fix(route): Hacker News (#8420)

docs/en/programming.md

@@ -137,26 +137,21 @@ For instance, the `/github/topics/framework/l=php&o=desc&s=stars` route will gen
 
 ### Section
 
-<RouteEn author="cf020031308" example="/hackernews/best/comments" path="/hackernews/:section/:type?" :paramsDesc="['Section', 'Link type']">
+<RouteEn author="cf020031308 nczitzk" example="/hackernews" path="/hackernews/:section?/:type?" :paramsDesc="['Section, see below, index by default', 'Link, see below, sources by default']">
 
-Website: https://news.ycombinator.com/
+Section
 
-| Section | section                             |
-| ------- | ----------------------------------- |
-| index   | https://news.ycombinator.com/       |
-| new     | https://news.ycombinator.com/newest |
-| past    | https://news.ycombinator.com/front  |
-| ask     | https://news.ycombinator.com/ask    |
-| show    | https://news.ycombinator.com/show   |
-| jobs    | https://news.ycombinator.com/jobs   |
-| best    | https://news.ycombinator.com/best   |
+| homepage                              | new                                           | past                                        | comments                                                | ask                                     | show                                      | jobs                                      | best                                      |
+| ------------------------------------- | --------------------------------------------- | ------------------------------------------- | ------------------------------------------------------- | --------------------------------------- | ----------------------------------------- | ----------------------------------------- | ----------------------------------------- |
+| [index](https://news.ycombinator.com) | [newest](https://news.ycombinator.com/newest) | [front](https://news.ycombinator.com/front) | [newcomments](https://news.ycombinator.com/newcomments) | [ask](https://news.ycombinator.com/ask) | [show](https://news.ycombinator.com/show) | [jobs](https://news.ycombinator.com/jobs) | [best](https://news.ycombinator.com/best) |
 
-> Official RSS：https://news.ycombinator.com/rss is same as `index` section
+Items link to
 
-| Link type | type                           |
-| --------- | ------------------------------ |
-| story     | Deault, link to shared address |
-| comments  | Link to Hacker News address    |
+| Source addresses shared by users | Comments on Hacker News |
+| -------------------------------- | ----------------------- |
+| sources                          | comments                |
+
+> Default RSS by the website: <https://news.ycombinator.com/rss>, same as `index` section, should be the first choice.
 
 </RouteEn>
 

docs/programming.md

@@ -219,28 +219,23 @@ GitHub 官方也提供了一些 RSS:
 
 ## Hacker News
 
-### 分类
+### 分区
 
-<Route author="cf020031308" example="/hackernews/best/comments" path="/hackernews/:section/:type?" :paramsDesc="['内容分区', '链接类型（可不填）']">
+<Route author="cf020031308 nczitzk" example="/hackernews" path="/hackernews/:section?/:type?" :paramsDesc="['内容分区，见下表，默认为 index', '链接类型，见下表，默认为 sources']">
 
-网站地址：<https://news.ycombinator.com/>
+内容分区
 
-| 内容分区 | section                               |
-| -------- | ------------------------------------- |
-| index    | <https://news.ycombinator.com/>       |
-| new      | <https://news.ycombinator.com/newest> |
-| past     | <https://news.ycombinator.com/front>  |
-| ask      | <https://news.ycombinator.com/ask>    |
-| show     | <https://news.ycombinator.com/show>   |
-| jobs     | <https://news.ycombinator.com/jobs>   |
-| best     | <https://news.ycombinator.com/best>   |
+| homepage                              | new                                           | past                                        | comments                                                | ask                                     | show                                      | jobs                                      | best                                      |
+| ------------------------------------- | --------------------------------------------- | ------------------------------------------- | ------------------------------------------------------- | --------------------------------------- | ----------------------------------------- | ----------------------------------------- | ----------------------------------------- |
+| [index](https://news.ycombinator.com) | [newest](https://news.ycombinator.com/newest) | [front](https://news.ycombinator.com/front) | [newcomments](https://news.ycombinator.com/newcomments) | [ask](https://news.ycombinator.com/ask) | [show](https://news.ycombinator.com/show) | [jobs](https://news.ycombinator.com/jobs) | [best](https://news.ycombinator.com/best) |
 
-> 网站有默认的 RSS：<https://news.ycombinator.com/rss> 内容同 index，应优先考虑
+条目指向链接类型
 
-| 链接类型 | type                          |
-| -------- | ----------------------------- |
-| story    | 默认值，链向用户分享的地址    |
-| comments | 链向 Hacker News 上的讨论页面 |
+| 用户分享的来源地址 | Hacker News 上的讨论页面 |
+| ------------------ | ------------------------ |
+| sources            | comments                 |
+
+> 网站有默认的 RSS：<https://news.ycombinator.com/rss> 内容同 homepage，应优先考虑。
 
 </Route>
 

lib/router.js

@@ -1434,7 +1434,7 @@ router.get('/aisixiang/ranking/:type?/:range?', lazyloadRouteHandler('./routes/a
 router.get('/aisixiang/thinktank/:name/:type?', lazyloadRouteHandler('./routes/aisixiang/thinktank'));
 
 // Hacker News
-router.get('/hackernews/:section/:type?', lazyloadRouteHandler('./routes/hackernews/story'));
+// router.get('/hackernews/:section/:type?', lazyloadRouteHandler('./routes/hackernews/story'));
 
 // LeetCode
 router.get('/leetcode/articles', lazyloadRouteHandler('./routes/leetcode/articles'));

lib/routes/hackernews/story.js

@@ -1,60 +0,0 @@
-const got = require('@/utils/got');
-const cheerio = require('cheerio');
-
-const host = 'https://news.ycombinator.com';
-const paths = {
-    index: '/',
-    new: '/newest',
-    past: '/front',
-    ask: '/ask',
-    show: '/show',
-    jobs: '/jobs',
-    best: '/best',
-};
-
-module.exports = async (ctx) => {
-    const section = ctx.params.section;
-    const showStory = ctx.params.type !== 'comments';
-    const path = paths[section];
-    const url = `${host}${path}`;
-    const response = await got.get(url);
-    const $ = cheerio.load(response.data);
-
-    const items = $('a.storylink')
-        .map(function () {
-            return {
-                title: $(this).text(),
-                link: $(this).attr('href'),
-                author: 'Hacker News',
-                description: '',
-            };
-        })
-        .get();
-
-    $('a.hnuser').each(function (i) {
-        items[i].author = $(this).text();
-    });
-
-    $('tr.athing').each(function (i) {
-        const commURL = `${host}/item?id=` + $(this).attr('id');
-        const item = items[i];
-
-        if (showStory) {
-            item.description = `Comments: <a href="${commURL}"> ${commURL} </a>`;
-        } else {
-            item.description = `Link: <a href="${item.link}"> ${item.title} </a>`;
-            item.link = commURL;
-        }
-    });
-
-    let title = `Hacker News: ${section}`;
-    if (!showStory) {
-        title += '/comments';
-    }
-
-    ctx.state.data = {
-        title,
-        link: url,
-        item: items,
-    };
-};

lib/v2/hackernews/index.js

@@ -0,0 +1,97 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const section = ctx.params.section ?? 'index';
+    const type = ctx.params.type ?? 'sources';
+
+    const rootUrl = 'https://news.ycombinator.com';
+    const currentUrl = `${rootUrl}${section === 'index' ? '' : `/${section}`}`;
+
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = cheerio.load(response.data);
+
+    const list = $('.athing')
+        .slice(0, ctx.query.limit ? parseInt(ctx.query.limit) : 30)
+        .map((_, thing) => {
+            thing = $(thing);
+
+            const item = {};
+
+            item.guid = thing.attr('id');
+            item.title = thing.find('.titlelink').text();
+            item.category = thing.find('.sitestr').text();
+            item.author = thing.next().find('.hnuser').text();
+            item.pubDate = parseDate(thing.next().attr('title'));
+
+            item.link = `${rootUrl}/item?id=${item.guid}`;
+            item.origin = thing.find('.titlelink').attr('href');
+
+            item.comments = thing.next().find('a').last().text().split(' comment')[0];
+            item.guid = type === 'sources' ? item.guid : `${item.guid}${item.comments === 'discuss' ? '' : `-${item.comments}`}`;
+
+            item.description = `<a href="${item.link}">Comments on Hacker News</a> | <a href="${item.origin}">Source</a>`;
+
+            return item;
+        })
+        .get();
+
+    const items = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.guid, async () => {
+                if (item.comments !== 'discuss' && type !== 'sources') {
+                    const detailResponse = await got({
+                        method: 'get',
+                        url: item.link,
+                    });
+
+                    const content = cheerio.load(detailResponse.data);
+
+                    content('.reply').remove();
+
+                    item.description = '';
+
+                    content('.comtr').each(function () {
+                        const author = content(this).find('.hnuser');
+                        const comment = content(this).find('.commtext');
+
+                        item.description +=
+                            `<div><div><small><a href="${rootUrl}/${author.attr('href')}">${author.text()}</a></small>` +
+                            `&nbsp&nbsp<small><a href="${rootUrl}/item?id=${content(this).attr('id')}">` +
+                            `${content(this).find('.age').attr('title')}</a></small></div>`;
+
+                        const commentText = comment.clone();
+
+                        commentText.find('p').remove();
+                        commentText.html(`<p>${commentText.text()}</p>`);
+                        commentText.append(
+                            comment
+                                .find('p')
+                                .toArray()
+                                .map((p) => `<p>${content(p).html()}</p>`)
+                        );
+
+                        item.description += `<div>${commentText.html()}</div></div>`;
+                    });
+                }
+
+                item.link = type === 'sources' ? item.origin : item.link;
+
+                delete item.origin;
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: $('title').text(),
+        link: currentUrl,
+        item: items,
+    };
+};

lib/v2/hackernews/maintainer.js

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

lib/v2/hackernews/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'ycombinator.com': {
+        _name: 'Hacker News',
+        '.': [
+            {
+                title: 'Section',
+                docs: 'https://docs.rsshub.app/programming.html#hacker-news',
+                source: ['/:section', '/'],
+                target: '/hackernews/:section?/:type?',
+            },
+        ],
+    },
+};

lib/v2/hackernews/router.js

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