Optimizing API Call Strategy with Caching and Rate Limiting in Retool

💬 Queries and Resources
, ,
9

I could not implement PQueue as per your suggestion @MikeCB because it did not seem compatible with Retool. The library is added via a CDN link, but is not accessible from queries. Maybe it's the same problem as described here. Anyways, instead I found an alternative called Bottleneck.

Here's my implementation for everyone interested in throttling POST requests:

const limiter = new Bottleneck({
  maxConcurrent: 1,
  minTime: 1000 // At most 1 request per second
});

const uniqueData = formatDataAsArray(getUniqueCurrencies.data)

const results = {};

// Use Promise.all to wait for all scheduled tasks to complete
await Promise.all(uniqueData.map(data => {
  const key = `${data.date}_${data.source}`;
  if (data.source === 'EUR') {
    results[key] = 1; // Directly assign value if source is 'EUR'
    return Promise.resolve(); // Immediately resolve for 'EUR'
  } else {
    // Schedule the task and update results within the promise
    return limiter.schedule(() => 
      getConversionRates.trigger({
      additionalScope: {
        conversionDate: data.date,
        sourceCurrency: data.source,
        targetCurrency: 'EUR',
      },
    }).then(res => {
        results[key] = res.data;
      }));
  }
}));

return results;

It works and it's faster than before. Sometimes I would notice "rate exceeded" warnings, but I guess I could increase the minTime to make sure it observes more than a second between requests. I'm sure this snippet could be improved.

Thanks for your help!

J.