R Shiny Object not found - Evaluating a strings as code - r

Question

I am trying to creat a shiny app which takes strings within a column of a data frame that are pieces of R code, and evaluate those against data frames which have been generated in the app. Below is a working reprex of the code outside of the shiny app:
## create df with eval expressions
code_df <- data.frame(desired_outcome = c("this should be true",
"this should be false",
"this will be true or false"),
code_string = c('nrow(random_df) > 0',
'nrow(random_df) == 0',
'nrow(random_df) >= 100'),
stringsAsFactors = F)
# generate a dataframe with 1-150 rows
random_df <- data.frame(rand_binary = sample(0:1,sample(1:150, 1),rep=TRUE))
## helper function for sapply
eval_parse <- function(x){
eval(parse(text = x))
}
## evaluate code strings
tf_vector <- sapply(code_df$code_string, eval_parse)
## add data to original df
code_df$nrow <- nrow(random_df)
code_df$tf <- tf_vector
code_df
If you run the code above, it will generate a 'random_df' with between 1-150 rows, then evaluate the code strings from code_df. This code works as intended.
The problem arises when I try to implement this in shiny (code below), the implementation returns "Error: object 'random_df' not found" when the action button is clicked.
One other wrinkle: If you run the non-shiny reprex code first, and do not clean the environment before you run the shiny app, the app will return the table, but it evaluates the code strings based on the non-shiny "random_df", not the newly randomly generated one from the shiny app. You can see this based on the fact that the 'nrow' column will change in value, while the 'tf' will not change.
server.R
library(shiny)
code_df <- data.frame(desired_outcome = c("this should be true", "this should be false", "this will be true or false"),
code_string = c('nrow(random_df) > 0', 'nrow(random_df) == 0', 'nrow(random_df) >= 100'),
stringsAsFactors = F)
## helper function for sapply
eval_parse <- function(x){
eval(parse(text = x))
}
# Define server logic required to draw a histogram
shinyServer(function(input, output) {
new_code_df <- eventReactive(input$newDF,{
# create data.frame
random_df <- data.frame(rand_binary = sample(0:1,sample(1:150, 1),rep=TRUE))
##
tf_vector <- sapply(code_df$code_string, eval_parse)
code_df$nrow <- nrow(random_df)
code_df$tf <- tf_vector
code_df
})
output$randomdf <- renderTable({new_code_df()})
})
ui.R
#
# This is the user-interface definition of a Shiny web application. You can
# run the application by clicking 'Run App' above.
#
# Find out more about building applications with Shiny here:
#
# http://shiny.rstudio.com/
#
library(shiny)
# Define UI for application that draws a histogram
shinyUI(fluidPage(
# Application title
titlePanel("Eval Code from Data Frame"),
sidebarLayout(
sidebarPanel(
actionButton("newDF","Generate New Dataframe")
),
mainPanel(
tableOutput('randomdf')
)
)
))

Answer 1

Functions in R (and therefore shiny) are lexically scoped. This mean that functions can only see the variables defined in the environment where they themselves are defined. You are defining eval_parse in the global environment but random_df is defined in the shiny server function. This the former cannot see the latter because random_df is not in the gloabl enviroment like it was in your non-shiny example.
If you want to make all the server variables available to your expression, you can specify an environment to eval(). First change the helper so you can pass an environment
eval_parse <- function(x, env=parent.frame()){
eval(parse(text = x), envir=env)
}
and then change your server code to pass along the function environment
tf_vector <- sapply(code_df$code_string, eval_parse, env=environment())