page fr osm wiki

This commit is contained in:
Tykayn 2025-01-12 12:28:59 +01:00 committed by tykayn
parent 400e1b2b42
commit 6b8e111d77
4 changed files with 667 additions and 589 deletions

View File

@ -6,43 +6,44 @@ import https from 'https'
import moment from 'moment'
import Parser from 'rss-parser'
import { load } from 'cheerio'
import {load} from 'cheerio'
let local_node_env_conf = dotenv.config()
const myArgs = process.argv.slice(2)
export const reallySendPost = hasCliArgument('--force')
console.log('reallySendPost', reallySendPost)
export const folderBlogPostsPreview = process.cwd() + '/assets/blog_posts_medias/'
export function randomIntFromInterval (min, max) { // min and max included
return Math.floor(Math.random() * (max - min + 1) + min) | 1
export function randomIntFromInterval(min, max) { // min and max included
return Math.floor(Math.random() * (max - min + 1) + min) | 1
}
export function getRandomElementOfArray (listItems) {
return listItems[Math.floor(Math.random() * listItems.length)]
export function getRandomElementOfArray(listItems) {
return listItems[Math.floor(Math.random() * listItems.length)]
}
let nowDate = new Date()
export let defaultConfigMasto = {
author: 'curator',
visibility: 'public',
language: 'fr',
sensitive: false,
disable_slugify: false,
reallySendPost,
image: '',
folder_image: process.cwd() + '/assets/blog_posts_medias/',
message: 'Hey coucou! on est le' + nowDate,
scheduled_at: '',
scheduled_at_bool: false,
content_type: 'text/markdown',
website: 'qzine',
slug: 'default_post_title',
postObject: {},
author: 'curator',
visibility: 'public',
language: 'fr',
sensitive: false,
disable_slugify: false,
reallySendPost,
image: '',
folder_image: process.cwd() + '/assets/blog_posts_medias/',
message: 'Hey coucou! on est le' + nowDate,
scheduled_at: '',
scheduled_at_bool: false,
content_type: 'text/markdown',
website: 'qzine',
slug: 'default_post_title',
postObject: {},
}
export function tokenForAuthorIsPresentInDotEnv (author) {
return process.env['TOKEN_' + author.toUpperCase()]
export function tokenForAuthorIsPresentInDotEnv(author) {
return process.env['TOKEN_' + author.toUpperCase()]
}
/**
@ -50,13 +51,13 @@ export function tokenForAuthorIsPresentInDotEnv (author) {
* @param userNickName
* @returns {Mastodon}
*/
export function createMastoFetcherWithAuthorLogin (userNickName) {
let accessToken = process.env['TOKEN_' + userNickName.toUpperCase()]
const masto = new Masto({
access_token: accessToken,
api_url: process.env.INSTANCE_MASTODON + '/api/v1/',
})
return masto
export function createMastoFetcherWithAuthorLogin(userNickName) {
let accessToken = process.env['TOKEN_' + userNickName.toUpperCase()]
const masto = new Masto({
access_token: accessToken,
api_url: process.env.INSTANCE_MASTODON + '/api/v1/',
})
return masto
}
/**
@ -64,125 +65,125 @@ export function createMastoFetcherWithAuthorLogin (userNickName) {
* @param config
* @returns {*}
*/
export default function sendPostMastodon (config) {
export default function sendPostMastodon(config) {
// console.log('send post', config.postObject.post_guid , config.postObject.guid )
// override defaults with input argument
config = {
...defaultConfigMasto,
...config,
}
// console.log('send post', config.postObject.post_guid , config.postObject.guid )
// override defaults with input argument
config = {
...defaultConfigMasto,
...config,
}
// console.log("sendPostMastodon config", config)
// console.log("sendPostMastodon config", config)
if (!config.reallySendPost) {
if (!config.reallySendPost) {
console.log('\n\n =========== le message ne sera PAS réellement posté sur le compte @' + config.author + '@' + process.env.INSTANCE_MASTODON + ' =========== \n')
// console.log('configPost.folder_image', config.folder_image)
console.log('config', config.message)
} else {
console.log('\n\n =========== le message ne sera PAS réellement posté sur le compte @' + config.author + '@' + process.env.INSTANCE_MASTODON + ' =========== \n')
// console.log('configPost.folder_image', config.folder_image)
console.log('config', config.message)
} else {
console.log(' ')
if (process.env.INSTANCE_MASTODON && tokenForAuthorIsPresentInDotEnv(config.author)) {
console.log(' ')
if (process.env.INSTANCE_MASTODON && tokenForAuthorIsPresentInDotEnv(config.author)) {
let visibility = 'public'
let language = 'fr'
let sensitive = false
let visibility = 'public'
let language = 'fr'
let sensitive = false
let accessToken = process.env['TOKEN_' + config.author.toUpperCase()]
const masto = new Masto({
access_token: accessToken,
api_url: process.env.INSTANCE_MASTODON + '/api/v1/',
})
let accessToken = process.env['TOKEN_' + config.author.toUpperCase()]
const masto = new Masto({
access_token: accessToken,
api_url: process.env.INSTANCE_MASTODON + '/api/v1/',
})
let params = {
status: config.message,
visibility,
language,
sensitive
}
if (config.cw) {
params['spoiler_text'] = config.cw
}
if (config.scheduled_at && config.scheduled_at_bool) {
let dateschedule = new Date(config.scheduled_at)
params['scheduled_at'] = dateschedule.toISOString()
}
let params = {
status: config.message,
visibility,
language,
sensitive
}
if (config.cw) {
params['spoiler_text'] = config.cw
}
if (config.scheduled_at && config.scheduled_at_bool) {
let dateschedule = new Date(config.scheduled_at)
params['scheduled_at'] = dateschedule.toISOString()
}
/**
* envoi sans fichier joint
*/
if (!config.image) {
console.log('pas d image dans la config')
/**
* envoi sans fichier joint
*/
if (!config.image) {
console.log('pas d image dans la config')
if (config.reallySendPost) {
if (config.reallySendPost) {
masto.post('statuses', params).then(rep => {
masto.post('statuses', params).then(rep => {
console.log('posté, yay!')
}, err => {
console.error(err)
})
}
}
console.log('posté, yay!')
}, err => {
console.error(err)
})
}
}
/**
* envoi avec fichier,
* on doit d'abord faire un upload du fichier,
* puis relier son id de média au nouveau post.
*/
else if (config.image) {
/**
* envoi avec fichier,
* on doit d'abord faire un upload du fichier,
* puis relier son id de média au nouveau post.
*/
else if (config.image) {
var id
console.log('envoi du média', config.image)
// upload new media
return masto.post('media', { file: fs.createReadStream(config.image) })
.then(resp => {
id = resp.data.id
params.media_ids = [id]
console.log('\n ✅ image, id', id)
masto.post('statuses', params).then(rep => {
// console.log('rep', rep)
console.log('\n ✅ posté avec une nouvelle image, WOOT')
}, err => {
console.error(err)
var id
console.log('envoi du média', config.image)
// upload new media
return masto.post('media', {file: fs.createReadStream(config.image)})
.then(resp => {
id = resp.data.id
params.media_ids = [id]
console.log('\n ✅ image, id', id)
masto.post('statuses', params).then(rep => {
// console.log('rep', rep)
console.log('\n ✅ posté avec une nouvelle image, WOOT')
}, err => {
console.error(err)
console.log('erreur T_T')
})
})
}
console.log('erreur T_T')
})
})
}
// }
// }
} else {
console.error(`pas de token pour l'auteur "${config.author}" ou pas d'instance mastodon définie`)
}
} else {
console.error(`pas de token pour l'auteur "${config.author}" ou pas d'instance mastodon définie`)
}
}
}
}
// Slugify a string
export function slugify (str) {
str = str.replace(/^\s+|\s+$/g, '')
export function slugify(str) {
str = str.replace(/^\s+|\s+$/g, '')
// Make the string lowercase
str = str.toLowerCase()
// Make the string lowercase
str = str.toLowerCase()
// Remove accents, swap ñ for n, etc
var from = 'ÁÄÂÀÃÅČÇĆĎÉĚËÈÊẼĔȆÍÌÎÏŇÑÓÖÒÔÕØŘŔŠŤÚŮÜÙÛÝŸŽáäâàãåčçćďéěëèêẽĕȇíìîïňñóöòôõøðřŕšťúůüùûýÿžþÞĐđßÆa·/_,:;'
var to = 'AAAAAACCCDEEEEEEEEIIIINNOOOOOORRSTUUUUUYYZaaaaaacccdeeeeeeeeiiiinnooooooorrstuuuuuyyzbBDdBAa------'
for (var i = 0, l = from.length; i < l; i++) {
str = str.replace(new RegExp(from.charAt(i), 'g'), to.charAt(i))
}
// Remove accents, swap ñ for n, etc
var from = 'ÁÄÂÀÃÅČÇĆĎÉĚËÈÊẼĔȆÍÌÎÏŇÑÓÖÒÔÕØŘŔŠŤÚŮÜÙÛÝŸŽáäâàãåčçćďéěëèêẽĕȇíìîïňñóöòôõøðřŕšťúůüùûýÿžþÞĐđßÆa·/_,:;'
var to = 'AAAAAACCCDEEEEEEEEIIIINNOOOOOORRSTUUUUUYYZaaaaaacccdeeeeeeeeiiiinnooooooorrstuuuuuyyzbBDdBAa------'
for (var i = 0, l = from.length; i < l; i++) {
str = str.replace(new RegExp(from.charAt(i), 'g'), to.charAt(i))
}
// Remove invalid chars
str = str.replace(/[^a-z0-9 -]/g, '')
// Collapse whitespace and replace by -
.replace(/\s+/g, '-')
// Collapse dashes
.replace(/-+/g, '-')
// Remove invalid chars
str = str.replace(/[^a-z0-9 -]/g, '')
// Collapse whitespace and replace by -
.replace(/\s+/g, '-')
// Collapse dashes
.replace(/-+/g, '-')
return str
return str
}
/**
@ -190,13 +191,13 @@ export function slugify (str) {
* lister les noms de fichier que l'on peut publier dans un dossier.
* retourne un tableau
*/
export function listFilesOfFolder (folderPath) {
let filesNames = []
fs.readdirSync(folderPath).map(fileName => {
return filesNames.push(fileName)
})
export function listFilesOfFolder(folderPath) {
let filesNames = []
fs.readdirSync(folderPath).map(fileName => {
return filesNames.push(fileName)
})
return filesNames
return filesNames
}
/**
@ -204,14 +205,14 @@ export function listFilesOfFolder (folderPath) {
* crée un dossier d'assets, avec ses sous dossiers not_published et published si ils manquent.
* une fois que l'on prendra une image dans le dossier non publié, on la déplacera dans le dossier des images publées.
*/
export function initializeFolderForPictures (folderName) {
try {
if (!fs.existsSync(folderName)) {
fs.mkdirSync(folderName)
}
} catch (err) {
console.error(err)
}
export function initializeFolderForPictures(folderName) {
try {
if (!fs.existsSync(folderName)) {
fs.mkdirSync(folderName)
}
} catch (err) {
console.error(err)
}
}
/**
@ -219,36 +220,36 @@ export function initializeFolderForPictures (folderName) {
* @param htmlContent
* @returns {string}
*/
export function findFirstImageInContent (htmlContent = '') {
let result = ''
let foundPictures = htmlContent.match(/<img\s[^>]*?src\s*=\s*['\"]([^'\"]*?)['\"][^>]*?>/)
let first = ''
if (foundPictures && foundPictures[0]) {
export function findFirstImageInContent(htmlContent = '') {
let result = ''
let foundPictures = htmlContent.match(/<img\s[^>]*?src\s*=\s*['\"]([^'\"]*?)['\"][^>]*?>/)
let first = ''
if (foundPictures && foundPictures[0]) {
first = foundPictures[0]
first = foundPictures[0]
} else {
console.log('pas d image trouvée dans le contenu ', htmlContent)
}
if (first) {
result = first.match(/src\=\"(.*)\"/i)
if (result.length && result[0]) {
result = result[0].split('"')
result = result[1]
}
}
result = clearLink(result)
console.log('clearLink', result)
return result
} else {
console.log('pas d image trouvée dans le contenu ', htmlContent)
}
if (first) {
result = first.match(/src\=\"(.*)\"/i)
if (result.length && result[0]) {
result = result[0].split('"')
result = result[1]
}
}
result = clearLink(result)
console.log('clearLink', result)
return result
}
function clearLink (linkString) {
linkString = linkString.replace('http:', 'https:')
linkString = linkString.replace('https://www.ailesse.info/~tykayn/bazar/kotlife', 'https://www.tykayn.fr/wp-content/uploads/i/kotlife')
linkString = linkString.replace('https://blog.artlemoine.com/public/i', 'https://www.tykayn.fr/wp-content/uploads/i')
linkString = linkString.replace('https://www.ailesse.com/%7Etykayn/bazar', 'https://www.tykayn.fr/wp-content/uploads/i/bazar')
function clearLink(linkString) {
linkString = linkString.replace('http:', 'https:')
linkString = linkString.replace('https://www.ailesse.info/~tykayn/bazar/kotlife', 'https://www.tykayn.fr/wp-content/uploads/i/kotlife')
linkString = linkString.replace('https://blog.artlemoine.com/public/i', 'https://www.tykayn.fr/wp-content/uploads/i')
linkString = linkString.replace('https://www.ailesse.com/%7Etykayn/bazar', 'https://www.tykayn.fr/wp-content/uploads/i/bazar')
return linkString
return linkString
}
/**
@ -260,25 +261,25 @@ function clearLink (linkString) {
* @param filepath
* @returns {Promise<unknown>}
*/
export function downloadImage (url, filepath) {
return new Promise((resolve, reject) => {
const options = {
headers: { 'User-Agent': 'Mozilla/5.0 (Macintosh; Intel Mac OS X 12_6) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/107.0.0.0 Safari/537.36 Edg/107.0.1418.52' }
}
export function downloadImage(url, filepath) {
return new Promise((resolve, reject) => {
const options = {
headers: {'User-Agent': 'Mozilla/5.0 (Macintosh; Intel Mac OS X 12_6) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/107.0.0.0 Safari/537.36 Edg/107.0.1418.52'}
}
https.get(url, options, (res) => {
if (res.statusCode === 200) {
res.pipe(fs.createWriteStream(filepath))
.on('error', reject)
.once('close', () => resolve(filepath))
} else {
// Consume response data to free up memory
res.resume()
reject(new Error(`Request Failed With a Status Code: ${res.statusCode}; \n ${res.statusMessage} `))
https.get(url, options, (res) => {
if (res.statusCode === 200) {
res.pipe(fs.createWriteStream(filepath))
.on('error', reject)
.once('close', () => resolve(filepath))
} else {
// Consume response data to free up memory
res.resume()
reject(new Error(`Request Failed With a Status Code: ${res.statusCode}; \n ${res.statusMessage} `))
}
})
})
}
})
})
}
/**
@ -289,29 +290,29 @@ export function downloadImage (url, filepath) {
* @returns {Promise<Object>}
* @constructor
*/
export function CropPicture (pictureName, width = 500, height = 300) {
export function CropPicture(pictureName, width = 500, height = 300) {
return sharp(pictureName)
return sharp(pictureName)
.extract({ left: 0, top: 0, width, height })
.toFile('thumb_' + pictureName, function (err) {
if (err) console.log(err)
})
.extract({left: 0, top: 0, width, height})
.toFile('thumb_' + pictureName, function (err) {
if (err) console.log(err)
})
}
/**
* prendre un post parmi tous ceux du blog, dans ceux qui ont été publiés
* @returns {*}
*/
export function getRandomLinkGeneral (tkpostsjson) {
let filteredLinks = []
if (tkpostsjson[0].post_status) {
export function getRandomLinkGeneral(tkpostsjson) {
let filteredLinks = []
if (tkpostsjson[0].post_status) {
filteredLinks = tkpostsjson.filter(elem => elem.post_status === 'publish')
} else if (tkpostsjson[0].status) {
filteredLinks = tkpostsjson.filter(elem => elem.status === 'publish')
}
return getRandomElementOfArray(filteredLinks)
filteredLinks = tkpostsjson.filter(elem => elem.post_status === 'publish')
} else if (tkpostsjson[0].status) {
filteredLinks = tkpostsjson.filter(elem => elem.status === 'publish')
}
return getRandomElementOfArray(filteredLinks)
}
/**
@ -319,44 +320,44 @@ export function getRandomLinkGeneral (tkpostsjson) {
* @param postContent
* @param configPost
*/
export function findPictureAndSendPost (postContent, configPost) {
export function findPictureAndSendPost(postContent, configPost) {
let firstPictureSource = findFirstImageInContent(postContent)
let firstPictureSource = findFirstImageInContent(postContent)
let filePathForDownloadedImage = `${configPost.folder_image}_${configPost.website}_media_post_${slugify(configPost.slug)}.jpg`
let filePathForDownloadedImage = `${configPost.folder_image}_${configPost.website}_media_post_${slugify(configPost.slug)}.jpg`
if (firstPictureSource) {
console.log('firstPictureSource found', firstPictureSource)
if (firstPictureSource) {
console.log('firstPictureSource found', firstPictureSource)
// check if picture already exist
console.log('on envoie le média et l image : ', filePathForDownloadedImage)
downloadImage(firstPictureSource, filePathForDownloadedImage)
.then((res) => {
// suite du poste avec upload d'image
// check if picture already exist
console.log('on envoie le média et l image : ', filePathForDownloadedImage)
downloadImage(firstPictureSource, filePathForDownloadedImage)
.then((res) => {
// suite du poste avec upload d'image
console.log('média téléchargé, on envoie le post')
configPost.image = filePathForDownloadedImage
console.log('média téléchargé, on envoie le post')
configPost.image = filePathForDownloadedImage
sendPostMastodon(configPost)
sendPostMastodon(configPost)
},
(err) => {
console.log('pas dimage trouvée pour l URL ', firstPictureSource, err)
sendPostMastodon(configPost)
}
)
.catch((err) => {
console.log('erreur avec cette URL ', firstPictureSource, err)
sendPostMastodon(configPost)
})
} else {
},
(err) => {
console.log('pas dimage trouvée pour l URL ', firstPictureSource, err)
sendPostMastodon(configPost)
}
)
.catch((err) => {
console.log('erreur avec cette URL ', firstPictureSource, err)
sendPostMastodon(configPost)
})
} else {
// no image provided
// no image provided
console.log('pas d\'image dans le corps du texte', configPost.image)
// on envoie avec l'image par défaut
sendPostMastodon(configPost)
}
console.log('pas d\'image dans le corps du texte', configPost.image)
// on envoie avec l'image par défaut
sendPostMastodon(configPost)
}
}
/**
@ -364,163 +365,163 @@ export function findPictureAndSendPost (postContent, configPost) {
* @param argument
* @returns {boolean}
*/
export function hasCliArgument (argument) {
return myArgs.indexOf(argument) !== -1
export function hasCliArgument(argument) {
return myArgs.indexOf(argument) !== -1
}
let parser = new Parser()
export function diffDaysBetweenTwoDates (date1, date2) {
export function diffDaysBetweenTwoDates(date1, date2) {
const a = moment(date1)
const b = moment(date2)
return a.diff(b, 'days')
const a = moment(date1)
const b = moment(date2)
return a.diff(b, 'days')
}
export function filterRegionAgendaDuLibreEvents (events_list, filter_critera) {
let selection = []
events_list.forEach(item => {
if (item.region_id == filter_critera) {
selection.push(item)
}
})
return selection
export function filterRegionAgendaDuLibreEvents(events_list, filter_critera) {
let selection = []
events_list.forEach(item => {
if (item.region_id == filter_critera) {
selection.push(item)
}
})
return selection
}
moment.locale('fr')
export function groupEventsByDay (events_list) {
let selection = {}
events_list.forEach(item => {
export function groupEventsByDay(events_list) {
let selection = {}
events_list.forEach(item => {
let formattedDay = moment(item.start_time).format('dddd DD')
let formattedDay = moment(item.start_time).format('dddd DD')
if (!selection[formattedDay]) {
selection[formattedDay] = []
}
selection[formattedDay].push(item)
})
return selection
if (!selection[formattedDay]) {
selection[formattedDay] = []
}
selection[formattedDay].push(item)
})
return selection
}
export function convertHTMLtoMD (htmlContent) {
const $ = load(htmlContent)
export function convertHTMLtoMD(htmlContent) {
const $ = load(htmlContent)
translateNode($)
translateNode($)
return $.html()
return $.html()
}
export function translateNode ($) {
const elementsToTranslate = [
{ tag: 'h1', mdTag: '##' },
{ tag: 'h2', mdTag: '###' },
{ tag: 'h3', mdTag: '####' },
{ tag: 'h4', mdTag: '#####' },
{ tag: 'h5', mdTag: '######' },
{ tag: 'p', mdTag: '\n\n' },
{ tag: 'ul', preserveChildren: true, transformChild: ($child) => '\n- ' + $.text($child) },
{ tag: 'ol', preserveChildren: true, transformChild: ($child, idx) => `\n${idx + 1}. ${$.text($child)}` },
{ tag: 'a', parseAttr: ('href', link => `[${link}](${link})`) },
{ tag: 'strong', mdFormat: '$&' },
{ tag: 'em', mdFormat: '_$&_' },
{ tag: 'code', mdFormat: '`$&`' },
]
export function translateNode($) {
const elementsToTranslate = [
{tag: 'h1', mdTag: '##'},
{tag: 'h2', mdTag: '###'},
{tag: 'h3', mdTag: '####'},
{tag: 'h4', mdTag: '#####'},
{tag: 'h5', mdTag: '######'},
{tag: 'p', mdTag: '\n\n'},
{tag: 'ul', preserveChildren: true, transformChild: ($child) => '\n- ' + $.text($child)},
{tag: 'ol', preserveChildren: true, transformChild: ($child, idx) => `\n${idx + 1}. ${$.text($child)}`},
{tag: 'a', parseAttr: ('href', link => `[${link}](${link})`)},
{tag: 'strong', mdFormat: '$&'},
{tag: 'em', mdFormat: '_$&_'},
{tag: 'code', mdFormat: '`$&`'},
]
elementsToTranslate.forEach(element => {
$(element.tag).each((_, el) => {
const $el = $(el)
if (element.parseAttr) {
const attrVal = $el.attr(element.parseAttr[0])
if (attrVal) {
$el.replaceWith(element.parseAttr[1](attrVal))
} else {
$el.remove()
}
} else if (element.transformChild) {
$el.contents().filter(() => !!this.parent()).each((_, childEl) => {
const transformedText = element.transformChild($(childEl), $el.index())
$el.before('\n' + transformedText)
})
$el.remove()
} else {
$el.replaceWith(element.mdTag + $el.text() + element.mdFormat)
}
})
})
elementsToTranslate.forEach(element => {
$(element.tag).each((_, el) => {
const $el = $(el)
if (element.parseAttr) {
const attrVal = $el.attr(element.parseAttr[0])
if (attrVal) {
$el.replaceWith(element.parseAttr[1](attrVal))
} else {
$el.remove()
}
} else if (element.transformChild) {
$el.contents().filter(() => !!this.parent()).each((_, childEl) => {
const transformedText = element.transformChild($(childEl), $el.index())
$el.before('\n' + transformedText)
})
$el.remove()
} else {
$el.replaceWith(element.mdTag + $el.text() + element.mdFormat)
}
})
})
}
export function getRequestOptions (host, port, path) {
const options = {
host: host,
port: port,
path: path,
method: 'GET',
headers: {
Accept: 'application/activity+json',
'User-Agent': 'MyApp/1.0.0 (contact@myemail.org)',
},
}
export function getRequestOptions(host, port, path) {
const options = {
host: host,
port: port,
path: path,
method: 'GET',
headers: {
Accept: 'application/activity+json',
'User-Agent': 'MyApp/1.0.0 (contact@myemail.org)',
},
}
return options
return options
}
export async function sendGetRequest (options) {
return new Promise((resolve, reject) => {
const req = http.request(options, (res) => {
let body = Buffer.alloc(0)
export async function sendGetRequest(options) {
return new Promise((resolve, reject) => {
const req = http.request(options, (res) => {
let body = Buffer.alloc(0)
res.on('data', (chunk) => {
body = Buffer.concat([body, chunk])
})
res.on('data', (chunk) => {
body = Buffer.concat([body, chunk])
})
res.on('end', () => {
resolve(JSON.parse(body.toString()))
})
})
res.on('end', () => {
resolve(JSON.parse(body.toString()))
})
})
req.on('error', (err) => {
reject(err)
})
req.on('error', (err) => {
reject(err)
})
req.end()
})
req.end()
})
}
const splitTextIntoChunks = (text, limit) => {
const words = text.trim().split(/\s+/g)
const chunks = []
let currentChunk = []
let wordCount = 0
const words = text.trim().split(/\s+/g)
const chunks = []
let currentChunk = []
let wordCount = 0
for (const word of words) {
if (wordCount + word.length > limit) {
chunks.push(currentChunk.join(' '))
currentChunk = [word]
wordCount = word.length
} else {
currentChunk.push(word)
wordCount += word.length + 1
}
}
for (const word of words) {
if (wordCount + word.length > limit) {
chunks.push(currentChunk.join(' '))
currentChunk = [word]
wordCount = word.length
} else {
currentChunk.push(word)
wordCount += word.length + 1
}
}
if (currentChunk.length > 0) {
const joinedWords = currentChunk.join(' ')
chunks.push(joinedWords.length <= limit ? joinedWords : joinedWords.substr(0, limit) + '...')
}
if (currentChunk.length > 0) {
const joinedWords = currentChunk.join(' ')
chunks.push(joinedWords.length <= limit ? joinedWords : joinedWords.substr(0, limit) + '...')
}
return chunks
return chunks
}
export function splitLongDescription (text, limit) {
let chunks = splitTextIntoChunks(text, limit)
if (chunks) {
if (chunks[0]) {
return chunks[0]
}
return ''
}
export function splitLongDescription(text, limit) {
let chunks = splitTextIntoChunks(text, limit)
if (chunks) {
if (chunks[0]) {
return chunks[0]
}
return ''
}
}

View File

@ -1,12 +1,8 @@
import fetch from "node-fetch"
import rp from "request-promise";
import $ from "cheerio";
import fs from "fs";
import path from 'path';
let url = 'https://wiki.openstreetmap.org/wiki/FR:Key:building:material';
let pictureDescrFinder = '.d_image img'
let pictureDescrFinder = '.d_image img'
// rp(url).then(function (html) {
// getSourceOfDescriptorPageContent(html)
@ -30,7 +26,7 @@ export function getSourceOfDescriptorPageContent(HTMLcontent) {
}
console.log(sourcesSet)
console.log(selectedPicture)
return selectedPicture;
return `https://wiki.openstreetmap.org${selectedPicture.trim()}`;
} else {
console.log("pas d'image de description dans le HTML")
}

View File

@ -1,7 +1,15 @@
/**
* Envoi d'une page du wiki au hasard parmi les pages en Français répertoriées dans le json all_wiki_osm.json
* Cette liste d'environ 3000 pages du wiki contient toutes les pages francophones, pas seulement les tags.
* Utiliser l'argument --force pour réellement envoyer le post avec le compte Curator.
* nécessite d'avoir le fichier .env rempli.
*/
import fs from 'fs';
import path, {dirname} from 'path';
import axios from 'axios';
import {fileURLToPath} from 'url';
import {getSourceOfDescriptorPageContent} from "./osm_get_description_picture.mjs";
import sendPostMastodon, {downloadImage, randomIntFromInterval} from "./libs/utils.mjs";
const __dirname = dirname(fileURLToPath(import.meta.url));
@ -13,10 +21,22 @@ const randomIndex = Math.floor(Math.random() * wikiArticles['elements'].length);
console.log('wikiArticles[\'elements\']', wikiArticles['elements'].length)
const selectedArticle = wikiArticles['elements'][randomIndex];
console.log('selectedArticle', selectedArticle)
// Récupérer le titre et la description de l'article
const title = selectedArticle.title;
const pageId = selectedArticle.pageid;
function makePostMessageFromObj(post_obj) {
return `
# Le tag OSM du jour : ${post_obj.key}=${post_obj.value} 🗺🏷
${post_obj.link}
${post_obj.description ? post_obj.description.trim() : ''}
${post_obj.long_desc ? post_obj.long_desc.trim() : ''}
#osm #openstreetmap #wiki #rtfw
`
}
// Récupérer le contenu de l'article via l'API de MediaWiki
const wikiApiUrl = `https://wiki.openstreetmap.org/w/api.php`;
const params = {
@ -26,10 +46,23 @@ const params = {
prop: 'text',
section: 0
};
// console.log('params', params)
let message = ''
let download_description_src = ''
axios.get(wikiApiUrl, {params})
.then(response => {
// console.log('wikiApiUrl', wikiApiUrl)
let keys = Object.keys(response.data)
// console.log('keys', keys)
const articleContent = response.data.parse.text['*'];
let resultpicture = getSourceOfDescriptorPageContent(articleContent)
// console.log('resultpicture', resultpicture)
if (resultpicture === "pas d'image de description dans le HTML") {
download_description_src = `${process.cwd()}/assets/blog_posts_medias/osm_wiki_description_page.jpg`
}
const firstParagraph = articleContent.split('<p>')[1].split('</p>')[0];
// Sanitizer le texte
@ -38,13 +71,67 @@ axios.get(wikiApiUrl, {params})
return String.fromCharCode(code);
});
console.log('image', download_description_src)
console.log('-----------------')
// Générer le message
const message = `Je vous recommande de lire l'article "${title}" sur le wiki d'OpenStreetMap :
message = `Le page #OSM du jour : ${title}.
https://wiki.openstreetmap.org/wiki/${title}
${sanitizedTextWithoutEntities}
Lire la suite sur : https://wiki.openstreetmap.org/wiki/${title}`;
#openstreetmap #wiki #RTFW
`;
console.log(message);
})
.catch(error => {
console.error(error);
});
});
export function sendMessageWikiOSMOfTheDay(message, download_description_src) {
let configPost = {
author: 'curator',
// document.querySelector('.d_image img').attributes['src']
// image: '/home/tykayn/www/multi-account-post-schedule-mastodon/assets/osm_post_' + randomIntFromInterval(1 , 5) + '.jpg',
image: `${process.cwd()}/assets/blog_posts_medias/assets/osm_post_${randomIntFromInterval(1, 5)}.jpg`,
message,
}
if (download_description_src) {
let filePathImage = `${process.cwd()}/assets/blog_posts_medias/osm_wiki_description_page.jpg`
if (download_description_src) {
console.log("firstPictureSource found", download_description_src)
// check if picture already exist
console.log('on récupère l image de description : ', filePathImage)
downloadImage(download_description_src, filePathImage)
.then((res) => {
// suite du poste avec upload d'image
console.log('média téléchargé, on envoie le post')
configPost.image = filePathImage;
sendPostMastodon(configPost)
},
(err) => {
console.log('pas dimage trouvée pour l URL ', download_description_src, err)
sendPostMastodon(configPost)
}
)
.catch((err) => {
console.log('erreur avec cette URL ', download_description_src, err)
sendPostMastodon(configPost)
})
}
} else {
console.log('no image description')
sendPostMastodon(configPost)
}
}
sendMessageWikiOSMOfTheDay(message, download_description_src)

View File

@ -1,6 +1,8 @@
// https://www.mediawiki.org/wiki/Manual:Random_page
/**
* Post de page aléatoire du wiki osm avec le compte curator
* utiliser l'argument --force pour réellement envoyer le post avec le compte Curator.
* nécessite d'avoir le fichier .env rempli.
*/
import fetch from "node-fetch"
import rp from "request-promise";
@ -24,188 +26,188 @@ let foundExistingWikiPageInFrench = false;
function getRandomWikiOSMPage() {
// get all FR tags
// const url = "https://wiki.openstreetmap.org/w/index.php?title=Special:Categories&offset=FR&limit=500"
const url = "https://wiki.openstreetmap.org/wiki/Special:AllPages?from=&to=&namespace=202&limit=500"
// get a random page wich is not a discussion
// const url = "https://wiki.openstreetmap.org/wiki/Special:Random?namespace=all-discussions%3B1%3B2%3B3%3B5%3B7%3B9%3B11%3B120%3B121%3B122%3B123%3B201%3B203&invert=1&limit=10&days=1&enhanced=1&namespace__202_color=c1&urlversion=2"
rp(url)
.then(function (html) {
//success!
const wikiUrls = [];
const filteredHtml = $('.mw-spcontent ul li a', html)
console.log("filteredHtml.length", filteredHtml.length)
// .filter(elem => {
// return (elem.attribs.title.indexOf('description for') === -1)
// })
for (let i = 0; i < filteredHtml.length; i++) {
// const contentCurated = filteredHtml[i]
const contentCurated = filteredHtml[i].attribs.href
wikiUrls.push(contentCurated);
}
console.log(wikiUrls);
return wikiUrls;
})
.catch(function (err) {
//handle error
console.error(err)
});
// get all FR tags
// const url = "https://wiki.openstreetmap.org/w/index.php?title=Special:Categories&offset=FR&limit=500"
const url = "https://wiki.openstreetmap.org/wiki/Special:AllPages?from=&to=&namespace=202&limit=500"
// get a random page wich is not a discussion
// const url = "https://wiki.openstreetmap.org/wiki/Special:Random?namespace=all-discussions%3B1%3B2%3B3%3B5%3B7%3B9%3B11%3B120%3B121%3B122%3B123%3B201%3B203&invert=1&limit=10&days=1&enhanced=1&namespace__202_color=c1&urlversion=2"
rp(url)
.then(function (html) {
//success!
const wikiUrls = [];
const filteredHtml = $('.mw-spcontent ul li a', html)
console.log("filteredHtml.length", filteredHtml.length)
// .filter(elem => {
// return (elem.attribs.title.indexOf('description for') === -1)
// })
for (let i = 0; i < filteredHtml.length; i++) {
// const contentCurated = filteredHtml[i]
const contentCurated = filteredHtml[i].attribs.href
wikiUrls.push(contentCurated);
}
console.log(wikiUrls);
return wikiUrls;
})
.catch(function (err) {
//handle error
console.error(err)
});
}
export default function getElementCartographique() {
if (foundExistingWikiPageInFrench) {
console.log("already found foundExistingWikiPageInFrench")
return;
}
const url = "https://wiki.openstreetmap.org/wiki/FR:%C3%89l%C3%A9ments_cartographiques"
const titleLink = "#toc a"
const listElementSelector = "#toc a"
// prendre un lien dans la table des matières de la page, extraire l'id du lien.
// prendre une ligne de tableau au hasard correspondant à l'ID de la section
const titleSelector = ""
return fs.readFile(__dirname + '/assets/documents/elements_cartographiques.html', 'utf8', function (err, html) {
if (err) {
return console.log(err);
if (foundExistingWikiPageInFrench) {
console.log("already found foundExistingWikiPageInFrench")
return;
}
// console.log(data);
const url = "https://wiki.openstreetmap.org/wiki/FR:%C3%89l%C3%A9ments_cartographiques"
const titleLink = "#toc a"
const listElementSelector = "#toc a"
// prendre un lien dans la table des matières de la page, extraire l'id du lien.
// prendre une ligne de tableau au hasard correspondant à l'ID de la section
const titleSelector = ""
//success!
const wikiUrls = [];
// const listOfLinks = $(titleLink, html)
// console.log("filteredHtml.length", listOfLinks.length)
// const selectedTocLink = getRandomElementOfArray(listOfLinks).attribs.href;
// console.log("selectedTocLink", selectedTocLink)
return fs.readFile(__dirname + '/assets/documents/elements_cartographiques.html', 'utf8', function (err, html) {
if (err) {
return console.log(err);
}
// console.log(data);
const listOfTableRows = $('.wikitable', html);
let keys = Object.keys(listOfTableRows);
let randNumber = 0;
if (selectionOverrideOfSectionTable) {
randNumber = selectionOverrideOfSectionTable
} else {
randNumber = randomIntFromInterval(0, keys.length)
}
//success!
const wikiUrls = [];
// const listOfLinks = $(titleLink, html)
// console.log("filteredHtml.length", listOfLinks.length)
// const selectedTocLink = getRandomElementOfArray(listOfLinks).attribs.href;
// console.log("selectedTocLink", selectedTocLink)
console.log("rand", randNumber)
const foundLine = $(listOfTableRows[randNumber]).find('tr');
const listOfTableRows = $('.wikitable', html);
let keys = Object.keys(listOfTableRows);
// console.log("randomTable",foundTable)
let randNumber = 0;
if (selectionOverrideOfSectionTable) {
randNumber = selectionOverrideOfSectionTable
} else {
randNumber = randomIntFromInterval(0, keys.length)
}
// on regarde les lignes du tableau
keys = Object.keys(foundLine);
if (selectionOverrideOfLineTable) {
randNumber = selectionOverrideOfLineTable
} else {
randNumber = randomIntFromInterval(0, keys.length)
}
console.log("rand", randNumber)
const foundLine = $(listOfTableRows[randNumber]).find('tr');
console.log("lignes de tableau", keys.length)
console.log("ligne sélectionnée", randNumber)
let cells = $(foundLine[randNumber]).find('td')
// console.log("randomTable",foundTable)
keys = Object.keys(cells);
// console.log("cellules trouvées", keys.length)
// on regarde les lignes du tableau
keys = Object.keys(foundLine);
if (selectionOverrideOfLineTable) {
randNumber = selectionOverrideOfLineTable
} else {
randNumber = randomIntFromInterval(0, keys.length)
}
console.log("\n")
let configPost = {
key: '',
value: '',
description: '',
long_desc: '',
}
cells.each((i, element) => {
// console.log("cell element", i, $(element).text().trim())
if (i === 0) {
configPost.key = $(element).text().trim();
}
if (i === 1) {
configPost.value = $(element).text().trim();
}
if (i === 3) {
configPost.description = $(element).text().replace(' ', '').replace(' ', '').trim();
}
})
console.log("lignes de tableau", keys.length)
console.log("ligne sélectionnée", randNumber)
let cells = $(foundLine[randNumber]).find('td')
if (!configPost.key) {
console.error("ERROR no key found", $(cells).html())
console.error('no key BOOOOOOOOOOH');
getElementCartographique();
return;
}
keys = Object.keys(cells);
// console.log("cellules trouvées", keys.length)
console.log("\n")
let configPost = {
key: '',
value: '',
description: '',
long_desc: '',
}
cells.each((i, element) => {
// console.log("cell element", i, $(element).text().trim())
if (i === 0) {
configPost.key = $(element).text().trim();
}
if (i === 1) {
configPost.value = $(element).text().trim();
}
if (i === 3) {
configPost.description = $(element).text().replace(' ', '').replace(' ', '').trim();
}
})
if (!configPost.key) {
console.error("ERROR no key found", $(cells).html())
console.error('no key BOOOOOOOOOOH');
getElementCartographique();
return;
}
// aller chercher le lien
// si la clé est une description avec des espaces, faire un lien vers la Key
let pageLink = `FR:Tag:${configPost.key}=${configPost.value}`
// si la clé est une description avec des espaces, faire un lien vers la Key
let pageLink = `FR:Tag:${configPost.key}=${configPost.value}`
if (configPost.value.indexOf(' ') !== -1) {
pageLink = `FR:Key:${configPost.key}`
}
configPost.link = `https://wiki.openstreetmap.org/wiki/${pageLink}`
if (configPost.value.indexOf(' ') !== -1) {
pageLink = `FR:Key:${configPost.key}`
}
configPost.link = `https://wiki.openstreetmap.org/wiki/${pageLink}`
const url = configPost.link
// get a random page wich is not a discussion
const url = configPost.link
// get a random page wich is not a discussion
// certaines valeurs ont des espaces, donc pas utilisable en url
if (configPost.value.indexOf(' ') === -1 || configPost.value.indexOf('<') === -1 || configPost.value.indexOf('>') === -1) {
// certaines valeurs ont des espaces, donc pas utilisable en url
if (configPost.value.indexOf(' ') === -1 || configPost.value.indexOf('<') === -1 || configPost.value.indexOf('>') === -1) {
console.log("url", url)
rp(url).then(function (html) {
//success!
const descriptionStrophe = '';
const filteredHtml = $('.mw-parser-output', html).find('p')
console.log("url", url)
rp(url).then(function (html) {
//success!
const descriptionStrophe = '';
const filteredHtml = $('.mw-parser-output', html).find('p')
// trouver l'image de description
configPost.download_description_src = getSourceOfDescriptorPageContent(html)
// trouver l'image de description
configPost.download_description_src = getSourceOfDescriptorPageContent(html)
console.log("filteredHtml", filteredHtml.length)
configPost.long_desc = $(filteredHtml).text().substring(0, 250)
console.log("filteredHtml", configPost.long_desc)
console.log("filteredHtml", filteredHtml.length)
configPost.long_desc = $(filteredHtml).text().substring(0, 250)
console.log("filteredHtml", configPost.long_desc)
// image de description:
// let imgSelector = ".description a.image img"
console.log("✅ cette page existe bien en Français sur le wiki OSM")
foundExistingWikiPageInFrench = true;
sendMessageWikiTagOfTheDay(makePostMessageFromObj(configPost), configPost.download_description_src)
// image de description:
// let imgSelector = ".description a.image img"
console.log("✅ cette page existe bien en Français sur le wiki OSM")
foundExistingWikiPageInFrench = true;
sendMessageWikiOSMOfTheDay(makePostMessageFromObj(configPost), configPost.download_description_src)
}, (err) => {
console.error(err.statusCode)
if (err.statusCode == 404) {
console.log("\n Cette page n'existe pas encore, c'est le moment de la créer.", `https://wiki.openstreetmap.org/w/index.php?title=FR:${configPost.key}=${configPost.value}`) //FR:Tag:boundary%3Dborder_zone&action=edit)
console.log("Essayez de voir si la version en Anglais existe.")
let englishPage = `https://wiki.openstreetmap.org/wiki/Tag:${configPost.key}=${configPost.value}`
let englishKey = `https://wiki.openstreetmap.org/wiki/Key:${configPost.key}`
console.log("-> ", englishPage)
checkExistenceOfWebPage(englishPage).catch(err => {
checkExistenceOfWebPage(englishKey)
});
console.log("bon spa tout ça on essaie une autre ligne")
getElementCartographique()
}, (err) => {
console.error(err.statusCode)
if (err.statusCode == 404) {
console.log("\n Cette page n'existe pas encore, c'est le moment de la créer.", `https://wiki.openstreetmap.org/w/index.php?title=FR:${configPost.key}=${configPost.value}`) //FR:Tag:boundary%3Dborder_zone&action=edit)
console.log("Essayez de voir si la version en Anglais existe.")
let englishPage = `https://wiki.openstreetmap.org/wiki/Tag:${configPost.key}=${configPost.value}`
let englishKey = `https://wiki.openstreetmap.org/wiki/Key:${configPost.key}`
console.log("-> ", englishPage)
checkExistenceOfWebPage(englishPage).catch(err => {
checkExistenceOfWebPage(englishKey)
});
console.log("bon spa tout ça on essaie une autre ligne")
getElementCartographique()
}
})
} else {
console.log("result", configPost)
rp(configPost.link).then((result) => {
console.log("oui cette page existe!", url)
sendMessageWikiOSMOfTheDay(makePostMessageFromObj(result))
}, (err) => {
console.log("hé non. WTF ?")
// getElementCartographique()
})
}
})
} else {
console.log("result", configPost)
rp(configPost.link).then((result) => {
console.log("oui cette page existe!", url)
sendMessageWikiTagOfTheDay(makePostMessageFromObj(result))
}, (err) => {
console.log("hé non. WTF ?")
// getElementCartographique()
})
return configPost;
}
return configPost;
});
});
}
@ -215,19 +217,19 @@ export default function getElementCartographique() {
* @returns {*}
*/
function checkExistenceOfWebPage(url) {
if (foundExistingWikiPageInFrench) {
console.log("already found foundExistingWikiPageInFrench, no need to check other page", url)
return;
}
return rp(url).then((html) => {
console.log("oui cette page existe!", url)
}, (err) => {
console.log("hé non. WTF ?")
})
if (foundExistingWikiPageInFrench) {
console.log("already found foundExistingWikiPageInFrench, no need to check other page", url)
return;
}
return rp(url).then((html) => {
console.log("oui cette page existe!", url)
}, (err) => {
console.log("hé non. WTF ?")
})
}
function makePostMessageFromObj(post_obj) {
return `
return `
# Le tag OSM du jour : ${post_obj.key}=${post_obj.value} 🗺🏷
${post_obj.link}
@ -238,93 +240,85 @@ ${post_obj.description ? post_obj.description.trim() : ''}
}
function getQuery() {
// var url = "https://wiki.openstreetmap.org/w/api.php";
var url = "https://fr.wikipedia.org/w/api.php";
var params = {
action: "query",
format: "json",
list: "random",
enhanced: "1",
hiderobot: "1",
urlversion: "2",
invert: "1",
rnnamespace: "all-discussions%3B1%3B2%3B3%3B5%3B7%3B9%3B11%3B120%3B121%3B122%3B123%3B201%3B203", // WIKI et FR:
lang: "FR",
rnlimit: "5"
};
// var url = "https://wiki.openstreetmap.org/w/api.php";
var url = "https://fr.wikipedia.org/w/api.php";
var params = {
action: "query",
format: "json",
list: "random",
enhanced: "1",
hiderobot: "1",
urlversion: "2",
invert: "1",
rnnamespace: "all-discussions%3B1%3B2%3B3%3B5%3B7%3B9%3B11%3B120%3B121%3B122%3B123%3B201%3B203", // WIKI et FR:
lang: "FR",
rnlimit: "5"
};
url = url + "?&origin=*";
Object.keys(params).forEach(function (key) {
url += "&" + key + "=" + params[key];
});
url = url + "?&origin=*";
Object.keys(params).forEach(function (key) {
url += "&" + key + "=" + params[key];
});
return fetch(url)
.then(function (response) {
return response.json();
})
.then(function (response) {
const randoms = response.query.random;
return fetch(url)
.then(function (response) {
return response.json();
})
.then(function (response) {
const randoms = response.query.random;
for (let ii = 0; ii < randoms.length; ii++) {
// console.log(randoms[ii].title);
console.log("randoms[ii]", randoms[ii])
}
})
.catch(function (error) {
console.log(error);
});
for (let ii = 0; ii < randoms.length; ii++) {
// console.log(randoms[ii].title);
console.log("randoms[ii]", randoms[ii])
}
})
.catch(function (error) {
console.log(error);
});
}
// run
const res = getElementCartographique()
// console.log("res", res)
export function sendMessageWikiOSMOfTheDay(message, download_description_src) {
function sendMessageWikiTagOfTheDay(message, download_description_src) {
let configPost = {
author: 'curator',
// document.querySelector('.d_image img').attributes['src']
// image: '/home/tykayn/www/multi-account-post-schedule-mastodon/assets/osm_post_' + randomIntFromInterval(1 , 5) + '.jpg',
image: `${process.cwd()}/assets/blog_posts_medias/assets/osm_post_${randomIntFromInterval(1, 5)}.jpg`,
message,
}
if (download_description_src) {
let filePathImage = `${process.cwd()}/assets/blog_posts_medias/osm_wiki_description_page.jpg`
if (download_description_src) {
console.log("firstPictureSource found", download_description_src)
// check if picture already exist
console.log('on récupère l image de description : ', filePathImage)
downloadImage(download_description_src, filePathImage)
.then((res) => {
// suite du poste avec upload d'image
console.log('média téléchargé, on envoie le post')
configPost.image = filePathImage;
sendPostMastodon(configPost)
},
(err) => {
console.log('pas dimage trouvée pour l URL ', download_description_src, err)
sendPostMastodon(configPost)
}
)
.catch((err) => {
console.log('erreur avec cette URL ', download_description_src, err)
sendPostMastodon(configPost)
})
let configPost = {
author: 'curator',
// document.querySelector('.d_image img').attributes['src']
// image: '/home/tykayn/www/multi-account-post-schedule-mastodon/assets/osm_post_' + randomIntFromInterval(1 , 5) + '.jpg',
image: `${process.cwd()}/assets/blog_posts_medias/assets/osm_post_${randomIntFromInterval(1, 5)}.jpg`,
message,
}
if (download_description_src) {
let filePathImage = `${process.cwd()}/assets/blog_posts_medias/osm_wiki_description_page.jpg`
if (download_description_src) {
console.log("firstPictureSource found", download_description_src)
} else {
console.log('no image description')
sendPostMastodon(configPost)
}
// check if picture already exist
console.log('on récupère l image de description : ', filePathImage)
downloadImage(download_description_src, filePathImage)
.then((res) => {
// suite du poste avec upload d'image
console.log('média téléchargé, on envoie le post')
configPost.image = filePathImage;
sendPostMastodon(configPost)
},
(err) => {
console.log('pas dimage trouvée pour l URL ', download_description_src, err)
sendPostMastodon(configPost)
}
)
.catch((err) => {
console.log('erreur avec cette URL ', download_description_src, err)
sendPostMastodon(configPost)
})
}
} else {
console.log('no image description')
sendPostMastodon(configPost)
}
}