Update distillability modeling scripts to predict long articles

heuristics/distillable/extract_features.js

 return (function() {
   function hasOGArticle() {
     var elems = document.head.querySelectorAll(
         'meta[property="og:type"],meta[name="og:type"]');
     for (var i in elems) {
       if (elems[i].content && elems[i].content.toUpperCase() == 'ARTICLE') {
         return true;
       }
     }
     return false;
@@ skipping 55 matching lines @@
       if (textContentLength < cut) {
         continue;
       }
 
       score += Math.pow(textContentLength - cut, power);
     }
     return score;
   }
 
   var body = document.body;
-  return {
+  var features = {
      'opengraph': hasOGArticle(),
      'url': document.location.href,
      'title': document.title,
      'numElements': body.querySelectorAll('*').length,
      'numAnchors': body.querySelectorAll('a').length,
      'numForms': body.querySelectorAll('form').length,
      'numTextInput': body.querySelectorAll('input[type="text"]').length,
      'numPasswordInput': body.querySelectorAll('input[type="password"]').length,
      'numPPRE': body.querySelectorAll('p,pre').length,
      'innerText': body.innerText,
      'textContent': body.textContent,
      'innerHTML': body.innerHTML,
-     'mozScore': _mozScore(true, 0.5, 140, true, 1e100),
-     'mozScoreAllSqrt': _mozScore(true, 0.5, 0, true, 1e100),
-     'mozScoreAllLinear': _mozScore(true, 1, 0, true, 1e100),
-     'mozScoreFast': _mozScore(false, 0.5, 140, true, 1000),
-     'mozScoreFastAllSqrt': _mozScore(false, 0.5, 0, true, 1000),
-     'mozScoreFastAllLinear': _mozScore(false, 1, 0, true, 1000),
+     'mozScore': Math.min(6 * Math.sqrt(1000 - 140), _mozScore(false, 0.5, 140, true, 1000)),
+     'mozScoreAllSqrt': Math.min(6 * Math.sqrt(1000), _mozScore(false, 0.5, 0, true, 1000)),
+     'mozScoreAllLinear': Math.min(6 * 1000, _mozScore(false, 1, 0, true, 1000)),
      'visibleElements': countVisible(body.querySelectorAll('*')),
      'visiblePPRE': countVisible(body.querySelectorAll('p,pre')),
   }
+  return features;
 })()