Merge main into sweep/json-support_2

packages/core/src/JsonDocument.ts

@@ -0,0 +1,26 @@
+import { Document } from './Document';
+
+/**
+ * JsonDocument is a class that extends the Document class to handle JSON data.
+ */
+export class JsonDocument extends Document {
+  jsonData: Object;
+
+  /**
+   * Constructor for the JsonDocument class.
+   * @param {Object} jsonData - The JSON data for the document.
+   */
+  constructor(jsonData: Object) {
+    super();
+    this.jsonData = jsonData;
+  }
+
+  /**
+   * Overrides the getText method of the Document class to return the JSON data.
+   * @returns {string} The JSON data of the document.
+   */
+  getText(): string {
+    return JSON.stringify(this.jsonData);
+  }
+}
+

packages/core/src/NodeParser.ts

@@ -1,4 +1,5 @@
 import { Document, NodeRelationship, TextNode } from "./Node";
+import { JsonDocument } from "./JsonDocument";
 import { SentenceSplitter } from "./TextSplitter";
 import { DEFAULT_CHUNK_OVERLAP, DEFAULT_CHUNK_SIZE } from "./constants";
 
@@ -27,18 +28,25 @@ export function getTextSplitsFromDocument(
  * @returns An array of nodes.
  */
 export function getNodesFromDocument(
-  document: Document,
+  document: Document | JsonDocument,
   textSplitter: SentenceSplitter,
   includeMetadata: boolean = true,
   includePrevNextRel: boolean = true
 ) {
   let nodes: TextNode[] = [];
 
-  const textSplits = getTextSplitsFromDocument(document, textSplitter);
+  let textSplits;
+  if (document instanceof JsonDocument) {
+    // Parse the JSON data and generate text splits
+    const jsonData = JSON.parse(document.getText());
+    textSplits = Array.isArray(jsonData) ? jsonData : [jsonData];
+  } else {
+    textSplits = getTextSplitsFromDocument(document, textSplitter);
+  }
 
   textSplits.forEach((textSplit) => {
     const node = new TextNode({
-      text: textSplit,
+      text: typeof textSplit === 'object' ? JSON.stringify(textSplit) : textSplit,
       metadata: includeMetadata ? document.metadata : {},
     });
     node.relationships[NodeRelationship.SOURCE] = document.asRelatedNodeInfo();

packages/core/src/ResponseSynthesizer.ts

@@ -33,7 +33,7 @@ interface BaseResponseBuilder {
    */
   getResponse(
     query: string,
-    textChunks: string[],
+    textChunks: (string | Object)[],
     parentEvent?: Event,
     prevResponse?: string
   ): Promise<string>;
@@ -295,9 +295,15 @@ export class ResponseSynthesizer {
   }
 
   async synthesize(query: string, nodes: NodeWithScore[], parentEvent?: Event) {
-    let textChunks: string[] = nodes.map((node) =>
-      node.node.getContent(MetadataMode.NONE)
-    );
+    let textChunks: string[] = nodes.map((node) => {
+      if (node.node instanceof JsonDocument) {
+        // Parse the JSON data and generate text chunks
+        const jsonData = node.node.getContent(MetadataMode.NONE);
+        return typeof jsonData === 'string' ? JSON.parse(jsonData) : jsonData;
+      } else {
+        return node.node.getContent(MetadataMode.NONE);
+      }
+    });
     const response = await this.responseBuilder.getResponse(
       query,
       textChunks,

packages/core/src/indices/list/ListIndex.ts

@@ -114,7 +114,7 @@ export class ListIndex extends BaseIndex<IndexList> {
   }
 
   static async fromDocuments(
-    documents: Document[],
+    documents: (Document | JsonDocument)[],
     args: {
       storageContext?: StorageContext;
       serviceContext?: ServiceContext;
@@ -130,7 +130,17 @@ export class ListIndex extends BaseIndex<IndexList> {
       docStore.setDocumentHash(doc.id_, doc.hash);
     }
 
-    const nodes = serviceContext.nodeParser.getNodesFromDocuments(documents);
+    let nodes = [];
+    for (const doc of documents) {
+      if (doc instanceof JsonDocument) {
+        // Handle JsonDocument instances
+        const jsonData = JSON.parse(doc.getText());
+        nodes = nodes.concat(Array.isArray(jsonData) ? jsonData : [jsonData]);
+      } else {
+        nodes = nodes.concat(serviceContext.nodeParser.getNodesFromDocuments([doc]));
+      }
+    }
+
     const index = await ListIndex.init({
       nodes,
       storageContext,
@@ -138,6 +148,9 @@ export class ListIndex extends BaseIndex<IndexList> {
     });
     return index;
   }
+    });
+    return index;
+  }
 
   asRetriever(options?: { mode: ListRetrieverMode }): BaseRetriever {
     const { mode = ListRetrieverMode.DEFAULT } = options ?? {};