splitPathFile: Analyze pathfile-strings

In OHDSI/ff: Memory-Efficient Storage of Large Data on Disk and Fast Access Functions

Description

splitPathFile splits a vector of pathfile-strings into path- and file-components without loss of information. unsplitPathFile restores the original pathfile-string vector. standardPathFile standardizes a vector of pathfile-strings: backslashes are replaced by slashes, except for the first two leading backslashes indicating a network share. tempPathFile returns - similar to tempfile - a vector of filenames given path(s) and file-prefix(es) and an optional extension. fftempfile returns - similar to tempPathFile - a vector of filenames following a vector of pathfile patterns that are intrepreted in a ff-specific way.

Usage

splitPathFile(x)
unsplitPathFile(splitted)
standardPathFile(x)
tempPathFile(splitted=NULL, path=splitted$path, prefix=splitted$file, extension=NULL)
fftempfile(x)

Arguments

x	a character vector of pathfile strings
splitted	a return value from splitPathFile
path	a character vector of path components
prefix	a character vector of file components
extension	optional extension like "csv" (or NULL)

Details

dirname and basename remove trailing file separators and therefore cannot distinguish pathfile string that contains ONLY a path from a pathfile string that contains a path AND file. Therefore file.path(dirname(pathfile), basename(pathfile)) cannot always restore the original pathfile string.
splitPathFile decomposes each pathfile string into three parts: a path BEFORE the last file separator, the file separator, the filename component AFTER the last file separator. If there is no file separator in the string, splitPathFile tries to guess whether the string is a path or a file component: ".", ".." and "~" are recognized as path components. No tilde expansion is done, see path.expand. Backslashes are converted to the current .Platform$file.sep using splitPathFile except for the first two leading backslashes indicating a network share.
unsplitPathFile restores the original pathfile-string vector up to translated backslashes.
tempPathFile internally uses tempfile to create its filenames, if an extension is given it repeats filename creation until none of them corresponds to an existing file.
fftempfile takes a path-prefix pattern as input, splits it, will replace an empty path by getOption("fftempdir") and will use getOption("ffextension") as extension.

Value

A list with components

path	a character vector of path components
fsep	a character vector of file separators or ""
file	a character vector of file components

Note

There is no gurantee that the path and file components contain valid path- or file-names. Like basename, splitPathFile can return ".", ".." or even "", however, all these make sense as a prefix in tempPathFile.

Author(s)

Jens Oehlschlägel

See Also

tempfile, dirname, basename, file.path

Examples

  pathfile <- c("", ".", "/.", "./", "./.", "/"
  , "a", "a/", "/a", "a/a", "./a", "a/.", "c:/a/b/c", "c:/a/b/c/"
  , "..", "../", "/..", "../..", "//", "\\\\a\\", "\\\\a/"
  , "\\\\a/b", "\\\\a/b/", "~", "~/", "~/a", "~/a/")
  splitted <- splitPathFile(pathfile)
  restored <- unsplitPathFile(splitted)
  stopifnot(all(gsub("\\\\","/",restored)==gsub("\\\\","/",pathfile)))

  dirnam <- dirname(pathfile)
  basnam <- basename(pathfile)

  db  <- file.path(dirnam,basnam)
  ident = gsub("\\\\","/",db) == gsub("\\\\","/",pathfile)
  sum(!ident)

  do.call("data.frame", c(list(ident=ident, pathfile=pathfile
   , dirnam=dirnam, basnam=basnam), splitted))

  ## Not run: 
    message("show the difference between tempfile and fftempfile")
    do.call("data.frame", c(list(ident=ident, pathfile=pathfile, dirnam=dirnam, basnam=basnam)
, splitted, list(filename=tempPathFile(splitted), fftempfile=fftempfile(pathfile))))

    message("for a single string splitPathFile is slower, 
for vectors of strings it scales much better than dirname+basename")

    system.time(for (i in 1:1000){
      d <- dirname(pathfile)
      b <- basename(pathfile)
    })
    system.time(for (i in 1:1000){
      s <- splitPathFile(pathfile)
    })

    len <- c(1,10,100,1000)
    timings <- matrix(0, 2, length(len), dimnames=list(c("dir.base.name", "splitPathFile"), len))
    for (j in seq(along=len)){
      l <- len[j]
      r <- 10000 / l
      x <- rep("\\\\a/b/", l)
      timings[1,j] <- system.time(for (i in 1:r){
          d <- dirname(x)
          b <- basename(x)
        })[3]
      timings[2,j] <- system.time(for (i in 1:r){
          s <- splitPathFile(x)
        })[3]
    }
    timings
  
## End(Not run)

OHDSI/ff documentation built on May 7, 2019, 8:30 p.m.