From 85ee7a24f50aa3a0c6d31a1a1a5dbf3f872075ec Mon Sep 17 00:00:00 2001 From: Tony Date: Thu, 16 Feb 2023 20:41:10 +0100 Subject: [PATCH] fix(route): phoronix (#11892) --- docs/en/new-media.md | 6 ++ docs/en/traditional-media.md | 6 -- docs/new-media.md | 6 ++ docs/traditional-media.md | 6 -- lib/router.js | 2 +- lib/{routes => v2}/phoronix/index.js | 90 +++++++++++++--------------- lib/v2/phoronix/maintainer.js | 3 + lib/v2/phoronix/radar.js | 13 ++++ lib/v2/phoronix/router.js | 3 + 9 files changed, 75 insertions(+), 60 deletions(-) rename lib/{routes => v2}/phoronix/index.js (51%) create mode 100644 lib/v2/phoronix/maintainer.js create mode 100644 lib/v2/phoronix/radar.js create mode 100644 lib/v2/phoronix/router.js diff --git a/docs/en/new-media.md b/docs/en/new-media.md index afade523a..e8cd57ac1 100644 --- a/docs/en/new-media.md +++ b/docs/en/new-media.md @@ -562,6 +562,12 @@ This route provides a flexible plan with full text content to subscribe specific +## Phoronix + +### News & Reviews + + + ## Polar ### Blog diff --git a/docs/en/traditional-media.md b/docs/en/traditional-media.md index 829a880dc..46a0f5350 100644 --- a/docs/en/traditional-media.md +++ b/docs/en/traditional-media.md @@ -361,12 +361,6 @@ Only `s00017` is in English. -## Phoronix - -### News & Reviews - - - ## Radio Free Asia (RFA) ### News diff --git a/docs/new-media.md b/docs/new-media.md index 2d5fb6a80..88b828952 100644 --- a/docs/new-media.md +++ b/docs/new-media.md @@ -1057,6 +1057,12 @@ IPFS 网关有可能失效，那时候换成其他网关。 +## Phoronix + +### 新闻与评测 + + + ## PMCAFF ### 今日推荐 / 精选 diff --git a/docs/traditional-media.md b/docs/traditional-media.md index 2b8f500d2..478769c8b 100644 --- a/docs/traditional-media.md +++ b/docs/traditional-media.md @@ -343,12 +343,6 @@ pageClass: routes -## Phoronix - -### 新闻与评测 - - - ## RTHK 傳媒透視 diff --git a/lib/router.js b/lib/router.js index f9c6f8d06..e87e376a1 100644 --- a/lib/router.js +++ b/lib/router.js @@ -3378,7 +3378,7 @@ router.get('/uisdc/topic/:title?/:sort?', lazyloadRouteHandler('./routes/uisdc/t router.get('/chinalaborwatch/reports/:lang?/:industry?', lazyloadRouteHandler('./routes/chinalaborwatch/reports')); // Phoronix -router.get('/phoronix/:page/:queryOrItem?', lazyloadRouteHandler('./routes/phoronix/index')); +// router.get('/phoronix/:page/:queryOrItem?', lazyloadRouteHandler('./routes/phoronix/index')); // 美国中央情报局 router.get('/cia/foia-annual-report', lazyloadRouteHandler('./routes/us/cia/foia-annual-report')); diff --git a/lib/routes/phoronix/index.js b/lib/v2/phoronix/index.js similarity index 51% rename from lib/routes/phoronix/index.js rename to lib/v2/phoronix/index.js index 87e223ddf..4b60e1433 100644 --- a/lib/routes/phoronix/index.js +++ b/lib/v2/phoronix/index.js @@ -2,14 +2,17 @@ const parser = require('@/utils/rss-parser'); const cheerio = require('cheerio'); const got = require('@/utils/got'); +const baseUrl = 'https://www.phoronix.com'; + module.exports = async (ctx) => { - const rssUrl = new URL('/rss.php', 'https://www.phoronix.com'); - rssUrl.searchParams.set('page', ctx.params.page); - if (ctx.params.queryOrItem) { - if (ctx.params.page === 'category') { - rssUrl.searchParams.set('item', ctx.params.queryOrItem); + const { page, queryOrItem } = ctx.params; + const rssUrl = new URL('/rss.php', baseUrl); + rssUrl.searchParams.set('page', page); + if (queryOrItem) { + if (page === 'category') { + rssUrl.searchParams.set('item', queryOrItem); } else { - rssUrl.searchParams.set('q', ctx.params.queryOrItem); + rssUrl.searchParams.set('q', queryOrItem); } } @@ -23,58 +26,50 @@ module.exports = async (ctx) => { const $ = cheerio.load(html); const content = $('.content'); - // Maybe it's paginated - const links = []; - $('.pagination > a').each((_i, pager) => { - const link = pager.attribs.href; - links.push(link); - }); + // Author + const authorSelector = $('.author > a'); + // thel last 2 are the category and comments + const author = authorSelector + .slice(0, authorSelector.length - 2) + .toArray() + .map((e) => $(e).text()); - const pages = await Promise.all( - links.map((link) => - ctx.cache.tryGet(link, async () => { - const response = await got(item.link); - const html = response.body; - const $$ = cheerio.load(html); - const page = $$('.content'); - return page.html(); - }) - ) - ); - content.append(pages); + // Maybe it's paginated + const links = $('.pagination > a') + .toArray() + .map((pager) => `${baseUrl}${pager.attribs.href}`) + .slice(0, -1); // the last one is "next page" + + if (links.length) { + const pages = await Promise.all( + links.map((link) => + ctx.cache.tryGet(link, async () => { + const response = await got(link); + const html = response.data; + const $$ = cheerio.load(html); + const page = $$('.content'); + return page.html(); + }) + ) + ); + content.append(pages); + } // Summary const summary = $('.content > p:nth-child(1)'); - // Author - const author = []; - const authorSelector = $('.author > a'); - // thel last 2 are the category and comments - authorSelector.slice(0, authorSelector.length - 2).each((_i, e) => author.push(e.attribs.content)); - - // Remove unwanted DOMs - const unwanted_element_selectors = [ - '[class*="-advert"]', - '.social-share', - '.article-body-after', - 'scmp-chinaR2-early-text', - '.newsletter-widget-wrapper', - '[id*="-tracker"]', - '[class^="advert-"]', - 'amp-list', - '.more-on-this', - ]; - unwanted_element_selectors.forEach((selector) => { - content.find(selector).each((_i, e) => { - $(e).remove(); - }); + // High res images + content.find('img').each((_, img) => { + if (img.attribs.src.endsWith('_med')) { + img.attribs.src = img.attribs.src.replace('_med', '_show'); + } }); return { title: item.title, id: item.guid, pubDate: item.pubDate, - author: author.join(), + author: author.join(', '), link: item.link, summary: summary.html(), description: content.html(), @@ -92,6 +87,7 @@ module.exports = async (ctx) => { item: items, language: feed.language, icon: 'https://www.phoronix.com/android-chrome-192x192.png', + image: 'https://www.phoronix.com/android-chrome-192x192.png', logo: 'https://www.phoronix.com/phxcms7-css/phoronix.png', // Copied from thier web page metadata category: [ diff --git a/lib/v2/phoronix/maintainer.js b/lib/v2/phoronix/maintainer.js new file mode 100644 index 000000000..0ad8666e5 --- /dev/null +++ b/lib/v2/phoronix/maintainer.js @@ -0,0 +1,3 @@ +module.exports = { + '/:page/:queryOrItem?': ['oppliate'], +}; diff --git a/lib/v2/phoronix/radar.js b/lib/v2/phoronix/radar.js new file mode 100644 index 000000000..96e1af860 --- /dev/null +++ b/lib/v2/phoronix/radar.js @@ -0,0 +1,13 @@ +module.exports = { + 'phoronix.com': { + _name: 'Phoronix', + '.': [ + { + title: '新闻与评测', + docs: 'https://docs.rsshub.app/new-media.html#phoronix', + source: ['/*'], + target: '/phoronix/news', + }, + ], + }, +}; diff --git a/lib/v2/phoronix/router.js b/lib/v2/phoronix/router.js new file mode 100644 index 000000000..90fbc5e2a --- /dev/null +++ b/lib/v2/phoronix/router.js @@ -0,0 +1,3 @@ +module.exports = (router) => { + router.get('/:page/:queryOrItem?', require('./index')); +};