multi-account-post-schedule.../helpers/wiki_osm.mjs

224 lines
7.8 KiB
JavaScript
Raw Normal View History

2022-08-04 14:16:43 +02:00
// https://www.mediawiki.org/wiki/Manual:Random_page
import fetch from "node-fetch"
import rp from "request-promise";
import $ from "cheerio";
2022-08-04 16:30:18 +02:00
import fs from "fs";
import path from 'path';
2022-08-04 23:07:59 +02:00
import {getRandomElementOfArray, randomIntFromInterval} from "./utils.js";
2022-08-04 16:30:18 +02:00
const __dirname = path.resolve();
2022-08-04 23:07:59 +02:00
2022-08-04 16:30:18 +02:00
function getRandomWikiOSMPage() {
2022-08-04 14:16:43 +02:00
// 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)
2022-08-04 16:30:18 +02:00
.then(function (html) {
2022-08-04 14:16:43 +02:00
//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;
})
2022-08-04 16:30:18 +02:00
.catch(function (err) {
2022-08-04 14:16:43 +02:00
//handle error
console.error(err)
});
}
2022-08-04 16:30:18 +02:00
export default function getElementCartographique() {
2022-08-04 14:16:43 +02:00
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 = ""
2022-08-05 21:30:04 +02:00
return fs.readFile(__dirname + '/assets/documents/elements_carto_osm.html', 'utf8', function (err, html) {
2022-08-04 16:30:18 +02:00
if (err) {
return console.log(err);
}
// console.log(data);
//success!
const wikiUrls = [];
const listOfLinks = $(titleLink, html)
console.log("filteredHtml.length", listOfLinks.length)
const selectedTocLink = getRandomElementOfArray(listOfLinks).attribs.href;
console.log("selectedTocLink", selectedTocLink)
const listOfTableRows = $('.wikitable', html);
let keys = Object.keys(listOfTableRows);
let randNumber = randomIntFromInterval(1, keys.length)
// let randNumber = 3
console.log("rand", randNumber)
const foundLine = $(listOfTableRows[randNumber]).find('tr');
// console.log("randomTable",foundTable)
// on regarde les lignes du tableau
keys = Object.keys(foundLine);
randNumber = randomIntFromInterval(0, keys.length)
console.log("lignes de tableau", keys.length)
console.log("ligne sélectionnée", randNumber)
let cells = $(foundLine[randNumber]).find('td')
keys = Object.keys(cells);
console.log("cellules trouvées", keys.length)
2022-08-04 16:30:18 +02:00
let result = {
key: '',
value: '',
description: ''
}
2022-08-04 23:07:59 +02:00
cells.each((i, element) => {
console.log("cell element", i, $(element).text().trim())
2022-08-04 23:07:59 +02:00
if (i === 0) {
result.key = $(element).text().trim();
}
2022-08-04 23:07:59 +02:00
if (i === 1) {
result.value = $(element).text().trim();
}
2022-08-04 23:07:59 +02:00
if (i === 3) {
2022-08-05 21:30:04 +02:00
result.description = $(element).text().replace(' ', '').replace(' ', '').trim();
}
2022-08-04 23:07:59 +02:00
// let cell = $(element).find("td");
// console.log($(cell[0]));
})
2022-08-05 21:30:04 +02:00
if (!result.key) {
console.log("ERROR no key found", $(cells).html())
throw Error('no key BOOOOOOOOOOH');
}
// const tableRow = foundTable[randNumber] // get one of all tables but the TOC
// console.log("keys", keys)
// console.log("foundTable[0]", foundTable[0].text())
// console.log("foundTable.attribs", foundTable.attribs)
// console.log("tableRow", $( 'td' ,foundTable).text())
// console.log("tableRow",tableRow.data)
// const tableCells = $('td', tableRow)
// console.log("tableRow.length", tableCells.length)
// console.log("tag key value", cells[1].text())
// console.log("Description", cells[3])
// // console.log("icone", tableCells[4])
// console.log("photo", cells[5].find('img')?.attribs.href)
2022-08-05 21:30:04 +02:00
// aller chercher le lien
// si la clé est une description avec des espaces, faire un lien vers la Key
let pageLink = `FR:Tag:${result.key}=${result.value}`
if (result.value.indexOf(' ') !== -1) {
pageLink = `FR:Key:${result.key}`
}
result.link = `https://wiki.openstreetmap.org/wiki/${pageLink}`
const url = result.link
// get a random page wich is not a discussion
// certaines valeurs ont des espaces, donc pas utilisable en url
if (result.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("filteredHtml", filteredHtml)
if (filteredHtml[0] && filteredHtml[0].text()) {
result.long_desc = filteredHtml[0].text();
} else if (filteredHtml[1] && filteredHtml[1].text()) {
result.long_desc = filteredHtml[1].text();
}
// console.log("result", result)
console.log("result.long_desc", result.long_desc)
// image de description:
let imgSelector = ".description a.image img"
const message = makePostMessageFromObj(result)
console.log("message", message)
}, (err)=>{
console.error(err)
})
} else {
console.log("result", result)
const message = makePostMessageFromObj(result)
console.log("message", message)
}
2022-08-04 23:07:59 +02:00
return result;
2022-08-04 16:30:18 +02:00
});
2022-08-04 14:16:43 +02:00
}
2022-08-05 15:19:45 +02:00
function makePostMessageFromObj(result) {
return `
# [Le tag OSM du jour : ${result.key}=${result.value}](${result.link})
> ${result.description.trim()}
#osm #openstreetmap #wiki #rtfw
`
}
2022-08-04 14:16:43 +02:00
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"
};
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;
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);
});
2022-08-05 15:19:45 +02:00
}
// run
const res = getElementCartographique()
console.log("res", res)