docxLoader not taking a blob

[latifs]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

import { DocxLoader } from '@langchain/community/document_loaders/fs/docx';

async function loadFileFromUrl(fileUrl: string) {
  try {
    const response = await fetch(fileUrl);

    if (!response.ok) {
      throw new Error(`HTTP error! status: ${response.status}`);
    }

    return await response.blob();
  } catch (error) {
    console.error('Error loading file from URL:', error);
    throw error;
  }
}


const fileBlob = await loadFileFromUrl(pathToRemoteFile);
const loader = new DocxLoader(fileBlob);
const documents = await loader.load();

console.log('documents:', documents); // <-- it errors out here

Description

Hey Guys,
trying to use the docx file loader here
with a remote file since it can take a string or blob. But it doesn't seem happy with the blob.
Here is the Class documentation

async function loadFileFromUrl(fileUrl: string) {
  try {
    const response = await fetch(fileUrl);

    if (!response.ok) {
      throw new Error(`HTTP error! status: ${response.status}`);
    }

    return await response.blob();
  } catch (error) {
    console.error('Error loading file from URL:', error);
    throw error;
  }
}


const fileBlob = await loadFileFromUrl(pathToRemoteFile);
const loader = new DocxLoader(fileBlob);
const documents = await loader.load();

console.log('documents:', documents); // <-- it errors out here

I'm obviously doing something wrong.
Thanks.

System Info

none

[dosubot[bot]]

The DocxLoader class in your TypeScript code is not accepting a Blob directly because it extends the BufferLoader class, which expects a Buffer object. To resolve this, you need to convert the Blob to a Buffer before passing it to the DocxLoader.

Here's how you can modify your code to convert the Blob to a Buffer:

import { DocxLoader } from '@langchain/community/document_loaders/fs/docx';

async function loadFileFromUrl(fileUrl: string): Promise<Blob> {
  try {
    const response = await fetch(fileUrl);

    if (!response.ok) {
      throw new Error(`HTTP error! status: ${response.status}`);
    }

    return await response.blob();
  } catch (error) {
    console.error('Error loading file from URL:', error);
    throw error;
  }
}

async function blobToBuffer(blob: Blob): Promise<Buffer> {
  const arrayBuffer = await blob.arrayBuffer();
  return Buffer.from(arrayBuffer);
}

const fileBlob = await loadFileFromUrl(pathToRemoteFile);
const buffer = await blobToBuffer(fileBlob);
const loader = new DocxLoader(buffer);
const documents = await loader.load();

console.log('documents:', documents);

Additionally, ensure that you have the mammoth module installed, as it is used internally by the DocxLoader to extract text content from the DOCX file. You can install it using:

npm install mammoth

This should resolve the issue and allow you to load the remote DOCX file successfully ^[1][2].

To continue talking to Dosu, mention @dosu.

[latifs]

Thanks @dosu,
the only issue with that and I've already tired it actually is that I'm getting this error now:

Argument of type 'Buffer' is not assignable to parameter of type 'string | Blob'.
  Type 'Buffer' is missing the following properties from type 'Blob': size, type, arrayBuffer, stream, textts(2345)

[dosubot[bot]]

To resolve the error "Argument of type 'Buffer' is not assignable to parameter of type 'string | Blob'", you need to ensure that the argument passed to the DocxLoader constructor is either a string (representing a file path) or a Blob, as the constructor is defined to accept only these types.

Here's how you can correctly instantiate DocxLoader with a Blob:

import { DocxLoader } from '@langchain/community/document_loaders/fs/docx';

async function loadFileFromUrl(fileUrl: string): Promise<Blob> {
  try {
    const response = await fetch(fileUrl);

    if (!response.ok) {
      throw new Error(`HTTP error! status: ${response.status}`);
    }

    return await response.blob();
  } catch (error) {
    console.error('Error loading file from URL:', error);
    throw error;
  }
}

const fileBlob = await loadFileFromUrl(pathToRemoteFile);
const loader = new DocxLoader(fileBlob);
const documents = await loader.load();

console.log('documents:', documents);

Ensure that pathToRemoteFile is the URL of the remote DOCX file you want to load. This should resolve the type error and allow you to load the remote DOCX file successfully ^[1].

To continue talking to Dosu, mention @dosu.

[latifs]

@dosu,
getting a different error now and it is a gnarly one:

Error: Can't find end of central directory : is this a zip file ? If it is, see https://stuk.github.io/jszip/documentation/howto/read_zip.html
    at ZipEntries.readEndOfCentral (webpack-internal:///(rsc)/./node_modules/.pnpm/[email protected]/node_modules/jszip/lib/zipEntries.js:167:23)
    at ZipEntries.load (webpack-internal:///(rsc)/./node_modules/.pnpm/[email protected]/node_modules/jszip/lib/zipEntries.js:255:14)
    at eval (webpack-internal:///(rsc)/./node_modules/.pnpm/[email protected]/node_modules/jszip/lib/load.js:48:24)
From previous event:
    at Object.openZip (webpack-internal:///(rsc)/./node_modules/.pnpm/[email protected]/node_modules/mammoth/lib/unzip.js:14:25)
    at extractRawText (webpack-internal:///(rsc)/./node_modules/.pnpm/[email protected]/node_modules/mammoth/lib/index.js:82:18)
    at DocxLoader.parse (webpack-internal:///(rsc)/./node_modules/.pnpm/@[email protected]_@[email protected][email protected][email protected][email protected][email protected]/node_modules/@langchain/community/dist/document_loaders/fs/docx.js:31:28)
    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)
    ```

[dosubot[bot]]

The "Can't find end of central directory: is this a zip file?" error typically occurs when the file being loaded is either corrupted or not a valid DOCX file. To resolve this, ensure that the file being fetched is indeed a valid DOCX file.

Here's a revised version of your code to ensure proper loading of a remote DOCX file using DocxLoader:

import { DocxLoader } from '@langchain/community/document_loaders/fs/docx';
import axios from 'axios';

async function loadRemoteDocx(url: string) {
  try {
    // Fetch the remote DOCX file
    const response = await axios.get(url, { responseType: 'arraybuffer' });
    const buffer = Buffer.from(response.data);

    // Create an instance of DocxLoader with the buffer
    const loader = new DocxLoader(buffer);

    // Parse the document
    const documents = await loader.parse(buffer, { source: url });

    // Handle the loaded documents
    console.log(documents);
  } catch (error) {
    console.error("Error loading DOCX file:", error);
  }
}

// Example usage
const pathToRemoteFile = 'https://example.com/path/to/your/document.docx';
loadRemoteDocx(pathToRemoteFile);

Make sure the URL points to a valid DOCX file to avoid the "Can't find end of central directory" error. Additionally, ensure that the mammoth module is installed, as it is used internally by DocxLoader to extract text from the DOCX file:

npm install mammoth

This should help you load the remote DOCX file successfully ^[1][2].

To continue talking to Dosu, mention @dosu.

[rossanodr]

Hey, I'm getting the same error.
Did you manage to solve it?