rssssing/lib.js

609 lines
14 KiB
JavaScript

// Ascii font used is "Shimrod"
import Path from "path"
import FS from "fs/promises"
import { JSDOM } from "jsdom"
let cache = await FS.readFile('./cache.json', { encoding: 'utf-8' })
.then(json => JSON.parse(json) )
// | o |
// . . |- . | ,-.
// | | | | | `-.
// `-` `-' ' ' `-'
export function PromiseBatch() {
let promises = []
this.add = (promise) =>
promises.push(promise)
this.complete = () =>
Promise.all(promises)
}
export const annotate = annotation =>
data => {
console.log(annotation),
data
}
export const write = async (path, content) => {
let dir = Path.dirname(path)
try {
await FS.access(dir)
} catch(e) {
await FS.mkdir(dir, { recursive: true })
}
return await FS.writeFile(path, content)
}
export const download = async (url, path, courtesyWait) => {
let response = await delayedFetch(url, {}, courtesyWait)
.catch(err => console.error(`Failed download of ${url}:`, err) )
if(response.ok) {
await write(path, response.body)
} else {
throw createNetworkingError(response)
}
}
export const createNetworkingError = response => {
return new Error(`Request failed, ${response.status}: ${response.statusText}`)
}
export const getLinkExtname = link =>
Path.extname(new URL(link).pathname)
export const buildImagePathHandler = (source, id) => (url, i, array) => {
let path = `images/${source.name}-${id}`
if(array.length > 1)
path += `-${i}`
return path + getLinkExtname(url)
}
export const writeStylesheet = (path, { directory, batch }) =>
batch.add(
FS.readFile(path)
.then(content => write(Path.join(directory, 'style.css'), content))
)
export const postIdFromPathname = post => {
let { pathname } = new URL(post.link)
return pathname.slice(pathname.lastIndexOf('/') + 1)
}
export const createLock = async (path) => {
let lockExists = false
try {
await FS.access(path)
lockExists = true
} catch(err) {
lockExists = false
}
let lock = {
sources: {},
lists: {}
}
if(lockExists) {
Object.assign(lock, JSON.parse(await FS.readFile(path, { encoding: 'utf8' })))
}
return lock
}
export const writeLock = (lock, path) =>
write(path, JSON.stringify(lock) )
export const testWhitelist = (array, whitelist) =>
whitelist.find(tag => !array.includes(tag)) !== undefined
export const testBlacklist = (array, blacklist) =>
blacklist.find(tag => array.includes(tag)) !== undefined
export const createView = (directory, pageSize, extra = {}) => {
return {
batch: new PromiseBatch(),
directory,
pageSize,
...extra
}
}
//
// ;-. ,-. ,-.
// | `-. `-.
// ' `-' `-'
export async function fetchRss(source) {
let { hostname } = source
let error
let response
try {
response = await delayedFetch(
new URL(source.pathname, 'https://' + hostname),
{},
source.courtesyWait ?? 5 * 1000
)
} catch(err) {
error = err
}
source.errored = error !== undefined || !response.ok
if(source.errored) {
source.error = error ?? createNetworkingError(response)
source.rss = ''
} else {
source.rss = await response.text()
}
return source
}
let waitingList = new Map()
export const sleep = delay => new Promise(resolve => setTimeout(() => resolve(), delay) )
export const delayedFetch = async (url, options, courtesyWait = 5 * 1000) => {
let [ domain ] = /[\w-]+.[\w-]+$/.exec(url.hostname)
let timeout = waitingList.get(domain) ?? 0
let now = Date.now()
if(timeout == null || timeout <= now) {
waitingList.set(domain, timeout + courtesyWait)
} else {
await sleep(timeout - now)
}
return await fetch(url, options)
}
class NoMatchesError extends Error {}
export const processRss = (source, fromDate, reducerCallback) => {
let { document } = new JSDOM(source.rss, { contentType: 'text/xml' }).window
let items = document.querySelectorAll('channel item')
if(items.length == 0) {
throw new NoMatchesError('Got no matches')
}
for(let item of items) {
let post = processRssItem(source, item, reducerCallback)
if(post && post.date > fromDate) {
source.posts.push(post)
}
}
return source
}
export const processRssItem = (source, item, reducerCallback) => {
let description = new JSDOM(item.querySelector('description').textContent).window.document
let dateString = item.querySelector('pubDate').textContent
let link = item.querySelector('link').textContent
let guid = item.querySelector('guid')?.textContent
let title = item.querySelector('title')?.textContent
let post = {
source,
item,
description,
dateString,
date: new Date(dateString).valueOf() ?? 0,
link,
guid,
title
}
return reducerCallback(post)
}
export const processCategories = (post) => {
let categoryMatches = post.item.querySelectorAll('category')
post.categories = []
for(let category of categoryMatches) {
post.categories.push(category.textContent)
}
return post
}
export const extractImages = (post, cache = true) => {
let images = post.description.querySelectorAll('img')
if(images) {
let imageUrls = []
for(let image of images) {
let { src } = image
if(!src) {
let finalSrc = image.srcset.split(', ').pop()
src = finalSrc.slice(0, finalSrc.indexOf(' ') )
}
// Sending through URL prevents potential XSS
imageUrls.push(new URL(src).href)
}
return imageUrls
}
}
// o
// . , . ,-. , , ,
// |/ | |-' |/|/
// ' ' `-' ' '
export const writePages = (list, { pageSize, header = '', directory, batch }) => {
let posts = []
let lastPageLink = 'about:blank'
let pageIndex = 0
list.posts.sort((a, b) => b.date - a.date)
for(let i = list.posts.length - 1; i >= 0; i--) {
posts.push(list.posts[i])
if(i % pageSize == 0) {
let title = getPageTitle(list, pageIndex)
let html = renderPage(title, posts.reverse(), header, renderNextPageLink(lastPageLink))
let filename = i < pageSize ? getFinalPageFilename(list) : getPageFilename(list, pageIndex)
let promise = write(Path.join(directory, filename), html)
batch.add(promise.then(annotate(`Created "${title}" (${filename})`)))
posts = []
lastPageLink = filename
pageIndex++
}
}
// lock.lists[list.name] = {
// pageIndex,
// lastPostDate: posts[0]?.date ?? lastPostDate
// }
}
export const getFinalPageFilename = list =>
(list.main ? 'index' : list.name) + '.html'
export const getPageFilename = (list, i) =>
list.name + '-' + i + '.html'
export const getPageTitle = (list, i) =>
list.displayName + ' - ' + (i + 1)
export const renderPage = (title, posts, header, footer) => `\
<html>
<head>
<title>${title}</title>
<meta name="viewport" content="width=device-width, initial-scale=1">
<link rel="stylesheet" href="./style.css">
</head>
<body>
<header>
${header}
</header>
<main>
${posts.map(renderPost).join('\n')}
</main>
<footer>
${footer}
</footer>
</body>
</html>`
export const renderPost = post => {
let date = new Date(post.date)
let details = [
[ 'source', `<a href="${post.source.hostname}">${post.source.hostname}</a>` ]
]
if(post.title)
details.push([ 'title', `"${post.title}"` ])
if(post.categories && post.categories.length > 0)
details.push([ 'categories', post.categories.map(name => `<mark>${name}</mark>`).join(', ') ])
return `\
${post.images.map(renderImage).join('\n')}
<details>
<summary><b>${post.source.displayName}</b> ${renderDate(date)} <a href="${post.link}">open</a></summary>
<ul>
${details.map(args => renderPostDetail(...args)).join('\n')}
<ul>
</details>
<hr>`
}
export const renderPostDetail = (name, value) =>
`<li><b>${name}</b> ${value}</li>`
export const renderImage = href => {
return `\
<a href="${href}"><img src="${href}" loading="lazy"></img></a>`
}
export const renderDate = date =>
(date.getMonth() + 1) + '.' + date.getDate() + '.' + date.getFullYear()
export const renderNextPageLink = link => `\
<a href="${link}">next</a>`
export const renderNav = (feeds, sources) => `\
<details>
<summary>Feeds</summary>
<section>
<ul>
${feeds.map(renderNavEntry).join('\n')}
</ul>
<hr>
<ul>
${sources.map(renderNavEntry).join('\n')}
</ul>
</section>
</details>
<hr>`
export const renderNavEntry = (list) => {
let extra = ''
if(list.errored) {
extra += ' (errored)'
} else if (list.posts.length == 0) {
extra += ' (empty)'
}
return `<li><a href="${getFinalPageFilename(list)}">${list.displayName}</a>${extra}</li>`
}
// | | | o
// ,-. ,-. | | ,-: |- . ,-. ;-.
// | | | | | | | | | | | | |
// `-' `-' ' ' `-` `-' ' `-' ' '
export const downloadImages = (images, getImagePath, courtesyWait, { directory, batch }) => {
let out = []
for(let i = 0; i < images.length; i ++) {
let url = images[i]
let relativePath = getImagePath(url, i, images)
let fullPath = Path.join(directory, relativePath)
let promise = FS.access(fullPath)
.catch(() =>
download(url, fullPath, courtesyWait)
.then(annotate( `Downloaded ${relativePath}`))
)
out.push(relativePath)
batch.add(promise)
}
return out
}
export const pullImages = (post, renderer, discardPostIfNoImages = false, getPostId = postIdFromPathname) => {
let images = extractImages(post)
if(!discardPostIfNoImages || images.length > 0) {
post.images = downloadImages(
images,
buildImagePathHandler(post.source, getPostId(post)),
post.source.courtesyWait,
renderer
)
return post
}
}
export const createFeed = (name, sources, main = false) => {
return {
name,
displayName: name,
main,
posts: sources.reduce((posts, source) => posts.concat(source.posts), [])
}
}
export const fetchRssFromInstances = async (source, lock) => {
let index = 0
let instances = source.instances
let lockHostname = lock.sources[source.name]?.hostname
if(lockHostname) {
instances.unshift(lockHostname)
}
while(!source.rss && index != instances.length) {
source.hostname = instances[index]
source = await fetchRss(source)
if(source.errored) {
console.error(`Failed to fetch ${source.name} from ${source.hostname}: `, source.error)
index++
} else {
break
}
}
return source
}
// const addPostsToLock = (source, renderer) => {
// (renderer.lock.sources[source.name] ??= {})
// .postData = source.posts.map(post => post.description)
// }
export const populateSource = (source, postReducerCallback, lock) => {
let sourceLock = lock.sources[source.name] ??= {}
source.posts = []
source = processRss(source, sourceLock.timestamp ?? 0, postReducerCallback, lock)
if(sourceLock.items) {
for(let itemText of sourceLock.items) {
let item = new JSDOM(itemText, { contentType: 'text/xml' }).window.document.documentElement
source.posts.push(processRssItem(source, item, postReducerCallback))
}
}
lock.sources[source.name] = sourceLock
lockSource(source, lock)
return source
}
export const lockSource = (source, lock) => {
let date = 0
let items = []
for(let post of source.posts) {
if(post.date > date)
date = post.date
items.push(post.item.outerHTML)
}
lock.sources[source.name] = {
hostname: source.hostname,
timestamp: date,
items
}
}
export const lockSources = (sources, lock) => {
sources.forEach(source => lockSource(source, lock))
}
export const writeView = (sources, feeds, renderer) => {
renderer.header = renderNav(feeds, sources)
for(let feed of feeds) {
writePages(feed, renderer)
}
for(let source of sources) {
writePages(source, renderer)
}
writeStylesheet(Path.join(import.meta.dirname, 'assets/style.css'), renderer)
}
// | | ,-
// ;-. | ,-: |- | ,-. ;-. ;-.-. ,-.
// | | | | | | |- | | | | | | `-.
// |-' ' `-` `-' | `-' ' ' ' ' `-'
// ' -'
export const tumblr = {
async createSource(user, courtesyWait, postReducerCallback, lock) {
let lowercaseUser = user.toLowerCase()
let source = {
hostname: lowercaseUser + '.tumblr.com',
pathname: 'rss',
courtesyWait,
name: `tumblr-${lowercaseUser}`,
displayName: user,
user: lowercaseUser,
}
source = await fetchRss(source)
source = populateSource(source, postReducerCallback, lock)
return source
},
createSources(users, ...args) {
return Promise.all(users.map(user => tumblr.createSource(user, ...args)))
},
isRepost(post) {
let reblog = post.description.querySelector('p > a.tumblr_blog')
return reblog && reblog.innerHTML !== post.source.user
},
matchesTags(post, whitelist, blacklist) {
if(whitelist && testWhitelist(post.categories, whitelist)) {
return false
}
if(blacklist && testBlacklist(post.categories, blacklist)) {
return false
}
return true
},
pullImages
}
export const nitter = {
async createSource(user, instances, courtesyWait, postReducerCallback, lock) {
let source = {
instances,
pathname: user + '/rss',
courtesyWait,
name: `nitter-${user}`,
displayName: user,
user
}
source = await fetchRssFromInstances(source, lock)
source = populateSource(source, postReducerCallback, lock)
return source
},
createSources(users, ...args) {
return Promise.all(users.map(user => nitter.createSource(user, ...args)))
},
isRepost(post) {
let creator = post.item.getElementsByTagName('dc:creator')[0]
return creator.innerHTML.slice(1) === post.source.user
},
pullImages
}
// TODO: Mastodon support
//
// "Turns out Mastodon has built-in RSS; your feed URL is [instance]/@[username].rss, so for example I'm
// https://mastodon.social/@brownpau.rss (note the "@")"
// - https://mastodon.social/@brownpau/100523448408374430
export const platforms = {
tumblr,
nitter
}