feat(route): add Distill (#8491)

docs/en/programming.md

@@ -16,6 +16,12 @@ pageClass: routes
 
 <RouteEn author="fengkx" example="/cve/search/PostgreSQL" path="/cve/search/:keyword" :paramsDesc="['keyword']" />
 
+## Distill
+
+### Latest
+
+<RouteEn author="nczitzk" example="/distill" path="/distill"/>
+
 ## GitHub
 
 ::: tip

docs/programming.md

@@ -58,6 +58,12 @@ pageClass: routes
 
 <Route author="nczitzk" example="/deeplearningai/thebatch" path="/deeplearningai/thebatch"/>
 
+## Distill
+
+### Latest
+
+<Route author="nczitzk" example="/distill" path="/distill"/>
+
 ## Dockone
 
 ### 周报

lib/v2/distill/index.js

@@ -0,0 +1,64 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const rootUrl = 'https://distill.pub';
+
+    const response = await got({
+        method: 'get',
+        url: rootUrl,
+    });
+
+    const $ = cheerio.load(response.data);
+
+    let items = $('.post-preview')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+
+            return {
+                title: item.find('.title').text(),
+                link: `${rootUrl}/${item.children('a').attr('href')}`,
+            };
+        });
+
+    items = await Promise.all(
+        items.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const detailResponse = await got({
+                    method: 'get',
+                    url: item.link,
+                });
+
+                const content = cheerio.load(detailResponse.data);
+
+                content('d-contents').remove();
+
+                content('img').each(function () {
+                    content(this).attr('src', `${item.link}/${content(this).attr('src')}`);
+                });
+
+                item.doi = content('meta[name="citation_doi"]').attr('content');
+                item.pubDate = parseDate(content('meta[property="article:published"]').attr('content'));
+                item.description = content('d-article')
+                    .children()
+                    .toArray()
+                    .map((c) => content(c).html())
+                    .join('');
+                item.author = content('meta[property="article:author"]')
+                    .toArray()
+                    .map((author) => content(author).attr('content'))
+                    .join(', ');
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: $('title').text(),
+        link: rootUrl,
+        item: items,
+    };
+};

lib/v2/distill/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/': ['nczitzk'],
+};

lib/v2/distill/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'distill.pub': {
+        _name: 'Distill',
+        '.': [
+            {
+                title: 'Latest',
+                docs: 'https://docs.rsshub.app/programming.html#distill',
+                source: ['/'],
+                target: '/distill',
+            },
+        ],
+    },
+};

lib/v2/distill/router.js

@@ -0,0 +1,3 @@
+module.exports = function (router) {
+    router.get('/', require('./index'));
+};