Editable Dataframe - feature requests and bugs

[karangattu]

Find below all the feature requests for future releases of editable dataframes

• Allow users to use Shift Key + multiple rows to highlight multiple rows when row_selection is configured to be multiple
• During edit mode, py-shiny should perform validation when user attempts to enter data that conflicts with the data type of the column in question.
• Allow users to move columns around for ease of use or better readability when working with dataframes
• Option for users to use pagination i.e. show 10, 25 or 50 entries in the table instead of infinite scroll if the data is being populated by calling an API.
• Allow users to undo or redo the changes made to a cell.
• Color in darkly mode is subtle and user might not recognize the highlighted cells.
• Have a way where users can move between different cells without editing each cell. In short when user clicks on a cell, do not turn on editing, only when user presses return/Enter key or double click on the cell.
• Hitting tab currently allows to shift focus to the next cell on the right in the same row. When it reaches end of line for that row, hitting Tab should allow to switch to the first cell in the next row.
• Allow copying across multiple cells by either dragging the selected cell or pasting when a selected row is in focus.

Find below all the issues found during the initial exploratory testing of editable dataframes

• When the user makes changes to a dataframe with mode=edit, scroll down the table till it is out of focus, then scroll back up. You will see all the edited cells highlighted with the green border animation.
  • Removed border animation in bug(data frame): Remove success animation, fix jumpy cells when editing, fix broken click on edited cell #1283; This removes the animation problem
• When a dataframe is sorted, hitting return/enter key will not move to the next row, but instead the unsorted order. (bug(data frame): Use sorted row data to find next row when hitting (shift+)enter #1291)
• Editing a cell in Firefox causes the border to bleed beyond the column area. (bug(data frame): Fix Firefox rendering of textarea in editing cell #1292)
  
  • top: 0;left: 0;width: 100%;height: 100%; fixes the issue in Firefox

[schloerke]

• Remove toggle rows and support only multi-native rows (bug(data frame): Replace "rows" (toggle) with "multi-native" selection #1286)
• Navigation like google sheets
  • Arrow keys selects cell
  • Tab is like arrow right; Shift+Tab is like arrow left
  • Enter is like arrow down; Shift+Enter is like arrow up
  • Any key but arrow or tab enters into edit mode
  • Double click enters into edit mode

[mdaeron]

May I suggest the following features?

• ability to paste multi-row, tab-delimited data (basically, enable straightforward copy-and-paste from Office-like spreadsheet software)
• automatically add rows when pasting such data

These two things would go a long way towards building apps that everyone can use, even without any coding background.

[nh-cadmus]

I would like to suggest the following additional functionality:

• Ability to define which columns should be editable (or which columns should be read-only)
• Different input controls for different data types, e.g. a boolean field should contain checkboxes, a timestamp field should contain a date picker, etc.
• Ability to define a custom dropdown list of available values for a column
• Data validation options - perhaps the ability to define a custom validation function for a column and/or a row, which gets called when data is added/updated and if it returns false shows an error message. Would allow for simple validation like a numeric range for number columns, and more complex custom validation for fields that might have special requirements.

[schloerke]

@nh-cadmus Data validation options - 🕺 Check out:

py-shiny/shiny/api-examples/data_frame_data_view/app-core.py

Lines 58 to 62 in 8787772

	@df_original.set_patch_fn
	def _(*, patch: render.CellPatch) -> render.CellValue:
	if patch["column_index"] in [0, 2]:
	return float(patch["value"])
	return int(patch["value"])

. If you need full control:

py-shiny/shiny/api-examples/data_frame_set_patches/app-core.py

Lines 50 to 66 in 8787772

	@my_data_frame.set_patches_fn
	def _(*, patches: list[render.CellPatch]) -> list[render.CellPatch]:
	for patch in patches:
	if patch["column_index"] in [0, 2]:
	patch["value"] = float(patch["value"])
	else:
	patch["value"] = int(patch["value"])
	# "Save to the database" by writing the edited data to a CSV file
	df = edited_df().copy()
	for patch in patches:
	df.iloc[patch["row_index"], patch["column_index"]] = patch["value"]
	edited_df.set(df)
	df.to_csv(here / "mtcars.csv", index=False)
	print("Saved the edited data to './mtcars.csv'")
	return patches

. Feel free to raise errors within either function.

[davidchall]

Feature requests to help end users create their own table, rather than editing an existing table:

• Ability to add row
• Ability to delete row(s)
• Ability to add column
• Ability to delete column(s)