feat(route): add zuzhirenshi 中国组织人事报 (DIYgod#12077)

* add zuzhirenshi * complete radar * change title format * Apply suggestion from code review
benzking · Mar 11, 2023 · 9dd7f0b · 9dd7f0b
1 parent 6fd8bdf
commit 9dd7f0b
Show file tree

Hide file tree

Showing 5 changed files with 118 additions and 0 deletions.
diff --git a/docs/traditional-media.md b/docs/traditional-media.md
@@ -2570,3 +2570,15 @@ category 对应的关键词有
 `https://www.rfa.org/cantonese/news/htm` 对应 `/rfa/cantonese/news/htm`
 
 </Route>
+
+## 组织人事报
+
+### 电子报
+
+<Route author="5upernove-heng" example="/zuzhirenshi" path="/zuzhirenshi/:id?" :paramsDesc="['报纸版号，默认为全部']" radar="1">
+
+| 第一版 要闻 | 第二版 要闻 | 第三版 人才 | 第四版 人社工作 | 第五版 基层党建 | 第六版 理论评论 | 第七版 史事通鉴 | 第八版 关注 |
+| ----------- | ----------- | ----------- | --------------- | --------------- | --------------- | --------------- | ----------- |
+| 1           | 2           | 3           | 4               | 5               | 6               | 7               | 8           |
+
+</Route>
diff --git a/lib/v2/zuzhirenshi/index.js b/lib/v2/zuzhirenshi/index.js
@@ -0,0 +1,87 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const timezone = require('@/utils/timezone');
+const { parseDate } = require('@/utils/parse-date');
+
+function parseJavascript(jsCode) {
+    // Parse Release Date, SubLanMuList, AreaInfoList
+    const ReleaseDateRegex = /var ReleaseDate\s*=\s*"([^"]+)";/;
+    const ReleaseDateMatch = jsCode.match(ReleaseDateRegex);
+    const ReleaseDate = ReleaseDateMatch ? ReleaseDateMatch[1] : null;
+
+    const SubLanMuListRegex = /var SubLanMuList\s*=\s*(\[.+?\]);/;
+    const SubLanMuListMatch = jsCode.match(SubLanMuListRegex);
+    const SubLanMuList = SubLanMuListMatch ? JSON.parse(SubLanMuListMatch[1]) : null;
+
+    const InfoListRegex = /var InfoList\s*=\s*(\[.+?\]);/;
+    const InfoListMatch = jsCode.match(InfoListRegex);
+    const InfoList = InfoListMatch ? JSON.parse(InfoListMatch[1]) : null;
+
+    return {
+        ReleaseDate,
+        SubLanMuList,
+        InfoList,
+    };
+}
+
+async function getArticle(link) {
+    const response = await got(link);
+    const $ = cheerio.load(response.data);
+    const title = $('div.listhottitle3').text().trim();
+    const pubDate = timezone(parseDate($('.innertop').clone().children().remove().end().text().trim(), 'YYYY/MM/DD'), +8);
+    const description = $('.innercontent').html();
+    return {
+        title,
+        link,
+        pubDate,
+        description,
+    };
+}
+
+async function getPageInfo(url) {
+    const response = await got(url);
+    const $ = cheerio.load(response.data);
+    const jsCode = $('script').text().split('$(function')[0];
+    return parseJavascript(jsCode);
+}
+
+async function getPageContents(releaseDate, id, ctx) {
+    const baseUrl = `http://www.zuzhirenshi.com/dianzibao`;
+    const link = `${baseUrl}/${releaseDate}/${id}/index.htm`;
+    const info = await getPageInfo(link);
+    const items = await Promise.all(
+        info.InfoList.map((article) => {
+            const link = `${baseUrl}/${releaseDate}/${id}/${article.infoid}.htm`;
+            const item = ctx.cache.tryGet(link, async () => {
+                const item = await getArticle(link);
+                return item;
+            });
+            return item;
+        })
+    );
+    return items;
+}
+
+module.exports = async (ctx) => {
+    const url = 'http://www.zuzhirenshi.com/dianzibao/index.htm';
+    const info = await getPageInfo(url);
+    const releaseDate = info.ReleaseDate;
+    const id = ctx.params.id || 0;
+    let items = [];
+    if (id === 0) {
+        const promises = [];
+        for (let i = 1; i <= info.SubLanMuList.length; i++) {
+            promises.push(getPageContents(releaseDate, i, ctx));
+        }
+        const results = await Promise.all(promises);
+        items = results.reduce((a, b) => a.concat(b), []);
+    } else {
+        items = await getPageContents(releaseDate, id, ctx);
+    }
+    const titleSuffix = id ? `${String(info.SubLanMuList[id - 1].lmmc)}` : '全部';
+    ctx.state.data = {
+        title: `组织人事报 - ${titleSuffix} - ${releaseDate}`,
+        link: 'https://www.zuzhirenshi.com/dianzibao/index.htm',
+        item: items,
+    };
+};
diff --git a/lib/v2/zuzhirenshi/maintainer.js b/lib/v2/zuzhirenshi/maintainer.js
@@ -0,0 +1,3 @@
+module.exports = {
+    '/:id?': ['5upernova-heng'],
+};
diff --git a/lib/v2/zuzhirenshi/radar.js b/lib/v2/zuzhirenshi/radar.js
@@ -0,0 +1,13 @@
+module.exports = {
+    'zuzhirenshi.com': {
+        _name: '组织人事报',
+        '.': [
+            {
+                title: '日报',
+                docs: 'https://docs.rsshub.app/traditional-media.html#zu-zhi-ren-shi-bao',
+                source: ['/dianzibao/*'],
+                target: '/zuzhirenshi',
+            },
+        ],
+    },
+};
diff --git a/lib/v2/zuzhirenshi/router.js b/lib/v2/zuzhirenshi/router.js
@@ -0,0 +1,3 @@
+module.exports = (router) => {
+    router.get('/:id?', require('.'));
+};