Finally implement row review #146

jthompson-arcus · 2024-11-21T20:46:16Z

Closes #99

LDSamson · 2024-11-22T12:08:04Z

@jthompson-arcus Maybe good to discuss the conceptual UI implementation together.

jthompson-arcus · 2024-11-22T13:02:53Z

@jthompson-arcus Maybe good to discuss the conceptual UI implementation together.

I am happy for all input on the UI implementation. There are several ways it could go and no clear best option.

Merge branch 'jt-99-review_by_row' into jt-99-review_by_row_for_reals # Conflicts: # R/mod_review_forms.R

LDSamson · 2024-12-23T13:17:16Z

R/mod_study_forms.R

+    observe({
+      reload_data(reload_data() + 1)
+      session$userData$update_checkboxes[[form]] <- NULL
+      session$userData$review_records[[form]] <- data.frame(id = integer(), reviewed = character())
+    }) |> 
+      bindEvent(r$subject_id, r$review_data,
+                ignoreInit = TRUE)


This feels a bit duplicated with the observeEvent in mod_review_forms and mod_common_forms. Placing a cat_dev in these three events, and updating subject_id fires multiple updates:

Yes, each form is it's own module. I would expect it to fire for every module whenever the subject is changed or the review data is updated. I think this can be moved outside of those modules though and be contained in the review forms module.

On further inspection, I don't think we really want to change any of this observer. There doesn't appear to be any advantage to further abstracting this logic. It's important to ensure the datatable's server-side object is getting reloaded whenever the subject is changed of the review data is updated.

LDSamson · 2024-12-23T13:23:23Z

R/mod_study_forms.R

+    observeEvent(input$table_review_selection, {
+      # Update review values for session's user data
+      session$userData$update_checkboxes[[form]] <- NULL
+      session$userData$review_records[[form]] <-


So, review_records are the records as per selection status in the checkboxes?

The object r$review_data was originally designed to pass through review status of each item. I am still struggling to see why we need so many review objects and can't simplify this. Can't o_reviewed be part of the review_data object instead?

In essence it is still being used in that way because that's where the field reviewed is coming from. The new field o_reviewed is necessary for the datatable object. Each datatable is pivoted, so the datatable itself would contain multiple records, this needs to be consolidated in some way for a row level review to even be possible.

LDSamson · 2024-12-23T13:32:26Z

R/mod_common_forms.R

+    }) |> 
+      bindEvent(r$subject_id, r$review_data,
+                ignoreInit = TRUE)
+


Just wondering: do you have a specific reason why you sometimes use observe/bindevent and sometimes observeEvent?

I prefer using bindEvent() when there are multiple triggers. In my opinion it is more readable.

LDSamson · 2024-12-23T13:47:43Z

R/mod_common_forms.R

+          dplyr::rows_upsert(
+            session$userData$review_records[[form]],
+            input[[review_selection]][c("id", "reviewed")],
+            by = "id"
+          ) |> 
+          dplyr::filter(!is.na(reviewed)) |> 
+          # Ensure that only the current subject is being reviewed
+          dplyr::semi_join(
+            subset(r$review_data, subject_id == r$subject_id & item_group == form),
+            by = "id"
          ) |> 
-          dplyr::select(-dplyr::starts_with("SAE"))
+          # Only update records where the review status is being changed
+          dplyr::anti_join(
+            subset(r$review_data, subject_id == r$subject_id & item_group == form),
+            by = c("id", "reviewed")
+          ) |> 
+          dplyr::arrange(id)


the upsert/filter/semi_join and anti_join are a bit hard to follow for me. Is the semi_join really needed for example? Can't you just add subject_id and form to the filter directly?

Also, I think this part could be a candidate to extract in a helper function since it is duplicated in mod_review_forms and is more or less 'business logic'

I will take a look.

I cannot add subject_id and item_group to the filter directly because input[[review_selection]] does not contain those values. It only contains the id's. This was intentional redundancy. Even though session$userData$review_records[[form]] gets reset whenever r$subject_id is changed, this ensures that no unintended records are being assessed. The overhead is very little because this is only triggered when a checkbox in a datatable is clicked.

clinsight/R/fct_form_helpers.R

Lines 1 to 39 in f5fa833

#' Update Review Records

#'

#' Updates the review records data frame when a datatable checkbox is clicked.

#'

#' @param review_records The review records data frame to update.

#' @param review_selection The review selection data frame input from the

#' datatable.

#' @param active_data The active review data frame.

#'

#' @return A data frame containing the updated records data.

#'

#' @details Three main steps are performed: UPSERT, SUBSET, and ANTI-JOIN The

#' UPSERT takes the review selection data frame and upserts it into the review

#' records data frame. (An upsert will insert a record if the unique

#' identifier is not yet present and update a record based on the unique

#' identifier if it already exists.) The SUBSET step removes an empty reviews

#' (partially review rows) and any records not part of the active review (as a

#' precautionary measure). The ANTI-JOIN step removes any records that match

#' the active review (records that will not be changing review status based on

#' user inputs).

#'

#' @noRd

update_review_records <- function(review_records, review_selection, active_data) {

if (is.null(review_records))

review_records <- data.frame(id = integer(), reviewed = character())

review_records |>

dplyr::rows_upsert(

review_selection,

by = "id"

) |>

# Remove empty reviews and inactive data IDs

subset(!is.na(reviewed) | !id %in% active_data$id) |>

# Only update records where the review status is being changed

dplyr::anti_join(

active_data,

by = c("id", "reviewed")

) |>

dplyr::arrange(id)

}

LDSamson

I added a view comments. I will have to stop now, and and probably can only continue in the new year. In general, I like the way it is implemented, but I am just wondering if we can make it a bit simpler and easier to follow. It is a bit hard to follow what happens when exactly, in what order.

Have a nice Christmas and new year!

jthompson-arcus · 2024-12-31T15:29:30Z

With all the requests for helper functions and such, I decided it was best to move all this logic to it's own module. Now all changes can be made in one location and the differences between the forms are essentially eliminated. The module itself still needs to be documented.

The PR grows and grows...

jthompson-arcus added 2 commits November 21, 2024 15:44

Add very basic checkbox column to tables

2d0d704

Merge branch 'jt-99-review_by_row' into jt-99-review_by_row_for_reals

d436332

jthompson-arcus added 13 commits November 22, 2024 09:10

Merge branch 'jt-99-review_by_row' into jt-99-review_by_row_for_reals

f56c780

"Highlight" rows when check status changed

67d817f

Grab id's for o_reviewed

7939b3c

Add input handler

e3bb3ed

Capture row index info

6e258a2

Standardize callbacks and renders

a42049b

Add missing observers

7ba17d9

Add study form observers

f0dfcc3

Add anti_join()s back

9f0e4de

First pass at integration in review module

b703bec

Fix couple of issues

5fd5747

Remove row selection

5956269

Fix bug on study forms

d9f49e8

jthompson-arcus mentioned this pull request Dec 2, 2024

Review by records instead of subject_id & form_id "rv_row" #137

Merged

jthompson-arcus added 6 commits December 2, 2024 09:07

Resolve merge conflict with jt-99-review_by_row

e2b9953

Merge branch 'jt-99-review_by_row' into jt-99-review_by_row_for_reals # Conflicts: # R/mod_review_forms.R

Integrate "form_reviewed" checkbox

2493779

Update tables from overall checkbox

0cb8710

Update app-feature-1 snapshots

2894351

Update app-feature-3-002.json

b27d95d

Update app_feature_04.md

36f843a

jthompson-arcus changed the base branch from dev to jt-99-review_by_row-devex December 2, 2024 19:11

jthompson-arcus changed the base branch from jt-99-review_by_row-devex to dev December 2, 2024 19:12

jthompson-arcus added 4 commits December 2, 2024 14:14

Create custom.js

e98afa0

Fix warnings and errors in mod_common_forms test

a353f5d

Repair part of mod_review_form tests

579d80e

Properly handle partially reviewed rows

a7fda21

LDSamson reviewed Dec 23, 2024

View reviewed changes

jthompson-arcus added 8 commits December 30, 2024 16:05

Fix bug on initialization

83642d8

Separate o_reviewed logic into helper function

ee4fdad

Create helper function for review selection UPSERT

fa9ded8

Document o_reviewed field

aa05fbb

Resolve documentation error

5176d1f

Add helper function to update reactive table from user selection

2feb93a

Move logic to module

103d282

Add table title back

fa06a00

jthompson-arcus added 15 commits December 31, 2024 10:57

Update documentation

f5fa833

Delete update_tbl_data_from_datatable.Rd

8a200f4

Update app feature 01 test JSONs and PNGs

dec6f34

Update app-feature-3-002.json

db1b0ce

Update app_feature_04.md

53eac61

Initialize disabled element on data creation

dbdfc20

Update app_feature_04.md

7c61998

Update test-mod_common_forms.R

b1b514a

Update mod_navigate_review JSONs

9481c0e

Repair part of study forms tests

0a96d60

Add back table check for common forms

add5001

Finish repairing study forms tests

1e94521

Update mod_report-001.json

10fe949

Update mod_review_form_tbl_server() documentation

6e03b44

Update global.R

3d14c53

jthompson-arcus marked this pull request as ready for review December 31, 2024 18:28

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Finally implement row review #146

Finally implement row review #146

jthompson-arcus commented Nov 21, 2024

LDSamson commented Nov 22, 2024

jthompson-arcus commented Nov 22, 2024

LDSamson Dec 23, 2024

jthompson-arcus Dec 30, 2024

jthompson-arcus Dec 30, 2024

LDSamson Dec 23, 2024

LDSamson Dec 23, 2024

jthompson-arcus Dec 30, 2024

LDSamson Dec 23, 2024

jthompson-arcus Dec 30, 2024

LDSamson Dec 23, 2024

jthompson-arcus Dec 30, 2024

jthompson-arcus Dec 30, 2024

jthompson-arcus Dec 31, 2024

LDSamson left a comment

jthompson-arcus commented Dec 31, 2024

	#' Update Review Records
	#'
	#' Updates the review records data frame when a datatable checkbox is clicked.
	#'
	#' @param review_records The review records data frame to update.
	#' @param review_selection The review selection data frame input from the
	#' datatable.
	#' @param active_data The active review data frame.
	#'
	#' @return A data frame containing the updated records data.
	#'
	#' @details Three main steps are performed: UPSERT, SUBSET, and ANTI-JOIN The
	#' UPSERT takes the review selection data frame and upserts it into the review
	#' records data frame. (An upsert will insert a record if the unique
	#' identifier is not yet present and update a record based on the unique
	#' identifier if it already exists.) The SUBSET step removes an empty reviews
	#' (partially review rows) and any records not part of the active review (as a
	#' precautionary measure). The ANTI-JOIN step removes any records that match
	#' the active review (records that will not be changing review status based on
	#' user inputs).
	#'
	#' @noRd
	update_review_records <- function(review_records, review_selection, active_data) {
	if (is.null(review_records))
	review_records <- data.frame(id = integer(), reviewed = character())
	review_records \|>
	dplyr::rows_upsert(
	review_selection,
	by = "id"
	) \|>
	# Remove empty reviews and inactive data IDs
	subset(!is.na(reviewed) \| !id %in% active_data$id) \|>
	# Only update records where the review status is being changed
	dplyr::anti_join(
	active_data,
	by = c("id", "reviewed")
	) \|>
	dplyr::arrange(id)
	}

Finally implement row review #146

Are you sure you want to change the base?

Finally implement row review #146

Conversation

jthompson-arcus commented Nov 21, 2024

LDSamson commented Nov 22, 2024

jthompson-arcus commented Nov 22, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

LDSamson left a comment

Choose a reason for hiding this comment

jthompson-arcus commented Dec 31, 2024