Using the cache

The cache is off by default. When you turn it on, every downloaded period is saved to disk, and later requests for the same instrument, timeframe and dates read it from there instead of the network. It speeds up repeated and overlapping downloads, and lets interrupted CLI downloads resume quickly.

npx dukascopy-node -i eurusd -from 2021-02-01 -to 2021-03-01 -t m1 -f json --cache

import { getHistoricalRates } from 'dukascopy-node';

const data = await getHistoricalRates({
  instrument: 'eurusd',
  dates: { from: '2021-02-01', to: '2021-03-01' },
  timeframe: 'm1',
  format: 'json',
  useCache: true
});

The first run downloads the data and fills the cache. Run it again and the data comes from disk, which is much faster and sends no requests to Dukascopy.

Cached data lives in .dukascopy-cache in the current working directory. Choose another folder with cacheFolderPath (Node.js) or --cache-path (CLI):

npx dukascopy-node -i eurusd -from 2021-02-01 -to 2021-03-01 -t m1 --cache --cache-path ~/market-data/cache

const data = await getHistoricalRates({
  instrument: 'eurusd',
  dates: { from: '2021-02-01', to: '2021-03-01' },
  timeframe: 'm1',
  useCache: true,
  cacheFolderPath: '/home/me/market-data/cache'
});

To clear the cache, delete the folder. Add it to .gitignore if it lives inside a repository.

The cache stores the raw responses for completed periods: one file per hour of ticks, day of minute candles, month of hour candles or year of day candles. The output is built from these, so a cached period serves every format and every timeframe built from the same source; for example, m1 to m30 all reuse cached minute candles. Bid and ask candles are cached separately.

Periods that contain the current time are not cached, because their data is still changing:

When a whole batch is served from the cache, the pause between batches is skipped too.