feat(route): techpowerup (#12551)

* feat(route): techpowerup

* feat(route): techpowerup
This commit is contained in:
Tony 2023-05-22 20:51:12 +05:00 committed by GitHub
parent 83476621d8
commit dd2be12722
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
7 changed files with 260 additions and 0 deletions

View File

@ -772,6 +772,16 @@ Compared to the official one, this feed:
<RouteEn author="NavePnow" example="/techcrunch/news" path="/techcrunch/news"/>
## TechPowerUp
### Latest Content
<RouteEn author="TonyRL" example="/techpowerup" path="/techpowerup" radar="1" />
### Reviews
<RouteEn author="TonyRL" example="/techpowerup/review/4090" path="/techpowerup/review/:keyword?" :paramsDesc="['Search Keyword']" radar="1" />
## The Brain
### Blog

View File

@ -0,0 +1,69 @@
const got = require('@/utils/got');
const cheerio = require('cheerio');
const { parseDate } = require('@/utils/parse-date');
const { baseUrl, headers, fixImages, parseReviews } = require('./utils');
module.exports = async (ctx) => {
const { data: response } = await got(baseUrl, {
headers,
});
const $ = cheerio.load(response);
const list = $('.newspost')
.toArray()
.map((item) => {
item = $(item);
const a = item.find('h1 a');
const date = item.find('time').attr('datetime');
return {
title: a.text(),
link: baseUrl + a.attr('href'),
pubDate: date ? parseDate(date) : null, // 2023-05-21T16:05:14+00:00
author: item.find('.byline address').text(),
category: item
.find('.byline .flags span')
.toArray()
.map((item) => $(item).text().trim()),
};
});
const items = await Promise.all(
list.map((item) =>
ctx.cache.tryGet(item.link, async () => {
const { data: response } = await got(item.link, {
headers,
});
const $ = cheerio.load(response);
fixImages($);
if (item.link.includes('/review/')) {
await parseReviews($, item);
return item;
}
// news
item.description = $('.newspost .text').html();
item.category = [
...new Set([
...item.category,
...$('.tags li a')
.toArray()
.map((item) => $(item).text()),
]),
];
return item;
})
)
);
ctx.state.data = {
title: 'TechPowerUp',
link: baseUrl,
language: 'en',
image: 'https://tpucdn.com/apple-touch-icon-v1684568903519.png',
item: items,
};
};

View File

@ -0,0 +1,5 @@
module.exports = {
'': ['TonyRL'],
'/': ['TonyRL'],
'/review/:keyword?': ['TonyRL'],
};

View File

@ -0,0 +1,19 @@
module.exports = {
'techpowerup.com': {
_name: 'TechPowerUp',
'.': [
{
title: 'Latest Content',
docs: 'https://docs.rsshub.app/en/new-media.html#techpowerup',
source: ['/'],
target: '/techpowerup',
},
{
title: 'Reviews',
docs: 'https://docs.rsshub.app/en/new-media.html#techpowerup',
source: ['/review/search', '/review'],
target: (_, url) => `/techpowerup/review${new URL(url).searchParams.has('q') ? `/${new URL(url).searchParams.get('q')}` : ''}`,
},
],
},
};

View File

@ -0,0 +1,69 @@
const got = require('@/utils/got');
const cheerio = require('cheerio');
const { parseDate } = require('@/utils/parse-date');
const { baseUrl, headers, fixImages, parseReviews } = require('./utils');
module.exports = async (ctx) => {
const { keyword } = ctx.params;
const url = new URL(`${baseUrl}/review/${keyword ? 'search/' : ''}`);
if (keyword) {
url.searchParams.set('q', keyword);
url.searchParams.set('_', Date.now());
}
const { data: response } = await got(url, {
headers,
});
const $ = cheerio.load(response);
const list = $('.reviewlist-bit')
.toArray()
.map((item) => {
item = $(item);
const a = item.find('.title a');
return {
title: a.text(),
link: baseUrl + a.attr('href'),
pubDate: parseDate(item.find('.date time').attr('datetime')), // 2023-05-21T16:05:14+00:00
author: item
.find('.author')
.contents()
.filter((_, c) => c.type === 'text')
.text()
.trim(),
category: item
.find('.category')
.contents()
.filter((_, c) => c.type === 'text')
.text()
.trim(),
};
});
const items = await Promise.all(
list.map((item) =>
ctx.cache.tryGet(item.link, async () => {
const { data: response } = await got(item.link, {
headers,
});
const $ = cheerio.load(response);
fixImages($);
await parseReviews($, item);
return item;
})
)
);
ctx.state.data = {
title: 'Reviews | TechPowerUp',
link: url.href,
language: 'en',
image: 'https://tpucdn.com/apple-touch-icon-v1684568903519.png',
item: items,
};
};

View File

@ -0,0 +1,4 @@
module.exports = (router) => {
router.get('/', require('./index'));
router.get('/review/:keyword?', require('./review'));
};

View File

@ -0,0 +1,84 @@
const got = require('@/utils/got');
const cheerio = require('cheerio');
const { parseDate } = require('@/utils/parse-date');
const md5 = require('@/utils/md5');
const baseUrl = 'https://www.techpowerup.com';
const headers = {
cookie: `botcheck=${md5(Date.now().toString())}`,
};
const fixImages = ($) => {
$('div.responsive-image-xx').each((_, d) => {
removeResponsiveStyle(d);
});
$('figure').each((_, f) => {
removeFigureStyle(f);
});
$('.newspost img').each((_, img) => {
hdImage(img);
});
};
const hdImage = (img) => {
img.attribs.src = img.attribs.src.replace('_thm', '').replace('_small', '');
if (img.parentNode.name === 'a' && img.parentNode.attribs['data-width'] && img.parentNode.attribs['data-height']) {
img.attribs.width = img.parentNode.attribs['data-width'];
img.attribs.height = img.parentNode.attribs['data-height'];
}
};
const removeFigureStyle = (f) => {
delete f.attribs.style;
};
const removeResponsiveStyle = (div) => {
delete div.attribs.style;
};
const parseReviews = async ($, item) => {
const { review } = JSON.parse($('script[type="application/ld+json"]').text());
const content = $('.text');
const nextPages = $('#pagesel option')
.toArray()
.map((a) => `${baseUrl}${a.attribs.value}`)
.slice(1, -1);
if (nextPages.length) {
const pages = await Promise.all(
nextPages.map(async (url) => {
const { data: response } = await got(url, {
headers,
});
const $ = cheerio.load(response);
$('.text div.responsive-image-xx').each((_, d) => {
removeResponsiveStyle(d);
});
$('.text figure').each((_, f) => {
removeFigureStyle(f);
});
$('.text img').each((_, img) => {
hdImage(img);
});
return $('.text').html();
})
);
content.append(pages);
}
item.author = review.author.name;
item.pubDate = parseDate(review.datePublished);
item.updated = parseDate(review.dateModified);
item.description = content.html();
};
module.exports = {
baseUrl,
headers,
fixImages,
hdImage,
parseReviews,
removeFigureStyle,
removeResponsiveStyle,
};