From 35b9fa168e2c9d33b321aa71a5386e829b5decb6 Mon Sep 17 00:00:00 2001 From: WilliamGates <3852641+williamgateszhao@users.noreply.github.com> Date: Fri, 12 Sep 2025 10:04:48 +0800 Subject: [PATCH] fix(route): uber (#20034) * fix route /uber/blog * refactor(route): Simplify JSON parsing and fix typo in description --- lib/routes/uber/blog.ts | 133 ++++++++++++++++++++++++++-------------- 1 file changed, 87 insertions(+), 46 deletions(-) diff --git a/lib/routes/uber/blog.ts b/lib/routes/uber/blog.ts index 1fbb51226..985e51d11 100644 --- a/lib/routes/uber/blog.ts +++ b/lib/routes/uber/blog.ts @@ -1,16 +1,17 @@ import { Route } from '@/types'; import cache from '@/utils/cache'; -import got from '@/utils/got'; +import ofetch from '@/utils/ofetch'; +import { load } from 'cheerio'; import { parseDate } from '@/utils/parse-date'; const rootURL = 'https://www.uber.com'; -const apiURL = 'https://blogapi.uber.com'; export const route: Route = { - path: '/blog/:maxPage?', + // `compat` is a never used parameter + // just for backward compatibility with the deprecated `:maxPage` parameter + path: '/blog/:compat?', categories: ['blog'], example: '/uber/blog', - parameters: { maxPage: 'max number of pages to retrieve, default to 1 page at most' }, features: { requireConfig: false, requirePuppeteer: false, @@ -21,65 +22,105 @@ export const route: Route = { }, radar: [ { - source: ['www.uber.com/:language/blog/engineering', 'www.uber.com/:language/blog'], + source: ['www.uber.com/:language/blog/engineering'], target: '/blog', }, ], name: 'Engineering', maintainers: ['hulb'], handler, - url: 'www.uber.com/blog/pittsburgh/engineering', + url: 'www.uber.com/en-HK/blog/engineering', + description: + "The English blog on any of Uber's regional sites (e.g., www.uber.com/en-JP/blog) is the same engineering blog provided by this route, so language selection is not supported. This route is not for the public news blog on specific regional sites (e.g., www.uber.com/ja-JP/blog).", + zh: { + description: 'uber的任何区域站点的英文blog（例如www.uber.com/en-JP/blog）都是相同的内容，正是本路由提供的engineering blog，因此本路由不提供语言选择；本路由不是uber在特定区域站点的公开新闻blog（例如www.uber.com/ja-JP/blog)', + }, }; -async function handler(ctx) { - let maxPage = Number(ctx.req.param('maxPage')); - if (Number.isNaN(maxPage)) { - maxPage = 1; - } - - let pages = await Promise.all( - [...Array.from({ length: maxPage }).keys()].map((pageIdx) => - got(`${apiURL}/wp-json/blog/v1/data`, { - searchParams: { - page: pageIdx + 1, - parent: 'pittsburgh', - slug: 'engineering', - }, - }) - ) - ); - pages = pages.map((page) => page.data); +async function handler() { + const response = await ofetch(`${rootURL}/en-HK/blog/engineering/rss/`, { + // The source site is misconfigured or intentionally blocking requests without a specific accept header + // Without this header, it will return an HTTP 406 error + // Note that the accept type must be 'text/html'; 'application/xml' or similar will get HTTP 404 error + headers: { + accept: 'text/html', + }, + // Without this, ofetch will parse the response as a blob instead of text, which cannot be loaded by cheerio + parseResponse: (txt) => txt, + }); + const $ = load(response, { xmlMode: true }); const result = await Promise.all( - pages.map((page) => - Promise.all( - page.posts.map((post) => - cache.tryGet(`${rootURL}${post.link}`, async () => { - let { data: article } = await got(`${apiURL}/wp-json/blog/v1/data`, { - searchParams: { - slug: post.link.replaceAll('/', '').replace('blog', ''), - }, - }); - article = article.article; + $('item') + .toArray() + .map((el) => + cache.tryGet($(el).find('link').text(), async () => { + const detailResponse = await ofetch($(el).find('link').text(), { + headers: { + accept: 'text/html', + }, + }); + const detail = load(detailResponse); - return { - link: article.link, - title: article.title, - description: article.content, - pubDate: parseDate(article.created), - author: article.author, - category: article.categories.map((category) => category.category_name), - }; - }) - ) + const scriptText = detail('script#__REDUX_STATE__').text().trim(); + // The json in the script element is over-encoded + // It needs to be decoded this way before it can be parsed by JSON.parse + const jsonText = decodeURIComponent(JSON.parse(`"${scriptText}"`)); + // Traverse the JSON to find the content node, which is more robust against format changes. + const contentHtml = findNode(JSON.parse(jsonText), { idKey: 'id', idValue: 'BlogArticleContent', siblingKey: 'props', childKey: 'content' }).replaceAll(String.raw`\n`, ''); + + return { + link: $(el).find('link').text(), + title: $(el).find('title').text(), + description: contentHtml, + pubDate: parseDate($(el).find('pubDate').text()), + category: $(el) + .find('category') + .toArray() + .map((item) => $(item).text()), + }; + }) ) - ) ); return { title: `Uber Engineering Blog`, link: rootURL + '/blog/engineering', description: 'The technology behind Uber Engineering', - item: result.flat(), + item: result, }; } + +function findNode( + json: any, + options: { + idKey?: string; + idValue: string; + siblingKey: string; + childKey: string; + } +): any { + const { idKey = 'id', idValue, siblingKey, childKey } = options; + + if (Array.isArray(json)) { + for (const item of json) { + const result = findNode(item, options); + if (result !== undefined) { + return result; + } + } + } else if (json && typeof json === 'object') { + if (json[idKey] === idValue) { + return json[siblingKey]?.[childKey]; + } + + for (const key in json) { + const result = findNode(json[key], options); + if (result !== undefined) { + return result; + } + } + } + + return undefined; +}