feat(route): 增加大数据专家委员会和中国人工智能学会新闻 (#11510)

* 增加大数据专家委员会和中国人工智能学会新闻

* 修复问题

* fix: add docs for `caai`

fix: maintainer does not with match with router

* fix: add docs for `ccf/tfbd`

fix: use raw output for `art-template`

fix: standardise use of `got`
not using `got()` in one place and `got.get()` in another place

* refactor: migrate to v2

---------

docs/new-media.md

@@ -4509,18 +4509,6 @@ wechat-feeds 来源[已停止更新](https://github.com/hellodword/wechat-feeds/
 
 </Route>
 
-## 中国计算机学会
-
-### 新闻
-
-<Route author="nczitzk" example="/ccf/news" path="/ccf/news/:category?" :paramsDesc="['分类，见下表，默认为 CCF 新闻']">
-
-| CCF 新闻     | CCF 聚焦 | ACM 信息   |
-| ---------- | ------ | -------- |
-| Media_list | Focus  | ACM_News |
-
-</Route>
-
 ## 中国科学院青年创新促进会
 
 ### 最新博文

docs/study.md

@@ -342,18 +342,6 @@ path="/ctfhub/upcoming/:limit?"
 
 <Route author="xfangbao" example="/jijitang/article/latest" path="/jijitang/article/:id" :paramsDesc="['类别，latest 或者 recommand']"/>
 
-## 计算机视觉专委
-
-### 学术动态 - 分类
-
-<Route author="elxy" example="/ccfcv/xsdt/xsqy" path="/ccfcv/:channel/:category" :paramsDesc="['频道，仅支持 `xsdt`', '分类，见下表，亦可在网站 url 里找到']">
-
-| 学术前沿 | 热点征文 | 学术会议 |
-| ---- | ---- | ---- |
-| xsqy | rdzw | xshy |
-
-</Route>
-
 ## 金山词霸
 
 ### 每日一句
@@ -659,12 +647,44 @@ path="/ctfhub/upcoming/:limit?"
 
 <Route author="xyqfer" example="/icourse163/newest" path="/icourse163/newest" />
 
+## 中国计算机学会
+
+### 新闻
+
+<Route author="nczitzk" example="/ccf/news" path="/ccf/news/:category?" :paramsDesc="['分类，见下表，默认为 CCF 新闻']">
+
+| CCF 新闻     | CCF 聚焦 | ACM 信息   |
+| ---------- | ------ | -------- |
+| Media_list | Focus  | ACM_News |
+
+</Route>
+
+### 计算机视觉专委会 - 学术动态 - 分类
+
+<Route author="elxy" example="/ccf/ccfcv/xsdt/xsqy" path="/ccf/ccfcv/:channel/:category" :paramsDesc="['频道，仅支持 `xsdt`', '分类，见下表，亦可在网站 url 里找到']">
+
+| 学术前沿 | 热点征文 | 学术会议 |
+| ---- | ---- | ---- |
+| xsqy | rdzw | xshy |
+
+</Route>
+
+### 大数据专家委员会
+
+<Route author="tudou027" example="/ccf/tfbd/xwdt/tzgg" path="/ccf/tfbd/:caty/:id" :paramsDesc="['主分类，可在 URL 找到', '子分类，可在 URL 找到']" radar="1"/>
+
 ## 中国留学网
 
 ### 通知公告
 
 <Route author="nczitzk" example="/cscse/tzgg" path="/cscse/tzgg"/>
 
+## 中国人工智能学会
+
+### 学会动态
+
+<Route author="tudou027" example="/caai/45" path="/caai/:caty" :paramsDesc="['分类 ID，可在 URL 找到']" radar="1"/>
+
 ## 中国人事考试网
 
 ### 通知公告

lib/router.js

@@ -3541,7 +3541,7 @@ router.get('/vgn/:platform?', lazyloadRouteHandler('./routes/vgn/index'));
 router.get('/iea/:category?', lazyloadRouteHandler('./routes/iea/index'));
 
 // 中国计算机学会
-router.get('/ccf/news/:category?', lazyloadRouteHandler('./routes/ccf/news'));
+// router.get('/ccf/news/:category?', lazyloadRouteHandler('./routes/ccf/news'));
 
 // The Brain
 router.get('/thebrain/:category?', lazyloadRouteHandler('./routes/thebrain/blog'));

lib/v2/caai/index.js

@@ -0,0 +1,22 @@
+const utils = require('./utils');
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+
+module.exports = async (ctx) => {
+    const base = utils.urlBase(ctx.params.caty);
+    const res = await got(base);
+    const info = utils.fetchAllArticles(res.data);
+    const $ = cheerio.load(res.data);
+
+    const details = await Promise.all(info.map((e) => utils.detailPage(e, ctx.cache)));
+
+    ctx.state.json = {
+        info,
+    };
+
+    ctx.state.data = {
+        title: '中国人工智能学会 - ' + $('.article-list h1').text(),
+        link: base,
+        item: details,
+    };
+};

lib/v2/caai/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/:caty': ['tudou027'],
+};

lib/v2/caai/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'caai.cn': {
+        _name: '中国人工智能学会',
+        '.': [
+            {
+                title: '学会动态',
+                docs: 'https://docs.rsshub.app/study.html#zhong-guo-ren-gong-zhi-neng-xue-hui',
+                source: ['/index.php'],
+                target: (_, url) => `/caai/${url.match(/\/(\d+)\.html/)[1]}`,
+            },
+        ],
+    },
+};

lib/v2/caai/router.js

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

lib/v2/caai/templates/description.art

@@ -0,0 +1 @@
+{{@ desc }}

lib/v2/caai/utils.js

@@ -0,0 +1,46 @@
+const cheerio = require('cheerio');
+const got = require('@/utils/got');
+const { art } = require('@/utils/render');
+const path = require('path');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+const base = 'http://www.caai.cn';
+
+const urlBase = (caty) => base + `/index.php?s=/home/article/index/id/${caty}.html`;
+
+const renderDesc = (desc) =>
+    art(path.join(__dirname, 'templates/description.art'), {
+        desc,
+    });
+
+const detailPage = (e, cache) =>
+    cache.tryGet(e.link, async () => {
+        const result = await got(e.link);
+        const $ = cheerio.load(result.data);
+        e.description = $('div.article').html();
+        return e;
+    });
+
+const fetchAllArticles = (data) => {
+    const $ = cheerio.load(data);
+    const articles = $('div.article-list > ul > li');
+    const info = articles.toArray().map((e) => {
+        const c = $(e);
+        const r = {
+            title: c.find('h3 a[href]').text().trim(),
+            link: base + c.find('h3 a[href]').attr('href'),
+            pubDate: timezone(parseDate(c.find('h4').text().trim(), 'YYYY-MM-DD'), +8),
+        };
+        return r;
+    });
+    return info;
+};
+
+module.exports = {
+    BASE: base,
+    urlBase,
+    fetchAllArticles,
+    detailPage,
+    renderDesc,
+};

lib/v2/ccfcv/index.js → lib/v2/ccf/ccfcv/index.js

@@ -4,7 +4,7 @@ const { parseDate } = require('@/utils/parse-date');
 const { art } = require('@/utils/render');
 const path = require('path');
 
-const rootUrl = 'http://ccfcv.ccf.org.cn';
+const rootUrl = 'https://tc.ccf.org.cn';
 
 const cateTitleMap = {
     xsdt: {
@@ -48,7 +48,7 @@ module.exports = async (ctx) => {
                     if (item.status !== 404) {
                         const content = cheerio.load(detailResponse.data);
                         const pdfUrl = content('div.g-box1 p a').attr('href');
-                        item.description = art(path.join(__dirname, 'templates/description.art'), {
+                        item.description = art(path.join(__dirname, '../templates/ccfcv/description.art'), {
                             pdfUrl,
                         });
                     }

lib/v2/ccf/maintainer.js

@@ -0,0 +1,5 @@
+module.exports = {
+    '/ccfcv/:channel/:category': ['elxy'],
+    '/news/:category?': ['nczitzk'],
+    '/tfbd/:caty/:id': ['tudou027'],
+};

lib/routes/ccf/news.js → lib/v2/ccf/news.js

@@ -1,42 +1,36 @@
 const got = require('@/utils/got');
 const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
 
 module.exports = async (ctx) => {
     const category = ctx.params.category || 'Media_list';
 
     const rootUrl = 'https://www.ccf.org.cn';
-    const currentUrl = `${rootUrl}/${category}`;
-    const response = await got({
-        method: 'get',
-        url: currentUrl,
-    });
+    const currentUrl = `${rootUrl}/${category}/`;
+    const response = await got(currentUrl);
 
     const $ = cheerio.load(response.data);
 
     const list = $('.tit a')
-        .slice(0, 10)
-        .map((_, item) => {
+        .toArray()
+        .map((item) => {
             item = $(item);
             return {
                 title: item.text(),
                 link: `${rootUrl}${item.attr('href')}`,
             };
-        })
-        .get();
+        });
 
     const items = await Promise.all(
         list.map((item) =>
             ctx.cache.tryGet(item.link, async () => {
-                const detailResponse = await got({
-                    method: 'get',
-                    url: item.link,
-                });
+                const detailResponse = await got(item.link);
                 const content = cheerio.load(detailResponse.data);
 
                 content('.new_info .num').remove();
 
                 item.description = content('.txt').html();
-                item.pubDate = new Date(content('.new_info span').text()).toUTCString();
+                item.pubDate = parseDate(content('.new_info span').text());
 
                 return item;
             })

lib/v2/ccf/radar.js

@@ -0,0 +1,41 @@
+module.exports = {
+    'ccf.org.cn': {
+        _name: '中国计算机学会',
+        '.': [
+            {
+                title: '新闻',
+                docs: 'https://docs.rsshub.app/study.html#zhong-guo-ji-suan-ji-xue-hui',
+                source: ['/:category', '/'],
+                target: '/ccf/news/:category',
+            },
+        ],
+        ccfcv: [
+            {
+                title: '计算机视觉专委会 - 学术动态 - 学术前沿',
+                docs: 'https://docs.rsshub.app/study.html#zhong-guo-ji-suan-ji-xue-hui',
+                source: ['/ccfcv/xsdt/xsqy/'],
+                target: '/ccf/ccfcv/xsdt/xsqy',
+            },
+            {
+                title: '计算机视觉专委会 - 学术动态 - 热点征文',
+                docs: 'https://docs.rsshub.app/study.html#zhong-guo-ji-suan-ji-xue-hui',
+                source: ['/ccfcv/xsdt/rdzw/'],
+                target: '/ccf/ccfcv/xsdt/rdzw',
+            },
+            {
+                title: '计算机视觉专委会 - 学术动态 - 学术会议',
+                docs: 'https://docs.rsshub.app/study.html#zhong-guo-ji-suan-ji-xue-hui',
+                source: ['/ccfcv/xsdt/xshy/'],
+                target: '/ccf/ccfcv/xsdt/xshy',
+            },
+        ],
+        tfbd: [
+            {
+                title: '大数据专家委员会',
+                docs: 'https://docs.rsshub.app/study.html#zhong-guo-ji-suan-ji-xue-hui',
+                source: ['/tfbd/:caty/:id', '/'],
+                target: '/ccf/tfbd/:caty/:id',
+            },
+        ],
+    },
+};

lib/v2/ccf/router.js

@@ -0,0 +1,5 @@
+module.exports = function (router) {
+    router.get('/ccfcv/:channel/:category', require('./ccfcv/index'));
+    router.get('/news/:category?', require('./news'));
+    router.get('/tfbd/:caty/:id', require('./tfbd/index'));
+};

lib/v2/ccf/templates/tfbd/description.art

@@ -0,0 +1 @@
+{{@ desc }}

lib/v2/ccf/tfbd/index.js

@@ -0,0 +1,22 @@
+const utils = require('./utils');
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+
+module.exports = async (ctx) => {
+    const base = utils.urlBase(ctx.params.caty, ctx.params.id);
+    const res = await got(base);
+    const info = utils.fetchAllArticles(res.data);
+    const $ = cheerio.load(res.data);
+
+    const details = await Promise.all(info.map((e) => utils.detailPage(e, ctx.cache)));
+
+    ctx.state.json = {
+        info,
+    };
+
+    ctx.state.data = {
+        title: '大数据专家委员会 - ' + $('.position a:last-child').text(),
+        link: base,
+        item: details,
+    };
+};

lib/v2/ccf/tfbd/utils.js

@@ -0,0 +1,47 @@
+const cheerio = require('cheerio');
+const got = require('@/utils/got');
+const { art } = require('@/utils/render');
+const path = require('path');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+const base = 'http://tfbd.ccf.org.cn';
+
+const urlBase = (caty, id) => base + `/tfbd/${caty}/${id}/`;
+
+const renderDesc = (desc) =>
+    art(path.join(__dirname, '../templates/tfbd/description.art'), {
+        desc,
+    });
+
+const detailPage = (e, cache) =>
+    cache.tryGet(e.link, async () => {
+        const result = await got(e.link);
+        const $ = cheerio.load(result.data);
+        e.description = $('div.articleCon').html();
+
+        return e;
+    });
+
+const fetchAllArticles = (data) => {
+    const $ = cheerio.load(data);
+    const articles = $('div.file-list div.article-item');
+    const info = articles.toArray().map((e) => {
+        const c = $(e);
+        const r = {
+            title: c.find('h3 a[href]').text().trim(),
+            link: base + c.find('h3 a[href]').attr('href'),
+            pubDate: timezone(parseDate(c.find('p').text().trim(), 'YYYY-MM-DD'), +8),
+        };
+        return r;
+    });
+    return info;
+};
+
+module.exports = {
+    BASE: base,
+    urlBase,
+    fetchAllArticles,
+    detailPage,
+    renderDesc,
+};