prev() wrongly skipping back an extra day on first call in specific case of time change

[ellis]

Here is sample code for v5.1.1 that demonstrates the skipped day:

import { CronExpressionParser } from 'cron-parser';
var interval = CronExpressionParser.parse("0 2 * * *", {tz: "America/New_York", currentDate: "2020-03-08T03:01:00.000-04:00"});
console.log(interval.prev().toISOString())
// 2020-03-07T07:00:00.000Z - ERROR: should be 2020-03-08T07:00:00.000Z
console.log(interval.next().toISOString())
// 2020-03-08T07:00:00.000Z

Here is another example where the error is even clearer:

import { CronExpressionParser } from 'cron-parser';
const interval = CronExpressionParser.parse("0 2 * * *", {tz: "America/New_York", currentDate: "2020-03-08T03:01:00.000-04:00"});
console.log(interval.next().toISOString())
// 2020-03-09T06:00:00.000Z
console.log(interval.prev().toISOString())
// 2020-03-07T07:00:00.000Z - ERROR: should be 2020-03-08T07:00:00.000Z

The equivalent code had previously worked in 4.9.0; my unit tests caught the error after I upgraded cron-parser today.

[harrisiirak]

Hi @ellis!

Thanks for reporting this. Strangely enough, v4.9.0 and the latest master (also tested with the initial v5.0.0 release) produce the same faulty result for me.

import { CronExpressionParser } from './src';
import cronParser from './benchmarks/versions/4.9.0/node_modules/cron-parser';

const interval = CronExpressionParser.parse('0 2 * * *', {
  tz: 'America/New_York',
  currentDate: '2020-03-08T03:01:00.000-04:00',
});
console.log(interval.next().toISOString());
console.log(interval.prev().toISOString());

// 2020-03-09T06:00:00.000Z
// 2020-03-07T07:00:00.000Z

const interval2 = cronParser.parseExpression('0 2 * * *', {
  tz: 'America/New_York',
  currentDate: '2020-03-08T03:01:00.000-04:00',
});
console.log(interval2.next().toISOString());
console.log(interval2.prev().toISOString());

// 2020-03-09T06:00:00.000Z
// 2020-03-07T07:00:00.000Z

However, this behavior is rather bizarre and unexpected. I have a feeling it might be related to issue #273, as the current problem also appears to involve DST.

Going to investigate this soon.

Best regards