Function getDocumentLoader

utils/langchain/documentLoader.ts:10–48

(fileType: string, filePath: string)

import {UnstructuredLoader} from "langchain/document_loaders/fs/unstructured";

export function getDocumentLoader(fileType: string, filePath: string): DocumentLoader {
    let loader;
    switch (fileType) {
        case "pdf":
            loader = new PDFLoader(filePath, {
                splitPages: false,
            });
            return loader;
        case "epub":
            loader = new EPubLoader(
                filePath, {
                    splitChapters: false,
                });
            return loader;
        case "docx":
            loader = new DocxLoader(
                filePath
            );
            return loader;
        case "txt":
            loader = new TextLoader(filePath);
            return loader;
        case "md":
            loader = new TextLoader(filePath);
            return loader;
        case "json":
            // JSONLoader is not implemented with split option
            loader = new TextLoader(filePath);
            return loader;
        case "csv":
            loader = new CSVLoader(filePath);
            return loader;
        case "zip":
            return getDirectoryLoader(filePath)
        default:
            loader = new UnstructuredLoader(filePath);
            return loader;
    }
}

export function getDirectoryLoader(path: string): DocumentLoader {
    const zipFilePath = path.split('.')[0];