deepseek_chatbot/scraper/getEntries.js

const fs = require('fs');
const path = require('path');

const IN_FILE = 'knowledgebase.html';
const OUT_DIR = './entries';

// read file
const html = fs.readFileSync(IN_FILE, 'utf8');

// MAGIC
const entryRegex = /<tr class="cb_pointer">([\s\S]*?<span name="(\d+)"[\s\S]*?<\/td>)[\s\S]*?<\/tr>/g;
const entries = {};
let match;

while ((match = entryRegex.exec(html)) !== null) {
    const id = match[2];
    const block = match[1].trim();
    entries[id] = block;
}

if (!fs.existsSync(OUT_DIR)) fs.mkdirSync(OUT_DIR);

function decodeHtmlEntities(str) {
    return str
        // .replace(/&lt;/g, "<")
        // .replace(/&gt;/g, ">")
        // .replace(/&quot;/g, "\"")
        // .replace(/&amp;/g, "&")
        // .replace(/<(?!br\s*\/?)[^>]+>/gi, '');
}

for (const [id, content] of Object.entries(entries)) {
    const filePath = path.join(OUT_DIR, `${id}.txt`);
    // fs.writeFileSync(filePath, content, 'utf8');
    // console.log(`📄 Gespeichert: ${filePath}`);
}

console.log(decodeHtmlEntities(entries[25]).trim());

console.log(`✅ ${Object.keys(entries).length} Einträge gespeichert.`);