Issue with digest function with files on network drive

[zross]

I'm getting an error running the digest() function on a file stored on a network drive from a Windows computer while no error occurs on the same file from the local drive.

So the following works perfectly and C is a local drive:

fname <- "C:/junk/all_sites.prj"
h2 <- digest(fname, algo="md5", file=TRUE)

With an identical file moved to a network drive I get an error:

fname <- "X:/junk/all_sites.prj"
h2 <- digest(fname, algo="md5", file=TRUE)
Error in digest(fname, algo = "md5", file = TRUE) : 
  The specified file is not readable: X:/junk/all_sites.prj

This is causing problems with the deployApp() function in Shiny and the document() function in devtools but report digest errors. I reported this to RStudio (thinking it was a Shiny issue).

It seems this may be due to the file.access function used by digest. The code:

file.access(fname, mode=4)

returns a 0 (success) using the local path and a -1 (failure) with the network path.

R version 3.1.2 (2014-10-31)
Platform: x86_64-w64-mingw32/x64 (64-bit)
digest_0.6.4

[eddelbuettel]

Methinks that is an OS issue. We do nothing but opening and reading / writiing files. We don't even know whether the file is local or not. So in brief if this fails you may just have to work on a local file and (as needed) move it.

The help page for file.access mentions no issues related to network drives, so maybe you need to sort this out at your end?

[zross]

Thanks for the quick response. As you suggest, I'll sort this out on my end.

[marcelboldt]

I have the same problem... Seems Windows related: if I call the windows method _access directly I get the same result:

library(inline)
inc <- "
#include <io.h>
"

src <- "
int res;
res = _access( \"d:\\path\\file.txt\" ,4);
return(wrap(res));
"

acc <- cxxfunction(signature(),
               body=src, 
               includes=inc,
               plugin="Rcpp")
acc()

# 0 for my C: drive
# -1 for a network drive (file system shown by Windows 7: CSC-Cache), currently in offline mode

[eddelbuettel]

Still don't think this is my issue. "Doctor it hurts when I do this ..." -- just don't place the to-be-digested files on a network drive if the OS does not fully support it.

[caerbannogwhite]

Unfortunately, there are circumstances in which it is impossible (or strongly discouraged) to move or copy the data from the network drive (ie, when working with sensible data).
Since it's still possible to read the files with any other R library, my suggestion (for MD5) here is:

content <- readBin(p, n = 1000000, what="raw")
md5 <- digest::digest(content, serialize = FALSE, algo = "md5")

edit: solution improved