From 8ed999f1b59d776f7a79032e5eaa8f9f58643e33 Mon Sep 17 00:00:00 2001 From: Tony Date: Fri, 9 Dec 2022 22:41:37 +0500 Subject: [PATCH] fix(route): researchgate (#11429) --- docs/en/new-media.md | 6 --- docs/en/study.md | 6 +++ docs/new-media.md | 6 --- docs/study.md | 6 +++ lib/router.js | 2 +- lib/routes/researchgate/publications.js | 60 --------------------- lib/v2/researchgate/maintainer.js | 3 ++ lib/v2/researchgate/publications.js | 70 +++++++++++++++++++++++++ lib/v2/researchgate/radar.js | 13 +++++ lib/v2/researchgate/router.js | 3 ++ 10 files changed, 102 insertions(+), 73 deletions(-) delete mode 100644 lib/routes/researchgate/publications.js create mode 100644 lib/v2/researchgate/maintainer.js create mode 100644 lib/v2/researchgate/publications.js create mode 100644 lib/v2/researchgate/radar.js create mode 100644 lib/v2/researchgate/router.js diff --git a/docs/en/new-media.md b/docs/en/new-media.md index 73dcb4548..e2b4dea48 100644 --- a/docs/en/new-media.md +++ b/docs/en/new-media.md @@ -581,12 +581,6 @@ Compared to the official one, this feed: -## Research Gate - -### Publications - - - ## RSS3 ### Blog diff --git a/docs/en/study.md b/docs/en/study.md index 589148f79..2e5367d93 100644 --- a/docs/en/study.md +++ b/docs/en/study.md @@ -108,6 +108,12 @@ pageClass: routes +## ResearchGate + +### Publications + + + ## X-MOL ### News diff --git a/docs/new-media.md b/docs/new-media.md index 9acdc7d1a..4b85817d8 100644 --- a/docs/new-media.md +++ b/docs/new-media.md @@ -1187,12 +1187,6 @@ IPFS 网关有可能失效，那时候换成其他网关。 -## Research Gate - -### Publications - - - ## RSS3 ### Blog diff --git a/docs/study.md b/docs/study.md index ae233e10a..db8106c84 100644 --- a/docs/study.md +++ b/docs/study.md @@ -234,6 +234,12 @@ path="/ctfhub/upcoming/:limit?" +## ResearchGate + +### Publications + + + ## X-MOL 平台 ### 新闻 diff --git a/lib/router.js b/lib/router.js index 1a6dfbbdc..8e7bd61cd 100644 --- a/lib/router.js +++ b/lib/router.js @@ -4028,7 +4028,7 @@ router.get('/sbs/chinese/:category?/:id?/:dialect?/:language?', lazyloadRouteHan // router.get('/asiantolick/:category?/:keyword?', lazyloadRouteHandler('./routes/asiantolick')); // Research Gate -router.get('/researchgate/publications/:id', lazyloadRouteHandler('./routes/researchgate/publications')); +// router.get('/researchgate/publications/:id', lazyloadRouteHandler('./routes/researchgate/publications')); // QuestMobile router.get('/questmobile/report/:category?/:label?', lazyloadRouteHandler('./routes/questmobile/report')); diff --git a/lib/routes/researchgate/publications.js b/lib/routes/researchgate/publications.js deleted file mode 100644 index 6d469ea92..000000000 --- a/lib/routes/researchgate/publications.js +++ /dev/null @@ -1,60 +0,0 @@ -const got = require('@/utils/got'); -const cheerio = require('cheerio'); - -module.exports = async (ctx) => { - const id = ctx.params.id; - - const rootUrl = 'https://www.researchgate.net'; - const currentUrl = `${rootUrl}/profile/${id}`; - const response = await got({ - method: 'get', - url: currentUrl, - }); - - const $ = cheerio.load(response.data); - - const list = $('div[itemprop="headline"] a') - .slice(0, 15) - .map((_, item) => { - item = $(item); - return { - title: item.text(), - link: item.attr('href'), - }; - }) - .get(); - - const items = await Promise.all( - list.map( - async (item) => - await ctx.cache.tryGet(item.link, async () => { - const detailResponse = await got({ - method: 'get', - url: item.link, - }); - const content = cheerio.load(detailResponse.data); - - item.doi = content('meta[property="citation_doi"]').attr('content'); - item.pubDate = Date.parse(content('meta[property="citation_publication_date"]').attr('content')); - - const authors = []; - - content('meta[property="citation_author"]').each(function () { - authors.push(content(this).attr('content')); - }); - - item.author = authors.join(', '); - - item.description = content('div[itemprop="description"]').html(); - - return item; - }) - ) - ); - - ctx.state.data = { - title: `${$('meta[property="profile:username"]').attr('content')}'s Publications - ResearchGate`, - link: currentUrl, - item: items, - }; -}; diff --git a/lib/v2/researchgate/maintainer.js b/lib/v2/researchgate/maintainer.js new file mode 100644 index 000000000..1422d9e67 --- /dev/null +++ b/lib/v2/researchgate/maintainer.js @@ -0,0 +1,3 @@ +module.exports = { + '/publications/:username': ['nczitzk'], +}; diff --git a/lib/v2/researchgate/publications.js b/lib/v2/researchgate/publications.js new file mode 100644 index 000000000..8451f6c67 --- /dev/null +++ b/lib/v2/researchgate/publications.js @@ -0,0 +1,70 @@ +const cheerio = require('cheerio'); +const { parseDate } = require('@/utils/parse-date'); + +module.exports = async (ctx) => { + const id = ctx.params.id; + + const rootUrl = 'https://www.researchgate.net'; + const currentUrl = `${rootUrl}/profile/${id}`; + const browser = await require('@/utils/puppeteer')(); + const page = await browser.newPage(); + await page.setRequestInterception(true); + page.on('request', (request) => { + request.resourceType() === 'document' || request.resourceType() === 'script' ? request.continue() : request.abort(); + }); + await page.goto(currentUrl); + const response = await page.evaluate(() => document.documentElement.innerHTML); + await page.close(); + + const $ = cheerio.load(response); + + const list = $('div[itemprop="headline"] a') + .toArray() + .slice(0, ctx.query.limit ? Number(ctx.query.limit) : 15) + .map((item) => { + item = $(item); + return { + title: item.text(), + link: item.attr('href'), + }; + }); + + const items = await Promise.all( + list.map((item) => + ctx.cache.tryGet(item.link, async () => { + const page = await browser.newPage(); + await page.setRequestInterception(true); + page.on('request', (request) => { + request.resourceType() === 'document' || request.resourceType() === 'script' ? request.continue() : request.abort(); + }); + await page.goto(item.link); + const detailResponse = await page.evaluate(() => document.documentElement.innerHTML); + await page.close(); + const content = cheerio.load(detailResponse); + + item.doi = content('meta[property="citation_doi"]').attr('content'); + item.pubDate = parseDate(content('meta[property="citation_publication_date"]').attr('content')); + + const authors = []; + + content('meta[property="citation_author"]').each(function () { + authors.push(content(this).attr('content')); + }); + + item.author = authors.join(', '); + + item.description = content('div[itemprop="description"]').html(); + + return item; + }) + ) + ); + + await browser.close(); + + ctx.state.data = { + title: `${$('meta[property="profile:username"]').attr('content')}'s Publications - ResearchGate`, + link: currentUrl, + item: items, + }; +}; diff --git a/lib/v2/researchgate/radar.js b/lib/v2/researchgate/radar.js new file mode 100644 index 000000000..4df44cdec --- /dev/null +++ b/lib/v2/researchgate/radar.js @@ -0,0 +1,13 @@ +module.exports = { + 'researchgate.net': { + _name: 'ResearchGate', + '.': [ + { + title: 'Publications', + docs: 'https://docs.rsshub.app/study.html#researchgate', + source: ['/profile/:username'], + target: '/researchgate/publications/:username', + }, + ], + }, +}; diff --git a/lib/v2/researchgate/router.js b/lib/v2/researchgate/router.js new file mode 100644 index 000000000..ae42543bf --- /dev/null +++ b/lib/v2/researchgate/router.js @@ -0,0 +1,3 @@ +module.exports = (router) => { + router.get('/publications/:id', require('./publications')); +};