[Bug]: Reporter fails with teal as a Shiny module

[mschaffer-incyte]

What happened?

Using the latest development version of teal, I'm running into issues trying to use teal as a Shiny module (ref: https://github.com/insightsengineering/teal/blob/main/vignettes/teal-as-a-shiny-module.Rmd). In the code below, a wrapper is created to dynamically use predefined teal data sets and teal modules based on a selected study name.

Code

library(shiny)
library(bslib)
library(teal)
library(teal.data)
library(teal.widgets)
library(teal.reporter)

options(shiny.legacy.datatable = TRUE)


# Define selectable studies with datasets and study-specific modules
studies <- c("Study1", "Study2")

datasets <- list(
  "Study1" = within(teal.data::teal_data(), { iris <- iris; mtcars <- mtcars}),
  "Study2" = within(teal.data::teal_data(), { df <- data.frame(a = 1:10, b = letters[1:10]) } )
)

mods <- list(
  "Study1" = teal::modules(teal::example_module(), label = "Study1 modules"),
  "Study2" = teal::modules(teal::example_module(), teal.modules.general::tm_variable_browser(), label="Study2 modules")
)


ui_app <- bslib::page_fluid(
  titlePanel("Dynamic teal modules"),
  selectInput("study_id", "Select study", choices = studies, multiple = FALSE),
  uiOutput("teal_ui"),
  ui_session_info("teal-footer-session_info")
)

srv_app <- function(input, output, session) {
  
  reporter <- teal.reporter::Reporter$new()
  
  current_modules <- reactive({
    req(input$study_id)
    mods[[input$study_id]]
  })
  
  current_datasets <- reactive({
    req(input$study_id)
    datasets[[input$study_id]]
  })
  
  output$teal_ui <- renderUI({
    mods <- current_modules()
    req(!is.null(mods))
    ui_teal(id = "teal", modules = mods)
  })
  
  observe({
    curr_mods <- current_modules()
    curr_ds <- current_datasets()
    req(!is.null(curr_mods), length(curr_ds) > 0)
    srv_teal("teal", data = curr_ds, modules = curr_mods, reporter = reporter)
  })
  
  srv_session_info("teal-footer-session_info")
}

shinyApp(ui_app, srv_app)

Issue

While this code works fine to dynamically switch between studies and load different data sets with study-specifc teal modules, using the report when toggling between studies leads to reporter issues and ultimately application crashes. I understand this is the development branch and reporter may be undergoing refactoring, but wanted to highlight the issue and/or get advice for correcting the code above.

Steps to reproduce

1. Run app.
2. With 'Study1' selected, click 'Add to Report' and name card.
3. Click 'Report' > 'Preview Report' to confirm card creation.
4. Switch to 'Study2'.

App crash message

Warning: Error in observe: Assertion on 'card' failed: Must inherit from class 'ReportCard'/'teal_card', but has class 'NULL'.
  88: <Anonymous>
  87: stop
  86: mstop
  85: makeAssertion
  84: checkmate::assert_multi_class
  83: observe
  82: <observer:observeEvent(input$add_card_ok)>
   3: runApp
   2: print.shiny.appobj
   1: <Anonymous>

sessionInfo()

R version 4.4.1 (2024-06-14)
Platform: aarch64-apple-darwin20
Running under: macOS Sonoma 14.7

Matrix products: default
BLAS:   /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib 
LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib;  LAPACK version 3.12.0

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

time zone: America/New_York
tzcode source: internal

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] teal.reporter_0.5.0.9009 teal.widgets_0.5.0.9000  teal_1.0.0.9009          teal.slice_0.7.0         teal.data_0.8.0.9000     teal.code_0.7.0          bslib_0.9.0             
[8] shiny_1.10.0            

loaded via a namespace (and not attached):
 [1] tidyr_1.3.1                     plotly_4.10.4                   sass_0.4.9                      generics_0.1.3                  digest_0.6.37                   magrittr_2.0.3                 
 [7] evaluate_1.0.3                  grid_4.4.1                      RColorBrewer_1.1-3              fastmap_1.2.0                   jsonlite_2.0.0                  ggrepel_0.9.6                  
[13] backports_1.5.0                 formatR_1.14                    promises_1.3.2                  httr_1.4.7                      purrr_1.0.4                     viridisLite_0.4.2              
[19] scales_1.4.0                    lazyeval_0.2.2                  jquerylib_0.1.4                 cli_3.6.5                       teal.transform_0.7.0            rlang_1.1.6                    
[25] withr_3.0.2                     cachem_1.1.0                    tools_4.4.1                     memoise_2.0.1                   checkmate_2.3.2                 dplyr_1.1.4                    
[31] ggplot2_3.5.2                   httpuv_1.6.15                   vctrs_0.6.5                     logger_0.4.1                    R6_2.6.1                        mime_0.12                      
[37] lifecycle_1.0.4                 fs_1.6.5                        htmlwidgets_1.6.4               shinycssloaders_1.1.0           fontawesome_0.5.3               shinyjs_2.1.0.9006             
[43] ggmosaic_0.3.3                  pkgconfig_2.0.3                 teal.logger_0.4.0               pillar_1.10.2                   later_1.4.1                     gtable_0.3.6                   
[49] rsconnect_1.3.4                 glue_1.8.0                      data.table_1.17.0               Rcpp_1.0.14                     teal.modules.general_0.5.0.9005 xfun_0.50                      
[55] tibble_3.2.1                    tidyselect_1.2.1                knitr_1.49                      rstudioapi_0.16.0               farver_2.1.2                    xtable_1.8-4                   
[61] bsicons_0.1.2                   htmltools_0.5.8.1               shinyWidgets_0.8.7              compiler_4.4.1

Relevant log output

Code of Conduct

• I agree to follow this project's Code of Conduct.

Contribution Guidelines

• I agree to follow this project's Contribution Guidelines.

Security Policy

• I agree to follow this project's Security Policy.

[vedhav]

Thank you @mschaffer-incyte for reporting this and providing a reprex. I was able reproduce this issue @insightsengineering/nest-core-dev

> sessionInfo()
R version 4.4.2 (2024-10-31 ucrt)
Platform: x86_64-w64-mingw32/x64
Running under: Windows 11 x64 (build 26100)

Matrix products: default


locale:
[1] LC_COLLATE=English_India.utf8  LC_CTYPE=English_India.utf8
[3] LC_MONETARY=English_India.utf8 LC_NUMERIC=C
[5] LC_TIME=English_India.utf8

time zone: Asia/Calcutta
tzcode source: internal

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base

other attached packages:
[1] teal.reporter_0.5.0.9009 teal.widgets_0.5.0.9000  teal_1.0.0.9009
[4] teal.slice_0.7.0.9000    teal.data_0.8.0.9000     teal.code_0.7.0.9001
[7] bslib_0.9.0              shiny_1.11.1

loaded via a namespace (and not attached):
 [1] tidyselect_1.2.1                viridisLite_0.4.2
 [3] dplyr_1.1.4                     farver_2.1.2
 [5] S7_0.2.0                        fastmap_1.2.0
 [7] lazyeval_0.2.2                  fontquiver_0.2.1
 [9] promises_1.3.2                  shinyjs_2.1.0
[11] digest_0.6.37                   mime_0.13
[13] lifecycle_1.0.4                 magrittr_2.0.3
[15] compiler_4.4.2                  rlang_1.1.5
[17] bsicons_0.1.2                   sass_0.4.10
[19] tools_4.4.2                     yaml_2.3.10
[21] data.table_1.17.8               knitr_1.50
[23] askpass_1.2.1                   htmlwidgets_1.6.4
[25] xml2_1.4.0                      RColorBrewer_1.1-3
[27] withr_3.0.2                     purrr_1.0.4
[29] shinyWidgets_0.9.0              rlistings_0.2.12.9001
[31] grid_4.4.2                      teal.logger_0.4.0.9001
[33] gdtools_0.4.3                   xtable_1.8-4
[35] learnr_0.11.5                   ggplot2_4.0.0.9000
[37] scales_1.4.0                    cli_3.6.4
[39] rmarkdown_2.30                  ragg_1.5.0
[41] generics_0.1.4                  sortable_0.5.0
[43] httr_1.4.7                      commonmark_2.0.0
[45] cachem_1.1.0                    teal.modules.general_0.5.0.9004
[47] assertthat_0.2.1                formatR_1.14
[49] rtables.officer_0.1.1.9000      vctrs_0.6.5
[51] jsonlite_2.0.0                  fontBitstreamVera_0.1.1
[53] ggrepel_0.9.6                   crosstalk_1.2.2
[55] systemfonts_1.2.3               plotly_4.11.0
[57] fontawesome_0.5.3               tidyr_1.3.1
[59] jquerylib_0.1.4                 rtables_0.6.13.9000
[61] formatters_0.5.11.9002          glue_1.8.0
[63] teal.transform_0.7.0.9000       DT_0.34.0
[65] stringi_1.8.7                   flextable_0.9.10
[67] gtable_0.3.6                    later_1.4.1
[69] shinycssloaders_1.1.0           tibble_3.3.0
[71] logger_0.4.1                    pillar_1.11.1
[73] ggmosaic_0.3.3                  htmltools_0.5.8.1
[75] openssl_2.3.4                   R6_2.6.1
[77] textshaping_1.0.3               rprojroot_2.1.1
[79] evaluate_1.0.5                  backports_1.5.0
[81] memoise_2.0.1                   httpuv_1.6.15
[83] fontLiberation_0.1.0            Rcpp_1.0.14
[85] zip_2.3.3                       uuid_1.2-1
[87] checkmate_2.3.3                 officer_0.7.0
[89] xfun_0.52                       fs_1.6.5
[91] pkgconfig_2.0.3

[gogonzo]

Hi @mschaffer-incyte , thanks for an issue.
Depending on how your use-case looks like I recommend to not call srv_teal multiple times with the same id within the same session. I wouldn't recommend to call any module in this way.

What I suggest is following. Call ui/srv_teal with different id, for example study_id. Thanks to this you'll have two independent teal modules with shared Reporter object. Remember to avoid calling server modules twice (see studies_called() in the code below)

  studies_called <- reactiveVal()
  observeEvent(input$study_id, {
    req(current_datasets(), current_modules())
    if (!input$study_id %in% studies_called()) {
      srv_teal(input$study_id, data = current_datasets, modules = current_modules(), reporter = reporter)
      studies_called(c(studies_called(), input$study_id))
    }
  })

working app code

library(shiny)
library(bslib)
library(teal)
library(teal.data)
library(teal.widgets)
library(teal.reporter)

options(shiny.legacy.datatable = TRUE)


# Define selectable studies with datasets and study-specific modules
studies <- c("Study1", "Study2")

datasets <- list(
  "Study1" = within(teal.data::teal_data(), { iris <- iris; mtcars <- mtcars}),
  "Study2" = within(teal.data::teal_data(), { df <- data.frame(a = 1:10, b = letters[1:10]) } )
)

mods <- list(
  "Study1" = teal::modules(teal::example_module(), label = "Study1 modules"),
  "Study2" = teal::modules(teal::example_module(), teal.modules.general::tm_variable_browser(), label="Study2 modules")
)

ui_app <- bslib::page_fluid(
  titlePanel("Dynamic teal modules"),
  selectInput("study_id", "Select study", choices = studies, multiple = FALSE),
  uiOutput("teal_ui"),
  ui_session_info("teal-footer-session_info")
)

srv_app <- function(input, output, session) {
  
  reporter <- teal.reporter::Reporter$new()
  
  current_modules <- reactive({
    req(input$study_id)
    mods[[input$study_id]]
  })
  
  current_datasets <- reactive({
    req(input$study_id)
    datasets[[input$study_id]]
  })
  
  output$teal_ui <- renderUI({
    mods <- current_modules()
    req(!is.null(mods))
    ui_teal(id = session$ns(input$study_id), modules = mods)
  })
  
  studies_called <- reactiveVal()
  observeEvent(input$study_id, {
    req(current_datasets(), current_modules())
    if (!input$study_id %in% studies_called()) {
      srv_teal(input$study_id, data = current_datasets, modules = current_modules(), reporter = reporter)
      studies_called(c(studies_called(), input$study_id))
    }
  })
  
  srv_session_info("teal-footer-session_info")
}

shinyApp(ui_app, srv_app)

Hope it helps 👍

[mschaffer-incyte]

Great! Thanks for looking into this and for the clear solution. It makes perfect sense.

[mschaffer-incyte]

@gogonzo , a follow up for this relates to defining a global filter per study. The code below works initially to set a filter but doesn't after switching studies. It seems global filters are managed at the application level. Is there a way to define filters at the study level when creating the srv_teal?

code below

library(shiny)
library(bslib)
library(teal)
library(teal.slice)
library(teal.data)
library(teal.widgets)
library(teal.reporter)


studies <- c("Study1", "Study2")

datasets <- list(
  "Study1" = within(teal.data::teal_data(), { iris <- iris; mtcars <- mtcars}),
  "Study2" = within(teal.data::teal_data(), { df <- data.frame(a = 1:10, b = letters[1:10]) } )
)

mods <- list(
  "Study1" = teal::modules(teal::example_module(), label = "Study1 modules"),
  "Study2" = teal::modules(teal::example_module(), teal.modules.general::tm_variable_browser(), label="Study2 modules")
)

filters <- list(
  "Study1" = teal_slices(teal_slice(dataname="iris", id = "Species setosa", expr = "Species == 'setosa'", title = "Species setosa")),
  "Study2" = teal_slices()
)



ui_app <- bslib::page_fluid(
  titlePanel("Dynamic teal modules"),
  selectInput("study_id", "Select study", choices = studies, multiple = FALSE),
  uiOutput("teal_ui"),
  ui_session_info("teal-footer-session_info")
)

srv_app <- function(input, output, session) {
  
  reporter <- teal.reporter::Reporter$new()
  
  current_datasets <- reactive({
    req(input$study_id)
    datasets[[input$study_id]]
  })
  
  current_modules <- reactive({
    req(input$study_id)
    mods[[input$study_id]]
  })
  
  current_filter <- reactive({
    req(input$study_id)
    filters[[input$study_id]]
  })
  
  output$teal_ui <- renderUI({
    mods <- current_modules()
    req(!is.null(mods))
    ui_teal(id = session$ns(input$study_id), modules = mods)
  })
  
  srv_teal_studies <- reactiveVal()
  
  observeEvent(input$study_id, {
    req(current_datasets(), current_modules())
    if (!input$study_id %in% srv_teal_studies()) {
      srv_teal(input$study_id, data = current_datasets, modules = current_modules(), filter = current_filter(), reporter = reporter)
      srv_teal_studies(c(srv_teal_studies(), input$study_id))
    }
  })
  
  srv_session_info("teal-footer-session_info")
}

shinyApp(ui_app, srv_app)

[mschaffer-incyte]

Ah, nevermind. The issue was the data = current_datasets should be data = current_datasets().

[mschaffer-incyte]

@gogonzo , there is some still unexpected behavior with this approach. Running the code below allows the app to initially switch between studies with corresponding datasets, modules and filters defined. However, switching from Study1 to Study 2, then back to Study1 leaves the plot and all configuration options blank when returning to Study1.

Any thoughts on why this may be happening when using data = current_datasets()? If I switch back to data = current_datasets, then the app crashes when switching studies.

It seems there are a lot of potential reactive complications with this pattern. Is there a better alternative for selecting and switching between studies with a clean teal environment?

code below

library(shiny)
library(bslib)
library(teal)
library(teal.slice)
library(teal.data)
library(teal.widgets)
library(teal.reporter)
library(teal.modules.general)

studies <- c("Study1", "Study2", "Study3")

datasets <- list(
  "Study1" = function() { 
    my_data <- within(teal_data(), {
      ADSL <- example_cdisc_data("ADSL")
      ADLB <- example_cdisc_data("ADLB")
      ADTTE <- example_cdisc_data("ADTTE")
    })
    
    datanames <- c("ADSL", "ADLB", "ADTTE")
    join_keys(my_data) <- default_cdisc_join_keys[datanames]
    my_data
  },
  "Study2" = function() { within(teal.data::teal_data(), { iris <- iris; mtcars <- mtcars}) },
  "Study3" = function() { within(teal.data::teal_data(), { mtcars <- data.frame(a = 1:10, b = letters[1:10]) } ) }
)

mods <- list(
  "Study1" = teal::modules(
    teal.modules.general::tm_g_scatterplot(
      label = "Scatterplot Choices",
      x = data_extract_spec(dataname = "ADSL",select = select_spec(label = "Select variable:",choices = variable_choices("ADSL", c("AGE", "BMRKR1", "BMRKR2")),selected = "AGE",multiple = FALSE,fixed = FALSE)),
      y = data_extract_spec(dataname = "ADSL",select = select_spec(label = "Select variable:",choices = variable_choices("ADSL", c("AGE", "BMRKR1", "BMRKR2")),selected = "BMRKR1",multiple = FALSE,fixed = FALSE))
    ),
    teal.modules.general::tm_variable_browser(),
    label="Study1 modules"),
  "Study2" = teal::modules(teal.modules.general::tm_variable_browser(), teal::example_module(), label = "Study2 modules"),
  "Study3" = teal::modules(teal.modules.general::tm_variable_browser(), teal::example_module(), label = "Study3 modules")
)

filters <- list(
  "Study1" = teal_slices(),
  "Study2" = teal_slices(teal_slice(dataname="iris", id = "Species setosa", expr = "Species == 'setosa'", title = "Species setosa"), module_specific = TRUE),
  "Study3" = teal_slices()
)



ui_app <- bslib::page_fluid(
  titlePanel("Dynamic teal modules"),
  selectInput("study_id", "Select study", choices = studies, multiple = FALSE),
  uiOutput("teal_ui"),
  ui_session_info("teal-footer-session_info")
)

srv_app <- function(input, output, session) {
  
  reporter <- teal.reporter::Reporter$new()
  
  current_datasets <- reactive({
    req(input$study_id)
    datasets[[input$study_id]]()
  })
  
  current_modules <- reactive({
    req(input$study_id)
    mods[[input$study_id]]
  })
  
  current_filter <- reactive({
    req(input$study_id)
    filters[[input$study_id]]
  })
  
  output$teal_ui <- renderUI({
    mods <- current_modules()
    req(!is.null(mods))
    ui_teal(id = session$ns(input$study_id), modules = mods)
  })
  
  srv_teal_studies <- reactiveVal()
  
  observeEvent(input$study_id, {
    req(current_datasets(), current_modules())
    if (!input$study_id %in% srv_teal_studies()) {
      srv_teal(input$study_id, data = current_datasets(), modules = current_modules(), filter = current_filter(), reporter = reporter)
      srv_teal_studies(c(srv_teal_studies(), input$study_id))
    }
  })
  
  srv_session_info("teal-footer-session_info")
}

shinyApp(ui_app, srv_app)

[gogonzo]

@mschaffer-incyte - I see know what is the problem here. ui_teal is re-rendered, so it will always reset the whole UI to default state. Default state is in most of the modules empty and eventually it would be set by server, but here server has been already run, so it won't happen.

I don't have a ready solution for you. I probably need to know more about your use-case. You wrote

use predefined teal data sets and teal modules based on a selected study name.

Are teal data and teal modules associated with study-name known before an app start? Is it something you can hardcode and just create an app containing static UI with conditional panels for study?

If not, then you probably need to create mechanism which doesn't re-render ui_teal(input$studyid), but adds this UI once via insertUI, and then switching between elements is done using toggle or show/hide or tabsetPanel.