Simple self-contained HTTP sanitizer

sanitize.js

var sanitize = (function() {
  var re = /(.*?)<(\/?)\s*([A-Za-z]+)[^>]*>/mg;

  var blockElements = ['br', 'div', 'p', 'blockquote'];
  var inlineBlockElements = ['img'];

  var sanitize = function(input) {
    input = input.replace(/\s+/g, ' ');
    var out = '';
    var index = 0;
    var match;
    while (null !== (match = re.exec(input))) {
      var before = match[1],
          close = match[2],
          tag = match[3].toLowerCase();
      out += before;
      if (!close) {
        if (blockElements.indexOf(tag) >= 0) {
          out += '\n';
        } else if (inlineBlockElements.indexOf(tag) >= 0) {
          out += ' ';
        }
      }
      index = re.lastIndex;
    }
    out += input.substr(index);
    return out;
  };

  return sanitize;
})();