Add sector classification module, integrate into all 7 scrapers, fix CF pagination

This commit is contained in:
Peter Foster
2026-02-14 17:12:51 +00:00
parent d1aa21c59f
commit 771fcf9d76
23 changed files with 2044 additions and 83 deletions

55
etendersni-api.mjs Normal file
View File

@@ -0,0 +1,55 @@
import axios from 'axios';
import * as cheerio from 'cheerio';
const searchUrl = 'https://etendersni.gov.uk/api/search';
(async () => {
try {
console.log('Fetching search page...');
const resp = await axios.get(searchUrl, { timeout: 10000 });
const $ = cheerio.load(resp.data);
console.log('=== PAGE STRUCTURE ===');
console.log('Title:', $('title').text());
console.log('\n=== FORMS ===');
$('form').each((i, el) => {
const action = $(el).attr('action');
const method = $(el).attr('method');
const id = $(el).attr('id');
console.log(`Form ${i}: method=${method} action=${action} id=${id}`);
// Look for inputs
$(el).find('input, select, textarea').each((j, inp) => {
const name = $(inp).attr('name');
const type = $(inp).attr('type');
const value = $(inp).attr('value');
console.log(` Input: name=${name} type=${type} value=${value}`);
});
});
console.log('\n=== TABLES/RESULTS ===');
const tables = $('table');
console.log('Found', tables.length, 'tables');
if (tables.length > 0) {
const firstTable = tables.eq(0);
console.log('\nFirst table rows:');
firstTable.find('tr').slice(0, 3).each((i, row) => {
const cells = $(row).find('td, th');
const text = cells.map((j, cell) => $(cell).text().trim().substring(0, 30)).get();
console.log(` Row ${i}:`, text);
});
}
console.log('\n=== LINKS IN RESULTS ===');
$('a[href*="view"], a[href*="notice"]').slice(0, 10).each((i, el) => {
const href = $(el).attr('href');
const text = $(el).text().trim().substring(0, 50);
console.log(` ${text} => ${href}`);
});
} catch (e) {
console.error('Error:', e.message);
}
})();