Function read_pdf

src/modules/loadFiles.ts:9–35

(file: File)

const pdfjsLib = pdfjsLibRaw as typeof pdfjsLibModule;

async function read_pdf(file: File) {
  const typedArray = await file.arrayBuffer();

  // Load the PDF file.
  //const loadingTask = PDFJS.getDocument(pdfData);
  //const pdf = await pdfjsLib.getDocument({ data: typedArray }).promise;
  //const pdf = await pdfjsLib.getDocument({ data: typedArray }).promise;
  const loadingTask = pdfjsLib.getDocument({ data: typedArray });
  const pdf = await loadingTask.promise;

  let textContent = '';

  for (let i = 1; i <= pdf.numPages; i++) {
    // Fetch the page
    const page = await pdf.getPage(i);

    // Fetch the text content
    const text = await page.getTextContent();

    // Concatenate the text
    textContent += text.items
      .map((item) => ('str' in item ? item.str : ''))
      .join('\n');
  }

  return textContent;
}

async function read_docx(file: File) {
  const arrayBuffer = await file.arrayBuffer();