{editbl}: DT extension for CRUD

editbl ('edit tibble') allows you to modify tables in a spreadsheet-like fashion. Not just in-memory data.frame objects, but also data living in a database.

Installation

From CRAN:

install.packages('editbl')

Latest development version:

remotes::install_github("https://github.com/openanalytics/editbl", ref = "main", subdir = "editbl")

Get started

Choose a dataset of your liking and use eDT to interactively explore and modify it!

modifiedData <- editbl::eDT(mtcars)
print(modifiedData)

Run some demo apps

editbl::runDemoApp()

More introductory examples can be found here. Advanced examples can be found in the vignettes.

Features

Overview of main features

Supporting multiple backends and in-place editing
Customizable (lightweight DT wrapper)
Easy integration in shiny apps
Undo/redo button
Copy rows
Drag cells
No need to have all data in-memory
Tackles challenges such as enforcing foreign keys and hiding of surrogate keys
Transactional commits (currently for tbl_dbi class and non in-place editing)
Default values for new rows (UUID's, 'current date', 'inserted by', ...)
Possible to set row level security

Constraints and normalized tables

Sometimes you want to restrict certain columns of your table to only contain specific values. Many of these restrictions would be implemented at database level through the use of foreign keys to other tables.

editbl allows you to specify similar rules through the use of foreignTbls as an argument to eDT(). Note that you can additionally hide surrogate keys by the use of naturalKey and columnDefs if you wish to.

a <- tibble::tibble(
    first_name = c("Albert","Donald","Mickey"),
    last_name_id = c(1,2,2)
  )

b <-  foreignTbl(
  a,
  tibble::tibble(
      last_name = c("Einstein", "Duck", "Mouse"),
      last_name_id = c(1,2,3)
    ),
  by = "last_name_id",
  naturalKey = "last_name"
)

eDT(a,
  foreignTbls = list(b),
  options = list(columnDefs = list(list(visible=FALSE, targets="last_name_id")))
)

Support for different backends

dplyr code is used for all needed data manipulations and it is recommended to pass on your data as a tbl. This allows editbl to support multiple backends through the usage of other packages like dtplyr, dbplyr etc.

In case you pass on other tabular objects like data.frame or data.table the function will internally automatically cast back and forth to tbl. Small side effects may occur because of this (like loosing rownames), so it might be better to cast yourself to tbl explicitly first.

# tibble support
modifiedData <- editbl::eDT(tibble::as_tibble(mtcars))

# data.frame support
modifiedData <- editbl::eDT(mtcars)

# data.table support
modifiedData <- editbl::eDT(data.table::data.table(mtcars))

# database support
tmpFile <- tempfile(fileext = ".sqlite")
file.copy(system.file("extdata", "chinook.sqlite", package = 'editbl'), tmpFile)
conn <- editbl::connectDB(dbname = tmpFile)
modifiedData <- editbl::eDT(dplyr::tbl(conn, "Artist"), in_place = TRUE)
DBI::dbDisconnect(conn)
unlink(tmpFile)

Note that there are some custom methods in the package itself for rows_update / rows_delete / rows_insert. The goal would be to fully rely on dplyr once these functions are not experimental anymore and support all needed requirements. These functions also explain the high amount of 'suggested' packages, while the core functionality of editbl has few dependencies.

Switching from DT

Let's say you already use DT::datatable() to display your data, but want to switch to editbl::eDT() to be able to edit it. Would this be a lot of effort? No! In fact, eDT() accepts the exact same arguments. So it is almost as easy as replacing the functions and you are done. Should you run into problems take a look here for some pointers to look out for.

Notes

https://github.com/tidyverse/dtplyr/issues/260 might cause errors / warnings when using eDT with dtplyr. If possible convert to normal tibble first.
editbl assumes that all rows in your table are unique. This assumption is the key (ba dum tss) to allow for only having the data partially in memory.
editbl does not attempt to detect/give notifications on concurrent updates by other users to the same data, nor does it 'lock' the rows you are updating.

It just sends its updates to the backend by matching on the keys of a row. If other users have in the meantime made conflicting adjustments, the changes you made might not be executed correctly or errors might be thrown.

General future goals for this package

Full dplyr compatibility so support for different backends is easily facilitated. Now there are 2 methods (e_rows_update, e_rows_insert) that need to be implemented to support a new backend.
Full DT compatibility, including all extensions.
Better editing / display options for time values. E.g. control over timezone and format of display / storage + nicer input forms.
Any addition that supports the concept of editing data as flexible/easy as possible while respecting backend schema's and constraints.