From 3e895734aafd79e6b144973be6650656ee5802f5 Mon Sep 17 00:00:00 2001
From: Ethan Shen <42264778+nczitzk@users.noreply.github.com>
Date: Mon, 16 Jun 2025 21:27:16 +0800
Subject: [PATCH] feat(route): add Mathpix Blog (#19382)

---
 lib/routes/mathpix/blog.ts                   | 155 +++++++++++++++++++
 lib/routes/mathpix/namespace.ts              |   9 ++
 lib/routes/mathpix/templates/description.art |  21 +++
 3 files changed, 185 insertions(+)
 create mode 100644 lib/routes/mathpix/blog.ts
 create mode 100644 lib/routes/mathpix/namespace.ts
 create mode 100644 lib/routes/mathpix/templates/description.art

diff --git a/lib/routes/mathpix/blog.ts b/lib/routes/mathpix/blog.ts
new file mode 100644
index 000000000..dffa0e5c9
--- /dev/null
+++ b/lib/routes/mathpix/blog.ts
@@ -0,0 +1,155 @@
+import { type Data, type DataItem, type Route, ViewType } from '@/types';
+
+import { art } from '@/utils/render';
+import cache from '@/utils/cache';
+import ofetch from '@/utils/ofetch';
+import { parseDate } from '@/utils/parse-date';
+
+import { type CheerioAPI, type Cheerio, load } from 'cheerio';
+import type { Element } from 'domhandler';
+import { type Context } from 'hono';
+import path from 'node:path';
+
+export const handler = async (ctx: Context): Promise<Data> => {
+    const limit: number = Number.parseInt(ctx.req.query('limit') ?? '30', 10);
+
+    const baseUrl: string = 'https://mathpix.com';
+    const targetUrl: string = new URL('blog', baseUrl).href;
+
+    const response = await ofetch(targetUrl);
+    const $: CheerioAPI = load(response);
+    const language = $('html').attr('lang') ?? 'en';
+
+    const categoryMap = {};
+
+    $('div.navbar-menu a.blog-category').each((_, el) => {
+        const $el: Cheerio<Element> = $(el);
+
+        const id: string = $el.attr('data-category');
+        const name: string = $el.text()?.trim();
+
+        if (id && name) {
+            categoryMap[id] = name;
+        }
+    });
+
+    let items: DataItem[] = [];
+
+    items = $('li.articles__item')
+        .slice(0, limit)
+        .toArray()
+        .map((el): Element => {
+            const $el: Cheerio<Element> = $(el);
+
+            const title: string = $el.find('a.articles__title').text();
+            const image: string | undefined = $el.find('div.articles__image img').attr('srcset') ? new URL($el.find('div.articles__image img').attr('srcset') as string, baseUrl).href : undefined;
+            const description: string | undefined = art(path.join(__dirname, 'templates/description.art'), {
+                images: image
+                    ? [
+                          {
+                              src: image,
+                              alt: title,
+                          },
+                      ]
+                    : undefined,
+                intro: $el.find('div.articles__text').text(),
+            });
+            const pubDateStr: string | undefined = $el.find('time.articles__date').attr('datetime');
+            const linkUrl: string | undefined = $el.find('a.articles__title').attr('href');
+            const categoryIds: string[] =
+                $el
+                    .attr('data-category')
+                    ?.split(/,\s/)
+                    .map((category) => category.trim()) ?? [];
+            const categories: string[] = categoryIds.map((id) => categoryMap[id]).filter(Boolean);
+            const upDatedStr: string | undefined = pubDateStr;
+
+            const processedItem: DataItem = {
+                title,
+                description,
+                pubDate: pubDateStr ? parseDate(pubDateStr) : undefined,
+                link: linkUrl ? new URL(linkUrl, baseUrl).href : undefined,
+                category: categories,
+                content: {
+                    html: description,
+                    text: description,
+                },
+                image,
+                banner: image,
+                updated: upDatedStr ? parseDate(upDatedStr) : undefined,
+                language,
+            };
+
+            return processedItem;
+        });
+
+    items = await Promise.all(
+        items.map((item) => {
+            if (!item.link) {
+                return item;
+            }
+
+            return cache.tryGet(item.link, async (): Promise<DataItem> => {
+                const detailResponse = await ofetch(item.link);
+                const $$: CheerioAPI = load(detailResponse);
+
+                const title: string = $$('h1.article__title').text();
+                const description: string | undefined = $$('div#setText').html();
+
+                const processedItem: DataItem = {
+                    title,
+                    description,
+                    content: {
+                        html: description,
+                        text: description,
+                    },
+                    language,
+                };
+
+                return {
+                    ...item,
+                    ...processedItem,
+                };
+            });
+        })
+    );
+
+    return {
+        title: $('title').text(),
+        description: $('meta[property="og:description"]').attr('content'),
+        link: targetUrl,
+        item: items,
+        allowEmpty: true,
+        image: $('meta[property="og:image"]').attr('content'),
+        language,
+        id: $('meta[property="og:url"]').attr('content'),
+    };
+};
+
+export const route: Route = {
+    path: '/blog',
+    name: 'Blog',
+    url: 'mathpix.com',
+    maintainers: ['nczitzk'],
+    handler,
+    example: '/mathpix/blog',
+    parameters: undefined,
+    description: undefined,
+    categories: ['blog'],
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportRadar: true,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['mathpix.com/blog'],
+            target: '/blog',
+        },
+    ],
+    view: ViewType.Articles,
+};
diff --git a/lib/routes/mathpix/namespace.ts b/lib/routes/mathpix/namespace.ts
new file mode 100644
index 000000000..07c3464ad
--- /dev/null
+++ b/lib/routes/mathpix/namespace.ts
@@ -0,0 +1,9 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: 'Mathpix',
+    url: 'mathpix.com',
+    categories: ['blog'],
+    description: '',
+    lang: 'en',
+};
diff --git a/lib/routes/mathpix/templates/description.art b/lib/routes/mathpix/templates/description.art
new file mode 100644
index 000000000..249654e7e
--- /dev/null
+++ b/lib/routes/mathpix/templates/description.art
@@ -0,0 +1,21 @@
+{{ if images }}
+  {{ each images image }}
+    {{ if image?.src }}
+      <figure>
+        <img
+          {{ if image.alt }}
+            alt="{{ image.alt }}"
+          {{ /if }}
+        src="{{ image.src }}">
+      </figure>
+    {{ /if }}
+  {{ /each }}
+{{ /if }}
+
+{{ if intro }}
+  <blockquote>{{ intro }}</blockquote>
+{{ /if }}
+
+{{ if description }}
+  {{@ description }}
+{{ /if }}
\ No newline at end of file