feat(route/cloudflare): add full-text blog feed and fix status feed

lib/routes/cloudflare/blog.ts

@@ -0,0 +1,106 @@
+import { load } from 'cheerio';
+import pMap from 'p-map';
+import Parser from 'rss-parser';
+
+import type { Data, Route } from '@/types';
+import { ViewType } from '@/types';
+import cache from '@/utils/cache';
+import ofetch from '@/utils/ofetch';
+import { parseDate } from '@/utils/parse-date';
+
+const parser = new Parser<Pick<Data, 'language'>, { creators?: string[] }>({
+    customFields: {
+        item: [['dc:creator', 'creators', { keepArray: true }]],
+    },
+});
+
+export const route: Route = {
+    path: '/blog',
+    name: 'Blog',
+    url: 'blog.cloudflare.com',
+    example: '/cloudflare/blog',
+    categories: ['programming'],
+    maintainers: ['ljh12138164'],
+    description:
+        'Uses the [official RSS feed](https://blog.cloudflare.com/rss/) for article metadata and fetches full articles from [Cloudflare Blog](https://blog.cloudflare.com/), including inline images, code blocks, and copyable prompts.',
+    zh: {
+        description: '来源：[Cloudflare Blog](https://blog.cloudflare.com/)。使用[官方 RSS](https://blog.cloudflare.com/rss/)获取文章列表、作者、日期和标签，再抓取并缓存网页正文，补齐图片、代码块和可复制的提示词。',
+    },
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportRadar: true,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['blog.cloudflare.com'],
+            target: '/blog',
+        },
+    ],
+    view: ViewType.Articles,
+    handler,
+};
+
+async function handler(): Promise<Data> {
+    const feedXml = await ofetch('https://blog.cloudflare.com/rss/', { responseType: 'text' });
+    const feed = await parser.parseString(feedXml);
+
+    return {
+        title: feed.title!,
+        description: feed.description,
+        link: feed.link,
+        image: feed.image?.url,
+        language: feed.language,
+        item: await pMap(
+            feed.items,
+            async (item) => ({
+                title: item.title!,
+                link: item.link,
+                description: await cache.tryGet(item.link!, () => fetchArticle(item.link!)),
+                pubDate: item.pubDate ? parseDate(item.pubDate) : undefined,
+                author: item.creators?.join(', '),
+                category: item.categories,
+                guid: item.guid,
+                enclosure_url: item.enclosure?.url,
+                enclosure_type: item.enclosure?.type,
+                // rss-parser returns XML strings despite declaring enclosure lengths as numbers.
+                // oxlint-disable-next-line typescript/no-unnecessary-type-conversion
+                enclosure_length: item.enclosure ? Number(item.enclosure.length) : undefined,
+            }),
+            { concurrency: 3 }
+        ),
+    };
+}
+
+async function fetchArticle(link: string): Promise<string> {
+    const response = await ofetch(link);
+    const $ = load(response);
+    const article = $('.article-content').first();
+    if (!article.length) {
+        throw new Error(`Cloudflare Blog article content not found: ${link}`);
+    }
+
+    article.find('[data-copy-prompt]').each((_, element) => {
+        // The copy button stores the full prompt as a JSON string.
+        const prompt = JSON.parse($(element).find('[data-prompt]').attr('data-prompt')!);
+        $(element).replaceWith($('<pre>').append($('<code>').text(prompt)));
+    });
+
+    article.find('script, style, button, img[data-image-placeholder]').remove();
+    article.find('[srcset]').removeAttr('srcset');
+    article.find('[src], [href], [poster]').each((_, element) => {
+        const $element = $(element);
+        for (const attribute of ['src', 'href', 'poster']) {
+            const value = $element.attr(attribute);
+            if (value) {
+                $element.attr(attribute, new URL(value, link).href);
+            }
+        }
+    });
+
+    return article.html()!;
+}

lib/routes/cloudflare/namespace.ts

@@ -0,0 +1,8 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: 'Cloudflare',
+    url: 'cloudflare.com',
+    categories: ['programming'],
+    lang: 'en',
+};

lib/routes/cloudflare/status.tsx

@@ -0,0 +1,130 @@
+import { load } from 'cheerio';
+import type { Context } from 'hono';
+import { raw } from 'hono/html';
+import { renderToString } from 'hono/jsx/dom/server';
+import MarkdownIt from 'markdown-it';
+
+import { config } from '@/config';
+import type { Data, DataItem, Route } from '@/types';
+import { ViewType } from '@/types';
+import ofetch from '@/utils/ofetch';
+import { parseDate } from '@/utils/parse-date';
+
+interface IncidentUpdate {
+    status: string;
+    body: string;
+    display_at: string;
+}
+
+interface Incident {
+    id: string;
+    name: string;
+    incident_updates: IncidentUpdate[];
+}
+
+interface StatusResponse {
+    page: { name: string };
+    incidents: Incident[];
+    scheduled_maintenances: Incident[];
+}
+
+const baseUrl = 'https://www.cloudflarestatus.com';
+const md = MarkdownIt({ html: true, linkify: true });
+
+function createItem(incident: Incident, update: IncidentUpdate, index: number) {
+    const previousUpdate = incident.incident_updates[index + 1];
+    const type = update.status === 'scheduled' && incident.incident_updates.length === 1 ? '' : update.status === previousUpdate?.status ? 'Update' : update.status.replaceAll('_', ' ').replace(/^./, (char) => char.toUpperCase());
+    const body = md.renderInline(update.body.replaceAll(/\r?\n/g, '<br>'));
+    const text = load(body, null, false).text();
+    // Statuspage's HTML timestamps use second precision; preserve existing GUIDs.
+    const timestamp = Math.floor(parseDate(update.display_at).getTime() / 1000) * 1000;
+    const pubDate = parseDate(timestamp, 'x');
+    const link = `${baseUrl}/incidents/${incident.id}`;
+    const guid = `${link}#${timestamp}`;
+    const description = renderToString(
+        <>
+            <h2>{incident.name}</h2>
+            {type ? (
+                <>
+                    <strong>{type}</strong>
+                    {' - '}
+                </>
+            ) : null}
+            <span class="whitespace-pre-wrap">{raw(body)}</span>
+            <br />
+            <small>
+                <span class="ago" data-datetime-unix={timestamp} />
+                {pubDate.toLocaleDateString('en-US', { month: 'short', day: 'numeric', year: 'numeric', timeZone: 'UTC' })}
+                {' - '}
+                {pubDate.toISOString().slice(11, 16)} UTC
+            </small>
+        </>
+    );
+
+    return {
+        title: `${type ? `${type} - ` : ''}${text}`,
+        description,
+        pubDate,
+        link,
+        category: type ? [type] : [],
+        guid,
+        id: guid,
+        content: { html: description, text: description },
+        updated: pubDate,
+        language: 'en',
+    } satisfies DataItem;
+}
+
+export const handler = async (ctx: Context): Promise<Data> => {
+    const limit = Number(ctx.req.query('limit') ?? '100');
+    const response = await ofetch<StatusResponse>(`${baseUrl}/api/v2/summary.json`, {
+        headers: { 'User-Agent': config.trueUA },
+    });
+    const items = [...response.incidents, ...response.scheduled_maintenances]
+        .flatMap((incident) => incident.incident_updates.map((update, index) => createItem(incident, update, index)))
+        .toSorted((a, b) => b.pubDate.getTime() - a.pubDate.getTime())
+        .slice(0, limit);
+
+    return {
+        title: response.page.name,
+        description: 'Real-time status and incident history for Cloudflare services, network locations, and scheduled maintenance.',
+        link: baseUrl,
+        item: items,
+        allowEmpty: true,
+        language: 'en',
+        id: baseUrl,
+    };
+};
+
+export const route: Route = {
+    path: '/status',
+    name: 'Status',
+    url: 'www.cloudflarestatus.com',
+    maintainers: ['nczitzk', 'ljh12138164'],
+    handler,
+    example: '/cloudflare/status',
+    parameters: undefined,
+    description:
+        'Uses the [official API](https://www.cloudflarestatus.com/api) to provide each update for ongoing incidents and upcoming or active maintenance from [Cloudflare Status](https://www.cloudflarestatus.com/) as a separate item. The legacy route `/cloudflarestatus` redirects here.',
+    zh: {
+        description:
+            '来源：[Cloudflare Status](https://www.cloudflarestatus.com/)。使用[官方 API](https://www.cloudflarestatus.com/api)获取未解决故障及计划中、进行中的维护通知，每次更新单独生成一条订阅内容。旧路由 `/cloudflarestatus` 会重定向到此路由。',
+    },
+    categories: ['programming'],
+    features: {
+        requireConfig: false,
+        requirePuppeteer: false,
+        antiCrawler: false,
+        supportRadar: true,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['www.cloudflarestatus.com'],
+            target: '/status',
+        },
+    ],
+    view: ViewType.Notifications,
+};

lib/routes/cloudflarestatus/index.tsx

@@ -1,105 +1,14 @@
-import type { Cheerio, CheerioAPI } from 'cheerio';
-import { load } from 'cheerio';
-import type { Element } from 'domhandler';
-import type { Context } from 'hono';
-import { raw } from 'hono/html';
-import { renderToString } from 'hono/jsx/dom/server';
-
-import type { Data, DataItem, Language, Route } from '@/types';
-import { ViewType } from '@/types';
-import ofetch from '@/utils/ofetch';
-import { parseDate } from '@/utils/parse-date';
-
-export const handler = async (ctx: Context): Promise<Data> => {
-    const limit = Number(ctx.req.query('limit') ?? '100');
-
-    const baseUrl = 'https://www.cloudflarestatus.com';
-
-    const response = await ofetch(baseUrl);
-    const $: CheerioAPI = load(response);
-    const language = ($('html').attr('lang') ?? 'en') as Language;
-
-    $('div.incidents-list').remove();
-
-    const items: DataItem[] = $('div.update')
-        .slice(0, limit)
-        .toArray()
-        .map((el) => {
-            const $el: Cheerio<Element> = $(el);
-
-            const $actualTitleEl: Cheerio<Element> = $el.parent().parent().find('a');
-            const actualTitle: string = $actualTitleEl.text();
-            const type: string = $el.find('strong').first().text();
-            const text: string = $el.find('span.whitespace-pre-wrap').first().text();
-
-            const title = `${type ? `${type} - ` : ''}${text}`;
-            const description: string | undefined = renderToString(
-                <>
-                    {actualTitle ? <h2>{actualTitle}</h2> : null}
-                    {$el.html() ? raw($el.html()) : null}
-                </>
-            );
-            const pubDateStr: string | undefined = $el.find('span.ago').attr('data-datetime-unix');
-            const href: string | undefined = $actualTitleEl.attr('href');
-            const linkUrl: string | undefined = href ? new URL(href, baseUrl).href : undefined;
-            const categories: string[] = [type].filter(Boolean);
-            const guid: string = linkUrl ? `${linkUrl}#${pubDateStr}` : '';
-            const upDatedStr: string | undefined = pubDateStr;
-
-            const processedItem: DataItem = {
-                title,
-                description,
-                pubDate: pubDateStr ? parseDate(pubDateStr, 'x') : undefined,
-                link: linkUrl,
-                category: categories,
-                guid,
-                id: guid,
-                content: {
-                    html: description,
-                    text: description,
-                },
-                updated: upDatedStr ? parseDate(upDatedStr, 'x') : undefined,
-                language,
-            };
-
-            return processedItem;
-        });
-
-    return {
-        title: $('title').text(),
-        description: $('meta[name="description"]').attr('content'),
-        link: baseUrl,
-        item: items,
-        allowEmpty: true,
-        language,
-        id: baseUrl,
-    };
-};
+import type { Route } from '@/types';
 
 export const route: Route = {
     path: '/',
-    name: 'Status',
+    name: 'Status (legacy)',
     url: 'www.cloudflarestatus.com',
-    maintainers: ['nczitzk'],
-    handler,
+    maintainers: ['nczitzk', 'ljh12138164'],
     example: '/cloudflarestatus',
-    parameters: undefined,
-    description: undefined,
-    categories: ['programming'],
-    features: {
-        requireConfig: false,
-        requirePuppeteer: false,
-        antiCrawler: false,
-        supportRadar: true,
-        supportBT: false,
-        supportPodcast: false,
-        supportScihub: false,
+    description: 'This route redirects to `/cloudflare/status`.',
+    zh: {
+        description: '此旧路由会重定向到 `/cloudflare/status`。',
     },
-    radar: [
-        {
-            source: ['www.cloudflarestatus.com'],
-            target: '/',
-        },
-    ],
-    view: ViewType.Notifications,
+    handler: (ctx) => ctx.set('redirect', `/cloudflare/status${new URL(ctx.req.url).search}`),
 };