feat(route): add 浙江在线浙报集团系列报刊 (DIYgod#12058)

* feat(route): add 浙江在线浙报集团系列报刊 * fix: duplicated content in jnyb
benzking · Mar 8, 2023 · 197610d · 197610d
1 parent fdb1e21
commit 197610d
Show file tree

Hide file tree

Showing 5 changed files with 167 additions and 0 deletions.
diff --git a/docs/traditional-media.md b/docs/traditional-media.md
@@ -2451,6 +2451,18 @@ category 对应的关键词有
 
 </Route>
 
+## 浙江在线
+
+### 浙报集团系列报刊
+
+<Route author="nczitzk" example="/zjol/paper/zjrb" path="/zjol/paper/:id?" :paramsDesc="['报纸 id，见下表，默认为 `zjrb`，即浙江日报']">
+
+| 浙江日报 | 钱江晚报 | 美术报 | 浙江老年报 | 浙江法制报 | 江南游报 |
+| ---- | ---- | --- | ----- | ----- | ---- |
+| zjrb | qjwb | msb | zjlnb | zjfzb | jnyb |
+
+</Route>
+
 ## 中国日报
 
 ### 英语点津

diff --git a/lib/v2/zjol/maintainer.js b/lib/v2/zjol/maintainer.js
@@ -0,0 +1,3 @@
+module.exports = {
+    '/paper/:id?': ['nczitzk'],
+};
diff --git a/lib/v2/zjol/paper.js b/lib/v2/zjol/paper.js
@@ -0,0 +1,88 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+module.exports = async (ctx) => {
+    const id = ctx.params.id ?? 'zjrb';
+    const limit = ctx.query.limit ? parseInt(ctx.query.limit) : 100;
+
+    const query = id === 'jnyb' ? 'map[name="PagePicMap"] area' : 'ul.main-ed-articlenav-list li a';
+
+    const rootUrl = id === 'qjwb' ? 'http://qjwb.thehour.cn' : `https://${id}.zjol.com.cn`;
+    let currentUrl = `${rootUrl}/paperindex.htm`;
+
+    let response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const url = response.data.match(/URL=(.*)"/)[1];
+    const pubDate = parseDate(url.match(/(\d{4}-\d{2}\/\d{2})/)[1], 'YYYY-MM/DD');
+
+    currentUrl = `${rootUrl}/${url.replace(`/${url.split('/').pop()}`, '')}`;
+
+    response = await got({
+        method: 'get',
+        url: `${rootUrl}/${url}`,
+    });
+
+    const $ = cheerio.load(response.data);
+
+    let items = $(query)
+        .toArray()
+        .map((a) => `${currentUrl}/${$(a).attr('href')}`);
+
+    await Promise.all(
+        $('#pageLink')
+            .slice(1)
+            .toArray()
+            .map((p) => `${currentUrl}/${$(p).attr('href')}`)
+            .map(async (p) => {
+                const pageResponse = await got({
+                    method: 'get',
+                    url: p,
+                });
+
+                const page = cheerio.load(pageResponse.data);
+
+                items.push(
+                    ...page(query)
+                        .toArray()
+                        .map((a) => `${currentUrl}/${page(a).attr('href')}`)
+                );
+            })
+    );
+
+    items = await Promise.all(
+        items
+            .filter((a) => (id === 'jnyb' ? /\?div=1$/.test(a) : true))
+            .slice(0, limit)
+            .map((link) =>
+                ctx.cache.tryGet(link, async () => {
+                    const detailResponse = await got({
+                        method: 'get',
+                        url: link,
+                    });
+
+                    const content = cheerio.load(detailResponse.data);
+
+                    const title = content('.main-article-title').text();
+
+                    content('.main-article-alltitle').remove();
+
+                    return {
+                        title,
+                        pubDate,
+                        link: link.split('?')[0],
+                        description: content('.main-article-content').html(),
+                    };
+                })
+            )
+    );
+
+    ctx.state.data = {
+        title: $('title').text(),
+        link: rootUrl,
+        item: items,
+    };
+};
diff --git a/lib/v2/zjol/radar.js b/lib/v2/zjol/radar.js
@@ -0,0 +1,61 @@
+module.exports = {
+    'zjol.com.cn': {
+        _name: '浙江在线',
+        '.': [
+            {
+                title: '浙报集团系列报刊',
+                docs: 'https://docs.rsshub.app/traditional-media.html#zhe-jiang-zai-xian-zhe-bao-ji-tuan-xi-lie-bao-kan',
+                source: ['/'],
+                target: (params, url) => `/zjol/paper/${new URL(url).toString().match(/\/\/(.*?)\.zjol/)[1]}`,
+            },
+        ],
+        zjrb: [
+            {
+                title: '浙江日报',
+                docs: 'https://docs.rsshub.app/traditional-media.html#zhe-jiang-zai-xian-zhe-bao-ji-tuan-xi-lie-bao-kan',
+                source: ['/'],
+                target: '/zjol/paper/zjrb',
+            },
+        ],
+        qjwb: [
+            {
+                title: '钱江晚报',
+                docs: 'https://docs.rsshub.app/traditional-media.html#zhe-jiang-zai-xian-zhe-bao-ji-tuan-xi-lie-bao-kan',
+                source: ['/'],
+                target: '/zjol/paper/qjwb',
+            },
+        ],
+        msb: [
+            {
+                title: '美术报',
+                docs: 'https://docs.rsshub.app/traditional-media.html#zhe-jiang-zai-xian-zhe-bao-ji-tuan-xi-lie-bao-kan',
+                source: ['/'],
+                target: '/zjol/paper/msb',
+            },
+        ],
+        zjlnb: [
+            {
+                title: '浙江老年报',
+                docs: 'https://docs.rsshub.app/traditional-media.html#zhe-jiang-zai-xian-zhe-bao-ji-tuan-xi-lie-bao-kan',
+                source: ['/'],
+                target: '/zjol/paper/zjlnb',
+            },
+        ],
+        zjfzb: [
+            {
+                title: '浙江法制报',
+                docs: 'https://docs.rsshub.app/traditional-media.html#zhe-jiang-zai-xian-zhe-bao-ji-tuan-xi-lie-bao-kan',
+                source: ['/'],
+                target: '/zjol/paper/zjfzb',
+            },
+        ],
+        jnyb: [
+            {
+                title: '江南游报',
+                docs: 'https://docs.rsshub.app/traditional-media.html#zhe-jiang-zai-xian-zhe-bao-ji-tuan-xi-lie-bao-kan',
+                source: ['/'],
+                target: '/zjol/paper/jnyb',
+            },
+        ],
+    },
+};
diff --git a/lib/v2/zjol/router.js b/lib/v2/zjol/router.js
@@ -0,0 +1,3 @@
+module.exports = function (router) {
+    router.get('/paper/:id?', require('./paper'));
+};