How to persistently store value of an object instead of its name?

问题

This is in continuation of my related question: Error when trying to interactively load data file saved by paused batch script. I decided to present my question with a reproducible example separately to avoid making the already large description in the previous question even bigger. In the following reproducible example, I expect to retrieve the value of the stored object ("Important data"), but instead, as you see, I retrieve the name of the object itself ("sf.data.devLinks"). I suspected that it could be because of me using as.name, but I tested additionally a primitive example in an interactive session and as.name worked fine. I also, as you see, have tried using eval and substitute, but it didn't help.

library(RCurl)

info <- "Important data"
ATTR <- "SQL"
request <- "SELECT info FROM topSecret"
dataName <- "sf.data.devLinks"
rdataFile <- "/tmp/testAttr.rds"

save <- TRUE

getData <- function() {
  return (info)
}

requestDigest <- base64(request)

# check if the archive file has already been processed
message("\nProcessing request \"", request, "\" ...\n")

# read back the object with the attribute
if (file.exists(rdataFile)) {
  # now check if request's SQL query hasn't been modified
  data <- readRDS(rdataFile)
  message("Retrieved object '", data, "', containing:\n")
  message(str(data))

  requestAttrib <- attr(data, ATTR, exact = TRUE)
  if (is.null(requestAttrib)) {
    message("Object '", data, "' doesn't have attribute \"",
            ATTR, "\"\n")
  }
  else {
    message("Object '", data, "' contains attribute ", ATTR, ":\n\"",
            base64(requestAttrib), "\"\n")

    if (identical(requestDigest, requestAttrib)) {
      message("Processing skipped: RDS file is up-to-date.\n")
      save <- FALSE
      return
    }
  }
  rm(data)
}

if (save) {
  message("Saving results of request \"",
          request, "\" as R data object ...\n")

  assign(dataName, getData())
  data <- as.name(dataName)
  #eval(substitute(assign(dataName, getData()),
  #                list(data <- as.name(dataName))))

  # save hash of the request's SQL query as data object's attribute,
  # so that we can detect when configuration contains modified query
  attr(data, ATTR) <- base64(request)

  # save current data frame to RDS file
  saveRDS(data, rdataFile)
}

Please note that testing this code requires running it twice (first run - to store the object, second - to retrieve).

回答1:

The problem is in your use of as.name, not in the code to save the object. This works perfectly fine:

data        <- 1:10
object.name <- 'data.name'
query       <- 'SELECT * FROM TABLE'
file        <- tempfile()
assign(object.name, structure(data, SQL = query))
saveRDS(get(object.name), file)
read.object <- readRDS(file)
identical(read.object, get(object.name))

You're creating a name object, and assigning attributes to it, but you're expecting the data to be there. It won't be, a symbol is just a pointer to the value. You need to use eval() or something similar to get the value from the symbol.

回答2:

Finally, I was able to return to this question. I have figured out the correct solution, so I'm answering my own question. The answer here is based on my reproducible example, but I have made corresponding changes in my more complex real-world R code. The solution is rather simple, but twofold, as follows.

Replace the original code data <- readRDS(rdataFile) with assign(dataName, readRDS(rdataFile)).
Replace the original code as.name(dataName) with get(dataName). An alternative to get() here is eval(parse(text=dataName)), which IMHO is more cumbersome.

Below I provide the solution's full source code, based on the original reproducible example. I don't provide the script's output, which is easy to reproduce (remember to run script at least twice). Again, thanks to everyone who helped with this question.

library(RCurl)

info <- "Important data"
ATTR <- "SQL"
request <- "SELECT info FROM topSecret"
dataName <- "sf.data.devLinks"
rdataFile <- "/tmp/testAttr.rds"

save <- TRUE

getData <- function() {
  return (info)
}

requestDigest <- base64(request)

# check if the archive file has already been processed
message("\nProcessing request \"", request, "\" ...\n")

# read back the object with the attribute
if (file.exists(rdataFile)) {
  # now check if request's SQL query hasn't been modified
  assign(dataName, readRDS(rdataFile))
  message("Retrieved object '", dataName, "', containing:\n")
  message(str(get(dataName)))

  requestAttrib <- attr(get(dataName), ATTR, exact = TRUE)
  if (is.null(requestAttrib)) {
    message("Object '", dataName, "' doesn't have attribute \"",
            ATTR, "\"\n")
  }
  else {
    message("Object '", dataName, "' contains attribute \"",
            ATTR, "\":\n\"", base64(requestAttrib), "\"\n")

    if (identical(requestDigest, requestAttrib)) {
      message("Processing skipped: RDS file is up-to-date.\n")
      save <- FALSE
      return
    }
  }
}

if (save) {
  message("Saving results of request \"",
          request, "\" as R data object ...\n")

  assign(dataName, getData())
  message(str(dataName))
  data <- get(dataName)
  # alternative to using get(), but more cumbersome:
  # data <- eval(parse(text=dataName))

  # save hash of the request's SQL query as data object's attribute,
  # so that we can detect when configuration contains modified query
  attr(data, ATTR) <- base64(request)

  message(str(data))

  # save current data frame to RDS file
  saveRDS(data, rdataFile)
}

来源：https://stackoverflow.com/questions/23739359/how-to-persistently-store-value-of-an-object-instead-of-its-name

标签

variables

object

persistence

metaprogramming