Document: Explicitly check that pageContents is not an empty string.

langchain/src/document.ts

@@ -8,8 +8,9 @@ export interface DocumentInput<
 }
 
 /**
- * Interface for interacting with a document.
+ * Class for creating and interacting with a document.
  */
+
 export class Document<
   // eslint-disable-next-line @typescript-eslint/no-explicit-any
   Metadata extends Record<string, any> = Record<string, any>
@@ -19,10 +20,14 @@ export class Document<
 
   metadata: Metadata;
 
+  /**
+   * Creates a new Document instance.
+   *
+   * {DocumentInput<Metadata>} fields - Takes `pageContent` as a string
+   * and optional `metadata` as Metadata.
+   */
   constructor(fields: DocumentInput<Metadata>) {
-    this.pageContent = fields.pageContent
-      ? fields.pageContent.toString()
-      : this.pageContent;
+    this.pageContent = fields.pageContent?.toString() ?? "";
     this.metadata = fields.metadata ?? ({} as Metadata);
   }
 }

langchain/src/tests/document.test.ts

@@ -0,0 +1,14 @@
+import { Document } from "../document.js";
+
+describe("Document", () => {
+  test("Constructing a document with an empty sting", async () => {
+    const doc = new Document({ pageContent: "" });
+    expect(doc.pageContent).toEqual("");
+  });
+
+  test("Constructing a document with an undefined at runtime", async () => {
+    const pageContent = undefined as unknown as string;
+    const doc = new Document({ pageContent });
+    expect(doc.pageContent).toEqual("");
+  });
+});