feat(route): add HAFU 河南财政金融学院 (#10330)

* feat(route): add HAFU 河南财政金融学院

* fix problems

1. fix route path in all files
2. fix radar target()
3. use camelCase
4. use ctx.cache.tryGet()
5. add "attachments" in description

* 1.run format, 2.add Promise.all()

docs/university.md

@@ -1252,6 +1252,18 @@ category 列表：
 
 <Route author="plusmultiply0" example="/hhu/libNewsc" path="/hhu/libNewsc"/>
 
+## 河南财政金融学院
+
+### 河南财政金融学院
+
+<Route author="WenPeiTung" example="/hafu/news/ggtz" path="/hafu/news/:type?" radar="1" rssbud="1" :paramsDesc="['分类，见下表（默认为 `ggtz`)']">
+
+| 校内公告通知 | 教务处公告通知 | 招生就业处公告通知 |
+| ------ | ------- | --------- |
+| ggtz   | jwc     | zsjyc     |
+
+</Route>
+
 ## 河南大学
 
 ### 河南大学

lib/v2/hafu/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/news/:tpye?': ['WenPeiTung'],
+};

lib/v2/hafu/news.js

@@ -0,0 +1,15 @@
+const parseList = require('./utils');
+
+module.exports = async (ctx) => {
+    // set default router type
+    const type = ctx.params.type ?? 'ggtz';
+
+    const { link, title, resultList } = await parseList(ctx, type);
+
+    ctx.state.data = {
+        title,
+        link,
+        description: '河南财政金融学院 - 公告通知',
+        item: resultList,
+    };
+};

lib/v2/hafu/radar.js

@@ -0,0 +1,23 @@
+module.exports = {
+    'hafu.edu.cn': {
+        _name: '河南财政金融学院',
+        www: [
+            {
+                title: '河南财政金融学院 - 通知公告',
+                docs: 'https://docs.rsshub.app/university.html#he-nan-cai-zheng-jin-rong-xue-yuan',
+                source: '/*',
+                target: (params, url) => {
+                    if (url.indexOf('www')) {
+                        return '/hafu/news/ggtz';
+                    }
+                    if (url.indexOf('jwc')) {
+                        return '/hafu/news/jwc';
+                    }
+                    if (url.indexOf('zsjyc')) {
+                        return '/hafu/news/zsjyc';
+                    }
+                },
+            },
+        ],
+    },
+};

lib/v2/hafu/router.js

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

lib/v2/hafu/templates/hafu.art

@@ -0,0 +1,3 @@
+{{ if articleBody }}
+    {{@ articleBody }}
+{{ /if }}

lib/v2/hafu/utils.js

@@ -0,0 +1,205 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+const { art } = require('@/utils/render');
+const path = require('path');
+
+const typeMap = {
+    ggtz: { url: 'https://www.hafu.edu.cn/index/ggtz.htm', root: 'https://www.hafu.edu.cn/', title: '河南财院 - 公告通知', parseFn: ggtzParse },
+    jwc: { url: 'https://jwc.hafu.edu.cn/tzgg.htm', root: 'https://jwc.hafu.edu.cn/', title: '河南财院 教务处 - 公告通知', parseFn: jwcParse },
+    zsjyc: { url: 'https://zsjyc.hafu.edu.cn/tztg.htm', root: 'https://zsjyc.hafu.edu.cn/', title: '河南财院 招生就业处 - 公告通知', parseFn: zsjycParse },
+};
+// Number of get articles
+let limit = 10;
+
+module.exports = async (ctx, type) => {
+    const link = typeMap[type].url;
+    const title = typeMap[type].title;
+
+    const response = await got(link);
+    const $ = cheerio.load(response.data);
+
+    limit = ctx.query.limit || limit;
+    const resultList = await typeMap[type].parseFn(ctx, $);
+
+    return {
+        title,
+        link,
+        resultList,
+    };
+};
+
+async function tryGetFullText(href, link, type) {
+    let articleData = '';
+    let description = '';
+    // for some unexpected href link
+    try {
+        const articleRes = await got(link);
+        articleData = cheerio.load(articleRes.data);
+        // fullText
+        let articleBody = articleData('div[class=v_news_content]').html();
+        // attachments
+        if (articleData('[id^=nattach]').length !== 0) {
+            articleBody = tryGetAttachments(articleData, articleBody, type);
+        }
+
+        description = art(path.join(__dirname, 'templates/hafu.art'), articleBody)();
+    } catch {
+        description = href;
+    }
+
+    return { articleData, description };
+}
+
+function tryGetAttachments(articleData, articleBody, type) {
+    if (type === 'ggtz') {
+        articleData(`[id^=nattach]`)
+            .prev()
+            .map((_, item) => {
+                const href = articleData(item).attr('href').slice(1);
+                const link = typeMap.ggtz.root + href;
+                const title = articleData(item).text();
+                articleBody += '<br/>';
+                articleBody += `<a href=${link}>${title}</a>`;
+                return null;
+            });
+    } else {
+        articleData('[id^=nattach]')
+            .parent()
+            .prev()
+            .map((_, item) => {
+                const href = articleData(item).find('a').attr('href').slice(1);
+                const link = typeMap[type].root + href;
+                const title = articleData(item).find('a').find('span').text();
+                articleBody += '<br/>';
+                articleBody += `<a href=${link}> ${title} </a>`;
+                return null;
+            });
+    }
+
+    return articleBody;
+}
+// A. got from hostPage     1.article(link), 2.article(title), 3.(pubDate)
+// B. got from articlePage  1.description(fullText), 2.article(author), 3.detailed(pubDate)
+async function ggtzParse(ctx, $) {
+    const data = $('a[class=c269582]').parent().slice(0, limit);
+    const resultItems = await Promise.all(
+        data
+            .map(async (_, item) => {
+                // .slice(3) for cut out str '../' in original link
+                const href = $(item).find('a[class=c269582]').attr('href').slice(3);
+                const link = typeMap.ggtz.root + href;
+                const title = $(item).find('a[class=c269582]').attr('title');
+
+                const result = await ctx.cache.tryGet(link, async () => {
+                    const { articleData, description } = await tryGetFullText(href, link, 'ggtz');
+                    let author = '';
+                    let pubDate = '';
+                    if (articleData instanceof Function) {
+                        const header = articleData('h1').next().text();
+                        const index = header.indexOf('日期');
+
+                        author = header.substring(0, index - 2) || '';
+
+                        const date = header.substring(index + 3, index + 19);
+                        pubDate = parseDate(date, 'YYYY-MM-DD HH:mm');
+                    } else {
+                        const date = $(item).find('a[class=c269582_date]').text();
+                        pubDate = parseDate(date, 'YYYY-MM-DD');
+                    }
+
+                    return {
+                        title,
+                        description,
+                        pubDate: timezone(pubDate, +8),
+                        link,
+                        author,
+                    };
+                });
+
+                return result;
+            })
+            .get()
+    );
+
+    return resultItems;
+}
+// A. got from hostPage     1.article(link), 2.article(title), 3.(pubDate)
+// B. got from articlePage  1.description(fullText), 2.article(author)
+async function jwcParse(ctx, $) {
+    const data = $('a[class=c259713]').parent().parent().slice(0, limit);
+    const resultItems = await Promise.all(
+        data
+            .map(async (_, item) => {
+                const href = $(item).find('a[class=c259713]').attr('href');
+                const link = typeMap.jwc.root + href;
+                const title = $(item).find('a[class=c259713]').attr('title');
+
+                const date = $(item).find('span[class=timestyle259713]').text();
+                const pubDate = parseDate(date, 'YYYY/MM/DD');
+
+                const result = await ctx.cache.tryGet(link, async () => {
+                    const { articleData, description } = await tryGetFullText(href, link, 'jwc');
+
+                    let author = '';
+                    if (articleData instanceof Function) {
+                        author = articleData('span[class=authorstyle259690]').text();
+                    }
+
+                    return {
+                        title,
+                        description,
+                        pubDate: timezone(pubDate, +8),
+                        link,
+                        author: '供稿单位：' + author,
+                    };
+                });
+
+                return result;
+            })
+            .get()
+    );
+
+    return resultItems;
+}
+// A. got from hostPage     1.article(link), 2.article(title), 3.(pubDate)
+// B. got from articlePage  1.description(fullText), 2.detailed(pubDate)
+async function zsjycParse(ctx, $) {
+    const data = $('a[class=c127701]').parent().parent().slice(0, limit);
+    const resultItems = await Promise.all(
+        data
+            .map(async (_, item) => {
+                const href = $(item).find('a[class=c127701]').attr('href');
+                const link = typeMap.zsjyc.root + href;
+
+                const title = $(item).find('a[class=c127701]').attr('title');
+
+                const result = await ctx.cache.tryGet(link, async () => {
+                    const { articleData, description } = await tryGetFullText(href, link, 'zsjyc');
+
+                    let pubDate = '';
+                    if (articleData instanceof Function) {
+                        const date = articleData('span[class=timestyle127702]').text();
+                        pubDate = parseDate(date, 'YYYY-MM-DD HH:mm');
+                    } else {
+                        const date = $(item).find('a[class=c269582_date]').text();
+                        pubDate = parseDate(date, 'YYYY-MM-DD');
+                    }
+
+                    return {
+                        title,
+                        description,
+                        pubDate: timezone(pubDate, +8),
+                        link,
+                        author: '供稿单位：招生就业处',
+                    };
+                });
+
+                return result;
+            })
+            .get()
+    );
+
+    return resultItems;
+}