Prefer JSON-LD metadata object, when present (mozilla#609)

* Prefer JSON-LD metadata object, when present

* Log JSON-LD parsing error

* Trim all JSON-LD fields

Readability.js

@@ -53,6 +53,7 @@ function Readability(doc, options) {
   this._serializer = options.serializer || function(el) {
     return el.innerHTML;
   };
+  this._disableJSONLD = !!options.disableJSONLD;
 
   // Start with all flags set
   this._flags = this.FLAG_STRIP_UNLIKELYS |
@@ -135,7 +136,9 @@ Readability.prototype = {
     whitespace: /^\s*$/,
     hasContent: /\S$/,
     srcsetUrl: /(\S+)(\s+[\d.]+[xw])?(\s*(?:,|$))/g,
-    b64DataUrl: /^data:\s*([^\s;,]+)\s*;\s*base64\s*,/i
+    b64DataUrl: /^data:\s*([^\s;,]+)\s*;\s*base64\s*,/i,
+    // See: https://schema.org/Article
+    jsonLdArticleTypes: /^Article|AdvertiserContentArticle|NewsArticle|AnalysisNewsArticle|AskPublicNewsArticle|BackgroundNewsArticle|OpinionNewsArticle|ReportageNewsArticle|ReviewNewsArticle|Report|SatiricalArticle|ScholarlyArticle|MedicalScholarlyArticle|SocialMediaPosting|BlogPosting|LiveBlogPosting|DiscussionForumPosting|TechArticle|APIReference$/
   },
 
   DIV_TO_P_ELEMS: [ "A", "BLOCKQUOTE", "DL", "DIV", "IMG", "OL", "P", "PRE", "TABLE", "UL", "SELECT" ],
@@ -244,6 +247,21 @@ Readability.prototype = {
     Array.prototype.forEach.call(nodeList, fn, this);
   },
 
+  /**
+   * Iterate over a NodeList, and return the first node that passes
+   * the supplied test function
+   *
+   * For convenience, the current object context is applied to the provided
+   * test function.
+   *
+   * @param  NodeList nodeList The NodeList.
+   * @param  Function fn       The test function.
+   * @return void
+   */
+  _findNode: function(nodeList, fn) {
+    return Array.prototype.find.call(nodeList, fn, this);
+  },
+
   /**
    * Iterate over a NodeList, return true if any of the provided iterate
    * function calls returns true, false otherwise.
@@ -1292,12 +1310,80 @@ Readability.prototype = {
     });
   },
 
+  /**
+   * Try to extract metadata from JSON-LD object.
+   * For now, only Schema.org objects of type Article or its subtypes are supported.
+   * @return Object with any metadata that could be extracted (possibly none)
+   */
+  _getJSONLD: function (doc) {
+    var scripts = this._getAllNodesWithTag(doc, ["script"]);
+
+    var jsonLdElement = this._findNode(scripts, function(el) {
+      return el.getAttribute("type") === "application/ld+json";
+    });
+
+    if (jsonLdElement) {
+      try {
+        // Strip CDATA markers if present
+        var content = jsonLdElement.textContent.replace(/^\s*<!\[CDATA\[|\]\]>\s*$/g, "");
+        var parsed = JSON.parse(content);
+        var metadata = {};
+        if (
+          !parsed["@context"] ||
+          !parsed["@context"].match(/^https?\:\/\/schema\.org$/)
+        ) {
+          return metadata;
+        }
+
+        if (!parsed["@type"] && Array.isArray(parsed["@graph"])) {
+          parsed = parsed["@graph"].find(function(it) {
+            return (it["@type"] || "").match(
+              this.REGEXPS.jsonLdArticleTypes
+            );
+          });
+        }
+
+        if (
+          !parsed ||
+          !parsed["@type"] ||
+          !parsed["@type"].match(this.REGEXPS.jsonLdArticleTypes)
+        ) {
+          return metadata;
+        }
+        if (typeof parsed.name === "string") {
+          metadata.title = parsed.name.trim();
+        } else if (typeof parsed.headline === "string") {
+          metadata.title = parsed.headline.trim();
+        }
+        if (parsed.author && typeof parsed.author.name === "string") {
+          metadata.byline = parsed.author.name.trim();
+        }
+        if (typeof parsed.description === "string") {
+          metadata.excerpt = parsed.description.trim();
+        }
+        if (
+          parsed.publisher &&
+          typeof parsed.publisher.name === "string"
+        ) {
+          metadata.siteName = parsed.publisher.name.trim();
+        }
+        return metadata;
+      } catch (err) {
+        this.log(err.message);
+      }
+    }
+    return {};
+  },
+
   /**
    * Attempts to get excerpt and byline metadata for the article.
    *
+   * @param {Object} jsonld — object containing any metadata that
+   * could be extracted from JSON-LD object.
+   *
    * @return Object with optional "excerpt" and "byline" properties
    */
-  _getArticleMetadata: function() {
+  _getArticleMetadata: function(jsonld) {
     var metadata = {};
     var values = {};
     var metaElements = this._doc.getElementsByTagName("meta");
@@ -1343,7 +1429,8 @@ Readability.prototype = {
     });
 
     // get title
-    metadata.title = values["dc:title"] ||
+    metadata.title = jsonld.title ||
+                     values["dc:title"] ||
                      values["dcterm:title"] ||
                      values["og:title"] ||
                      values["weibo:article:title"] ||
@@ -1356,12 +1443,14 @@ Readability.prototype = {
     }
 
     // get author
-    metadata.byline = values["dc:creator"] ||
+    metadata.byline = jsonld.byline ||
+                      values["dc:creator"] ||
                       values["dcterm:creator"] ||
                       values["author"];
 
     // get description
-    metadata.excerpt = values["dc:description"] ||
+    metadata.excerpt = jsonld.excerpt ||
+                       values["dc:description"] ||
                        values["dcterm:description"] ||
                        values["og:description"] ||
                        values["weibo:article:description"] ||
@@ -1370,7 +1459,8 @@ Readability.prototype = {
                        values["twitter:description"];
 
     // get site name
-    metadata.siteName = values["og:site_name"];
+    metadata.siteName = jsonld.siteName ||
+                        values["og:site_name"];
 
     // in many sites the meta value is escaped with HTML entities,
     // so here we need to unescape it
@@ -2029,12 +2119,15 @@ Readability.prototype = {
     // Unwrap image from noscript
     this._unwrapNoscriptImages(this._doc);
 
+    // Extract JSON-LD metadata before removing scripts
+    var jsonLd = this._disableJSONLD ? {} : this._getJSONLD(this._doc);
+
     // Remove script tags from the document.
     this._removeScripts(this._doc);
 
     this._prepDocument();
 
-    var metadata = this._getArticleMetadata();
+    var metadata = this._getArticleMetadata(jsonLd);
     this._articleTitle = metadata.title;
 
     var articleContent = this._grabArticle();

test/test-pages/aclu/expected-metadata.json

@@ -2,7 +2,7 @@
   "title": "Facebook Is Tracking Me Even Though I’m Not on Facebook",
   "byline": "By Daniel Kahn Gillmor, Senior Staff Technologist, ACLU Speech, Privacy, and Technology Project",
   "dir": "ltr",
-  "excerpt": "I don't use Facebook. I'm not technophobic — I'm a geek. I've been using email since the early 1990s, I have accounts on hundreds of services around the net, and I do software development and internet protocol design both for work and for fun. I believe that a globe-spanning communications network like the internet can be a positive social force, and I publish much of my own work on the open web.",
+  "excerpt": "Facebook collects data about people who have never even opted in. But there are ways these non-users can protect themselves.",
   "readerable": true,
   "siteName": "American Civil Liberties Union"
 }

test/test-pages/bbc-1/expected-metadata.json

@@ -1,5 +1,5 @@
 {
-  "title": "Obama admits US gun laws are his 'biggest frustration' - BBC News",
+  "title": "Obama admits US gun laws are his 'biggest frustration'",
   "byline": null,
   "excerpt": "President Barack Obama tells the BBC his failure to pass \"common sense gun safety laws\" is the greatest frustration of his presidency.",
   "readerable": true,

test/test-pages/folha/expected-metadata.json

@@ -2,7 +2,7 @@
   "title": "Tite diz que errou ao levar taça da Libertadores a Lula em 2012",
   "byline": "21.dez.2018 às 10h55",
   "dir": null,
-  "excerpt": "Após rechaçar um encontro da seleção brasileira com o presidente eleito Jair Bolsonaro, o técnico Tite declarou que errou ao levar a taça da Copa Libertadores de 2012, conquistada pelo Corinthians, ao ex-presidente Luiz Inácio Lula da Silva.",
+  "excerpt": "Na ocasião, técnico do Corinthians entregou réplica do troféu ao ex-presidente",
   "siteName": "Folha de S.Paulo",
   "readerable": true
 }

test/test-pages/folha/source.html

@@ -204,7 +204,7 @@
         "description": "Na ocasião, técnico do Corinthians entregou réplica do troféu ao ex-presidente",
         "datePublished": "2018-12-21T12:55:00Z",
 
-        "image": { "@type": "ImageObject", "url": "https://f.i.uol.com.br/fotografia/2018/12/21/15454034955c1cfc67131dc_1545403495_3x2_md.jpg", "width": "768", "height": "512" }
+        "image": { "@type": "ImageObject", "url": "https://f.i.uol.com.br/fotografia/2018/12/21/15454034955c1cfc67131dc_1545403495_3x2_md.jpg", "width": "768", "height": "512" },
 
         "contentLocation": {
         "@type": "Place",

test/test-pages/lazy-image-1/expected-metadata.json

@@ -2,7 +2,7 @@
   "title": "Node.js and CPU profiling on production (in real-time without downtime)",
   "byline": "Vincent Vallet",
   "dir": null,
-  "excerpt": "Why CPU monitoring is important?",
-  "siteName": "Medium",
+  "excerpt": "How to run a CPU profiling with Node.js on your production in real-time and without interruption of service.",
+  "siteName": "Voodoo Engineering",
   "readerable": true
 }

test/test-pages/videos-1/expected-metadata.json

@@ -1,8 +1,8 @@
 {
-  "title": "The 21 best movies of 2017",
-  "byline": "By Alissa Wilkinson@alissamarie\n                                 Updated Jul 24, 2018, 2:15pm EDT",
+  "title": "How to watch the 21 best films of 2017",
+  "byline": "Alissa Wilkinson",
   "dir": null,
-  "excerpt": "How to watch the greatest movies of the year, from Lady Bird and Dunkirk to Get Out and The Big Sick.",
+  "excerpt": "It was an extraordinary year for movies.",
   "siteName": "Vox",
   "readerable": true
 }

test/test-pages/videos-2/expected-metadata.json

@@ -2,7 +2,7 @@
   "title": "Screenshot : «Vape Wave», «6 Days», «Alphonse Président»…",
   "byline": "Par Alexandre Hervaud et Jérémy Piette",
   "dir": null,
-  "excerpt": "Séries, documentaires, programmes jeunesse… Retrouvez les recommandations de Libération pour savoir quoi regarder sur vos écrans cette semaine. Pour dépasser...",
-  "siteName": "Libération.fr",
+  "excerpt": "Séries, documentaires, programmes jeunesse… Retrouvez les recommandations de Libération pour savoir quoi regarder sur vos écrans cette semaine.\nPour dépasser...",
+  "siteName": "Libération",
   "readerable": true
 }

test/test-pages/wikipedia-2/expected-metadata.json

@@ -1,8 +1,8 @@
 {
-  "title": "New Zealand - Wikipedia",
-  "byline": "Authority control",
+  "title": "New Zealand",
+  "byline": "Contributors to Wikimedia projects",
   "dir": "ltr",
   "excerpt": "Coordinates: 42°S 174°E / 42°S 174°E",
-  "siteName": null,
+  "siteName": "Wikimedia Foundation, Inc.",
   "readerable": true
 }

test/test-pages/wikipedia-3/expected-metadata.json

@@ -1,8 +1,8 @@
 {
-  "title": "Hermitian matrix - Wikipedia",
-  "byline": null,
+  "title": "Hermitian matrix",
+  "byline": "Contributors to Wikimedia projects",
   "dir": "ltr",
   "excerpt": "In mathematics, a Hermitian matrix (or self-adjoint matrix) is a complex square matrix that is equal to its own conjugate transpose—that is, the element in the i-th row and j-th column is equal to the complex conjugate of the element in the j-th row and i-th column, for all indices i and j:",
-  "siteName": null,
+  "siteName": "Wikimedia Foundation, Inc.",
   "readerable": true
 }

test/test-readability.js

@@ -157,23 +157,23 @@ function runTestsWithItems(label, domGenerationFn, source, expectedContent, expe
     });
 
     it("should extract expected title", function() {
-      expect(expectedMetadata.title).eql(result.title);
+      expect(result.title).eql(expectedMetadata.title);
     });
 
     it("should extract expected byline", function() {
-      expect(expectedMetadata.byline).eql(result.byline);
+      expect(result.byline).eql(expectedMetadata.byline);
     });
 
     it("should extract expected excerpt", function() {
-      expect(expectedMetadata.excerpt).eql(result.excerpt);
+      expect(result.excerpt).eql(expectedMetadata.excerpt);
     });
 
     it("should extract expected site name", function() {
-      expect(expectedMetadata.siteName).eql(result.siteName);
+      expect(result.siteName).eql(expectedMetadata.siteName);
     });
 
     expectedMetadata.dir && it("should extract expected direction", function() {
-      expect(expectedMetadata.dir).eql(result.dir);
+      expect(result.dir).eql(expectedMetadata.dir);
     });
   });
 }