UltraEdit - Compare two documents

// An UltraEdit script which compares two documents
// Much simpler than UltraCompare, it only extracts a RegExp from each line (default: \w+)
// Then compares the two sets of extracted strings
// Output:
// - Extracted strings of first doc
// - Extracted strings of second doc
// - Strings contained in the first but not in the second doc
// - Strings contained in the second but not in the first doc
// - Strings contained in both docs
start();

function start() {
  if (UltraEdit.document.length >= 2) {
    var param = getParameters();
    if (param.first && param.second) {
      compareDocuments(param.first, param.second, param.norm);
      }
    }
  else {
    UltraEdit.outputWindow.write( "Es müssen mindestens zwei Textfenster offen sein" );
    }
  }

function compareDocuments(first,second,norm) {
  var set1 = getLines( first, norm);
  output( "Erstes Dokument", getKeys( set1 ) );
  var set2 = getLines( second, norm);
  output( "Zweites Dokument", getKeys( set2 ) );
  compareSets( set1, set2 );
  }

function getParameters() {

  var msg="[0 1 \\w+] ",i,fileName, param;

  for (i=0;i<UltraEdit.document.length;i++) {
    if (UltraEdit.document[i].path.match(/\\([^\\]*)$/)) {
      fileName = RegExp.$1;
      }
    else {
      fileName = UltraEdit.document[i].path;
      }
    msg += i + ":" + fileName + ",";
    }

  var inp = UltraEdit.getString( msg, 1 );
  if (inp.match(/(\d+)\D+(\d+)(?:\s+(.*))?/)) {
    var re = RegExp.$3 || "\\w+";
    re = "(" + re + ")";
    var ret = {};
    ret.norm = new RegExp( re );
    ret.first = UltraEdit.document[1*RegExp.$1];
    ret.second = UltraEdit.document[1*RegExp.$2];
    return ret;
    }
  else
    return { norm:/(\w+)/, first:UltraEdit.document[0], second:UltraEdit.document[1]};

  }

function getLines(doc,re) {
  doc.selectAll();
  var selection = doc.selection
                     .replace( /^\xEF\xBB\xBF/,"")  // Unicode-BOM
                     .replace( /\x00/g,"");         // Unicode-Bytes enfernen
  var result = {};
  var lineEnd   = selection.match( /[\r\n]{1,2}/ ) || "\r\n" ;
  var lines = selection.split( lineEnd );
  for (var i=0; i<lines.length;i++) {
    if (lines[i].match(re)) {
      if (result[RegExp.$1]) result[RegExp.$1]++;
      else result[RegExp.$1] = 1;
      }
    }
  return result;
  }

function compareSets( set1, set2 ) {

  var result;

  result = getDifference(set1, set2);
  output( "Nur im ersten Dok, nicht im zweiten", result);

  result = getDifference(set2, set1);
  output( "Nur im zweiten Dok, nicht im ersten", result);

  result = getIntersection(set1, set2);
  output( "Gemeinsame Felder", result);

  }

function output(title, result) {
  beginSection(title);
  showArray(result);
  endSection();
  }

function beginSection(title) {
  UltraEdit.outputWindow.write("");
  UltraEdit.outputWindow.write("* " + title);
  }

function endSection() {
  UltraEdit.outputWindow.write("* ---");
  UltraEdit.outputWindow.write("");
  }

function showArray(lines) {
  for (var i=0;i<lines.length;i++)
    UltraEdit.outputWindow.write( lines[i] );
  }

function getDifference(set1, set2) {
  var result = [];
  for (var x in set1) {
    if (! (x in set2) ) result.push(x);
    }
  return result;
  }

function getIntersection(set1, set2) {
  var result = [];
  for (var x in set1) {
    if (x in set2) result.push(x);
    }
  return result;
  }

function getKeys( obj ) {
  var result = [];
  for (var x in obj) result.push( x );
  return result;
  }