Date cells

Excel supports cells that display formatted dates. For testing purposes I created a file with only one cell (A1) and entered the value 2014-09-28. The value was immediately formatted as a date and displayed as 9/28/14.

I wanted to see how I can reliably extract back the original date. I used the following program:

var xlsx = require('xlsx');

var workbook = xlsx.readFile(process.argv[2]);
for (var i = 0; i < workbook.SheetNames.length; i++){
    var worksheet = workbook.Sheets[workbook.SheetNames[i]];
    console.log(worksheet);
}

And got the following output (after running it with node myfile.xlsx);

{ A1: { t: 'n', v: 41910, w: '9/28/14' }, '!ref': 'A1' }
{}
{}

I see several problems here:

• The number 41910 is the number of days from some epoch, either 1900 or 1904, depending on where the file was created (Windows or Mac). There should be a way to tell how to treat this value.
• The t field (for type) indicates this is a number. Can it instead indicate it’s a date? or should there be a different property?
• There’s no way to retrieve the actual value that was entered - 2014-09-28 - so the original date, along with a reliable way of retrieving it, seems to be lost.

All of the above also applies for time fields.

Am I missing something? Is there a way to reliably get 2014-09-28 from the cell data?

[SheetJSDev]

My comment should not be construed to imply that I think any of this should be the “right” way to do things, and I’m sure we’ll come up with a better solution.

The t field (for type) indicates this is a number. Can it instead indicate it’s a date? or should there be a different property?

Quick review: the real thorn here is the bad date 2/29/1900 (1900 was not a leap year yet Excel recognizes it as a valid date). This is a bug from Lotus 1-2-3 that Excel replicated (decades later, it still haunts us). There is another issue regarding how to handle the day of week.

We could decide to throw away the corner case entirely and just store dates. In fact, the XLSX format allows for a cell type d for a date string. There would be problems at date code 60, but otherwise would be fine.

There should be a way to tell how to treat this value

The date1904 mode is stored in the (admittedly awkward) location Workbook.WBProps.date1904. Using the roo_190*_base files:

> require('xlsx').readFile('test_files/roo_1900_base.xlsx').Workbook.WBProps.date1904
'0'
> require('xlsx').readFile('test_files/roo_1904_base.xlsx').Workbook.WBProps.date1904
'1'

the original date, along with a reliable way of retrieving it, seems to be lost.

To convert the date, the SSF library has a handy parse_date_code function:

> require('xlsx').SSF.parse_date_code(41910, {date1904:false})
{ D: 41910,
  T: 0,
  u: 0,
  y: 2014,
  m: 9,
  d: 28,
  H: 0,
  M: 0,
  S: 0,
  q: 0 }
> require('xlsx').SSF.parse_date_code(41910, {date1904:true})
{ D: 41910,
  T: 0,
  u: 0,
  y: 2018,
  m: 9,
  d: 29,
  H: 0,
  M: 0,
  S: 0,
  q: 6 }

Those fields are described here: https://github.com/SheetJS/ssf/#usage

To recover a JS rendering of the date, take a look at the js-harb helper for this: https://github.com/SheetJS/js-harb/blob/master/bits/22_helpers.js#L7-L17

It probably makes sense for parse_date_code to return the original date object as well, as that would obviate the need for an additional function.

@hmalphettes @notatestuser @vratiu @wbrandongeorge thoughts?

[ligang-code2020]

Now I want to turn off the date conversion. For example, I want to export the string ‘1.9.3’, and sheet.js is automatically installed and replaced with 2003/1/9. How to turn it off

well. i find way to resolve my question.just to set up {type:“array”,raw:“true”}