feat: add 羊城晚报金羊网 (DIYgod#11995)

* feat: add 羊城晚报金羊网 * Fix example link of ycwb * fix(route): 羊城晚报金羊网 * fix(route): Update lib/v2/ycwb/index.js * fix(route): Update lib/v2/ycwb/radar.js * fix(route): Update lib/v2/ycwb/maintainer.js ---------
benzking · Mar 2, 2023 · 85eab96 · 85eab96
1 parent be06aeb
commit 85eab96
Show file tree

Hide file tree

Showing 6 changed files with 127 additions and 0 deletions.
diff --git a/docs/traditional-media.md b/docs/traditional-media.md
@@ -2433,6 +2433,26 @@ category 对应的关键词有
 
 <Route author="nczitzk" example="/cctv/photo/jx" path="/cctv/photo/jx" radar="1" rssbud="1"/>
 
+## 羊城晚报金羊网
+
+### 新闻
+
+<Route author="TimWu007" example="/ycwb/1" path="/ycwb/:node" :paramsDesc="['栏目 id']">
+
+注：小部分栏目的 URL 会给出 nodeid。如未给出，可打开某条新闻链接后，查看网页源代码，搜索 nodeid 的值。
+
+常用栏目节点：
+
+| 首页 | 中国 | 国际 | 体育 | 要闻 | 珠江评论 | 民生观察  | 房产  | 金羊教育 | 金羊财富 | 金羊文化 | 金羊健康  | 金羊汽车 |
+| -- | -- | -- | -- | -- | ---- | ----- | --- | ---- | ---- | ---- | ----- | ---- |
+| 1  | 14 | 15 | 16 | 22 | 1875 | 21773 | 222 | 5725 | 633  | 5281 | 21692 | 223  |
+
+| 广州 | 广州 - 广州要闻 | 广州 - 社会百态 | 广州 - 深读广州 | 广州 - 生活服务 | 今日大湾区  | 广东 - 政经热闻 | 广东 - 民生视点 | 广东 - 滚动新闻 |
+| -- | --------- | --------- | --------- | --------- | ------ | --------- | --------- | --------- |
+| 18 | 5261      | 6030      | 13352     | 83422     | 100418 | 13074     | 12252     | 12212     |
+
+</Route>
+
 ## 中国日报
 
 ### 英语点津

diff --git a/lib/v2/ycwb/index.js b/lib/v2/ycwb/index.js
@@ -0,0 +1,82 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const timezone = require('@/utils/timezone');
+const { parseDate } = require('@/utils/parse-date');
+const { art } = require('@/utils/render');
+const path = require('path');
+
+module.exports = async (ctx) => {
+    const node = ctx.params.node ?? 1;
+    const currentUrl = `https://6api.ycwb.com/app_if/jy/getArticles?nodeid=${node}&pagesize=15`;
+
+    const { data: response } = await got(currentUrl);
+
+    const list = response.artiles.map((item) => ({
+        title: item.TITLE,
+        description: art(path.join(__dirname, 'templates/description.art'), {
+            thumb: item.PICLINKS,
+            description: item.ABSTRACT,
+        }),
+        pubDate: timezone(parseDate(item.PUBTIME), +8),
+        link: item.PUBURL,
+        nodeName: item.NODENAME,
+    }));
+
+    let nodeName = '';
+    let nodeLink = '';
+
+    const items = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const detailResponse = await got({
+                    method: 'get',
+                    url: item.link,
+                });
+
+                const content = cheerio.load(detailResponse.data);
+
+                const $comments = content('.main_article')
+                    .contents()
+                    .filter(function () {
+                        return this.nodeType === 8;
+                    });
+                $comments.each(function () {
+                    // Remove useless comments
+                    if (this.data.match(/audioPlayer|audio-box/)) {
+                        this.data = '';
+                    }
+                    // Filter author from comments
+                    if (this.data.match(/author/)) {
+                        item.author = this.data.split('<author>')[1].split('</author>')[0];
+                    }
+                });
+
+                nodeName = nodeName === '' ? item.nodeName : nodeName;
+
+                // Filter node link from content('.path a')
+                const children = content('.path').children('a');
+                for (const child of children) {
+                    if (content(child).text() === nodeName && nodeLink === '') {
+                        nodeLink = content(child).attr('href');
+                    }
+                }
+
+                content('span').removeAttr('style').removeAttr('class');
+                content('img').removeAttr('style').removeAttr('class').removeAttr('placement').removeAttr('data-toggle').removeAttr('trigger').removeAttr('referrerpolicy');
+                content('br').removeAttr('style').removeAttr('class');
+                content('p').removeAttr('style').removeAttr('class');
+                content('.space10, .ddf').remove();
+
+                item.description += content('.main_article').html() ?? '';
+
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: `羊城晚报金羊网 - ${nodeName}`,
+        link: String(nodeLink === '' ? 'https://www.ycwb.com/' : nodeLink),
+        item: items,
+    };
+};
diff --git a/lib/v2/ycwb/maintainer.js b/lib/v2/ycwb/maintainer.js
@@ -0,0 +1,3 @@
+module.exports = {
+    '/:node': ['TimWu007'],
+};
diff --git a/lib/v2/ycwb/radar.js b/lib/v2/ycwb/radar.js
@@ -0,0 +1,12 @@
+module.exports = {
+    'xwlb.com.cn': {
+        _name: '羊城晚报金羊网',
+        '.': [
+            {
+                title: '新闻',
+                docs: 'https://docs.rsshub.app/traditional-media.html#yang-cheng-wan-bao-jin-yang-wang',
+                source: ['/'],
+            },
+        ],
+    },
+};
diff --git a/lib/v2/ycwb/router.js b/lib/v2/ycwb/router.js
@@ -0,0 +1,3 @@
+module.exports = (router) => {
+    router.get('/:node', require('./index'));
+};
diff --git a/lib/v2/ycwb/templates/description.art b/lib/v2/ycwb/templates/description.art
@@ -0,0 +1,7 @@
+{{ if thumb }}
+<img src="{{ thumb }}"><br>
+{{ /if }}
+{{ if description }}
+    <blockquote><p>{{ description }}</p></blockquote>
+    <br>
+{{ /if }}