fix(route): improve economist espresso parser (#11599)

* Extract Espresso articles from Next.js data payload

* Set espresso gobbet headline as title

lib/v2/economist/espresso.js

@@ -15,36 +15,66 @@ module.exports = async (ctx) => {
         false
     );
 
-    const gobbets = $('._gobbets ._gobbet p')
-        .toArray()
-        .map((item) => {
-            item = $(item);
-            return {
-                title: item.text(),
-                description: item.html(),
-                link,
-                guid: item.text(),
-            };
-        });
+    const nextData = JSON.parse($('script#__NEXT_DATA__').text());
+    const parts = nextData.props.pageProps.content.hasPart.parts[0].hasPart.parts.filter((part) => part.type.includes('Article'));
 
-    const articles = $('._articles ._article')
-        .toArray()
-        .map((item) => {
-            item = $(item);
-            const title = item.find('h3').text();
-            item.find('h3').remove();
-            return {
-                title,
-                description: item.html(),
-                link,
-                guid: item.text(),
-            };
-        });
+    const renderHTML = (node) => {
+        let el;
+        switch (node.type) {
+            case 'tag':
+                el = $(`<${node.name}>`);
+                for (const name in node.attribs) {
+                    el.attr(name, node.attribs[name]);
+                }
+                if (node.children) {
+                    el.append(node.children.map(renderHTML));
+                }
+                return el;
+            case 'text':
+                return node.data;
+            default:
+                el = $('<div>');
+                el.append(node.map(renderHTML));
+                return el;
+        }
+    };
+
+    const gobbetPart = parts.shift();
+    const gobbets = gobbetPart.text.map((node, index) => {
+        const html = renderHTML(node).html();
+        return {
+            guid: `${gobbetPart.id}/${index}`,
+            link,
+            title: gobbetPart.headline,
+            pubDate: gobbetPart.datePublished,
+            description: html,
+        };
+    });
+
+    const articles = parts.map((part) => {
+        const image = part.image.main;
+        const imgNode = {
+            type: 'tag',
+            name: 'img',
+            attribs: { width: image.width, height: image.height, src: image.url.canonical },
+        };
+
+        const html = renderHTML([imgNode, ...part.text]).html();
+
+        return {
+            link,
+            guid: part.id,
+            title: part.headline,
+            pubDate: part.datePublished,
+            description: html,
+        };
+    });
 
     ctx.state.data = {
         title: $('head title').text(),
         link,
         description: $('meta[property="og:description"]').attr('content'),
+        language: 'en-gb',
         item: [...gobbets, ...articles],
     };
 };