opensource/RSSHub
1
0
Fork 0
mirror of https://github.com/DIYgod/RSSHub.git synced 2025-12-01 09:38:04 +08:00
Code Issues Packages Projects Releases Wiki Activity

feat(route): DevolverDigital Blog: Parse per page instead of full page, and add cache support. (#12560)

Browse Source 
* get each post instead of the whole page

* add cache support

* Update blog.js

* Update blog.js
This commit is contained in:
TheLittle_Yang
2023-05-26 12:18:28 +08:00
committed by GitHub
parent d824a628a3
commit 4e2c3efe2d
1 changed files with 31 additions and 23 deletions
Download Patch File Download Diff File Expand all files Collapse all files

54
lib/v2/devolverdigital/blog.js
View File

@@ -3,39 +3,47 @@ const cheerio = require('cheerio');
module.exports = async (ctx) => {
const baseUrl = 'https://www.devolverdigital.com/blog';
const { data: response } = await got(baseUrl);
const $ = cheerio.load(response);
$('noscript').remove();
const nextData = JSON.parse($('#__NEXT_DATA__').text());
const allBlogContents = $('div.flex > div > div > div > div:not([class])');
const items = nextData.props.pageProps.posts.map((post, postIndex) => {
// img resource redirection and
// clean up absolute layouts for img and span
const items = await Promise.all(
nextData.props.pageProps.posts.map((postData) => {
const postUrl = `${baseUrl}/post/${postData.id}`;
return ctx.cache.tryGet(postUrl, async () => {
const { data: postPage } = await got(postUrl);
const imageUrls = post.body.filter((item) => item.type === 'upload' && item.value.cloudinary.resource_type === 'image').map((item) => item.value.cloudinary.secure_url);
const $page = cheerio.load(postPage);
$page('noscript').remove();
const postContent = $page('div.flex > div > div > div > div:not([class])');
const allImageSpans = $(allBlogContents[postIndex]).find('span > img').parent();
// img resource redirection and
// clean up absolute layouts for img and span
const imageUrls = postData.body.filter((item) => item.type === 'upload' && item.value.cloudinary.resource_type === 'image').map((item) => item.value.cloudinary.secure_url);
const allImageSpans = postContent.find('span > img').parent();
allImageSpans.each((spanIndex, span) => {
$(span).attr('style', $(span).attr('style').replace('position:absolute', ''));
const img = $(span).find('img');
img.attr('src', imageUrls[spanIndex]);
img.attr('style', img.attr('style').replace('position:absolute', '').replace('width:0', '').replace('height:0', ''));
});
allImageSpans.each((spanIndex, span) => {
$(span).attr('style', $(span).attr('style').replace('position:absolute', ''));
const img = $(span).find('img');
img.attr('src', imageUrls[spanIndex]);
img.attr('style', img.attr('style').replace('position:absolute', '').replace('width:0', '').replace('height:0', ''));
});
return {
title: post.title,
author: post.author,
pubDate: Date.parse(post.createdAt),
description: $(allBlogContents[postIndex]).html(),
};
});
return {
title: postData.title,
link: postUrl,
author: postData.author,
pubDate: Date.parse(postData.createdAt),
updated: Date.parse(postData.updatedAt),
description: postContent.html(),
};
});
})
);
ctx.state.data = {
title: 'DevolverDigital Blog',
language: 'en-us',
link: 'https://www.devolverdigital.com/blog',
item: items,
};