当需要清除状态时，如何优雅地处理R中的错误？ [英] How to elegantly handle errors in R when state needs to be cleaned up?

问题描述

处理R中的错误的最佳方法是什么？我希望既可以从脚本的最终用户那里提取堆栈跟踪信息，又可以清理我可能正在使用的所有临时变量和状态。

What is the best was to handle errors in R? I want to be able to both abstract away stack traces from the end user of my script, as well as clean up any temporary variables and state i may be working with.

因此，我想我的问题有两个方面：

So i suppose my question is two fold:

1. 最重要的是，在处理错误时如何正确处理清理状态？


2. 如何在不吐出内部
   东西的情况下从R的错误消息中提取有用的好信息？

此刻，我有一些类似的东西，但是（1）似乎太糟糕了，（2）仍然给了我可怕的错误消息。

At the moment, I have something which looks like this, but (1) it seems horribly inelegant, and (2) still gives me horrible error messages.

# Create some temporary working state and variables to clean up
file <- "somefile"
working.dir <- getwd()
setwd("../")  # Go somewhere else
saf.default <- getOption("stringsAsFactors")
options(stringsAsFactors = FALSE)

# Now lets try to open the file, but it doesn't work (for whatever reason)
# I want to clean up the state, stop, and wrap the error string with a nicer
# message
tryCatch({
  # Need to embed the tryCatch because we still need some of the state variables
  # for the error message
  tryCatch({
    f <- read.table(file)
  }, error = function(err.msg) {
    # Can't clean up here, we still need the `file variable!
    stop("Could not open file '", file, "' with error message:\n", print(err.msg), 
         call.=FALSE)
  })
}, error = function(err.msg) {
  # Now we can clean up state
  setwd(working.dir)
  options(stringsAsFactors = saf.default)
  rm(file, working.dir, saf.default, 
     envir=globalenv())  # This also seems awful?
  stop(print(err.msg), call.=FALSE)
})

# Do more stuff, get more state, handle errors, then clean up.
# I.e can't use `finally` in previous tryCatch!

由此产生的错误消息仍然是许多丑陋的内部元素：

The error message from this comes out as, still lots of ugly internals:

# <simpleError in file(file, "rt"): cannot open the connection>
# <simpleError: Could not open file 'somefile' with error message:
# Error in file(file, "rt"): cannot open the connection
>
# Error: Could not open file 'somefile' with error message:
# Error in file(file, "rt"): cannot open the connection
# In addition: Warning messages:
# 1: In file(file, "rt") :
#   cannot open file 'somefile': No such file or directory
# 2: In stop(print(err.msg), call. = FALSE) :
#   additional arguments ignored in stop()
>

推荐答案

我会将任何状态更改代码隔离到自己的代码中函数，并使用 on.exit 。这样可以保证无论是否发生错误，都将进行清理。

I would isolate any state-changing code into its own function, and use on.exit. This guarantees that cleanup will happen, no matter if an error occurs.

readFile <- function(.....)
{
    on.exit({
        setwd(cur.dir)
        options(stringsAsFactors=saf)
    })
    cur.dir <- getwd()
    saf <- getOption("stringsAsFactors")
    setwd("new/dir")
    options(stringsAsFactors=FALSE)
    ....
}

这篇关于当需要清除状态时，如何优雅地处理R中的错误？的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！