x

parse.js

@@ -0,0 +1,31 @@
+const cheerio = require('cheerio');
+const fs = require('fs');
+const html = fs.readFileSync('mega01.html', 'utf8');
+const $ = cheerio.load(html);
+
+console.log('--- Page Title ---');
+console.log($('title').text());
+
+console.log('\n--- Main Content Sections ---');
+const sections = [];
+$('section, .section, [data-section]').each((i, el) => {
+  sections.push({
+    index: i + 1,
+    id: $(el).attr('id') || 'no-id',
+    class: $(el).attr('class') || 'no-class',
+    text: $(el).text().replace(/\s+/g, ' ').substring(0, 500)
+  });
+});
+if (sections.length === 0) {
+  // If no semantic tags, maybe just print divs with specific classes
+  $('div').each((i, el) => {
+    if($(el).attr('class') && $(el).attr('class').includes('py-')) {
+       sections.push({
+         index: i + 1,
+         class: $(el).attr('class') || 'no-class',
+         text: $(el).text().replace(/\s+/g, ' ').substring(0, 500)
+       });
+    }
+  });
+}
+console.log(JSON.stringify(sections, null, 2));