feat: add AEON & SA 60-Second Science (#3213)

docs/multimedia.md

@@ -4,6 +4,12 @@ pageClass: routes
 
 # 音视频
 
+## 60-Second Science - Scientific American
+
+### Transcript
+
+<Route author="emdoe" example="/60s-science/transcript" path="/60s-science/transcript"/>
+
 ## 99% Invisible
 
 ### Transcript

docs/new-media.md

@@ -27,6 +27,20 @@ pageClass: routes
 
 </Route>
 
+## AEON
+
+<Route author="emdoe" example="/aeon/ideas" path="/aeon/:category" :paramsDesc="['类别']"></Route>
+
+支持以文体分类：
+| Ideas | Essays | Videos |
+| ----- | ------ | ------ |
+| ideas | essays | videos |
+
+同样支持以话题分类：
+| Culture | Philosophy | Psychology | Society | Science |
+| ------- | ---------- | ---------- | ------- | ------- |
+| culture | philosophy | psychology | society | science |
+
 ## BOF
 
 ### 首页

lib/router.js

@@ -782,6 +782,9 @@ router.get('/laosiji/hot', require('./routes/laosiji/hot'));
 router.get('/laosiji/feed', require('./routes/laosiji/feed'));
 router.get('/laosiji/hotshow/:id', require('./routes/laosiji/hotshow'));
 
+// Scientific American 60-Second Science
+router.get('/60s-science/transcript', require('./routes/60s-science/transcript'));
+
 // 99% Invisible
 router.get('/99percentinvisible/transcript', require('./routes/99percentinvisible/transcript'));
 
@@ -1422,6 +1425,9 @@ router.get('/ningmeng/song', require('./routes/ningmeng/song'));
 // 紫竹张
 router.get('/zzz', require('./routes/zzz/index'));
 
+// AEON
+router.get('/aeon/:cid', require('./routes/aeon/category'));
+
 // AlgoCasts
 router.get('/algocasts', require('./routes/algocasts/all'));
 

lib/routes/60s-science/transcript.js

@@ -0,0 +1,48 @@
+const cheerio = require('cheerio');
+const got = require('@/utils/got');
+
+module.exports = async (ctx) => {
+    const url = `https://www.scientificamerican.com/podcast/60-second-science/`;
+
+    const res = await got.get(url);
+    const $ = cheerio.load(res.data);
+    const list = $('.underlined_text.t_small').get();
+
+    const out = await Promise.all(
+        list.map(async (item) => {
+            const $ = cheerio.load(item);
+            const title = $(item).attr('aria-label');
+            const address = $(item).attr('href');
+            const cache = await ctx.cache.get(address);
+            if (cache) {
+                return Promise.resolve(JSON.parse(cache));
+            }
+            const res = await got.get(address);
+            const capture = cheerio.load(res.data);
+
+            const tStr = '.article-header__divider > ul > li > span:nth-child(2)';
+            const time = capture(tStr).text();
+            const aStr = '.article-header__divider > ul > li > span:nth-child(1) > span';
+            const author = capture(aStr).text();
+
+            capture('.transcript__close').remove();
+            const intro = capture('.article-media__object').html() + '<br><br>';
+            const contents = intro + capture('.transcript__inner').html();
+            const single = {
+                title,
+                description: contents,
+                link: address,
+                guid: address,
+                author: author,
+                pubDate: new Date(time).toUTCString(),
+            };
+            ctx.cache.set(address, JSON.stringify(single));
+            return Promise.resolve(single);
+        })
+    );
+    ctx.state.data = {
+        title: 'Transcripts of SA 60-Second Science',
+        link: url,
+        item: out,
+    };
+};

lib/routes/aeon/category.js

@@ -0,0 +1,86 @@
+const cheerio = require('cheerio');
+const got = require('@/utils/got');
+
+module.exports = async (ctx) => {
+    const url = `https://aeon.co/${ctx.params.cid}`;
+
+    const res = await got.get(url);
+    const $ = cheerio.load(res.data);
+    const list = $('.article-card').get();
+
+    const out = await Promise.all(
+        list.map(async (item) => {
+            const $ = cheerio.load(item);
+            const title = $('.article-card__title').text();
+            const partial = $('.article-card__title').attr('href');
+            const address = `https://aeon.co${partial}`;
+            const author = $('.article-card__authors__inner').text();
+
+            const cache = await ctx.cache.get(address);
+            if (cache) {
+                return Promise.resolve(JSON.parse(cache));
+            }
+
+            const res = await got.get(address);
+            const capture = cheerio.load(res.data);
+            const time = capture('.article__date.published').text();
+
+            capture('div.newsletter-signup').remove();
+            capture('div.pullquote').remove();
+            capture('div.article__related').remove();
+            capture('div.article__shares').remove();
+
+            let bio;
+            const authorInfo = '.article__body__author-details > p:nth-child(2)';
+            if (capture(authorInfo).html() !== null) {
+                bio =
+                    `<br><br><br><br><hr><br><b>` +
+                    author.trim() +
+                    '</b> ' +
+                    capture(authorInfo)
+                        .html()
+                        .trim() +
+                    `<p><hr>`;
+                capture('.article__inline-sidebar').remove();
+            } else {
+                bio = '';
+            }
+
+            let banner;
+            if (capture('.article-card__image-wrap').html() !== null) {
+                const imgSty = capture('figure.responsive-image').attr('style');
+                const imgLink = imgSty.match(/url\(["']?([^"']*)["']?\)/)[1];
+                banner = `<img src="${imgLink}">`;
+            }
+            if (capture('.article__image__wrapper').html() !== null) {
+                banner = '';
+                capture('img').each((index, item) => {
+                    item = $(item);
+                    item.removeAttr('alt');
+                });
+                capture('.display-article').remove;
+            }
+            if (capture('.video__embed').html() !== null) {
+                banner = capture('div.video__embed').html();
+                capture('.article__header__subtitle').remove();
+            }
+
+            const contents = banner + capture('.article__body__content').html() + bio;
+            const single = {
+                title,
+                author: author,
+                pubDate: new Date(time).toUTCString(),
+                description: contents,
+                link: address,
+                guid: address,
+            };
+            ctx.cache.set(address, JSON.stringify(single));
+            return Promise.resolve(single);
+        })
+    );
+    ctx.state.data = {
+        title: $('title').text(),
+        link: url,
+        item: out,
+    };
+};