Ods Reader Style Support (Last) - Column Styles (Partial)

[oleibman]

Continuing the work of PR #4810 and PR #4813 and PR #4814. According to the spec for LibreOffice, the xml for both columns and rows allow a default-cell-style-name attribute. However, my testing indicates that using that attribute with a row doesn't result in the results I would expect; furthermore, LibreOffice doesn't seem to use that attribute in association with a row (it does use it for columns). For rows, it fills the unpopulated rows in the cell with a repeated table cell with the appropriate style. These filler cells have been a source of problems for us, and I'm not willing to add them at this time (maybe in future). So, if a style is applied to a row in PhpSpreadsheet, it will be reflected in all cells in that row until highestDataColumn, but no further.

Since LibreOffice does support adding a style to a column both in the UI and the Xml, it should be a little easier. But its support is imcomplete as well - when reading a spreadsheet, the UI will apply the style only to the rows that are defined in the xml. Again, this requires filler rows for the rows beyond highestDataRow, and, again, I am not willing to add those. However, it is important to process the column style when reading because, unlike the case with rows, the cells which are defined in the column may omit style data altogether if the style matches the column style.

There has till now been no support for ReadDataOnly (default false) and ReadEmptyCells (default true) for Ods Reader. These are now added. ReadEmptyCells is especially important and it would be sensible to usually set it to false. The reason is that LibreOffice usually supplies a width for all the columns mentioned in its row fillers, and PhpSpreadsheet needs to create ColumnDimensions for each of those as a result, resulting in much larger xml than one would expect when the spreadsheet is saved. This arose when I was testing this change - it has always been the case, but nobody has raised an issue about it. We avoid this problem by ignoring the extra columns for ReadEmptyCells false. But this requires a little redundancy in the code. We need to define styles for an entire column before we define styles for the cells in the column, but we don't need to define column widths beforehand. So, if ReadEmptyCells is false, we need to process the columns for styles, then process the cell data (so that we know what highDataColumn is), then process the columns again but this time for widths (ignoring columns beyond highDataColumn). If ReadEmptyCells is true, we can process column styles and widths in a single pass.

This is:

• a bugfix
• a new feature
• refactoring
• additional unit tests

Checklist:

• Changes are covered by unit tests
  • Changes are covered by existing unit tests
  • New unit tests have been added
• Code style is respected
• Commit message explains why the change is made (see https://github.com/erlang/otp/wiki/Writing-good-commit-messages)
• CHANGELOG.md contains a short summary of the change and a link to the pull request if applicable
• Documentation is updated as necessary