From 288f1d1fd745a72f0bdcce5a596d74cea2738d75 Mon Sep 17 00:00:00 2001
From: Enoch Ma <emdoe@users.noreply.github.com>
Date: Tue, 26 Mar 2024 15:33:48 +0100
Subject: [PATCH] feat(route): support category and fix time (twreporter)
 (#14974)

* support category and fix time(twreporter)

* html -> text

* simplification

* update list of entries

---------
---
 lib/routes/twreporter/category.ts      | 63 ++++++++++++++++++++++++++
 lib/routes/twreporter/fetch-article.ts | 14 +++---
 2 files changed, 70 insertions(+), 7 deletions(-)
 create mode 100644 lib/routes/twreporter/category.ts

diff --git a/lib/routes/twreporter/category.ts b/lib/routes/twreporter/category.ts
new file mode 100644
index 0000000000..c09e8b92dc
--- /dev/null
+++ b/lib/routes/twreporter/category.ts
@@ -0,0 +1,63 @@
+import { Route } from '@/types';
+import cache from '@/utils/cache';
+import { load } from 'cheerio';
+import got from '@/utils/got';
+
+import fetch from './fetch-article';
+
+export const route: Route = {
+    path: '/category/:category',
+    categories: ['new-media'],
+    example: '/twreporter/category/world',
+    parameters: { category: 'Category' },
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['twreporter.org/:category'],
+        },
+    ],
+    name: '分類',
+    maintainers: ['emdoe'],
+    handler,
+    url: 'twreporter.org/',
+};
+
+async function handler(ctx) {
+    const baseURL = 'https://www.twreporter.org';
+    const url = baseURL + `/categories/${ctx.req.param('category')}`;
+    const res = await got(url);
+    const $ = load(res.data);
+
+    const regexp = /^window\.__REDUX_STATE__=(.*);$/gm;
+    const raw = $('script[charset="UTF-8"]').text().replaceAll(regexp, '$1');
+    const posts = JSON.parse(raw).entities.posts;
+
+    const list = posts.allIds.map((id) => ({
+        link: baseURL + '/a/' + posts.byId[id].slug,
+        title: posts.byId[id].title,
+    }));
+    const category = posts.byId[posts.allIds[0]].category_set[0].category.name;
+
+    const out = await Promise.all(
+        list.map((item) =>
+            cache.tryGet(item.link, async () => {
+                const single = await fetch(item.link);
+                single.title = item.title;
+                return single;
+            })
+        )
+    );
+
+    return {
+        title: `報導者 | ${category}`,
+        link: url,
+        item: out,
+    };
+}
diff --git a/lib/routes/twreporter/fetch-article.ts b/lib/routes/twreporter/fetch-article.ts
index ea635bbd23..a9bca8d4c0 100644
--- a/lib/routes/twreporter/fetch-article.ts
+++ b/lib/routes/twreporter/fetch-article.ts
@@ -7,6 +7,12 @@ export default async function fetch(address) {
     const capture = load(res.data);
     capture('.gIMvvS').remove();
 
+    let time = capture('.gzaDTq').text();
+    // For `photography`
+    if (!time) {
+        time = capture('.fEZfRw').text();
+    }
+
     let metaInfoBox = capture('.ffAPnj')
         .filter((index) => index === 0)
         .get();
@@ -19,12 +25,6 @@ export default async function fetch(address) {
     }
 
     const acquire = load(metaInfoBox);
-    let time = acquire('.gimsRe').text();
-
-    // For `photography`
-    if (!time) {
-        time = acquire('.kHluJP').text();
-    }
     // # Author(s) of the article
     //
     // There exists two formats for this section.
@@ -62,6 +62,6 @@ export default async function fetch(address) {
         description: contents,
         link: address,
         guid: address,
-        pubDate: parseDate(time, 'M/D/YYYY'),
+        pubDate: parseDate(time, 'YYYY/M/D'),
     };
 }