[Datatable] missing percentage column and wrong headers on export formatted csv

[stratoula]

Summary

Fixes #49149. Percentage column is missing in formatted csv.
Fixes #57096. Wrong formatted CSV when splitting tables.

Have tested it on Chrome, Safari and FF. Both raw and formatted CSVs are exported as expected.
Moreover the order of the header columns was different from the rows so the data were not displayed correctly in some cases (under the corresponding header).

Preview of formatted csv with percentages:

Preview of formatted csv when splitting tables:

[elasticmachine]

Pinging @elastic/kibana-app (Team:KibanaApp)

[stratoula]

@elasticmachine merge upstream

[stratoula]

@elasticmachine merge upstream

[kertal]

Quick question while testing, could you take a look at the split table output, there seems to be a difference to current master, since columns 1 isn't displayed in master, I think it should only be part of the csv export?

Master:

PR (Note the redundant first column):

thx!

[stratoula]

I will! Thanx @kertal I wasn't sure about it tbh ;)

[kertal]

Have a question to understand this part. The change in this implementation is, if there's no column available for a rows's property, don't add it to the column. This sounds valid, and an improvement. Just try to understand, how could that be the case? Are there row properties with no columns? Just checking to prevent some kind of regression here. Because then there is a nice potential to simplify the code here in a way like this

for (const col of columns) {
   const value = row[col.id];
   const formattedValue = formatted && col.formatter
      ? escape(col.formatter.convert(v))
      : escape(v);
   rowArray.push(formattedValue);
}

[stratoula]

Yes there is a case where the user has selected two buckets and has ticked the Show partial rows. On the table you can't see any change but when the CSV is downloaded an extra column is added but as there is no extra header, it is messed up. So the question here is which is the expected behavior? As far as I can't see any change on the table, I expect to see the same on the CSV, right? If not this is another bug that has to do with the partial rows

[kertal]

synced with @stratoula, and she showed me another issue in the current implementation, fixed by this PR, so here the example described in the recent comment,

Exporting this table led to an formatted CSV export with an extra column

"extension.keyword: Descending","geo.dest: Descending",Count
,624,122,CN
,624,96,IN
,624,51,US
,624,23,ID
,624,22,PK
gz,285,63,CN
gz,285,52,IN
gz,285,16,US
gz,285,9,NG
gz,285,7,ID

This is also fixed by this PR, and therefore it's safe to refactor the existing code in the suggested way

[stratoula]

@elasticmachine merge upstream

[kibanamachine]

💚 Build Succeeded

• continuous-integration/kibana-ci/pull-request
• Commit: 4297c42

History

• 💔 Build #49770 failed b9b0de8
• 💚 Build #49761 succeeded ba9c1a8318e610d7992b16f77a3433bc8a09def9
• 💚 Build #49617 succeeded ac99273
• 💚 Build #49522 succeeded ac428ef
• 💚 Build #49133 succeeded 3312df7105b60c0f06130ff33bac313ddac06afb

To update your PR or re-run it, just comment with:
@elasticmachine merge upstream

[kertal]

Code LGTM. 2 + 1 issues solved by a single PR 👏 . Tested locally with Safari, Firefox, Chrome. Works!