Closes #131 and closes #142 and closes #161 and closes #162


Co-authored-by: Dawid Kałędkowski <dawid.kaledkowski@gmail.com>

This reverts commit 73c5ce9.

…@main

Merging `main` branch into `filter_panel_refactor`  following #189.

---------

Co-authored-by: chlebowa <chlebowa@users.noreply.github.com>
Co-authored-by: Dawid Kałędkowski <dawid.kaledkowski@gmail.com>

Allows filter state inputs to collapse and expand, defaulting to
collapsed. Only 1 filter state per data set can be open at a time.

- For a given data set, filter information is in an accordion of
collapsible cards
- Only 1 filter state (per data set) can be open at a time.
- Card headers show a summary of the filter state.

Logical and choices filter state UIs don't display correctly, at least
on Bootstrap 4.


Closes #129.

---------

Co-authored-by: Dawid Kałędkowski <dawid.kaledkowski@gmail.com>

# Pull Request

Fixes
#[183](#183)

---------

Signed-off-by: Marek Blazewicz <110387997+BLAZEWIM@users.noreply.github.com>
Co-authored-by: Mahmoud Hallal <86970066+mhallal1@users.noreply.github.com>
Co-authored-by: Dawid Kałędkowski <6959016+gogonzo@users.noreply.github.com>

# Pull Request

Fixes
#[158](#158)


1. At initialization, when values are not valid, corrections will be
performed with warnings.
2. Later, when changed by the user, shinyvalidate is used.
 
Example to test:

```
adsl <- synthetic_cdisc_dataset("latest", "adsl")

app <- init(
  data = teal_data(dataset("ADSL", adsl)),
  modules = example_module(),
  filter = list(
    "ADSL" = list(
      RANDDT = list(selected = c("2021-02-16", "2021-02-17")),
      TRT01SDTM = list(selected = c("2021-02-11 17:09:18", "2021-02-10 20:42:27"))
    )
  )
)
runApp(app)
```

---------

Signed-off-by: Marek Blazewicz <110387997+BLAZEWIM@users.noreply.github.com>
Co-authored-by: Dawid Kałędkowski <6959016+gogonzo@users.noreply.github.com>

closes #173 #174 #167 

Making reactive filter counts triggered only if the filter is visible

Co-authored-by: Andrew Bates <andrew.bates@atorusresearch.com>

# Pull Request

Fixes
#[140](#140)

---------

Signed-off-by: Marek Blazewicz <110387997+BLAZEWIM@users.noreply.github.com>
Co-authored-by: Blazewicz <blazewim@emea.roche.com>
Co-authored-by: Dawid Kałędkowski <6959016+gogonzo@users.noreply.github.com>

Closes #166 
 implements filtered counts and prepares for the hierarchical filtering

Co-authored-by: Andrew Bates <andrew.bates@atorusresearch.com>

remove CDISCFilteredData and methods. Tests for FilteredData

fixes @asbates comment
fix lintr

Expands the clickable area of filter cards so they expand/collapse when
clicking anywhere in the card header. (except remove icon)

Closes #203

This reverts commit bd5f0d7.

Moves the filter panel disable/enable switch into the active variables
section.

Closes #219

---------

Co-authored-by: Nikolas Burkoff <nikolas.burkoff@roche.com>

This PR contains work towards a new filter panel API as outlined in
#187.

**TESTING class methods**:

1. preparations
```
# get FilteredData object on demand
utils::data(miniACC, package = "MultiAssayExperiment")
get_fd <- function() {
  init_filtered_data(
    x = list(
      iris = list(dataset = iris),
      mtcars = list(dataset = mtcars),
      mae = list(dataset = miniACC)
    )
  )
}


# specify filter states (old way)
fss <- list(
  iris = list(
    "Species" = list(selected = "setosa"),
    "Sepal.Length" = list(selected = c(5, 6))
  ),
  mtcars = list(
    "disp" = list(selected = c(0, 280)),
    "cyl" = list(selected = 6)
  ),
  mae = list(
    subjects = list(
      years_to_birth = list(selected = c(30, 50), keep_na = TRUE, keep_inf = FALSE),
      vital_status = list(selected = "1", keep_na = FALSE),
      gender = list(selected = "female", keep_na = TRUE)
    ),
    RPPAArray = list(
      subset = list(
        ARRAY_TYPE = list(selected = "", keep_na = TRUE)
      )
    )
  )
)

# specify filter states (new way)
tss <- filter_settings(
  filter_var("iris", "Species", selected = "setosa"),
  filter_var("iris", "Sepal.Length", selected = c(5, 6)),
  filter_var("mtcars", "disp", selected = c(0, 280)),
  filter_var("mtcars", "cyl", selected = 6),
  filter_var("mae", "years_to_birth", selected = c(30, 50), keep_na = TRUE, keep_inf = FALSE, datalabel = "subjects", target = "y"),
  filter_var("mae", "vital_status", selected = "1", keep_na = FALSE, datalabel = "subjects", target = "y"),
  filter_var("mae", "gender", selected = "female", keep_na = TRUE, datalabel = "subjects", target = "y"),
  filter_var("mae", "ARRAY_TYPE", selected = "", keep_na = TRUE, datalabel = "RPPAArray", target = "subset")
)
```

2. convert old states specification to new one 
```
fss_new <- as.teal_slices(fss)
identical(fss, tss)
identical(fss_new, tss)
```
:point_up: this happens in `FilteredData$set_filter_state` with a
warning
Note `as.teal_slices` does not perform any validation, so a list-like
filter state that specifies filters on columns of `MAE@colData` that is
not wrapped as `list(MAE = list(...))` but is only `list(var =
list(...))` will be interpreted as a `data.frame` filter.

3. set states as list
```
# create FilteredData
fd <- get_fd()
# apply filter states
fd$set_filter_state(fss)
# see calls
fd$get_call("iris") %>% isolate
fd$get_call("mtcars") %>% isolate
fd$get_call("mae") %>% isolate
#recover filter states
fd$get_filter_state() %>% isolate
```

4. set states as `teal_slices`
```
# create FilteredData
fd <- get_fd()
# apply filter states
fd$set_filter_state(tss)
# see calls
fd$get_call("iris") %>% isolate
fd$get_call("mtcars") %>% isolate
fd$get_call("mae") %>% isolate
#recover filter states
fd$get_filter_state() %>% isolate
```
Note that calls are not generated. This is these filters are
instantiated and constructors don't know how to handle choices yet, so
by default they are created with everything selected, hence no calls.

5. modify states as `teal_slices`
```
fd$set_filter_state(tss)
# see calls
fd$get_call("iris") %>% isolate
fd$get_call("mtcars") %>% isolate
fd$get_call("mae") %>% isolate
#recover filter states
fd$get_filter_state() %>% isolate
```


**TESTING wrapper functions**:
```
datasets <- init_filtered_data(
  x = list(
    iris = list(dataset = iris),
    mae = list(dataset = miniACC)
  )
)
fs <- filter_settings(
  filter_var("iris", "Species", selected = c("setosa", "versicolor")),
  filter_var("iris", "Sepal.Length", selected = c(5.1, 6.4)),
  filter_var("mae", "years_to_birth", selected = c(30, 50),
             keep_na = TRUE, keep_inf = FALSE, datalabel = "subjects", target = "y"),
  filter_var("mae", "vital_status", selected = "1", keep_na = FALSE, datalabel = "subjects", target = "y"),
  filter_var("mae", "gender", selected = "female", keep_na = TRUE, datalabel = "subjects", target = "y"),
  filter_var("mae", "ARRAY_TYPE", selected = "", keep_na = TRUE, datalabel = "RPPAArray", target = "subset")
)

# set initial filter state
set_filter_state(datasets, filter = fs)

fd$.__enclos_env__$private$get_filter_count() %>% isolate

# get filter state
get_filter_state(datasets)

fd$.__enclos_env__$private$get_filter_count() %>% isolate

# modify filter state
set_filter_state(
  datasets,
  filter_settings(
    filter_var("iris", "Species", selected = "setosa", keep_na = TRUE)
  )
)

fd$.__enclos_env__$private$get_filter_count() %>% isolate

# remove specific filters
remove_filter_state(
  datasets,
  filter_settings(
    filter_var("iris", "Species"),
    filter_var("mae", "years_to_birth"),
    filter_var("mae", "vital_status")
  )
)

fd$.__enclos_env__$private$get_filter_count() %>% isolate

# remove all states
clear_filter_states(datasets)

fd$.__enclos_env__$private$get_filter_count() %>% isolate

```

---------

Signed-off-by: Marek Blazewicz <110387997+BLAZEWIM@users.noreply.github.com>
Co-authored-by: Marek Blazewicz <110387997+BLAZEWIM@users.noreply.github.com>
Co-authored-by: Blazewicz <blazewim@emea.roche.com>
Co-authored-by: Dawid Kałędkowski <6959016+gogonzo@users.noreply.github.com>

Signed-off-by: Aleksander Chlebowski <114988527+chlebowa@users.noreply.github.com>

…@main

Co-authored-by: Dawid Kałędkowski <dawid.kaledkowski@gmail.com>

Fixes a bug in the count labels for `ChoicesFilterState`. Uses the x
with any unused levels removed to set choices counts.

Fixes #260

filterable through appi
- removed set_filteravle_varnames. Setting filterable possible only through teal_slices
- moved get_fillter_state to abstract class

Updates the summary row of the card header.

- Removes "Include" for NA/Inf
- Move NA/Inf contents further to the right of the card summary.
- NA/Inf displays "NA/Inf" with a check or X. Check if yes, X if no. If
no NA or Inf are present in the column, nothing is displayed
- For value/selected portion of summary, the text is automatically
truncated depending on screen width.
- For categorical variables, the selected/value display is as follows
- if the total length of "level 1, level 2, level 3" is more than 40
characters, the text "N levels selected" is used.
  - otherwise, the selected levels are displayed
- 40 characters is a magic number and is an approximation. I found it to
be ~300px which is the total text limit. But different font sizes,
capitalization vs. not, browser zoom, etc. will all affect the actual
number of characters that = Xpx. So there's never going to be a "right"
value for this
- For numeric and dates, an endash is used instead of "-" to separate
the selected values
- For numeric variables, if there are > 4 significant digits the number
is formatted in scientific notation with 4 significant digits. This is
related to #251. If the current implementation is not exactly what we
want, it should affect this PR. We can refine the method with #251.
- The font size of some peripherals of input elements were adjusted to
match the font size in the card header.

Fixes #220

related to #244 
closes #141 

This version provides a richer UI for a fixed filter state card compared
to #244

FilterState has a private$fixed field that holds a logical flag. In this PR:

* a fixed FilterState has a lock icon added to the header part of the UI
* a fixed FilterState displays visualizations of the selection state as usual, but without input widgets
* FilterState$disable/enable are rewritten to accommodate for the need to disable/enable a fixed filter state

---------

Signed-off-by: Aleksander Chlebowski <114988527+chlebowa@users.noreply.github.com>
Co-authored-by: Andrew Bates <andrew.bates@atorusresearch.com>

…@main

Closes #206.

Changes the bahavior of `FilteredData$filter_panel_disable/enable`.
Filter states are no longer removed and re-instantiated. Rather, their
state is changed to disabled/enabled by sending a filter panel API
command down the class structure.

* the global disable/enable method is modified appropriately
* caching individual filter states is removed; caching states at the
global level is necessary as some states should be restored disabled
* an observer is added to `FilterState$server` to update the disable
switch accordingly

Fixes #271

Adds a numeric range input to range filter state cards so range can be
set manually. This is a redo of #226


Fixes #133