User:Polygnotus/Scripts/GetContext.js: Difference between revisions

Revision as of 23:07, 8 September 2024

Revision as of 23:10, 8 September 2024

Line 1:

function getWikipediaContextgetWikipediaContent(articleName, word, n) {

return new Promise((resolve, reject) => {

const url = `https://en.wikipedia.org/w/api.php?action=parsequery&pageprop=revisions&titles=${encodeURIComponent(articleName)}&rvprop=content&format=json&propformatversion=text2&origin=*`;

fetch(url)

.then(response => response.json())

.then(data => {

const htmlContentwikitext = data.parsequery.textpages['*'0].revisions[0].content;

const textContent = htmlContent.replace(/<[^>]+>/g, ' ').replace(/\s+/g, ' ').trim();

// Remove comments and nowiki tags

const textContentcleanText = htmlContentwikitext.replace(/<!--[^>\s\S]+*?-->/g, ' ').replace(/\s+/g, ' ').trim();

.replace(/<nowiki>[\s\S]*?<\/nowiki>/g, '');

const wordRegex = new RegExp(`\\b${word}\\b`, 'gi');

Line 13 ⟶ 16:

let index = -1;

while (count < n && (match = wordRegex.exec(textContentcleanText)) !== null) {

count++;

if (count === n) {

Line 26 ⟶ 29:

}

const words = textContentcleanText.split(/\s+/);

const wordIndex = words.findIndex((_, i) => words.slice(0, i + 1).join(' ').length >= index);

Line 39 ⟶ 42:

}

// Run the example automatically

getWikipediaContextgetWikipediaContent('Horseshoe crab', 'trilobite', 3)

.then(result => {

console.log("Result:");