From 93e22bba30acccd9faa01493077a7a6ebf073481 Mon Sep 17 00:00:00 2001
From: BenYip <yebenmy@gmail.com>
Date: Wed, 19 Jul 2023 20:25:17 +0800
Subject: [PATCH] feat(route): add NPR (#12827)

* feat(route): add NPR

* fix(route/npr): docs

* fix(route/npr): remove duplicate images

* fix(route/npr): ignore item until audio is available

* fix(route/npr): remove duplicate captions

* fix(route/npr): caption

* fix(route/npr): handle multiple audios
---
 docs/en/traditional-media.md | 12 +++++
 docs/traditional-media.md    | 10 ++++
 lib/v2/npr/full.js           | 89 ++++++++++++++++++++++++++++++++++++
 lib/v2/npr/maintainer.js     |  3 ++
 lib/v2/npr/radar.js          | 11 +++++
 lib/v2/npr/router.js         |  3 ++
 6 files changed, 128 insertions(+)
 create mode 100644 lib/v2/npr/full.js
 create mode 100644 lib/v2/npr/maintainer.js
 create mode 100644 lib/v2/npr/radar.js
 create mode 100644 lib/v2/npr/router.js

diff --git a/docs/en/traditional-media.md b/docs/en/traditional-media.md
index 8c0b9e838..f5c01348c 100644
--- a/docs/en/traditional-media.md
+++ b/docs/en/traditional-media.md
@@ -784,3 +784,15 @@ Free articles only.
 | 皇室           | koushitsu |
 
 </RouteEn>
+
+## NPR
+
+### News
+
+<RouteEn author="bennyyip" example="/npr/1001" path="/npr/:endpoint?" :paramsDesc="['Channel ID, can be found in Official RSS URL, `1001` by default']">
+
+Provide full article RSS for CBC topics.
+
+</RouteEn>
+
+
diff --git a/docs/traditional-media.md b/docs/traditional-media.md
index 800d0e7e9..c3f361d4c 100644
--- a/docs/traditional-media.md
+++ b/docs/traditional-media.md
@@ -2715,3 +2715,13 @@ category 对应的关键词有
 | 1           | 2           | 3           | 4               | 5               | 6               | 7               | 8           |
 
 </Route>
+
+## NPR
+
+### 新闻
+
+<Route author="bennyyip" example="/npr/1001" path="/npr/:endpoint?" :paramsDesc="['频道 ID，可在官方频道 RSS URL 中找到，默认为 `1001`']">
+
+通过提取文章全文，以提供比官方源更佳的阅读体验。
+
+</Route>
diff --git a/lib/v2/npr/full.js b/lib/v2/npr/full.js
new file mode 100644
index 000000000..faf84a4ef
--- /dev/null
+++ b/lib/v2/npr/full.js
@@ -0,0 +1,89 @@
+const parser = require('@/utils/rss-parser');
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+
+const getArticleDetail = (link, ctx) =>
+    ctx.cache.tryGet(link, async () => {
+        const response = await got(link);
+        const $ = cheerio.load(response.data);
+
+        const categories = $('.slug').text().trim();
+
+        // ignore item until audio is available
+        if ($('.audio-availability-message').length > 0) {
+            return null;
+        }
+
+        // handle multiple audios
+        $('.audio-module')
+            .toArray()
+            .forEach((x) => {
+                const audioLink = $($(x).find('.audio-tool-download a')).attr('href');
+                const audio_tag = `<audio controls><source src="${audioLink}" type="audio/mp3"></audio>`;
+                $(x).replaceWith(audio_tag);
+            });
+
+        // primaryaudio is not in `.storytext`, prepend to it
+        const primaryaudio = $('#primaryaudio');
+        const audio = primaryaudio.length > 0 ? $('#primaryaudio').html() : '';
+
+        // replace video
+        const regex = /\?storyId=(\d+)&amp;mediaId=(\d+)/;
+        const m = $.html().match(regex);
+        if (m) {
+            const storyId = m[1];
+            const mediaId = m[2];
+            const video_tag = `<iframe width="740" height="416" src="https://www.npr.org/embedded-video?storyId=${storyId}&mediaId=${mediaId}&jwMediaType=music" frameborder="0" scrolling="no"></iframe>`;
+            const nprVideo = $('.npr-video');
+            if (nprVideo.length === 1) {
+                nprVideo.replaceWith(video_tag);
+            }
+        }
+
+        // remove duplicate caption and images
+        $('.enlarge_measure').remove();
+        $('.enlarge_html').remove();
+        $('.hide-caption').remove();
+        $('.toggle-caption').remove();
+        $('.credit-caption b.credit').remove();
+        $('.credit-caption span.credit').wrap('<figcaption></figcaption>');
+        $('.caption > p').wrap('<figcaption></figcaption>');
+
+        const article =
+            audio +
+            $('.storytext')
+                .toArray()
+                .map((el) => $(el).html())
+                .join('');
+
+        return { article, categories };
+    });
+
+module.exports = async (ctx) => {
+    const endpoint = ctx.params.endpoint || '1001';
+    const feed = await parser.parseURL(`https://feeds.npr.org/${endpoint}/rss.xml`);
+
+    const items = (
+        await Promise.all(
+            feed.items.map(async (item) => {
+                const itemDetails = await getArticleDetail(item.link, ctx);
+                if (itemDetails === null) {
+                    return null;
+                }
+                return {
+                    ...item,
+                    description: itemDetails.article,
+                    category: itemDetails.categories,
+                };
+            })
+        )
+    ).filter(Boolean);
+
+    ctx.state.data = {
+        title: feed.title,
+        link: feed.link,
+        description: feed.description,
+        icon: 'https://media.npr.org/images/podcasts/primary/npr_generic_image_300.jpg?s=200',
+        item: items,
+    };
+};
diff --git a/lib/v2/npr/maintainer.js b/lib/v2/npr/maintainer.js
new file mode 100644
index 000000000..4d5d4a2ce
--- /dev/null
+++ b/lib/v2/npr/maintainer.js
@@ -0,0 +1,3 @@
+module.exports = {
+    '/:endpoint?': ['bennyyip'],
+};
diff --git a/lib/v2/npr/radar.js b/lib/v2/npr/radar.js
new file mode 100644
index 000000000..3f753077a
--- /dev/null
+++ b/lib/v2/npr/radar.js
@@ -0,0 +1,11 @@
+module.exports = {
+    'npr.org': {
+        _name: 'NPR',
+        '.': [
+            {
+                title: '全文 RSS',
+                docs: 'https://docs.rsshub.app/traditional-media.html#npr',
+            },
+        ],
+    },
+};
diff --git a/lib/v2/npr/router.js b/lib/v2/npr/router.js
new file mode 100644
index 000000000..ba56db90b
--- /dev/null
+++ b/lib/v2/npr/router.js
@@ -0,0 +1,3 @@
+module.exports = (router) => {
+    router.get('/:endpoint?', require('./full'));
+};