The Tor Browser / file revision

 /*

  * Copyright (c) 2010 Arc90 Inc

  *

  * Licensed under the Apache License, Version 2.0 (the "License");

  * you may not use this file except in compliance with the License.

  * You may obtain a copy of the License at

  *

  *     http://www.apache.org/licenses/LICENSE-2.0

  *

  * Unless required by applicable law or agreed to in writing, software

  * distributed under the License is distributed on an "AS IS" BASIS,

  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.

  * See the License for the specific language governing permissions and

  * limitations under the License.

  */

 /*

  * This code is heavily based on Arc90's readability.js (1.7.1) script

  * available at: http://code.google.com/p/arc90labs-readability

  */

 var Readability = function(uri, doc) {

   const ENABLE_LOGGING = false;

   this._uri = uri;

   this._doc = doc;

   this._biggestFrame = false;

   this._articleByline = null;

   this._articleDir = null;

   // Start with all flags set

   this._flags = this.FLAG_STRIP_UNLIKELYS |

                 this.FLAG_WEIGHT_CLASSES |

                 this.FLAG_CLEAN_CONDITIONALLY;

   // The list of pages we've parsed in this call of readability,

   // for autopaging. As a key store for easier searching.

   this._parsedPages = {};

   // A list of the ETag headers of pages we've parsed, in case they happen to match,

   // we'll know it's a duplicate.

   this._pageETags = {};

   // Make an AJAX request for each page and append it to the document.

   this._curPageNum = 1;

   // Control whether log messages are sent to the console

   if (ENABLE_LOGGING) {

     this.log = function (msg) {

       dump("Reader: (Readability) " + msg);

     };

   } else {

     this.log = function () {};

   }

 }

 Readability.prototype = {

   FLAG_STRIP_UNLIKELYS: 0x1,

   FLAG_WEIGHT_CLASSES: 0x2,

   FLAG_CLEAN_CONDITIONALLY: 0x4,

   // The number of top candidates to consider when analysing how

   // tight the competition is among candidates.

   N_TOP_CANDIDATES: 5,

   // The maximum number of pages to loop through before we call

   // it quits and just show a link.

   MAX_PAGES: 5,

   // All of the regular expressions in use within readability.

   // Defined up here so we don't instantiate them repeatedly in loops.

   REGEXPS: {

     unlikelyCandidates: /combx|comment|community|disqus|extra|foot|header|menu|remark|rss|shoutbox|sidebar|sponsor|ad-break|agegate|pagination|pager|popup|tweet|twitter/i,

     okMaybeItsACandidate: /and|article|body|column|main|shadow/i,

     positive: /article|body|content|entry|hentry|main|page|pagination|post|text|blog|story/i,

     negative: /hidden|combx|comment|com-|contact|foot|footer|footnote|masthead|media|meta|outbrain|promo|related|scroll|shoutbox|sidebar|sponsor|shopping|tags|tool|widget/i,

     extraneous: /print|archive|comment|discuss|e[\-]?mail|share|reply|all|login|sign|single|utility/i,

     byline: /byline|author|dateline|writtenby/i,

     replaceFonts: /<(\/?)font[^>]*>/gi,

     trim: /^\s+|\s+$/g,

     normalize: /\s{2,}/g,

     videos: /http:\/\/(www\.)?(youtube|vimeo)\.com/i,

     nextLink: /(next|weiter|continue|>([^\|]|$)|»([^\|]|$))/i,

     prevLink: /(prev|earl|old|new|<|«)/i,

     whitespace: /^\s*$/

   },

   DIV_TO_P_ELEMS: [ "A", "BLOCKQUOTE", "DL", "DIV", "IMG", "OL", "P", "PRE", "TABLE", "UL", "SELECT" ],

   /**

    * Run any post-process modifications to article content as necessary.

    *

    * @param Element

    * @return void

   **/

   _postProcessContent: function(articleContent) {

     // Readability cannot open relative uris so we convert them to absolute uris. 

     this._fixRelativeUris(articleContent);

   },

   /**

    * Converts each <a> and <img> uri in the given element to an absolute URI.

    *

    * @param Element

    * @return void

    */

   _fixRelativeUris: function(articleContent) {

     let scheme = this._uri.scheme;

     let prePath = this._uri.prePath;

     let pathBase = this._uri.pathBase;

     function toAbsoluteURI(uri) {

       // If this is already an absolute URI, return it.

       if (/^[a-zA-Z][a-zA-Z0-9\+\-\.]*:/.test(uri))

         return uri;

       // Scheme-rooted relative URI.

       if (uri.substr(0, 2) == "//")

         return scheme + "://" + uri.substr(2);

       // Prepath-rooted relative URI.

       if (uri[0] == "/")

         return prePath + uri;

       // Standard relative URI; add entire path. pathBase already includes a

       // trailing "/".

       return pathBase + uri;

     }

     function convertRelativeURIs(tagName, propName) {

       let elems = articleContent.getElementsByTagName(tagName);

       for (let i = elems.length; --i >= 0;) {

         let elem = elems[i];

         let relativeURI = elem.getAttribute(propName);

         if (relativeURI != null)

           elems[i].setAttribute(propName, toAbsoluteURI(relativeURI));

       }

     }

      // Fix links.

     convertRelativeURIs("a", "href");

      // Fix images.

     convertRelativeURIs("img", "src");

   },

   /**

    * Get the article title as an H1.

    *

    * @return void

    **/

   _getArticleTitle: function() {

     let doc = this._doc;

     let curTitle = "";

     let origTitle = "";

     try {

       curTitle = origTitle = doc.title;

       // If they had an element with id "title" in their HTML

       if (typeof curTitle !== "string")

         curTitle = origTitle = this._getInnerText(doc.getElementsByTagName('title')[0]);

     } catch(e) {}

     if (curTitle.match(/ [\|\-] /)) {

       curTitle = origTitle.replace(/(.*)[\|\-] .*/gi,'$1');

       if (curTitle.split(' ').length < 3)

         curTitle = origTitle.replace(/[^\|\-]*[\|\-](.*)/gi,'$1');

     } else if (curTitle.indexOf(': ') !== -1) {

       curTitle = origTitle.replace(/.*:(.*)/gi, '$1');

       if (curTitle.split(' ').length < 3)

         curTitle = origTitle.replace(/[^:]*[:](.*)/gi,'$1');

     } else if (curTitle.length > 150 || curTitle.length < 15) {

       let hOnes = doc.getElementsByTagName('h1');

       if (hOnes.length === 1)

         curTitle = this._getInnerText(hOnes[0]);

     }

     curTitle = curTitle.replace(this.REGEXPS.trim, "");

     if (curTitle.split(' ').length <= 4)

       curTitle = origTitle;

     return curTitle;

   },

   /**

    * Prepare the HTML document for readability to scrape it.

    * This includes things like stripping javascript, CSS, and handling terrible markup.

    *

    * @return void

    **/

   _prepDocument: function() {

     let doc = this._doc;

     // In some cases a body element can't be found (if the HTML is

     // totally hosed for example) so we create a new body node and

     // append it to the document.

     if (doc.body === null) {

       let body = doc.createElement("body");

       try {

         doc.body = body;

       } catch(e) {

         doc.documentElement.appendChild(body);

         this.log(e);

       }

     }

     // Remove all style tags in head

     let styleTags = doc.getElementsByTagName("style");

     for (let st = 0; st < styleTags.length; st += 1) {

       styleTags[st].textContent = "";

     }

     this._replaceBrs(doc.body);

     let fonts = doc.getElementsByTagName("FONT");

     for (let i = fonts.length; --i >=0;) {

       this._setNodeTag(fonts[i], "SPAN");

     }

   },

   /**

    * Finds the next element, starting from the given node, and ignoring

    * whitespace in between. If the given node is an element, the same node is

    * returned.

    */

   _nextElement: function (node) {

     let next = node;

     while (next

         && (next.nodeType != Node.ELEMENT_NODE)

         && this.REGEXPS.whitespace.test(next.textContent)) {

       next = next.nextSibling;

     }

     return next;

   },

   /**

    * Replaces 2 or more successive <br> elements with a single <p>.

    * Whitespace between <br> elements are ignored. For example:

    *   <div>foo<br>bar<br> <br><br>abc</div>

    * will become:

    *   <div>foo<br>bar<p>abc</p></div>

    */

   _replaceBrs: function (elem) {

     let brs = elem.getElementsByTagName("br");

     for (let i = 0; i < brs.length; i++) {

       let br = brs[i];

       let next = br.nextSibling;

       // Whether 2 or more <br> elements have been found and replaced with a

       // <p> block.

       let replaced = false;

       // If we find a <br> chain, remove the <br>s until we hit another element

       // or non-whitespace. This leaves behind the first <br> in the chain

       // (which will be replaced with a <p> later).

       while ((next = this._nextElement(next)) && (next.tagName == "BR")) {

         replaced = true;

         let sibling = next.nextSibling;

         next.parentNode.removeChild(next);

         next = sibling;

       }

       // If we removed a <br> chain, replace the remaining <br> with a <p>. Add

       // all sibling nodes as children of the <p> until we hit another <br>

       // chain.

       if (replaced) {

         let p = this._doc.createElement("p");

         br.parentNode.replaceChild(p, br);

         next = p.nextSibling;

         while (next) {

           // If we've hit another <br><br>, we're done adding children to this <p>.

           if (next.tagName == "BR") {

             let nextElem = this._nextElement(next);

             if (nextElem && nextElem.tagName == "BR")

               break;

           }

           // Otherwise, make this node a child of the new <p>.

           let sibling = next.nextSibling;

           p.appendChild(next);

           next = sibling;

         }

       }

     }

   },

   _setNodeTag: function (node, tag) {

     node.localName = tag.toLowerCase();

     node.tagName = tag.toUpperCase();

   },

   /**

    * Prepare the article node for display. Clean out any inline styles,

    * iframes, forms, strip extraneous <p> tags, etc.

    *

    * @param Element

    * @return void

    **/

   _prepArticle: function(articleContent) {

     this._cleanStyles(articleContent);

     // Clean out junk from the article content

     this._cleanConditionally(articleContent, "form");

     this._clean(articleContent, "object");

     this._clean(articleContent, "h1");

     // If there is only one h2, they are probably using it as a header

     // and not a subheader, so remove it since we already have a header.

     if (articleContent.getElementsByTagName('h2').length === 1)

       this._clean(articleContent, "h2");

     this._clean(articleContent, "iframe");

     this._cleanHeaders(articleContent);

     // Do these last as the previous stuff may have removed junk

     // that will affect these

     this._cleanConditionally(articleContent, "table");

     this._cleanConditionally(articleContent, "ul");

     this._cleanConditionally(articleContent, "div");

     // Remove extra paragraphs

     let articleParagraphs = articleContent.getElementsByTagName('p');

     for (let i = articleParagraphs.length - 1; i >= 0; i -= 1) {

       let imgCount = articleParagraphs[i].getElementsByTagName('img').length;

       let embedCount = articleParagraphs[i].getElementsByTagName('embed').length;

       let objectCount = articleParagraphs[i].getElementsByTagName('object').length;

       if (imgCount === 0 &&

         embedCount === 0 &&

         objectCount === 0 &&

         this._getInnerText(articleParagraphs[i], false) === '')

         articleParagraphs[i].parentNode.removeChild(articleParagraphs[i]);

     }

     let brs = articleContent.getElementsByTagName("BR");

     for (let i = brs.length; --i >= 0;) {

       let br = brs[i];

       let next = this._nextElement(br.nextSibling);

       if (next && next.tagName == "P")

         br.parentNode.removeChild(br);

     }

   },

   /**

    * Initialize a node with the readability object. Also checks the

    * className/id for special names to add to its score.

    *

    * @param Element

    * @return void

   **/

   _initializeNode: function(node) {

     node.readability = {"contentScore": 0};

     switch(node.tagName) {

       case 'DIV':

         node.readability.contentScore += 5;

         break;

       case 'PRE':

       case 'TD':

       case 'BLOCKQUOTE':

         node.readability.contentScore += 3;

         break;

       case 'ADDRESS':

       case 'OL':

       case 'UL':

       case 'DL':

       case 'DD':

       case 'DT':

       case 'LI':

       case 'FORM':

         node.readability.contentScore -= 3;

         break;

       case 'H1':

       case 'H2':

       case 'H3':

       case 'H4':

       case 'H5':

       case 'H6':

       case 'TH':

         node.readability.contentScore -= 5;

         break;

     }

     node.readability.contentScore += this._getClassWeight(node);

   },

   /***

    * grabArticle - Using a variety of metrics (content score, classname, element types), find the content that is

    *         most likely to be the stuff a user wants to read. Then return it wrapped up in a div.

    *

    * @param page a document to run upon. Needs to be a full document, complete with body.

    * @return Element

   **/

   _grabArticle: function (page) {

     let doc = this._doc;

     let isPaging = (page !== null ? true: false);

     page = page ? page : this._doc.body;

     let pageCacheHtml = page.innerHTML;

     // Check if any "dir" is set on the toplevel document element

     this._articleDir = doc.documentElement.getAttribute("dir");

     while (true) {

       let stripUnlikelyCandidates = this._flagIsActive(this.FLAG_STRIP_UNLIKELYS);

       let allElements = page.getElementsByTagName('*');

       // First, node prepping. Trash nodes that look cruddy (like ones with the

       // class name "comment", etc), and turn divs into P tags where they have been

       // used inappropriately (as in, where they contain no other block level elements.)

       //

       // Note: Assignment from index for performance. See http://www.peachpit.com/articles/article.aspx?p=31567&seqNum=5

       // TODO: Shouldn't this be a reverse traversal?

       let node = null;

       let nodesToScore = [];

       // Let each node know its index in the allElements array.

       for (let i = allElements.length; --i >= 0;) {

         allElements[i]._index = i;

       }

       /**

        * JSDOMParser returns static node lists, not live ones. When we remove

        * an element from the document, we need to manually remove it - and all

        * of its children - from the allElements array.

        */

       function purgeNode(node) {

         for (let i = node.childNodes.length; --i >= 0;) {

           purgeNode(node.childNodes[i]);

         }

         if (node._index !== undefined && allElements[node._index] == node)

           delete allElements[node._index];

       }

       for (let nodeIndex = 0; nodeIndex < allElements.length; nodeIndex++) {

         if (!(node = allElements[nodeIndex]))

           continue;

         let matchString = node.className + node.id;

         if (matchString.search(this.REGEXPS.byline) !== -1 && !this._articleByline) {

           this._articleByline = node.textContent;

           node.parentNode.removeChild(node);

           purgeNode(node);

           continue;

         }

         // Remove unlikely candidates

         if (stripUnlikelyCandidates) {

           if (matchString.search(this.REGEXPS.unlikelyCandidates) !== -1 &&

             matchString.search(this.REGEXPS.okMaybeItsACandidate) === -1 &&

             node.tagName !== "BODY") {

             this.log("Removing unlikely candidate - " + matchString);

             node.parentNode.removeChild(node);

             purgeNode(node);

             continue;

           }

         }

         if (node.tagName === "P" || node.tagName === "TD" || node.tagName === "PRE")

           nodesToScore[nodesToScore.length] = node;

         // Turn all divs that don't have children block level elements into p's

         if (node.tagName === "DIV") {

           // Sites like http://mobile.slate.com encloses each paragraph with a DIV

           // element. DIVs with only a P element inside and no text content can be

           // safely converted into plain P elements to avoid confusing the scoring

           // algorithm with DIVs with are, in practice, paragraphs.

           let pIndex = this._getSinglePIndexInsideDiv(node);

           if (pIndex >= 0 || !this._hasChildBlockElement(node)) {

             if (pIndex >= 0) {

               let newNode = node.childNodes[pIndex];

               node.parentNode.replaceChild(newNode, node);

               purgeNode(node);

             } else {

               this._setNodeTag(node, "P");

               nodesToScore[nodesToScore.length] = node;

             }

           } else {

             // EXPERIMENTAL

             for (let i = 0, il = node.childNodes.length; i < il; i += 1) {

               let childNode = node.childNodes[i];

               if (!childNode)

                 continue;

               if (childNode.nodeType === 3) { // Node.TEXT_NODE

                 let p = doc.createElement('p');

                 p.textContent = childNode.textContent;

                 p.style.display = 'inline';

                 p.className = 'readability-styled';

                 childNode.parentNode.replaceChild(p, childNode);

               }

             }

           }

         }

       }

       /**

        * Loop through all paragraphs, and assign a score to them based on how content-y they look.

        * Then add their score to their parent node.

        *

        * A score is determined by things like number of commas, class names, etc. Maybe eventually link density.

       **/

       let candidates = [];

       for (let pt = 0; pt < nodesToScore.length; pt += 1) {

         let parentNode = nodesToScore[pt].parentNode;

         let grandParentNode = parentNode ? parentNode.parentNode : null;

         let innerText = this._getInnerText(nodesToScore[pt]);

         if (!parentNode || typeof(parentNode.tagName) === 'undefined')

           continue;

         // If this paragraph is less than 25 characters, don't even count it.

         if (innerText.length < 25)

           continue;

         // Initialize readability data for the parent.

         if (typeof parentNode.readability === 'undefined') {

           this._initializeNode(parentNode);

           candidates.push(parentNode);

         }

         // Initialize readability data for the grandparent.

         if (grandParentNode &&

           typeof(grandParentNode.readability) === 'undefined' &&

           typeof(grandParentNode.tagName) !== 'undefined') {

           this._initializeNode(grandParentNode);

           candidates.push(grandParentNode);

         }

         let contentScore = 0;

         // Add a point for the paragraph itself as a base.

         contentScore += 1;

         // Add points for any commas within this paragraph.

         contentScore += innerText.split(',').length;

         // For every 100 characters in this paragraph, add another point. Up to 3 points.

         contentScore += Math.min(Math.floor(innerText.length / 100), 3);

         // Add the score to the parent. The grandparent gets half.

         parentNode.readability.contentScore += contentScore;

         if (grandParentNode)

           grandParentNode.readability.contentScore += contentScore / 2;

       }

       // After we've calculated scores, loop through all of the possible

       // candidate nodes we found and find the one with the highest score.

       let topCandidates = [];

       for (let c = 0, cl = candidates.length; c < cl; c += 1) {

         let candidate = candidates[c];

         // Scale the final candidates score based on link density. Good content

         // should have a relatively small link density (5% or less) and be mostly

         // unaffected by this operation.

         let candidateScore = candidate.readability.contentScore * (1 - this._getLinkDensity(candidate));

         candidate.readability.contentScore = candidateScore;

         this.log('Candidate: ' + candidate + " (" + candidate.className + ":" +

           candidate.id + ") with score " + candidateScore);

         for (let t = 0; t < this.N_TOP_CANDIDATES; t++) {

           let aTopCandidate = topCandidates[t];

           if (!aTopCandidate || candidateScore > aTopCandidate.readability.contentScore) {

             topCandidates.splice(t, 0, candidate);

             if (topCandidates.length > this.N_TOP_CANDIDATES)

               topCandidates.pop();

             break;

           }

         }

       }

       let topCandidate = topCandidates[0] || null;

       let lastTopCandidate = (topCandidates.length > 3 ? topCandidates[topCandidates.length - 1] : null);

       // If we still have no top candidate, just use the body as a last resort.

       // We also have to copy the body node so it is something we can modify.

       if (topCandidate === null || topCandidate.tagName === "BODY") {

         // Move all of the page's children into topCandidate

         topCandidate = doc.createElement("DIV");

         let children = page.childNodes;

         for (let i = 0; i < children.length; ++i) {

           topCandidate.appendChild(children[i]);

         }

         page.appendChild(topCandidate);

         this._initializeNode(topCandidate);

       }

       // Now that we have the top candidate, look through its siblings for content

       // that might also be related. Things like preambles, content split by ads

       // that we removed, etc.

       let articleContent = doc.createElement("DIV");

       if (isPaging)

         articleContent.id = "readability-content";

       let siblingScoreThreshold = Math.max(10, topCandidate.readability.contentScore * 0.2);

       let siblingNodes = topCandidate.parentNode.childNodes;

       for (let s = 0, sl = siblingNodes.length; s < sl; s += 1) {

         let siblingNode = siblingNodes[s];

         let append = false;

         this.log("Looking at sibling node: " + siblingNode + " (" + siblingNode.className + ":" + siblingNode.id + ")" + ((typeof siblingNode.readability !== 'undefined') ? (" with score " + siblingNode.readability.contentScore) : ''));

         this.log("Sibling has score " + (siblingNode.readability ? siblingNode.readability.contentScore : 'Unknown'));

         if (siblingNode === topCandidate)

           append = true;

         let contentBonus = 0;

         // Give a bonus if sibling nodes and top candidates have the example same classname

         if (siblingNode.className === topCandidate.className && topCandidate.className !== "")

           contentBonus += topCandidate.readability.contentScore * 0.2;

         if (typeof siblingNode.readability !== 'undefined' &&

           (siblingNode.readability.contentScore+contentBonus) >= siblingScoreThreshold)

           append = true;

         if (siblingNode.nodeName === "P") {

           let linkDensity = this._getLinkDensity(siblingNode);

           let nodeContent = this._getInnerText(siblingNode);

           let nodeLength = nodeContent.length;

           if (nodeLength > 80 && linkDensity < 0.25) {

             append = true;

           } else if (nodeLength < 80 && linkDensity === 0 && nodeContent.search(/\.( |$)/) !== -1) {

             append = true;

           }

         }

         if (append) {

           this.log("Appending node: " + siblingNode);

           // siblingNodes is a reference to the childNodes array, and

           // siblingNode is removed from the array when we call appendChild()

           // below. As a result, we must revisit this index since the nodes

           // have been shifted.

           s -= 1;

           sl -= 1;

           if (siblingNode.nodeName !== "DIV" && siblingNode.nodeName !== "P") {

             // We have a node that isn't a common block level element, like a form or td tag.

             // Turn it into a div so it doesn't get filtered out later by accident. */

             this.log("Altering siblingNode of " + siblingNode.nodeName + ' to div.');

             this._setNodeTag(siblingNode, "DIV");

           }

           // To ensure a node does not interfere with readability styles,

           // remove its classnames.

           siblingNode.className = "";

           // Append sibling and subtract from our list because it removes

           // the node when you append to another node.

           articleContent.appendChild(siblingNode);

         }

       }

       // So we have all of the content that we need. Now we clean it up for presentation.

       this._prepArticle(articleContent);

       if (this._curPageNum === 1) {

         let div = doc.createElement("DIV");

         div.id = "readability-page-1";

         div.className = "page";

         let children = articleContent.childNodes;

         for (let i = 0; i < children.length; ++i) {

           div.appendChild(children[i]);

         }

         articleContent.appendChild(div);

       }

       // Now that we've gone through the full algorithm, check to see if

       // we got any meaningful content. If we didn't, we may need to re-run

       // grabArticle with different flags set. This gives us a higher likelihood of

       // finding the content, and the sieve approach gives us a higher likelihood of

       // finding the -right- content.

       if (this._getInnerText(articleContent, true).length < 500) {

         page.innerHTML = pageCacheHtml;

         if (this._flagIsActive(this.FLAG_STRIP_UNLIKELYS)) {

           this._removeFlag(this.FLAG_STRIP_UNLIKELYS);

         } else if (this._flagIsActive(this.FLAG_WEIGHT_CLASSES)) {

           this._removeFlag(this.FLAG_WEIGHT_CLASSES);

         } else if (this._flagIsActive(this.FLAG_CLEAN_CONDITIONALLY)) {

           this._removeFlag(this.FLAG_CLEAN_CONDITIONALLY);

         } else {

           return null;

         }

       } else {

         if (lastTopCandidate !== null) {

           // EXPERIMENTAL: Contrast ratio is how we measure the level of competition between candidates in the

           // readability algorithm. This is to avoid offering reader mode on pages that are more like

           // a list or directory of links with summaries. It takes the score of the last top candidate

           // (see N_TOP_CANDIDATES) and checks how it compares to the top candidate's. On pages that are not

           // actual articles, there will likely be many candidates with similar score (i.e. higher contrast ratio).

           let contrastRatio = lastTopCandidate.readability.contentScore / topCandidate.readability.contentScore;

           if (contrastRatio > 0.45)

             return null;

         }

         return articleContent;

       }

     }

   },

   /**

    * Attempts to get the excerpt from these

    * sources in the following order:

    * - meta description tag

    * - open-graph description

    * - twitter cards description

    * - article's first paragraph

    * If no excerpt is found, an empty string will be

    * returned.

    *

    * @param Element - root element of the processed version page

    * @return String - excerpt of the article

   **/

   _getExcerpt: function(articleContent) {

     let values = {};

     let metaElements = this._doc.getElementsByTagName("meta");

     // Match "description", or Twitter's "twitter:description" (Cards)

     // in name attribute.

     let namePattern = /^\s*((twitter)\s*:\s*)?description\s*$/gi;

     // Match Facebook's og:description (Open Graph) in property attribute.

     let propertyPattern = /^\s*og\s*:\s*description\s*$/gi;

     // Find description tags.

     for (let i = 0; i < metaElements.length; i++) {

       let element = metaElements[i];

       let elementName = element.getAttribute("name");

       let elementProperty = element.getAttribute("property");

       let name;

       if (namePattern.test(elementName)) {

         name = elementName;

       } else if (propertyPattern.test(elementProperty)) {

         name = elementProperty;

       }

       if (name) {

         let content = element.getAttribute("content");

         if (content) {

           // Convert to lowercase and remove any whitespace

           // so we can match below.

           name = name.toLowerCase().replace(/\s/g, '');

           values[name] = content.trim();

         }

       }

     }

     if ("description" in values) {

       return values["description"];

     }

     if ("og:description" in values) {

       // Use facebook open graph description.

       return values["og:description"];

     }

     if ("twitter:description" in values) {

       // Use twitter cards description.

       return values["twitter:description"];

     }

     // No description meta tags, use the article's first paragraph.

     let paragraphs = articleContent.getElementsByTagName("p");

     if (paragraphs.length > 0) {

       return paragraphs[0].textContent;

     }

     return "";

   },

   /**

    * Removes script tags from the document.

    *

    * @param Element

   **/

   _removeScripts: function(doc) {

     let scripts = doc.getElementsByTagName('script');

     for (let i = scripts.length - 1; i >= 0; i -= 1) {

       scripts[i].nodeValue="";

       scripts[i].removeAttribute('src');

       if (scripts[i].parentNode)

           scripts[i].parentNode.removeChild(scripts[i]);

     }

   },

   /**

    * Get child index of the only P element inside a DIV with no

    * text content. Returns -1 if the DIV node contains non-empty

    * text nodes or if it contains other element nodes.

    *

    * @param Element

   **/

   _getSinglePIndexInsideDiv: function(e) {

     let childNodes = e.childNodes;

     let pIndex = -1;

     for (let i = childNodes.length; --i >= 0;) {

       let node = childNodes[i];

       if (node.nodeType === Node.ELEMENT_NODE) {

         if (node.tagName !== "P")

           return -1;

         if (pIndex >= 0)

           return -1;

         pIndex = i;

       } else if (node.nodeType == Node.TEXT_NODE && this._getInnerText(node, false)) {

         return -1;

       }

     }

     return pIndex;

   },

   /**

    * Determine whether element has any children block level elements.

    * 

    * @param Element

    */

   _hasChildBlockElement: function (e) {

     let length = e.childNodes.length;

     for (let i = 0; i < length; i++) {

       let child = e.childNodes[i];

       if (child.nodeType != 1)

         continue;

       if (this.DIV_TO_P_ELEMS.indexOf(child.tagName) !== -1 || this._hasChildBlockElement(child))

         return true;

     }

     return false;

   },

   /**

    * Get the inner text of a node - cross browser compatibly.

    * This also strips out any excess whitespace to be found.

    *

    * @param Element

    * @return string

   **/

   _getInnerText: function(e, normalizeSpaces) {

     let textContent = e.textContent.replace(this.REGEXPS.trim, "");

     normalizeSpaces = (typeof normalizeSpaces === 'undefined') ? true : normalizeSpaces;

     if (normalizeSpaces) {

       return textContent.replace(this.REGEXPS.normalize, " ");

     } else {

       return textContent;

     }

   },

   /**

    * Get the number of times a string s appears in the node e.

    *

    * @param Element

    * @param string - what to split on. Default is ","

    * @return number (integer)

   **/

   _getCharCount: function(e,s) {

     s = s || ",";

     return this._getInnerText(e).split(s).length - 1;

   },

   /**

    * Remove the style attribute on every e and under.

    * TODO: Test if getElementsByTagName(*) is faster.

    *

    * @param Element

    * @return void

   **/

   _cleanStyles: function(e) {

     e = e || this._doc;

     let cur = e.firstChild;

     if (!e)

       return;

     // Remove any root styles, if we're able.

     if (typeof e.removeAttribute === 'function' && e.className !== 'readability-styled')

       e.removeAttribute('style');

     // Go until there are no more child nodes

     while (cur !== null) {

       if (cur.nodeType === 1) {

         // Remove style attribute(s) :

         if (cur.className !== "readability-styled")

           cur.removeAttribute("style");

         this._cleanStyles(cur);

       }

       cur = cur.nextSibling;

     }

   },

   /**

    * Get the density of links as a percentage of the content

    * This is the amount of text that is inside a link divided by the total text in the node.

    *

    * @param Element

    * @return number (float)

   **/

   _getLinkDensity: function(e) {

     let links = e.getElementsByTagName("a");

     let textLength = this._getInnerText(e).length;

     let linkLength = 0;

     for (let i = 0, il = links.length; i < il; i += 1) {

       linkLength += this._getInnerText(links[i]).length;

     }

     return linkLength / textLength;

   },

   /**

    * Find a cleaned up version of the current URL, to use for comparing links for possible next-pageyness.

    *

    * @author Dan Lacy

    * @return string the base url

   **/

   _findBaseUrl: function() {

     let uri = this._uri;

     let noUrlParams = uri.path.split("?")[0];

     let urlSlashes = noUrlParams.split("/").reverse();

     let cleanedSegments = [];

     let possibleType = "";

     for (let i = 0, slashLen = urlSlashes.length; i < slashLen; i += 1) {

       let segment = urlSlashes[i];

       // Split off and save anything that looks like a file type.

       if (segment.indexOf(".") !== -1) {

         possibleType = segment.split(".")[1];

         // If the type isn't alpha-only, it's probably not actually a file extension.

         if (!possibleType.match(/[^a-zA-Z]/))

           segment = segment.split(".")[0];

       }

       // EW-CMS specific segment replacement. Ugly.

       // Example: http://www.ew.com/ew/article/0,,20313460_20369436,00.html

       if (segment.indexOf(',00') !== -1)

         segment = segment.replace(',00', '');

       // If our first or second segment has anything looking like a page number, remove it.

       if (segment.match(/((_|-)?p[a-z]*|(_|-))[0-9]{1,2}$/i) && ((i === 1) || (i === 0)))

         segment = segment.replace(/((_|-)?p[a-z]*|(_|-))[0-9]{1,2}$/i, "");

       let del = false;

       // If this is purely a number, and it's the first or second segment,

       // it's probably a page number. Remove it.

       if (i < 2 && segment.match(/^\d{1,2}$/))

         del = true;

       // If this is the first segment and it's just "index", remove it.

       if (i === 0 && segment.toLowerCase() === "index")

         del = true;

       // If our first or second segment is smaller than 3 characters,

       // and the first segment was purely alphas, remove it.

       if (i < 2 && segment.length < 3 && !urlSlashes[0].match(/[a-z]/i))

         del = true;

       // If it's not marked for deletion, push it to cleanedSegments.

       if (!del)

         cleanedSegments.push(segment);

     }

     // This is our final, cleaned, base article URL.

     return uri.scheme + "://" + uri.host + cleanedSegments.reverse().join("/");

   },

   /**

    * Look for any paging links that may occur within the document.

    *

    * @param body

    * @return object (array)

   **/

   _findNextPageLink: function(elem) {

     let uri = this._uri;

     let possiblePages = {};

     let allLinks = elem.getElementsByTagName('a');

     let articleBaseUrl = this._findBaseUrl();

     // Loop through all links, looking for hints that they may be next-page links.

     // Things like having "page" in their textContent, className or id, or being a child

     // of a node with a page-y className or id.

     //

     // Also possible: levenshtein distance? longest common subsequence?

     //

     // After we do that, assign each page a score, and

     for (let i = 0, il = allLinks.length; i < il; i += 1) {

       let link = allLinks[i];

       let linkHref = allLinks[i].href.replace(/#.*$/, '').replace(/\/$/, '');

       // If we've already seen this page, ignore it.

       if (linkHref === "" ||

         linkHref === articleBaseUrl ||

         linkHref === uri.spec ||

         linkHref in this._parsedPages) {

         continue;

       }

       // If it's on a different domain, skip it.

       if (uri.host !== linkHref.split(/\/+/g)[1])

         continue;

       let linkText = this._getInnerText(link);

       // If the linkText looks like it's not the next page, skip it.

       if (linkText.match(this.REGEXPS.extraneous) || linkText.length > 25)

         continue;

       // If the leftovers of the URL after removing the base URL don't contain

       // any digits, it's certainly not a next page link.

       let linkHrefLeftover = linkHref.replace(articleBaseUrl, '');

       if (!linkHrefLeftover.match(/\d/))

         continue;

       if (!(linkHref in possiblePages)) {

         possiblePages[linkHref] = {"score": 0, "linkText": linkText, "href": linkHref};

       } else {

         possiblePages[linkHref].linkText += ' | ' + linkText;

       }

       let linkObj = possiblePages[linkHref];

       // If the articleBaseUrl isn't part of this URL, penalize this link. It could

       // still be the link, but the odds are lower.

       // Example: http://www.actionscript.org/resources/articles/745/1/JavaScript-and-VBScript-Injection-in-ActionScript-3/Page1.html

       if (linkHref.indexOf(articleBaseUrl) !== 0)

         linkObj.score -= 25;

       let linkData = linkText + ' ' + link.className + ' ' + link.id;

       if (linkData.match(this.REGEXPS.nextLink))

         linkObj.score += 50;

       if (linkData.match(/pag(e|ing|inat)/i))

         linkObj.score += 25;

       if (linkData.match(/(first|last)/i)) {

         // -65 is enough to negate any bonuses gotten from a > or » in the text,

         // If we already matched on "next", last is probably fine.

         // If we didn't, then it's bad. Penalize.

         if (!linkObj.linkText.match(this.REGEXPS.nextLink))

           linkObj.score -= 65;

       }

       if (linkData.match(this.REGEXPS.negative) || linkData.match(this.REGEXPS.extraneous))

         linkObj.score -= 50;

       if (linkData.match(this.REGEXPS.prevLink))

         linkObj.score -= 200;

       // If a parentNode contains page or paging or paginat

       let parentNode = link.parentNode;

       let positiveNodeMatch = false;

       let negativeNodeMatch = false;

       while (parentNode) {

         let parentNodeClassAndId = parentNode.className + ' ' + parentNode.id;

         if (!positiveNodeMatch && parentNodeClassAndId && parentNodeClassAndId.match(/pag(e|ing|inat)/i)) {

           positiveNodeMatch = true;

           linkObj.score += 25;

         }

         if (!negativeNodeMatch && parentNodeClassAndId && parentNodeClassAndId.match(this.REGEXPS.negative)) {

           // If this is just something like "footer", give it a negative.

           // If it's something like "body-and-footer", leave it be.

           if (!parentNodeClassAndId.match(this.REGEXPS.positive)) {

             linkObj.score -= 25;

             negativeNodeMatch = true;

           }

         }

         parentNode = parentNode.parentNode;

       }

       // If the URL looks like it has paging in it, add to the score.

       // Things like /page/2/, /pagenum/2, ?p=3, ?page=11, ?pagination=34

       if (linkHref.match(/p(a|g|ag)?(e|ing|ination)?(=|\/)[0-9]{1,2}/i) || linkHref.match(/(page|paging)/i))

         linkObj.score += 25;

       // If the URL contains negative values, give a slight decrease.

       if (linkHref.match(this.REGEXPS.extraneous))

         linkObj.score -= 15;

       /**

        * Minor punishment to anything that doesn't match our current URL.

        * NOTE: I'm finding this to cause more harm than good where something is exactly 50 points.

        *     Dan, can you show me a counterexample where this is necessary?

        * if (linkHref.indexOf(window.location.href) !== 0) {

        *  linkObj.score -= 1;

        * }

       **/

       // If the link text can be parsed as a number, give it a minor bonus, with a slight

       // bias towards lower numbered pages. This is so that pages that might not have 'next'

       // in their text can still get scored, and sorted properly by score.

       let linkTextAsNumber = parseInt(linkText, 10);

       if (linkTextAsNumber) {

         // Punish 1 since we're either already there, or it's probably

         // before what we want anyways.

         if (linkTextAsNumber === 1) {

           linkObj.score -= 10;

         } else {

           linkObj.score += Math.max(0, 10 - linkTextAsNumber);

         }

       }

     }

     // Loop thrugh all of our possible pages from above and find our top

     // candidate for the next page URL. Require at least a score of 50, which

     // is a relatively high confidence that this page is the next link.

     let topPage = null;

     for (let page in possiblePages) {

       if (possiblePages.hasOwnProperty(page)) {

         if (possiblePages[page].score >= 50 &&

           (!topPage || topPage.score < possiblePages[page].score))

           topPage = possiblePages[page];

       }

     }

     if (topPage) {

       let nextHref = topPage.href.replace(/\/$/,'');

       this.log('NEXT PAGE IS ' + nextHref);

       this._parsedPages[nextHref] = true;

       return nextHref;

     } else {

       return null;

     }

   },

   _successfulRequest: function(request) {

     return (request.status >= 200 && request.status < 300) ||

         request.status === 304 ||

          (request.status === 0 && request.responseText);

   },

   _ajax: function(url, options) {

     let request = new XMLHttpRequest();

     function respondToReadyState(readyState) {

       if (request.readyState === 4) {

         if (this._successfulRequest(request)) {

           if (options.success)

             options.success(request);

         } else {

           if (options.error)

             options.error(request);

         }

       }

     }

     if (typeof options === 'undefined')

       options = {};

     request.onreadystatechange = respondToReadyState;

     request.open('get', url, true);

     request.setRequestHeader('Accept', 'text/html');

     try {

       request.send(options.postBody);

     } catch (e) {

       if (options.error)

         options.error();

     }

     return request;

   },

   _appendNextPage: function(nextPageLink) {

     let doc = this._doc;

     this._curPageNum += 1;

     let articlePage = doc.createElement("DIV");

     articlePage.id = 'readability-page-' + this._curPageNum;

     articlePage.className = 'page';

     articlePage.innerHTML = '<p class="page-separator" title="Page ' + this._curPageNum + '">&sect;</p>';

     doc.getElementById("readability-content").appendChild(articlePage);

     if (this._curPageNum > this.MAX_PAGES) {

       let nextPageMarkup = "<div style='text-align: center'><a href='" + nextPageLink + "'>View Next Page</a></div>";

       articlePage.innerHTML = articlePage.innerHTML + nextPageMarkup;

       return;

     }

     // Now that we've built the article page DOM element, get the page content

     // asynchronously and load the cleaned content into the div we created for it.

     (function(pageUrl, thisPage) {

       this._ajax(pageUrl, {

         success: function(r) {

           // First, check to see if we have a matching ETag in headers - if we do, this is a duplicate page.

           let eTag = r.getResponseHeader('ETag');

           if (eTag) {

             if (eTag in this._pageETags) {

               this.log("Exact duplicate page found via ETag. Aborting.");

               articlePage.style.display = 'none';

               return;

             } else {

               this._pageETags[eTag] = 1;

             }

           }

           // TODO: this ends up doubling up page numbers on NYTimes articles. Need to generically parse those away.

           let page = doc.createElement("DIV");

           // Do some preprocessing to our HTML to make it ready for appending.

           // - Remove any script tags. Swap and reswap newlines with a unicode

           //   character because multiline regex doesn't work in javascript.

           // - Turn any noscript tags into divs so that we can parse them. This

           //   allows us to find any next page links hidden via javascript.

           // - Turn all double br's into p's - was handled by prepDocument in the original view.

           //   Maybe in the future abstract out prepDocument to work for both the original document

           //   and AJAX-added pages.

           let responseHtml = r.responseText.replace(/\n/g,'\uffff').replace(/<script.*?>.*?<\/script>/gi, '');

           responseHtml = responseHtml.replace(/\n/g,'\uffff').replace(/<script.*?>.*?<\/script>/gi, '');

           responseHtml = responseHtml.replace(/\uffff/g,'\n').replace(/<(\/?)noscript/gi, '<$1div');

           responseHtml = responseHtml.replace(this.REGEXPS.replaceFonts, '<$1span>');

           page.innerHTML = responseHtml;

           this._replaceBrs(page);

           // Reset all flags for the next page, as they will search through it and

           // disable as necessary at the end of grabArticle.

           this._flags = 0x1 | 0x2 | 0x4;

           let nextPageLink = this._findNextPageLink(page);

           // NOTE: if we end up supporting _appendNextPage(), we'll need to

           // change this call to be async

           let content = this._grabArticle(page);

           if (!content) {

             this.log("No content found in page to append. Aborting.");

             return;

           }

           // Anti-duplicate mechanism. Essentially, get the first paragraph of our new page.

           // Compare it against all of the the previous document's we've gotten. If the previous

           // document contains exactly the innerHTML of this first paragraph, it's probably a duplicate.

           let firstP = content.getElementsByTagName("P").length ? content.getElementsByTagName("P")[0] : null;

           if (firstP && firstP.innerHTML.length > 100) {

             for (let i = 1; i <= this._curPageNum; i += 1) {

               let rPage = doc.getElementById('readability-page-' + i);

               if (rPage && rPage.innerHTML.indexOf(firstP.innerHTML) !== -1) {

                 this.log('Duplicate of page ' + i + ' - skipping.');

                 articlePage.style.display = 'none';

                 this._parsedPages[pageUrl] = true;

                 return;

               }

             }

           }

           this._removeScripts(content);

           thisPage.innerHTML = thisPage.innerHTML + content.innerHTML;

           // After the page has rendered, post process the content. This delay is necessary because,

           // in webkit at least, offsetWidth is not set in time to determine image width. We have to

           // wait a little bit for reflow to finish before we can fix floating images.

           setTimeout((function() {

             this._postProcessContent(thisPage);

           }).bind(this), 500);

           if (nextPageLink)

             this._appendNextPage(nextPageLink);

         }

       });

     }).bind(this)(nextPageLink, articlePage);

   },

   /**

    * Get an elements class/id weight. Uses regular expressions to tell if this 

    * element looks good or bad.

    *

    * @param Element

    * @return number (Integer)

   **/

   _getClassWeight: function(e) {

     if (!this._flagIsActive(this.FLAG_WEIGHT_CLASSES))

       return 0;

     let weight = 0;

     // Look for a special classname

     if (typeof(e.className) === 'string' && e.className !== '') {

       if (e.className.search(this.REGEXPS.negative) !== -1)

         weight -= 25;

       if (e.className.search(this.REGEXPS.positive) !== -1)

         weight += 25;

     }

     // Look for a special ID

     if (typeof(e.id) === 'string' && e.id !== '') {

       if (e.id.search(this.REGEXPS.negative) !== -1)

         weight -= 25;

       if (e.id.search(this.REGEXPS.positive) !== -1)

         weight += 25;

     }

     return weight;

   },

   /**

    * Clean a node of all elements of type "tag".

    * (Unless it's a youtube/vimeo video. People love movies.)

    *

    * @param Element

    * @param string tag to clean

    * @return void

    **/

   _clean: function(e, tag) {

     let targetList = e.getElementsByTagName(tag);

     let isEmbed = (tag === 'object' || tag === 'embed');

     for (let y = targetList.length - 1; y >= 0; y -= 1) {

       // Allow youtube and vimeo videos through as people usually want to see those.

       if (isEmbed) {

         let attributeValues = "";

         for (let i = 0, il = targetList[y].attributes.length; i < il; i += 1) {

           attributeValues += targetList[y].attributes[i].value + '|';

         }

         // First, check the elements attributes to see if any of them contain youtube or vimeo

         if (attributeValues.search(this.REGEXPS.videos) !== -1)

           continue;

         // Then check the elements inside this element for the same.

         if (targetList[y].innerHTML.search(this.REGEXPS.videos) !== -1)

           continue;

       }

       targetList[y].parentNode.removeChild(targetList[y]);

     }

   },

   /**

    * Clean an element of all tags of type "tag" if they look fishy.

    * "Fishy" is an algorithm based on content length, classnames, link density, number of images & embeds, etc.

    *

    * @return void

    **/

   _cleanConditionally: function(e, tag) {

     if (!this._flagIsActive(this.FLAG_CLEAN_CONDITIONALLY))

       return;

     let tagsList = e.getElementsByTagName(tag);

     let curTagsLength = tagsList.length;

     // Gather counts for other typical elements embedded within.

     // Traverse backwards so we can remove nodes at the same time

     // without effecting the traversal.

     //

     // TODO: Consider taking into account original contentScore here.

     for (let i = curTagsLength-1; i >= 0; i -= 1) {

       let weight = this._getClassWeight(tagsList[i]);

       let contentScore = 0;

       this.log("Cleaning Conditionally " + tagsList[i] + " (" + tagsList[i].className + ":" + tagsList[i].id + ")");

       if (weight + contentScore < 0) {

         tagsList[i].parentNode.removeChild(tagsList[i]);

       } else if (this._getCharCount(tagsList[i],',') < 10) {

         // If there are not very many commas, and the number of

         // non-paragraph elements is more than paragraphs or other

         // ominous signs, remove the element.

         let p = tagsList[i].getElementsByTagName("p").length;

         let img = tagsList[i].getElementsByTagName("img").length;

         let li = tagsList[i].getElementsByTagName("li").length-100;

         let input = tagsList[i].getElementsByTagName("input").length;

         let embedCount = 0;

         let embeds = tagsList[i].getElementsByTagName("embed");

         for (let ei = 0, il = embeds.length; ei < il; ei += 1) {

           if (embeds[ei].src.search(this.REGEXPS.videos) === -1)

             embedCount += 1;

         }

         let linkDensity = this._getLinkDensity(tagsList[i]);

         let contentLength = this._getInnerText(tagsList[i]).length;

         let toRemove = false;

         if (img > p) {

           toRemove = true;

         } else if (li > p && tag !== "ul" && tag !== "ol") {

           toRemove = true;

         } else if ( input > Math.floor(p/3) ) {

           toRemove = true;

         } else if (contentLength < 25 && (img === 0 || img > 2) ) {

           toRemove = true;

         } else if (weight < 25 && linkDensity > 0.2) {

           toRemove = true;

         } else if (weight >= 25 && linkDensity > 0.5) {

           toRemove = true;

         } else if ((embedCount === 1 && contentLength < 75) || embedCount > 1) {

           toRemove = true;

         }

         if (toRemove)

           tagsList[i].parentNode.removeChild(tagsList[i]);

       }

     }

   },

   /**

    * Clean out spurious headers from an Element. Checks things like classnames and link density.

    *

    * @param Element

    * @return void

   **/

   _cleanHeaders: function(e) {

     for (let headerIndex = 1; headerIndex < 3; headerIndex += 1) {

       let headers = e.getElementsByTagName('h' + headerIndex);

       for (let i = headers.length - 1; i >= 0; i -= 1) {

         if (this._getClassWeight(headers[i]) < 0 || this._getLinkDensity(headers[i]) > 0.33)

           headers[i].parentNode.removeChild(headers[i]);

       }

     }

   },

   _flagIsActive: function(flag) {

     return (this._flags & flag) > 0;

   },

   _addFlag: function(flag) {

     this._flags = this._flags | flag;

   },

   _removeFlag: function(flag) {

     this._flags = this._flags & ~flag;

   },

   /**

    * Runs readability.

    *

    * Workflow:

    *  1. Prep the document by removing script tags, css, etc.

    *  2. Build readability's DOM tree.

    *  3. Grab the article content from the current dom tree.

    *  4. Replace the current DOM tree with the new one.

    *  5. Read peacefully.

    *

    * @return void

    **/

   parse: function () {

     // Remove script tags from the document.

     this._removeScripts(this._doc);

     // FIXME: Disabled multi-page article support for now as it

     // needs more work on infrastructure.

     // Make sure this document is added to the list of parsed pages first,

     // so we don't double up on the first page.

     // this._parsedPages[uri.spec.replace(/\/$/, '')] = true;

     // Pull out any possible next page link first.

     // let nextPageLink = this._findNextPageLink(doc.body);

     this._prepDocument();

     let articleTitle = this._getArticleTitle();

     let articleContent = this._grabArticle();

     if (!articleContent)

       return null;

     this._postProcessContent(articleContent);

     // if (nextPageLink) {

     //   // Append any additional pages after a small timeout so that people

     //   // can start reading without having to wait for this to finish processing.

     //   setTimeout((function() {

     //     this._appendNextPage(nextPageLink);

     //   }).bind(this), 500);

     // }

     let excerpt = this._getExcerpt(articleContent);

     return { title: articleTitle,

              byline: this._articleByline,

              dir: this._articleDir,

              content: articleContent.innerHTML,

              length: articleContent.textContent.length,

              excerpt: excerpt };

   }

 };