oadf-pdf-converter/lib/index.js

A small utility tool to convert PDF files into a data structure which is used by OADF parsers

'use strict';

Object.defineProperty(exports, "__esModule", {
  value: true
});

var _pdfjsDist = require('pdfjs-dist');

var _pdfjsDist2 = _interopRequireDefault(_pdfjsDist);

var _Element = require('./Element');

var _Element2 = _interopRequireDefault(_Element);

function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }

exports.default = function (data) {
  return _pdfjsDist2.default.getDocument(data).then(function (doc) {
    var numPages = doc.numPages;
    var pagePromises = [];

    for (var i = 1; i <= numPages; i++) {
      var promise = doc.getPage(i).then(function (page) {
        return page.getTextContent().then(function (content) {
          var items = content.items;
          return _Element2.default.orderElements(items.map(function (item) {
            return new _Element2.default(item);
          }));
        });
      });
      pagePromises.push(promise);
    }

    return Promise.all(pagePromises);
  });
};