extract-classifications.js

const fs = require('fs');
const path = require('path');
const { parse } = require('node-html-parser');

// Read the HTML file
const htmlPath = path.join(__dirname, 'classification.html');
const htmlContent = fs.readFileSync(htmlPath, 'utf-8');

// Parse HTML
const root = parse(htmlContent);

// Find all elements to get both <p> tags and <tr> tags in order
const allElements = root.querySelectorAll('p, tr');

const classifications = [];
let currentCategory = null;

allElements.forEach(element => {
    // Check if it's a <p> tag with category information
    if (element.tagName === 'P') {
        const fonts = element.querySelectorAll('font');
        if (fonts.length >= 2) {
            // First font might have the code in <strong>
            const codeStrong = fonts[0].querySelector('strong');
            // Second font might have the category name in <strong>
            const nameStrong = fonts[1].querySelector('strong');
            
            if (codeStrong && nameStrong) {
                const code = codeStrong.text.trim();
                const name = nameStrong.text.trim();
                if (code && name) {
                    currentCategory = { code, name };
                }
            }
        }
    }
    
    // Check if it's a <tr> tag with classification information
    if (element.tagName === 'TR') {
        const cells = element.querySelectorAll('td');
        
        if (cells.length === 3) {
            const codeCell = cells[1].querySelector('font');
            const nameCell = cells[2].querySelector('font');
            
            if (codeCell && nameCell) {
                const code = codeCell.text.trim();
                const name = nameCell.text.trim();
                
                if (code && name) {
                    // If the name is "General" and we have a current category, use the category name
                    if (name === 'General' && currentCategory) {
                        classifications.push({ code, name: "General " + currentCategory.name });
                    } else if (name !== 'General') {
                        classifications.push({ code, name });
                    }
                }
            }
        }
    }
});

// Write to JSON file
const jsonPath = path.join(__dirname, 'classifications.json');
fs.writeFileSync(jsonPath, JSON.stringify(classifications, null, 2));

// Also write to CSV
const csvPath = path.join(__dirname, 'classifications.csv');
const csvContent = 'code,name\n' + 
    classifications.map(c => `"${c.code}","${c.name}"`).join('\n');
fs.writeFileSync(csvPath, csvContent);

console.log(`✓ Extracted ${classifications.length} classifications`);
console.log(`  Output: ${jsonPath}`);
console.log(`  Output: ${csvPath}`);
console.log('\nFirst 5 entries:');
classifications.slice(0, 5).forEach(c => {
    console.log(`  ${c.code} → ${c.name}`);
});