Ability to export data as excel or csv

[chriddyp]

• For Excel files, only XLSX (not XLS) will be supported
• Only the data will be exported, formatting will not be exported
• The export will include the data in the current view. For example, if columns are hidden, sorted, or filtered, then the exported file will display the current view.
• Export will not protect users from “CSV Injection” attacks (https://www.owasp.org/index.php/CSV_Injection)
  Exporting Excel files may require a large 3rd party open source library. This library is too large to include in the default version of dash-table, so we will need to engineer dynamic JavaScript module loading as part of this requirement.

We will consider the UI for this feature in a separate issue (we need to design a UI needs encompasses all of the new features that we're adding).

[Marc-Andre-Rivet]

XLSX: One option that seems well supported https://github.com/SheetJS/js-xlsx -- also has a pro/commercial tier with unknown conditions. The community version is under Apache License; the implications of that might need to be checked out to be safe.

CSV: Many options exist Including SheetJS above.

As for other optional operations on the table, implement this in the most isolated manner possible to keep it well isolated from the table's implementation. Internally should be its own component.

The feature's implementation must be purely front-end and will involve some JS magic to initiate what looks like a file download -- this is a known problem / been solved may times.

The implementation should take into consideration the possibility of exporting the data as it is displayed in the table for future feature/improvement.

Lazy-loading would be great but is not needed for this feature to be implemented. Will increase the table's bundle size significantly though.

With a purely functional UI, this shouldn't take more than a few days to implement and test.

[Marc-Andre-Rivet]

Just dropping this here, the pro version of xlsx-js offers styling support
https://sheetjs.com/pro

No idea about licensing.

[Marc-Andre-Rivet]

@alinastarkov From our previous discussion, I'd say the CSV should export with the column id's and the XLSX should export with the column header names (multi-row / merged if required)

The idea being that if the user does a CSV export I would expect him to be somewhat aware of the data structure or planning on further processing the data in system x-y-z. Exporting headers would potentially cause the CSV to have multiple header rows which is not processed by default (in pandas, can do headers=[0,1] for rows 0 and 1 being used as headers) -- also, exporting the headers might result in duplicated names.

For XLSX, the user would arguably want to present or further process the data manually, not programmatically, and in this case the header names are more relevant.

@alexcjohnson, additional insight?