migration wip

test-decode44.js

@@ -0,0 +1,29 @@
+// Test what's in the file
+const fs = require('fs');
+const content = fs.readFileSync('data/processed/pages.json', 'utf8');
+const pages = JSON.parse(content);
+const start = pages.find(p => p.slug === 'start' && p.locale === 'de');
+const html = start.contentHtml;
+
+// Find bg_image patterns
+const matches = html.match(/bg_image=[^>]+/g);
+if (matches) {
+  console.log('Found bg_image patterns:');
+  matches.slice(0, 2).forEach(m => {
+    console.log('  ', m.substring(0, 80));
+  });
+}
+
+// Check what's actually between bg_image and the number
+const test = html.match(/bg_image=([^>]+)/);
+if (test) {
+  const after = test[1];
+  console.log('\nAfter bg_image=:', JSON.stringify(after.substring(0, 30)));
+  
+  // Check if it starts with ”
+  if (after.startsWith('”')) {
+    console.log('✓ Has ” (decimal entity)');
+    const id = after.substring(7, 12); // Skip ” and get 5 digits
+    console.log('  ID would be:', id);
+  }
+}