migration wip

This commit is contained in:
2025-12-30 00:06:54 +01:00
parent 3efbac78cb
commit 89dbf8af87
94 changed files with 5674 additions and 308 deletions

View File

@@ -0,0 +1,125 @@
#!/usr/bin/env node
// Test the final function with actual raw data
const fs = require('fs');
const path = require('path');
// Load the actual raw data
const rawData = JSON.parse(fs.readFileSync('data/raw/2025-12-27T21-26-12-521Z/pages.en.json', 'utf8'));
const testExcerpt = rawData[0].excerptHtml;
console.log('=== Testing Final Function ===');
console.log('Raw excerpt (first 200 chars):');
console.log(testExcerpt.substring(0, 200));
console.log('');
// The function from process-data.js
function processExcerptShortcodes(excerptHtml) {
if (!excerptHtml) return '';
let processed = excerptHtml;
// First, decode HTML entities to regular characters
// Handle both numeric entities (”) and named entities (")
processed = processed
// Decode numeric HTML entities first
.replace(/&#(\d+);/g, (match, dec) => String.fromCharCode(dec))
// Then handle any remaining Unicode characters
.replace(/”/g, '"') // ” - Right double quote
.replace(/“/g, '"') // “ - Left double quote
.replace(/„/g, ',') // „ - Low double quote
.replace(/‟/g, '"') // ‟ - High double quote
.replace(//g, "'") // - Left single quote
.replace(//g, "'") // - Right single quote
.replace(//g, '-') // - En dash
.replace(/—/g, '—') // — - Em dash
.replace(/…/g, '…') // … - Ellipsis
.replace(/″/g, '"') // ″ - Inches/Prime
.replace(//g, "'") // - Feet/Prime
.replace(//g, ',') // - Single low quote
.replace(//g, '`') // - Single high reversed quote
.replace(/•/g, '•') // • - Bullet
.replace(/€/g, '€') // € - Euro
// Named HTML entities
.replace(/"/g, '"')
.replace(/'/g, "'")
.replace(//g, "'")
.replace(//g, "'")
.replace(/“/g, '"')
.replace(/”/g, '"')
.replace(//g, '-')
.replace(/—/g, '—')
.replace(/…/g, '…')
.replace(/•/g, '•')
.replace(/€/g, '€');
// Process WPBakery shortcodes with HTML entities
processed = processed
// vc_row - convert to div with classes
.replace(/\[vc_row([^\]]*)\]/gi, (match, attrs) => {
const classes = ['vc-row'];
if (attrs.includes('full_width_background')) classes.push('full-width-bg');
if (attrs.includes('in_container')) classes.push('in-container');
if (attrs.includes('full_width_content')) classes.push('full-width-content');
return `<div class="${classes.join(' ')}">`;
})
.replace(/\[\/vc_row\]/gi, '</div>')
// vc_column - convert to div with classes
.replace(/\[vc_column([^\]]*)\]/gi, (match, attrs) => {
const classes = ['vc-column'];
if (attrs.includes('1/1')) classes.push('col-1-1');
if (attrs.includes('1/2')) classes.push('col-1-2');
if (attrs.includes('1/3')) classes.push('col-1-3');
if (attrs.includes('2/3')) classes.push('col-2-3');
if (attrs.includes('1/4')) classes.push('col-1-4');
if (attrs.includes('3/4')) classes.push('col-3-4');
if (attrs.includes('5/12')) classes.push('col-5-12');
if (attrs.includes('7/12')) classes.push('col-7-12');
return `<div class="${classes.join(' ')}">`;
})
.replace(/\[\/vc_column\]/gi, '</div>')
// vc_column_text - convert to div
.replace(/\[vc_column_text([^\]]*)\]/gi, '<div class="vc-column-text">')
.replace(/\[\/vc_column_text\]/gi, '</div>');
// Remove any remaining shortcodes
processed = processed.replace(/\[.*?\]/g, '');
// Clean up any HTML that might be broken
processed = processed.replace(/<p[^>]*>\s*<\/p>/gi, '');
processed = processed.replace(/<div[^>]*>\s*<\/div>/gi, '');
// Normalize whitespace
processed = processed.replace(/\s+/g, ' ').trim();
return processed;
}
const result = processExcerptShortcodes(testExcerpt);
console.log('After processing:');
console.log(result);
console.log('');
// Check for entities
const hasEntities = /[”“‘’–—]/.test(result);
const hasNumericEntities = /&#\d+;/.test(result);
const hasShortcodes = /\[vc_row|\[vc_column/.test(result);
console.log('=== Verification ===');
console.log('Has Unicode entities:', hasEntities);
console.log('Has numeric entities:', hasNumericEntities);
console.log('Has shortcodes:', hasShortcodes);
console.log('Has proper HTML:', result.includes('<div class="vc-row"') || result.includes('<div class="vc-column"'));
console.log('');
if (!hasEntities && !hasNumericEntities && !hasShortcodes && result.includes('<div class="vc-row"')) {
console.log('✅ SUCCESS: Function works correctly!');
} else {
console.log('❌ Issues found');
}