migration wip

test-decode7.js

@@ -0,0 +1,29 @@
+// Test decoding
+const fs = require('fs');
+
+// Read actual file
+const content = fs.readFileSync('data/processed/pages.json', 'utf8');
+const idx = content.indexOf('bg_image=');
+const snippet = content.substring(idx, idx + 30);
+
+console.log('File snippet:', snippet);
+console.log('Bytes:', Buffer.from(snippet).toString('hex'));
+
+// The file has: bg_image=”10432″
+// Which is: bg_image= + ” + 10432 + ″
+
+// Test decoding
+const test = 'bg_image=”45569″';
+console.log('\nTest:', test);
+
+// Decode decimal entities
+const decoded = test
+  .replace(/”/g, '"')  // ” -> "
+  .replace(/″/g, '"'); // ″ -> "
+
+console.log('Decoded:', decoded);
+console.log('Match:', decoded === 'bg_image="45569"');
+
+// Check what the file actually has
+console.log('\nFile has literal ”:', snippet.includes('”'));
+console.log('File has literal ”:', snippet.includes('”'));