get_ids.js

function test() {
  var ids = getIds();
  var batches = createBatches(ids);

  for (var id in batches) {
    if (batches.hasOwnProperty(id)) {
      var pages = batches[id];
      var start = pages[0];
      var end   = pages[pages.length -1];

      savePDFSplit(id, start, end);
    }
  }
  
  app.beep(3);
  console.println("Done!");
}

function savePDFSplit(id, start, end) {
  var filename = id + ".pdf" 
  this.extractPages(start, end).saveAs(filename);
}

// list of ids to batches -> {ids: [<pageNums>], ...}
function createBatches(ids) {
  return ids.reduce(function (acc, x, i) {
    if (acc[x]) {
      acc[x].push(i)
    } else {
      acc[x] = [i]
    }
  
    return acc;
  }, {});
}

function getIds() {
  var ids = [];
  var numPages = this.numPages;

  for (var i = 0; i < numPages; i++) {
    var txt = pageText(i);
    var id = parseId(txt);

    app.beep(i % 2 === 0 ? 1 : 2);
    ids.push(id);
  }

  return ids;
}

function parseId(txt) {
  // Tests for LabCorp style ids of the form:
  // xxxx, ID
  // Assumes ids will allways start with a digit and be ended with ", ID"
  var idTest1 = /\d.*(?=, ID)/.exec(txt);
  
  if (idTest1 !== null) {
    return idTest1;
  }
  
  // Tests for LABS style ids of the form:
  // Additional ID 1
  // ...
  // xx-xx
  // Assumes ids will be within 100 characters of "Additional ID 1", and will contain at least one dash.
  var subsection = /Additional ID 1[\s\S]{100}/m.exec(txt);
  var idTest2 = /\w*\-\S*/.exec(subsection);

  if (idTest2 !== null) {
    return idTest2;
  }

  throw new Error("Unable to find id");
}

function pageText(n) {
  var numWords = this.getPageNumWords(n)

  var txt = "";
  for (var i = 0; i < numWords; i++) {
    txt += this.getPageNthWord(n, i, false);
  }

  return txt;
}