开发者

Using tryCatch and source

开发者 https://www.devze.com 2023-02-15 18:49 出处:网络
Suppose I have two R files: correct.R and broken.R. What would be the best way of using tryCatch to check for errors?

Suppose I have two R files: correct.R and broken.R. What would be the best way of using tryCatch to check for errors?

Currently, I have

> x = tryCatch(source("broken.R"), error=function(e) e)
> x
 <simpleError in source("broken.R"): test.R:2:0: unexpected end of input
  1: x = {
     ^>
> y = tryCatch(source("correct.R"), error=function(e) e)
> y
 $value
 [1] 5

 $visible
 [1] FA开发者_StackOverflowLSE

However, the way I've constructed the tryCatch means that I have to interrogate the x and y objects to determine if there has been an error.

Is there a better way of doing this?


The question comes from teaching. 100 students upload their R scripts and I run the scripts. To be nice, I'm planning on creating a simple function that determines if their function sources correctly. It only needs to return TRUE or FALSE.


Try this:

> tryCatch(stop("foo"), error = function(e) {
+ cat(e$message, "\n")
+ FALSE
+ })
foo 
[1] FALSE

Alternatively, you should consider Hadley's testthat package:

> expect_that(stop("foo"), is_a("numeric"))
Error in is.vector(X) : foo


To expand upon mdsumner's point, this is a simple implementation.

sources_correctly <- function(file)
{
  fn <- try(source(file))
  !inherits((fn, "try-error"))
}


Maybe I'm underthinking this, but since you're just looking for the boolean, you can just test for the existence of the $visible:

y <- tryCatch(source("broken.R"), error=function(e) e)
works <- !is.null(y$visible) #y$visible would be null if there were an error

Does that solve what you're looking for? You could wrap it in a loop (or use lapply) like:

for(i in 1:length(students)) {
  works[i] <- !is.null(tryCatch(source(student_submissions[i]), error=function(e) e)$visible)
}
0

精彩评论

暂无评论...
验证码 换一张
取 消