dashboard/scripts/azure/update-data

#!/usr/bin/env node

/**
 * This script reads the Azure website and updates our hard-coded file it needed
 */

const fs = require('fs');
const path = require('path');
const axios = require('axios');

console.log('Updating Azure regions in Availability Zones');
console.log('============================================');
console.log('');

const SCRIPT = process.argv[1];
const DIR = path.resolve(SCRIPT, '../../..');
const ENDPOINTS_URL = 'https://learn.microsoft.com/en-us/azure/reliability/regions-list';
const TS_FILE = path.resolve(DIR, 'pkg/aks/util/aks-regions.ts');
const COLUMN_REGEX = /<td>(.*)<\/td>/;
const US_GOV = 'usgov';

function findRegions(body) {
  const regions = [];
  let inTable = false;
  let inRow = false;
  let inColumn = false;
  let row = [];
  let column = ''

  body.split('\n').forEach((line) => {
    line = line.trim();

    if (line.indexOf('<table') >= 0) {
      inTable = true;
    } else if (inTable) {
      if (line.indexOf('</table>') >= 0) {
        inTable = false;
      } else if (inRow) {
        if (!inColumn && line.indexOf('<td>') === 0) {
          inColumn = true;
          column = line;

          if (line.endsWith('</td>')) {
            row.push(column);
            inColumn = false;
          }
        } else if (inColumn && line.endsWith('</td>')) {
          column += line;
          row.push(column);
          inColumn = false;
        } else if (line.indexOf('</tr>') === 0) {
          inRow = false;

          if (row.length === 5) {
            if (!row[0].startsWith('<td><img') && !row[1].startsWith('<td></td>')) {
              const m = row[0].match(COLUMN_REGEX);

              if (m && m.length === 2) {
                let region = m[1].toLowerCase();
                region = region.replaceAll(' ', '');            
                regions.push(region);
              }
            }
          }
        }
      } else if (!inRow && line.indexOf('<tr>') == 0) {
        inRow = true;
        row = [];
      }
    }
  });

  return regions.sort();
}

function cleanItem(item) {
  const p = item.split(':');

  if (p.length === 2) {
    return p[0];
  }

  return undefined;
}

function readExisting() {
  const data = fs.readFileSync(TS_FILE).toString();
  const items = [];

  let processing = false;
  let done = false;

  data.split(/\r?\n/).forEach(line =>  {
    line = line.trim();

    if (!done) {
      if (!processing) {
        if (line.startsWith('export const ')) {
          processing = true;
        }
      } else {
        if (line === '} as any;') {
          processing = false;
          done = true;
        } else {
          const r = cleanItem(line);

          if (r) {
            items.push(r);
          }
        }
      }
    }
  });

  return items.sort();
}

function writeNewFile(regions) {
  let lines = ['export const regionsWithAvailabilityZones = {'];

  let maxLength = 1;

  regions.forEach((r) => {
    if (r.length > maxLength) {
      maxLength = r.length + 1; // Add room for :
    }
  });

  regions.forEach((r) => {
    const key = `${r}:`.padEnd(maxLength);

    lines.push(`  ${ key } true,`)
  });

  lines.push('} as any;');
  lines.push('');

  fs.writeFileSync(TS_FILE, lines.join('\n'));
}

const existing = readExisting();

axios.get(ENDPOINTS_URL, { responseType: 'document' }).then((res) => {
  if (res.status !== 200 || !res.data) {
    console.error(`Could not fetch Azure data - status ${ res.statusCode } ${ res.statusMessage }`);

    process.exit(1);
  }

  const regions = findRegions(res.data);
  const govRegions = existing.filter((r) => r.startsWith(US_GOV));
  const existingCount = existing.length;

  if (regions.length === 0) {
    console.error('Could not fetch data from Azure - no regions found');
    process.exit(2);
  }

  console.log(`Existing regions in our codebase: ${ existingCount - govRegions.length } (Excluding US Government regions)`);
  console.log(existing);
  console.log('');

  console.log(`Regions from Azure: ${ regions.length }`);
  console.log(regions);
  console.log('');

  const removed = [];

  existing.forEach((r) => {
    if (r.startsWith(US_GOV)) {
      regions.push(r);
    } else if (!regions.includes(r)) {
      removed.push(r);
    }
  });

  console.log('');
  console.log(`US Government regions to be retained: ${ govRegions.join(', ') }`);
  console.log('');

  if (removed.length) {
    console.log(`These region(s) have been removed: ${ removed.join(', ') }`);
  }

  const added = [];

  regions.forEach((r) => {
    if (!existing.includes(r)) {
      added.push(r);
    }
  });

  if (added.length) {
    console.log(`These region(s) have been added: ${ added.join(', ') }`);
  }

  if (added.length || removed.length) {
    // Sort the regions (needs a resort if we added back in usgov regions)
    writeNewFile(regions.sort());

    console.log('');
    console.log(`Regions updated from ${ existingCount } to ${ regions.length }`);
    console.log('');
  } else {
    console.log('No changes to the Azure region data');
  }
});