-        mustConditions.push({
-          key: filter.key,
-          match: {
-            gt: filter.value,
-            lt: filter.value,
+  for (const subfilter of subFilters.filters) {
+    if (subfilter.operator === FilterOperator.EQ) {
+      if (typeof subfilter.value === "number") {
+        conditions.push({
+          key: subfilter.key,
+          range: {
+            gte: subfilter.value,
+            lte: subfilter.value,
           },
         });
       } else {
-        mustConditions.push({
-          key: filter.key,
-          match: {
-            value: filter.value,
-          },
+        conditions.push({
+          key: subfilter.key,
+          match: { value: subfilter.value },
         });
       }
+    } else if (subfilter.operator === FilterOperator.LT) {
+      conditions.push({
+        key: subfilter.key,
+        range: { lt: subfilter.value },
+      });
+    } else if (subfilter.operator === FilterOperator.GT) {
+      conditions.push({
+        key: subfilter.key,
+        range: { gt: subfilter.value },
+      });
+    } else if (subfilter.operator === FilterOperator.GTE) {
+      conditions.push({
+        key: subfilter.key,
+        range: { gte: subfilter.value },
+      });
+    } else if (subfilter.operator === FilterOperator.LTE) {
+      conditions.push({
+        key: subfilter.key,
+        range: { lte: subfilter.value },
+      });
+    } else if (subfilter.operator === FilterOperator.TEXT_MATCH) {
+      conditions.push({
+        key: subfilter.key,
+        match: { text: subfilter.value },
+      });
+    } else if (subfilter.operator === FilterOperator.NE) {
+      conditions.push({
+        key: subfilter.key,
+        match: { except: [subfilter.value] },
+      });
+    } else if (subfilter.operator === FilterOperator.IN) {
+      const values = Array.isArray(subfilter.value)
+        ? subfilter.value.map(String)
+        : String(subfilter.value).split(",");
+      conditions.push({
+        key: subfilter.key,
+        match: { any: values },
+      });
+    } else if (subfilter.operator === FilterOperator.NIN) {
+      const values = Array.isArray(subfilter.value)
+        ? subfilter.value.map(String)
+        : String(subfilter.value).split(",");
+      conditions.push({
+        key: subfilter.key,
+        match: { except: values },
+      });
+    } else if (subfilter.operator === FilterOperator.IS_EMPTY) {
+      conditions.push({
+        is_empty: { key: subfilter.key },
+      });
     }
-
-    return {
-      must: mustConditions,
-    };
   }
+
+  const filter: QdrantFilter = {};
+  if (subFilters.condition === FilterCondition.OR) {
+    filter.should = conditions;
+  } else {
+    filter.must = conditions;
+  }
+
+  return filter;
 }

packages/llamaindex/tests/indices/VectorStoreIndex.test.ts

@@ -27,7 +27,7 @@ describe("VectorStoreIndex", () => {
       runs: number = 2,
     ): Promise<Array<number>> => {
       const documents = [new Document({ text: "lorem ipsem", id_: "1" })];
-      const entries: number[] = [];
+      const entries = [];
       for (let i = 0; i < runs; i++) {
         await VectorStoreIndex.fromDocuments(documents, {
           serviceContext,
@@ -43,7 +43,7 @@ describe("VectorStoreIndex", () => {
 
   test("fromDocuments stores duplicates without a doc store strategy", async () => {
     const entries = await testStrategy(DocStoreStrategy.NONE);
-    expect(entries[0]).toBe(entries[1]);
+    expect(entries[0]! + 1).toBe(entries[1]);
   });
 
   test("fromDocuments ignores duplicates with upserts doc store strategy", async () => {

packages/llm/groq/CHANGELOG.md

@@ -1,5 +1,23 @@
 # @llamaindex/groq
 
+## 0.0.14
+
+### Patch Changes
+
+- @llamaindex/openai@0.1.15
+
+## 0.0.13
+
+### Patch Changes
+
+- @llamaindex/openai@0.1.14
+
+## 0.0.12
+
+### Patch Changes
+
+- @llamaindex/openai@0.1.13
+
 ## 0.0.11
 
 ### Patch Changes

packages/llm/groq/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@llamaindex/groq",
   "description": "Groq Adapter for LlamaIndex",
-  "version": "0.0.11",
+  "version": "0.0.14",
   "type": "module",
   "main": "./dist/index.cjs",
   "module": "./dist/index.js",

packages/llm/ollama/CHANGELOG.md

@@ -1,5 +1,27 @@
 # @llamaindex/ollama
 
+## 0.0.7
+
+### Patch Changes
+
+- Updated dependencies [5f67820]
+  - @llamaindex/core@0.2.12
+
+## 0.0.6
+
+### Patch Changes
+
+- Updated dependencies [ee697fb]
+  - @llamaindex/core@0.2.11
+
+## 0.0.5
+
+### Patch Changes
+
+- Updated dependencies [3489e7d]
+- Updated dependencies [468bda5]
+  - @llamaindex/core@0.2.10
+
 ## 0.0.4
 
 ### Patch Changes

packages/llm/ollama/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@llamaindex/ollama",
   "description": "Ollama Adapter for LlamaIndex",
-  "version": "0.0.4",
+  "version": "0.0.7",
   "type": "module",
   "main": "./dist/index.cjs",
   "module": "./dist/index.js",

packages/llm/openai/CHANGELOG.md

@@ -1,5 +1,27 @@
 # @llamaindex/openai
 
+## 0.1.15
+
+### Patch Changes
+
+- Updated dependencies [5f67820]
+  - @llamaindex/core@0.2.12
+
+## 0.1.14
+
+### Patch Changes
+
+- Updated dependencies [ee697fb]
+  - @llamaindex/core@0.2.11
+
+## 0.1.13
+
+### Patch Changes
+
+- Updated dependencies [3489e7d]
+- Updated dependencies [468bda5]
+  - @llamaindex/core@0.2.10
+
 ## 0.1.12
 
 ### Patch Changes

packages/llm/openai/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@llamaindex/openai",
   "description": "OpenAI Adapter for LlamaIndex",
-  "version": "0.1.12",
+  "version": "0.1.15",
   "type": "module",
   "main": "./dist/index.cjs",
   "module": "./dist/index.js",