feat: add abc news (#5776)

Co-authored-by: codefactor-io <support@codefactor.io>
Co-authored-by: Henry Wang <hi@henry.wang>

docs/en/traditional-media.md

@@ -4,6 +4,20 @@ pageClass: routes
 
 # News
 
+## ABC News
+
+### Site
+
+<Route author="nczitzk" example="/abc" path="/abc/:site?" :paramsDesc="['Site, see below']">
+
+Site
+
+| Just In | Politics | World | Business | Analysis | Sport | Science | Health | Arts | Fact Check | 中文新闻 | Berita Bahasa Indonesia | Tok Pisin |
+| - | - | - | - | - | - | - | - | - | - | - | - | - | - |
+| justin | politics | world | business | analysis-and-opinion | sport | science | health | arts-culture | factcheck | chinese | indonesian | tok-pisin |
+
+</Route>
+
 ## AP News
 
 ### Topics

docs/traditional-media.md

@@ -10,6 +10,20 @@ pageClass: routes
 
 <Route author="brilon" example="/21caijing/channel/readnumber" path="/21caijing/channel/:name" :paramsDesc="['频道名称，可在[https://m.21jingji.com/](https://m.21jingji.com/)页面URL中找到']"/>
 
+## ABC News
+
+### 子站
+
+<Route author="nczitzk" example="/abc/chinese" path="/abc/:site?" :paramsDesc="['子站，见下表']">
+
+子站
+
+| Just In | Politics | World | Business | Analysis             | Sport | Science | Health | Arts         | Fact Check | 中文新闻 | Berita Bahasa Indonesia | Tok Pisin |
+| ------- | -------- | ----- | -------- | -------------------- | ----- | ------- | ------ | ------------ | ---------- | -------- | ----------------------- | --------- |
+| justin  | politics | world | business | analysis-and-opinion | sport | science | health | arts-culture | factcheck  | chinese  | indonesian              | tok-pisin |
+
+</Route>
+
 ## AP News
 
 ### 话题

lib/router.js

@@ -3260,6 +3260,9 @@ router.get('/appsales/:caty?/:time?', require('./routes/appsales/index'));
 // CGTN
 router.get('/cgtn/top', require('./routes/cgtn/top'));
 
+// ABC News
+router.get('/abc/:site?', require('./routes/abc/index.js'));
+
 // 台湾中央通讯社
 router.get('/cna/:id?', require('./routes/cna/index'));
 

lib/routes/abc/index.js

@@ -0,0 +1,86 @@
+const url = require('url');
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+
+module.exports = async (ctx) => {
+    ctx.params.site = ctx.params.site || '';
+
+    const rootUrl = 'https://mobile.abc.net.au';
+    const currentUrl = `${rootUrl}/news/${ctx.params.site}`;
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+    const $ = cheerio.load(response.data);
+
+    $('._3zQMR, .ticker').remove();
+
+    const list = $('h3')
+        .slice(0, 10)
+        .map((_, item) => {
+            item = $(item);
+            return {
+                title: item.text(),
+                link: url.resolve(rootUrl, item.parents('a').attr('href') || item.find('a').attr('href')),
+            };
+        })
+        .get();
+
+    const items = await Promise.all(
+        list.map(
+            async (item) =>
+                await ctx.cache.tryGet(item.link, async () => {
+                    const detailResponse = await got({
+                        method: 'get',
+                        url: item.link,
+                    });
+                    const content = cheerio.load(detailResponse.data);
+
+                    content('.modal-content').remove();
+
+                    content('img').each(function () {
+                        content(this).attr('src', content(this).attr('data-src'));
+                    });
+
+                    let coverImage, contentBody;
+                    if (content('div[data-component="FeatureMedia"]').html() !== null) {
+                        coverImage = content('div[data-component="FeatureMedia"]').html();
+                    } else {
+                        coverImage = content('.view-hero-media').html();
+                    }
+                    if (content('#body').html() !== null) {
+                        contentBody = content('#body').html();
+                    } else {
+                        contentBody = content('.article-text').html();
+                    }
+
+                    const authorsArray = [];
+                    const authorsMatch = detailResponse.data.match(/author":(.*),"dateModified/);
+                    if (authorsMatch === null) {
+                        authorsArray.push(content('meta[name="DCTERMS.contributor"]').attr('content'));
+                    } else {
+                        const authors = JSON.parse(authorsMatch[1]);
+                        try {
+                            for (const author of authors) {
+                                authorsArray.push(author.name);
+                            }
+                        } catch (e) {
+                            authorsArray.push(authors.name);
+                        }
+                    }
+
+                    item.description = coverImage + contentBody;
+                    item.pubDate = new Date(content('meta[property="article:published_time"]').attr('content')).toUTCString();
+                    item.author = authorsArray.join(', ');
+
+                    return item;
+                })
+        )
+    );
+
+    ctx.state.data = {
+        title: $('title').text(),
+        link: currentUrl,
+        item: items,
+    };
+};