Formatted cells in `ztable` lose text formatting.

[Alex-Muirhead]

When applying formatting to cells (i.e. bold weight) using a show rule, this formatting isn't passed through for cells that are formatted within ztable. In the example below, the last row is meant to have a bold weight.

For other modules I've written, I found this snippet to be quite useful to taking an existing cell and applying functions to the body and/or applying parameters to the cell.

#import "@preview/t4t:0.3.2" : is, is-none

#let modify-cell(it, func: none, ..style) = {
  // Default is identity mapping
  if is-none(func) { func = { body => body } }

  if is.elem(table.cell, it) {

    // Reconstruct the modified cell
    let attrs = it.fields()
    let body = attrs.remove("body")
    // Default to individual cells overriding
    table.cell(func(body), ..style, ..attrs)

  } else if is.elem(text, it) or is.elem("sequence", it) {

    // Construct a new cell for basic text
    table.cell(func(it), ..style)

  } else {

    // Let everything else (i.e. hline & vline) pass through
    it

  }
}

[Mc-Zen]

Hmm, I will need to investigate that. Thank you for reporting.

For other modules I've written, I found this snippet to be quite useful to taking an existing cell and applying functions to the body and/or applying parameters to the cell.

Can you explain how you use this snippet? Is it for preprocessing all ..children before passing them on to std.table? Because, the way zero works is different:

• all children (table cells) are passed straight to std.table
• they are then modified depending on content and column just through a show rule on table.cell. This is important because only then is the (row and mainly the) column known for all cells.

[Mc-Zen]

Okay, so I have tracked the issue and the good news is that there is a solution.

I will explain it here for the posterity but you don't need to read it ^^

At the heart of ztable lies something like

#show table.cell: it => {
  if is-not-numeral(it..) { it }
  else { transform-into-aligned-number(it, ..) }
}    
#table(..children)

This means, that

• the cell remains unchanged if it does not contain a number or is not in a column for which number alignment has been activated,
• and otherwise the cell is replaced with something else. Importantly, in this case, it is not a table.cell that is returned which is perfectly legal as I make sure that it behaves just like the cell that was there before. The reason is that otherwise I would get recursive show rules.

Since in the second case, there is no table.cell, it cannot be styled with set and show rules.

Now interestingly, with the current design of the package I can somehow wrap the content into a table.cell without getting a recursive show rule after all.... not exactly sure why.

Edit: It seems that returning a cell with a body that is a context expression is the key here.

But it works and now things like

#show table.cell.where(y: 4): strong

behave as they should.

[Alex-Muirhead]

Fantastic, thank you for the fix and the explanation! In regards to the snippet, yes it's used to preprocessing ..children or a subset thereof. I can see now how zero relies on a bit more context.

[Alex-Muirhead]

Interestingly this doesn't work with the 0.12 release that's on the online editor. I'll give it another go offline when I get a change, with the master version.

Edit: On closer inspection, it doesn't work since call-num returns a context, which no formatting can be applied to. I think the context keywords might need to be moved higher up the call-chain, but I might be missing another solution.

[Mc-Zen]

Interestingly this doesn't work with the 0.12 release that's on the online editor.

Can you specify what "this" is which does not work? I did not understand 😅

[Mc-Zen]

Taking a guess:

It seems that making the numbers bold does not work in the web app. But this is a different issue, not connected to zero at all.

Using strong with equations just does nothing

#strong[$12a^2+b$]

I wonder why since it works with tinymist ....

Try #show table.cell.where(y: 1): text.with(blue), this works fine.

[Alex-Muirhead]

Ah, whoops, yes that's the cause! I was also using a label to add formatting to table headers and footers, which was being lost, but that feels like more of a usage issue.