feat(route): the atlantic (DIYgod#12092)

* feat(route): the atlantic * fix: docs * chore: discard async-pool * fix: docs * feat: utils update * fix: typo * Update lib/v2/theatlantic/utils.js * Update lib/v2/theatlantic/templates/article-description.art ---------
benzking · Mar 14, 2023 · afc93ac · afc93ac
1 parent b33996b
commit afc93ac
Show file tree

Hide file tree

Showing 8 changed files with 152 additions and 0 deletions.
diff --git a/docs/en/traditional-media.md b/docs/en/traditional-media.md
@@ -574,6 +574,20 @@ Language
 
 <RouteEn author="TonyRL" example="/taiwannews/hot" path="/taiwannews/hot/:lang?" :paramsDesc="['Language, `en` or `zh`, `en` by default']" radar="1" rssbud="1"/>
 
+## The Atlantic
+
+### News
+
+<RouteEn author="NavePnow" example="/theatlantic/latest" path="/theatlantic/:category" :paramsDesc="['category, see below']">
+
+| Popular      | Latest | Politics | Technology | Business |
+| ------------ | ------ | -------- | ---------- | -------- |
+| most-popular | latest | politics | technology | business |
+
+More categories (except photo) can be found within the navigation bar at <https://www.theatlantic.com/>
+
+</RouteEn>
+
 ## The Economist
 
 ### Category

diff --git a/docs/traditional-media.md b/docs/traditional-media.md
@@ -468,6 +468,20 @@ Solidot 提供的 feed:
 
 </Route>
 
+## The Atlantic
+
+### News
+
+<Route author="NavePnow" example="/theatlantic/latest" path="/theatlantic/:category" :paramsDesc="['分类, 见下表']">
+
+| Popular      | Latest | Politics | Technology | Business |
+| ------------ | ------ | -------- | ---------- | -------- |
+| most-popular | latest | politics | technology | business |
+
+More categories (except photo) can be found within the navigation bar at <https://www.theatlantic.com/>
+
+</Route>
+
 ## The Economist
 
 ### 分类

diff --git a/lib/v2/theatlantic/maintainer.js b/lib/v2/theatlantic/maintainer.js
@@ -0,0 +1,3 @@
+module.exports = {
+    '/:category': ['NavePnow'],
+};
diff --git a/lib/v2/theatlantic/news.js b/lib/v2/theatlantic/news.js
@@ -0,0 +1,33 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { getArticleDetails } = require('./utils');
+module.exports = async (ctx) => {
+    const host = 'https://www.theatlantic.com';
+    const category = ctx.params.category;
+    const url = `${host}/${category}/`;
+    const response = await got({
+        method: 'get',
+        url,
+    });
+    const $ = cheerio.load(response.data);
+    const contents = JSON.parse($('script#__NEXT_DATA__').text()).props.pageProps.urqlState;
+    const firstKey = Object.keys(contents)[0];
+    const data = JSON.parse(contents[firstKey].data);
+    let list = Object.values(data)[0].river.edges;
+    list = list.filter((item) => !item.node.url.startsWith('https://www.theatlantic.com/photo'));
+    list = list.map((item) => {
+        const data = {};
+        data.title = item.node.title;
+        data.link = item.node.url;
+        data.pubDate = item.node.datePublished;
+        return data;
+    });
+    const items = await getArticleDetails(list, ctx);
+
+    ctx.state.data = {
+        title: `The Atlantic - ${category.toUpperCase()}`,
+        link: url,
+        description: `The Atlantic - ${category.toUpperCase()}`,
+        item: items,
+    };
+};
diff --git a/lib/v2/theatlantic/radar.js b/lib/v2/theatlantic/radar.js
@@ -0,0 +1,13 @@
+module.exports = {
+    'theatlantic.com': {
+        _name: 'The Atlantic',
+        www: [
+            {
+                title: '新闻',
+                docs: 'https://docs.rsshub.app/traditional-media.html#the-atlantic',
+                source: '/:category',
+                target: '/theatlantic/:category',
+            },
+        ],
+    },
+};
diff --git a/lib/v2/theatlantic/router.js b/lib/v2/theatlantic/router.js
@@ -0,0 +1,3 @@
+module.exports = (router) => {
+    router.get('/:category', require('./news'));
+};
diff --git a/lib/v2/theatlantic/templates/article-description.art b/lib/v2/theatlantic/templates/article-description.art
@@ -0,0 +1,19 @@
+<div>
+    {{@ item.caption}}
+    <br>
+    {{ if item.imgUrl }}
+    <figure>
+        <img src="{{item.imgUrl}}" alt="{{item.imgAlt}}">
+        <figcaption>{{item.imgCaption}}</figcaption>
+    </figure>
+    {{ /if }}
+    {{each item.content}}
+        {{@ $value.innerHtml}}
+        {{if !$value.tagName}}
+        <br>
+        <br>
+        {{/if}}
+    {{/each}}
+
+</div>
+
diff --git a/lib/v2/theatlantic/utils.js b/lib/v2/theatlantic/utils.js
@@ -0,0 +1,53 @@
+const cheerio = require('cheerio');
+const got = require('@/utils/got');
+const { parseDate } = require('@/utils/parse-date');
+const { art } = require('@/utils/render');
+const path = require('path');
+const UA = require('@/utils/rand-user-agent')({ browser: 'chrome', os: 'android', device: 'mobile' });
+
+const getArticleDetails = async (items, ctx) => {
+    const list = await Promise.all(
+        items.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const url = item.link;
+                const response = await got({
+                    url,
+                    method: 'get',
+                    headers: {
+                        'User-Agent': UA,
+                    },
+                });
+                const html = response.data;
+                const $ = cheerio.load(html);
+                let data = JSON.parse($('script#__NEXT_DATA__').text());
+
+                const list = data.props.pageProps.urqlState;
+                const keyWithContent = Object.keys(list).filter((key) => list[key].data.includes('content'));
+                data = JSON.parse(list[keyWithContent].data).article;
+                item.category = data.categories.map((category) => category.slug);
+                data.channels.forEach((channel) => {
+                    item.category.push(channel.slug);
+                });
+                item.content = data.content.filter((item) => item.innerHtml !== undefined && item.innerHtml !== '');
+                item.caption = data.dek;
+                item.imgUrl = data.leadArt.image?.url;
+                item.imgAlt = data.leadArt.image?.altText;
+                item.imgCaption = data.leadArt.image?.attributionText;
+                item.description = art(path.join(__dirname, 'templates/article-description.art'), {
+                    item,
+                });
+                return {
+                    title: item.title,
+                    pubDate: parseDate(item.pubDate),
+                    link: item.link,
+                    description: item.description,
+                };
+            })
+        )
+    );
+    return list;
+};
+
+module.exports = {
+    getArticleDetails,
+};