Bugfix new yorker wired extractors (#604)

* www.newyorker.com: add updated fixtures and fix extractors

* www.wired.com: add updated fixtures and fix extractors

Co-authored-by: John Holdun <john@johnholdun.com>

src/extractors/custom/www.newyorker.com/index.js

@@ -5,6 +5,7 @@ export const NewYorkerExtractor = {
   domain: 'www.newyorker.com',
   title: {
     selectors: [
+      'h1[class^="content-header"]',
       'h1[class^="ArticleHeader__hed"]',
       ['meta[name="og:title"]', 'value'],
     ],
@@ -12,13 +13,17 @@ export const NewYorkerExtractor = {
 
   author: {
     selectors: [
+      ['meta[name="author"]', 'value'],
       'div[class^="ArticleContributors"] a[rel="author"]',
       'article header div[class*="Byline__multipleContributors"]',
     ],
   },
 
   content: {
-    selectors: ['main[class^="Layout__content"]'],
+    selectors: [
+      'article.article.main-content',
+      'main[class^="Layout__content"]',
+    ],
 
     // Is there anything in the content you selected that needs transformed
     // before it's consumable content? E.g., unusual lazy loaded images
@@ -31,8 +36,10 @@ export const NewYorkerExtractor = {
   },
 
   date_published: {
-    selectors: [['meta[name="pubdate"]', 'value']],
-    format: 'YYYYMMDD',
+    selectors: [
+      'time.content-header__publish-date',
+      ['meta[name="pubdate"]', 'value'],
+    ],
     timezone: 'America/New_York',
   },
 
@@ -41,7 +48,7 @@ export const NewYorkerExtractor = {
   },
 
   dek: {
-    selectors: ['h2[class^="ArticleHeader__dek"]'],
+    selectors: ['div.content-header__dek', 'h2[class^="ArticleHeader__dek"]'],
   },
 
   next_page_url: null,

src/extractors/custom/www.newyorker.com/index.test.js

@@ -16,7 +16,7 @@ describe('NewYorkerExtractor', () => {
       url =
         'http://www.newyorker.com/tech/elements/hacking-cryptography-and-the-countdown-to-quantum-computing';
       const html = fs.readFileSync(
-        './fixtures/www.newyorker.com/1557138180688.html'
+        './fixtures/www.newyorker.com/1611473608343.html'
       );
       result = Mercury.parse(url, { html, fallback: false });
     });
@@ -73,7 +73,7 @@ describe('NewYorkerExtractor', () => {
       // the article.
       assert.equal(
         lead_image_url,
-        'https://media.newyorker.com/photos/59097a5e8b51cf59fc4239f5/16:9/w_1200,h_630,c_limit/Hutchinson-Quantum-Computing.jpg'
+        'https://media.newyorker.com/photos/59097a5e8b51cf59fc4239f5/16:9/w_1280,c_limit/Hutchinson-Quantum-Computing.jpg'
       );
     });
 
@@ -97,7 +97,7 @@ describe('NewYorkerExtractor', () => {
       // the article.
       assert.equal(
         first13,
-        'In a laboratory in Shanghai, researchers work on developing a quantum computer—a new'
+        'Given the recent ubiquity of cyber-scandals—Colin Powell’s stolen e-mails, Simone Biles’s leaked medical'
       );
     });
   });
@@ -109,7 +109,7 @@ describe('NewYorkerExtractor', () => {
       url =
         'http://www.newyorker.com/magazine/2016/12/05/lessons-from-my-mother';
       const html = fs.readFileSync(
-        './fixtures/www.newyorker.com/1557145645680.html'
+        './fixtures/www.newyorker.com/1611475571383.html'
       );
       result = Mercury.parse(url, { html, fallback: false });
     });

src/extractors/custom/www.wired.com/index.js

@@ -5,6 +5,7 @@ export const WiredExtractor = {
   domain: 'www.wired.com',
   title: {
     selectors: [
+      'h1.content-header__hed',
       'h1.post-title',
       // enter title selectors
     ],
@@ -12,6 +13,7 @@ export const WiredExtractor = {
 
   author: {
     selectors: [
+      ['meta[name="author"]', 'value'],
       'a[rel="author"]',
       // enter author selectors
     ],
@@ -19,6 +21,7 @@ export const WiredExtractor = {
 
   content: {
     selectors: [
+      'article.article.main-content',
       'article.content',
       // enter content selectors
     ],
@@ -34,7 +37,10 @@ export const WiredExtractor = {
   },
 
   date_published: {
-    selectors: [['meta[itemprop="datePublished"]', 'value']],
+    selectors: [
+      'time.content-header__publish-date',
+      ['meta[itemprop="datePublished"]', 'value'],
+    ],
   },
 
   lead_image_url: {

src/extractors/custom/www.wired.com/index.test.js

@@ -17,7 +17,7 @@ describe('WiredExtractor', () => {
       url =
         'https://www.wired.com/2016/09/ode-rosetta-spacecraft-going-die-comet/';
       const html = fs.readFileSync(
-        './fixtures/www.wired.com/1475256747028.html'
+        './fixtures/www.wired.com/1611475755063.html'
       );
       result = Mercury.parse(url, { html, fallback: false });
     });
@@ -62,7 +62,7 @@ describe('WiredExtractor', () => {
 
       // Update these values with the expected values from
       // the article.
-      assert.equal(date_published, '2016-09-30T07:00:12.000Z');
+      assert.equal(date_published.split('T')[0], '2016-09-30');
     });
 
     it('returns the lead_image_url', async () => {
@@ -74,7 +74,7 @@ describe('WiredExtractor', () => {
       // the article.
       assert.equal(
         lead_image_url,
-        'https://www.wired.com/wp-content/uploads/2016/09/Rosetta_impact-1-1200x630.jpg'
+        'https://media.wired.com/photos/5926b676af95806129f50602/191:100/w_1280,c_limit/Rosetta_impact-1.jpg'
       );
     });
 
@@ -98,7 +98,7 @@ describe('WiredExtractor', () => {
       // the article.
       assert.equal(
         first13,
-        'Today, the European Space Agency’s Rosetta spacecraft will engage its thrusters for one'
+        "Today, the European Space Agency's Rosetta spacecraft will engage its thrusters for one"
       );
     });
   });