Add to docs, update arg

langchain-ai · Mar 28, 2023 · 4c71c14 · 4c71c14
1 parent 1b3771a
commit 4c71c14
Show file tree

Hide file tree

Showing 3 changed files with 20 additions and 5 deletions.
diff --git a/docs/docs/modules/indexes/document_loaders/examples/file_loaders/pdf.md b/docs/docs/modules/indexes/document_loaders/examples/file_loaders/pdf.md
@@ -4,15 +4,15 @@ hide_table_of_contents: true
 
 # PDF files
 
-This example goes over how to load data from PDF files. One document will be created for each PDF file.
+This example goes over how to load data from PDF files. By default, one document will be created for each page in the PDF file, you can change this behavior by setting the `splitPages` option to `false`.
 
 # Setup
 
 ```bash npm2yarn
-npm install pdf-parse
+npm install pdfjs-dist
 ```
 
-# Usage
+# Usage, one document per page
 
 ```typescript
 import { PDFLoader } from "langchain/document_loaders";
@@ -21,3 +21,15 @@ const loader = new PDFLoader("src/document_loaders/example_data/example.pdf");
 
 const docs = await loader.load();
 ```
+
+# Usage, one document per file
+
+```typescript
+import { PDFLoader } from "langchain/document_loaders";
+
+const loader = new PDFLoader("src/document_loaders/example_data/example.pdf", {
+  splitPages: false,
+});
+
+const docs = await loader.load();
+```
diff --git a/langchain/src/document_loaders/pdf.ts b/langchain/src/document_loaders/pdf.ts
@@ -3,8 +3,11 @@ import { Document } from "../document.js";
 import { BufferLoader } from "./buffer.js";
 
 export class PDFLoader extends BufferLoader {
-  constructor(filePathOrBlob: string | Blob, public splitPages = true) {
+  private splitPages: boolean;
+
+  constructor(filePathOrBlob: string | Blob, { splitPages = true } = {}) {
     super(filePathOrBlob);
+    this.splitPages = splitPages;
   }
 
   public async parse(

diff --git a/langchain/src/document_loaders/tests/pdf.test.ts b/langchain/src/document_loaders/tests/pdf.test.ts
@@ -20,7 +20,7 @@ test("Test PDF loader from file to single document", async () => {
     path.dirname(url.fileURLToPath(import.meta.url)),
     "./example_data/1706.03762.pdf"
   );
-  const loader = new PDFLoader(filePath, false);
+  const loader = new PDFLoader(filePath, { splitPages: false });
   const docs = await loader.load();
 
   expect(docs.length).toBe(1);