feat: add Soomal (DIYgod#5396)

Co-authored-by: Henry Wang <[email protected]>

docs/en/new-media.md

@@ -13,7 +13,7 @@ pageClass: routes
 Supported sub-sites：
 | 9To5Mac | 9To5Google | 9To5Toys |
 | ------- | ---------- | -------- |
-| Mac | Google | Toys |
+| Mac     | Google     | Toys     |
 
 </RouteEn>
 
@@ -152,6 +152,32 @@ Compared to the official one, this feed:
 
 <RouteEn author="kt286" example="/sixthtone/news" path="/sixthtone/news"/>
 
+## Soomal
+
+### 话题
+
+<RouteEn author="zoenglinghou" example="/soomal/topics/Phone/en" path="/soomal/topics/:category/:language?" :paramsDesc="['Topic, found on the top menu bar', 'locale, default to simplified Chinese']">
+
+- Available languages：
+
+| Simplified Chinese | Traditional Chinese | English |
+| ------------------ | ------------------- | ------- |
+| zh                 | zh_tw               | en      |
+
+- Available topics by locale：
+
+| Languages           |          |       |          |          |          |
+| ------------------- | -------- | ----- | -------- | -------- | -------- |
+| Simplified Chinese  | 最新文章 | 科普  | 测评报告 | 发烧入门 | 摄影入门 | 古典音乐入门 |
+| Traditional Chinese | 最新文章 | 科普  | 測評報告 | 發燒入門 | 攝影入門 | 古典音樂入門 |
+| English             | Phone    | Audio | Album    | Review   |
+
+- Soomal offers official RSS subscriptions
+  - Soomal website：[http://www.soomal.com/doc/101.rss.xml](http://www.soomal.com/doc/101.rss.xml)
+  - Soomal forum and comments：[http://www.soomal.com/bbs/101.rss.xml](http://www.soomal.com/bbs/101.rss.xml)
+
+</RouteEn>
+
 ## The Verge
 
 <RouteEn author="HenryQW" example="/verge" path="/verge">

docs/new-media.md

@@ -316,6 +316,32 @@ pageClass: routes
 
 <Route author="nczitzk" example="/socialbeta/hunt" path="/socialbeta/hunt"/>
 
+## Soomal
+
+### 话题
+
+<Route author="zoenglinghou" example="/soomal/topics/最新文章" path="/soomal/topics/:category/:language?" :paramsDesc="['话题，可在顶部菜单找到对应名称', '语言，默认为简体中文']">
+
+-   可选语言：
+
+| 简体中文 | 正体中文 | 英语 |
+| -------- | -------- | ---- |
+| zh       | zh_tw    | en   |
+
+-   可选话题（按语言分类）：
+
+| 语言     |          |       |          |          |          |              |
+| -------- | -------- | ----- | -------- | -------- | -------- | ------------ |
+| 简体中文 | 最新文章 | 科普  | 测评报告 | 发烧入门 | 摄影入门 | 古典音乐入门 |
+| 正体中文 | 最新文章 | 科普  | 測評報告 | 發燒入門 | 攝影入門 | 古典音樂入門 |
+| 英语     | Phone    | Audio | Album    | Review   |          |              |
+
+-   Soomal 提供官方 RSS 订阅
+    -   Soomal 网站更新：<http://www.soomal.com/doc/101.rss.xml>
+    -   Soomal 论坛与留言系统的更新：<http://www.soomal.com/bbs/101.rss.xml>
+
+</Route>
+
 ## The Verge
 
 ### The Verge

lib/router.js

@@ -3088,6 +3088,9 @@ router.get('/souyun/today', require('./routes/souyun/today'));
 // 生物谷
 router.get('/bioon/latest', require('./routes/bioon/latest'));
 
+// soomal
+router.get('/soomal/topics/:category/:language?', require('./routes/soomal/topics'));
+
 // NASA
 router.get('/nasa/apod', require('./routes/nasa/apod'));
 router.get('/nasa/apod-ncku', require('./routes/nasa/apod-ncku'));

lib/routes/soomal/topics.js

@@ -0,0 +1,130 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const url = require('url');
+const iconv = require('iconv-lite');
+
+module.exports = async (ctx) => {
+    const category = ctx.params.category;
+    const lang = ctx.params.language;
+
+    let index_url;
+    let character_set;
+    switch (lang) {
+        case 'zh_tw':
+            index_url = 'http://www.soomal.com/doc/index201000_0001_00.htm';
+            character_set = 'big5';
+            break;
+        case 'en':
+            index_url = 'http://eng.soomal.com/edoc/index101000_0001_00.htm';
+            character_set = 'ISO-8859-5';
+            break;
+        default:
+            index_url = 'http://www.soomal.com/doc/index101000_0001_00.htm';
+            character_set = 'gbk';
+            break;
+    }
+
+    let category_list;
+    let url_list;
+
+    const category_url = await ctx.cache.tryGet([lang, category], async () => {
+        const front_page = await got.get(index_url, {
+            responseType: 'buffer',
+        });
+
+        const front_data = iconv.decode(front_page.data, character_set);
+
+        const $ = cheerio.load(front_data);
+        if (lang !== 'en') {
+            const container_list = $('div.container.cf').slice(0, 2);
+            category_list = container_list
+                .find('ul.ul-reset')
+                .find('h3')
+                .children()
+                .get()
+                .map((item) => $(item).text());
+            url_list = container_list
+                .find('ul.ul-reset')
+                .find('h3')
+                .children()
+                .get()
+                .map((item) => $(item).attr('href'));
+        } else {
+            category_list = $('div.menu')
+                .find('a')
+                .get()
+                .map((item) => $(item).text());
+            url_list = $('div.menu')
+                .find('a')
+                .get()
+                .map((item) => $(item).attr('href'));
+        }
+
+        return url.resolve(index_url, url_list[category_list.indexOf(category)]);
+    });
+
+    const response = await got.get(category_url, {
+        responseType: 'buffer',
+    });
+
+    const data = iconv.decode(response.data, character_set);
+    const $ = cheerio.load(data);
+
+    const list = $('div.item')
+        .find('div.title')
+        .children('a')
+        .get()
+        .map((item) => url.resolve(category_url, $(item).attr('href')));
+
+    const out = await Promise.all(
+        list.map(async (link) => {
+            const [title, author, pubDate, description] = await ctx.cache.tryGet(link, async () => {
+                const response = await got.get(link, {
+                    responseType: 'buffer',
+                });
+
+                const data = iconv.decode(response.data, character_set);
+                const $ = cheerio.load(data);
+
+                let [title, author, pubDate, description] = '';
+                if (lang !== 'en') {
+                    title = $('div.titlebox').find('div.title').text();
+                    author = $('div.titlebox').find('div.info').children('a').text();
+                    pubDate = Date.parse(
+                        $('div.titlebox')
+                            .find('div.info')
+                            .text()
+                            .match(/\d{4}([.\-/ ])\d{2}\1\d{2}\s\d{2}[:]\d{2}[:]\d{2}/)[0]
+                    );
+
+                    $('div.toolbar,div.grade').remove();
+                    description = $('div.Doc').html();
+                } else {
+                    title = $('div.titlebox').find('div.title').text();
+                    author = $('div.titlebox').find('div.infobox').children('a').text();
+                    pubDate = new Date();
+
+                    $('div.grade').remove();
+                    description = $('div.textbox').html();
+                }
+
+                return [title, author, pubDate, description];
+            });
+
+            const item = {
+                title: title,
+                description: description,
+                pubDate: pubDate,
+                link: link,
+                author: author,
+            };
+            return Promise.resolve(item);
+        })
+    );
+
+    ctx.state.data = {
+        title: 'Soomal.com - ' + category,
+        link: category_url,
+        item: out,
+    };
+};