import path from "path";
import fs from "fs";
import { DocumentConverter, DocumentType } from "@mendable/firecrawl-rs";
describe("Document Converter tests", () => {
const samplesDir = path.join(process.cwd(), "samples");
const expectedHtmlBase = (documentText: string) =>
`DocumentHello!
${documentText} file to test the Firecrawl Document Converter.
Italic
Bold
Underlined
Strikethrough
| Header 1 | Header 2 |
| Value 1 | Value 2 |
`;
const sampleFiles = [
{ file: "sample.docx", type: DocumentType.Docx, name: "DOCX" },
{ file: "sample.odt", type: DocumentType.Odt, name: "ODT" },
{ file: "sample.rtf", type: DocumentType.Rtf, name: "RTF" },
];
const converter = new DocumentConverter();
describe.each(sampleFiles)(
"$name document conversion",
({ file, type, name }) => {
const filePath = path.join(samplesDir, file);
beforeAll(() => {
if (!fs.existsSync(filePath)) {
throw new Error(`Sample file ${filePath} does not exist`);
}
});
it.concurrent(
`should convert ${name} document and return expected HTML`,
async () => {
const fileBuffer = fs.readFileSync(filePath);
const html = await converter.convertBufferToHtml(
new Uint8Array(fileBuffer),
type,
);
expect(html).toBe(expectedHtmlBase(name));
},
10000,
);
},
);
describe("XLSX document conversion", () => {
const expectedHtml = `DocumentSheet1
| sample file | test |
| |
| Name | Price |
| iPhone | 1000 |
| iPad | 800 |
| Macbook | 1200 |
Sheet2
| other tab | |
| |
| Name | Price |
| ChatGPT | 20 |
| Claude | 17 |
| Perplexity | 20 |
`;
it("should convert XLSX document and return expected HTML", async () => {
const filePath = path.join(samplesDir, "sample.xlsx");
const fileBuffer = fs.readFileSync(filePath);
const html = await converter.convertBufferToHtml(
new Uint8Array(fileBuffer),
DocumentType.Xlsx,
);
expect(html).toBe(expectedHtml);
});
});
});